JOT June 2019 PDF

IN MEMORIAM
In Memoriam: AONA Tribute to Jeffrey

Welling Mast, MD (1940–2019)
spending 2 years in Bern, Switzerland. The time he spent
in Bern resulted in a career-long friendship with Reinhold
Ganz. This friendship resulted in development of a safe
method of performing periacetabular osteotomy and bio-
Downloaded from https://journals.lww.com/jorthotrauma by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3LZS6UAPyeTOgLA+NOSjkP/NRV4u7Qy0q7mdYbyTOsGoP5l/nLr+/tw== on 06/01/2019
regenerative surgical treatment of hip disease, which

created tremendous opportunities to improve patients with
congenital hip dysplasia and Perthes disease. As a result of
the friendship and collaboration with Ganz, these 2
surgeons became the top 2 contemporary hip preservation
and reconstructive surgeons in the world. Jeff’s career
was focused on developing and learning new techniques
and perfecting them. Jeff also had a close friendship with
another contemporary orthopaedic icon, the esteemed ace-
tabular surgeon, Emile Letournel. He would travel exten-
sively with Emile and his “Bande of Letournel” worldwide,
teaching their craft.
In 1989, together with Drs. Ganz and Jakob, Jeff
wrote the book “Planning and Reduction Techniques in
As we reflect on important people we encounter in Fracture Surgery.” This book singularly changed orthopae-
our lives, one singular individual usually stands out above dic trauma surgery. The book’s deviation from rigid AO
all others, as having the most dramatic effect on the course doctrine was initially sharply criticized by previous AO
and direction we have chosen in our professional careers. generations as heresy. These authors, using incredibly
To many of us who have spent our careers in the AO detailed imaging, promoted the concepts of preoperative
Foundation, that individual was Jeffrey Welling Mast, MD. planning, using the plate implant as a reduction instrument,
It is a great honor to write this tribute for our long-term developing the idea of minimally invasive surgery, resto-
mentor, critic, teacher, and most important of all, our very ration of extremity length and axis implants were used to
close friend. perform indirect reduction and bridging of comminuted
Jeffrey Welling Mast was born on September 7, fractures, emphasizing soft tissue preservation. Today,
1940, in Easton, New York. Jeff was the son of a renowned the study of this book is essential to the training of the
artist/educator, Gerald Mast. Jeff’s legendary preopera- modern orthopaedic trauma surgeon.
tive planning and operative skills reflected this artistic Jeff’s uncanny ability to draw the skeleton, the perfect
background. An outstanding athlete, he was an expert osteotomy angle of correction, the appropriate implant size
skier and was a walk-on to the University of Colorado and dimension (usually a blade plate), and then be able to
football team. His career included an early Peace Corps overlay his preoperative plan on the postoperative radiograph
participation in Vietnam and Tibet into which he enlisted to be exactly perfect to his final surgical result, without using
following a medicine internship. A random medical en- templates, was mesmerizing.
counter with the wife of a hotel owner in Kathmandu was The consummate educator, Jeff was frequently
responsible for Jeff finding his place in surgery. As head of observed to be the last person to leave the classroom or
the Peace Corps in the Southeast Asia area, and later direc- teaching laboratory, where he had spent time addressing a
tor of health care efforts in Tibet, Jeff was asked to accom- range of questions, from the easiest to the most complex
pany the wife of a local Kathmandu Russian hotel owner on operative technique. He was more often than not the teacher’s
an air evacuation medical mission to seek specialized care teacher. In his later years, he would not hesitate to travel any
in Bern, Switzerland. Jeff was introduced to Maurice distance to help his disciples with difficult surgical cases and
Muller, where he became fascinated with Muller’s surgical preoperative planning. These types of visits were unique to
skills, ultimately influencing him to enter into orthopaedic Jeff and an invaluable surgical and social experience for all of
surgery and specialize in fracture fixation, nonunion, bio- us lucky enough to have “The Master” present in our oper-
regenerative hip salvage, and osteotomy of hip and knee ating rooms. We, who knew him personally, as well as the
deformities. thousands of students he taught throughout his professional
After completing orthopaedic training at University life, are indebted to this man for his erudite, visionary teach-
of Southern California Los Angeles County Hospital, he ing, his pursuit in the achievement of the perfect operative
had various practice locations and training, eventually result, and his unending courage to take on increasingly more
J Orthop Trauma Volume 33, Number 6, June 2019 www.jorthotrauma.com | 267
Copyright © 2019 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Johnson et al J Orthop Trauma Volume 33, Number 6, June 2019
difficult cases. His impact on the advancement of orthopaedic Eric E. Johnson, MD*
trauma and reconstructive surgery cannot be measured. Keith A. Mayo, MD†
With Jeff’s passing, we realize that the AO Founda- Joel M. Matta, MD‡
tion and AO North America’s “Camelot Era” has lost its Brett Bolhofner, MD§
“Knight in Shining Armor.” This is an end of an era when *Department of Orthopaedic Surgery,
absolute excellence was always expected. Every implant, Davis Geffen School of Medicine,
screw, plate, or nail had to be in the correct position, the University of California,
correct axis with the correct amount of compression to Los Angeles, Los Angeles CA;
obtain the approval of “The Master.” It is our wish that †Hansjoerg Wyss Hip and Pelvis Center,
his memory be never forgotten, that his preoperative plan- Swedish Hospital, Seattle, WA;
ning skills still be taught at every AO course, and the ‡The Steadman Cinic,
recognition of his accomplishments and techniques be pre- The Steadman Philippon,
served for future generations. We all stand on Jeffrey Research Center, Vail, CO; and
Mast’s shoulders when we treat every orthopaedic trauma §All Florida Orthopaedic Associates,
and reconstructive patient. St. Petersburg, FL.
268 | www.jorthotrauma.com Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.
ORIGINAL ARTICLE
Is the Caprini Score Predictive of Venothromboembolism

Events in Orthopaedic Fracture Patients?
Jesse Dashe, MD, Robert L. Parisien, MD, Matthew Pina, MD, Anthony F. De Giacomo, MD,
and Paul Tornetta III, MD
type. The Caprini score may help identify patients who may require
Objective: Explore the validity of the Caprini Score in orthopaedic increased protection.
patients with lower-extremity fractures.
Key Words: VTE, PE, DVT, Caprini Score, fracture, anticoagulation
Design: Retrospective cohort study.
Level of Evidence: Prognostic Level III. See Instructions for
Setting: Level I trauma academic medical center. Authors for a complete description of levels of evidence.
Patients/Participants: Eight hundred forty-eight patients with (J Orthop Trauma 2019;33:269–276)
lower-extremity fractures from 2002 to 2015 with exclusion criteria:
minors, follow-up less than 30 days. INTRODUCTION
The topic of venothromboembolisms (VTEs) in ortho-
Intervention: Stratify patients into 2 groups: high-risk (pelvic and paedic surgery is controversial, in that there are differing
acetabular fractures) and low-risk groups (isolated foot and ankle expert opinions and no validated universal risk assessment
fractures). tool.1 Clinical practice varies by orthopaedic provider to
determine the risk of developing a VTE and appropriate
Main Outcome: Caprini Score, fracture classification, length of
selection of prophylactic treatment regimen. The provider
follow-up, deep vein thrombosis (DVT) chemoprophylaxis, and
variation ranges from intermittent pneumatic compressive
venothromboembolism (VTE) events [DVT and/or pulmonary
devices to various pharmaceutical agents: aspirin, low
embolism (PE)] diagnosed with objective testing.
molecular weight heparin, warfarin, factor Xa inhibitors,
Results: Eight hundred forty-eight patients (499 M; 349 F) 18– and other novel anticoagulants.
93 years of age (average 43.7) with average body mass index of In general, orthopaedic surgery patients with lower-
29. Three hundred high-risk and 548 low-risk patients with no differ- extremity injuries are at increased risk for VTEs.1–7 Despite
ences in demographics with average follow-up of 288 days. There the fact that these patients are recognized to be at higher risk
were 33 (3.9%) VTE events, which were more common in the high- for VTE, no clear metric currently exists to distinguish or
risk group (8%: 9 DVT, 15 PE) than the low-risk group (1.6%: 8 stratify these orthopaedic patients relative to one other. There
DVT, 1 PE) (P , 0.0001). The cutoff that best-predicted VTE events is published evidence identifying specific risk factors associ-
based on receiver-operating curves was 12 (c = 0.74) in the high-risk ated with a higher risk of developing a VTE, but there have
group, 11 (c = 0.79) in the low-risk group, and 12 (c = 0.83) overall. been no clear universal recommendations presented on how
to use these factors to guide clinical decision-making.3,8,9
Conclusion: There was a significant lower VTE rate found in the However, there is one widely accepted model named
low-risk group, but the Caprini prediction model was not significantly the Caprini Score with an established history and utilization
different between the 2 groups. This displays that patient factors play as a reliable predictive VTE risk assessment tool outside the
a large role in the development of VTE events independent of injury field of orthopaedics. The Caprini11 Score is calculated by
adding together the point values for various patient risk
factors; the assigned point value for each risk factor is
Accepted for publication January 16, 2019. derived from previous research that investigated each risk
From the Boston Medical Center, Boston University, Boston, MA.
Funding from quality improvement grants at Boston Medical Center. factor and its association of developing a VTE (Table 1). It
The authors report no conflict of interest. has been validated and is regularly used in numerous other
Presented as poster at Annual Meeting of the American Orthopaedic surgical fields including general, vascular, plastic, urologic,
Association, June 26–29, 2017, Charlotte, NC, and the Annual Meeting and head and neck surgery and may be applicable to ortho-
of the Orthopaedic Trauma Association, October 5–8, 2016, National paedic surgery as well. This risk assessment model is
Harbor, MD.
Supplemental digital content is available for this article. Direct URL citations unique, in that it classifies an individual into a specific risk
appear in the printed text and are provided in the HTML and PDF group, identifying those who are at an increased risk of
versions of this article on the journal’s Web site (www.jorthotrauma. developing a VTE based on specific unique patient factors
com). and injury type. The Caprini Score also includes an associ-
Reprints: Jesse Dashe, MD, Boston Medical Center, 850 Harrison Avenue,
Dowling 2 North, Boston, MA 02118 (e-mail: jessedashe@gmail.com).
ated chemical prophylaxis recommendation.10–15 There are
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. specific patient factors that place one at higher risk for
DOI: 10.1097/BOT.0000000000001451 developing a VTE (ie, undiagnosed clotting disorder and
Dashe et al J Orthop Trauma Volume 33, Number 6, June 2019
family history of VTE) and should thus be taken into computed tomography scan.3 No routine screening was per-
account and carefully considered when assessing a patient’s formed to assess for VTE in the absence of clinical findings.
requirement for VTE prophylaxis. Inpatient hospital charts and outpatient records were
Given the broad and reliable utilization of the Caprini reviewed to assess injuries sustained, treatments rendered,
Score in other surgical fields, the purpose of this study was to past medical and surgical history, length of follow-up, use,
explore the validity of the Caprini Score in the orthopaedic and type of chemoprophylaxis, type and development of
fracture population by comparing 2 groups of patients deemed VTE, diagnosis of VTE and time from injury, and/or surgery
high risk versus low risk for thrombosis based on past literature. to VTE. Patient age at the time of injury/encounter, sex, body
mass index (BMI), and the patient’s medical history and
injury characteristics necessary to perform a Caprini Score
PATIENTS AND METHODS were recorded. Caprini Scores were subsequently performed
using data obtained from the chart review.
Study Design Patients were stratified as both an overall group and as 2
Following approval from our Institutional Review distinct subgroups: perihip fracture (PHF) with N = 300 and
Board, a retrospective review of 1310 patients at a single periankle fracture (PAF) with N = 548 (Fig. 1), based on
institution from 2002 to 2015 with lower-extremity frac- previously published rates of VTE. The PHF group consisted
tures. Exclusion criteria included the following: age less of patients having sustained pelvic (any Young-Burgess clas-
than 18 years old, follow-up less than 30 days, death sification) and/or acetabular (any Letournel classification)
secondary to non-VTE–related complications, and/or incom- fractures with or without additional injuries, representing
plete medical records (medical records that did not allow for injury patterns with historically higher rates of VTE. The
calculation of the Caprini Score and/or other pertinent med- PAF group consisted of patients who sustained isolated foot,
ical or injury history). Four hundred sixty-two patients did ankle, and/or pilon fractures that required a period of immo-
not meet inclusion criteria resulting in a final study cohort of bilization, representing injury patterns with historically lower
848 patients. rates of VTE. Patients had a mixture of both high- and low-
The American College of Chest Physician guidelines on energy mechanisms of trauma.
when to investigate/workup for VTEs were followed and only The use of VTE chemoprophylaxis was recorded but
included thromboembolic events that were objectively diag- not used as an exclusion criterion for the study. Most of
nosed through venous duplex ultrasound, ventilation– the high-risk group patients (PHF) received chemoprophy-
perfusion (V/Q) scan, and/or pulmonary embolism protocol laxis, which included warfarin, low molecular weight heparin,
TABLE 1. Caprini Score Risk Factors

Caprini Score Value Risk Factors
1 Age 41–60 y Acute myocardial infarction
Current swollen legs Congestive heart failure (,1 mo)
Varicose veins Medical patient currently at bed rest
Obesity (BMI .25) History of inflammatory bowel disease
Minor surgery planned History of prior major surgery (,1 mo)
Sepsis (,1 mo) Abnormal pulmonary function (COPD)
Pregnancy or postpartum (,1 mo) Serious lung disease including pneumonia (,1 mo)
Oral contraceptives or hormone replacement therapy History of unexplained stillborn infant, recurrent spontaneous abortion
(3 or more), premature birth with toxemia, or growth-restricted infant
2 Age 61–74 y Patient confined to bed (.72 h)
Arthroscopic surgery Immobilizing plaster cast (,1 mo)
Malignancy (present or previous) Central venous access
Laparoscopic surgery (.45 min) Major surgery (.45 min)
3 Age 75 y or older Positive prothrombin 20210A
History of DVT/pulmonary embolism Positive lupus anticoagulant
Positive Factor V Leiden Elevated anticardiolipin antibodies
Heparin-induced thrombocytopenia Elevated serum homocysteine
Other congenital or acquire thrombophilia Family history of thrombosis* (often missed factor)
5 Stroke (,1 mo) Acute spinal cord injury (paralysis) (,1 mo)
Elective major lower-extremity arthroplasty Multiple trauma (,1 mo)
Hip, pelvis, or leg fracture (,1 mo)**
(Needs further clarification for orthopaedics)
COPD, chronic obstructive pulmonary disease.
J Orthop Trauma Volume 33, Number 6, June 2019 Caprini Score Predicts VTE for Fractures
heparin, or dalteparin (87%). Seventy-two percent of the low- RESULTS

risk group patients (PAF) were treated with aspirin (36%) or Of the 1310 patients whose charts were initially reviewed,
no chemoprophylaxis (36%) with 28% in the low-risk group 462 were excluded for incomplete chart data, follow-up less than
being treated with warfarin, low molecular weight heparin, 30 days, and death secondary to non-VTE–related complica-
heparin, or dalteparin. tions. A total of 848 patients were included in the final analysis.
The average age of the overall population in the non-VTE
Outcome Measures group was 44 years as compared to 51 years in the VTE group,
The primary outcome of the study assessed the Caprini P = 0.025. However, when subgrouped as PHF and PAF, age
Score cutoff associated with development of VTE using was only found to be significantly different in the PAF group
a receiver-operating curve and performing a c-statistic. Sec- (P = 0.034). Both BMI and sex were not significantly different
ondary outcomes included differences in the Caprini Score between the non-VTE and VTE groups (Table 2).
between the PHF and PAF groups, time to VTE, and specific
risk factors associated with development of VTE. VTE Characteristics
Of the 848 eligible patients, a total of 32 patients (3.8%)
Statistical Methods were diagnosed with a VTE: 17 (2%) DVT events and 16
A biomedical statistician was used to evaluate the (1.8%) pulmonary embolism (PE) events (one patient was
Caprini Score data. To evaluate the cutoff for those who diagnosed with both a PE and DVT). In comparison of the
developed a VTE based on Caprini Score, a receiver- PHF versus PAF subgroups, there was a statistically signif-
operating curve and c-statistic were calculated for each icant difference in the number of VTE events overall
population. For evaluation of patient characteristics, VTE [P , 0.0001, odds ratio (OR): 5.2 with 95% confidence inter-
characteristics, and Caprini Score characteristics, normally val (CI) 2.4–11.4] with the breakdown: n = 9 (3%) versus 8
distributed continuous variables were analyzed with a Student (1.5%) DVT (P = 0.13, OR: 2.1 with 95% CI 0.8–5.5) and n =
t test between 2 groups, and categorical variables were ana- 15 (5%) versus 1 (0.2%) PE (P , 0.0001, OR: 28.8 with 95%
lyzed using Fisher exact test. These evaluations were per- CI 3.8–219) in PHF versus PAF groups, respectively. There
formed using the GraphPad online calculator. Significant was no significant difference (P = 0.095) regarding time to
difference was considered for P , 0.05. VTE between the PHF and PAF groups (Table 3).
Funding Caprini Score Characteristics

Funding from quality improvement grants were used to Each identifiable risk factor in the Caprini Score was
obtain a biostatistician for statistical analysis. assessed as an independent risk factor for development of VTE.
FIGURE 1. Patient cohort
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. www.jorthotrauma.com | 271
TABLE 2. VTE Rates in Study Population

No VTE VTE No VTE VTE No VTE VTE
Population (Combined) (Combined) (Perihip) (Perihip) (Periankle) (Periankle)
Characteristic (n = 816) (n = 32) P (n = 277) (n = 23) P (n = 539) (n = 9) P
Mean age (y) (SD) 44 (16.62) 51 (17.38) 0.025 43 (18.08) 49 (15.33) 0.138 44 (15.81) 55 (21.21) 0.034
BMI (kg/m2) (SD) 29 (6.63) 28.5 (6.94) 0.744 28.1 (6.82) 28.3 (7.25) 0.887 29.3 (6.52) 29.2 (6.52) 0.937
Gender 0.469 0.8141 1
Male (%) 478 (58.6) 21 (65.6) 198 (71.5) 16 (69.6) 280 (51.9) 5 (55.6)
Female (%) 338 (41.4) 11 (34.4) 79 (28.5) 7 (30.4) 259 (48.1) 4 (44.4)
Bolded values have P-values ,0.05.
The factors found to be significantly different in the PHF group which patients were at an increased risk of developing a VTE.
were age 41–60 years (P = 0.019), BMI . 25 (P = 0.031), acute The c-statistic was 0.83 with sensitivity = 90.6%, specificity =
lung pathology (P = 0.007), minor surgery performed (P = 73.9%, positive predictive value = 12%, and negative predictive
0.006), patient confined to bed rest (P = 0.003), cast immobili- value = 99.5%. The Caprini Score cutoff for the PHF group was
zation (P = 0.0031), and patients with central venous access (P = also 12 with a c-statistic of 0.74 (sensitivity = 100%, specificity =
0.001). The factors that were found to be significantly different 48.6%, positive predictive value = 13.9%, and negative predictive
in the PAF group were personal or family history of VTE, value = 100%). Finally, the PAF group had a Caprini Score cutoff
history of thrombosis (P = 0.001), and congenital or acquired of 11 with a c-statistic of 0.79 (sensitivity = 88.9%, specificity =
thrombophilia (P = 0.033). In combining both populations (PHF 68.8%, positive predictive value = 4.6%, and negative predictive
and PAF), the factors found to be significantly different included value = 99.7%). This data are demonstrated in Supplemental
acute lung pathology (P = 0.001), minor surgery performed Digital Content 1 (see Table, http://links.lww.com/JOT/A674).
(P = 0.000), patients confined to bed rest (P = 0.000), patients
with central venous access (P = 0.000), personal or family his-
tory of VTE (P = 0.005), and polytrauma patients (P = 0.000) DISCUSSION
(Table 4). Caprini Score risk factors not identified in any of the Identifying those who are at higher risk of developing
patients in this retrospective review included history of unex- VTEs is important because it has the potential to prevent fatal
plained still infant or other birth complications, arthroscopic outcomes. The purpose of this study was to act as a pilot
surgery, positive Factor V Leiden, elevated serum homocysteine, study to evaluate the applicability of the Caprini VTE risk
elevated anticardiolipin antibodies, prothrombin 20210A, stroke, stratification tool in orthopaedic fracture patients by compar-
or lower-extremity arthroplasty. ing the scores for historically low-rate and high-rate VTE
A receiver-operating curve was created to determine populations. In this study, it was determined that a Caprini
a Caprini Score cutoff for incidence of VTE. The overall study Score of 11–12 correlated with a higher rate of VTEs in both
population demonstrated a Caprini Score cutoff of 12, above the high- and low-risk populations, which suggests that this
TABLE 3. VTE Characteristics

VTE Characteristics Combined Group (n = 33) Perihip Fracture Group (n = 24) Periankle Fracture Group (n = 9) P
VTE Event* 24 (8%)† 9 (1.6%)† ,0.0001
DVT 17 (52%) 9 (3%)† 8 (1.5%)† 0.13
PE 16 (48%) 15 (5%)† 1 (0.2%)† ,0.0001
Days to VTE from injury (range) 21 (2–75) 16 (2–75) 31 (2–58) 0.095
VTE diagnosis
DVT on U/S below the knee 4 (12%) 3 (13%) 1 (11%)
DVT on U/S at or above the knee 13 (39%) 6 (25%) 7 (78%)
Subsegmental PE on CTPA 4 (12%) 4 (17%) 0
Significant PE on CTPA 11 (33%) 11 (46%) 0
PE on V/Q scan 1 (3%) 0 1 (11%)
Chemoprophylaxis
None or aspirin 5 (15%) 3 (12%) 2 (22%)
Acceptable chemoprophylaxis for 28 (85%) 21 (88%) 7 (78%)
high risk‡
Reported P-values compare the perihip versus periankle population.
*One patient in the perihip group had a DVT above the knee and subsegmental PE
†Percent of DVT or PE in total perihip or periankle populations.
‡Warfarin; heparin 5000u TID; dalteparin; low molecular weight heparin (mg): 30 qday, 40 qday, 30 bid, 40 bid, therapeutic.
CTPA, computed tomography pulmonary angiogram; U/S, ultrasound.
TABLE 4. Caprini Score Factors Correlated With Development of VTE in Study Population
Caprini Score Characteristics Caprini Score Point Value Overall, P Perihip Trauma, P Periankle Trauma, P
Age 41–60 years old 1 0.129 0.019 1.000
Obesity (BMI .25) 1 0.440 0.031 0.701
Oral contraceptive or HRT 1 0.230 0.147 1.000
History of previous major surgery 1 0.019 0.430 1.000
(,1 mo)
Serious lung disease (include 1 0.001 0.007 0.182
pneumonia ,1 mo)
Minor surgery performed (,45 min) 1 0.000 0.006 1.000
Medical patient at bed rest at baseline 1 0.021 0.265 1.000
Age 61–74 years old 2 0.195 0.149 0.258
Malignancy (previous or current) 2 1.000 0.274 1.000
Patient confined to bed for .72 h 2 0.000 0.003 1.000
Immobilizing plaster cast within 1 mo 2 0.102 0.031 1.000
Major surgery (.45 min) 2 0.849 0.235 0.329
Central venous access 2 0.000 0.001 1.000
Age 75 y old or older 3 0.185 1.000 0.055
Personal history of VTE 3 0.029 0.382 0.017
Personal or family history of VTE 3 0.005 0.382 0.001
Congenital or acquired thrombophilia 3 0.072 1.000 0.033
History of thrombosis 3 0.230 1.000 1.000
Acute spine cord injury (paralysis) 5 0.072 0.147 1.000
(,1 mo)
Multiple trauma (,1 mo)* 5 0.000 0.084 1.000
Bolded values have P-values ,0.05.
*Multiple trauma = lower-extremity injury + trauma to additional organ (example: thoracic, abdominal, and head).
HRT, hormone replacement therapy.
screening tool may have utility in stratifying patients for VTE finding because it implies that a Caprini Score of 5 or greater
risk. does in fact have a uniform increased risk of VTE across different
Our study found that, for the entire study population, surgical patient populations, displaying the importance of patient
a Caprini Score $11 correlated with higher rates of VTEs factors being associated with VTEs. This was additionally con-
with Caprini Score cutoff values being $11 in the PAF group firmed by Saragas et al (n = 216) in foot and ankle patients and
and $12 in the PHF group. This means that a Caprini Score required below the knee casting for at least 4 weeks and a period
of ,11 was associated with a significantly lower incidence of of non–weight-bearing. The study found that, using a risk score
VTE for the total study population. To maintain assessment similar to the Caprini Score, 90.9% of patient who had a VTE
sensitivity, we would recommend a conservative Caprini had a score of 5 or greater as compared to 73.7% of patients who
Score cutoff of .10 in lower-extremity fracture patients to did not have a VTE had a score of 5 or greater with an average
define those who are at greatest risk for VTE, as this was the score of 7.7 in the VTE group.17
lower of both thresholds determined in the study. Our study’s This speaks directly to the historical skepticism in
findings are consistent with previous literature by Luksa- adopting the Caprini Score in orthopaedic surgery, as the
meearunothai et al16 (n = 92) who found that a Caprini Score current score does not stratify those that have sustained lower-
of 12 or great had a significantly higher relative risk of pre- extremity fractures. A representative example is that both an
operative DVT versus no DVT in hip fracture patients. ankle and a pelvic fracture would have a Caprini Score of 5 for
Based on significant findings in other surgical fields having “hip, pelvic, or leg fracture ,1 month.” Given that patients
previously validated the Caprini Score, chemoprophylactic with lower-extremity trauma have drastically different rates of
medication is suggested for individuals with a Caprini Score of VTE: 0.6%–30% for fractures below the knee versus 61% for
5 or greater as these individuals are at higher risk of VTEs. The pelvic fractures, the criteria of “hip, pelvis, or leg fracture ,1
discrepancy between other studies having a Caprini cutoff of 5, month” may need to be more clearly defined.2,18–21
and our finding of having a Caprini cutoff of 11–12 is most likely Anecdotally, current orthopaedic practice across the
attributed to the fact that all the patients in the current study have United States is to provide chemical VTE prophylaxis to
an additional 5 points for sustaining “hip, pelvic, or leg fracture individuals with trauma at or above the knee such as femoral
,1 month.” These additional 5 points were likely not factors shaft, hip, and pelvic fractures. The main discrepancy in
added in the previous studies that validated the Caprini Score practice within the orthopaedic community is for individuals
because they were in a variety of surgical populations that did not having sustained isolated trauma below the knee. It is this
have concomitant lower-extremity fractures. This is an important particular orthopaedic patient population who could be at
greatest benefit of the Caprini risk stratification tool in helping VTEs can occur as late as 60–90 days after injury because some
to discern those requiring VTE prophylaxis. patients may have developed VTE at a later period and did not
There are other risk assessment tools in orthopaedic present to our institution.3 This was chosen as many ankle frac-
surgery, which are not ubiquitously used as they are specific tures are only followed for 6 weeks clinically, and we wanted to
to various patient populations or rely on factors that are not make sure that these patients were represented and included in
unique to the patient’s medical history. One such tool is called the analysis. Despite the limitations of the study, we were able to
the risk assessment profile that was validated for the trauma demonstrate a direct correlation of a higher Caprini Score asso-
population based on a study of 53 patients.22 Factors included ciated with a greater risk of VTE.
in this score are obesity, history of malignancy, abnormal In conclusion, this study confirms that patient factors
coagulation profile at admission, history of VTE, central line play a large role in the development of VTE events indepen-
access, number of transfusions, surgical procedures, injury to dent of injury type, and that the Caprini Score may help to
major blood vessels, injury to various organ systems, and age. identify these patients who may require increased protection.
Many of these factors are also found within the Caprini Score; We found that a Caprini Score greater than 10 is associated
however, the Caprini Score is more exhaustive with regard to with a higher incidence of VTE. As discussed previously,
patient factors and has been validated in a greater number of removing the “hip, pelvic, or leg fracture ,1 month” criteria
patients across multiple surgical specialties. revises the Caprini Score to 6, which is similar to the Caprini
An additional risk assessment tool created by Parkland Score cutoff of 5 that has previously been associated with
Orthopaedics attempts to predict the risk of a pulmonary a higher risk of VTEs in other surgical specialties.
embolism in trauma patients.23 This scoring system includes It is our recommendation that additional research is
factors such as age, occurrence of motorcycle accident, method warranted in a prospective fashion where patient history is
of arrival to hospital, admission to the intensive care unit, pre- specifically asked for each factor of the Caprini Score. We
senting heart rate, BMI, and organ systems involved in the believe that with further strategic investigation, the Caprini
trauma. Although this scoring system has some utility and was Score may become an important tool for the orthopaedic
created based on a population of 38,000 patients, it relies on surgeon in guiding VTE risk stratification and management.
factors that are surrogates of the patient’s injuries and lacks
specific patient characteristics that would predispose the patient
to development of VTE. Such additional factors not considered
include a history of Factor V Leiden and history of malignancy.
ACKNOWLEDGMENTS
It is important to note that the Caprini Score risk
The authors would like to acknowledge Dr. Caprini for
stratification tool currently includes anticoagulation recom-
his support and expertise in the development of this study.
mendations associated with particular Caprini Scores based
on studies in other surgical fields. However, this current study
did not follow these recommendations, nor does it endorse
REFERENCES
such recommendations. The reason for this is that there are
1. Sagi HC, Ahn J, Ciesla D, et al. Venous thromboembolism prophylaxis
a variety of methods to provide preventative VTE anti- in orthopaedic trauma patients: a survey of OTA member practice pat-
coagulation with no concrete evidence as to the best method terns and OTA expert panel recommendations. J Orthop Trauma. 2015;
or dosage of medication, for example, low molecular weight 29:e355–e362.
heparin 40 mg daily versus 30 mg twice a day.24–28 VTE 2. Abelseth G, Buckley RE, Pineo GE, et al. Incidence of deep-vein throm-
bosis in patients with fractures of the lower extremity distal to the hip. J
chemical prophylactic recommendations would need to con- Orthop Trauma. 1996;10:230–235.
sider clinician experience and current practices based on evi- 3. Falck-Ytter Y, Francis CW, Johanson NA, et al. Prevention of VTE
dence that is currently available on this topic. More in orthopaedic surgery patients: antithrombotic therapy and preven-
importantly, this study supports that the Caprini Score factors, tion of thrombosis, 9th ed: American College of Chest Physicians
Evidence-Based Clinical Practice Guidelines. Chest. 2012;141:
excluding “hip, pelvic, or leg fracture ,1 month”, are asso-
e278S–325S.
ciated with a higher rate of VTE with an inability to distin- 4. Lapidus LJ, Ponzer S, Pettersson H, et al. Symptomatic venous throm-
guish between low-risk and high-risk populations. boembolism and mortality in orthopaedic surgery—an observational
Our study is inherently limited by its retrospective nature. study of 45 968 consecutive procedures. BMC Musculoskelet Disord.
The lack of prospectively collected data limits the accuracy of 2013;14:177.
5. Schiff RL, Kahn SR, Shrier I, et al. Identifying orthopedic patients at
identified risk factors such as varicose veins, history of stillbirth, high risk for venous thromboembolism despite thromboprophylaxis.
and family history of VTE, etc, and is only as good as the Chest. 2005;128:3364–3371.
thoroughness of the original questioner. In addition, VTE 6. Whiting PS, Jahangir AA. Thromboembolic disease after orthopedic
chemical prophylactic medication was not controlled for, and trauma. Orthop Clin North Am. 2016;47:335–344.
multiple electronic medical records were used to collect data 7. Whiting PS, White-Dzuro GA, Greenberg SE, et al. Risk factors for deep
venous thrombosis following orthopaedic trauma surgery: an analysis of
given the large time frame used for inclusion. Furthermore, the 56,000 patients. Arch Trauma Res. 2016;23:e32915.
retrospective nature of the study only allowed for the chart 8. Jacobs JJ, Mont MA, Bozic KJ, et al. American Academy of Orthopaedic
reviewers to identify those who developed VTEs at the chart Surgeons clinical practice guideline on: preventing venous thromboem-
level and does not capture those who presented elsewhere or bolic disease in patients undergoing elective hip and knee arthroplasty. J
Bone Joint Surg Am. 2012;94:746–747.
who did not have accurate documentation of a VTE. Another 9. Parvizi J, Huang R, Raphael IJ, et al. Symptomatic pulmonary embolus
limit of the study is that the minimum included follow-up was after joint arthroplasty: stratification of risk factors. Clin Orthop Relat
30 days. This exclusion criteria could have been made longer as Res. 2014;472:903–912.
10. Bahl V, Hu HM, Henke PK, et al. A validation study of a retrospective 19. Pelet S, Roger ME, Belzile EL, et al. The incidence of thromboembolic
venous thromboembolism risk scoring method. Ann Surg. 2010;251: events in surgically treated ankle fracture. J Bone Joint Surg Am. 2012;
344–350. 94:502–506.
11. Caprini JA. Risk assessment as a guide for the prevention of the many 20. Selby R. Symptomatic venous thromboembolism uncommon without
faces of venous thromboembolism. Am J Surg. 2010;199:S3–S10. thromboprophylaxis After isolated lower-limb fracture. J Bone Joint Surg
12. Pannucci CJ, Bailey SH, Dreszer G, et al. Validation of the Caprini risk Am. 2014;96:e83.
assessment model in plastic and reconstructive surgery patients. J Am 21. Geerts WH, Code KI, Jay RM, et al. A prospective study of venous throm-
Coll Surg. 2011;212:105–112. boembolism after major trauma. N Engl J Med. 1994;331:1601–1606.
13. Pannucci CJ, Dreszer G, Wachtman CF, et al. Postoperative enoxaparin 22. Greenfield LJ, Proctor MC, Rodriguez JL, et al. Post trauma thrombo-
prevents symptomatic venous thromboembolism in high-risk plastic sur- embolism prophylaxis. J Trauma. 1997;42:100–103.
gery patients. Plast Reconstr Surg. 2011;128:1093–1103. 23. Parkland PE Risk Assessment. Available at: http://www.parklandhospital.
14. Shuman AG, Hu HM, Pannucci CJ, et al. Stratifying the risk of venous com/phhs/orthopaedic-research.aspx. Accessed April 2016.
thromboembolism in otolaryngology. Otolaryngol Head Neck Surg. 24. Geerts WH, Jay RM, Code KI, et al. A comparison of low-dose heparin
2012;146:719–724. with low-molecular-weight heparin as prophylaxis against venous throm-
15. Zhou HX, Peng LQ, Yan Y, et al. Validation of the Caprini risk assess- boembolism after major trauma. N Engl J Med. 1996;335:701–707.
ment model in Chinese hospitalized patients with venous thromboembo- 25. Al-Sallami H, Jordan S, Ferguson R, et al. Current enoxaparin dosing
lism. Thromb Res. 2012;130:735–740. guidelines have dubious credibility. N Z Med J. 2010;123:62–67.
16. Luksameearunothai K, Sa-ngasoongsong P, Kulachote N, et al. 26. Riha GM, Van PY, Differding JA, et al. Incidence of deep vein throm-
Usefulness of clinical predictors for preoperative screening of deep bosis is increased with 30 mg twice daily dosing of enoxaparin compared
vein thrombosis in hip fractures. BMC Musculoskelet Disord. 2017; with 40 mg daily. Am J Surg. 2012;203:598–602.
18:208. 27. Rutherford EJ, Schooler WG, Sredzienski E, et al. Optimal dose of
17. Saragas NP, Ferrao PN, Saragas E, et al. The impact of risk assessment enoxaparin in critically ill trauma and surgical patients. J Trauma.
on the implementation of venous thromboembolism prophylaxis in foot 2005;58:1167–1170.
and ankle surgery. Foot Ankle Surg. 2014;20:85–89. 28. Bush S, LeClaire A, Hampp C, et al. Review of a large clinical series:
18. Lassen MR, Borris LC, Nakov RL. Use of the low-molecular-weight once- versus twice-daily enoxaparin for venous thromboembolism pro-
heparin reviparin to prevent deep-vein thrombosis after leg injury requir- phylaxis in high-risk trauma patients. J Intensive Care Med. 2011;26:
ing immobilization. N Engl J Med. 2002;347:726–730. 111–115.
Invited Commentary
Risk Assessment After Orthopaedic Trauma: Coming of Age
A fatal pulmonary embolus is the number 1 preventable

cause of death after surgery.1 It has been shown that using
traditional anticoagulants prevent almost all these deaths and
associated with a much lower incidence of thrombosis than
those involving the proximal leg and/or pelvis.
The key element in applying the Caprini score to a
sublethal thrombotic events which on occasion may be life- given population is to identify the cutoff between low- and
changing. To preserve good outcomes, surgeons must care- high-risk individuals using the score rather than the type of
fully weigh the thrombotic versus bleeding risk of each fracture. This enables selection of the appropriate prophylaxis
patient. Traditional anticoagulation may cause bleeding carefully balancing the risks of bleeding versus thrombosis.
that jeopardizes a well-performed procedure. The current study was performed to explore the validity of the
The incidence of thrombotic events is related to the Caprini score in the orthopaedic fracture population. They
presence and number of thrombosis risk factors in addition to also compared the score between 2 groups of fracture patients
the type and extent of injury. The Caprini score has been believed to be high versus low risk of thrombosis.
validated in multiple specialties, and when the data are One of the features of the Caprini score is to look at the
properly collected, every study shows a statistically signifi- entire risk profile of the patient independent of the type of
cant correlation between the score and the incidence of surgery or fracture. A low-risk fracture may be associated
venous thromboembolism (VTE). This is most obvious when with a high risk of thrombosis when additional risk factors are
the group tested does not receive thrombosis prophylaxis.2 present. Patients with high-risk fractures but no additional risk
The authors report that the Caprini score represents a factors may have a lower risk than average for that type of
thorough history and physical. This score, in fact, provides a injury. The bottom line is to look at all the risks associated
profile of the patient and allows for a selection of the type, with an individual patient not just the type of fracture or
strength, and duration of prophylaxis. One of the main procedure.
criticisms of the Caprini score after Orthopaedic trauma is The authors found that the low-risk fracture group had
that all fractures are scored the same. The authors correctly fewer deep vein thrombosis (DVT) events than seen in those
point out that there is a disparity between different types of with high-risk fractures. The Caprini score was not statisti-
fractures depending on location. Ankle fractures overall are cally different between these groups when looking at fracture
ORIGINAL ARTICLE
Radiographic Healing of Far Cortical Locking

Constructs in Distal Femur Fractures:
A Comparative Study With Standard Locking Plates
Yanin Plumarom, MD,*† Brandon G. Wilkinson, MD,* J. Lawrence Marsh, MD,*
Michael C. Willey, MD,* Qiang An, MPH,* Yubo Gao, PhD,* and Matthew D. Karam, MD*
rates between the 2 groups. The union rate was 90.5% and 82.1% for
Objectives: To investigate the radiographic healing of far cortical the FCL and LP groups, respectively, at final follow-up.
locking (FCL) construct fixation in distal femur fractures compared
with traditional locking plate (LP) constructs. Conclusions: To our knowledge, this is the first comparative study
between FCL and LP constructs. The FCL group was noted to have
Design: A retrospective cohort of 143 consecutive patients with 146 significantly higher mRUST scores at all periods indicating
distal femur fractures. increased callus formation, but the differences seen were small and
there were no differences in healing rates or complications between
Setting: Level I trauma center.
the 2 groups, thus bringing the clinical benefit of FCL into question.
Patients/Participants: After excluding patients with OTA/AO Further prospective study designs are needed to compare FCL with
type B fractures, referred nonunion cases, patients younger than 16 LP constructs and to investigate the role of interfragmentary motion
years, and patients with less than 24 weeks of follow-up, 69 patients on callus formation in distal femur fractures.
with 70 total fractures were included for analysis.
Key Words: far cortical locking, locked plating, modified RUST
Intervention: AP and lateral knee radiographs were blinded to score, fracture healing, distal femur fracture
type of screws and individually reviewed by 3 orthopaedic trauma
Level of Evidence: Therapeutic Level III. See Instructions for
surgeons.
Authors for a complete description of levels of evidence.
Outcome Measurements: The modified RUST (mRUST) score (J Orthop Trauma 2019;33:277–283)
was our primary outcome measure. mRUST scores were assigned at
6, 12, and 24 weeks and final follow-up based on AP and lateral
INTRODUCTION
radiographs and compared between FCL and LP groups as a tool for
Distal femur fractures are commonly treated by peri-
evaluating fracture healing. Secondary outcomes compared between
articular locking plates (LPs),1 which have largely replaced
FCL and LP included union rate and postoperative complications.
intramedullary nails, blade plates, and condylar screws. These
Results: Statistically significant differences in mRUST scores were fractures are frequently comminuted, and LPs are placed
noted between FCL and LP groups at 6 weeks (P = 0.040), 12 weeks using the bridging technique and depend on some degree of
(P = 0.034), 24 weeks (0.044), and final follow-up (P = 0.048). interfragmentary motion to stimulate osseous union. Periar-
There was no significant difference in union or specific complication ticular LPs are stiffer than previously used implants,2–8 and
construct rigidity may delay fracture healing. Clinical studies
looking at LPs in distal femur fractures have reported fracture
Accepted for publication February 13, 2019.
From the *Department of Orthopedics and Rehabilitation, University of Iowa healing complications, including delayed union, implant fail-
Hospitals and Clinics, Iowa City, IA; and †Department of Orthopaedic ure, loss of alignment, and nonunion.9–16
Surgery, Phramongkutklao Hospital, Phramongkutklao College of Medi- Locking plates with far cortical locking (FCL) screws
cine, Ratchathewi, Bangkok, Thailand. permit controlled interfragmentary motion leading to increased
The authors report no conflict of interest.
Presented in part at the Annual Meeting of Orthopaedic Trauma Association, callus formation in animal models when compared with
October 14, 2017, Vancouver, BC, Canada; Annual Meeting of American standard LP constructs.17,18 Biomechanical studies suggest that
of Orthopaedic Surgeons, March 8, New Orleans, LA; and Annual FCL constructs significantly reduce the axial stiffness of
Meeting of the Mid-America Orthopaedic Association, April 19, 2018, LPs.2,18–20 FCL constructs provide flexible fixation and nearly
San Antonio, TX. parallel interfragmentary motion.2,17–20 The results of FCL
Supplemental digital content is available for this article. Direct URL citations
appear in the printed text and are provided in the HTML and PDF constructs for distal femur fractures have been reported in case
versions of this article on the journal’s Web site (www.jorthotrauma. series with favorable results21–23; however, they have not been
com). directly compared with traditional locked plating techniques.
Reprints: Yanin Plumarom, MD, Department of Orthopedics and Rehabili- The purpose of this study was to assess the timing and
tation, University of Iowa Hospitals and Clinics, 200 Hawkins Drive,
Iowa City, IA 52242 (e-mail: yaninyo24@gmail.com).
degree of fracture healing using the modified RUST
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. (mRUST) score in FCL compared with LP constructs in
DOI: 10.1097/BOT.0000000000001464 distal femur fractures. We also assessed the ultimate union
Plumarom et al J Orthop Trauma Volume 33, Number 6, June 2019
and complication rates between the 2 groups. We hypothe- implant type, time to full weight bearing, revision surgery,
sized that FCL constructs would result in earlier callus complications, periprosthetic fracture, date of surgery, and
formation and higher union rates. duration of follow-up. Complications were defined as implant
failure (broken plate and screws), nonunion (pain with weight
bearing, limp, tender to palpation at the fracture site, and the
PATIENTS AND METHODS absence of fracture healing progression or bridging callus on
After obtaining IRB approval (IRB ID #201612761), serial radiographs), surgical site infection, revision surgery,
a cohort of 143 consecutive adult patients with 146 distal and implant irritation.
femur fractures, including periprosthetic fractures, treated Nonperiprosthetic fractures were classified according to
from 2011 to 2016 at a single institution were identified ret- the OTA/AO universal fracture classification,25 and open
rospectively. Patients with partial articular fractures (OTA/ fractures were classified using the Gustilo–Anderson classi-
AO24 type B fractures) (n = 15), intramedullary nail con- fication.26 All fractures were divided into those with meta-
structs (n = 13), use of Smith and Nephew Peri-Loc plate physeal comminution (33A3, C2, and C3) and those without
(n = 5), previous nonunion (n = 5), missing follow-up radio- metaphyseal comminution (33A1, A2, and C1).
graphs (n = 3), cases using allograft (n = 1), and follow-up
less than 6 months (n = 34) were excluded. After applying our Radiographic Assessment of Healing
exclusion criteria, 69 patients with 70 fractures (42 FCL and Anteroposterior (AP) and lateral radiographs of the
28 LP) were included for analysis. Of these, 21 of the FCL knee and femur were independently reviewed by 3 inves-
cases and 15 of the LP cases were periprosthetic fractures. tigators (Y.P., M.C.W., and M.D.K.), and callus formation of
Although some of the plates overlapped portions of the stem, each cortex was evaluated and scored using the mRUST
unicortical locking screws were not placed and all the prox- score.27 The mRUST score was applied using both AP and
imal fixation was with capped FCL screws. All periprosthetic lateral radiographs to assess all 4 cortices as follows: 1 = no
fractures were distal femur periprosthetic fractures. So, callus, 2 = callus present, 3 = bridging callus, and 4 = remod-
although the plate overlapped the stem in some cases, there eled, fracture not visible. Low scores indicate poor fracture
was enough room to place 3–4 screws distal to the tip of the healing and callus formation, and high scores correlate with
stem, and screw fixation at the level of the stem was not fracture healing and remodeling. The mRUST score was
performed. totaled for each cortex to equal a minimum score of 4 or
All surgeons used standard locked plating techniques. a maximum score of 16. mRUST scores were collected from
All periprosthetic fractures, type A fractures, and type C each investigator on radiographs at 6, 12, and 24 weeks and
fractures with a simple articular line were treated with a small final follow-up, when available. Before radiographic assess-
distal incision for plate insertion, indirect reduction of the ment, all 3 investigators reviewed mock radiographs together
fracture, and percutaneous insertion of proximal shaft screws. and came to a consensus on how to apply the mRUST score
The remaining type C fractures were treated with a lateral to each cortex in efforts to decrease the learning curve of
arthrotomy, +/2 lag screw (articular portions only), and per- applying the mRUST score.27–30 In cases of obstructed visu-
cutaneous insertion of proximal shaft screws. The metaphy- alization of the lateral cortex, our observers were instructed to
seal portion of the fracture was not directly reduced. Hybrid use consolidation of the fracture line at the lateral cortex to
constructs were used in a minority of cases in which a tradi- best apply the mRUST score. Investigators were blinded to
tional nonlocking screw was placed proximally to approxi- type of screws used (Fig. 1), but were allowed to compare
mate the plate to bone, followed by locking screws/caps in the radiographs to previous radiographs to better exemplify incre-
remainder of the proximal holes. All 42 FCL constructs used mental differences in fracture healing.
locking caps for all proximal screws. Operative fixation was Reliability of the mRUST score was assessed by
performed by 3 fellowship-trained orthopaedic trauma sur- comparing the average of the 4 mRUST subscores from 3
geons with the choice of implant dictated by the individual independent observers at each time point using intraclass
surgeons’ preference. Implant type and number used are correlation coefficient (ICC), as previously described by
shown in a table (see Table, Supplemental Digital Content Landis and Koch.31 The mRUST score was used solely as
1, http://links.lww.com/JOT/A678). a tool to compare early callus formation between fractures
Standard follow-up was at 2, 6, 12, and 24 weeks or treated with FCL versus LP and was not used to define frac-
until the fracture was completely healed. Clinical and physical ture union.
examination was performed at each follow-up visit and Fracture union was assessed for the 69 patients with 70
recorded in the electronic medical record. All patients were fractures. Union was determined by the operating surgeon as
uniformly made touchdown weight bearing for a period of 4– documented in the medical record and was based on clinical
8 weeks postoperatively. The exact timing of transition to full history and examination findings (patient reported absence of
weight bearing; however, was at the discretion of the surgeon pain, painless ambulation, absence of limp, and no tenderness
based on clinical and radiographic findings at follow-up. to palpation of fracture site) and evaluation of radiographs
Medical records were reviewed to determine patient age, lat- (cortical continuity, bridging cortices, and visibility of
erality, body mass index, smoking status, diabetes mellitus, fracture line) at the time of follow-up. The radiographs of
fracture classification, open or closed fracture, mechanism of all cases documented as united in the medical record were
injury, high (ie, motor vehicle collisions, fall .10 feet, etc.) independently reviewed and unanimously agreed upon by 3
or low (ie, ground-level fall) energy injury, type of treatment, authors (Y.P., B.G.W., and J.L.M.) to confirm union.
J Orthop Trauma Volume 33, Number 6, June 2019 Radiographic Healing
FIGURE 1. Blinded radiographs shown to in-

vestigators. Editor’s Note: A color image
accompanies the online version of this article.
Similarly, nonunion was dictated by the operating smoking status, open fracture, or diabetes (see Table, Sup-
surgeon as documented in the medical chart based on clinical plemental Digital Content 2, http://links.lww.com/JOT/
and radiographic findings. All nonunion cases were indepen- A679). There were no significant differences in OTA/AO
dently reviewed by 3 authors (Y.P., B.G.W., and J.L.M.) to fracture classification or distribution of periprosthetic frac-
confirm nonunion and determine construct condition, time at tures between the 2 groups (see Table, Supplemental Digital
nonunion, time of revision, and revision type for nonunion. Content 3, http://links.lww.com/JOT/A680). Mean follow-up
The average length of follow-up for both union and nonunion for all patients (n = 69) was 15.99 months (range 6–55
cases was collected from the medical record. months).
Subgroup analysis was performed comparing FCL and
LP constructs in periprosthetic fractures (Table 2). Likewise, Assessment of Radiographic Callus by
in attempts to avoid inherent plate rigidity as a confounder, Modified RUST Score
subgroup analysis was performed for similar metals by Intraclass correlation coefficient of the mRUST scores
excluding stainless steel constructs (Table 4). Different metals showed moderate agreement with ICCs of 0.62, 0.61, 0.55,
were not directly compared. and 0.57 at 6, 12, and 24 weeks, and final follow-up,
respectively (see Table, Supplemental Digital Content 4,
Statistical Analysis http://links.lww.com/JOT/A681). Statistically significant dif-
The study was designed in conjunction with a PhD ferences in mean mRUST scores were noted at 6 weeks (P =
statistician. All data are reported as mean with associated SD. 0.040), 12 weeks (P = 0.034), 24 weeks (P = 0.044), and final
Data analysis was performed using SAS software, version 9.3 follow-up (P = 0.048) postoperatively, with higher mean
(SAS Institute, Inc of Cary, NC. Group comparisons for scores in the FCL group (Table 1).
continuous variables were analyzed using a 2 sample indepen- Subgroup analysis using the mRUST score for peri-
dent t test. Chi-square analysis was used to determine differ- prosthetic fractures (FCL n = 21, LP n = 15) showed signif-
ences between categorical variables. ICC was used to evaluate icant differences between FCL and LP constructs at 6 weeks
the measurement reliability among investigators using the postoperatively (P = 0.023), with FCL constructs noted to
mRUST scores. Significance was defined as P , 0.05. have earlier callus formation (Table 2). All other time points
showed no difference.
There was a significant difference in mRUST scores at
RESULTS 24 weeks in nonperiprosthetic fractures (FCL: n = 21; LP:
n = 13) treated with FCL compared with LP (P = 0.048)
Demographic Data (Table 3).
A total of 69 patients with 70 fractures (42 FCL and 28 Subgroup analysis of the mRUST score for similar
LP) were included for analysis. There was no significant metals (FCL: n = 42; LP: n = 24) showed significant differ-
difference in demographic data between the FCL and LP ences favoring FCL at 6 weeks (P = 0.044), but showed no
groups including age, sex, body mass index, energy of injury, significant differences at the remaining time points (Table 4).
TABLE 1. Comparison of Radiographic Findings Between FCL and LP Constructs Using the mRUST Score
Mean mRUST Score
Locked Plating Constructs
Timing FCL Constructs (Mean 6 SD) (Mean 6 SD) P
6 wk 6.90 6 1.83 (n = 41) 6.01 6 1.65 (n = 28) 0.0403
12 wk 9.90 6 2.03 (n = 42) 8.86 6 1.81 (n = 27) 0.0337
24 wk 12.22 6 2.01 (n = 36) 11.12 6 2.20 (n = 26) 0.0439
Final follow-up 14.18 6 1.87 (n = 33) 13.03 6 2.39 (n = 24) 0.0476
Time of transition to full weight bearing was not implant (n = 1), and symptomatic nonunion with intact im-
significantly different between FCL (12.33 weeks) and LP plants (n = 1). The average time of revision surgery was 23.3
(15.79 weeks) cohorts (P = 0.08). weeks.
Union Rate and Revision Surgery

Mean follow-up for patients with fracture union (n = DISCUSSION
61) was 16.2 months (range 6–55 months) and did not differ Treating distal femur fractures with LPs has become
significantly between FCL and LP groups (P = 0.321). There increasingly popular. Initial reports indicated that the fracture
was no significant difference in the overall union rate at final healed on almost all cases.10–16 However, more recent reports
follow-up in FCL (90.48%) compared with LP (82.14%) con- have demonstrated that nonunion may be more common than
structs (P = 0.47). previously identified. Ricci et al32 found a 14% nonunion rate
Mean follow-up for patients with nonunion (n = 9) was despite a minimally invasive insertion technique for the treat-
14.6 months (range 7–33 months). There was no significant ment of periprosthetic supracondylar femur fractures. Hender-
difference in nonunion rates for FCL (9.52%) and LP son et al33 found an even higher rate of nonunion (20%) in
(17.86%) at final follow-up (P = 0.468) (see Table, Supple- nonperiprosthetic distal femur fractures treated with LPs. It
mental Digital Content 5, http://links.lww.com/JOT/A682). has been suggested that slow healing or failure to heal may
Of the 9 total nonunions, 6 were revised and 3 were not result from excessive construct stiffness,2,17,18,34 which
revised. The 3 nonunions that were not revised were judged causes inconsistent and asymmetric formation of periosteal
to not be healed at 32, 33, and 64 weeks, respectively, and callus.35
were lost to follow-up. The 61 cases that were judged to be Far cortical locking screw constructs increase parallel
healed had 1 reoperation for infection at 24 weeks and sub- interfragmentary motion, and both biomechanical and animal
sequently healed and 1 interprosthetic fracture at 3 weeks studies suggest that this may lead to clinical advantage
treated nonoperatively in a brace that then went on to heal through earlier and stronger fracture repair.2,17,18 Bottlang
at 5 months. et al2 showed that compared with standard locked plating
There were no significant differences in revision constructs, the initial stiffness of FCL constructs was 88%
surgery, infection, implant irritation, or nonunion between lower in axial compression, 58% lower in torsion, and 29%
FCL and LP constructs (see Table, Supplemental Digital lower in bending. Studies in sheep with a gap osteotomy
Content 5, http://links.lww.com/JOT/A682). There were 6 model have shown increased callus formation in FCL con-
revision surgeries—all for nonunion (FCL 2 and LP 4). In structs compared with LP constructs and that increased callus
the FCL group, both revisions were secondary to implant resulted in stronger torsional strength.2,17
failure at 24 and 28 weeks (broken plates in the middle hole Despite the benefits demonstrated in these experimental
at the level of fracture site). In the LP group, revisions were studies, only a few clinical studies on patients with distal
performed for implant failure (n = 2) (1 plate broke at 12 femur fractures treated with FCL constructs have been
weeks and the other bent at 28 weeks both at the middle hole reported. Adams et al21 found no nonunions or implant fail-
at the level of fracture site), refracture after removal of ures in their small cohort of distal femur fractures. Bottlang
TABLE 2. Comparison of Radiographic Findings Between FCL and LP Constructs Using the mRUST Score in Periprosthetic
Fractures (N = 36)
Mean mRUST Score
Locked Plating Constructs
Timing FCL Constructs (Mean 6 SD) (Mean 6 SD) P
6 wk 7.00 6 1.97 (n = 21) 5.60 6 1.32 (n = 15) 0.0225
12 wk 9.92 6 2.23 (n = 21) 8.98 6 1.58 (n = 14) 0.1805
24 wk 11.98 6 2.11 (n = 18) 11.29 6 2.33 (n = 16) 0.3829
TABLE 3. Comparison of Radiographic Findings Between FCL and LP Constructs Using the mRUST Score in Nonperiprosthetic
Fractures (N = 34)
Mean mRUST Score
Locked Plating Constructs (Mean
Timing FCL Constructs (Mean 6 SD) 6 SD) P
6 wk 6.80 6 1.71 (n = 20) 6.49 6 1.91 (n = 13) 0.6280
12 wk 9.89 6 1.86 (n = 21) 8.74 6 2.09 (n = 13) 0.1052
24 wk 12.46 6 1.94 (n = 18) 10.92 6 2.12 (n = 12) 0.0484
et al22 reported the results of a prospective observational of scores during the crucial time of healing and formation of
study of 33 distal femur fractures (OTA/AO types 33A and bridging callus.27
33C) with no implant or fixation failures. They suggested that We acknowledge that there are several limitations to
dynamic fixation of distal femur fractures with FCL screws this study. One limitation is the task of assigning an mRUST
provides safe and effective fixation. In a retrospective review, score to a lateral cortex either partly or fully obscured by the
Ries et al reported an 88.9% healing rate using FCL con- implant. Litrenta et al27 noted similar difficulties and showed
structs for fixation of periprosthetic distal femur fractures.23,36 superior agreement of ICCs in intramedullary nail constructs
There have been no previous studies that have com- compared with lateral plates—likely secondary to lack of
pared clinical and radiographic outcomes between FCL and agreement at the lateral cortex. Their study demonstrated that
LP constructs. The current study demonstrated no differences ICC of the mRUST scores in the plate constructs was 0.59,
in nonunion rate, revisions, or other specific complications which was moderate agreement and was similar to our study
between FCL and LP cohort. However, there were significant (see Table, Supplemental Digital Content 4, http://links.
differences in the mRUST scores between FCL and LP lww.com/JOT/A681).
groups at all time points—suggesting that in these patients, Another limitation is uncontrolled mechanical variables
FCL constructs formed callus earlier than LP constructs. between groups. These variables affect construct stiffness and
These differences were also present at the 6-week time point include working length, number of screws, and bridge span
in subgroup analysis of periprosthetic fractures and in those and fracture and bone quality variability. These could
cases with similar metals. potentially confound the results. Theoretically, FCL does
In clinical studies, demonstrating differences in fracture not rely on the bridging distance or flexibility of spanning
healing is challenging and is a limitation to comparative ratio because its mechanics are directly related to the distance
studies. Various methods have been used to evaluate from the screw shaft to the near cortex of the bone.19,40 There
radiographic fracture union; however, no uniform method were also multiple surgeons using differing surgical techni-
or definitions have been widely accepted.37,38 Criteria for ques leading to some variations in fracture fixation techni-
radiographic fracture healing include cortical continuity, vis- ques. Risk factors for nonunion such as soft tissue
ibility of the fracture line, number of bridging cortices, and stripping, quality of reduction, degree of comminution,
the surgeon’s general impression.28 The RUST score substan- patient age, and general health status are difficult to assess
tially improves the reliability of assessment of fracture heal- between groups. Although our analysis indicates that the 2
ing compared with previously published methods and reliable groups were similar, systematic bias on how the surgeons
across a variety of experience levels.29,39 A second score, the chose cases for the 2 different constructs is possible. These
mRUST score, was subsequently created to specifically eval- limitations are inherent in retrospectively assessed groups of
uate metadiaphyseal fractures. The mRUST score has been patients.
shown to have slightly higher reliability than the standard We further acknowledge that weight bearing as an
RUST in metadiaphyseal fractures with plate fixation and outcome measure may be imprecise because all patients were
has shown moderate observer agreement.27 One advantage uniformly made touchdown weight bearing for a period of 4–8
of the mRUST score is its focus on gaining a greater range weeks. However, it remains an outcome measure because the
TABLE 4. Comparison of Radiographic Findings Between FCL and LP Constructs Excluding Stainless Steel Using the mRUST Score
(N = 66)
Mean mRUST Score
Locked Plating Constructs (Mean
Timing FCL Constructs (Mean 6 SD) 6 SD) P
6 wk 6.90 6 1.83 (n = 41) 5.96 6 1.71 (n = 24) 0.0442
12 wk 9.90 6 2.03 (n = 42) 8.96 6 1.83 (n = 23) 0.0671
24 wk 12.22 6 2.01 (n = 36) 11.30 6 2.19 (n = 23) 0.1038
exact timing of release to full weight bearing was dictated inde- 11. Gaines RJ, Sanders R, Sagi HC, et al. Titanium versus Stainless Steel
pendently by the operative surgeon based on clinical history and Locked Plates for Distal Femur Fractures: Is There Any Difference?
OTA Abstract; 2008. Paper Number 5.
examination and radiographic findings at the time of follow-up. 12. Kayali C, Agus H, Turgut A. Successful results of minimally invasive
Blinding is also another potential limitation because surgery for comminuted supracondylar femoral fractures with LISS:
despite attempts to blind the reviewers from the plate construct, comparative study of multiply injured and isolated femoral fractures. J
some constructs could potentially be identified and could lead Orthop Sci. 2007;12:458–465.
13. Kregor PJ, Stannard JA, Zlowodzki M, et al. Treatment of distal femur
to bias when assigning mRUST scores in the study. In addition, fractures using the less invasive stabilization system: surgical experience
at first glance (Table 1), mRUST scores separated by 1 point and early clinical results in 103 fractures. J Orthop Trauma. 2004;18:
would not seem to delineate significant difference at all time 509–520.
points postoperatively; however, because these measurements 14. Schandelmaier P, Partenheimer A, Koenemann B, et al. Distal femoral
were observed over numerous cases, even small differences in fractures and LISS stabilization. Injury. 2001;32(suppl 3):SC55–SC63.
15. Schutz M, Muller M, Regazzoni P, et al. Use of the less invasive stabilization
mRUST scores proved significant. system (LISS) in patients with distal femoral (AO33) fractures: a prospective
Follow-up duration is another limitation to the study. multicenter study. Arch Orthop Trauma Surg. 2005;125:102–108.
To strengthen the union comparison, subgroup analysis was 16. Schutz M, Muller M, Krettek C, et al. Minimally invasive fracture sta-
performed to include only patients with minimum 6-month bilization of distal femoral fractures with the LISS: a prospective multi-
center study. Results of a clinical study with special emphasis on difficult
follow-up. In addition, 3 authors independently reviewed the cases. Injury. 2001;32(suppl 3):SC48–SC54.
radiographs of each case deemed united in the medical record 17. Bottlang M, Lesser M, Koerber J, et al. Far cortical locking can improve
to unanimously confirm radiographic union. Despite this, it is healing of fractures stabilized with locking plates. J Bone Joint Surg Am.
possible that there are fractures that would be found to be not 2010;92:1652–1660.
united with longer follow-up. 18. Bottlang M, Tsai S, Bliven EK, et al. Dynamic stabilization with active
locking plates delivers faster, stronger, and more symmetric fracture-
healing. J Bone Joint Surg Am. 2016;98:466–474.
19. Bottlang M, Feist F. Biomechanics of far cortical locking. J Orthop
CONCLUSIONS Trauma. 2011;25:S21–S28.
To our knowledge, this is the first comparative study 20. Bottlang M, Doornink J, Fitzpatrick DC, et al. AAOS Scientific Exhibit
SE 81, 2010 Annual Meeting. New Orleans, LA.
between FCL and LP constructs. The FCL group was noted to 21. Adams JD Jr, Tanner SL, Jeray KJ. Far cortical locking screws in distal
have significantly higher mRUST scores indicating increased femur fractures. Orthopedics. 2015;38:e153–e156.
fracture callus at all periods. However, the differences seen 22. Bottlang M, Fitzpatrick DC, Sheerin D, et al. Dynamic fixation of distal
were small, and there was no significant difference in union femur fractures using far cortical locking screws: a prospective observa-
rates of FCL versus LP groups at a minimum 6-month follow- tional study. J Orthop Trauma. 2014;28:181–188.
23. Ries Z, Hansen K, Bottlang M, et al. Healing results of periprosthetic-
up, thus bringing the benefit of FCL into question. Further distal femur fractures treated with far cortical locking technology: a pre-
prospective study designs are needed to compare FCL to LP liminary retrospective study. Iowa Orthop J. 2013;33:7–11.
constructs and to investigate the role of interfragmentary 24. Meinberg EG, Agel J, Roberts CS, et al. Fracture and Dislocation Clas-
motion on callus formation in distal femur fractures. sification Compendium - 2018. J Orthop Trauma. 2018;32 Supplement 1
pp: S1–S170.
25. Orthopaedic Trauma Association Committee for Coding and Classifica-
REFERENCES tion. Fracture and dislocation compendium. J Orthop Trauma. 1996;
1. Kubiak EN, Fulkerson E, Strauss E, et al. The evolution of locked plates. 10(suppl 1):v–ix, 1–154.
J Bone Joint Surg Am. 2006;88(suppl 4):189–200. 26. Gustilo RB, Anderson JT. Prevention of infection in the treatment of one
2. Bottlang M, Doornink J, Fitzpatrick DC, et al. Far cortical locking can thousand and twenty-five open fractures of long bones: retrospective and
reduce stiffness of locked plating constructs while retaining construct prospective analyses. J Bone Joint Surg Am. 1976;58:453–458.
strength. J Bone Joint Surg Am. 2009;91:1985–1994. 27. Litrenta J, Tornetta PIII, Mehta S, et al. Determination of radiographic
3. Markmiller M, Konrad G, Sudkamp N. Femur-LISS and distal femoral healing: an assessment of consistency using RUST and modified RUST
nail for fixation of distal femoral fractures: are there differences in out- in metadiaphyseal fractures. J Orthop Trauma. 2015;29:516–520.
come and complications? ClinOrthopRelatRes. 2004;426:252–257. 28. Kooistra BW, Dijkman BG, Busse JW, et al. The radiographic union
4. Ristevski B, Nauth A, Williams DS, et al. Systematic review of the scale in tibial fractures: reliability and validity. J Orthop Trauma. 2010;
treatment of periprosthetic distal femur fractures. J Orthop Trauma. 24:S81–S86.
2014;28:307–312. 29. Whelan DB, Bhandari M, Stephen D, et al. Development of the radio-
5. Zlowodzki M, Williamson S, Cole PA, et al. Biomechanical evaluation graphic union score for tibial fractures for the assessment of tibial frac-
of the less invasive stabilization system, angled blade plate, and retro- ture healing after intramedullary fixation. J Orthop Trauma. 2010;68:
grade intramedullary nail for the internal fixation of distal femur frac- 629–632.
tures. J Orthop Trauma. 2004;18:494–502. 30. Leow JM, Clement ND, Tawonsawatruk T, et al. The radiographic union
6. Marsh JL, Jansen H, Yoong HK, et al. Supracondylar fractures of the scale in tibial (RUST) fractures. Bone Joint Res. 2016;5:116–121.
femur treated by external fixation. J Orthop Trauma. 1997;11:405–410. 31. Landis JR, Koch GG. The measurement of observer agreement for cat-
7. Bedes L, Bonnevialle P, Ehlinger M, et al. External fixation of distal egorical data. Biometrics. 1977;33:159–174.
femoral fractures in adults’ multicenter retrospective study of 43 patients. 32. Ricci WM, Loftus T, Cox C, et al. Locked plates combined with mini-
Orthop Traumatology Surg Res. 2014;100:867–872. mally invasive insertion technique for the treatment of periprosthetic
8. KregorPJ, Zlowodzki M, Stannard J, et al. Submuscular plating of the supracondylar femur fractures above a total knee arthroplasty. J Orthop
distal femur. Oper Tech Orthop. 2003;13:85–95. Trauma. 2006;20:190–196.
9. Hoffmann MF, Jones CB, Sietsema DL, et al. Clinical outcomes of 33. Henderson CE, Lujan TJ, Kuhl LL, et al. 2010 Mid-America Orthopaedic
locked plating of distal femoral fractures in a retrospective cohort. J Association Physician in Training Award: healing complications are
Orthop Surg Res. 2013;8:43. common after locked plating for distal femur fractures. ClinOrthopRelat
10. Fankhauser F, Gruber G, Schippinger G, et al. Minimal-invasive treat- Res. 2011;469:1757–1765.
ment of distal femoral fractures with the LISS (less invasive stabilization 34. Fitzpatrick DC, Doornink J, Madey SM, et al. Relative stability of con-
system): a prospective study of 30 fractures with a follow up of 20 ventional and locked plating fixation in a model of the osteoporotic
months. Acta Orthop Scand. 2004;75:56–60. femoral diaphysis. ClinBiomech (Bristol, Avon). 2009;24:203–209.
35. Lujan TJ, Henderson CE, Madey SM, et al. Locked plating of distal 38. Kooistra BW, Sprague S, Bhandari M, et al. Outcomes assessment in fracture
femur fractures leads to inconsistent and asymmetric callus formation. healing trials: a primer. J Orthop Trauma. 2010;24(suppl 1):S71–S75.
J Orthop Trauma. 2010;24:156–162. 39. Hammer RR, Hammerby S, Lindholm B. Accuracy of radiologic assess-
36. Ries ZG, Marsh JL. Far cortical locking technology for fixation of peripros- ment of tibial shaft fracture union in humans. ClinOrthopRelat Res.
thetic distal femur fractures: a surgical technique. J Knee Surg. 2013;26:15–18. 1985;199:233–238.
37. Bhandari M, Guyatt GH, Swiontkowski MF, et al. A lack of consensus in 40. Fitzpatrick D, Hansen K, Madey S, et al. The use of far cortical locking
the assessment of fracture healing among orthopaedic surgeons. J Orthop constructs for fixation of periprosthetic fractures. Tech Orthop. 2013;28:
Trauma. 2002;16:562–566. 260–264.
ORIGINAL ARTICLE
Hypoalbuminemia Is an Independent Risk Factor

for 30-Day Mortality, Postoperative Complications,
Readmission, and Reoperation in the Operative Lower
Extremity Orthopaedic Trauma Patient
Jacob M. Wilson, MD, Matthew P. Lunati, MD, Zachary J. Grabel, MD, Christopher A. Staley, BA,
Andrew M. Schwartz, MD, and Mara L. Schenker, MD
Key Words: malnutrition, orthopaedic trauma, complications, albu-

Introduction: Malnutrition, as indicated by hypoalbuminemia, is min, hypoalbuminemia, outcomes
known to have detrimental effects on outcomes after arthroplasty,
geriatric hip fractures, and multiple general surgeries. Hypoalbumi- Level of Evidence: Prognostic Level III. See Instructions for
nemia has been examined in the critically ill but has largely been Authors for a complete description of levels of evidence.
ignored in the orthopaedic trauma literature. We hypothesized that (J Orthop Trauma 2019;33:284–291)
admission albumin levels would correlate with postoperative course
in the nongeriatric lower extremity trauma patient.
INTRODUCTION
Methods: Patients with lower extremity (including pelvis and Malnutrition is common in surgical and hospitalized
acetabulum) fracture who underwent operative intervention were patients and is a well-established risk factor for morbidity.1–4
collected from the ACS-NSQIP database. Patients younger than 65 Albumin is often used as a marker for chronic nutritional sta-
years were included. Patient demographic data, complications, tus.5 Hypoalbuminemia has been associated with poor out-
length of stay, reoperation rate, and readmission rate were collected, comes in both orthopaedic and general surgery.6–9 Within
and patient modified frailty index scores were calculated. Poisson orthopaedics, extensive literature in the spine, total joint arthro-
regression with robust error variance was then conducted, controlling plasty, and geriatric hip fracture realms supports the idea that
for potential confounders. hypoalbuminemia (albumin ,3.5 g/dL) is associated with
postoperative complications.5,10–22 Recently, Blevins et al13
Results: Five thousand six hundred seventy-three patients with demonstrated that of 5 commonly used nutrition biomarkers,
albumin available were identified, and 29.6% had hypoalbuminemia.
low albumin had the highest specificity and positive predictive
Hypoalbuminemic patients had higher rates of postoperative com-
value for the development of prosthetic joint infection. From
plications [9.3% vs. 2.6%; relative risk (RR) 1.63] including increased
a biological perspective, malnutrition and protein deficiency
rates of: mortality (3.2% vs. 0.4%; RR 4.86, 95% confidence interval
are known to have detrimental effects on fracture healing.23,24
2.66–8.87), sepsis (1.5% vs. 0.5%, RR 2.35), and reintubation (2.3%
Perhaps, most pertinent to this discussion, it is known that
vs. 0.4%; RR 3.84). Reoperation (5.5% vs. 2.6%, RR 1.74) and read-
hypoalbuminemia increases both complications and mortality
mission (11.4% vs. 4.1%; RR 2.53) rates were also higher in patients
after surgery for hip fracture in the geriatric cohort.21,22,25,26
with low albumin.
However, the data regarding the effect of hypoalbuminemia on
Conclusion: Hypoalbuminemia is a powerful predictor of acute outcomes after surgery for orthopaedic trauma are lacking.
postoperative course and mortality after surgical fixation in non- A recent systematic review on the effect of malnutrition
geriatric, lower extremity orthopaedic trauma patients. Admission in the orthopaedic trauma patient suggested that malnutrition
albumin should be a routine part of the orthopaedic trauma workup. may be associated with increased wound complications,
Further study into the utility of supplementation is warranted, as this fracture nonunion, and immobility-associated decubitus ulcer-
may represent a modifiable risk factor. ation.26 However, most of the studies included only patients
with hip fracture, which generally consists of a frail, geriatric
cohort.27,28 Ultimately, the authors noted high-quality litera-
ture on the subject is scarce and concluded that further study
was needed.
Accepted for publication January 8, 2019.
From the Department of Orthopaedic Surgery, Emory University School of The purpose of this investigation was to determine the
Medicine, Atlanta, GA. impact of hypoalbuminemia on postoperative complication
The authors report no conflict of interest. rate, length of stay (LOS), reoperation rate, and readmission
Reprints: Jacob M. Wilson, MD, Department of Orthopaedic Surgery, Emory rate in nongeriatric (,65 years) patients with lower extremity
University School of Medicine, 59 Executive Park South, Atlanta, GA
30329 (e-mail: Jmwils8@emory.edu).
orthopaedic trauma. We hypothesize that hypoalbuminemia is
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. associated with adverse outcomes in patients who undergo
DOI: 10.1097/BOT.0000000000001448 surgery for lower extremity orthopaedic trauma.
284 | www.jorthotrauma.com J Orthop Trauma Volume 33, Number 6, June 2019
J Orthop Trauma Volume 33, Number 6, June 2019 Hypoalbuminemia and Orthopedic Trauma
METHODS the principle procedure. In all cases, this study used the
albumin level drawn most proximally to the procedure. Based
Data Collection on extensive precedence in the literature,14,21,22 albumin lev-
Patients included in this study were collected from the els were categorically defined as hypoalbuminemia (,3.5 g/
American College of Surgeons—National Surgery Quality dL), normal albumin ($3.5 g/dL), or missing albumin if
Improvement Program (ACS-NSQIP) database. The NSQIP unavailable.
database is a widely used,29,30 prospectively collected data-
base that collects preoperative and 30-day postoperative out- Outcome and Complication Data
come data for patients undergoing surgical operations. This To assess outcomes, 30-day postoperative complication
includes patients across multiple subspecialties who have data were collected for each patient. The collected complica-
undergone procedures performed at both academic and pri- tions included anemia requiring transfusion, cardiac arrest
vate institutions. The database has excellent follow-up and requiring cardiopulmonary resuscitation, unplanned intuba-
captures 95% of 30-day outcomes by observing in-hospital tion, cerebrovascular accident, deep vein thrombosis, pulmo-
morbidity and mortality through contacting patients via writ- nary embolism, MI, pneumonia, sepsis, surgical site infection,
ing and phone call at the end of the 30-day period. urinary tract infection (UTI), renal insufficiency, readmission,
In this study, the NSQIP database was searched for and reoperation. In addition, data for total complications were
appropriate patients using Current Procedural Terminology analyzed and Clavien-Dindo IV complications (life-
codes. Patients undergoing operative intervention for lower threatening complications with end-organ dysfunction).33
extremity orthopaedic trauma (pelvis and acetabulum to For this study, a patient was considered to have incurred
ankle) were included. This included the following Current a Clavien–Dindo IV complication if they had cardiac arrest,
Procedural Terminology codes grouped into 6 broad catego- MI, sepsis, pulmonary embolism, or renal failure. Finally,
ries for some aspects of analysis: hip (27235, 27236, 27244, LOS was collected.
27245, 27253, 27269); ankle/pilon (27766, 27769, 27784,
27792, 27814, 27822, 27823, 27826, 27827, 27828, 27829, Statistical Analysis
27846, 27848); femoral shaft (27506, 27507, 27509, 27511); Statistical analysis was performed using IBM SPSS
tibial shaft (27756, 27758, 27759); knee periarticular (27513, (IBM Corporation, Armonk, NY) statistical software. For the
27514, 27524, 27535, 27536); and pelvis and acetabulum purposes of analysis, we used Poisson regression with robust
(27215, 27217, 27218, 27226, 27227, 27228, 27254). error variance as an alternative to typical multivariate
Geriatric patients (.65 years old) were excluded from anal- regression as initially described by Zou34 and recently used
ysis. Finally, patients meeting sepsis or presepsis criteria by Bohl et al.21 This method allows for direct reporting of
before surgery were excluded. relative risks (RRs) and avoids the potential for overestima-
tion of risk as readers frequently misinterpret odds ratios as
Patient Demographic Information RRs. This method of analysis as an alternative to standard
Patient demographic and comorbid data were collected linear regression of binary data has become widely accepted
and compiled. These included sex, age, race, American and used.21,34 After identifying those patients with and with-
Society of Anesthesiology classification, body mass index out available albumin levels, bivariate Poisson regression
(BMI), wound classification, and smoking status. The fol- with robust error variance was conducted to assess association
lowing comorbidities were also collected for each patient and between availability of albumin and baseline patient charac-
were included if present within the 30 days preceding teristics and subsequently to assess the association of hypo-
surgery: dyspnea on exertion (DOE), diabetes mellitus albuminemia with baseline characteristics. This multivariate
(DM), congestive heart failure (CHF), anemia (defined as analysis controlled for all baseline characteristics including
hematocrit ,41% in men and ,36% in women), hyperten- age, sex, BMI, mFI, DM, CHF, DOE, HTN, open wound
sion (HTN), acute renal failure, and chronic obstructive pul- or infection, COPD, current smoking status, anemia, acute
monary disease (COPD). In addition, given the known renal failure, and region of injury. A P value of less than
association with postoperative complications in this cohort,31 0.05 was considered significant for this analysis. Analysis
we calculated the 11-item modified frailty index (mFI) scores was then performed comparing postoperative complication
as previously described27,32 for each patient. Briefly, this rates between patients with hypoalbuminemia and normal
index includes the following patient history items: DM, albumin levels again using multivariate Poisson regression
CHF, HTN, COPD, transient ischemic attack or cerebrovas- with robust error variance. In addition to the control variables
cular incident, nonindependent functional status, myocardial mentioned above, LOS and operative time were also con-
infarction (MI), peripheral vascular disease, cerebrovascular trolled for in this analysis, as they were felt to possibly affect
incident with neurological deficit, angina or previous percu- complication rates. Risk of death was then analyzed using the
taneous coronary intervention, and impaired sensorium. The same method comparing patients with hypoalbuminemia to
number of items present for each patient was then tabulated those with normal albumin, and then another analysis exam-
and divided by 11 to calculate each patient’s mFI score. These ined albumin as a continuous variable. All patients were
data were compiled and used in multivariate analysis. included in every analysis, and patients without available
In addition, preoperative albumin levels were collected albumin were coded using a missing variable. LOS was ana-
for each patient. Serum albumin levels were available for lyzed using a nonparametric t test, as this variable was not
patients who had an albumin level drawn within 90 days of normally distributed.
Wilson et al J Orthop Trauma Volume 33, Number 6, June 2019
RESULTS extremity fracture were identified from the NSQIP database.

The average patient in our cohort was 46.69 years old
Patient Demographics and Baseline (613.62 years; $65 years old excluded) and was overweight
Characteristics by BMI (average BMI 29.3 6 7.62). There was a slight
From 2006 to 2014, 17,510 patients younger than 65 female predominance at 52.9%. Most of the patients had zero
years who underwent operative intervention of a lower or one frailty comorbidity (Table 1).
TABLE 1. Analysis of the Rate at Which Albumin Was Available for Analysis
# of Patients with Available
Demographic Characteristic # of Patients Initially Identified Albumin Level (%) RR P
Overall 17,510 5673 (32.4)
Age 0.001
18–25 1739 236 (13.6) Reference
26–35 2613 451 (17.3) 1.27 (1.10–1.47)
36–45 2895 745 (25.7) 1.90 (1.66–2.17)
46–55 4175 1487 (35.6) 2.62 (2.32–2.98)
56–65 6088 2754 (45.2) 3.33 (2.95–3.77)
Sex ,0.001
Male 8248 2485 (30.1) Reference
Female 9252 3186 (34.4) 1.14 (1.10–1.19)
BMI ,0.001
,18 kg/m2 487 288 (59.1) 1.98 (1.82–2.15)
18–19.9 kg/m2 4581 1540 (33.6) 1.12 (1.06–1.19)
20–24.9 kg/m2 5661 1694 (29.9) Reference
25–29.9 kg/m2 3562 1067 (30.0) 1.00 (0.94–1.07)
30–34.9 kg/m2 1814 569 (31.4) 1.05 (0.97–1.13)
$35 kg/m2 1405 515 (36.7) 1.23 (1.13–1.33)
mFI ,0.001
0.00 11,191 2704 (24.2) Reference
0.09 4017 1660 (41.3) 1.71 (1.63–1.80)
0.18 1764 956 (54.2) 2.24 (2.13–2.37)
0.27 398 250 (62.8) 2.60 (2.39–2.82)
$0.36 140 103 (73.6) 3.05 (2.74–3.38)
Dyspnea on exertion ,0.001
No 11,593 5411 (31.8) Reference
Yes 244 262 (51.8) 1.63 (1.49–1.78)
Open wound or infection ,0.001
No 16,827 5359 (31.8) Reference
Yes 683 314 (46.0) 1.44 (1.33–1.57)
Current smoker ,0.001
No 11,912 3672 (30.8) Reference
Yes 5598 2001 (35.7) 1.16 (1.11–1.21)
Anemia ,0.001
No 16,829 5259 (31.2) Reference
Yes 681 414 (60.8) 1.95 (1.82–2.08)
Acute renal failure ,0.001
No 17,461 5634 (32.3) Reference
Yes 49 39 (79.6) 2.47 (2.14–2.85)
Region of injuries ,0.001
Hip 3456 1996 (57.8) 2.52 (2.41–2.64)
Ankle 9467 2167 (22.9) Reference
Knee periarticular 2301 682 (29.6) 1.30 (1.20–1.39)
Femur 657 325 (49.5) 2.16 (1.98–2.35)
Tibia 1398 409 (29.3) 1.28 (1.17–1.40)
Pelvic and acetabulum 231 94 (40.7) 1.78 (1.52–2.09)
RRs are presented with the 95% CIs in parentheses.
TABLE 2. Analysis of Risk Factors for Hypoalbuminemia (Serum Albumin #3.5 g/dL)
Percentage of Patients With Multivariate Comparisons
Risk Factors Hypoalbuminemia (%) RR (95% CI) P
Overall 29.6
Age 0.002
18–25 12.3 Reference
26–35 15.1 1.21 (0.82–1.80)
36–45 19.7 1.43 (0.99–2.05)
46–55 29.4 1.63 (1.15–2.31)
56–65 36.2 1.67 (1.18–2.36)
Sex 0.10
Male 27.4 Reference
Female 31.2 1.11 (1.03–1.20)
BMI ,0.001
,18 kg/m2 55.6 1.61 (1.39–1.87)
18–19.9 kg/m2 35.0 1.28 (1.16–1.43)
20–24.9 kg/m2 23.6 Reference
25–29.9 kg/m2 23.6 1.01 (0.88–1.15)
30–34.9 kg/m2 27.1 1.13 (0.97–1.32)
$35 kg/m2 33.8 1.24 (1.07–1.44)
mFI ,0.001
0.00 19.5 Reference
0.09 33.0 1.63 (1.43–1.85)
0.18 44.5 1.99 (1.64–2.41)
0.27 48.8 1.90 (1.47–2.47)
$0.36 56.3 1.95 (1.43–2.68)
DM 0.19
No 26.1 Reference
Yes 44.2 1.16 (1.03–1.30)
CHF 0.12
No 29.2 Reference
Yes 61.4 1.20 (0.96–1.51)
Dyspnea on exertion 0.76
No 28.7 Reference
Yes 47.3 1.13 (0.99–1.30)
Hypertension 0.001
No 24.1 Reference
Yes 37.4 0.81 (0.72–0.92)
Open wound or infection ,0.001
No 28.4 Reference
Yes 49.7 1.42 (1.26–1.60)
COPD 0.25
No 27.9 Reference
Yes 49.5 0.92 (0.81–1.06)
Current smoker ,0.001
No 27.0 Reference
Yes 34.4 1.22 (1.12–1.32)
Anemia ,0.001
No 27.7 Reference
Yes 53.1 1.41 (1.27–1.56)
Acute renal failure ,0.001
No 29.3 Reference
Yes 74.4 1.54 (1.22–1.94)
Region of injuries ,0.001
Hip 43.6 1.79 (1.60–2.01)
(continued on next page )
TABLE 2. (Continued ) Analysis of Risk Factors for Hypoalbuminemia (Serum Albumin #3.5 g/dL)
Percentage of Patients With Multivariate Comparisons
Risk Factors Hypoalbuminemia (%) RR (95% CI) P
Ankle 17.8 Reference
Knee periarticular 27.0 1.34 (1.16–1.56)
Femur 37.5 1.67 (1.41–1.97)
Tibia 20.3 1.12 (0.91–1.37)
Pelvic and acetabulum 35.1 1.62 (1.20–2.17)
Prevalence of hypoalbuminemia = patients with serum albumin ,3.5 g/dL divided by the # of patients with available serum albumin. Adjusted for all other risk factors listed in this
table.
Availability of Albumin Levels and Risk Factors risk of mortality [3.2% (albumin #3.5 g/dL) vs. 0.4%
for Hypoalbuminemia (albumin .3.5 g/dL), RR 3.8, 95% CI 2.09–6.74, P ,
Of the 17,510 patients identified, 5673 (32.4%) had 0.001]. This association was also found when examining
albumin levels available for analysis. Our analysis demonstrated albumin as a continuous variable as risk of mortality was
that patients were more likely to have albumin levels available if inversely correlated with serum albumin concentration (RR
they were older, female, had a very low (,20 kg/m2) or very 0.25, 95% CI 0.18–0.34, P , 0.001) (Fig. 1C).
high ($35 kg/m2) BMI, or had a higher mFI. Specifically,
albumin was more commonly available for patients with the Hypoalbuminemia, Readmission, Reoperation,
following comorbidities: DM, CHF, DOE, HTN, open and Length of Stay
wound or infection, COPD, current smoking status, anemia, Patients with serum albumin #3.5 g/dL, when com-
or acute renal failure. Of the regions of injury, patients with pared to normoalbuminemic patients, were also found to have
ankle fracture had the lowest proportion of patients with increased risk of readmission within 30 days (11.4% vs.
albumin available, and all 4 other regions (pelvis and ace- 4.1%, respectively, RR 2.0, 95% CI 1.55–2.57, P , 0.001)
tabulum, femur, tibia, and hip) had a significantly higher and reoperation (5.5% vs. 2.6%, respectively, RR 1.52, 95%
proportion of patients with available albumin when com- CI 1.11–2.07, P = 0.009) (Fig. 1D). In addition, hypoalbumi-
pared with patients with ankle fracture (Table 1). nemia was associated with longer lengths of stay as those
Multivariate analysis revealed that patients who were with albumin #3.5 g/dL had a mean LOS of 7.5 (610.45)
46 years and older, had a BMI ,20 kg/m2 or $35 kg/m2, days while those with normal albumin levels had a mean LOS
had an mFI score of $0.09, had an open wound or infection, of 3.57 (65.0) days (P , 0.001) (Fig. 1E).
were active smokers, were anemic, had acute renal failure, or
had a fracture other than an ankle fracture or tibial shaft
fracture were more likely to have hypoalbuminemia (Table
DISCUSSION
2). In this cohort, HTN was inversely associated with hypo- Hypoalbuminemia is often used as a surrogate marker
albuminemia [RR 0.81, 95% confidence interval (CI) for malnutrition and is a common risk factor in surgical and
hospitalized patients.1–3,5 Total joint arthroplasty, hip frac-
0.72–0.92, P , 0.001].
ture, and spine literature suggest low albumin levels are asso-
ciated with adverse postoperative outcomes.10–20 A recent
Hypoalbuminemia, Postoperative study investigated the effect of hypoalbuminemia in patients
Complications, and Risk of Mortality with total joint arthroplasty and showed it was associated with
Overall, 1678 (29.6%) of patients with albumin avail- higher intensive care admission rate, hospital readmission
able for analysis were found to have hypoalbuminemia. The rate, and emergency department visits resulting in a mean
mean albumin level was 3.72 (SD 0.63). The distribution of 90-day charge increase of $9270.35 Koval et al22 demon-
albumin levels can be seen in Fig. 1A. Patients with hypo- strated that hypoalbuminemia is associated with increased
albuminemia were found to have higher rates of incurring at risk of mortality, increased LOS, and inability to regain pre-
least one postoperative complication when compared to pa- operative independence in patients with hip fracture. Bohl
tients with serum albumin .3.5 g/dL (9.3% vs. 4.3%, respec- et al21 subsequently demonstrated an association with 30-
tively, RR 1.46, 95% CI 1.30–1.64, P , 0.001) (Fig. 1B). day mortality and complications in patients with hip fracture.
Specifically, these patients had significantly increased rates of Similar relationships exist in patients after undergoing spine
the following complications: anemia requiring transfusion, surgery.12 However, the relationship between hypoalbumine-
cardiac arrest requiring resuscitation, renal insufficiency, sep- mia and postoperative course in a young orthopaedic trauma
sis, unplanned intubation, and UTI (P , 0.05, Table 3). cohort is not well reported.
Patients with hypoalbuminemia had higher rates of Malnutrition is an important risk factor to recognize, as
Clavien–Dindo IV complications when compared to those it is potentially modifiable. Although the nature of trauma
with normal albumin (4.1% vs 1%, respectively, RR 2.0, 95% precludes the ability to prehabilitate36 or correct nutritional
CI 1.3–3.09, P = 0.002). This translated into an increased status preoperatively, there has been suggestion that
FIGURE 1. A, Distribution of serum albumin concentrations for patients included in study, B, risk of complication by preoperative
serum albumin, C, risk of mortality by preoperative serum albumin, D, risk of reoperation by preoperative serum albumin, and E,
LOS by preoperative serum albumin.
TABLE 3. Hypoalbuminemia and Morbidity After Lower Extremity Orthopaedic Trauma

Normal Albumin Hypoalbuminemia Missing Albumin
Complications (n = 3995) (%) (n = 1678) (%) (n = 11,837) (%) RR* P
Significant association:
Anemia requiring transfusion 322 (8.1) 362 (21.6) 394 (3.3) 1.47 (1.28–1.69) ,0.001
Cardiac arrest requiring 6 (0.2) 16 (1.0) 9 (0.1) 3.45 (1.20–9.93) 0.022
cardiopulmonary resuscitation
Renal insufficiency 2 (0.1) 11 (0.7) 3 (0.0) 6.10 (1.19–31.28) 0.030
Sepsis 18 (0.5) 26 (1.5) 23 (0.2) 1.99 (1.03–3.86) 0.041
Unplanned intubation 15 (0.4) 39 (2.3) 19 (0.2) 2.95 (1.49–5.84) 0.002
UTI 51 (1.3) 57 (3.4) 54 (0.5) 1.58 (1.04–2.41) 0.033
No significant association:
Cerebrovascular accident 2 (0.1) 1 (0.1) 4 (0.0) 0.62 (0.02–24.3) 0.801
Deep vein thrombosis 28 (0.7) 16 (1.0) 51 (0.4) 1.01 (0.52–1.94) 0.989
MI 10 (0.3) 6 (0.4) 10 (0.1) 0.65 (0.18–2.38) 0.520
Pneumonia 27 (0.7) 37 (2.2) 34 (0.3) 1.75 (0.98–3.13) 0.059
Pulmonary embolism 15 (0.4) 10 (0.6) 47 (0.4) 0.86 (0.35–2.08) 0.732
Infection 72 (1.8) 56 (3.3) 169 (1.4) 1.38 (0.94–2.02) 0.101
All values are presented as n (%).
*Multivariate analysis adjusted for region of injury, BMI, LOS, total operative time, sex, age, smoking status, modified frailty index, and the presence of an open wound, dyspnea,
diabetes, CHF, anemia, hypertension, acute renal failure, and COPD. The values are reported as RR, with the 95% CI in parentheses. P , 0.05 considered significant.
postoperative supplementation can reduce the incidence of In conclusion, this investigation demonstrates hypoal-
complications in patients with hip fractures.26,37–41 Beyond buminemia was associated with a significant increase in
short-term complications, correction of malnutrition is likely complications, including mortality, increased hospital LOS,
important for achieving bony union in the setting of fracture readmission rates, and reoperation rates for patients who
care, and reversal of malnutrition could conceivably reverse underwent surgery for lower extremity orthopaedic trauma.
its detrimental effects.23,24 However, there is a limited evi- Although this study cannot determine whether albumin is
dence for or against nutritional supplementation after ortho- a reliable marker for nutrition status in the orthopaedic trauma
paedic trauma. patient, this study reveals it has strong prognostic implica-
Our patient population had an average age of 46 years, tions and may be a useful indicator of patients requiring
an average BMI of 29, and most patients had zero or one frailty specialized postoperative pathways to prevent undue out-
comorbidity (Table 1). This distinguishes our group from pre- comes. Although previous studies have established a decrease
vious studies.12,21,22,26 In addition, the relatively younger age in complications due to nutritional supplementation after
and healthy nature of the study cohort diminish potential con- geriatric hip fracture,44,45 further work is needed to delineate
founding factors that may affect patient outcomes. whether albumin is a modifiable risk factor in the setting of
The results of this investigation demonstrate that orthopaedic trauma.
hypoalbuminemia is independently associated with increased
postoperative morbidity. This includes increased rates of
overall complications (including significantly increased rates REFERENCES
of anemia requiring transfusion, cardiac arrest requiring 1. Leandro-Merhi VA, de Aquino JL. Determinants of malnutrition and
resuscitation, renal insufficiency, sepsis, unplanned intuba- post-operative complications in hospitalized surgical patients. J Health
Popul Nutr. 2014;32:400–410.
tion, and UTI), Clavien–Dindo IV (life-threatening) compli- 2. Sungurtekin H, Sungurtekin U, Balci C, et al. The influence of nutritional
cations, and mortality. Hypoalbuminemia was also associated status on complications after major intraabdominal surgery. J Am Coll
with increased 30-day readmission rate, reoperation rate, and Nutr. 2004;23:227–232.
hospital LOS. These relationships were strong, and even 3. Burden S, Todd C, Hill J, et al. Pre-operative nutrition support in patients
when controlling for confounders, hypoalbuminemia had undergoing gastrointestinal surgery. Cochrane Database Syst Rev. 2012;
11:CD008879.
a RR of 3.8 for mortality. 4. Sung J, Bochicchio GV, Joshi M, et al. Admission serum albumin is
It is worth noting that although hypoalbuminemia is predictive of outcome in critically ill trauma patients. Am Surg. 2004;70:
typically associated with malnutrition, it also acts as an acute- 1099–1102.
phase reactant. Both trauma and surgery cause an increase in the 5. Cross MB, Yi PH, Thomas CF, et al. Evaluation of malnutrition in
orthopaedic surgery. J Am Acad Orthop Surg. 2014;22:193–199.
body’s inflammatory products and can lead to a reduction in 6. Morisaki K, Yamaoka T, Iwasa K. Risk factors for wound complications
protein in the acute phase due to an overly catabolic and hyper- and 30-day mortality after major lower limb amputations in patients with
metabolic response to stress.26,42,43 These acute changes in peripheral arterial disease. Vascular. 2018;26:12–17.
serum marker levels may challenge the validity of the “tradi- 7. Wohlauer M, Brier C, Kuramochi Y, et al. Preoperative hypoalbumine-
tional” methods for malnutrition assessment, such as albumin. mia is a risk factor for early and late mortality in patients undergoing
endovascular juxtarenal and thoracoabdominal aortic aneurysm repair.
However, our data demonstrate that, although we cannot empir- Ann Vasc Surg. 2017;42:198–204.
ically prove that albumin is an accurate marker of nutrition in the 8. Inagaki E, Farber A, Eslami MH, et al. Preoperative hypoalbuminemia is
acute trauma setting, it is still reliably prognostic for morbidity. associated with poor clinical outcomes after open and endovascular
There are several limitations in this study, most of abdominal aortic aneurysm repair. J Vasc Surg. 2017;66:53–63 e1.
9. Peacock MR, Farber A, Eslami MH, et al. Hypoalbuminemia predicts
which are inherent to large database studies and the NSQIP perioperative morbidity and mortality after infrainguinal lower
database, in particular. The NSQIP database only includes extremity bypass for critical limb ischemia. Ann Vasc Surg. 2017;
complications observed within 30 days of surgery. This may 41:169–175 e4.
falsely lower the number of complications for analysis as late 10. Adogwa O, Elsamadicy AA, Mehta AI, et al. Preoperative nutritional
status is an independent predictor of 30-day hospital readmission after
complications are not available. In addition, many orthopae- elective spine surgery. Spine (Phila Pa 1976). 2016;41:1400–1404.
dic outcomes of interest are not collected by the database, and 11. Adogwa O, Martin JR, Huang K, et al. Preoperative serum albumin level
functional outcome data are not available and cannot be as a predictor of postoperative complication after spine fusion. Spine
analyzed in this study. In addition, only 32.4% of patients (Phila Pa 1976). 2014;39:1513–1519.
studied had albumin levels available for analysis. This is 12. Phan K, Kim JS, Xu J, et al. Nutritional insufficiency as a predictor for
adverse outcomes in adult spinal deformity surgery. Global Spine J.
a significantly lower percentage of recorded albumin levels 2018;8:164–171.
than the hip fracture population studied by Bohl et al16 where 13. Blevins K, Aalirezaie A, Shohat N, et al. Malnutrition and the develop-
60.1% of patients had available albumin levels. We speculate ment of periprosthetic joint infection in patients undergoing primary
that the lower percent of recorded albumin is likely due to the elective total joint arthroplasty. J Arthroplasty. 2018;33:2971–2975.
study population being a younger, generally healthier popu- 14. Bohl DD, Shen MR, Kayupov E, et al. Hypoalbuminemia independently
predicts surgical site infection, pneumonia, length of stay, and readmis-
lation in which preoperative albumin levels are not standard sion after total joint arthroplasty. J Arthroplasty. 2016;31:15–21.
of care. This created a potential selection bias, as those with 15. Bohl DD, Shen MR, Mayo BC, et al. Malnutrition predicts infectious and
increasing frailty (as indicated by mFI) were more likely to wound complications following posterior lumbar spinal fusion. Spine
have albumin available for analysis. However, patients with (Phila Pa 1976). 2016;41:1693–1699.
16. Bohl DD, Shen MR, Kayupov E, et al. Is hypoalbuminemia associated
all mFI scores were well represented in the group analyzed, with septic failure and acute infection after revision total joint arthro-
and frailty was controlled for in the multivariable analysis to plasty? A study of 4517 patients from the National Surgical Quality
help mitigate this influence. Improvement Program. J Arthroplasty. 2016;31:963–967.
17. Garcia GH, Fu MC, Dines DM, et al. Malnutrition: a marker for dic injuries, 2018. Presented at the American Academy of Orthopedic
increased complications, mortality, and length of stay after total shoulder Surgeons, June 03, 2018; New Orleans, LA.
arthroplasty. J Shoulder Elbow Surg. 2016;25:193–200. 32. Farhat JS, Velanovich V, Falvo AJ, et al. Are the frail destined to fail?
18. Greene KA, Wilde AH, Stulberg BN. Preoperative nutritional status of Frailty index as predictor of surgical morbidity and mortality in the elderly.
total joint patients. Relationship to postoperative wound complications. J J Trauma Acute Care Surg. 72:1526–1530, 2012; discussion 1530–1.
Arthroplasty. 1991;6:321–325. 33. Dindo D, Demartines N, Clavien PA. Classification of surgical compli-
19. Lavernia CJ, Sierra RJ, Baerga L. Nutritional parameters and short term cations: a new proposal with evaluation in a cohort of 6336 patients and
outcome in arthroplasty. J Am Coll Nutr. 1999;18:274–278. results of a survey. Ann Surg. 2004;240:205–213.
20. Huang R, Greenky M, Kerr GJ, et al. The effect of malnutrition on 34. Zou G. A modified Poisson regression approach to prospective studies
patients undergoing elective joint arthroplasty. J Arthroplasty. 2013;28: with binary data. Am J Epidemiol. 2004;159:702–706.
21–24. 35. Shroer WC, Diesfeld PJ, LeMarr AR, et al. Modifiable risk factors in
21. Bohl DD, Shen MR, Hannon CP, et al. Serum albumin predicts survival primary joint arthroplasty increase 90-day cost of care. J Arthroplasty.
and postoperative course following surgery for geriatric hip fracture. J 2018;33:2740–2744.
Bone Joint Surg Am. 2017;99:2110–2118. 36. Le Roy B, Selvy M, Slim K. The concept of prehabilitation: what the
22. Koval KJ, Maurer SG, Su ET, et al. The effects of nutritional status on surgeon needs to know? J Visc Surg. 2016;153:109–112.
outcome after hip fracture. J Orthop Trauma. 1999;13:164–169. 37. Anbar R, Beloosesky Y, Cohen J, et al. Tight calorie control in geriatric
23. Day SM, DeHeer DH. Reversal of the detrimental effects of chronic patients following hip fracture decreases complications: a randomized,
protein malnutrition on long bone fracture healing. J Orthop Trauma. controlled study. Clin Nutr. 2014;33:23–28.
2001;15:47–53. 38. Bell JJ, Bauer JD, Capra S, et al. Multidisciplinary, multi-modal nutri-
24. Jones CB. Biological basis of fracture healing. J Orthop Trauma. 2005; tional care in acute hip fracture inpatients—results of a pragmatic inter-
19:S1–S3. vention. Clin Nutr. 2014;33:1101–1107.
25. O’Daly BJ, Walsh JC, Quinlan JF, et al. Serum albumin and 39. Eneroth M, Olsson UB, Thorngren KG. Nutritional supplementation
total lymphocyte count as predictors of outcome in hip fractures. Clin decreases hip fracture-related complications. Clin Orthop Relat Res.
Nutr. 2010;29:89–93. 2006;451:212–217.
26. Ernst A, Wilson JM, Ahn J, et al. Malnutrition and the orthopaedic 40. Hommel A, Bjorkelund KB, Thorngren KG, et al. Nutritional status
trauma patient: a systematic review of the literature. J Orthop Trauma. among patients with hip fracture in relation to pressure ulcers. Clin Nutr.
2018;32:491–499. 2007;26:589–596.
27. Vu CCL, Runner RP, Reisman WM, et al. The frail fail: increased 41. Olofsson B, Stenvall M, Lundstrom M, et al. Malnutrition in hip fracture
mortality and post-operative complications in orthopaedic trauma pa- patients: an intervention study. J Clin Nurs. 2007;16:2027–2038.
tients. Injury. 2017;48:2443–2450. 42. Desborough JP. The stress response to trauma and surgery. Br J Anaesth.
28. Patel KV, Brennan KL, Brennan ML, et al. Association of a modified 2000;85:109–117.
frailty index with mortality after femoral neck fracture in patients aged 60 43. Monk DN, Plank LD, Franch-Arcas G, et al. Sequential changes in the
years and older. Clin Orthop Relat Res. 2014;472:1010–1017. metabolic response in critically injured patients during the first 25 days
29. Bohl DD, Singh K, Grauer JN. Nationwide Databases in Orthopaedic after blunt trauma. Ann Surg. 1996;223:395–405.
Surgery Research. J Am Acad Orthop Surg. 2016;24:673–682. 44. Lawson RM, Doshi MK, Barton JR, et al. The effect of unselected post-
30. Molina CS, Thakore RV, Blumer A, et al. Use of the National Surgical operative nutritional supplementation on nutritional status and clinical
Quality Improvement Program in orthopaedic surgery. Clin Orthop Relat outcome of orthopaedic patients. Clin Nutr. 2003;22:39–46.
Res. 2015;473:1574–1581. 45. Delmi M, Rapin CH, Bengoa JM, et al. Dietary supplementation in
31. Rege RM, Staley CA, Runner R, et al. Frailty predicts mortality and elderly patients with fractured neck of the femur. Lancet. 1990;335:
complications in chronologically young patients with traumatic orthope- 1013–1016.
ORIGINAL ARTICLE
Loss of Independence After Operative Management of

Femoral Neck Fractures
Emil H. Schemitsch, MD, FRCSC,* Sheila Sprague, PhD,†‡ Martin J. Heetveld, MD,§
Sofia Bzovsky, MSc,‡ Diane Heels-Ansdell, MSc,† Qi Zhou, PhD,† Marc Swiontkowski, MD,║
and Mohit Bhandari, MD, PhD, FRCSC†‡ on behalf of the FAITH Investigators
Methods: We conducted a descriptive analysis to quantify patients’

Objectives: The FAITH trial evaluated effects of sliding hip screws changes in living status, use of aids, and used multivariable Cox
versus cancellous screws in femoral neck fracture patients. Using regression analyses to determine factors associated with living and
FAITH trial data, we quantified changes in living status, use of aids, walking independently after fracture.
and investigated factors associated with living and walking inde-
pendently 12 months after fracture. Results: Of patients who lived independently before hip fracture,
3.07% (50–80 years old) and 19.81% (.80 years old) were institu-
Accepted for publication December 29, 2018. tionalized 12 months after injury. Of patients who were walking
From the *Department of Surgery, University of Western Ontario, London, independently before injury, 33.62% (50–80 years old) and
ON, Canada; †Department of Health Research Methods, Evidence, and 69.34% (.80 years old) required a walking aid 12 months after
Impact, McMaster University, Hamilton, ON, Canada; ‡Division of Ortho-
paedic Surgery, Department of Surgery, McMaster University, Hamilton,
injury. Factors associated with higher chances of living indepen-
ON, Canada; §Department of Surgery, Spaarne Gasthuis, Haarlem, The dently included the following: being between 50 and 80 years, hav-
Netherlands; and ║Department of Orthopaedic Surgery, University of Min- ing a class I American Society of Anesthesiologists classification,
nesota, Minneapolis, MN. not using a walking aid before fracture, being a nonsmoker or former
The FAITH study was supported by research grants from the Canadian smoker, and having an acceptable quality of implant placement.
Institutes of Health Research (MOP-106630 and MCT-87771), the
National Institutes of Health (1R01AR055267-01A1), Stichting NutsOhra Factors associated with higher chances of walking without an aid
(SNO-T-0602-43), the Netherlands Organization for Health Research and included the following: being between 50 and 80 years, having
Development (80-82310-97-11032), and Physicians’ Services Incorpo- a class I American Society of Anesthesiologists classification, living
rated. independently before fracture, being a nonsmoker or former smoker,
E. H. Schemitsch reports personal fees from Stryker, personal fees from
Smith & Nephew, personal fees from Zimmer, personal fees from
having an undisplaced fracture, and not requiring revision surgery.
Acumed, personal fees from Amgen, personal fees from Sanofi, and
personal fees from Pendopharm, outside the submitted work. S. Sprague
Conclusions: Femoral neck fracture patients suffer great losses of
reports employment/salary from McMaster University, and other from the independence. Identifying factors associated with living and walking
Global Research Solutions, outside the submitted work. M. J. Heetveld independently after hip fracture may help surgeons better identify
reports grants from Stichting NutsOhra, grants from the Netherlands which patients are at risk and optimize care of patients with this
Organization for Health Research and Development, during the conduct injury.
of the study. M. Swiontkowski reports grants from the National Institutes
of Health (NIH)/National Institute of Arthritis and Musculoskeletal and Key Words: femoral neck fractures, cancellous screws, sliding hip
Skin Diseases (NIAMS), during the conduct of the study; other from JBJS
Editor, outside the submitted work. M. Bhandari reports grants from the
screws, patient autonomy, fragility fracture, mobility, activities of
Canadian Institutes of Health Research, grants from the National Insti- daily living, functional recovery, aged
tutes of Health, grants from Stichting NutsOhra, grants from the Nether-
lands Organization for Health Research and Development, grants from Level of Evidence: Prognostic Level II. See Instructions for
Physicians’ Services Incorporated, and grants from Stryker Inc, during the Authors for a complete description of levels of evidence.
conduct of the study; grants and personal fees from Stryker Inc, personal
fees from Smith & Nephew, grants and personal fees from Amgen, grants (J Orthop Trauma 2019;33:292–300)
from DePuy, grants and personal fees from Eli Lilly, grants and personal
fees from DJO Global Inc, personal fees from Zimmer, personal fees from INTRODUCTION
Ferring, and grants from the Canada Research Chair in Musculoskeletal
Trauma, outside the submitted work. The remaining authors have no Hip fractures are a common type of fragility fracture
conflicts of interest. that impacts approximately 1.6 million individuals globally
Presented in part at the Annual Meeting of the Orthopaedic Trauma each year and can lead to significant morbidity and mortality
Association, October 11, 2017, Vancouver, BC, Annual Meeting of the in older patients.1,2 The associated morbidity can oftentimes
American Academy of Orthopaedic Surgeons, March 8, 2018, New
Orleans, LA, and at the Annual Meeting of the Canadian Orthopaedic lead to decreased mobility resulting in the need for a walking
Association, June 20–23, 2018, Victoria, BC. aid, a loss of independence, change in patient living status,
Members of FAITH Investigators are listed in Appendix 1. and an inability to accomplish activities of daily living.
Reprints: Emil H. Schemitsch, MD, FRCSC, Department of Surgery, Recovery of walking ability after a hip fracture is required
University of Western Ontario, 339 Windermere Rd, Box 5339, London,
ON N6A 5A5 (e-mail: Emil.Schemitsch@lhsc.on.ca).
for reestablishing patients into their normal preinjury environ-
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. ment, but at this time, the causes of poor recovery are not
DOI: 10.1097/BOT.0000000000001444 fully understood.3
J Orthop Trauma Volume 33, Number 6, June 2019 Loss of Independence
The current literature suggests that only 40% of ankle, tibia, fibula, knee, or femur; dislocations of the ankle,
recovering hip fracture patients are able to achieve their knee, or hip; or femoral head defects or fracture); (3) retained
prefracture ambulatory level and that only 20% reach their hardware around the affected hip; (4) infection around the hip
prefracture level in advanced mobility activities.4,5 Examin- (ie, soft tissue or bone); (5) patients with disorders of known
ing independent patient living status, the need for walking bone metabolism except osteoporosis (ie, Paget disease, renal
aids, and precisely the type of aids most commonly used after osteodystrophy, and osteomalacia); (6) patients with a history
a hip fracture, in a large cohort of patients, can help surgeons of frank dementia that would interfere with assessment of the
understand the natural history of functional recovery after hip primary outcome (ie, revision surgery at 24 months); (7)
fracture surgery and, ultimately, help surgeons to inform their likely problems, in the judgment of the investigators, with
patients and caregivers of the risk of loss of independence maintaining follow-up (ie, patients with no fixed address,
after a hip fracture. In addition, understanding which factors report a plan to move out of town, or intellectually challenged
are associated with living and walking independently after patients without adequate family support); and (8) exclusion
a hip fracture can help surgeons better identify which patients of a patient because of enrollment in another ongoing drug or
are at risk of loss of independence and mobility after a hip surgical intervention trial will be left to the discretion of the
fracture. Ultimately, being aware of these factors will aid attending surgeon, on a case-by-case basis.7 Patients were
surgeons in making treatment decisions and optimize the care assessed clinically at 1 and 10 weeks and 6, 9, 12, 18, and
of hip fracture patients. 24 months after surgery. The primary outcome of the FAITH
The recently completed fixation using alternative im- trial was revision surgery to promote healing, relieve pain,
plants for the treatment of hip fractures (FAITH)-randomized treat infection, or improve function over 24 months after
controlled trial evaluated the effects of sliding hip screws surgery.6,7 All revision surgeries were reviewed by a Central
versus cancellous screws in 1079 patients 50 years of age or Adjudication Committee. The trial protocol and results have
older with a low-energy femoral neck fracture with any been previously published.6,7 The trial was approved by the
degree of displacement.6 Using data from this multicenter Hamilton Integrated Research Ethics Board (#06–402) and by
trial, we quantified how patients’ living status and use of aids all participating clinical sites’ research ethics boards/
change over the 12 months after their femoral neck fracture. institutional review boards.
Changes in Living Status

PATIENTS AND METHODS We conducted a secondary analysis of the FAITH data
to descriptively quantify patients’ changes in living status
FAITH Study Overview over the 12 months after their femoral neck fracture. Specif-
The FAITH trial was an international, concealed, ically, for patients who were living independently before hip
randomized controlled trial (Clinical Trials Identification fracture, we determined the proportion of patients who were
Number: NCT00761813) that enrolled 1079 patients 50 years living independently versus institutionalized at 10 weeks and
of age or older with a low-energy femoral neck fracture with at 6 and 12 months after fracture by age groups 50–80 years
any degree of displacement requiring fracture fixation from old and greater than 80 years old. We categorized patients by
81 clinical sites in the United States, Canada, Australia, the these 2 age groups in our analysis because in the FAITH trial,
Netherlands, Norway, Germany, the United Kingdom, and minimization was used to ensure balance between interven-
India. Using central randomization, patients were allocated to tion groups for several patient factors, including age (50–80
receive surgical treatment with cancellous screws or a sliding years old or greater than 80 years old) during patient
hip screw. Patients were eligible if they met all the following allocation.3
inclusion criteria: (1) men or women 50 years of age and
older (with no upper age limit); (2) fracture of the femoral
Changes in Walking Aids
neck confirmed with anteroposterior and lateral hip radio-
We conducted a secondary analysis of the FAITH data
graphs, computed tomography, or magnetic resonance imag-
to descriptively quantify changes in the use of walking aids in
ing; (3) operative treatment of displaced fractures within 4
patients who were walking independently before hip fracture
days of presenting to the emergency department, or operative
over the 12 months after their femoral neck fracture by age
treatment of undisplaced fractures within 7 days of presenting
groups 50–80 years old and greater than 80 years old. In
to the emergency department; (4) patients were ambulatory
addition, we descriptively outlined the types of walking aids
before fracture, although they may have used an aid such as
used over the 12-month follow-up by age groups 50–80 years
a cane or a walker; (5) anticipated medical optimization for
old and greater than 80 years old.
operative fixation of the hip; (6) provision of informed
consent by the patient or a legally appointed representative;
(7) no other major trauma (defined as an injury severity score Determining Factors Associated With Living
.16); and (8) low-energy fracture, in the judgment of the Independently Within 12 Months After Hip
attending surgeon.7 Patients meeting any of the following Fracture
criteria were excluded: (1) patients not suitable for internal We used a multivariable Cox regression analysis to
fixation (ie, severe osteoarthritis, rheumatoid arthritis, or path- determine factors associated with time to living independently
ologic fracture); (2) associated major injuries of the lower within 12 months after fracture, with living independently
extremity (ie, ipsilateral or contralateral fractures of the foot, within 12 months as the dependent variable. Although the
Schemitsch et al J Orthop Trauma Volume 33, Number 6, June 2019
total follow-up period for the FAITH trial was 24 months aids over 12 months were reported as frequency counts and
after fracture, we selected 12 months for this analysis because percentages. For each of the aforementioned Cox regression
participants’ living and ambulation statuses did not change models, factors were included as independent variables.
greatly between the 12-month and 24-month follow-up visits, Before entering the potential factors into the multivariable
and more data were available at 12 months. The coauthors models, we looked at all pairwise correlations between
identified 15 factors (independent variables) a priori from the them. For continuous variables, we looked at Pearson
baseline, fracture characteristics, and surgical data collected correlation coefficient, and for the binary and categorical
as part of the FAITH trial.6 For each selected factor, we pro- variables, we looked at Cramer’s V statistics. For the binary
vided a rationale and proposed a hypothesized effect. We had and the continuous variables, we looked at point-biserial
planned to include quality of reduction within the model; correlation. If 2 variables were highly correlated (.0.7),
however, only 5 participants had unacceptable quality of we included only 1 of them in the multivariable models.
reduction. Therefore, this factor was not included in this All multivariable models also included the randomized
model. Patients who were institutionalized before their hip treatment. All tests were 2-tailed with alpha = 0.05. Results
fracture were excluded from this analysis. were reported as adjusted hazard ratios (HR), 95% confi-
dence intervals (CIs), and associated P values. We per-
Determining Factors Associated With Walking formed all analyses using SAS software (version 9.4: SAS
Without an Aid Within 12 Months After Hip Institute, Cary, NC).
Fracture
We used a multivariable Cox regression analysis to
RESULTS
determine factors associated with time to walking indepen-
dently within 12 months after fracture, with walking inde- Changes in Living Status
pendently within 12 months as the dependent variable. Seven hundred forty-seven FAITH participants between
Although the total follow-up period for the FAITH trial was the age of 50 and 80 years were living independently before
24 months after fracture, we selected 12 months for this their hip fracture. By the 12-month follow-up visit, data were
analysis because this is the time point that functional recovery available for 619 of the 747 patients; 594 of these 619 patients
is expected to return. The coauthors identified 15 factors were living independently (594/619 = 95.96%), 19 were insti-
(independent variables) a priori from the baseline, fracture tutionalized (eg, rehabilitation facilities) (19/619 = 3.07%), and
characteristics, and surgical data collected as part of the 36 had died (36/747 = 4.82%) (Table 1).
FAITH trial.6 For each selected factor, we provided a rationale Two hundred seventy-one patients older than 80 years
and proposed a hypothesized effect. As mentioned above, we were living independently before their fracture. At the 12-
had intended to include quality of reduction within the model; month follow-up visit, data were available for 207 of the 271
however, very few patients had unacceptable quality of reduc- patients; 156 of these 207 patients were living independently
tion. Therefore, this factor was not included in this model. (156/207 = 75.36%), 41 were institutionalized (41/207 =
Patients who were using walking aids before their hip fracture 19.81%), and 44 had died (44/271 = 16.24%) (Table 1).
were excluded from this analysis.
Changes in Use of Walking Aids
Data Analysis Six hundred seventy-eight patients between the age of
All FAITH patients with available data regarding their 50–80 years could ambulate independently before their frac-
living status and use of walking aids were included in the ture. At the 12-month follow-up visit, data were available for
analyses for changes in living status and walking aids. 580 of the 678 patients, and 195 of these 580 patients required
Results on changes in living status and changes in walking a walking aid (195/580 = 33.62%). A worsening of
TABLE 1. Changes in Living Status for Patients

Living Location Before Hip Fracture, n (%) 10 Weeks, n (%) 6 Months, n (%) 12 Months, n (%) P
Patients between the age of 50 and 80 N = 747 N = 690* N = 628* N = 619*
years living independently at
baseline
Not living independently N/A 56 (8.12) 21 (3.34) 19 (3.07) Overall ,0.0001
Living independently 747 (100) 620 (89.86) 595 (94.75) 594 (95.96)
Cumulative deaths N/A 14 (1.87) 26 (3.48) 36 (4.82)
Patients older than 80 years living N = 271 N = 256* N = 215* N = 207*
independently at baseline
Not living independently N/A 69 (26.95) 44 (20.47) 41 (19.81) Overall = 0.32
Living independently 271 (100) 174 (67.97) 162 (75.35) 156 (75.36)
*N decreased due to missing data.
TABLE 2. Changes in the Use of Aids for Patients

Use of Aids Before Hip Fracture, n (%) 10 Weeks, n (%) 6 Months, n (%) 12 Months, n (%) P
Patients between the age of 50 and 80 N = 678 N = 623* N = 575* N = 580*
years who were not using walking
aids at baseline
Using walking aids N/A 455 (73.03) 244 (42.43) 195 (33.62) Overall ,0.0001
Not using walking aids 678 (100) 164 (26.32) 323 (56.17) 380 (65.52)
Worsening of ambulatory status N/A 23 (3.73), n = 616 30 (5.32), n = 564 29 (5.08), n = 571
Patients older than 80 years who were N = 170 N = 160* N = 138* N = 137*
not using walking aids at baseline
Using walking aids N/A 140 (87.50) 105 (76.09) 95 (69.34) Overall = 0.002
Not using walking aids 170 (100) 16 (10.00) 29 (21.01) 39 (28.47)
Worsening of ambulatory status N/A 15 (9.80), n = 153 9 (6.72), n = 134 11 (8.33), n = 132
*N decreases due to missing data.
ambulatory status was also observed in 29 patients (29/571 = follow-up visit (Table 2). Table 3 quantifies these changes in
5.08%) at the 12-month follow-up visit (Table 2). Table 3 ambulatory aids over time.
quantifies these changes over time.
One hundred seventy patients older than 80 years were Factors Associated With Living Independently
walking independently before their fracture. At the 12-month Within 12 Months of a Hip Fracture
follow-up visit, data were available for 137 of the 170 Seven hundred eighty-six of 1079 FAITH participants
patients. Of these 137 patients, 95 required a walking aid (mean age: 72.81 years; 64.12% women) were included in
(95/137 = 69.34%). A worsening of ambulatory status was this analysis. Sixty-one FAITH participants were excluded
observed in 11 patients (11/132 = 8.33%) at the 12-month from this analysis because they were not living independently
TABLE 3. Breakdown of Aids Used by Nonindependent Ambulators

Breakdown of Aids Used 10 Weeks, n (%) 6 Months, n (%) 12 Months, n (%) P
Overall N = 182 N = 142* N = 124*
Bedridden 4 (2.2) 1 (0.7) 1 (0.8) Overall = 0.07
Wheelchair 45 (24.7) 38 (26.8) 27 (21.8)
Walker 119 (65.4) 84 (59.2) 72 (58.1)
2 crutches 0 (0.0) 1 (0.7) 0 (0.0)
1 crutch 1 (0.5) 0 (0.0) 0 (0.0)
Cane 13 (7.1) 18 (12.7) 23 (18.5)
Other 0 (0.0) 0 (0.0) 1 (0.8)
50–80 years old N = 77 N = 59* N = 52*
Wheelchair 18 (23.4) 14 (23.7) 10 (19.2)
Walker 48 (62.3) 35 (59.3) 28 (53.8)
2 crutches 0 (0.0) 1 (1.7) 0 (0.0)
1 crutch 1 (1.3) 0 (0.0) 0 (0.0)
Cane 8 (10.4) 9 (15.3) 13 (25.0)
Other 0 (0.0) 0 (0.0) 0 (0.0)
.80 years old N = 105 N = 83* N = 72*
Wheelchair 27 (25.7) 24 (28.9) 17 (23.6)
Walker 71 (67.6) 49 (59.0) 44 (61.1)
2 crutches 0 (0.0) 0 (0.0) 0 (0.0)
1 crutch 0 (0.0) 0 (0.0) 0 (0.0)
Cane 5 (4.8) 9 (10.8) 10 (13.9)
Other 0 (0.0) 0 (0.0) 1 (1.4)
*N decreases due to missing data.
before their fracture, and an additional 232 participants were

TABLE 4. Factors Associated With Living Independently
excluded from this analysis due to missing data. Being
Within 12 Months After Fracture (n = 786)
between the age of 50–80 years compared with an older
age (HR 1.70; 95% CI, 1.39–2.08; P , 0.0001), having Variable, n = 786 HR (95% CI) P
a class I American Society of Anesthesiologists (ASA) clas- Age
sification compared with a higher classification (class II, III, 50–80 years vs. .80 years 1.70 (1.39–2.08) ,0.0001
or IV) (P , 0.0001), not using a walking aid before fracture ASA classification Overall: ,0.0001
(HR 1.85; 95% CI, 1.43–2.44; P , 0.0001), being a non- Class I vs. class II 1.52 (1.18–1.96) 0.001
smoker or former smoker (HR 1.59; 95% CI, 1.25–2.00; P = Class I vs. class III 1.72 (1.27–2.38) 0.001
0.0001), and having an acceptable quality of implant place- Class I vs. class IV 3.33 (1.96–5.56) ,0.0001
ment (HR 1.77; 95% CI, 1.19–2.64; P = 0.01) were associ- Prefracture functional status
ated with a higher chance of living independently 12 months Independent ambulator vs. using 1.85 (1.43–2.44) ,0.0001
after fracture (Table 4). Gender, body mass index, diabetes, ambulation aid
fracture displacement, level of fracture line, Pauwels classifi- Smoking status
cation, type of reduction, treatment group, and the absence of Other vs. current 1.59 (1.25–2.00) 0.0001
revision surgery were not associated with living indepen- Quality of implant placement
dently 12 months after fracture (P . 0.05). Acceptable vs. unacceptable 1.77 (1.19–2.64) 0.01
To note, of the 786 participants included in this analysis Gender
of determining factors associated with living independently Male vs. female 1.01 (0.84–1.20) 0.93
within 12 months of a hip fracture, 119 had a class I ASA Body mass index (change in 5 1.07 (0.98–1.17) 0.13
points)
classification. The majority of these participants were
Diabetes
between the age of 50 and 80 years (96.7%) and women
Yes vs. no 1.01 (0.81–1.25) 0.94
(59.2%) with an undisplaced fracture (59.2%). 99.2% of these
Fracture displacement
participants were living independently before their fracture,
Displaced vs. undisplaced 0.97 (0.78–1.2) 0.77
and 98.3% were independent ambulators before their fracture.
Level of the fracture line Overall: 0.19
By the 12-month follow-up visit, of the 106 participants who
Midcervical vs. subcapital 1.16 (0.95–1.42) 0.14
attended their appointment, the majority were full weight-
Basal vs. subcapital 0.90 (0.58–1.4) 0.63
bearing (98.1%), living at home (97.2%), and not using any
Pauwels classification Overall: 0.97
walking aids (85.8%).
Type I vs. type III 1.05 (0.71–1.54) 0.82
Type II vs. type III 1.02 (0.77–1.36) 0.88
Factors Associated With Walking Type of reduction Overall: 0.19
Independently (Not Using a Walking Aid) None vs. open 0.76 (0.49–1.16) 0.20
Closed vs. open 0.68 (0.44–1.05) 0.08
Within 12 Months of a Hip Fracture
Treatment group
Six hundred eighty-eight of 1079 FAITH participants
Sliding hip screws vs. 1.03 (0.88–1.20) 0.76
(mean age: 72.82 years; 63.08% women) were included in cancellous screws
this analysis. Two hundred thirty-one FAITH participants Time from injury to surgery 1.02 (0.99–1.05) 0.26
were excluded because they were using a walking aid before Revision surgery
their fracture, and an additional 160 FAITH participants were Yes vs. no 0.92 (0.75–1.14) 0.46
excluded from this analysis as a result of missing data. Being
between the age of 50–80 years (compared with an older age)
(HR 3.27; 95% CI, 2.10–5.10; P , 0.0001), having a class I between the age of 50 and 80 years (96.7%) and women
ASA classification (compared with class II, III, or IV) (P , (58.5%) with an undisplaced fracture (59.3%). 96.7% of these
0.0001), living independently (HR 14.29; 95% CI, 1.79– participants were living independently before their fracture,
100.00; P = 0.01), being a nonsmoker or former smoker and 98.4% were independent ambulators before their fracture.
(HR 1.59; 95% CI, 1.19–2.13; P = 0.002), having an undis- By the 12-month follow-up visit, of the 108 participants who
placed fracture (HR 1.39; 95% CI, 1.02–1.89; P = 0.04), and attended their appointment, the majority were full weight-
not requiring a revision surgery (HR 1.79; 95% CI, 1.35– bearing (98.1%), living at home (97.2%), and not using any
2.38; P , 0.0001) were associated with a higher chance of walking aids (84.3%).
walking without an aid within 12 months after fracture (Table
5). Quality of implant placement, gender, body mass index,
diabetes, level of fracture line, Pauwels classification, type of
reduction, treatment group, and time from injury to surgery DISCUSSION
were not associated with walking independently 12 months Using data from the FAITH trial, we quantified the
after fracture (P . 0.05). changes in living status in patients who were living indepen-
To note, of the 688 participants included in this analysis dently before hip fracture and changes in the use of walking
of determining factors associated with walking independently aids in patients who were walking independently before hip
within 12 months of hip fracture, 122 had a class I ASA fracture over the 12 months after their femoral neck fracture.
classification. The majority of these participants were Overall, we found that the majority of patients between the
in our analyses. Our results were found to be fairly consistent

TABLE 5. Factors Associated With Walking Independently
with the current published literature, which followed a similar
(Not Using a Walking Aid) Within 12 Months After Fracture
(n = 688) trend.7,9–13 For consistency with the a priori minimization
factors in the FAITH trial, we categorized patients by 50–
Variable, n = 688 HR (95% CI) P
80 years and .80 years age groups in our analysis.7
Age Being a nonsmoker or former smoker and lower ASA
50–80 years vs. .80 years 3.27 (2.10–5.10) ,0.0001 classification (class I vs. classes II, III, and IV) were other factors
ASA classification Overall:,0.0001 found in our analyses to be associated with both living and
Class I vs. class II 1.43 (1.04–2.00) 0.03 walking independently 12 months after fracture. Semel et al12
Class I vs. class III 3.33 (2.13–5.26) ,0.0001 also examined smoking status in their study but found that this
Class I vs. class IV 33.33 (4.17–333.33) 0.001 variable was not associated with living status at hospital dis-
Prefracture living setting charge (P = 0.77). The general trend seen when examining
Not institutionalized vs. 14.29 (1.79–100.00) 0.01 ASA classification as a predictive factor for outcomes related
institutionalized
to a hip fracture was similar to our analyses, where the higher the
Smoking status
ASA score, the greater the risk of a negative outcome.14,15
Other vs. current 1.59 (1.19–2.13) 0.002
A large and diverse sample of over 1000 femoral neck
Fracture displacement
fracture patients from sites in the United States, Canada,
Undisplaced vs. displaced 1.39 (1.02–1.89) 0.04
Australia, the Netherlands, Norway, Germany, the United
Reoperation
Kingdom, and India were enrolled in the FAITH trial, which
No vs. yes 1.79 (1.35–2.38) ,0.0001
enhances the external validity and generalizability of our
Quality of implant placement
findings. Completing our analysis on the changes in living
Acceptable vs. unacceptable 2.10 (0.99–4.48) 0.05
status and the use of walking aids by age groups 50–80 years
Gender 0.55
old and greater than 80 years old allowed us to find important
Male vs. female 0.92 (0.71–1.20)
differences between the 2 groups, which had not been eval-
Body mass index (change in 0.97 (0.85–1.11) 0.71
5 points)
uated previously in the current literature. In addition, there
Diabetes
were numerous follow-up visits during the 12-month follow-
Yes vs. no 1.14 (0.77–1.69) 0.52
up period, which allowed for a frequent and ongoing assess-
Level of the fracture line Overall: 0.71
ment of the use of walking aids and living status after surgery.
Midcervical vs. subcapital 1.03 (0.76–1.39) 0.85
For our second analysis, the use of a multivariable Cox
Basal vs. subcapital 0.77 (0.36–1.64) 0.50
regression assisted us in controlling for several potential con-
Pauwels classification Overall: 0.96
founders that could have skewed the interpretation of our
Type I vs. type III 0.94 (0.53–1.66) 0.83
results. The major limitation we experienced was not being
Type II vs. type III 1.01 (0.71–1.43) 0.97
able to include all 1079 FAITH trial patients in this particular
Type of reduction Overall: 0.83
analysis as a result of missing data. In addition, only factors
None vs. open 1.17 (0.61–2.23) 0.64
that were gathered as part of the FAITH trial could be used in
Closed vs. open 1.07 (0.56–2.02) 0.84
our second analysis.
Treatment group
Identifying factors associated with living and walking
Sliding hip screws vs. 0.94 (0.74–1.19) 0.60
independently after a hip fracture may help the orthopaedic
cancellous screws community better identify which patients are at risk of loss of
Time from injury to surgery 1.01 (0.95–1.09) 0.70 independence and mobility after a hip fracture and, ulti-
mately, help to optimize the care of patients with this type of
injury. Clinicians will be able to inform hip fracture patients
and their families of the risks of loss of independence and
age of 50–80 years were able to regain their mobility and mobility when explaining treatment options and prognosis. In
independence by 12 months after their injury. For those older addition, the variables identified in our analysis may allow for
than 80 years, most patients were also able to regain their surgeons to consider different care options, such as joint
independence by 12 months after injury, but not their replacement, for those who are at an increased risk of loss of
mobility. independence. Finally, our findings may also inform future
Our analysis on the changes in living status and the use research by identifying high-risk patients who may benefit
of walking aids is consistent with the limited literature from new interventions and adjuncts to care.
evaluating the living status and walking ability in older hip
fracture patients. For example, a Swedish study that assessed REFERENCES
the rehabilitation of 3053 consecutive patients with a femoral 1. Johnell O, Kanis JA. An estimate of the worldwide prevalence and
neck fracture found a similar trend in living status after disability associated with osteoporotic fractures. Osteoporos Int. 2006;
surgical treatment for those patients who were living 17:1726–1733.
independently before fracture.8 2. Florschutz AV, Langford JR, Haidukewych GJ, et al. Femoral neck
fractures. J Orthop Trauma. 2015;29:121–129.
Younger age (50–80 years old vs. .80 years old) was 3. Salpakoski A, Törmäkangas T, Edgren J, et al. Walking recovery after
found to be the factor most strongly associated with both a hip fracture: a prospective follow-up study among community-dwelling
living and walking independently 12 months after fracture over 60-year-old men and women. Biomed Res Int. 2014;2014:289549.
4. Magaziner J, Hawkes W, Hebel JR, et al. Recovery from hip fracture in 9. Hagino T, Sato E, Tonotsuka H, et al. Prediction of ambulation prognosis
eight areas of function. J Gerontol A Biol Sci Med Sci. 2000;55:M498– in the elderly after hip fracture. Int Orthop. 2006;30:315–319.
M507. 10. Michel J-P, Hoffmeyer P, Klopfenstein C, et al. Prognosis of functional
5. Visser M, Harris TB, Fox KM, et al. Change in muscle mass and muscle recovery 1 year after hip fracture: typical patient profiles through cluster
strength after a hip fracture: relationship to mobility recovery. J Gerontol analysis. J Gerontol. 2000;55:508–515.
A Biol Sci Med Sci. 2000;55:M434–M440. 11. Ingemarsson AH, Frändin K, Mellström D, et al. Walking ability and
6. Fixation Using Alternative Implants for the Treatment of Hip fractures activity level after hip fracture in the elderly—a follow-up. J Rehabil
(FAITH) Investigators. Fracture fixation in the operative management of Med. 2003;35:76–83.
12. Semel J, Gray J, Ahn HJ, et al. Predictors of outcome following hip
hip fractures (FAITH): an international, multicentre, randomised con-
fracture rehabilitation. PM R. 2010;2:799–805.
trolled trial. Lancet. 2017;389:1519–1527. 13. Lee D, Jo JY, Jung JS, et al. Prognostic factors predicting early recovery
7. FAITH Investigators. Fixation using alternative implants for the treat- of pre-fracture functional mobility in elderly patients with hip fracture.
ment of hip fractures (FAITH): design and rationale for a multi-centre Ann Rehabil Med. 2014;38:827–835.
randomized trial comparing sliding hip screws and cancellous screws on 14. Kastanis G, Topalidou A, Alpantaki K, et al. Is the ASA score in geriatric
revision surgery rates and quality of life in the treatment of femoral neck hip fractures a predictive factor for complications and readmission? Sci-
fractures. BMC Musculoskelet Disord. 2014;15:219. entifica (Cairo). 2016;2016:1–6.
8. Holmberg S, Thorngren KG. Rehabilitation after femoral neck fracture. 15. Yeoh CJC, Fazal MA. ASA grade and elderly patients with femoral neck
3053 patients followed for 6 years. Acta Orthop Scand. 1985;56:305–308. fracture. Geriatr Orthop Surg Rehabil. 2014;5:195–199.
APPENDIX 1. FAITH Investigators

Steering Committee Participating Clinical Sites
M. Bhandari (Chair, McMaster University), M. Swiont- Canada: Robert McCormack, Kelly Apostle, Dory
kowski (University of Minnesota), P. J. Devereaux (McMas- Boyer, Farhad Moola, Bertrand Perey, Trevor Stone, Darius
ter University), Gordon Guyatt (McMaster University), M. J. Viskontas, H. Michael Lemke, Mauri Zomar, Karyn Moon,
Heetveld (Spaarne Gasthuis, Haarlem), Kyle Jeray (Green- Raely Moon, and Amber Oatt (Royal Columbian Hospital/
ville Health System), Susan Liew (The Alfred), Martin Fraser Health Authority/University of British Columbia);
Richardson (University of Melbourne), E. H. Schemitsch Richard E. Buckley, Paul Duffy, Robert Korley, Shannon
(University of Western Ontario), Lehana Thabane (McMaster Puloski, James Powell, Kelly Johnston, Kimberly Carcary,
University), Paul Tornetta III (Boston University Medical Melissa Lorenzo, and Ross McKercher (Foothills Medical
Center), and Stephen D. Walter (McMaster University). Centre); David Sanders, Mark MacLeod, Abdel-Rahman
Lawendy, and Christina Tieszer (London Health Sciences
Global Methods Center
Centre); David Stephen, Hans Kreder, Richard Jenkinson,
M. Bhandari (Principal Investigator); S. Sprague
Markku Nousiainen, Terry Axelrod, John Murnaghan, Diane
(Research Methodologist); Paula McKay (Manager); Taryn
Nam, Veronica Wadey, Albert Yee, Katrine Milner, Monica
Scott, Alisha Garibaldi, Helena Viveiros, and Marilyn
Kunz, and Wesley Ghent (Sunnybrook Health Sciences
Swinton (Research Coordination); Mark Gichuru and S.
Centre); E. H. Schemitsch, Michael D. McKee, Jeremy A.
Bzovsky (Adjudication Coordination); D. Heels-Ansdell and
Hall, Aaron Nauth, Henry Ahn, Daniel B. Whelan, Milena R.
Q. Zhou (Statistical Analysis); Lisa Buckingham and Aravin
Vicente, Lisa M. Wild, Ryan M. Khan, and Jennifer T. Hidy
Duraikannan (Data Management); and Deborah Maddock
(St. Michael’s Hospital); Chad Coles, Ross Leighton, Michael
and Nicole Simunovic (Grants Management) (McMaster
Biddulph, David Johnston, Mark Glazebrook, David Alexan-
University).
der, Catherine Coady, Michael Dunbar, J. David Amirault,
US Method Center Michael Gross, William Oxner, Gerald Reardon, C. Glen
M. Swiontkowski (Principal Investigator) and Julie Richardson, J. Andrew Trenholm, Ivan Wong, Kelly Trask,
Agel (Research Coordination) (University of Minnesota). Shelley MacDonald, and Gwendolyn Dobbin (Queen Eliza-
beth II Health Sciences Centre); Ryan Bicknell, Jeff Yach,
The Netherlands Method Center Davide Bardana, Gavin Wood, Mark Harrison, David Yen,
M. J. Heetveld (Principal Investigator); Esther M. M. Sue Lambert, Fiona Howells, and Angela Ward (Human
Van Lieshout (Research Coordination); and Stephanie M. Mobility Research Centre, Queen’s University and Kingston
Zielinski (Trial Coordination) (Erasmus MC, University General Hospital); Paul Zalzal, Heather Brien, V. Naumetz,
Medical Center Rotterdam). Brad Weening, and Nicole Simunovic (Oakville Trafalgar
UK Method Center Memorial Hospital); Eugene K. Wai, Steve Papp, Wade T.
Amar Rangan (Principal Investigator); Birgit C. Hanusch, Gofton, Allen Liew, Stephen P. Kingwell, Garth Johnson,
Lucksy Kottam, and Rachel Clarkson (Research Coordination) Joseph O’Neil, Darren M. Roffey, and Vivian Borsella (The
(The James Cook University Hospital). Ottawa Hospital); and Victoria Avram (Juravinski Hospital
and Cancer Centre).
Adjudication Committee The United States: Todd M. Oliver, Vicki Jones, and
Gregory J. Della Rocca (Chair) (Duke University), Michelle Vogt (Boone Hospital Center—Columbia Orthopae-
Robert Haverlag (Onze Lieve Vrouwe Gasthuis), Susan dic Group); Clifford B. Jones, James R. Ringler, Terrence J.
Liew (The Alfred), Gerard P. Slobogean (University of Endres, Debra L. Sietsema, and Jane E. Walker (Orthopaedic
Maryland, Baltimore), and Kyle Jeray (Greenville Health Associates of Michigan); Kyle J. Jeray, J. Scott Broderick,
System). David R. Goetz, Thomas B. Pace, Thomas M. Schaller,
Scott E. Porter, Michael L. Beckish, John D. Adams, Benja- and Roger B. Huff (Kaiser Permanente); Joseph Baele, Tim-
min B. Barden, Aaron T. Creek, Stephen H. Finley, Jonathan othy Weber, and Matt Edison (OrthoIndy Trauma St. Vincent
L. Foret, Garland K. Gudger Jr, Richard W. Gurich Jr, Austin Trauma Center); Jessica Cooper McBeth (Santa Clara Valley
D. Hill, Steven M. Hollenbeck, Lyle T. Jackson, Kevin K. Medical Center); Karl Shively, Janos P. Ertl, Brian Mullis, J.
Kruse III, Wesley G. Lackey, Justin W. Langan, Julia Lee, Andrew Parr, Ripley Worman, Valda Frizzell, and Molly M.
Lauren C. Leffler, Timothy J. Miller, R. Lee Murphy, Jr, Moore (Indiana University—Eskenazi Health Services);
Lawrence K. O’Malley II, Melissa E. Peters, Dustin M. Price, Charles J. DePaolo, Rachel Alosky, Leslie E. Shell, Lynne
John A. Tanksley, Jr, Erick T. Torres, Dylan J. Watson, Scott Hampton, Stephanie Shepard, Tracy Nanney, and Claudine
T. Watson, Stephanie L. Tanner, Rebecca G. Snider, Lauren Cuento (Mission Hospital Research Institute); Robert V. Can-
A. Nastoff, Shea A. Bielby, and Robert J. Teasdall (Green- tu, Eric R. Henderson, and Linda S. Eickhoff (Dartmouth-
ville Health System); Julie A. Switzer, Peter A. Cole, Sarah Hitchcock Medical Center); E. Mark Hammerberg, Philip
A. Anderson, Paul M. Lafferty, Mengnai Li, Thuan V. Ly, Stahel, David Hak, Cyril Mauffrey, Corey Henderson, Han-
Scott B. Marston, Amy L. Foley, Sandy Vang, and David M. nah Gissel, and Douglas Gibula (Denver Health and Hospital
Wright (Regions Hospital-University of Minnesota); Andrew Authority); David P. Zamorano, Martin C. Tynan, Deeba
J. Marcantonio, Michael S. H. Kain, Richard Iorio, Lawrence Pourmand, and Deanna Lawson (University of California Ir-
M. Specht, John F. Tilzey, Margaret J. Lobo, and John S. vine Medical Center); Gregory J. Della Rocca, Brett D. Crist,
Garfi (Lahey Hospital & Medical Center); Heather A. Vallier, Yvonne M. Murtha, and Linda K. Anderson (University of
Andrea Dolenc, and Mary Breslin (MetroHealth Medical Missouri Health Care); Colleen Linehan and Lindsey Pilling
Center); Michael J. Prayson, Richard Laughlin, L. Joseph (Covenant Healthcare of Saginaw); Courtland G. Lewis, Ste-
Rubino, Jedediah May, Geoffrey Ryan Rieser, Liz Dulaney- phanie Caminiti, Raymond J. Sullivan, and Elizabeth Roper
Cripe, and Chris Gayton (Miami Valley Hospital); James (Hartford Hospital); William Obremskey, Philip Kregor, Jus-
Shaer, Tyson Schrickel, and Barbara Hileman (St. Elizabeth tin E. Richards, and Kenya Stringfellow (Vanderbilt Univer-
Youngstown Hospital); John T. Gorczyca, Jonathan M. sity Medical Center); and Michael P. Dohm and Abby Zellar
Gross, Catherine A. Humphrey, Stephen Kates, John P. Ketz, (Western Slope Study Group).
Krista Noble, Allison W. McIntyre, and Kaili Pecorella (Uni- The Netherlands: Michiel J. M. Segers, Jacco A. C. Zijl,
versity of Rochester Medical Center); Craig A. Davis, Stuart Bart Verhoeven, Anke B. Smits, Jean Paul P. M. de Vries,
Weinerman, Peter Weingarten, Philip Stull, Stephen Linden- Bram Fioole, Henk van der Hoeven, Evert B. M. Theunissen,
baum, Michael Hewitt, John Schwappach, Janell K. Baker, Tammo S. de Vries Reilingh, Lonneke Govaert, Philippe
Tori Rutherford, Heike Newman, Shane Lieberman, Erin Wittich, Maurits de Brauw, Jan Wille, Peter M. N. Y. M. Go,
Finn, Kristin Robbins, Meghan Hurley, Lindsey Lyle, Khalis Ewan D. Ritchie, Ronald N. Wessel, and Eric R. Hammacher
Mitchell, Kieran Browner, Erica Whatley, Krystal Payton, (St. Antonius Ziekenhuis); M. J. Heetveld, Gijs A. Visser,
and Christina Reeves (Colorado Orthopedic Consultants); Li- Heyn Stockmann, Rob Silvis, Jaap P. Snellen, Bram Rij-
sa K. Cannada, David E. Karges, and Sarah A. Dawson (St. broek, Joris J. G. Scheepers, Erik G. J. Vermeulen, Michiel P.
Louis University Hospital); Samir Mehta, John Esterhai, Jai- C. Siroen, Ronald Vuylsteke, Hans L. F. Brom, and Herman
mo Ahn, Derek Donegan, Annamarie D. Horan, Patrick J. Rijna (Kennemer Gasthuis); Piet A. R. de Rijcke, Cees L.
Hesketh, Evan R. Bannister (University of Pennsylvania); Koppert, Steven E. Buijk, Richard P. R. Groenendijk, Imro
Jonathan P. Keeve, Christopher G. Anderson, Michael D. Dawson, Geert W. M. Tetteroo, Milko M. M. Bruijninckx,
McDonald, and Jodi M. Hoffman (Northwest Orthopaedic Pascal G. Doornebosch, and Eelco J. R. de Graaf (IJsselland
Specialists); Ivan Tarkin, Peter Siska, Gary Gruen, Andrew Ziekenhuis); Maarten van der Elst, Carmen C. van der Pol,
Evans, Dana J. Farrell, James Irrgang, and Arlene Luther Martijne van ’t Riet, Tom M. Karsten, Mark R. de Vries,
(University of Pittsburgh Medical Center); William W. Cross Laurents P. S. Stassen, Niels W. L. Schep, G. Ben Schmidt,
III, Joseph R. Cass, Stephen A. Sems, Michael E. Torchia, and W. H. Hoffman (Reinier de Graaf Gasthuis); Rudolf W.
Tyson Scrabeck (Mayo Clinic); Mark Jenkins, Jules Dumais, Poolman, Maarten P. Simons, Frank H. W. M. van der Heij-
and Amanda W. Romero (Texas Tech University Health Sci- den, W. Jaap Willems, Frank R. A. J. de Meulemeester, Cor
ences Center—Lubbock); Carlos A. Sagebien, Mark S. But- P. van der Hart, Kahn Turckan, Sebastiaan Festen, Frank de
ler, James T. Monica, and Patricia Seuffert (University Nies, Robert Haverlag, Nico J. M. Out, and Jan Bosma (Onze
Orthopaedic Associates, LLC); Joseph R. Hsu, Daniel Stin- Lieve Vrouwe Gasthuis); Albert van Kampen, Jan Biert, Arie
ner, James Ficke, Michael Charlton, Matthew Napierala, and B. van Vugt, Michael J. R. Edwards, Taco J. Blokhuis, Jan
Mary Fan (US Army Institute of Surgical Research); Paul Paul M. Frölke, Leo M. G. Geeraedts, Jean W. M. Gardeniers,
Tornetta III, Chadi Tannoury, Hope Carlisle, and Heather Edward C. T. H. Tan, Lodewijk M. S. J. Poelhekke, Maarten
Silva (Boston University Medical Center); Michael Archdea- C. de Waal Malefijt, and Bart Schreurs (University Medical
con, Ryan Finnan, Toan Le, John Wyrick, and Shelley Hess Center St. Radboud); Gert R Roukema, Hong A. Josaputra,
(UC Health/University of Cincinnati Medical Center); Paul Keller, Peter D. de Rooij, Hans Kuiken, Han Boxma,
Michael L. Brennan, Robert Probe, Evelyn Kile, Kelli Mills, Berry I. Cleffken, and Ronald Liem (Maasstad Ziekenhuis);
Lydia Clipper, Michelle Yu, and Katie Erwin (Scott and Steven J. Rhemrev, Coks H. R. Bosman, Alexander de Mol
White Memorial Hospital); Daniel Horwitz, Kent Strohecker, van Otterloo, Jochem Hoogendoorn, Alexander C. de Vries,
and Teresa K. Swenson (Geisinger Medical Center); Andrew and Sven A. G. Meylaerts (Medisch Centrum Haaglanden);
H. Schmidt and Jerald R. Westberg (Hennepin County Med- Michiel H. J. Verhofstad, Joost Meijer, Teun van Egmond,
ical Center); Kamran Aurang, Gary Zohman, Brett Peterson, Frank H. W. M. van der Heijden, and Igor van der Brand
(St. Elisabeth Ziekenhuis); Peter Patka, Martin G. Eversdijk, Farah Khan (Nirmal Hospital, India); Ateet Sharma, Amir San-
Rolf Peters, Dennis Den Hartog, Oscar J. F. Van Waes, and ghavi, and Mittal Trivedi (Satellite Orthopaedic Hospital and
Pim Oprel (Erasmus MC, University Medical Center Rotter- Research Centre, India); Anil Rai, Subash, and Kamal Rai
dam); Harm M. van der Vis, Martin Campo, Ronald Verha- (Highway Hospital, India); Vineet Yadav, Sanjay Singh, and
gen, G. H. Robert Albers, and Arthur W. Zurcher (Tergooi Kamal Rai (Popular Hospital, India); Kevin Tetsworth, Geoff
Ziekenhuizen); Rogier K. J. Simmermacher, Jeroen van Donald, Patrick Weinrauch, Paul Pincus, Steven Yang, Brett
Mulken, Karlijn van Wessem, Taco J. Blokhuis, Steven M. Halliday, Trevor Gervais, Michael Holt, and Annette Flynn
van Gaalen, and Luke P. H. Leenen (University Medical (Royal Brisbane and Women’s Hospital, Australia); Amal
Center Utrecht); Maarten W. G. A. Bronkhorst and Onno Shankar Prasad and Vimlesh Mishra (Madhuraj Nursing
R. Guicherit (Bronovo Ziekenhuis); J. Carel Goslings, Robert Home, India); D. C. Sundaresh and Angshuman Khanna (M.
Haverlag, and Kees Jan Ponsen (Academic Medical Center). S. Rammaiah Medical College & Hospital, India); Joe Joseph
International: Mahesh Bhatia, Vinod Arora, and Vivek Cherian, Davy J. Olakkengil, and Gaurav Sharma (St John’s
Tyagi (RLB Hospital and Research Center, India); Susan Liew, Medical College Hospital, India); Marinis Pirpiris, David Love,
Harvinder Bedi, Ashley Carr, Hamish Curry, Andrew Chia, Andrew Bucknill, and Richard J. Farrugia (Royal Melbourne
Steve Csongvay, Craig Donohue, Stephen Doig, Elton Ed- Hospital, Australia); Hans-Christoph Pape, Matthias Knobe,
wards, Greg Etherington, Max Esser, Andrew Gong, Arvind and Roman Pfeifer (University of Aachen Medical Center,
Jain, Doug Li, Russell Miller, Ash Moaveni, Matthias Russ, Lu Germany); Peter Hull, Sophie Lewis, and Simone Evans (Cam-
Ton, Otis Wang, Adam Dowrick, Zoe Murdoch, and Claire bridge University Hospitals, England); Rajesh Nanda, Rajani-
Sage (The Alfred, Australia); Frede Frihagen, John Clarke- kanth Logishetty, Sanjeev Anand, and Carol Bowler
Jenssen, Geir Hjorthaug, Torben Ianssen, Asgeir Amundsen, (University Hospital of North Tees, England); Akhil Dadi,
Jan Egil Brattgjerd, Tor Borch, Berthe Bøe, Bernhard Flatøy, Naveen Palla, and Utsav Ganguly (Sunshine Hospital, India);
Sondre Hasselund, Knut Jørgen Haug, Kim Hemlock, Tor B. Sachidananda Rai and Janakiraman Rajakumar (Unity
Magne Hoseth, Geir Jomaas, Thomas Kibsgård, Tarjei Lona, Health Complex, India); Andrew Jennings, Graham Chuter,
Gilbert Moatshe, Oliver Müller, Marius Molund, Tor Nicolais- Glynis Rose, and Gillian Horner (University Hospital of North
en, Fredrik Nilsen, Jonas Rydinge, Morten Smedsrud, Are Durham and Darlington Memorial Hospital, England); Callum
Stødle, Axel Trommer, Stein Ugland, Anders Karlsten, Guri Clark and Kate Eke (Wexham Park Hospital, England); Mike
Ekås, Elise Berg Vesterhus, and Anne Christine Brekke (Oslo Reed, Dominic Inman, Chris Herriott, and Christine Dobb
University Hospital, Norway); Ajay Gupta, Neeraj Jain, and (Northumbria Healthcare NHS Foundation Trust, England).
ORIGINAL ARTICLE
Inter-Rater Reliability of the Modified Radiographic Union

Score for Diaphyseal Tibial Fractures With Bone Defects
Stuart L. Mitchell, MD,*† William T. Obremskey, MD, MPH, MMHC,‡ Jason Luly, MS,†
Michael J. Bosse, MD,§ Katherine P. Frey, PhD, MPH, RN,† Joseph R. Hsu, MD,§
Ellen J. MacKenzie, PhD,† Saam Morshed, MD, PhD,║ Robert V. O’Toole, MD,¶
Daniel O. Scharfstein, ScD,** Paul Tornetta III, MD†† and the Major Extremity Trauma Rehabilitation
Consortium (METRC)
Main Outcome Measurements: Inter-rater reliability was

Objectives: To evaluate inter-rater reliability of the modified Radio- assessed using Krippendorff’s alpha (KA) statistic; intraclass corre-
graphic Union Score for Tibial (mRUST) fractures among patients with lation coefficient (ICC) is presented for comparison with previous
open, diaphyseal tibia fractures with a bone defect treated with publications.
intramedullary nails (IMNs), plates, or definitive external fixation (ex-fix).
Results: Two hundred thirteen patients met inclusion criteria
Design: Retrospective cohort study. including 115 IMNs, 24 plates, 29 ex-fixes, and 45 cases that no
Setting: Fifteen-level one civilian trauma centers; 2 military longer had instrumentation at evaluation. All reviewers agreed on the
treatment facilities. pattern of scoreable cortices for 90.4% of IMNs, 88.9% of those
without instrumentation, 44.8% of rings, and 20.8% of plates.
Patients/Participants: Patients $18 years old with open, diaph- Thirty-one (15%) cases, primarily plates and ex-fixes, did not
yseal tibia fractures with a bone defect $1 cm surgically treated contribute to KA and ICC estimates because ,2 raters scored all
between 2007 and 2012. cortices. The overall KA for the 85% that could be analyzed was
0.64 (ICC 0.71). For IMNs, plates, ex-fixes, and no instrumentation,
Intervention: Three of 6 orthopedic traumatologists reviewed and KA (ICC) was 0.65 (0.75), 0.88 (0.90), 0.47 (0.62), and 0.48 (0.57),
applied mRUST scoring criteria to radiographs from the last clinical respectively.
visit within 13 months of injury.
Conclusions: In tibia fractures with bone defects, the mRUST
seems similarly reliable to previous work in patients treated with
Accepted for publication December 29, 2018. IMN but is less reliable in those with plates or ex-fixes, or after
From the *Department of Orthopaedic Surgery, Johns Hopkins University removal of instrumentation.
School of Medicine, Baltimore, MD; †Department of Health Policy and
Management, Johns Hopkins Bloomberg School of Public Health, Balti- Key Words: radiographic union, bone defect, mRUST score, inter-
more, MD; ‡Department of Orthopaedic Surgery, Vanderbilt University,
Nashville, TN; §Department of Orthopaedic Surgery, Carolinas Medical
rater reliability, tibia fracture, fracture healing
Center, Charlotte, NC; ║Department of Orthopaedic Surgery, University of (J Orthop Trauma 2019;33:301–307)
California San Francisco, San Francisco, CA; ¶Department of Orthopaedic
Surgery, University of Maryland School of Medicine, Baltimore, MD;
**Department of Biostatistics, Johns Hopkins Bloomberg School of Public INTRODUCTION
Health, Baltimore, MD; and ††Department of Orthopaedic Surgery, Boston Limited consensus exists regarding the exact definition
University, Boston, MA.
Grant supported by Department of Defense USAMRMC, Contract Number: of a “healed fracture.”1–3 Radiographic evaluation of union is
W81XWH-09-20108, and by T32 AR067708 (S.M.) from the National used to augment observed clinical outcomes such as presence
Institutes of Health. of pain with bearing weight, or tenderness with palpation of
P. Tornetta receives Smith-Nephew and Wolters-Kluwer royalties. The the fracture site. In an attempt to make this assessment more
remaining authors report no conflict of interest.
Presented as a poster at the Annual Meeting of the Orthopaedic Trauma objective, Whelan et al4 developed a scoring system based
Association, October 2016, National Harbor, MD, as a poster at the on the appearance of each of the cortices on the anteropos-
Annual Meeting of MHSRS 2016, and as a podium presentation at American terior (AP) and lateral radiographic views of tibial diaphy-
Academy of Orthopaedic Surgery Annual Meeting 2018. seal fractures treated with intramedullary nails (IMN), which
Supplemental digital content is available for this article. Direct URL citations they referred to as the Radiographic Union Score for Tibial
appear in the printed text and are provided in the HTML and PDF
versions of this article on the journal’s Web site (www.jorthotrauma. (RUST) fractures. Two orthopedic residents, 2 community
com). orthopedic surgeons, and 3 orthopedic traumatologists rated
METRC Corporate Authors are listed in Appendix 1. 45 sets of radiographs, and they reported an intraclass cor-
Reprints: Paul Tornetta, MD III, Department of Orthopaedic Surgery, Boston relation coefficient (ICC), a measure of inter-rater reliability,
Medical Center, Boston University, 850 Harrison Ave, Dowling 2 North,
Boston, MA 02118 (e-mail: ptornetta@gmail.com).
of 0.86. To expand utilization of this scoring system, Litren-
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. ta et al5 analyzed its reliability and applicability to patients
DOI: 10.1097/BOT.0000000000001445 with metadiaphyseal fractures. This application included an
Mitchell et al J Orthop Trauma Volume 33, Number 6, June 2019
important change to the range of the score where the inves- the time elapsed between injury and radiographs. Raters were
tigators subdivided the category “no fracture line, bridging instructed to score each case using mRUST criteria. Each of
callus” into 2 distinct categories: “callus present” and the cortices (4 total) on the AP and lateral radiographs were
“bridging callus.” The resulting modified RUST (mRUST) graded as: 1 = no callus; 2 = callus present without bridging;
has a maximum score of 16 compared with the maximum 3 = bridging callus, but visible fracture line; and 4 = fracture
standard RUST score of 12. Using 27 sets of radiographs of line not visible (remodeled). The mRUST score is the sum of
metadiaphyseal distal femur fractures rated by 12 orthopedic the 4 cortical scores (range, 4–16). Each cortex was scored
trauma surgeons, the authors found slightly higher inter-rater individually. In some cases, assigned reviewers did not score
reliability for the mRUST than the standard RUST (ICC a cortex because they felt they could not adequately assess the
0.68 vs. 0.63). This finding held for the subgroup of patients injury due to problems such as an obstructed view of the
treated with nails (mRUST ICC 0.74; RUST ICC 0.67) and fracture site. At least 2 of the 3 assigned raters for each case
for the subgroup treated with plates (mRUST ICC 0.59; must have scored all 4 cortices in order for the scoring to be
RUST ICC 0.53). included in sum total mRUST analysis. In the event of
Several studies assessing the utility of the RUST
and/or mRUST have considered them to be “valid” or
“valid and reliable” in adult patients with tibia diaphyseal
fractures treated with IMN.4–10 RUST and/or mRUST cri-
teria have also been applied in other patient populations
such as in pediatric osteogenesis imperfecta patients,11
pediatric congenital tibial pseudoarthrosis (neurofibromato-
sis type I) patients,12 and adult high-tibial osteotomy pa-
tients (modification of RUST criteria).13 However, no
published studies have examined the reliability of any cor-
tical scoring system in patients with bone defects, nor in
a cohort of patients treated with definitive external fixation.
The current study evaluated inter-rater reliability of the
mRUST at both the composite and cortex level in a cohort
of patients with operatively treated open tibia shaft fractures
with associated bone loss.
PATIENTS AND METHODS

There were 739 patients initially identified in a retro-
spective review of long bone fractures treated with definitive
surgical stabilization between June 2007 and May 2012 from
15-level one civilian trauma centers and 2 military treatment
facilities as part of the Major Extremity Trauma and
Rehabilitation Consortium14 RETRODefect study. Included
were skeletally mature patients ($18 years old) with a bone
defect of $1 cm in length and $50% cortical loss. For the
purpose of segmenting the results, the defect size was defined
as the average bone loss measurement across the anterior,
posterior, medial, and lateral cortices on the first postfixation
radiograph. Inclusion for this current study was limited to
patients with open, diaphyseal tibia shaft fractures (OTA/
AO 42)15 who had final AP and lateral radiographs between
3 and 13 months after definitive fixation and who did not
undergo an amputation before the final radiograph. Patients
with more than one type of definitive fixation instrumentation
(eg, both IMN and plating) seen on the final (rated) radio-
graph were excluded. Radiographs for all potential cases were
reviewed by a senior orthopedic traumatologist to confirm
that selection criteria were met.
Six experienced, fellowship-trained orthopedic trauma
surgeons participated in the mRUST scoring process. Each
case was reviewed by exactly 3 surgeons, assigned at random,
from this pool of 6. The raters were blinded to the initial
radiographs, defect size, and bone grafting status. Reviewers
were provided information on injury date, fixation date, and FIGURE 1. Flow diagram of patient selection.
J Orthop Trauma Volume 33, Number 6, June 2019 Inter-Rater Reliability of the mRUST
multiple, discrete defects, reviewers were instructed to score Sum Total mRUST Analyses
the area of the largest defect. Table 2 summarizes inter-rater reliability of the sum
Inter-rater reliability was assessed for the sum total total mRUST scores, and Figure 2 provides a visual display
mRUST score and for the score given to each cortex. Results of the concordance of mRUST scores given to each individual
were stratified by the type of instrumentation seen on the case. The overall KA was 0.64 (95% CI, 0.54–0.71). The
radiographs at the time of evaluation: IMN, plate, definitive KAs for the IMN, plate, ring, and no instrumentation groups
external fixation, or none (indicating previous removal of all were 0.65 (95% CI, 0.53–0.74), 0.88 (95% CI, 0.60–0.94),
instrumentation). Reliability of the sum total mRUST score 0.47 (95% CI, 0.13–0.69), and 0.48 (95% CI, 0.28–0.63),
was also assessed based on (1) presence/absence of bone respectively.
grafting, (2) time from grafting to scored radiograph (30–90
days and .90 days), and (3) defect size: small (,2.5 cm),
medium (2.5–5.0 cm), and large (.5.0 cm).
Cortex-Level Analyses
Krippendorff’s Alpha (KA) statistic, appropriate for Among the 213 cases in the initial cohort, all 3
ordinal data, was the primary method used to assess inter- reviewers agreed on the pattern of scoreable cortices (ie,
rater reliability.16,17 ICC, appropriate for continuous data, is which cortices can or cannot be scored) for 90.4% of cases
presented to enable comparison with previous work.18 For with IMNs, 88.9% of cases with no instrumentation,
both KA and ICC, a value of 1 indicates complete agreement 44.8% of cases still in rings, and 20.8% of cases with
among raters. Visual displays of agreement/disagreement of plates. A supplemental table (see Table, Supplemental
raters within individuals are also presented to aid interpreta- Digital Content 1, http://links.lww.com/JOT/A663) dis-
tion. Analyses were conducted in R 3.4.2 (R Foundation for plays inter-rater reliability estimates by cortex, overall
Statistical Computing, Vienna, Austria) and STATA (Stata-
Corp LLC, College Station, TX).
TABLE 1. Treatment and Injury Characteristics Among
Patients With and Without at Least Two Valid mRUST Scores
RESULTS Entire ,2 Valid ‡2 Valid
Among the 739 patients identified as eligible for the Cohort mRUST Scores mRUST Scores
RETRODefect study, 213 patients (Fig. 1) including 183 men Total, N 213 31 (15%)* 182 (85%)*
and 30 women, with an average age of 34.3 years (range, 18– Instrumentation on
68) met the selection criteria for this analysis. The average rated X-rays
time between initial definitive fixation and the rated radio- IMN 115 5 (4%) 110 (96%)
graphs was 294 6 85 days. The average bone defect size Defect size
was 3.6 6 2.9 cm. Initial definitive fixation was accomplished ,2.5 cm 73 3 (4%) 70 (96%)
using an IMN in 118 (55%) cases, plate and screws in 24 2.5–5.0 cm 23 2 (9%) 21 (91%)
(11%), or multiplanar (ring) external fixator in 71 (33%). .5.0 cm 19 0 (0%) 19 (100%)
There were 63 cases (30%) in which the final radiograph ORIF w/plate 24 14 (58%) 10 (42%)
had either no fixation instrumentation (n = 45) or different Defect size
instrumentation from the initial definitive fixation (n = 18). Of ,2.5 cm 12 7 (58%) 5 (42%)
the 45 patients who had no instrumentation, 41 were origi- 2.5–5.0 cm 9 5 (56%) 4 (44%)
nally treated with definitive external fixation. Bone grafting .5.0 cm 3 2 (67%) 1 (33%)
was performed in 109 (51%) subjects at an average of 188 6 External fixator 29 11 (38%) 18 (62%)
90 days before the scored radiographs. Defect size
Of the 213 patients who met selection criteria, all 4 ,2.5 cm 9 2 (22%) 7 (78%)
cortices could be scored by all 3 reviewers for 158 (74%) 2.5–5.0 cm 6 4 (67%) 2 (33%)
cases, by 2 of 3 reviewers for 24 (11%) cases, by one of 3 .5.0 cm 14 5 (36%) 9 (64%)
reviewers for 21 (10%), and by zero of 3 reviewers for 10 None 45 1 (2%) 44 (98%)
(5%) cases. Thus, there were 182 (85%) cases contributing to Defect size
the estimation of KA and ICC scored by $2 raters. There ,2.5 cm 15 1 (7%) 14 (93%)
were 31 (15%) cases scored by ,2 raters that could not 2.5–5.0 cm 14 0 (0%) 14 (100%)
contribute to the estimate of KA and ICC (Table 1). Of these .5.0 cm 16 0 (0%) 16 (100%)
excluded cases, 14 patients had a plate (58% of plated pa- Time to X-rays 294 6 85 299 6 88 293 6 85
tients), 11 had a ring (38% of rings), 5 had an IMN (4% of Bone defect size
IMNs), and 1 patient had no instrumentation (2% of patients ,2.5 cm 109 13 (12%) 96 (88%)
with no instrumentation). Thus, most of the noncontributing 2.5–5.0 cm 52 11 (21%) 41 (79%)
cases had either a plate or ring seen on the rated radiograph .5.0 cm 52 7 (13%) 45 (87%)
(25 of 31, 81%), suggesting an obstructed view of the fracture Bone grafting status
callous secondary to the instrumentation. Among the 182 No 104 13 (12%) 91 (88%)
contributing cases, 104 of 110 IMNs (95%), 3 of 10 plates Yes 109 18 (17%) 91 (83%)
(30%), 11 of 18 rings (61%), and 40 of 44 cases with no *Row %.
instrumentation (91%) could be scored by all 3 reviewers.
TABLE 2. Inter-rater Reliability of the mRUST Score, Overall and Stratified by Final Instrumentation, Defect Size, and Bone
Grafting Status
Not Used in KA and ICC Used in KA and ICC
0/3 Scores* 1/3 Scores* 2/3 Scores* 3/3 Scores* KA (95% CI) ICC (95% CI)
Overall 10 (5%) 21 (10%) 24 (11%) 158 (74%) 0.64 (0.54–0.71) 0.71 (0.65–0.77)
By instrumentation
IMN 1 (1%) 4 (3%) 6 (5%) 104 (90%) 0.65 (0.53–0.74) 0.75 (0.68–0.81)
Plate 6 (25%) 8 (33%) 7 (29%) 3 (13%) 0.88 (0.60–0.94) 0.90 (0.72–0.97)
Ring 3 (10%) 8 (28%) 7 (24%) 11 (38%) 0.47 (0.13–0.69) 0.62 (0.36–0.83)
None 0 (0%) 1 (2%) 4 (9%) 40 (89%) 0.48 (0.28–0.63) 0.57 (0.41–0.72)
Defect size (cm)
,2.5 5 (5%) 8 (7%) 9 (8%) 87 (80%) 0.60 (0.46–0.71) 0.70 (0.61–0.78)
2.5–5.0 3 (6%) 8 (15%) 6 (12%) 35 (67%) 0.77 (0.60–0.85) 0.85 (0.77–0.91)
.5.0 2 (4%) 5 (10%) 9 (17%) 36 (69%) 0.57 (0.36–0.71) 0.67 (0.53–0.79)
Grafted before X-ray?
No 3 (3%) 10 (10%) 10 (10%) 81 (78%) 0.61 (0.47–0.72) 0.70 (0.61–0.78)
Yes 7 (6%) 11 (10%) 14 (13%) 77 (71%) 0.66 (0.54–0.75) 0.73 (0.64–0.80)
Yes, days prior:
,30 0 (0%) 0 (0%) 1 (33%) 2 (67%) — —
30–90 0 (0%) 3 (21%) 1 (7%) 10 (71%) 0.60 (0.19–0.79) 0.67 (0.31–0.90)
.90 7 (8%) 8 (9%) 12 (13%) 65 (71%) 0.66 (0.52–0.76) 0.74 (0.65–0.82)
*The number of raters (0, 1, 2, or 3) out of 3 who provided an mRUST score for all 4 cortices of a given case.
and stratified by treatment instrumentation. Overall, the was lowest for the lateral cortex in the Ring (KA 0.04)
inter-rater reliability was lower for the lateral and posterior and the “None” groups (KA 0.36), for the anterior cortex
cortices compared with medial and anterior cortices. When in the Plate group (KA 0.37) and for the posterior cortex in
stratified by instrumentation type, inter-rater reliability the IMN group (KA 0.59).
FIGURE 2. Visualizing disagreement in mRUST by type, or lack, of instrumentation visible on final radiograph applied to 213
cases. Each rater is represented by a symbol with “/”, “X”, and “ ” representing agreement among the 3 raters. Presence of the
“X” symbol indicates that 2 (of 3) raters were in perfect agreement for a given case, and presence of the “ ” symbol indicates
that all 3 raters were in perfect agreement. The “/” symbol indicates that a single rater provided a different assessment than the
other raters. For included cases, the vertical line segment connects the minimum and maximum scores for a single patient. The
figure is stratified into 4 panels by instrumentation group. Within each panel, scoring disagreement (as measured by the length of
the line segments) increases from left to right; noncontributing cases ($2 N/As) are placed furthest to the right. Multiple cases
straddle a previously defined score indicating union [mRUST = 13 (horizontal line)]. Thirty-nine (18%) cases had $1 score above
and another score below this threshold (IMN: 15%, none: 38%, plate: 0%, ring: 17%). “IMN” indicates intramedullary nail;
“None” indicates cases with no instrumentation seen on rated radiographs; “N/A” indicates that a score could not be computed;
Alpha indicates Krippendorff’s alpha. Editor’s Note: A color image accompanies the online version of this article.
Other Subgroup Analyses an abnormal appearance to the tibia with the potential for less
Inter-rater reliability did not differ substantially by discrete cortices (Fig. 3). Furthermore, the mRUST scores can
presence or absence of bone grafting (KA 0.66 vs. 0.61) or be influenced by the amount of fracture callus present, and so,
the time from grafting to scored radiograph (KA 0.60 for 30– differences in agreement may also be due to the different
90 days vs. 0.66 for grafting .90 days). Inter-rater reliability biomechanical environments seen in patients treated with
for cases with small (,2.5 cm) and large (.5.0 cm) defects IMN versus plates versus ring fixators.5
was comparable (KA 0.60 vs. 0.57); agreement for medium Nearly all previous RUST and mRUST studies exam-
(2.5–5.0 cm) defects was higher (KA 0.77). ined narrowly defined patient populations.4–7,10–12 In the cur-
rent study, there was relatively greater heterogeneity in
treatment. Specifically, inclusion was not limited to a specific
DISCUSSION fixation device at the time of initial definitive fixation surgery.
Cortical scoring systems have become common tools in Some subjects in this study underwent revision fixation sur-
reporting radiographic progression toward union in lower- gery and either had additional, or entirely different, instru-
extremity fracture trials. Since the initial description by Whelan mentation seen on the final rated radiograph. In nearly one-
et al,4 the RUST and/or mRUST has been applied to, and third of cases (n = 63/213), the instrumentation present on the
reliability assessed in, multiple patient populations5–7,10–12 rated radiographs was different than the initial definitive fix-
and in animal studies.19–21 Among adult patients, the RUST ation (eg, patient initially treated with plating was later con-
has been judged to be most reliable in the assessment of diaph- verted to IMN, patient initially treated with a ring fixator had
yseal tibia fractures that are treated with an IMN, with reported it removed, etc.). Although patient selection criteria was lim-
ICCs ranging from 0.67 to 0.87.4,6,7,10 The inter-rater reliability ited to only 3 types of definitive fixation, there were 4 differ-
of mRUST in humans was first evaluated by Litrenta et al,5 ent groups based on the type(s) of, or lack of, instrumentation
who reported ICCs of 0.74 (95% CI, 0.68–0.81) and 0.59 (95% present on the rated radiographs. Thus, the results of this
CI, 0.51–0.67) for metadiaphyseal distal femur fractures treated study may be more generalizable to real-world orthopedic
with IMNs and plates, respectively.5 By contrast, the current trauma populations.
study focused on a population of patients with open, diaphyseal
tibia fractures with a bone defect of at least 1 centimeter with at
least 50% cortical loss treated operatively with IMN, plating, or
definitive external fixation, with or without bone grafting. Rel-
ative to the Litrenta study, inter-rater reliability, as measured by
ICCs, was nearly identical for IMNs (ICC 0.75), but higher for
plates (ICC 0.90).5 However, the inter-rater reliability results
for plates and rings must be viewed with great caution as 58%
(n = 14) of plate and 38% (n = 11) of ring cases could not
contribute to the estimation of KA and ICC because they could
not be scored by $2 raters. In addition, only 13% of plates and
only 38% of rings had all 4 cortices scored by all 3 reviewers,
substantially reducing the utility of this measure in tibial frac-
tures with a bone defect and treated with plates or rings.
The current study introduces 2 new subgroups to
analysis of the inter-rater reliability of the mRUST: patients
treated with definitive multiplanar external fixation and cases
with no instrumentation present due to removal before the
radiographic evaluation. Metallic instrumentation can block
a raters’ view(s) of the cortices; inter-rater reliability for cases
with external fixators in place might be expected to be similar
to the Litrenta study of plating (ICC 0.59), although they did
not report difficulty with scoring radiographs.5 In fact, the
ICC for external fixators was 0.62, but this is based on a subset
of only 62% cases that could be scored by $2 raters. Con-
versely, inter-rater reliability for cases with no instrumenta-
tion was expected to be at least as high as IMNs (ICC 0.74)5
because there were no implants blocking the view of the
cortices. However, the inter-rater reliability was lower than
anticipated with an ICC of 0.57. Notably, most of the cases
with no instrumentation (n = 40/45) were initially treated with
definitive external fixation. We hypothesize that the relatively FIGURE 3. Radiographs from a patient with a clinically healed
low level of reliability may be due to the severity of the tibia fracture showing persistent abnormal appearance of the
original injury in patients treated with definitive external fix- bone. This is an example of a case that may be unreliably
ation because these were frequently grafted and can heal with scored under the mRUST protocol.
Cortical scoring has been shown to correlate both with REFERENCES

clinical markers of fracture healing8 and with biomechanical 1. Cook GE, Bates BD, Tornetta P, et al. Assessment of fracture repair. J
healing.20,21 Previous analyses5,21 suggest an mRUST thresh- Orthop Trauma. 2015;29(suppl 12):S57–S61. Accessed November 27,
2017.
old of 13 to define radiographic union. Of the 182 included 2. Bhandari M, Guyatt GH, Swiontkowski MF, et al. A lack of consensus in
patients, 39 (18%) had $1 score above and another score the assessment of fracture healing among orthopaedic surgeons. J Orthop
strictly below this threshold (IMN: 15%, none: 38%, plate: Trauma. 2002;16:562–566. Accessed November 21, 2017.
0%, ring: 17%) (Fig. 2). These results show marked variabil- 3. Kooistra BW, Sprague S, Bhandari M, et al. Outcomes assessment in
ity in mRUST scoring around the union threshold based on fracture healing trials: a primer. J Orthop Trauma. 2010;24(suppl 1):
S71–S75. Accessed November 21, 2017.
one set of radiographs per patient, measured generally late in 4. Whelan DB, Bhandari M, Stephen D, et al. Development of the radio-
the healing process (mean/median 294/321 days after initial graphic union score for tibial fractures for the assessment of tibial frac-
definitive fixation). Given that mRUST is used for assessing ture healing after intramedullary fixation. J Trauma. 2010;68:629–632.
progress toward union, our study would have greater gener- 5. Litrenta J, Tornetta P, Mehta S, et al. Determination of radiographic
healing: an assessment of consistency using RUST and modified RUST
alizability for the assessment of fracture union in this popu- in metadiaphyseal fractures. J Orthop Trauma. 2015;29:516–520. Ac-
lation if the reviewed radiographs were more uniformly cessed November 21, 2017.
distributed along the healing continuum. 6. Azevedo FA, Cotias RB, Azi ML, et al. Reliability of the radiographic
The limitations in this study include those inherent to union scale in tibial fractures (RUST). Rev Bras Ortop. 2017;52:35–39.
any uncontrolled, retrospective study: heterogeneity of the Accessed November 21, 2017.
7. Leow JM, Clement ND, Tawonsawatruk T, et al. The radiographic union scale
patient population, variability in length of follow-up, and in tibial (RUST) fractures: reliability of the outcome measure at an indepen-
challenges with collection of data from multiple centers by dent centre. Bone Joint Res. 2016;5:116–121. Accessed November 21, 2017.
multiple research staff. There were 31 (15%) subjects who 8. Cekiç E, Alıcı E, Yeşil M. Reliability of the radiographic union score for
were excluded from our estimates of inter-rater reliability tibial fractures. Acta Orthop Traumatol Turc. 2014;48:533–540. Ac-
cessed November 21, 2017.
analysis because their radiographs could not be scored by $ 2 9. Kooistra BW, Dijkman BG, Busse JW, et al. The radiographic union
raters. We were unable to account for this disagreement in scale in tibial fractures: reliability and validity. J Orthop Trauma. 2010;
reviewers’ ability to score radiographs because typical con- 24(suppl 1):S81–S86. Accessed November 21, 2017.
sensus models do not account for raters that abstain from 10. Bhandari M, Kooistra BW, Busse J, et al. Radiographic union scale for
rating. If accounted for, this disagreement would certainly tibial (r.u.s.t.) fracture healing assessment: preliminary validation. Orthop
Proc. 2011;93-B (supp IV):575.
have lowered inter-rater reliability estimates. Thus, the re- 11. Franzone JM, Finkelstein MS, Rogers KJ, et al. Evaluation of fracture
ported KA and ICC from our study are not fully representa- and osteotomy union in the setting of osteogenesis imperfecta: reliability
tive of the true inter-rater reliability, particularly for patients of the modified radiographic union score for tibial fractures (RUST).
with plates or multiplanar external fixators. The high exclu- J Pediatr Orthop. 2017. Accessed November 21, 2017. [epub ahead of print].
12. Richards BS, Wilkes D, Dempsey M, et al. A radiographic scoring
sion rates in these subgroups also limited interpretation and system to assess healing in congenital pseudarthrosis of the tibia. J Pe-
strength of conclusions because of the wide confidence inter- diatr Orthop B. 2015;24:118–122. Accessed November 21, 2017.
vals observed. In addition, cases that underwent bone grafting 13. van Houten AH, Heesterbeek PJC, van Heerwaarden RJ, et al. Medial
were not excluded from analysis: 17 (16% of grafted cases) of open wedge high tibial osteotomy: can delayed or nonunion be pre-
these cases were within 90 days of grafting surgery, 14 of dicted? Clin Orthop Relat Res. 2014;472:1217–1223. Accessed Novem-
ber 21, 2017.
which were included in the inter-rater reliability analysis. 14. Building a clinical research network in trauma orthopaedics: the major
Although we did not find a difference in the inter-rater reli- extremity trauma research consortium (METRC). J Orthop Trauma.
ability of grafted cases between 30–90 and .90 days after 2016;30:353–361. Accessed January 4, 2018.
grafting, some surgeons may find it difficult to accurately 15. Marsh JL, Slongo TF, Agel J, et al. Fracture and dislocation classification
compendium—2007: orthopaedic trauma association classification, data-
score radiographs close to the time of graft surgery. Finally, base and outcomes committee. J Orthop Trauma. 2007;21(10 suppl):1.
the raters in our study were highly experienced orthopedic Accessed January 12, 2018.
traumatologists, suggesting that our estimates of inter-rater 16. Krippendorff K. Computing Krippendorff’s Alpha-Reliability; 2011.
reliability could be higher than might be expected when Available at: https://repository.upenn.edu/asc_papers/43. Accessed Feb-
applied to the broader community of surgeons. ruary 13, 2018.
17. Krippendorff K. Content Analysis: An Introduction to its Methodology.
In this retrospective study, the mRUST seems similarly 2nd ed. Thousand Oaks, CA: SAGE; 2004:211–251. Accessed February 13,
reliable to previous studies in the evaluation of patients with 2018.
open, diaphyseal tibia fractures with associated bone defects 18. Tawonsawatruk T, Hamilton DF, Simpson A, et al. Validation of the use
treated with intramedullary nails, but the use of this cortical of radiographic fracture-healing scores in a small animal model. J Orthop
Res. 2014;32:1117–1119. Accessed February 8, 2018.
scoring either in patients treated with plates and with ring 19. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reli-
fixators or in cases with removal of all instrumentation is less ability. Psych Ol Bull. 1979;86:420–428.
reliable. Because the assessment of fracture union continues 20. Litrenta J, Tornetta P, Ricci W, et al. In vivo correlation of radiographic
to be a challenging task for orthopedic surgeons, especially scoring (radiographic union scale for tibia fractures) and biomechanical
among patients with open, diaphyseal tibia fractures with data in a sheep osteotomy model: can we define union radiographically?
J Orthop Trauma. 2017;31:127–130. Accessed November 21, 2017.
associated bone defects, a prospective, controlled study is 21. Cooke ME, Hussein AI, Lybrand KE, et al. Correlation between RUST
needed to definitively assess the utility of this tool in patients assessments of fracture healing to structural and biomechanical proper-
with bone defects. ties. J Orthop Res. 2017. Accessed November 21, 2017.
APPENDIX 1. Major Extremity Trauma and Rehabilitation Consortium Corporate Authors

Brooke Army Medical Center: Christina M. Hylden, Howe, BS; University of Oklahoma: David Teague, MD;
MD, and Daniel J. Stinner, MD; Carolinas Medical Center: University of California San Francisco: Theodore Miclau,
Erica Andrews McArthur, MD; Kaitlyn M. Hurst, MD; and MD; Vanderbilt University Medical Center: Kristin R.
Katherine Sample, MD; Denver Health & Hospital Author- Archer, PhD, DPT; Basem Attum, MD, MS; Robert H.
ity: Corey Henderson, MS; Hennepin County Medical Cen- Boyce, MD; Vamshi Gajari, MBBS; A. Alex Jahangir, MD;
ter: Andrew H. Schmidt, MD, and Jerald R. Westberg, BA; Andres Rodriguez-Buitrago, MD; and Manish K. Sethi, MD;
Inova Fairfax Hospital: Robert A. Hymes, MD, and Lolita Wake Forest Baptist University Medical Center: Eben A.
Ramsey, PhD, RN, CCRC; MetroHealth Medical Center: Carroll, MD; Holly Pilson, MD; Robert C. Weinschenk,
Heather A. Vallier, MD, and Mary A. Breslin, BA; University MD; and Martha B. Holden, AAS, AA; Walter Reed
of Miami Ryder Trauma Center: Gregory A. Zych, DO, and National Military Medical Center: Benjamin K. Potter,
Gabriela M. Zych, BS, CCRC; St. Louis University Hospital: MD; Xochitl Ceniceros, PhD, RN; Jean-Claude G. D’Alleyr-
Lisa K. Cannada, MD; Tampa General Hospital: Hassan R. and, MD; Wade T. Gordon, MD; Whitley B. Lucio; and
Mir, MD, MBA, FACS; Rafael Serrano, MD; and Barbara Sandra L. Waggoner, BA; METRC Coordinating Center at
Steverson, MHA; University of Maryland R Adams Cowley the Johns Hopkins Bloomberg School of Public Health:
Shock Trauma Center: Andrew G. Dubina, MD, and Andrea Renan C. Castillo, PhD, and Tara Taylor, MPH.
ORIGINAL ARTICLE
Knee Disarticulations Versus Transfemoral Amputations:

Functional Outcomes
Elizabeth M. Polfer, MD,*† Benjamin W. Hoyt, MD,†‡ Adam J. Bevevino, MD,*†
Jonathan A. Forsberg, MD, PhD,†‡ and Benjamin K. Potter, MD, FACS†‡
Conclusions: We detected no functional differences measured on

Objectives: To determine whether there is a patient-reported the PEQ, LLQ, SF-36, and Tegner Activity Scale scores between
functional difference between combat-related knee disarticulations KDs and TFAs. In the absence of a proven functional difference, we
(KDs) and transfemoral amputations (TFAs). advocate performing trauma-related amputations at the most distal
Setting: Role 3 Military Trauma Centers. level the osseous and soft tissue injuries permit.
Key Words: amputation, combat, transfemoral amputation, knee
Patients: We identified and contacted all KDs and TFAs performed
disarticulation, functional outcome
at the Walter Reed National Military Medical Center, Walter Reed
Army Medical Center, and National Naval Medical Center from Level of Evidence: Therapeutic Level III. See Instructions for
January 2003 until July 2012 to participate in a retrospective Authors for a complete description of levels of evidence.
functional cohort analysis. Ten KD patients were available for study
completion and were matched against 18 patients in the TFA group. (J Orthop Trauma 2019;33:308–311)
Intervention: Knee disarticulation versus transfemoral amputation. INTRODUCTION

Knee disarticulations (KDs) for lower extremity trauma
Main Outcome Measurements: The following surveys were have been performed since 1581 and have been described in
obtained from the participants—AAOS Lower Limb Outcome Ques- the American literature with varying functional results since
tionnaire (LLQ), Tegner Activity Scale, SF-36, and Prosthetic Eval- 1825.1 Initial concerns with this level of amputation included
uation Questionnaires (PEQs). bulbous distal portion due to the flaring of the condyles, diffi-
Results: Ten KD patients agreed to participate in the study, and 18 culty with prosthetic fit, and uneven knee hinge level when the
TFA matched controls were interviewed. Patients were followed up contralateral extremity is intact. However, some have stated
at an average of 66 months (interquartile range 50–79 months) after that a knee disarticulation is quick, safe, and less traumatic
injury. There were no significant differences with regard to the SF- than a transfemoral amputation (TFA).1 Furthermore, the bul-
36, PEQ, LLQ, and Tegner Activity Scale scores. bous distal residual limb provides for an end-bearing prosthetic
socket fit and superior suspension.2 Currently, there is limited
evidence differentiating which amputation level between KD
and TFA provides better long-term functional outcomes.
Accepted for publication December 29, 2018. Although some gait analysis studies suggest the longer
From the *Department of Orthopaedics, William Beaumont Army Medical the residual limb, the more advantageous the gait,3 other
Center, El Paso, TX; †Department of Surgery, Uniformed Services Uni- studies would suggest that KDs result in a worse gait com-
versity, Bethesda, MD; and ‡Department of Orthopaedics, Walter Reed pared with both transtibial amputations and TFAs.4 Previous
National Military Medical Center, Bethesda, MD.
studies have consistently demonstrated that with more prox-
Each author certifies that he or she has no commercial associations (eg, imal amputations, there is a higher energy expenditure with
consultancies, stock ownership, equity interest, and patent/licensing ambulation, an increase in gait asymmetries, and a decrease in
arrangements) that might pose a conflict of interest in connection with self-selected walking speeds.3,5–10 The longer residual limb
the submitted article. Each author certifies that all investigations were provides a longer moment arm for gait and maintains the
conducted in conformity with ethical principles of research. The authors
are employees of the US Government. This work was prepared as part of native insertion of the adductor musculature, preserving my-
their official duties. Title 17 USC §105 provides that “Copyright pro- odesis function in leg swing and sitting positioning. The pur-
tection under this title is not available for any work of the US Govern- pose of this study was to determine whether there was
ment.” Title 17 USC §101 defined a US Government work as a work a patient-reported functional difference between KD and
prepared by a military service member or employees of the US Govern-
ment as part of that person’s official duties. The opinions or assertions
TFA. Our null hypothesis was that there were no significant
contained in this article are the private views of the authors and are not to differences between the KD and TFA cohorts.
be construed as reflecting the views, policy or positions of the Depart-
ments of the Navy, Department of Defense nor the US Government.
Reprints: Elizabeth M. Polfer, MD, Department of Orthopaedics, William PATIENTS AND METHODS
Beaumont Army Medical Center, El Paso, TX 79920 (e-mail: Elizabeth.
polfer@gmail.com).
After institutional review board approval, we retrospec-
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. tively reviewed the surgery scheduling system at our
DOI: 10.1097/BOT.0000000000001440 institution and included all KDs and TFAs performed at our
J Orthop Trauma Volume 33, Number 6, June 2019 Knee Disarticulations vs. Transfemoral Amputations
institutions from January 2003 to August 2012. We collected The hamstrings are then sewn into the myodesis, and the
patient demographics to include date of injury, level of flaps and skin are closed.
amputation, residual length from the lesser trochanter, and
contralateral lower extremity adjusted injury score (AIS). All Statistical Analysis
patients were greater than 1 year from amputation at the time Data analysis was performed with the use of SPSS
of the study. Patients were excluded for contralateral lower 22.0.0.0 software (SPSS, Chicago, IL). A one-way analysis of
limb injury with AIS greater than 3 in an effort to minimize variance and a t test were used to calculate significant differ-
the confounding influence of combined extremity injuries and ences between the 2 groups, and the Pearson bivariate corre-
amputations. We also excluded TFAs with residual length lation test was performed to assess for associated variables.
less than 21 cm from the lesser trochanter to limit the Multiple linear regression analysis was performed to control
confounding effect of a short residuum on patient outcome. for potential confounding variables to include age, time to
The level of final amputation was determined by the operative follow-up, contralateral AIS pain, and general health score,
surgeon based on the availability of appropriate soft tissue and an ANOVA was performed to determine the effect of
coverage. amputation type and residual limb length on functional out-
Once a patient was deemed to meet the inclusion come measures. Our null hypothesis was that there were no
criteria, they were contacted via telephone or mail to obtain significant differences between the KD and TFA cohorts. We
consent for participation in the study. To ensure that the defined statistical significance as a 2-tailed a of ,0.05. Power
reported outcomes reflected a thorough assessment of func- analysis was performed post hoc using G*Power 3.1 software
tional and psychosocial outcomes after amputation, we (Heinrich-Heine-Universität Düsseldorf 2017) with the effect
looked at the difference between the AAOS Lower Limb size calculated from patient-reported overall function scores
Outcome Questionnaire (LLQ), change in the Tegner Activity as reported by Bosse et al.15 Post hoc power analysis demon-
Scale (TAS) score, Short Form-36 (SF-36) general health strated that the sample size was sufficient to detect clinically
scores, and Prosthetic Evaluation Questionnaires (PEQ). The significant differences between groups (1 2 B = 0.97).
AAOS LLQ and SF-36 are both general outcome scores for
postsurgical patients. Johanson et al11 demonstrated that the
LLQ was reliable and sensitive to acute changes in patient
functional and satisfaction status. They also reported that it RESULTS
might be applied to any level of the lower extremity without The knee disarticulation cohort consisted of 10 patients
sacrificing its reliability, presumably making it useful in com- injured at an average age of 22.8 (20–26) years and inter-
paring amputations performed at different levels. The PEQ is viewed at an average follow-up of 69 (25.9–128; SD 37)
a tool for measurement of perceived disability of amputation months since initial injury. Their contralateral lower limb
commonly used in the rehabilitation literature.12 Finally, the AIS was 1.10 (0–3, SD 1.197), and preinjury Tegner scores
TAS is a standardized scale of activity as determined by in this cohort were average 7.15 6 1.6. The transfemoral
participation in work and sporting activities that is reliable amputation cohort consisted of 18 patients injured at an aver-
for measurement in patients with orthopaedic pathology about age age of 25.4 (19–37) years and interviewed at an average
the knee and responsive to acute change.13 Once patients follow-up of 63.8 (33.8–103.4; SD 19.5) months. Their con-
were consented, these outcome scoring tools were either tralateral AIS was 1.94 (0–3, SD 0.87), and mean reported
mailed to the patient to complete or completed over telephone preinjury Tegner scores in this cohort was 7.53 6 1.17. The
interview. The questionnaires were scored, and the statistical average residual limb length for the TFA group was 28.9
analysis was performed per the guidance from the specific (21.45–36.19, SD 3.95) cm. There was no significant differ-
questionnaires. ence between groups in time to follow-up interview (P =
0.680), preinjury Tegner score (P = 0.24), or contralateral
Surgical Technique AIS (P = 0.070). The age at the time of initial injury was
The authors’ preferred technique for knee disarticula- significantly older for the TFA group (P = 0.049).
tion relies on adequate coverage using a gastrocnemius my- Change in the Tegner score was calculated from pre-
ofasciocutaneous flap. To achieve adequate coverage, the and post-injury Tegner scores. There was no significant
posterior flap should be equal or greater in length to the difference in the change in the Tegner score between cohorts
diameter of the knee joint. The authors retain the patella (P = 0.28). A summary score was calculated using the scoring
and stabilize the patellar tendon to the cruciate ligaments. tool for the LLQ. There was no significant difference in the
Finally, the flap is brought anteriorly, and the gastrocnemius LLQ score between groups (P = 0.35). Patient-reported pros-
fascia is sewn into the anterior knee joint retinaculum. thetic and overall outcomes were calculated using the scoring
For transfemoral amputation, the authors’ preferred tools for the PEQ and SF-36, and no significant differences in
technique is to develop a long medial myofasciocutaneous any outcome measure calculated by the PEQ or SF-36 were
flap sewn into a lateral flap through drill holes in the distal found (Table 1).
femur.14 The adductor magnus is brought over the cut end Within the long transfemoral group, there was no
and sutured to the lateral femoral cortex under tension while significant association between the residual limb length and
the leg is held in full adduction. The quadriceps are then LLQ (P = 0.64), change in the Tegner score (P = 0.77),
sewn over the cut bone end into the posterior cortex while physical function (P = 0.40), energy (P = 0.40), utility (P =
the leg is held in extension to prevent contracture. 0.91), ambulation (P = 0.84), or well-being (P = 0.57).
Polfer et al J Orthop Trauma Volume 33, Number 6, June 2019
study, which demonstrated that in patients who underwent

TABLE 1. Summary of Reported Outcome Scores by
KD with universal gastrocnemius myofasciocutaneous flaps
Amputation Level
for distal soft tissue coverage, there was no difference in
Questionnaire Outcome Measure KD TFA P patient-perceived functional outcome between the 2 groups.
Tegner Preinjury 7.15 7.53 0.52 Although no statistical significance was identified, the differ-
Postinjury 4.65 4.17 0.52 ences, which may have been detected with greater patient
Score change 22.50 23.36 0.28 numbers, seem to favor knee disarticulation.
LLQ LLQ score 88.38 85.06 0.45 Matsen et al19 examined the patients’ perspective after
PEQ Utility 75.33 71.35 0.35 a lower extremity amputation and determined that it was not
Appearance 64.43 67.21 0.70 the level of the amputation that effected their perceived result,
Sounds 61.50 65.32 0.77 but rather it was dependent on comfort of the residual limb,
Residual limb health 65.68 57.57 0.26 condition of the contralateral limb, prosthetic comfort, func-
Perceived response 89.93 93.00 0.38 tion, and appearance, social factors, and ability to exercise
Frustration 60.50 54.83 0.65 recreationally. This gives more credence to the concept that
Social burden 80.57 78.80 0.82 regardless of the level of amputation, the most important factor
Ambulation 62.00 66.27 0.40 when determining amputation level is the soft tissue enve-
Well-being 83.75 73.06 0.15 lope.18,20 This was also articulated when Pinzur et al17 recom-
SF-36v2 Physical function 59.00 57.50 0.85 mended against a knee disarticulation in trauma unless there
Role limitation due to 70.63 64.93 0.47 was sufficient gastrocnemius to allow for appropriate padding.
physical health They went on to describe using either a long posterior myo-
Role limitation due to 76.67 76.85 0.98 fasciocutaneous flap encompassing the gastrocnemius or a sep-
emotional health
arate gastrocnemius flap if using sagittal skin flaps.
Energy 57.00 53.06 0.50
A benefit of performing a knee disarticulation is that it
Emotional well-being 64.80 66.00 0.82
is a less traumatic amputation level with less soft tissue
Social function 75.00 72.92 0.80
dissection, less blood loss, and no osseous cuts.1,14,17 The
Pain 67.00 64.58 0.74
follow-up surgical care also demonstrates resistance to infec-
General health 68.00 67.50 0.93
tion and postoperative complications when performed with
posterior myocutaneous flaps.21 Tintle et al22 performed an
analysis on the revisions of traumatic lower extremity ampu-
DISCUSSION tations. The follow-up surgical revisions included infection,
Although some studies have speculated that patients with wound dehiscence, heterotopic ossification, neuromas, scar
KDs do worse than those with TFAs,4,15 there are other studies revisions, and myodesis failure. KDs underwent significantly
that suggest that KDs do better. The frequently cited reasons fewer revision surgeries than either transtibial amputations or
for KDs being favored are due to the end-bearing nature of the TFAs. It was also noted that there is less incidence of hetero-
amputation level, the soft tissue balancing about the femur, the topic ossification in KDs, possibly because of the intact cor-
long lever arm for wheelchair transfers, the ease of surgical tices and (with viable gastrocnemius being a prerequisite for
procedure, and the favorable platform for wheelchair sit- the amputation level) less injured muscle.
ting.10,16–18 The favorable soft tissue balancing results in fewer Gait analysis studies have been performed analyzing
hip flexion contractures and less abductor drift.16,17 This study the gaits of various amputation levels (ankle disarticulation,
demonstrated no difference between the 2 groups. transtibial amputations, KDs, TFAs, and hip disarticulations).
The Lower Extremity Assessment Project (LEAP) Self-selected walking speeds decrease as amputation levels
study15 evaluated a cohort of patients with severe lower move more proximal, resulting from decreased cadence and
extremity injuries over 2 years, of which 149 underwent reduced stride length.5 A KD provides decreased metabolic
amputation during the initial hospitalization and an additional cost of ambulating as compared to a higher level of amputa-
25 underwent amputation on subsequent stays. In this group, tion while allowing for direct load transfer of body weight
initial analysis demonstrated frequently cited, but statistically into the prosthesis.8,10 As a result, as the amputation level
nonsignificant difference in patient-reported Sickness Impact moves distal in the lower extremity, functional outcome and
Profile (SIP) between patients who underwent KD or TFA. A walking independence improve.18
later analysis of 2-year data looking at functional outcomes There are several notable limitations to the current
after trauma-related amputations in 161 patients using the study. First, it is a survey-based study and therefore subject to
Sickness Impact Profile (SIP) suggested that KDs did worse some degree of bias and inconsistent reporting of current and
than both transtibial and TFAs with the overall SIP scores past events and symptoms. Second, this study is limited by
significantly higher in the knee disarticulation group.4 How- the number of patients in the knee disarticulation cohort. In
ever, those authors noted that 17 of the 18 amputations at this the decade encompassed in this study, there were only 21
level occurred through the zone of injury and 12 of these patients identified who underwent a KD, whereas there were
patients lacked a gastrocnemius myofasciocutaneous flap for over a 100 patients identified who had undergone unilateral
distal soft tissue coverage, which may have produced a sub- TFA. Third, to avoid consideration of the confounding
optimal bone–soft tissue interface at the end-bearing surface. deleterious effects of decreased limb length, we specifically
These functional conclusions are in contradistinction to our only looked at longer TFAs, restricting generalizability to all
J Orthop Trauma Volume 33, Number 6, June 2019 Knee Disarticulations vs. Transfemoral Amputations
TFAs. Fourth, the 2 amputation cohorts were not randomly 3. Bell JC, Wolf EJ, Schnall BL, et al. Transfemoral amputations: the effect
selected for inclusion into their respective groups, but rather of residual limb length and orientation on gait analysis outcome meas-
ures. J Bone Joint Surg Am. 2013;95:408–414.
underwent amputation at the level that the treating surgeon 4. MacKenzie EJ, Bosse MJ, Castillo RC, et al. Functional outcomes fol-
determined to be most appropriate based on injuries. Fifth, the lowing trauma-related lower-extremity amputation. J Bone Joint Surg.
outcome measures used may be imperfect for appreciating 2004;86-A:1636–1645.
a significant difference in patient outcomes. Legro et al23 did 5. Bell JC, Wolf EJ, Schnall BL, et al. Transfemoral amputations: is there
an effect of residual limb length and orientation on energy expenditure?
not appreciate significant differences between PEQ scores Clin Orthop Relat Res. 2014;472:3055–3061.
after transfemoral and transtibial amputations despite the 6. Genin JJ, Bastien GJ, Franck B, et al. Effect of speed on the energy cost
large functional discrepancy between these levels. However, of walking in unilateral traumatic lower limb amputees. Eur J Appl
this same difficulty was also noted for Sickness Impact Profile Physiol. 2008;103:655–663.
scores in the Lower Extremity Assessment Project (LEAP) 7. Göktepe AS, Cakir B, Yilmaz B, et al. Energy expenditure of walking
with prostheses: comparison of three amputation levels. Prosthetics Or-
study, in which transfemoral amputees scored better than thot Int. 2010;34:31–36.
transtibial patients, albeit not by a significant margin.4 In 8. Jeans KA, Browne RH, Karol LA. Effect of amputation level on energy
addition, the TAS can only reach 6 or higher with participa- expenditure during overground walking by children with an amputation.
tion in sporting activities, potentially limiting its ability to J Bone Joint Surg. 2011;93:49–56.
appreciate differences at higher levels of function. We also 9. Schmalz T, Blumentritt S, Jarasch R. Energy expenditure and biome-
chanical characteristics of lower limb amputee gait: the influence of
asked patients to rate the preinjury TAS, potentially exposing prosthetic alignment and different prosthetic components. Gait Posture.
the change to recall bias. Last, as with other studies that report 2002;16:255–263.
on active duty military, the potential lack of generalizability 10. Pinzur MS. Gait analysis in peripheral vascular insufficiency through-
of our results is an inherent limitation. The patients included knee amputation. J Rehabil Res Dev. 1993;30:388–392.
11. Johanson NA, Liang MH, Daltroy L, et al. American academy of ortho-
in this study were young, previously healthy, and had excel- paedic surgeons lower limb outcomes assessment instruments. J Bone
lent rehabilitation and prosthetic resources. Joint Surg. 2004;86:902–909.
Potential strengths of this study include the reliable 12. Boone DA, Coleman KL. Use of the prosthesis evaluation questionnaire
access to a robust network of rehabilitative care experts that is (PEQ). JPO J Prosthetics Orthot. 2006;18:P68–P79.
consistent between all amputations performed in this study 13. Briggs KK, Lysholm J, Tegner Y, et al. The reliability, validity, and
responsiveness of the Lysholm score and Tegner activity scale for ante-
group. A second potential strength, which differentiates our rior cruciate ligament injuries of the knee: 25 years later. Am J Sports
findings from those in the LEAP cohort, is the consistent Med. 2009;37:890–897.
application of a surgical technique for knee disarticulation 14. Morse BC, Cull DL, Kalbaugh C, et al. Through-knee amputation in
that preserves a durable myofasciocutaneous flap for terminal patients with peripheral arterial disease: a review of 50 cases. J Vasc
Surg. 2008;48:638–643; discussion 643.
padding and for transfemoral amputation that applies an 15. Bosse MJ, MacKenzie EJ, Kellam JF, et al. An analysis of outcomes of
appropriately tensioned myodesis of the adductor muscula- reconstruction or amputation after leg-threatening injuries. New Engl J
ture. In our cohorts, all KDs were performed with a gastroc- Med. 2002;347:1924–1931.
nemius myofasciocutaneous flap for the residual limb 16. Hagberg E, Berlin OK, Renström P. Function after through-knee com-
coverage and all TFAs underwent an adductor myodesis. pared with below-knee and above-knee amputation. Prosthetics Orthot
Int. 1992;16:168–173.
In conclusion, we found no functional differences 17. Pinzur MS, Bowker JH. Knee disarticulation. Clin Orthop Relat Res.
between a well-padded KD and long transfemoral amputa- 1999:23–28.
tion, as measured on the PEQ, LLQ, SF-36, and Tegner 18. Pinzur MS, Gottschalk FA, Pinto MA, et al; American Academy of
Activity Scale scores. Successful outcomes in any amputation Orthopaedic Surgeons. Controversies in lower-extremity amputation. J
Bone Joint Surg. 2007;89:1118–1127.
rely on appropriate patient selection, creating adequate distal 19. Matsen SL, Malchow D, Matsen FA, III. Correlations with patients’
padding for the bearing surface, and functionally appropriate perspectives of the result of lower-extremity amputation. J Bone Joint
myodesis for terminal control. As a result, we advocate Surg. 2000;82:1089–1095.
performing definitive combat- and trauma-related amputa- 20. Polfer EM, Tintle SM, Forsberg JA, et al. Skin grafts for residual limb
tions at the most distal level the soft tissue and osseous coverage and preservation of amputation length. Plastic Reconstr Surg.
2015;136:603–609.
injuries permit. 21. Nijmeijer R, Voesten HG, Geertzen JHB, et al. Disarticulation of the
knee: analysis of an extended database on survival, wound healing,
REFERENCES and ambulation. J Vasc Surg. 2017;66:866–874.
1. Batch JW, Spittler AW, McFaddin JG. Advantages of the knee disartic- 22. Tintle SM, Shawen SB, Forsberg JA, et al. Reoperation after combat-
ulation over amputations through the thigh. J Bone Jt Surg. 1954;36: related major lower extremity amputations. J Orthop Trauma. 2014;28:
921–930. 232–237.
2. Cull DL, Taylor SM, Hamontree SE, et al. A reappraisal of a modified 23. Legro MW, Reiber G, del Aguila M, et al. Issues of importance reported
through-knee amputation in patients with peripheral vascular disease. Am by persons with lower limb amputations and prostheses. J Rehabil Res
J Surg. 2001;182:44–48. Dev. 1999;36:155–163.
ORIGINAL ARTICLE
Ability of a Risk Prediction Tool to Stratify Quality and Cost

for Older Patients With Operative Ankle Fractures
Ariana Lott, MD,* Kenneth A. Egol, MD,* Thomas Lyon, MD,† and Sanjit R. Konda, MD*‡
Key Words: ankle fracture, risk stratification, triage, geriatric, cost

Objectives: To investigate the ability of a validated geriatric of care, hospital quality measures
trauma risk prediction tool to stratify hospital quality metrics and
inpatient cost for middle-aged and geriatric patients admitted from Level of Evidence: Prognostic Level III. See Instructions for
the emergency department for operative treatment of an ankle Authors for a complete description of levels of evidence.
fracture. (J Orthop Trauma 2019;33:312–317)
Design: Prospective cohort study.
Setting: Single Academic Medical Center. INTRODUCTION
Although there remains controversy over the preferred
Patients: Patients 55 years of age and older who sustained treatment of ankle injuries in geriatric patients, a significant
a rotational ankle fracture and who were treated operatively during portion of these patients have surgical fixation.1–3 Analysis of
their index hospitalization. the Medicare population reveals that 34% of ankle fractures
undergo surgical fixation, and even 25% of patients 85 years
Intervention: Calculation of validated trauma triage score, Score of age or older with ankle fractures undergo surgical fixation.4
for Trauma Triage in Geriatric and Middle Aged (STTGMA), using
The financial burden of these injuries is enormous with stud-
patient demographics, injury severity, and functional status. Patients
ies estimating that $11 billion is spent on foot and ankle
were stratified into groups based on scores to create a minimal-, low-
surgeries in the Medicare population annually.5
, moderate-, and high-risk cohort.
In our groups’ experience, younger patients with ankle
Main Outcome Measurements: Length of stay, complications, fractures are more easily discharged from the emergency
need for intensive care unit–/step-down unit–level care, discharge department setting and can be optimized for surgery in the
location, and index admission costs. outpatient setting. Older patients, however, are less likely to
be safely discharged from the emergency department setting
Results: Fifty ankle fracture patients met inclusion criteria. The and require inpatient admission for a variety of reasons includ-
mean length of stay was 7.8 6 5.2 days with a significant difference ing inability to safely mobilize with an assistive device, inad-
among the 4 risk groups (4.6-day difference between low and high equate home assistance to perform activities of daily living,
risk). 73.1% of minimal-risk patients were discharged home com- and/or active medical comorbidities or associated traumatic
pared with 0% of high-risk patients. There was no difference in injuries that require inpatient optimization. In addition, in the
complication rate or in need for intensive care unit–level care immediate postoperative setting, it is our groups’ experience
between groups. However, high-risk patients had a mean total inpa- that older patients are less likely to be discharged on the same
tient cost 2 times greater than that of minimal-risk patients. day as their surgery when compared with their young adult
counterparts. The discrepancy in perioperative management is
Conclusion: The Score for Trauma Triage in Geriatric and Middle-
skewed toward more costly care for older patients.
Aged tool is able to meaningfully stratify older patients with ankle
Recently, there has been a transition from a fee-for-
fracture who require operative fixation regarding hospital quality
service model to a bundled payment of care model in the area
metrics and cost. This information may allow for efficient targeted
of orthopaedic trauma. As such, it is important for providers,
reductions in costs while optimizing outcomes.
hospitals, and insurers to understand and predict patient
Accepted for publication January 2, 2019. outcomes (eg, hospital quality metrics) and cost to optimize
From the *Department of Orthopedics, NYU Langone Orthopedic Hospital, value-based care. Several studies have analyzed factors
NYU Langone Medical Center, New York, NY; †Department of Orthope- related to length of stay (LOS), complications, and inpatient
dics, NYU Lutheran Hospital, NYU Langone Medical Center, Brooklyn, mortality including American Society of Anesthesiologists
NY; and ‡Department of Orthopedics, Jamaica Hospital Medical Center, (ASA) scores and comorbidities.6–9 However, to the best of
Queens, NY.
The authors report no conflict of interest. No sources of funding have been our knowledge, no group has sought to prospectively stratify
granted for this study. Kenneth Egol, MD, is a consultant for Exactech, ankle fracture patients with a validated risk-stratification tool
and Sanjit Konda, MD, is a consultant for Stryker. to provide surgeons with actionable data regarding hospital
Reprints: Sanjit R. Konda, MD, Department of Orthopedic Surgery, NYU quality metrics and cost before surgical treatment. By identi-
Langone Medical Center, 301 East 17th St, New York, NY 10003
(e-mail: Sanjit.konda@nyumc.org).
fying patients on admission who are both low and high risks
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. for these aforementioned outcomes, surgeons will not only
DOI: 10.1097/BOT.0000000000001446 be able to provide their patients with better expectations
J Orthop Trauma Volume 33, Number 6, June 2019 Risk Prediction Ankle Fracture
regarding the expected hospital course, but also they will be considered significant. All statistical analyses were performed
able to better direct less or more resources to low- and high- using SPSS (Version 23; Chicago, IL).
risk patients, respectively, in an attempt to provide higher
value-based care. The purpose of this study is to investigate
whether a validated geriatric trauma risk tool can be used to RESULTS
stratify middle-aged and geriatric patients with ankle fracture
A total of 50 patients were consecutively enrolled in our
who require operative fixation regarding hospital quality met-
prospective registry and met inclusion criteria and were
rics and cost. Secondarily, the study aims to characterize
included in this analysis. The mean age of the cohort was
hospital quality metrics and costs of care for this cohort of
67.6 6 9.0 years. Of these 50 patients, 37 (74.0%) patients’
patients.
injuries were secondary to low-energy mechanisms and 13
(26.0%) patients were secondary to high-energy mechanisms.
PATIENTS AND METHODS Eight sustained (16.0%) open fractures. Most patients had
isolated ankle injuries. The average GCS was 14.9 6 0.9.
Between October 2014 and September 2016, patients
The mean AIS scores were as follows: 0.2 6 0.5 for the
55 years of age and older who presented to 1 academic
AIS head/neck, 0.1 6 0.2 for the AIS chest, and 3.2 6 0.4
medical center and sustained an ankle fracture that required
for AIS extremities. 12.0% of patients sustained a head/neck
inpatient admission and subsequent operative fixation during
injury and had an AIS head/neck score of greater than 0. The
their index admission were included in this study. An ankle
mean CCI was 0.9 6 1.4 with 62% of patients having a CCI
fracture is defined as a fracture involving the medial,
of 0. Using these variables, the mean STTGMA score was 1.6
posterior, and/or lateral malleolus of the ankle. Patients who
6 4.2%. Based on STTGMA score, 25 (50.0%) patients were
underwent surgery for other injuries were excluded. Pilon
determined to be minimal risk, 14 (28.0%) were low risk, 9
fracture patients involving the weight-bearing portion of the
(16.0%) were moderate risk, and 2 (6.0%) were high risk.
tibial plafond were excluded. On initial evaluation in the ED,
Baseline study characteristics stratified by the risk groups
patient demographics, injury severity, and functional status
are described in Table 1.
were recorded. Study variables included patient age; Glasgow
The mean LOS for the cohort was 7.8 6 5.2 days with
Coma Scale (GCS) on initial evaluation; mechanisms of
a significant difference between the 4 risk groups (P = 0.005)
injury; Abbreviated Injury Severity (AIS) subscores for the
(Table 2). Patients in the minimal-risk cohort had the mean
head and neck (AIS-HN), chest (AIS-CHS), and pelvis and
LOSs of 5.4 6 2.9 days compared with 11.6 6 7.4 days in the
extremity body regions (AIS-EXT); and Charlson Comorbid-
moderate-risk group. The mean number of complications in
ity Index (CCI) score. These variables were used to calculate
the cohort was 0.2 6 0.6 with no significant difference
a Score for Trauma Triage in Geriatric and Middle Aged
between the risk groups. However, there was a difference in
(STTGMA), which represents the predicted risk (0%–100%)
need for advanced level of care between the risk groups (P =
of inpatient mortality during the index hospitalization.
0.049). In the total cohort, 8.0% of patients required ICU/
STTGMA is a validated and published tool to calculate inpa-
SDU care. In the high-risk cohort, 50.0% of patients required
tient mortality risk.10 Patients were stratified into 4 groups
this advanced level of care compared with 0.0% of patients in
based on these scores to create a minimal-risk, low-risk,
the minimal-risk group requiring this care. This difference
moderate-risk, and high-risk cohort ranging from ,0.6%,
between the risk groups was also seen in the need for post-
0.6%–1.2%, 1.2%–4.0%, and .4.0%, respectively.
acute facility care, in which 72.0% of minimal-risk patients
Information on LOS, complications, need for intensive
were discharged home, whereas 0% of high-risk patients were
care unit (ICU)-/step-down unit (SDU)-level care, and
discharged home (Table 2).
discharge location was collected. Complications included
With respect to inpatient admission costs, there was
acute renal failure, surgical site infection, decubitus ulcer,
a large difference between the total inpatient costs among the
urinary tract infections, acute anemia, sepsis, pneumonia,
different STTGMA risk groups. The mean total index
acute respiratory failure, acute myocardial infarction, deep
admission cost in the high-risk cohort was double than that
vein thrombus, pulmonary embolism, cardiac arrest, and
of the minimal-risk cohort. This cost difference was also seen
stroke. Patients were followed up for 30 days to observe
in many subdivisions of care including room and board,
readmissions. Total costs of these hospitalizations were
laboratory/pathology, radiology, and allied health costs
obtained from the hospital finance department from their cost
(Figure 1). The relative comparative cost of care between
accounting system (EPSI, NY) and categorized into the
patients in the cohort is demonstrated in Figure 2 with each
following groups: room/board, ED, pharmacy, laboratory/
patient’s cost of care presented as a percentage of total cost of
pathology, radiology, dialysis, cardiology, procedure, allied
the entire cohort.
health costs, and others (eg, blood products). Our institution
considered this financial data proprietary information. Thus,
all cost data are reported as a proportion of the mean total
direct variable cost of care for the entire cohort of patients DISCUSSION
(“x”). One patient was excluded from the cost analysis This study demonstrates the variance in operative-
because the hospital was unable to obtain their direct variable treated ankle fractures and the ability of a middle-aged and
cost information. Analysis of variance was used to determine geriatric trauma triage score to risk-stratify patients. As would
differences between the different risk cohorts with P , 0.05 be expected, patients denoted as high risk were more likely to
Lott et al J Orthop Trauma Volume 33, Number 6, June 2019
TABLE 1. STTGMA Score Variables for Minimal-Risk, Low-Risk, Moderate-Risk, and High-Risk Cohorts
Minimal Risk Low Risk Moderate Risk High Risk Total Cohort
(n = 25) (n = 14) (n = 9) (n = 2) (n = 50) P
Age (y) (mean 6 SD) 61.7 6 4.2 70.7 6 7.7 76.4 6 7.9 81.0 6 14.5 67.6 6 9.0 ,0.001
GCS (mean 6 SD) 15.0 6 0.0 15.0 6 0.0 15.0 6 0.0 11.5 6 3.5 14.9 6 0.9 ,0.001
AIS H/N (mean 6 SD) 0.0 6 0.0 0.1 6 0.4 0.3 6 0.5 1.5 6 2.1 0.2 6 0.5 ,0.001
AIS Chest (mean 6 SD) 0.0 6 0.0 0.1 6 0.3 0.2 6 0.4 0.0 6 0.0 0.1 6 0.2 0.117
AIS E/P (mean 6 SD) 3.2 6 0.4 3.0 6 0.0 3.3 6 0.5 3.5 6 0.7 3.2 6 0.4 0.095
CCI (mean 6 SD) 0.2 6 0.4 0.8 6 1.1 2.4 6 1.8 4.0 6 0.0 0.9 6 1.4 ,0.001
STTGMA (%) (mean 6 SD) 0.4 6 0.1 0.8 6 0.2 2.8 6 1.2 17.3 6 17.1 1.6 6 4.2 ,0.001
have longer, costlier admissions requiring ICU-level care and fixation. Interestingly, they did not find that medical comor-
postacute care. bidities played a large role.
The middle-aged and geriatric trauma triage score used The high complication rate (2%–21.5%) seen in geri-
in this study was developed to predict inpatient mortality and atric patients who undergo operative fixation of their ankle
has been validated both in the National Trauma Databank fracture is one of the reasons that some argue against surgi-
(.100,000 patients) and prospectively at a Level 1 trauma cal fixation of these injuries in this elderly cohort.8 Wound
center.10,11 However, this study demonstrates its ability to complications are the most common complication in this
risk-stratify geriatric ankle fracture patients with respect to cohort.8 In an analysis of complications in Medicare popu-
hospital quality measures and costs. This score incorporates lation with ankle fractures, Koval et al8 cite that patients
a patient’s age, physiologic status, injury status, mechanism who underwent operative treatment had a 16 times greater
of injury, and comorbid status all of which are factors that risk of developing a medical and operative complication
have been shown to independently predict risk and poor out- than those who underwent nonoperative treatment. Patients
comes in the middle-aged and geriatric population.12–14 Ben- with a greater number of comorbidities and the presence of
efits of this tool include the ease of collection at the time of diabetes or peripheral vascular disease were at greater risk of
admission so that providers can use this risk-stratification tool complications. Given the large number of middle-aged and
to make care decisions even before admission. geriatric patients with these injuries, identifying high
Hospital LOS is not only a way to assess the resource utilization and high cost patients at the time of
effectiveness of operative treatment, but it is also a large admission would be helpful to guide these patients into
determinant of cost. Therefore, understanding the factors that value-based pathways designed to minimize extraneous cost
influence LOS is important not only in trying to improve and optimize outcomes. Although the incidence of compli-
outcomes in patients undergoing surgical fixation of ankle cations in this study cohort was low in comparison with
injuries but also in reducing cost. A recent study of 2008 previous studies, these patients were only followed through
Medicare claims data identified that the mean LOS for the their hospitalization. Therefore, complications such as
operative treatment of ankle fractures was 4.6 days.15 Another wound complications, the majority of which occur after hos-
analysis of over 600 patients treated between 2004 and 2010 pitalization, are not captured in this study.
at 1 academic medical center cited LOSs for patients under- The high usage of postacute facilities in the Medicare
going ankle fractures ranging from 3 to 12 days depending on population is widely documented with estimates citing that
ASA status.6 This range in LOSs among ankle fractures is 42% of the Medicare population receives care in the postacute
consistent with the range seen in our analysis. McDonald care setting.16 Analysis of the 2008 Medicare claims data has
et al6 also demonstrated that ASA status is a powerful pre- demonstrated that 59% of patients admitted for ankle frac-
dictor of LOS for patients who underwent surgical fixation of tures were admitted to nursing facilities after hospitalization.
their ankle fracture citing that each increase in ASA corre- Notably, these patients had higher Elixhauser and Deyo–
sponded to a 3.42-day increase in LOS. Although this study Charlson comorbidity scores.17 This high utilization of
does include all adult patients, it does suggest that physical postacute care in geriatric ankle fracture patients has been
status before admission plays a role in LOS after surgical demonstrated by other groups with Anderson et al citing that
TABLE 2. Hospital Quality Measures and Discharge Disposition of Minimal-Risk, Low-Risk, Moderate-Risk, and High-Risk Cohorts
Minimal Risk Low Risk Moderate Risk High Risk Total Cohort
(n = 25) (n = 14) (n = 9) (n = 2) (n = 50) P
LOS (d) (mean 6 SD) 5.4 6 2.9 9.4 6 5.1 11.6 6 7.4 10.0 6 5.7 7.8 6 5.2 0.005
Complications (mean 6 SD) 0.2 6 0.6 0.4 6 0.6 0.1 6 0.3 0.0 6 0.0 0.2 6 0.6 0.457
Patients who required ICU/SDU care, n (%) 0 (0%) 2 (14.3%) 1 (11.1%) 1 (50.0%) 4 (8.0%) 0.049
Patients discharged home, n (%) 18 (72.0%) 6 (42.9%) 1 (11.1%) 0 (0.0%) 25 (50.0%) 0.006
FIGURE 1. Index admission costs of care of the minimal-risk, low-risk, moderate-risk, and high-risk cohorts. All costs have been
referenced to the mean total cost of care for total cohort of patients, which is now equivalent to 1. Editor’s Note: A color image
accompanies the online version of this article.
64% of patients older than 65 years who underwent operative Identifying patients who are at risk of using postacute
treatment of their ankle fractures required nursing home care is useful for multiple reasons. First, identifying patients
placement. Furthermore, these patients who required place- that are likely to need postacute care as early as possible is
ment had significantly greater rates of postoperative medical helpful to ensure early referrals to facilities and reduce costly
and surgical complications.18 This high utilization of post- inpatient LOS. Second, with the current optional bundled
acute care facilities was also seen in our cohort. payment systems as proposed by CMS, hospitals are
FIGURE 2. Cost of care of each patient as a percentage of total cost of the cohort. Patients are listed by their cohort (0: minimal, 1:
low, 2: moderate, and 3: high) followed by a number designating in which number the patients are in the cohort; example: the
minimal-risk group contains patients 0/1–0/25. Editor’s Note: A color image accompanies the online version of this article.
Lott et al J Orthop Trauma Volume 33, Number 6, June 2019
responsible for costs of care through 90 days after discharge insurers on value-based care. Therefore, it is important for
including readmissions and postacute care costs. Therefore, orthopaedic surgeons, particularly those who treat orthopae-
hospitals have been developing strategies to reduce utilization dic trauma injuries, to better understand the outcomes and
of high-cost postacute care services. Identifying which costs of commonly treated fractures. It will be imperative to
patients are at risk of high utilization of these high-cost identify ways to predict patients who are prone to fall outside
services makes it easier for hospitals to devote their resources of the planned reimbursement so as to devote more resources
to the most at-risk patients. Third, it has been demonstrated to these patients and to create programs aimed at improving
that patients that are discharged to postacute care facilities outcomes while decreasing costs. The authors feel that the
have more complications and increased readmission trauma triage tool used in this study is one that hospitals can
rates.19,20 This bodes poorly for patients due to the negative use to prepare for the changing medical economic landscape.
consequences on overall outcome; however, it also bodes Strengths of this study include the use of direct variable
poorly for hospitals if, in the future, ankle fractures become cost data because many similar studies are limited by their use
incorporated into a bundled payment system because these of hospital charge data or LOS as a proxy for cost. By using
poor outcomes will ultimately decrease the overall patient direct cost data, we were also able to identify areas of high
contribution margin realized from the bundled payment. utilization, which is essential when developing plans to
The direct variable costs associated with the fixation of deliver more cost-effective care. Weaknesses of the study
ankle fractures are not widely reported. McDonald et al’s include the limited number of patients included in this study.
analysis of operative ankle fractures cited that the cost of Although we did include patients over a 2-year period, given
treatment in this patient population ranged from $12,000 to our exclusion criteria so as to study a uniform population, the
$54,000 depending on the ASA physical status of the total number of patients was limited. As such, the low number
patient.6 However, these costs were indirectly calculated costs of complications observed in the study was not large enough
of admission using an average per-day inpatient cost for that to ascertain differences among the risk groups. Limitations to
particular hospital system. A larger analysis of over 15,000 the study include the fact that the cost data may not be
Medicare patients who underwent operative fixation of ankle applicable to all populations because this study took place at
fractures cited a cost for operative management of $8798.21 an academic medical center. Second, socioeconomic factors
However, these costs were determined using the Medicare were not controlled for in this analysis. Further analysis of
reimbursement data that are not equivalent to the hospital these variables may prove that they are confounding variables
direct costs, which are used in this study. In our analysis, in this analysis.
which obtained cost data directly from a validated cost
accounting system, room and board, and procedure costs
comprised 75% of the index admission costs, which is con-
sistent with the cost data reported for the treatment of other CONCLUSIONS
lower-extremity trauma injuries.22 There were also significant In this study, the STTGMA score is able to risk-stratify
differences seen in several subdivisions of cost including operative ankle fracture patients for cost of care and is
room/board, allied health, laboratory/pathology, and radiol- therefore ideally suited to aid hospitals in their efforts to
ogy costs. This increase in room/board costs in the higher- provide more cost-effective care.
risk cohorts is most likely due to the longer LOSs seen in the
higher-risk cohort groups in addition to the increase in need
REFERENCES
for advanced level of care. Similarly, longer LOSs will con- 1. Sporer SM, Weinstein JN, Koval KJ. The geographic incidence and
tribute to greater allied health costs because the patient will treatment variation of common fractures of elderly patients. J Am Acad
receive more physical and occupational therapy sessions. As Orthop Surg. 2006;14:246–255.
expected, higher-risk patients with higher STTGMA scores 2. Court-Brown CM, McBirnie J, Wilson G. Adult ankle fractures—an
will have more comorbidities and therefore greater pharmacy increasing problem? Acta Orthop Scand. 1998;69:43–47.
3. Kannus P, Palvanen M, Niemi S, et al. Increasing number and incidence
need. They may also have other injuries such as head or chest of low-trauma ankle fractures in elderly people: Finnish statistics during
injuries that require additional imaging. As demonstrated in 1970–2000 and projections for the future. Bone. 2002;31:430–433.
Figure 2, there was an overall trend of higher costs of care in 4. Koval KJ, Lurie J, Zhou W, et al. Ankle fractures in the elderly: what you
the more high-risk STTGMA groups with one patient not get depends on where you live and who you see. J Orthop Trauma. 2005;
19:635–639.
solely responsible for the increased cost of care in the more 5. Belatti DA, Phisitkul P. Economic burden of foot and ankle surgery in
high-risk groups. the US Medicare population. Foot Ankle Int. 2014;35:334–340.
The trend of replacing the fee-for-service model to the 6. McDonald MR, Sathiyakumar V, Apfeld JC, et al. Predictive factors of
bundled payment model is only increasing. Orthopaedic hospital length of stay in patients with operatively treated ankle fractures.
trauma care has become increasingly incorporated into the J Orthop Traumatol. 2014;15:255–258.
7. Regan DK, Manoli A III, Hutzler L, et al. Impact of diabetes mellitus on
bundled payment model with the mandatory inclusion of surgical quality measures after ankle fracture surgery: implications for
certain hip fractures into the Comprehensive Joint Replace- “Value-Based” compensation and “pay for performance.” J Orthop
ment bundle. In addition, starting from October 1, 2018, CMS Trauma. 2015;29:e483–e486.
is allowing for voluntary enrollment into a bundled payment 8. Koval KJ, Zhou W, Sparks MJ, et al. Complications after ankle fracture
in elderly patients. Foot Ankle Int. 2007;28:1249–1255.
program for all operatively repaired hip and femur fractures. 9. Shivarathre DG, Chandran P, Platt SR. Operative fixation of unstable
Although the nature of these bundled payment programs may ankle fractures in patients aged over 80 years. Foot Ankle Int. 2011;32:
shift with the political climate, there is a clear emphasis from 599–602.
10. Konda SR, Seymour R, Manoli A, et al. Development of a middle-age 16. Medicare Payment Advisory Committee. Report to the Congress:
and geriatric trauma mortality risk score a tool to guide palliative care Medicare and the Health Care Delivery System. 2016. Available at:
consultations. Bull Hosp Jt Dis (2013). 2016;74:298–305. http://medpac.gov/docs/default‐source/reports/june‐2016‐report‐to‐the‐
11. Konda SR, Lott A, Selah H, et al. How does frailty factor into mortality congress‐medicare‐and‐the‐health‐care‐delivery‐system.pdf. Accessed
risk-assessment of a middle-aged and geriatric trauma population? Ger- December 10, 2018.
iatr Orthop Surg Rehabil. 2017;8:225–230. 17. Kadakia RJ, Hsu RY, Hayda R, et al. Evaluation of one-year mortality
12. Earl-Royal E, Kaufman EJ, Hsu JY, et al. Age and preexisting conditions after geriatric ankle fracture in patients admitted to nursing homes.
as risk factors for severe adverse events and failure to rescue after injury. Injury. 2015;46:2010–2015.
J Surg Res. 2016;205:368–377. 18. Anderson SA, Li X, Franklin P, et al. Ankle fractures in the elderly:
13. Maxwell CA, Mion LC, Mukherjee K, et al. Preinjury physical frailty initial and long-term outcomes. Foot Ankle Int. 2008;29:1184–1188.
and cognitive impairment among geriatric trauma patients determine 19. Dailey EA, Cizik A, Kasten J, et al. Risk factors for readmission of ortho-
postinjury functional recovery and survival. J Trauma Acute Care Surg. paedic surgical patients. J Bone Joint Surg Am. 2013;95:1012–1019.
2016;80:195–203. 20. Pollock FH, Bethea A, Samanta D, et al. Readmission within 30 days of
14. Hwabejire JO, Kaafarani HM, Lee J, et al. Patterns of injury, outcomes, discharge after hip fracture care. Orthopedics. 2015;38:e7–e13.
and predictors of in-hospital and 1-year mortality in nonagenarian and 21. Kadakia RJ, Ahearn BM, Tenenbaum S, et al. Costs associated with
centenarian trauma patients. JAMA Surg. 2014;149:1054–1059. geriatric ankle fractures. Foot Ankle Spec. 2017;10:26–30.
15. Bariteau JT, Hsu RY, Mor V, et al. Operative versus nonoperative treat- 22. Kleweno CP, O’Toole RV, Ballreich J, et al. Does fracture care make
ment of geriatric ankle fractures: a Medicare part A claims database money for the hospital? An analysis of hospital revenues and costs for
analysis. Foot Ankle Int. 2015;36:648–655. treatment of common fractures. J Orthop Trauma. 2015;29:e219–e224.
ORIGINAL ARTICLE
Effect of Mental Health Conditions on Complications,

Revision Rates, and Readmission Rates Following Femoral
Shaft, Tibial Shaft, and Pilon Fracture
Brock D. Foster, MD, Hyunwoo Paco Kang, MD, Zorica Buser, PhD, John Carney, BS,
Jeffrey C. Wang, MD, and Geoffrey S. Marecek, MD
Conclusions: Comorbid MH conditions are associated with higher

Objectives: To evaluate the effect of pre-existing mental health postoperative complication, readmission, and revision surgery rates
(MH) conditions on 90-day complication, 90-day readmission, and for treated femoral, tibial, and pilon fractures.
all-time revision surgical intervention rates following femoral, tibial,
and pilon fractures. Key Words: femur fracture, tibia fracture, pilon fracture, mental
health, complications, revision, readmission
Design: Data were collected using a commercially available
database software for which Current Procedural Terminology codes Level of Evidence: Prognostic Level III. See Instructions for
were used to identify patients who underwent surgical treatment of Authors for a complete description of levels of evidence.
tibial, femoral, or pilon fractures. These patients were then subdi- (J Orthop Trauma 2019;33:e210–e214)
vided into those with and without pre-existing MH condition using
International Classification of Disease, Ninth Edition codes. Ninety- INTRODUCTION
day postoperative complications, revision surgery, and 90-day read- Mental health (MH) conditions are increasingly recog-
mission rates were then compared between those with and without nized as predictive of objective and subjective outcomes.1–13
MH conditions. The association has been evaluated among patients undergo-
ing elective orthopaedic interventions, including joint replace-
Setting: National databases of 70 million combined patients from
ment and spine surgery.1,2,7–9,11 Moreover, MH disorders,
2007 to 2015.
including depression and posttraumatic stress disorder
Patients/Participants: Humana and Medicare insured patients. (PTSD), commonly develop in the setting of traumatic inju-
ries.12,14–22 Depression rates have been reported to be as high
Intervention: Surgical treatment of tibial, femoral, and pilon as 42% in general trauma populations.16 By comparison,
fractures. inpatient medical populations have a depression rate of
Main Outcome Measurements: Ninety-day readmission, 90- 34%, outpatient medical populations of 2%–12%, and cardiac
day complications, and all-time revision surgical intervention. disease population of 32%.23,24 In the Lower Extremity
Assessment Project (LEAP) population, the incidence of
Results: The total number of patients for femoral, tibial, and pilon moderate-to-severe depression was 41.8%, and the incidence
treatment, respectively, included 6207, 6253, and 5940 without of severe depression was 15.3%.16 Thus, trauma patients are
MH conditions and 4879, 5247, and 2911 with MH conditions. at risk of developing postinjury depression, anxiety, and
Femoral, tibial, and pilon readmission rates, revision rates, and PTSD, which has been associated with poorer outcomes.
complication rates were significantly higher among patients with Complications, hospital readmission, and need for revi-
MH disorders in matched cohorts after controlling for medical sion surgery impact physician and patient satisfaction as well
comorbidities (P # 0.05 for all). as reimbursement. Few studies have evaluated the effect of
comorbid MH conditions on revision surgery, postoperative
complications, and hospital readmission.1–7,11,12,17,25 Fewer
Accepted for publication December 28, 2018.
From the Department of Orthopaedic Surgery, Keck School of Medicine of studies have assessed these outcomes among trauma patients.5,12
USC, Los Angeles, CA. As such, the purpose of this study is to assess the impact of MH
The authors report no conflict of interest. conditions on rates of revision surgery, complications, and hos-
Presented in part at the Annual Meeting of the American Academy of pital readmission in orthopaedic trauma interventions when
Orthopaedic Surgeons, March 7, 2018, New Orleans, LA.
compared to a population without MH conditions.
appear in the printed text and are provided in the HTML and PDF
versions of this article on the journal’s Web site (www.jorthotrauma.
com). METHODS
Reprints: Brock D. Foster, MD, Department of Orthopaedic Surgery, Data were collected using the commercially available
Vanderbilt University Medical Center, 1215 21st Avenue South, Suite
4200, Nashville, TN 37232 (e-mail: brockdfoster@gmail.com).
PearlDiver software [national database of Humana and
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. Medicare insured patients, PearlDiver (www.pearldiverinc.
DOI: 10.1097/BOT.0000000000001438 com) Colorado Springs, CO] for which Current Procedural
e210 | www.jorthotrauma.com J Orthop Trauma Volume 33, Number 6, June 2019
J Orthop Trauma Volume 33, Number 6, June 2019 Effect of Mental Health Conditions
exact-match basis for CCI and gender; age was matched by

TABLE 1. Demographic Information Among Patients With
5-year groupings (eg, 65–69).
and Without MH Disorders for All Fracture Surgery Types,
Before Matching Revision surgery, 90-day postoperative complications,
and 90-day readmission rates were then compared between
All Fractures
those with and without MH conditions. Revision surgery was
Control MH based on CPT codes for revision surgery for femoral and
Age tibial fractures as well as fusion for pilon fractures (see
64 and under 6388 (34.7%) 5027 (38.6%) Appendix, Supplemental Digital Content 1, http://links.
65–69 3057 (16.6%) 1653 (12.7%) lww.com/JOT/A656). Postoperative complications were ob-
70–74 2342 (12.7%) 1659 (12.7%) tained using ICD-9 codes and included pneumonia, myocar-
75–79 1879 (10.2%) 1444 (11.1%) dial infarction, pulmonary embolism, transfusion, infection,
80–84 1734 (9.4%) 1284 (9.8%) wound complications, transfusion, deep vein thrombosis,
85 and over 2933 (15.9%) 1912 (14.7%) stroke, mechanical complication of implant, and infection of
Unknown 67 (0.4%) 58 (0.4%) orthopaedic implant. Complication, revision, and readmission
Region rates were then calculated and compared using odds
Midwest 4769 (25.9%) 3249 (24.9%) ratios.26,27 All analyses were computed using Microsoft Excel
Northeast 1916 (10.4%) 1363 (10.5%) (Microsoft Corp, Redmond, WA).
South 9147 (49.7%) 6919 (53.1%) Of note, as a mechanism to ensure patient anonymity, if the
West 2566 (13.9%) 1506 (11.6%) number of patients in a group is 10 or less, the database is unable
Unknown 2 (0.0%) 0 (0.0%) to provide the exact number of patients in the group. In these
Gender cases, the data cannot be examined and are reported as N/A.
Female 10,070 (54.7%) 9352 (71.7%)
Male 8263 (44.9%) 3627 (27.8%)
Unknown 67 (0.4%) 58 (0.4%) RESULTS
Total 18,400 13,037 The total number of patients with treated femoral, tibial,
and pilon fractures were 6207, 6253, and 5940 without MH
conditions, and 4879, 5247, and 2911 with MH conditions,
respectively (Table 1, see Appendix, Supplemental Digital
Terminology (CPT) codes were used to identify patients who Content 2, http://links.lww.com/JOT/A657).
underwent treatment of tibial, femoral, or pilon fractures In matched cohorts, after controlling for medical
between 2007 and 2015 (see Appendix, Supplemental comorbidities, age and gender, 90-day all-cause readmission
Digital Content 1, http://links.lww.com/JOT/A656). These rates for femoral, tibial, and pilon fractures were 33.9%,
patients were then subdivided into those with and without 18.0%, and 17.1% without MH conditions and 42.6%, 24.9%,
pre-existing MH conditions using International Classification and 23.3% with MH conditions, respectively (P , 0.05,
of Disease, ninth Edition (ICD-9) codes. Anxiety, depression, Table 2). Revision rates in matched cohorts for femoral, tibial,
chronic pain syndrome, fibromyalgia, bipolar disorder, PTSD, and pilon fractures, respectively, were 1.4%, 1.4%, and 1.3%
schizophrenia, and psychosis were included in the analysis. for those without MH conditions and 2.0%, 2.5%, and 2.2%
After identification of the study populations, 2 matching for those with MH conditions (P # 0.05, Table 2). There
groups were created based on medical comorbidities as mea- were no significant differences in time between index sur-
sured by the Charlson Comorbidity Index (CCI), age, and gery and revision for any fracture type between MH and
gender, to reduce confounding by these factors. Matching control groups (see Appendix, Supplemental Digital
was performed by the PearlDiver software system on a 1:1, Content 3, http://links.lww.com/JOT/A658). Rates of several
TABLE 2. Ninety-Day All-Cause Readmission and All-Time Revision Surgery Among Patients With and Without MH Disorders for
All Fracture Types
Without MH Condition With MH Condition
(% of Cases) (% of Cases) Odds Ratio P
Femur (n = 7662)
90-day all-cause hospital readmission 33.9 42.6 1.5 0.047
Surgical revision rate 1.4 2.0 1.4 0.052
Without MH With MH Without MH With MH
Condition Condition Odds Condition Condition Odds
(% of Cases) (% of Cases) Ratio P (% of Cases) (% of Cases) Ratio P
Tibia (n = 7820) Pilon (n = 4596)
90-day all-cause hospital 18 24.9 1.5 ,0.001 17.1 23.3 1.5 ,0.001
readmission
Surgical revision rate 1.4 2.5 1.8 ,0.001 1.3 2.2 1.6 0.034
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. www.jorthotrauma.com | e211
Foster et al J Orthop Trauma Volume 33, Number 6, June 2019
TABLE 3. Ninety-Day Postoperative Complication Rates Among Patients With and Without MH Disorders Based on the Fracture
Type (Femur, Tibia, and Pilon)
Without MH Condition With MH Condition
90-Day Postoperative (% of Cases) (% of Cases) Odds Ratio P
Complications Femur (n = 7662)
Pneumonia 8.4 10.3 1.2 0.005
Myocardial infarction 2.2 2.9 1.3 0.043
Pulmonary embolism 2.1 3.0 1.5 0.009
Transfusion 14.3 19.2 1.4 ,0.001
Deep vein thrombosis 6.4 7.4 1.2 0.080
Stroke 3.4 3.2 0.9 0.607
Infection/sepsis 2.3 3.9 1.7 ,0.001
Wound complication 2.2 2.2 1.1 0.476
Mechanical complication of implant 10.6 16.2 1.6 ,0.001
Infection of orthopaedic implant 1.0 2.3 2.4 ,0.001
Without MH With MH Without MH With MH
Condition Condition Odds Condition Condition Odds
90-Day Postoperative (% of Cases) (% of Cases) Ratio P (% of Cases) (% of Cases) Ratio P
Complications Tibia (n = 7820) Pilon (n = 4596)
Pneumonia 4.6 6.4 1.4 0.001 2.7 3.9 1.5 0.021
Myocardial infarction 1.2 1.1 0.9 0.525 N/A N/A N/A N/A
Pulmonary embolism 1.2 1.2 1.0 0.836 N/A N/A N/A N/A
Transfusion 3.6 5.1 1.5 0.001 2.4 3.6 1.5 0.016
Deep vein thrombosis 3.3 3.9 1.2 0.164 1.8 2.2 1.2 0.293
Stroke 1.9 2.1 1.1 0.520 N/A N/A N/A N/A
Infection/sepsis 2.4 3.3 1.4 0.018 2.9 4.0 1.4 0.030
Wound complication 3.1 3.4 1.1 0.372 2.7 4.8 1.8 ,0.001
Mechanical complication of 1.8 2.8 1.6 0.004 2.6 3.4 1.4 0.085
implant
Infection of orthopaedic N/A 0.6 N/A N/A N/A N/A N/A N/A
implant
When the number of patients is less than 11, the database is unable to provide the exact number of cases. As such, the term “N/A” is used because these data are unable to be
accurately analyzed.
orthopaedic and nonorthopaedic complications were higher periprosthetic infection, periprosthetic fracture, dislocation, and
among those with MH conditions (Table 3 and Figs. 1A–C). revision rates.
Trauma patients represent a group with a baseline
increased risk of complication, readmission, and revision
DISCUSSION rates.4,10,12,13,21 As such, identification and modification of
MH conditions, including depression, anxiety, PTSD, known risk factors may improve outcomes and reduce mon-
chronic pain syndrome, fibromyalgia, bipolar disorder, and etary impediments to trauma care. Few studies have evaluated
schizophrenia, were found to be associated with increased the effect of preinjury MH disorders on outcomes in an ortho-
readmission, complications, and revision surgery rates among paedic trauma population.5 Furthermore, these patients are
patients with treated tibia, femur, and pilon fractures. These likely at higher risk of these complications, as they are prone
data would indicate that MH disorders are associated with to develop depression, PTSD, and anxiety following the
worse outcomes among common fractures encountered by injury.14–20,22,29 Menendez et al5 described the influence of
orthopaedic surgeons. psychiatric illness on inpatient complications and discharge in
Several prior studies have evaluated the effect of a population of lower-extremity fracture patients. They found
depression on similar metrics in joint replacement, chronic that depression and anxiety were associated with shorter hos-
obstructive pulmonary disease, and spine surgery.2,3,7–11 For pital stay, fewer in hospital adverse events, and no difference
example, 5-year survival rates for unicondylar knee arthro- in death. Depression was also associated with increased rate
plasty was found to be negatively influenced by the presence of transfusion. Similarly, Vallier et al found that in a popula-
of depression along with complicated diabetes, low volume tion of polytrauma patients, orthopaedic surgeons were less
hospitals, obesity, and age .74.8 Similarly, Klement et al28 likely to prescribe inpatient psychiatric medications and pro-
found in a population of total hip arthroplasty patients, psychi- vide follow-up psychiatric instructions.12 Furthermore,
atric illness is associated with increased 90-day complication, depression was an independent predictor of increased
e212 | www.jorthotrauma.com Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.
J Orthop Trauma Volume 33, Number 6, June 2019 Effect of Mental Health Conditions
FIGURE 1. A, Forest plot of odds ratios for

90-day complications in femur fractures. B,
Forest plot of odds ratios for 90-day complica-
tions in tibia fractures. C, Forest plot of odds
ratios for 90-day complications in pilon frac-
tures. DVT, deep vein thrombosis; Mech
Comp, mechanical complication; MI, myocar-
dial infarction; PE, pulmonary embolism; PNA,
pneumonia; Readm, readmission; Wound
Comp, wound complication.
complications (odds ratio 2.96). The authors concluded that how this affects short- and long-term outcomes. Although this
greater awareness of MH disorders is needed in trauma study examines comorbid MH conditions at the time of
populations. intervention for tibia, femur, or pilon fractures, it is possible
Substantial evidence exists describing the increased those who develop MH symptoms following a traumatic
incidence of MH conditions after trauma, but little is known event would follow a similar trajectory as those in this study.
Foster et al J Orthop Trauma Volume 33, Number 6, June 2019
Perhaps, such outcomes can be averted with appropriate MH procedures in the UK National Health Service. JAMA Surg. 2017;152:
assessments and treatment. e173949.
8. Jeschke E, Gehrke T, Günster C, et al. Five-year survival of 20,946
This study has several weaknesses. First, this is unicondylar knee replacements and patient risk factors for failure: an
a retrospective database review that relies on accurate coding analysis of German insurance data. J Bone Joint Surg Am. 2016;98:
for the injury, treatment, and associated conditions. However, 1691–1698.
Humana conducts patient record audits, provider payment 9. Mollon B, Mahure SA, Ding DY, et al. The influence of a history of
clinical depression on peri-operative outcomes in elective total shoulder
integrity audits, and on-site audits to ensure claim accuracy. arthroplasty: a ten-year national analysis. Bone Joint J. 2016;98-B:818–
Humana is also required to contract with an independent third 824.
party to conduct validity and reliability audits on an annual 10. Knutsen EJ, Paryavi E, Castillo RC, et al. Is satisfaction among ortho-
basis. Medicare claims are also subject to validation and paedic trauma patients predicted by depression and activation levels? J
review under Centers for Medicare and Medicaid Services. Orthop Trauma. 2015;29:e183–187.
11. Ricciardi BF, Oi KK, Daines SB, et al. Patient and perioperative varia-
The validation process requires hospital claims match both bles affecting 30-day readmission for surgical complications after hip and
physician’s description and the beneficiary’s medical record. knee arthroplasties: a matched cohort study. J Arthroplasty. 2017;32:
This is further supported by a Veterans Affairs study that 1074–1079.
found PTSD diagnoses were well corroborated by the medical 12. Weinberg DS, Narayanan AS, Boden KA, et al. Psychiatric illness is
common among patients with orthopaedic polytrauma and is linked with
record.30 Second, this study may not be applicable to other poor outcomes. J Bone Joint Surg Am. 2016;98:341–348.
patient populations, for example, Medicaid and uninsured 13. Yeoh JC, Pike JM, Slobogean GP, et al. Role of depression in outcomes
patients, or other diagnoses. As this study included femoral of low-energy distal radius fractures in patients older than 55 years. J
shaft, pilon, and tibial shaft fractures among those insured Orthop Trauma. 2016;30:228–233.
with Humana or Medicare coverage policies, other diagnoses 14. Crichlow RJ, Andres PL, Morrison SM, et al. Depression in orthopaedic
trauma patients. Prevalence and severity. J Bone Joint Surg Am. 2006;88:
or insurance carriers (or lack thereof) may not have display 1927–1933.
similar results with regard to MH conditions. Third, the data- 15. Nota SP, Bot AG, Ring D, et al. Disability and depression after ortho-
base identifies all-cause readmission, not just orthopedic- paedic trauma. Injury. 2015;46:207–212.
related readmissions. Specifically, in patients with known 16. Butcher JL, MacKenzie EJ, Cushing B, et al. Long-term outcomes after
lower extremity trauma. J Trauma. 1996;41:4–9.
MH conditions, it is very possible that subsequent readmis- 17. Sutherland AG, Suttie S, Alexander DA, et al. The mind continues to
sions may be primarily due to MH-related issues as opposed matter: psychologic and physical recovery 5 years after musculoskeletal
to orthopaedic ones. Fourth, severity of injury cannot be con- trauma. J Orthop Trauma. 2011;25:228–232.
trolled. As such, if patients with MH conditions suffer higher 18. Sanders MB, Starr AJ, Frawley WH, et al. Posttraumatic stress symptoms
energy, more complicated injuries, this association cannot be in children recovering from minor orthopaedic injury and treatment. J
Orthop Trauma. 2005;19:623–628.
identified and may lead to bias. Finally, the database is unable 19. Muscatelli S, Spurr H, OʼHara NN, et al. Prevalence of depression and
to identify number of cases below 10, and these groups were posttraumatic stress disorder after acute orthopaedic trauma: a systematic
not assessed. review and meta-analysis. J Orthop Trauma. 2017;31:47–55.
In conclusion, MH conditions were found to be associ- 20. McCarthy ML, MacKenzie EJ, Edwin D, et al. Psychological distress
associated with severe lower-limb injury. J Bone Joint Surg Am. 2003;
ated with increased readmission, complication, and revision 85-A:1689–1697.
surgery rates among patients with tibia, femur, and pilon 21. Archer KR, Abraham CM, Obremskey WT. Psychosocial factors predict
fractures after matching for CCI, age, and gender. Proper pain and physical health after lower extremity trauma. Clin Orthop Relat
screening is recommended to identify this patient population Res. 2015;473:3519–3526.
given the above data. Higher quality studies are needed to 22. Starr AJ, Smith WR, Frawley WH, et al. Symptoms of posttraumatic
stress disorder after orthopaedic trauma. J Bone Joint Surg Am. 2004;
assess the validity and applicability of these data. 86-A:1115–1121.
23. Kessler RC, Berglund P, Demler O, et al. The epidemiology of major
depressive disorder: results from the National Comorbidity Survey Rep-
REFERENCES lication (NCS-R). JAMA. 2003;289:3095–3105.
1. Adogwa O, Elsamadicy AA, Mehta AI, et al. Association between base- 24. Oliver JM, Simmons ME. Affective disorders and depression as mea-
line affective disorders and 30-day readmission rates in patients under- sured by the diagnostic interview schedule and the Beck Depression
going elective spine surgery. World Neurosurg. 2016;94:432–436. Inventory in an unselected adult population. J Clin Psychol. 1985;41:
2. Gold HT, Slover JD, Joo L, et al. Association of depression with 90-day 469–477.
hospital readmission after total joint arthroplasty. J Arthroplasty. 2016; 25. Sarmiento A, Sharpe FE, Ebramzadeh E, et al. Factors influencing the
31:2385–2388. outcome of closed tibial fractures treated with functional bracing. Clin
3. Singh G, Zhang W, Kuo YF, et al. Association of psychological disor- Orthop Relat Res. 1995:8–24.
ders with 30-day readmission rates in patients with COPD. Chest. 2016; 26. Bland JM, Altman DG. Statistics notes. The odds ratio. BMJ. 2000;320:
149:905–915. 1468.
4. Berges IM, Amr S, Abraham DS, et al. Associations between depressive 27. Altman DG, Bland JM. How to obtain the P value from a confidence
symptoms and 30-day hospital readmission among older adults. J interval. BMJ. 2011;343:d2304.
Depress Anxiety. 2015;4:185. 28. Klement MR, Bala A, Blizzard DJ, et al. Should we think twice about
5. Menendez ME, Neuhaus V, Bot AGJ, et al. Do psychiatric comorbidities psychiatric disease in total hip arthroplasty? J Arthroplasty. 2016;31:
influence inpatient death, adverse events, and discharge after lower 221–226.
extremity fractures? Clin Orthop. 2013;471:3336–3348. 29. Starr AJ. Fracture repair: successful advances, persistent problems, and
6. Roh YH, Lee BK, Noh JH, et al. Effect of anxiety and catastrophic pain the psychological burden of trauma. J Bone Joint Surg Am. 2008;
ideation on early recovery after surgery for distal radius fractures. J Hand 90(suppl 1):132–137.
Surg Am. 2014;39:2258–e2. 30. McCarron KK, Reinhard MJ, Bloeser KJ, et al. PTSD diagnoses among
7. Ali AM, Loeffler MD, Aylin P, et al. Factors associated with 30-day Iraq and Afghanistan veterans: comparison of administrative data to chart
readmission after primary total hip arthroplasty: analysis of 514 455 review. J Trauma Stress. 2014;27:626–629.
ORIGINAL ARTICLE
Fewer Hospitals Provide Operative Fracture Care to

Medicaid Patients Than Otherwise-Insured Patients in 4
Large States
Dane J. Brodke, MD, MPH* and Saam Morshed, MD, PhD, MPH†
Conclusions: Fewer hospitals provide operative fracture care to

Objective: To determine whether Medicaid patients receive Medicaid patients than otherwise-insured patients, but Medicaid
operative fracture care at an equal number of hospitals as patients do not travel longer distances to the hospital on a population
otherwise-insured patients and to compare travel distances between level.
Medicaid and otherwise-insured patients.
Key Words: Medicaid, health services research, access disparities,
Design: Retrospective, population-based cohort study of adminis- population health
trative health data.
Level of Evidence: Prognostic Level III. See Instructions for
Setting: One thousand seventy-five hospitals in California, Florida, Authors for a complete description of levels of evidence.
New York, and Texas.
(J Orthop Trauma 2019;33:e215–e222)
Participants: Two hundred forty thousand three hundred seventy-
six patients who underwent open reduction and internal fixation of INTRODUCTION
a fracture of the radius/ulna, tibia/fibula, or humerus between 2006 Over the past few decades, the Medicaid program has
and 2010 in Texas or New York, or between 2010 and 2014 in grown substantially in both size and importance to the US
California or Florida. health care system. Established by the 1965 Social Security
Act, the program has grown from a 1980 enrollment of 20
Intervention: Open reduction and internal fixation of the radius/ million (9% of the population) to a 2018 enrollment of 74
ulna, tibia/fibula, or humerus.
million (23% of the population) and has become a key source
Main Outcome Measurements: The number of unique hospi- of financing for safety-net health systems.1–3 The Medicaid
tals visited and the distance traveled for care were compared by expansion in 31 states enabled by the Patient Protection and
payer status and admission acuity. The distance traveled was also Affordable Care Act increased enrollment by 16 million,
stratified by urban versus rural geographic area. reducing the uninsured population and increasing access to
primary and specialty care.2,4,5 Now that Medicaid covers
Results: In nonemergent settings, 7%–16% fewer hospitals saw almost one-quarter of the US population, one challenge that
Medicaid patients than otherwise-insured patients. In emergent set- remains is ensuring that Medicaid is a sustainable component
tings, the gap between the number of hospitals seeing Medicaid and of the US health care system going forward. As Medicaid
otherwise-insured patients was less than 5% in every state except matures from a program for restricted population groups into
Texas, where the gap was 11%–14%. The Medicaid and Medicare one of America’s most significant insurers, it becomes
groups had longer travel distances in the nonemergent setting than in increasingly important to consider the access to care it pro-
the emergent setting. Medicaid patients did not travel longer distan- vides for beneficiaries in relation to other payers.
ces than otherwise-insured patients except in Texas, where they The American Academy of Orthopaedic Surgeons has
traveled 3–5 miles further than otherwise-insured patients in the affirmed the importance of universal access to specialty care,
nonemergent, urban setting. stating that “everyone within the United States should receive
access to health care coverage—including specialty care—
without financial barriers or undue burdens placed on the
Accepted for publication December 28, 2018. patient or physician.”6 Despite this, researchers have identified
From the *Department of Orthopaedic Surgery, University of California, Los disproportionate challenges faced by Medicaid patients in ac-
Angeles, Los Angeles, CA; and †Department of Orthopaedic Surgery, cessing orthopaedic surgical care including reduced appoint-
University of California, San Francisco, San Francisco, CA. ment availability and greater distances traveled for care.7–24
Presented as a poster at the Annual Meeting of the American Academy of It remains unanswered in the orthopaedic trauma
Orthopaedic Surgeons, March 2018, New Orleans, LA. population whether Medicaid patients have access to care
Reprints: Saam Morshed, MD, PhD, MPH, Department of Orthopaedic equal to that of otherwise-insured patients on a population
Surgery, University of California, San Francisco, 500 Parnassus Avenue, level. Building on previous work that focused on individual
MU 320-W, San Francisco, CA 94143 (e-mail: Saam.Morshed@ucsf.
edu).
states8 and institutions,7 this study examined 2 measures of
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. access to care across 4 large states constituting 33% of the US
DOI: 10.1097/BOT.0000000000001439 population. We hypothesized that fewer hospitals in these
J Orthop Trauma Volume 33, Number 6, June 2019 www.jorthotrauma.com | e215
Brodke and Morshed J Orthop Trauma Volume 33, Number 6, June 2019
states would be found to provide operative fracture care to admission type variable. In the California data, admission
Medicaid patients compared with otherwise-insured patients, type was coded as “scheduled,” which was categorized as
after adjusting for the size of the patient groups. We further nonemergent in this analysis, or “unscheduled,” which was
hypothesized that Medicaid patients traveled greater average categorized as emergent. New York, Florida, and Texas data
distances from their homes to the hospital to undergo fracture had 3 distinct admission type values: “emergent,” “urgent,”
repair surgery. Because previous work showed greater dispar- and “elective.” For this analysis, records coded as “emergent”
ities in travel distances for nonemergent procedures than for and “elective” were categorized as emergent and nonemer-
emergent procedures,7 we stratified our analysis by the acuity gent, respectively. As for the “urgent” designation, the US
of the admission for surgery (emergent or nonemergent) and Agency for Healthcare Research and Quality states that
predicted that access disparities would be larger in nonemer- “urgent” implies the patient was admitted to “the first avail-
gent settings. We also predicted that travel distance disparities able and suitable accommodation,” whereas “emergent” im-
would be most pronounced in rural areas, where the density plies “the patient was admitted through the emergency
of hospitals, especially safety-net hospitals, is lower than that room.”26 Because “urgent” hospitalization was not always
in urban areas. immediately after injury and, not originating from the emer-
gency department, could plausibly be denied on the basis of
insurance status, we grouped “urgent” admission records with
METHODS “elective” admissions in the nonemergent category.
Patients 65 years of age and older were excluded from
Design the Medicaid group to avoid including patients dually eligible
This study was a retrospective, population-based cohort for Medicaid and Medicare, as recommended by the US
study. Agency for Healthcare Research and Quality.27
Data Sources Outcomes
We used statewide administrative databases from the 4 The primary outcome was the total number of unique
most populous US states, California, Florida, New York, and hospitals visited by patients in each subgroup of Medicaid,
Texas, which together represent 38%25 of US Medicaid Medicare, or private insurance and nonemergent or emergent
enrollment. Data sources and time frames were selected based admission acuity. The secondary outcome was the straight-
on the availability and completeness of the data, including the line distance between the center of the patient’s home zip
following key variables: facility name or ID, International code and the hospital’s location.
Classification of Diseases, ninth revision (ICD-9) procedure
code, admission type, primary payer, and patient zip code. Exposures
We used Healthcare Cost and Utilization Project (HCUP) The exposures of interest in this study were payer
State Inpatient Databases for Florida (2010–2014) and New status as Medicaid, Medicare, or private coverage and
York (2006–2010). For Texas, we accessed the Texas Depart- admission acuity as either emergent or nonemergent. For
ment of State Health Services’ Inpatient Public Use Data Files the travel distance outcome, whether the patient lived in an
(2006–2010), and for California, the analysis was performed urban or rural area was an additional exposure of interest.
on the Office of Statewide Health Planning and Develop- This was determined by the patient’s home zip code. Zip
ment’s Patient Discharge Data (2010–2014). codes identified by the census bureau as overlapping an
urban area were designated urban.28 Other zip codes were
Cohort Inclusion designated as rural.
To examine the interaction of admission acuity and
payer status on measures of access to care, the study cohort Statistical Methods
included patients who underwent frequently billed long bone The number of unique hospitals visited was determined
fracture repair procedures that are associated with both for each subgroup of payer status and admission acuity in
emergent and nonemergent admissions. The population- each state. Because the cohort included fewer Medicaid
based study cohort included patients insured by Medicaid, patients than Medicare and privately insured patients, the
Medicare, or a private payer who underwent a procedure hospital counts for Medicare and private groups were adjusted
coded by one of the following ICD-9 procedure codes: 79.31 to account for their larger relative size. The group-size
[open reduction and internal fixation (ORIF), humerus], 79.32 adjustment involved drawing 1000 repeated random samples
(ORIF, radius or ulna), or 79.36 (ORIF, tibia or fibula). To with replacement from each Medicare and private coverage
limit the comparison with Medicaid to other major payers, group of the size of the corresponding Medicaid group.
patients whose insurance status was coded as “self-pay,” “no Within the 1000 samples drawn, the median sample hospital
charge,” “other,” or “missing” were excluded. Patients who count was used as the adjusted hospital count, and the sample
underwent ORIF of the femur (ICD-9 code 79.35) were hospital counts at the 2.5 and 97.5 percentiles were used as
excluded because these operations were almost exclusively bounds for the 95% confidence interval. In this way, we
associated with emergent admissions. To compare access to estimated how many hospitals Medicare and privately insured
care in the most acute situations to access in less acute sit- groups would have visited if they were as small as the Med-
uations, admission records were stratified into emergent and icaid groups. This nonparametric statistical boot-strapping
nonemergent categories. This was determined by the method had the advantage of not imposing any assumptions
J Orthop Trauma Volume 33, Number 6, June 2019 Medicaid Access to Operative Fracture Care
about the relationship between group size and number of Medicaid patients was interpreted to be significantly smaller
hospitals visited. Mean travel distances between payer sub- than the number seeing Medicare or privately insured patients
groups were compared with a linear regression of travel dis- if the Medicaid hospital counts were below the lower bound
tance on payer status, stratified by admission acuity and urban of the 95% confidence interval for the adjusted Medicare/
versus rural. All records in the database were included in the private hospital counts (Fig. 1). Fewer hospitals saw Medicaid
analysis of hospital visitation patterns. Records that did not patients than otherwise-insured patients in nonemergent set-
include a patient zip code were excluded from the travel tings across every state. In emergent settings, Medicaid pa-
distance analysis. Patient zip code data were missing in less tients were seen at fewer hospitals than otherwise-insured
than 1% of records. The analysis was performed with R Stu- patients in New York and Texas, but were seen by a similar
dio software (Boston, MA) and SAS software (Cary, NC). number of hospitals as otherwise-insured patients in Califor-
nia and Florida. In every state, the disparity in the number of
hospitals seeing Medicaid patients compared with hospitals
RESULTS seeing otherwise-insured patients was greater in nonemergent
A population-based sample of 240,376 hospital admis- settings than in emergent settings.
sions from the 4 most populous US states—California, Flor- In California, 8%–16% fewer hospitals saw Medicaid
ida, New York, and Texas—was included in the study (Table patients compared with otherwise-insured patients. In Florida,
1). This included 73,159 hospital admissions in California, 7% fewer hospitals saw Medicaid patients, whereas in New
80,854 admissions in Florida, 45,388 admissions in New York 8%–13% fewer hospitals saw Medicaid patients, and in
York, and 40,975 admissions in Texas. Across all states, Texas 14%–16% fewer hospitals saw Medicaid patients. Con-
the average ages of Medicaid, Medicare, and privately insured trary to the disparity in the nonemergent setting, the gap
patients were 33 years, 73 years, and 44 years, respectively. between the number of hospitals seeing Medicaid and
The cohort was 61% females, 69% white, 7.6% black, 17% otherwise-insured patients in the emergent setting was less
Hispanic, and 6.9% other or unknown race or ethnicity. Pa- than 5% in every state except Texas, where the gap was
tients requiring emergent ORIF procedures for radius/ulna, 11%–14%.
tibia/fibula, or humerus fractures were compared with patients Estimated travel distances for patients in the cohort
in the same state requiring similar procedures on a nonemer- were highly variable. Box-and-whisker plots demonstrated
gent basis (Table 2). The number of hospitals seeing that the median and 75th percentile travel distance was greater
TABLE 1. Demographic Characteristics of Study Cohort

Admission # Age, Mean Sex Race/Ethnicity
State Acuity Payer Patients 6 SD Female Male White Black Hispanic Other Unknown
California Nonemergent Medicaid 2436 37 6 16 45% 55% 34% 10% 48% 8% 0.3%
Medicare 6863 72 6 11 75% 25% 74% 4% 15% 7% 0.4%
Private 9314 45 6 17 52% 48% 67% 5% 19% 9% 0.6%
Emergent Medicaid 11,771 33 6 18 41% 59% 33% 11% 48% 7% 0.6%
Medicare 19,584 73 6 12 74% 26% 73% 4% 15% 7% 0.5%
Private 23,348 41 6 19 47% 53% 64% 4% 20% 11% 0.9%
Florida Nonemergent Medicaid 1948 34 6 18 51% 49% 55% 19% 21% 4% 0.5%
Medicare 13,087 73 6 11 78% 22% 87% 4% 7% 2% 0.7%
Private 9980 46 6 17 57% 43% 79% 7% 9% 3% 1.1%
Emergent Medicaid 6461 35 6 20 51% 49% 46% 22% 26% 5% 0.7%
Medicare 27,220 74 6 12 78% 22% 81% 5% 11% 2% 0.5%
Private 22,158 45 6 18 54% 46% 72% 9% 14% 4% 0.9%
New York Nonemergent Medicaid 1422 37 6 17 48% 52% 44% 18% 23% 15% 1.3%
Medicare 4244 72 6 11 76% 24% 81% 5% 6% 7% 1.8%
Private 6903 44 6 18 54% 46% 73% 7% 7% 10% 2.6%
Emergent Medicaid 5150 37 6 17 44% 56% 35% 23% 26% 15% 1.0%
Medicare 9958 74 6 12 76% 24% 81% 7% 6% 5% 0.8%
Private 17,711 44 6 18 51% 49% 70% 11% 8% 9% 1.3%
Texas Nonemergent Medicaid 1249 25–29* 54% 46% 33% 12% 50% 4% 0.6%
Medicare 6849 70–74* 79% 21% 73% 4% 18% 5% 0.4%
Private 9231 45–49* 57% 43% 70% 6% 17% 6% 0.5%
Emergent Medicaid 2572 18–19* 44% 56% 33% 13% 49% 5% 0.2%
Medicare 8364 70–74* 78% 22% 75% 5% 15% 4% 0.2%
Private 12,710 45–49* 53% 47% 68% 8% 19% 5% 0.2%
*Age in Texas data was recorded in ranges. Median age range is given.
TABLE 2. Number of Hospitals Visited and Distance Traveled to Hospital by Payer Status and Admission Acuity
Avg. Avg.
# (%) Distance to Distance to
# Hospitals 95% CI for Hospital in P for Hospital in P for
Hospitals Seeing Any Comparison Urban Comparison Rural Comparison
Admission in State Patient in With Areas, Miles With Areas, Miles With
State Acuity Payer Cohort This group* Medicaid (N) Medicaid (N) Medicaid
California Nonemergent Medicaid 343 224 (65%) — 13.6 (2379) — 40.2 (56) —
Medicare 267 (78%) 259 (76%)– 10.9 (6691) P , 0.001 44.0 (163) P = 0.69
275 (80%)
Private 245 (71%) 236 (69%)– 13.9 (9139) P = 0.53 42.8 (164) P = 0.79
255 (74%)
Emergent Medicaid 343 307 (90%) — 13.5 (11,441) — 40.0 (295) —
Medicare 305 (89%) 301 (88%)– 10.8 (19,083) P , 0.001 35.1 (472) P = 0.18
309 (90%)
Private 305 (89%) 300 (87%)– 20.9 (22,746) P , 0.001 39.3 (530) P = 0.84
309 (90%)
Florida Nonemergent Medicaid 206 166 (81%) — 12.7 (1872) — 28.6 (76) —
Medicare 180 (87%) 174 (84%)– 8.9 (12,782) P , 0.001 30.8 (305) P = 0.56
186 (90%)
Private 180 (87%) 174 (84%)– 12.9 (9737) P = 0.80 32.3 (243) P = 0.34
186 (90%)
Emergent Medicaid 206 189 (92%) — 8.9 (6317) — 30.2 (144) —
Medicare 191 (93%) 188 (91%)– 6.8 (26,815) P , 0.001 27.9 (405) P = 0.47
194 (94%)
Private 191 (93%) 189 (92%)– 11.7 (21,722) P , 0.001 27.8 (436) P = 0.45
194 (94%)
New Nonemergent Medicaid 190 138 (73%) — 6.7 (1355) — 24.4 (67) —
York
Medicare 159 (84%) 153 (81%)– 7.8 (3997) P = 0.01 28.2 (247) P = 0.34
165 (87%)
Private 150 (79%) 142 (75%)– 10.0 (6438) P , 0.001 28.8 (465) P = 0.25
157 (83%)
Medicare 179 (94%) 177 (93%)– 5.7 (9348) P = 0.16 21.2 (610) P = 0.02
181 (95%)
Private 181 (95%) 177 (93%)– 9.1 (16,458) P , 0.001 27.4 (1253) P = 0.26
183 (96%)
Texas Nonemergent Medicaid 336 195 (58%) — 21.1 (1169) — 32.0 (80) —
Medicare 232 (69%) 222 (66%)– 16.4 (6243) P , 0.001 37.4 (606) P = 0.25
241 (72%)
Private 226 (67%) 217 (65%)– 17.6 (8576) P = 0.002 37.3 (655) P = 0.26
239 (71%)
Medicare 227 (68%) 220 (65%)– 14.2 (7727) P = 0.01 33.9 (637) P = 0.31
234 (70%)
Private 221 (66%) 213 (63%)– 20.9 (11,851) P , 0.001 41.3 (859) P = 0.41
229 (68%)
*Medicare/private hospital counts were adjusted to account for the larger size of these groups relative to Medicaid groups. Details are provided in the text.
in nonemergent settings than in emergent settings in most patients was greater than for both Medicare and private cov-
cases, including for Medicaid and Medicare in every state and erage patients. In the nonemergent, urban setting in California
for private coverage in 2 states (Figs. 2 and 3). Travel dis- and Florida, Medicaid patients had travel distances greater
tances were also universally greater in rural areas than in than those of Medicare patients but similar to those of private
urban areas (Table 2). In the nonemergent, urban setting in patients. In the emergent, urban setting in California, Florida,
Texas, Medicaid patients had a mean estimated travel distance and Texas, average Medicaid travel distances were signifi-
of 21.1 miles, which was significantly greater than the mean cantly greater than Medicare travel distances, but significantly
travel distance for Medicare patients (16.4 miles) and pri- smaller than private coverage travel distances. The only rural
vately insured patients (17.6 miles, Table 2). There was no setting associated with a significant difference in travel dis-
other setting in which the mean travel distance for Medicaid tances was the emergent, rural setting in New York, in which
FIGURE 1. Bar chart showing proportion of hospitals in each state that saw each payer/acuity group. Medicaid patients tended to
be cared for in fewer facilities than otherwise-insured patients, particularly in nonemergent settings. Medicare and private hospital
counts were adjusted for comparability with the smaller Medicaid groups (details in the text). Error bars represent 95% confidence
intervals for this adjustment.
Medicaid patients traveled significantly further than Medicare for Medicaid patients in the majority of settings studied. We
patients. expected any disparity in travel burden to be particularly
pronounced in rural areas, and travel distances were indeed
much longer in rural areas, but we saw no difference in travel
DISCUSSION distance for Medicaid versus otherwise-insured patients in
This large multistate analysis showed that fewer most rural settings studied.
hospitals provide operative fracture care to Medicaid patients This result confirms and expands on findings from prior
than otherwise-covered patients in 4 large states. This gap is analyses, which have tended to focus on a single state or
present in nonemergent settings in all 4 states, and, to a lesser institution. Researchers have demonstrated increased wait
extent, in emergent settings in New York and Texas. Contrary times and travel distances for Medicaid patients seeking
to our second hypothesis, this disparity in hospital visitation orthopaedic care in the areas of trauma,7–9 sports medi-
patterns did not manifest as a notably increased travel burden cine,11,29 arthroplasty,12–17 hand,18,19 foot and ankle,20
FIGURE 2. Box-and-whisker plots showing travel distances from home to hospital for each payer/acuity group in urban areas. The
shaded box represents the interquartile range (IQR) from the first to the third quartile, and whiskers are drawn to the most
extreme values within 1.5 · IQR from the edge of the box. The line within the box represents the median, and diamonds represent
means. Outliers beyond the range of the whiskers were included in the calculations of means and quartiles but are not visualized in
the plot.
FIGURE 3. Box-and-whisker plots showing travel distances from home to hospital for each payer/acuity group in rural areas. The
shaded box represents the interquartile range (IQR) from the first to the third quartile, and whiskers are drawn to the most
extreme values within 1.5 · IQR from the edge of the box. The line within the box represents the median, and diamonds represent
means. Outliers beyond the range of the whiskers were included in the calculations of means and quartiles but are not visualized in
the plot.
spine,21,22 and pediatric orthopaedics.9,11,23,24 Such barriers Another question raised is whether Medicaid patients
have been shown to influence Medicaid patient outcomes are adversely affected, on a population level, by the fact that
by delaying time-sensitive diagnoses15,30 and surgeries,31 as they are cared for in a smaller number of hospitals. Does
well as impairing follow-up.8,32 Adding to these results, the traveling out of their way to find a hospital that will accept
Medicaid access disparity in this multistate analysis suggests their coverage impose a disadvantage that affects their
a more widespread problem than previously demonstrated. outcomes? One such burden would be greater travel distances
Our analysis raises the question of whether the smaller associated with Medicaid coverage that lead to delays in care,
number of hospitals providing care to Medicaid versus but our population-based study did not find convincing
otherwise-insured patients is a result of coverage-related evidence of greater travel distances for Medicaid patients
access or a result of patient preference based on location of than otherwise-insured patients in the majority of settings
residence. The stratification of hospital visitation patterns and studied, contrary to prior research.7 Another possible disad-
travel distances by admission acuity helps to address this vantage would involve Medicaid patients having less access
question. It could be argued that emergent admissions often to high-volume hospitals, as has been demonstrated in the
occur near the location of injury, whereas nonemergent joint replacement population.34 However, unlike in total joint
admissions, with a delay after the injury, are more likely to arthroplasty, the volume–outcome relationship for common
occur near the patient’s residence. With fewer sites offering fracture repair surgeries has not been conclusively demon-
orthopaedic care in areas with high Medicaid enrollment, strated.35,36 Delays in care Medicaid patients could face in
location of residence could thereby explain the larger dispar- having access to fewer hospitals may have a greater adverse
ity in hospital visitation we see in nonemergent settings ver- impact on clinical outcomes than the specific procedure vol-
sus emergent settings. However, this explanation does not ume or other characteristics of the hospitals themselves.
hold up to the finding that Medicaid patients travel further Understanding the overall quality of care Medicaid patients
to receive nonemergent care than emergent care. Together, receive in a reduced number of facilities therefore requires
the findings of a greater decline, relative to other payers, in further study, looking at measures such as the frequency of
the number of hospitals visited in nonemergent settings and a delay in seeking care and patient-reported outcomes after
longer travel distances in nonemergent versus emergent set- treatment. Nevertheless, the multistate finding of a disparity
tings strongly suggest that some Medicaid patients are trav- in hospital visitation patterns presents a problem for the
eling out of their way to find hospitals that will accept their AAOS stated commitment to achieving universal access to
insurance. When making decisions regarding where to specialty care.
undergo an elective joint replacement operation, FitzGerald The limitations of this study were mostly those of large
and colleagues demonstrated that socioeconomic factors actu- database studies in general and those related to the constraints
ally dominate geographic factors for urban patients.33 It is of the specific databases used. Large database studies in
reasonable to assume that socioeconomic factors would sim- general can be subject to coding errors and the lack of ability
ilarly enter into the decision of where to receive nonemergent to prospectively capture outcome indicators. The state data-
operative fracture care. bases used in this study were chosen because they represent
complete samples across all payers and inpatient facilities, The implications of this study for the practicing
allowing for a comprehensive analysis of hospital visitation orthopaedic surgeon are 2-fold. First, orthopaedic surgeons
patterns. However, these sources lacked the longitudinal can take steps in their own communities to identify and
clinical data and injury characteristics that would have address barriers Medicaid patients face in receiving optimal
permitted a more detailed breakdown of delays in care and orthopaedic trauma care. This could take the form of quality
patient outcomes. In addition, although the availability of zip initiatives that target the increased rates of inappropriate self-
code data allowed us to estimate travel distance, straight-line discharge from the emergency department40 and follow-up
distances from zip code centers are an imperfect measure of noncompliance41 in the Medicaid population. In addition to
travel time and burden. Travel times are difficult to estimate this local response, orthopaedic surgeons can advocate for the
with zip code data given varying road layouts across a zip policy changes their patients need through professional soci-
code, and straight-line distance is a reasonable proxy,37 but eties and state medical boards. Medicaid access barriers cor-
future studies would benefit from considering actual travel relate with reimbursement rates,24,42–45 and the program is
times by mapping directions between addresses. Overall, associated with increased billing complexity relative to other
given the long timeframes and large numbers of patients, payers.46 On a federal level, relevant policy goals would be
our results are likely generalizable to the experience of similar linking Medicaid reimbursements to Medicare rates, which
patients cared for at other time points in California, Florida, could be done through executive action rather than legisla-
New York, and Texas, but may be less generalizable to the tion,47 or expanding the Medicare program to replace Med-
experience of similar patients in other states. icaid and cover low income people, which has been proposed
One limitation of the time frames selected was that in the US Congress.
different 5-year periods were studied in Florida and California This study has highlighted, through an analysis of state-
(2010–2014) and New York and Texas (2006–2010). This was level hospitalization data, disparities in Medicaid access to
due to the availability of data but could have made Florida and operative fracture repair. These results illuminate unique
California results less comparable with New York and Texas challenges Medicaid patients can face when seeking essential
results, especially considering that Medicaid expansion began in fracture care and suggest that both providers and policymakers
California and New York in January 2014. Nevertheless, re- have a role to play in improving access in this population.
searchers studying the change in coverage-related access for
Medicaid patients over time have found very small changes even REFERENCES
in the primary care setting, where incentives for providers to see 1. Burwell SM. 2016 Actuarial Report on the Financial Outlook for Med-
Medicaid patients have changed more than in the specialty set- icaid. 2016:i–15. Available at: https://www.cms.gov/Research-Statistics-
Data-and-Systems/Research/ActuarialStudies/MedicaidReport.html. Ac-
ting.38,39 Given this limited change in access over time, the use cessed April 2, 2017.
of different time frames is unlikely to undermine the validity of 2. Centers for Medicare & Medicaid Services. January 2018 Medicaid and
our results as long as the results are considered generalizable to CHIP Enrollment Data Highlights. Available at: https://www.medicaid.
the experience of other patients in the same state. gov/medicaid/program-information/medicaid-and-chip-enrollment-data/
An additional limitation was the lack of ability to more report-highlights/index.html. Accessed October 28, 2018.
3. Chokshi DA, Chang JE, Wilson RM. Health reform and the changing
conclusively identify the cause of increased traveled distances safety net in the United States. N Engl J Med. 2016;375:1790–1796.
for some patients. As mentioned above, the combination of 4. Sommers BD, Gunja MZ, Finegold K, et al. Changes in self-reported
greater travel distances and fewer hospitals visited in non- insurance coverage, access to care, and health under the Affordable Care
emergent settings suggests coverage-related access is an Act. JAMA. 2015;314:366–369.
5. Wherry LR, Miller S. Early coverage, access, utilization, and health
issue, but the databases do not shed light on whether an effects associated with the Affordable Care Act Medicaid expansions.
individual patient is traveling further out of preference or out Ann Intern Med. 2016;164:795–817.
of need. We saw in many settings that private patients 6. Principles of Health Care Reform and Specialty Care. American Asso-
traveled as far or further than Medicaid patients, suggesting ciation of Orthopaedic Surgeons. 2015:1–9. Available at: http://www.
that although some patients may have lacked local access, aaos.org/About/Statements/Position/. Accessed October 28, 2018.
7. Lack WD, Carlo JO, Marsh JL. Payer status and increased distance traveled
others merely preferred a more distant hospital, and these for fracture care in a rural state. J Orthop Trauma. 2013;27:113–118.
issues were not readily separable. To focus more on coverage- 8. Medford-Davis LN, Lin F, Greenstein A, et al. “I broke my ankle”:
related access, it would have been beneficial to identify access to orthopedic follow-up care by insurance status. Acad Emerg
patients who came through the emergency department rather Med. 2017;24:98–105.
9. Potak H, A Iobst C. Influence of insurance type on the access to pediatric
than through clinic. Unfortunately, the HCUP State Inpatient care for children with distal radius torus fractures. J Pediatr Orthop.
Databases do not contain this level of detail. The State 2017:1 [epub ahead of print].
Emergency Department Databases would be a good source 10. Patterson BM, Draeger RW, Olsson EC, et al. A regional assessment of
for this type of population-level investigation, but are not Medicaid access to outpatient orthopaedic care: the influence of popula-
available for California or Texas, the 2 most populous states. tion density and proximity to academic medical centers on patient access.
J Bone Joint Surg Am. 2014;96:e156.
Overall, the results of this study should be interpreted as 11. Pierce TR, Mehlman CT, Tamai J, et al. Access to care for the adolescent
identifying the breadth, across different clinical sites and state anterior cruciate ligament patient with Medicaid versus private insurance.
policy environments, of a potential access problem that J Pediatr Orthop. 2012;32:245–248.
orthopaedic surgeons are equipped to respond to and mitigate. 12. Hinman A, Bozic KJ. Impact of payer type on resource utilization, outcomes
and access to care in total hip arthroplasty. J Arthroplasty. 2008;23:9–14.
Population-level studies of access problems can motivate 13. Kim CY, Wiznia DH, Hsiang WR, et al. The effect of insurance type on
related research with smaller but more detailed registry and patient access to knee arthroplasty and revision under the Affordable
clinical consortium data sets. Care Act. J Arthroplasty. 2015;30:1498–1501.
14. Lavernia CJ, Contreras JS, Alcerro JC. Access to arthroplasty in South 30. Baraga MG, Smith MK, Tanner JP, et al. Anterior cruciate ligament
Florida. J Arthroplasty. 2012;27:1585–1588. injury and access to care in South Florida: does insurance status play
15. Martin CT, Callaghan JJ, Liu SS, et al. Disparity in total joint arthro- a role? J Bone Joint Surg. 2012;94:e183.
plasty patient comorbidities, demographics, and postoperative outcomes 31. Dy CJ, Baty J, Saeed MJ, et al. A population-based analysis of time to
based on insurance payer type. J Arthroplasty. 2012;27:1761–1765.e1. surgery and travel distances for brachial plexus surgery. J Hand Surg.
16. Schwarzkopf R, Phan DL, Hoang M, et al. Do patients with income- 2016;41:903–909.e3.
based insurance have access to total joint arthroplasty? J Arthroplasty. 32. Rosenthal BD, Hulst JB, Moric M, et al. The effect of payer type on clinical
2014;29:1083–1086.e1. outcomes in total knee arthroplasty. J Arthroplasty. 2014;29:295–298.
17. Warth LC, Callaghan JJ, Wells CW, et al. Demographic and comorbid 33. FitzGerald JD, SooHoo NF, Losina E, et al. Potential impact on patient
disparities based on payer type in a total joint arthroplasty cohort: residence to hospital travel distance and access to care under a policy of
implications in a changing health care arena. Iowa Orthop J. 2011; preferential referral to high-volume knee replacement hospitals. Arthritis
31:64–68. Care Res. 2012;64:890–897.
18. Draeger RW, Patterson BM, Olsson EC, et al. The influence of patient 34. Catanzano AA, Hutzler LH, Bosco JA. The relationship between hospital
insurance status on access to outpatient orthopedic care for flexor tendon payer mix and volume growth in total joint arthroplasty: a 12-year anal-
lacerations. J Hand Surg. 2014;39:527–533. ysis. J Arthroplasty. 2016;31:1641–1644.
19. Kim C-Y, Wiznia DH, Wang Y, et al. The effect of insurance type on 35. Okike K, Chan PH, Paxton EW. Effect of surgeon and hospital volume
patient access to carpal tunnel release under the Affordable Care Act. J on morbidity and mortality after hip fracture. J Bone Joint Surg. 2017;99:
Hand Surg. 2016;41:503–509.e1. 1547–1553.
20. Kim C-Y, Wiznia DH, Roth AS, et al. Survey of patient insurance status 36. SooHoo NF, Krenek L, Eagan MJ, et al. Complication rates following
on access to specialty foot and ankle care under the Affordable Care Act. open reduction and internal fixation of ankle fractures. J Bone Joint Surg.
Foot Ankle Int. 2016;37:776–781. 2009;91:1042–1049.
21. Anandasivam NS, Wiznia DH, Kim CY, et al. Access of patients with 37. Phibbs CS, Luft HS. Correlation of travel time on roads versus straight
lumbar disc herniations to spine surgeons: the effect of insurance type line distance. Med Care Res Rev. 1995;52:532–542.
under the Affordable Care Act. Spine. 2017;42:1179–1183. 38. Neprash HT, Zink A, Gray J, et al. Physicians’ participation in Medicaid
22. Weiner BK, Black KP, Gish J. Access to spine care for the poor and near increased only slightly following expansion. Health Aff. 2018;37:1087–1091.
poor. Spine J. 2009;9:221–224. 39. Decker SL. No association found between the Medicaid primary care fee
23. Sabatini CS, Skaggs KF, Kay RM, et al. Orthopedic surgeons are less bump and physician-reported participation in Medicaid. Health Aff. 2018;
likely to see children now for fracture care compared with 10 years ago. J 37:1092–1098.
Pediatr. 2012;160:505–507. 40. Menendez ME, van Dijk CN, Ring D. Who leaves the hospital against
24. Skaggs DL, Lehmann CL, Rice C, et al. Access to orthopaedic care for medical advice in the orthopaedic setting? Clin Orthop Relat Res. 2014;
children with Medicaid versus private insurance: results of a national 473:1140–1149.
survey. J Pediatr Orthop. 2006;26:400–404. 41. Stone ME Jr, Marsh J, Cucuzzo J, et al. Factors associated with trauma
25. Total Monthly Medicaid and CHIP Enrollment. 2018. Available at: https:// clinic follow-up compliance after discharge. J Trauma Acute Care Surg.
www.kff.org/health-reform/state-indicator/total-monthly-medicaid-and-chip- 2014;76:185–190.
enrollment/. Accessed October 28, 2018. 42. Decker SL. In 2011 nearly one-third of physicians said they would not
26. Agency for Healthcare Research and Quality (AHRQ). Admission type accept new Medicaid patients, but rising fees may help. Health Aff (Mill-
data element. Available at: https://ushik.ahrq.gov/ViewItemDetails? wood). 2012;31:1673–1679.
system=mdr&itemKey=74569000. Accessed October 28, 2018. 43. Decker SL. Changes in Medicaid physician fees and patterns of ambu-
27. Barrett M, Hines A, Andrews R, et al. An Examination of Expected Payer latory care. Inquiry. 2009;46:291–304.
Coding in HCUP Databases. U.S. Agency for Healthcare Research and 44. Polsky D, Richards M, Basseyn S, et al. Appointment availability after increases
Quality; 2014. Available at: http://www.hcup-us.ahrq.gov/reports/methods/ in Medicaid payments for primary care. N Engl J Med. 2015;372:537–545.
methods.jsp. Accessed October 28, 2018. 45. White C. A comparison of two approaches to increasing access to care:
28. United States Census Bureau. 2010 Urban Area to ZIP Code Tabulation expanding coverage versus increasing physician fees. Health Serv Res.
Area (ZCTA) Relationship File. Available at: https://www.census.gov/ 2012;47(3 pt 1):963–983.
geo/maps-data/data/ua_rel_download.html. Accessed October 28, 2018. 46. Gottlieb JD, Shapiro AH, Dunn A. The complexity of billing and paying
29. Patterson BM, Spang JT, Draeger RW, et al. Access to outpatient care for for physician care. Health Aff. 2018;37:619–626.
adult rotator cuff patients with private insurance versus Medicaid in 47. Rosenbaum S. Medicaid payments and access to care. N Engl J Med.
North Carolina. J Shoulder Elbow Surg. 2013;22:1623–1627. 2014;371:2345–2347.
ORIGINAL ARTICLE
Standardized Hospital-Based Care Programs Improve

Geriatric Hip Fracture Outcomes: An Analysis of the ACS
NSQIP Targeted Hip Fracture Series
Armin Arshi, MD,* Kevin Rezzadeh, BA,† Alexandra I. Stavrakis, MD,* Susan V. Bukata, MD,*
and Erik N. Zeegen, MD*
Key Words: hip fractures, standardized care program, organized

Objective: To determine relative complication rates and outcome care program, medical comanagement, geriatric, mortality, com-
measures in patients treated under a standardized hip fracture plication
program (SHFP).
Level of Evidence: Therapeutic Level III. See Instructions for
Methods: The American College of Surgeons National Surgical Authors for a complete description of levels of evidence.
Quality Improvement Program was queried to identify patients who
underwent operative fixation of femoral neck, intertrochanteric hip, (J Orthop Trauma 2019;33:e223–e228)
and subtrochanteric hip fractures in 2016. Cohorts of patients who
were and were not treated under a documented SHFP were INTRODUCTION
identified. Relevant perioperative clinical and outcomes data were Hip fractures represent a considerable source of mor-
collected. Multivariate regression was used to assess risk-adjusted bidity and mortality in patients. More than 300,000 hip
complication rates and outcomes for patients treated in SHFPs. fractures occur annually, with 86% of patients sustaining
these hip fractures being over the age of 65 years with
Results: A total of 9360 hip fracture patients were identified of associated medical comorbidities that complicate care and
whom 5070 (54.2%) were treated under a documented SHFP.
recovery.1,2 Furthermore, hip fractures represent the costliest
Median age was 84 years, and 69.9% of patients were women.
fractures to care for in our current health care system. The
Patients in an SHFP had a lower risk-adjusted incidence of
current estimated annual cost for these fractures in the United
postoperative deep vein thrombosis [odds ratio (OR) 0.48 (0.32–
States is $12–15 billion, and that number is projected to
0.72), P , 0.001]. Rates of other medical and surgical complications
increase over the coming decades.3,4 Given the high cost of
and 30-day mortality were statistically comparable. Risk-adjusted
care for hip fractures and the anticipated growth in the num-
evaluation showed that SHFP patients were less likely to be dis-
ber of patients over the age of 65 years in the United States,
charged to an inpatient facility versus home [OR 0.72 (0.63–0.81),
special consideration to improve health care delivery and out-
P , 0.001] and had a lower 30-day readmission rate [OR 0.83
comes given to geriatric patients in particular is merited.
(0.71–0.97), P = 0.023]. Furthermore, the SHFP patients had higher
At the forefront of efforts to optimize care for elderly
rates of immediate postoperative weight-bearing as tolerated [OR
patients with hip fractures is the implementation of standardized
1.23 (1.10–1.37), P , 0.001], adherence to deep vein thrombosis
hip fracture programs (SHFPs). These comanaged, protocol-
prophylaxis at 28 days [OR 1.27 (1.16–1.38), P , 0.001], and
driven programs comprise standardized measures including
initiation of bone protective medications [OR 1.79 (1.64–1.96),
admission checklists, perioperative order sets, multidisciplinary
P , 0.001].
evaluation and management, postoperative rehabilitation goals,
Conclusions: Care in a modern hospital-based SHFP is associated and discharge criteria.5–7 Despite significant variability in the
with improved short-term outcome measures. Further development features of SHFPs, studies from institutions that have imple-
and widespread implementation of organized, multidisciplinary mented them have consistently shown this approach to be effi-
orthogeriatric hip fracture protocols is recommended. cacious in curbing both complications and costs associated with
hip fractures.8–10 Individual centers have specifically reported
improved time to surgery, hospital total length of stay (LOS),
total costs, perioperative complications, and short-term mortal-
Accepted for publication December 29, 2018. ity in these programs.5–12 However, no studies have reported
From the *Department of Orthopaedic Surgery, David Geffen School of Med- multicenter comparative data on the relative efficacy of these
icine at UCLA, Los Angeles, CA; and †New York University School of
Medicine, New York, NY. orthogeriatric SHFPs as they are practiced presently.
The authors report no conflict of interest. The purpose of this study was to evaluate the 30-day
Reprints: Armin Arshi, MD, Department of Orthopaedic Surgery, David perioperative outcomes and complications in patients receiv-
Geffen School of Medicine at UCLA, 10833 Le Conte Avenue, 76-143 ing care at institutions with an SHFP relative to patients who
CHS, Los Angeles, CA 90095-6902 (e-mail: arminarshi@mednet.ucla.
edu)
received care at institutions without SHFPs in place in 2016.
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. Our initial hypothesis was that patients receiving hip fracture
DOI: 10.1097/BOT.0000000000001443 care at institutions with a standardized care program would
Arshi et al J Orthop Trauma Volume 33, Number 6, June 2019
have lower short-term postoperative morbidity and mortality time. Data were also collected on whether patients were
rates, as well as improved rates of hip fracture–specific out- weight-bearing as tolerated (WBAT) on POD1, had a pre-
come measures such as osteoporosis treatment, early mobi- scription for DVT prophylaxis (eg, aspirin, low-molecular-
lization, discharge to home, and hospital LOS. weight heparin or unfractionated heparin, warfarin, fondapar-
inux, factor Xa inhibitors, or direct thrombin inhibitors) at 28
days, and had prescription for postoperative bone protective
MATERIALS AND METHODS pain medications at 30 days (eg, vitamin D, bisphosphonates,
We performed a retrospective review of the American teriparatide, denosumab, and raloxifene).
College of Surgeons (ACS) National Surgical Quality Descriptive statistics and comparison of baseline char-
Improvement Program (NSQIP) database (https://www.facs. acteristics were performed using x2 testing, independent-
org/quality-programs/acs-nsqip). The ACS NSQIP is a pro- samples t test, and independent-samples median test with
spective, multi-institutional program that collects periopera- Yates’ continuity. Multivariate logistic regression was used
tive data on more than 150 patient variables from more than to determine the relationship between SHFPs and medical
500 NSQIP-participating hospitals in the United States.13 Re- complications, surgical complications, and hip fracture out-
ported data are acquired from medical records, operative re- comes. Baseline age, sex, race, BMI, functional status, ASA
ports, and patient interviews by trained clinical reviewers and classification, and comorbidities were used as covariates to
are compliant with the Health Insurance Portability and compute risk-adjusted odds ratio (OR) for each variable
Accountability Act. The series undergoes routine auditing, with patients in an SHFP treated as the exposed group.
which ensures high-quality data with a reported inter-rater Multivariate linear regression was similarly used to assess
disagreement of 2.3% for all variables. Data are collected the risk-adjusted relationship between SHFPs and time from
through postoperative day (POD) 30 and include information admission to surgery, operative time, and hospital LOS.
after hospital discharge. Standardized OR, 95% confidence intervals, and P values
Patients who underwent operative fixation of femoral were computed using standard methods. Statistical signifi-
neck, intertrochanteric hip, and subtrochanteric hip fractures cance was defined as P , 0.05, and all statistical analyses
(OTA/AO classification 31 and 32) in 2016 were identified were performed using SPSS 21 software (IBM Corp, Ar-
using the targeted hip fracture participant utilization file, monk, NY).
which collects hip fracture–specific data in addition to the
standard reported variables. Hip fracture patients were
divided into 2 cohorts based on whether or not they under- RESULTS
went treatment at an institution with an SHFP as reported by The query identified a total of 9360 patients who had
the participant utilization file. SHFPs were defined as in- operative fixation of femoral neck, intertrochanteric hip, and
stitutions reported by NSQIP to have a standard hip fracture subtrochanteric hip fractures in 2016 in the NSQIP series. Of
management protocol and/or care pathway used for all hip these, 5070 patients were treated in an SHFP and 4290 patients
fracture patients, which may include formal protocols for were not treated. The median age in the 2 cohorts was 84 and
standardized admission and perioperative order sets, preoper- 83 years, respectively (Table 1). Of the total cohort, 69.9%
ative multidisciplinary evaluation and screening, daily post- were women, 90.0% of patients were white, and 78.7% were
operative milestones and discharge criteria, and structured independent in their activities of daily living at baseline.
care coordination.13 Patient characteristics collected from Among fractures treated, 38.4% were femoral neck fractures,
the registry included demographics, smoking history, Amer- 54.9% were intertrochanteric fractures, and 6.7% were subtro-
ican Society of Anesthesiologists (ASA) class, and medical chanteric fractures. Baseline age (P = 0.023), sex (P = 0.034),
comorbidities including diabetes, dementia, chronic obstruc- race (P , 0.001), functional health status (P , 0.001), and
tive pulmonary disorder, liver disease with ascites, congestive ASA classification (P = 0.011) of the 2 groups were not sta-
heart failure, hypertension, and dialysis-dependent kidney tistically equivalent between the 2 cohorts at baseline. Baseline
disease. Body mass index (BMI) was calculated for each BMI was comparable between the 2 cohorts at baseline (P =
patient’s height and weight. Functional status was defined 0.270). Baseline medical comorbidities were not statistically
as the patient’s ability to perform the activities of daily living equivalent between the 2 cohorts (Table 1).
either independently or in a partially or completely dependent Multivariate logistic regression analysis showed that
manner within the 30 days before admission. patients in an SHFP had a lower risk-adjusted incidence of
Data on postoperative medical complications within 30 postoperative DVT within 30 days [0.8% vs. 1.7%, OR 0.48
days were collected: deep vein thrombosis (DVT), pulmonary (0.32–0.72), P , 0.001]. All other 30-day medical complica-
embolism, pneumonia, acute renal failure, urinary tract tions occurred at statistically equivalent risk-adjusted rates
infection, cardiac arrest, myocardial infarction, cerebrovascu- between the 2 cohorts, and there was no statistically signifi-
lar accidents, and postoperative delirium. Surgical complica- cant difference in 30-day mortality (Table 2). Similarly, sur-
tions data on postoperative superficial and deep surgical site gical complications, including superficial (P = 0.409) and
infection (SSI), wound dehiscence, reoperation, and blood deep SSI (P = 0.137), wound dehiscence (P = 0.261), all-
transfusion were also collected for both cohorts. Hip fracture– cause reoperation (P = 0.862), and transfusion (P = 0.582)
specific outcomes data collected included discharge destina- occurred at statistically comparable rates.
tion (eg, home vs. facility), hospital readmission, time to Multivariate logistic regression showed that patients in
surgery from admission, total hospital LOS, and operative an SHFP had a lower likelihood of being discharged to an
J Orthop Trauma Volume 33, Number 6, June 2019 Standardized Hip Fracture Care Programs
TABLE 1. Comparative Analysis of Baseline Demographics and Characteristics of Patients Treated in Standardized Hip Fracture
Care Programs
Standardized Hip Fracture No Hip Fracture Program
Baseline Characteristic Program (n = 5070) (n = 4290) P
Age (y, median)* 84 83 0.023
Female (%) 70.8 68.8 0.034
Race ,0.001
White (%) 88.9 91.3
Black (%) 3.5 3.5
Hispanic (%) 4.8 2.8
Asian (%) 2.5 2.2
Other (%) 0.2 0.2
BMI† 25.0 6 5.7 25.1 6 6.0 0.270
Functional health status ,0.001
Independent 80.6% 76.4%
Partially dependent 17.3% 19.9%
Totally dependent 2.1% 3.7%
ASA classification (continuous)* 3.07 6 0.64 3.03 6 0.63 0.003
ASA classification (ordinal) 0.011
1 0.7% 0.6%
2 15.2% 16.4%
3 60.7% 62.5%
4 23.2% 20.4%
5 0.2% 0.1%
Dementia (%) 30.3 26.9 ,0.001
Diabetes (%) 16.8 19.1 0.004
Smoker, current (%) 10.6 9.9 0.009
COPD (%) 10.4 11.0 0.345
Ascites (%) 0.4 0.4 0.745
CHF (%) 3.9 3.8 0.791
Hypertension (%) 64.1 69.0 ,0.001
Dialysis (%) 1.6 2.4 0.013
Hip fracture pattern 0.005
Femoral neck 39.5% 37.1%
Intertrochanteric 53.4% 56.7%
Subtrochanteric 7.2% 6.2%
All other comparisons of categorical variables performed using the chi-square analysis.
*Statistical comparison performed using the independent-samples median test with Yates’ continuity correction.
†Statistical comparison performed using the independent-samples t test.
CHF, congestive heart failure; COPD, chronic obstructive pulmonary disorder.
inpatient facility instead of home [77.3% vs. 81.5%, OR 0.72 DISCUSSION

(0.63–0.81), P , 0.001] (Table 3). There was no significant SHFPs have been adopted in the United States and
difference in inpatient facility utilization rates at 30 days after abroad as a mechanism to improve outcomes and potentially
surgery. Rates of hospital readmission within 30 days was curtail costs associated with geriatric hip fractures. Individual
lower among patients in an SHFP [7.1% vs. 9.1%, OR 0.83 centers with comprehensive, protocol-driven pathways have
(0.71–0.97), P = 0.023]. Care under an SHFP also predicted reported reduced in-hospital mortality and complications and
other favorable hip fracture outcome measures including readmission rates.14 The multi-institutional data presented
prescription of bone protective medications at 30 days here demonstrate that organized orthogeriatric SHFPs
[54.5% vs. 36.4%, OR 1.79 (1.64–1.96), P , 0.001], WBAT improve short-term outcome measures specific to hip fracture
on POD1 [75.6% vs. 71.4%, OR 1.23 (1.10–1.37), P , care with several important practical implications.
0.001], and DVT chemoprophylaxis at 4 weeks [60.8% vs. Hip fractures are associated with an estimated 2.2%
53.7%, OR 1.27 (1.16–1.38), P , 0.001]. Patients in incidence of symptomatic DVT, with 85% of these events
an SHFP also had longer mean hospital LOS [6.9 days vs. occurring within 5 weeks postoperatively.15 In this study,
6.0 days, P , 0.001] (Table 4). Time from admission to OR patients in SHFPs had a lower risk-adjusted incidence of post-
(P = 0.536) and operative time (P = 0.876) was not signifi- operative DVT than patients who underwent surgery at insti-
cantly different for patients in an SHFP. tutions without SHFPs. This reduction in DVT prevalence in
TABLE 2. Comparison of Risk-Adjusted Postoperative Medical and Surgical Complications for Patients Treated in Standardized
Hip Fracture Care Programs Using Multivariate Logistic Regression
Complications Standardized Hip Fracture No Hip Fracture Program Adjusted OR (for Patients in Hip
(within 30 d) Program (n = 5070), % (n = 4290), % Fracture program)* P
Medical complications
Death 6.6 6.5 0.97 (0.81–1.18) 0.777
DVT 0.8 1.7 0.48 (0.32–0.72) ,0.001
Pulmonary embolism 0.8 0.7 1.01 (0.60–1.69) 0.975
Pneumonia 4.0 4.3 0.82 (0.65–1.03) 0.093
Acute renal failure 0.2 0.4 0.70 (0.32–1.52) 0.365
Urinary tract infection 5.2 3.9 1.21 (0.97–1.51) 0.090
Cerebrovascular accident 0.9 1.0 0.89 (0.56–1.41) 0.620
Cardiac arrest 0.6 0.8 0.76 (0.45–1.28) 0.301
Myocardial infarction 3.4 2.0 0.97 (0.71–1.34) 0.863
Postoperative delirium 28.8 27.9 0.91 (0.82–1.01) 0.080
Surgical complications
Superficial SSI 0.6 0.3 1.37 (0.65–2.89) 0.409
Deep SSI 0.1 0.2 0.30 (0.06–1.46) 0.137
Wound dehiscence 0.0 0.1 0.27 (0.03–2.64) 0.261
Reoperation 2.2 2.3 1.03 (0.76–1.38) 0.862
Transfusion 26.8 27.9 0.97 (0.88–1.08) 0.582
*Adjusted OR computed using multivariate logistic regression with age, sex, BMI, functional status, comorbidities, and ASA classification as covariates.
the SHFP cohort has been previously reported in the literature between the 2 groups likely reflects differences in established
and may be explained by many of the preventive measures institution-specific protocols.
used as part of geriatric hip fracture programs such as DVT The patients in the intervention group showed higher
thromboprophylaxis order sets, early mobilization after sur- rates of immediate postoperative WBAT. The increased rate
gery, and clear postoperative physical rehabilitation goals.16,17 of early weight-bearing can be attributed to the interdisci-
Although many of these factors are not exclusive to the SHFP plinary nature of SHFPs, which typically involve preordered
intervention group, they are more likely to be completed as intensive physical therapy with protocol-driven criteria. It
a result of being in standardized program. The lower rate of may also reflect increased institutional emphasis on immedi-
DVTs in this study may also be explained in part by patients in ate postoperative weight-bearing as an important tenet of
the intervention group having significantly higher rates of DVT geriatric hip fracture care. Not only does early mobility
prophylaxis compliance at 28 days. Despite chemoprophylaxis decrease the risk of delirium and other postoperative compli-
being generally recommended in the month after hip fracture cations developing in elderly patients, but also unrestricted
surgery to reduce DVT incidence, there is no consensus on the weight-bearing has been shown to reduce the incidence of
optimal anticoagulation agent or timing for its use in the pres- fragility fractures in geriatric populations and is considered
ent American Academy of Orthopaedic Surgeons Clinical standard of care in postoperative management of most
Practice Guideline.18 Nevertheless, this observed difference geriatric hip fractures.19,20 This may also partially account
in rates of continued DVT chemoprophylaxis at 28 days for the observed differences in postoperative DVT rates.
TABLE 3. Comparison of Risk-Adjusted Outcomes Measures for Patients Treated in Standardized Hip Fracture Care Programs
Using Multivariate Logistic Regression
Standardized Hip Fracture No Hip Fracture Program Adjusted OR (for Patients in
Outcome Measure Program (n = 5070), % (n = 4290), % Hip Fracture Program)* P
Discharge to inpatient facility 77.3 81.5 0.72 (0.63–0.81) ,0.001
Inpatient facility at 30 days 43.7 40.0 0.96 (0.86–1.06) 0.416
Hospital readmission within 30 days 7.1 9.1 0.83 (0.71–0.97) 0.023
Bone protective medications initiated 54.5 36.4 1.79 (1.64–1.96) ,0.001
WBAT on POD1 75.6 71.4 1.23 (1.10–1.37) ,0.001
DVT prophylaxis at 28 days 60.8 53.7 1.27 (1.16–1.38) ,0.001
*Adjusted OR computed using multivariate logistic regression with age, sex, BMI, functional status, comorbidities, and ASA classification as covariates.
J Orthop Trauma Volume 33, Number 6, June 2019 Standardized Hip Fracture Care Programs
TABLE 4. Comparison of Risk-Adjusted Perioperative Outcomes Measures for Patients Treated in Standardized Hip Fracture Care
Programs Using Multivariate Linear Regression
Standardized Hip Fracture No Hip Fracture Program Coefficient (B) (for Patient in
Outcome Measure Program (n = 5070) (n = 4290) Hip Fracture program)* P
Time to OR from admission (d) 1.1 6 1.2 1.1 6 1.2 0.01 (20.04 to 0.07) 0.536
Total hospital LOS (d) 6.9 6 4.8 6.0 6 3.9 0.55 (0.38 to 0.72) ,0.001
Operative time (minutes) 64.0 6 40.5 65.2 6 39.6 0.14 (21.58 to 1.86) 0.876
*Independent correlation computed using multivariate linear regression with age, sex, BMI, functional status, comorbidities, and ASA classification as covariates.
Given that hip fractures are most prevalent in elderly complications as a result of regular geriatric consultation and
patients who often have concomitant osteoporosis, referral of standardization of hip fracture patient management.
patients with hip fractures for osteoporosis evaluation and The previous meta-analysis on SHFPs mentioned also
treatment may help reduce subsequent fragility fractures and acknowledged mixed results regarding implementation of
is recommended by the American Academy of Orthopaedic geriatric hip fracture programs and LOS, with a shortened
Surgeons.18 The interdisciplinary nature of hip fracture pro- LOS being reported in most studies.10,25,26 One possible
grams, in which geriatricians are incorporated into or lead explanation for a longer hospital LOS in the intervention
care of the hip fracture patient, could explain higher rates group is that standardized multidisciplinary care and
of bone protective medication initiation in the intervention protocol-driven guidelines may delay discharge in exchange
group. Although up to 10% of geriatric hip fracture patients for patient optimization. Indeed, despite longer hospital LOS,
suffer another hip fracture within 1–5 years, an analysis of patients in the SHFP group showed lower risk-adjusted rates
osteoporosis medication use after hip fracture in the United of discharge to an inpatient facility and 30-day hospital read-
States between 2002 and 2011 showed an annual decrease in mission. This may also reflect an institutional preference for
medication prescription attributable to communication lapses patient discharge to home as opposed to rehabilitation centers,
between orthopaedic surgeons and other health care providers which may require extra in-hospital therapy or perioperative
invested in longitudinal patient care.21 The interdisciplinary optimization. Such findings have been previously cited as
management mandated by SHFPs in acute care settings a primary advantage of widespread implementation of
should result in improved long-term care and follow-up on SHFPs. Furthermore, these lower rates of costly postdi-
these finer points of orthogeriatric hip fracture management.22 scharge inpatient facility utilization and hospital readmissions
We highlight that the utilization rate among patients in SHFPs may partially explain and countermand cost concerns associ-
in this study (54.5%) is significantly higher than the estimate ated with the observed longer hospital LOS.
reported by Solomon et al21 in 2014 (20.5%). Such a marked Notably, this study found that the 30-day mortality and
improvement in osteoporosis treatment may have significant morbidity profiles were not significantly different for patients
impact on the rates of morbid and costly repeat fragility treated in SHFPs. Previous studies from single institutions
fractures.23 have reported improved short-term mortality and morbidity
This study also found that SHFPs were associated with rates on implementation of SHFPs.7,12,14,24,27 In their 2008
no significant improvement in time from admission to surgery series of 535 patients treated in an SHFP, Pedersen et al24
and a longer hospital LOS, contradictory to our initial reported a significant decrease in postoperative delirium,
hypothesis. Previous studies have shown a reduction in time pneumonia, and urinary tract infection, as well as hospital
to surgery after the implementation of a hip fracture pro- LOS. Similarly, Kates et al have consistently reported signif-
gram.7,24 The rapid transition to the operating room after icantly improved mortality rates in the Rochester multidisci-
SHFP implementation reported in the literature has been plinary care model.5,9,11,17 We attribute the findings of this
attributed to timely medical assessment and early medical study to the fact that hip fracture care has improved as a whole
management geriatric syndromes by geriatricians working across the United States and that institutions without formal
in conjunction with orthopaedic surgeons. However, a recent programs have begun to achieve similar outcomes to those
meta-analysis reported that time to hip fracture surgery varied with standardized programs with respect to many, although
depending on the model of orthogeriatric intervention imple- not all, metrics. However, it is also possible that these dispa-
mented in individual studies; routine geriatric consultation rate findings may reflect publication bias in the literature on
resulted in a significant decrease in time to surgery, while this topic, to which such a study design using anonymized,
the geriatric ward and shared care models of hip fracture care multi-institutional data would be inherently less susceptible.25
programs resulted in no significant difference in time to sur- The findings of this study may also have significant
gery.25 The finding of equivalent times to surgery between the policy implications. As bundled payment models emphasize
control and interventional group in our study may be ex- an increasingly popular value-based approach to medical
plained by more rigorous preoperative clearance in the SHFP reimbursement, the focus of the health care community has
relative to their nonstandardized counterparts, which may off- shifted from solely addressing acute issues to preventing
set the time saved by the avoidance of preoperative medical costly complications and readmissions after discharge.
Indeed, this study found that 54.2% of patients treated at an 6. Kates SL, Mendelson DA, Friedman SM. The value of an organized
NSQIP-participating hospital are treated in an SHFP. Hip fracture program for the elderly: early results. J Orthop Trauma. 2011;
25:233–237.
fracture programs have existed for some time and have 7. Miura LN, DiPiero AR, Homer LD. Effects of a geriatrician-led hip
proven to be both efficacious and cost-effective in single fracture program: improvements in clinical and economic outcomes. J
institutions reports and meta-analyses derived from Am Geriatr Soc. 2009;57:159–167.
them.7,12,28 The findings of this study provide greater incen- 8. Clement RC, Ahn J, Mehta S, et al. Economic viability of geriatric hip
tive and support for these programs from both patient care fracture centers. Orthopedics. 2013;36:e1509–14.
9. Kates SL, Mendelson DA, Friedman SM. Co-managed care for fragility
and cost saving perspectives. hip fractures (Rochester model). Osteoporos Int. 2010;21(suppl 4):621–
This study design has several limitations that also merit 625.
discussion. First, patient outcomes and complication data in 10. Soong C, Cram P, Chezar K, et al. Impact of an integrated hip fracture
the ACS NSQIP series are limited to 30 days postoperatively. inpatient program on length of stay and costs. J Orthop Trauma. 2016;
Other series have reported follow-up data for patients treated 30:647–652.
11. Schnell S, Friedman SM, Mendelson DA, et al. The 1-year mortality of
in SHFPs up to 1 year after admission, and this may be an patients treated in a hip fracture program for elders. Geriatr Orthop Surg
important consideration for hip fracture care that has long- Rehabil. 2010;1:6–14.
term sequelae that warrant investigation. Second, there are no 12. Swart E, Vasudeva E, Makhni EC, et al. Dedicated perioperative hip
data describing the specific protocols and criteria (eg, DVT fracture comanagement programs are cost-effective in high-volume cen-
prophylaxis and discharge milestones) used in the SHFPs ters: an economic analysis. Clin Orthop Relat Res. 2016;474:222–233.
13. ACS NSQIP: User Guide for the 2016 ACS NSQIP Procedure Targeted
described here. Therefore, we conclude from this study that Participant Use Data File (PUF). Chicago, IL: American College of
institution-specific comprehensive care pathways in general Surgeons; 2017:1–83.
improve short-term hip fracture outcomes and cannot advo- 14. Kates SL. Hip fracture programs: are they effective? Injury. 2016;47:
cate for one protocol in particular. This is an important S25–S27.
deficiency of the literature on this topic in general, which is 15. McNamara I, Sharma A, Prevost T, et al. Symptomatic venous throm-
boembolism following a hip fracture. Acta Orthop. 2009;80:687–692.
highly heterogeneous, and underscores the need for further 16. Geerts WH, Bergqvist D, Pineo GF, et al. Prevention of venous throm-
study to delineate what perioperative care pathways are boembolism: American College of Chest Physicians Evidence-Based
critical to achieving desired outcomes. Third, this was Clinical Practice Guidelines (8th edition). Chest. 2008;133(6 suppl):
a retrospective review and as such the 2 groups are not 381S–453S.
equivalent and matched groups, with differences in baseline 17. O’Malley NT, Blauth M, Suhm N, et al. Hip fracture management, before
and beyond surgery and medication: a synthesis of the evidence. Arch
comorbidities, functional health status, and race. However, Orthop Trauma Surg. 2011;131:1519–1527.
these baseline differences were accounted for using multivar- 18. Roberts KC, Brox WT. AAOS Clinical Practice Guideline. J Am Acad
iate logistic regression to compute risk-adjusted ORs in Orthop Surg. 2015;23:138–140.
driving conclusions. Another important consideration is the 19. Koval KJ, Friend KD, Aharonoff GB, et al. Weight bearing after hip
potential selection bias associated with such a study using the fracture: a prospective series of 596 geriatric hip fracture patients. J
Orthop Trauma. 1996;10:526–530.
NSQIP database. Indeed, the population of NSQIP- 20. Kubiak EN, Beebe MJ, North K, et al. Early weight bearing after lower
participating institutions are disproportionately academic extremity fractures in adults. J Am Acad Orthop Surg. 2013;21:727–738.
and/or Level 1 trauma centers,29 and these academic centers 21. Solomon DH, Johnston SS, Boytsov NN, et al. Osteoporosis medication
are more likely to have established SHFPs. This may con- use after hip fracture in U.S. patients between 2002 and 2011. J Bone
Min Res. 2014;29:1929–1937.
found and partially explain the outcome differences noted in
22. Kandel L, Schler D, Brezis M, et al. A simple intervention for improving
this study. Because the database provides deidentified data the implementation rate of a recommended osteoporosis treatment after
without reference to the characteristics of the treating hospi- hip fracture. Endocr Pract. 2013;19:46–50.
tal, further assessment of these potential confounders could 23. Sander B, Elliot-Gibson V, Beaton DE, et al. A coordinator program in
not be performed. Finally, ACS NSQIP reports only specific post-fracture osteoporosis management improves outcomes and saves
costs. J Bone Joint Surg. 2008;90:1197–1205.
perioperative data and complications. Although the data are 24. Pedersen SJ, Borgbjerg FM, Schousboe B, et al. A comprehensive hip
reliable and closely audited, relevant patient reported out- fracture program reduces complication rates and mortality. J Am Geriatr
comes are not available through this database. Soc. 2008;56:1831–1838.
25. Grigoryan KV, Javedan H, Rudolph JL. Orthogeriatric care models and
REFERENCES outcomes in hip fracture patients: a systematic review and meta-analysis.
1. Brauer CA, Coca-Perraillon M, Cutler DM, et al. Incidence and mortality J Orthop Trauma. 2014;28:e49–55.
of hip fractures in the United States. JAMA. 2009;302:1573–1579. 26. Bracey DN, Kiymaz TC, Holst DC, et al. An orthopedic-hospitalist
2. Hoang-Kim A, Busse JW, Groll D, et al. Co-morbidities in elderly pa- comanaged hip fracture service reduces inpatient length of stay. Geriatr
tients with hip fracture: recommendations of the ISFR-IOF hip fracture Orthop Surg Rehabil. 2016;7:171–177.
outcomes working group. Arch Orthop Trauma Surg. 2014;134:189– 27. Neuman MD, Archan S, Karlawish JH, et al. The relationship between
195. short-term mortality and quality of care for hip fracture: a meta-analysis
3. Schousboe JT, Paudel ML, Taylor BC, et al. Pre-fracture individual of clinical pathways for hip fracture. J Am Geriatr Soc. 2009;57:2046–
characteristics associated with high total health care costs after hip frac- 2054.
ture. Osteoporos Int. 2017;28:889–899. 28. Nikitovic M, Wodchis WP, Krahn MD, et al. Direct health-care costs
4. Burge R, Dawson-Hughes B, Solomon DH, et al. Incidence and eco- attributed to hip fractures among seniors: a matched cohort study. Os-
nomic burden of osteoporosis-related fractures in the United States, teoporos Int. 2013;24:659–669.
2005–2025. J Bone Miner Res. 2007;22:465–475. 29. Auspitz M, Cleghorn MC, Tse A, et al. Understanding quality issues in
5. Kates SL, Blake D, Bingham KW, et al. Comparison of an Organized your surgical department: comparing the ACS NSQIP with traditional
Geriatric Fracture Program to United States Government Data. Geriatr morbidity and mortality conferences in a Canadian academic hospital. J
Orthop Surg Rehabil. 2010;1:15–21. Surg Educ. 2015;72:1272–1277.
ORIGINAL ARTICLE
Factors Predictive of Blocking Screw Placement in

Retrograde Nailing of Distal Femur Fractures
Adam P. Schumaier, MD, Brendan R. Southam, MD, Frank R. Avilucea, MD, Ryan P. Finnan, MD,
John D. Wyrick, MD, Michael T. Archdeacon, MD, MSE, and Theodore T. Le, MD
addition, the BLOCK group had larger RABGs (BLOCK: 8.2 mm,
Objectives: (1) Identify factors that predict blocking screw NO BLOCK: 3.6 mm, P = 0.02), more SAFN (BLOCK: 51.4 mm,
placement in the treatment of a distal femur fracture with retrograde NO BLOCK: 39.8 mm, P = 0.02), and shorter distal segments
nail fixation and (2) determine whether acceptable alignment and (BLOCK: 1.7 · BCW, NO BLOCK: 2.0 · BCW, P = 0.01). In
stability were achieved in fractures that received blocking screws. a multivariable logistic regression, the combination of these factors
was significantly predictive of blocking screw placement with a large
Design: Retrospective Comparative Study.
effect size (R2 = 0.36, P , 0.01). A distal segment length # ·2
Setting: Level I Trauma Center. BCW was 77% sensitive for blocking screw placement, and a BMI
$25 kg/m2 was 70% sensitive. Negative predictive values for block-
Patients/Participants: Between 2011 and 2017, we identified 84 ing screw placement were distal segment length . ·2 BCW (79%),
patients with distal third femur fractures treated with a retrograde BMI ,25 kg/m2 (77%), RABG ,4 mm (76%), and SAFN ,50 mm
femoral nail. Data were analyzed according to those who did (71%). Patients that received blocking screws had acceptable post-
(BLOCK, n = 30) and did not (NO BLOCK, n = 54) receive block- operative alignment and stability, similar to fractures that did not
ing screws. Patients in both groups were treated by orthopaedic receive blocking screws (P . 0.05).
trauma surgeons; postoperative weight-bearing radiographs were ob-
tained of every patient. Conclusions: This retrospective study of distal femur fractures
treated with retrograde nails identified several factors that can be
Intervention: Fixation using a retrograde femoral nail with or without used to predict when blocking screw placement may be useful
blocking screws. Blocking screws were placed at the discretion of the for increasing stability and maintaining alignment in distal third
treating surgeon to reduce malaligned fractures or improve stability. femur fractures treated with retrograde IM nails. Patients treated
Main Outcome Measurements: (1) Demographics, radio- with blocking screws had a higher BMI, greater cortical bone
graphic apparent bone gap (RABG), space available for the nail loss, more SAFN, and shorter distal segments. There was no
(SAFN), and distal segment length [as a ratio of bicondylar width difference in postoperative alignment or stability between the 2
(BCW)]; and (2) post-operative alignment and stability (change in groups.
alignment over time). Key Words: distal femur fractures, intramedullary nails, blocking
screws, poller screws
Results: Patients treated with blocking screws had a higher body
mass index (BMI) (BLOCK: 29.0, NO BLOCK 25.7, P = 0.03). In Level of Evidence: Prognostic Level III. See Instructions for
Accepted for publication January 10, 2019. (J Orthop Trauma 2019;33:e229–e233)
From the Department of Orthopaedics and Sports Medicine, University of
Cincinnati, Cincinnati, OH.
B. R. Southam receives research support from DePuy; F. R. Avilucea is INTRODUCTION
a paid educational speaker for Zimmer-Biomet; J. D. Wyrick is a paid Modern intramedullary nailing (IMN) techniques began
speaker for Smith and Nephew and is a paid consultant for Stryker; M. T. in Germany in 1940s and have progressively become the
Archdeacon receives royalties and is a paid consultant for Stryker, standard of care for various long bone fractures.1 By virtue of
receives royalties, financial support, or material support from SLACK
incorporated, and is a board or committee member of the Orthopaedic the frictional forces generated at the bone-implant interface,
Trauma Association. The remaining authors report no conflict of interest. IMNs not only stabilize against bending and rotation but also
Presented in part at the Annual Meeting of the American Academy of impart fracture reduction when the IMN fills the canal at the
Orthopaedic Surgeons, New Orleans, Louisiana, March 8, 2018, and at fracture site. Conversely, malalignment is more likely when
the Annual Meeting of the MAOA, April 2018, San Antonio, TX. canal diameter is larger than nail diameter, a relationship
appear in the printed text and are provided in the HTML and PDF often encountered in metaphyseal fractures.2 Popularized by
versions of this article on the journal’s Web site (www.jorthotrauma. Krettek et al3 in 1999, blocking screws (Poller screws) can be
com). used as an adjunctive reduction tool to properly align an IMN
Reprints: Adam P. Schumaier, MD, Department of Orthopaedics and Sports in tibial and femoral metaphyseal fractures (Fig. 1, see Figure
Medicine, University of Cincinnati, PO Box 67012, Cincinnati, OH
45267 (e-mail: adam.schumaier@uc.edu).
1, Supplemental Digital Content 1, http://links.lww.com/
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. JOT/A670 for an additional figure demonstrating blocking
DOI: 10.1097/BOT.0000000000001450 screws stabilizing distal femur fractures).
Schumaier et al J Orthop Trauma Volume 33, Number 6, June 2019
Reporting of Observational Studies in Epidemiology).9 Over

a 6-year period (2011–2017), using current procedural termi-
nology code 27506, we identified 602 femoral shaft fractures
treated with an IMN. An initial review of the 602 fractures
revealed that 34 were proximal, 51 mid-proximal, 354 mid,
113 mid-distal, and 50 were distal. From the group of 163
mid-distal and distal fractures, 137 were treated with a retro-
grade nail. After full-length measurements of the 137 femurs,
we identified 124 distal third fractures that were treated with
a retrograde femoral nail. Of these 124, an additional 40
patients were excluded for the following reasons: treatment
by nontrauma surgeons, inadequate post–weight-bearing im-
ages, or supplemental plate fixation at distal fracture site. The
remaining 84 patients were subsequently grouped based on
those who did (BLOCK, n = 30) and did not (NO BLOCK, n
= 54) receive blocking screws. There were 4 patients initially
in the NO BLOCK group who subsequently developed
nonunions and were eventually treated with blocking screws.
Radiographically, these 4 patients were analyzed in the
BLOCK group. Reoperations for nonunions before blocking
screw placement were tallied in the NO BLOCK group, and
those after blocking screw placement were tallied in the
BLOCK group.
The electronic medical record was queried for several
variables including demographic information, injury specific
data, complications, and radiographs (postoperative and final
follow-up). Fractures were classified using the Association for
the Study of Internal Fixation/Orthopaedic Trauma Association
(OTA/AO) classification.10 The number of blocking screws
used and plane of placement was collected for each case using
the postoperative radiographs and operative notes. Per opera-
tive dictations, all blocking screws were placed with the intent
to correct malalignment or enhance stability. A threshold for
malalignment was not predetermined.
FIGURE 1. Blocking screws can be placed either before (top)
or after (bottom) placement of the intramedullary nail. The Radiographic Analysis
arrows signify that the fluoroscopic images were taken before The following radiographic parameters were measured:
and after blocking screw placement. 1. Radiographic apparent bone gap (RABG) (see Figure 2,
Supplemental Digital Content 2, http://links.lww.com/
Selective blocking screw placement has been reported as JOT/A671 for a figure that demonstrates how the RABG
a useful method for preventing or correcting malalignment in was measured): This is a measure of the average cortical
distal femur fractures treated with retrograde nails,4–6 a tech- bone loss and is a nonunion predictor.11,12 Full-thickness
nique that has led to development of blocking screw targeting cortical bone loss was measured on the medial, lateral,
guides.7,8 However, there remains a paucity of data reporting anterior, and posterior cortices and then averaged. Visible
patient or fracture characteristics that should hasten blocking fracture lines were measured as a 2-mm minimum.
screw placement. The primary purpose of this study was to 2. Space available for the nail (SAFN) (see Figure 3, Sup-
identify factors where blocking screws were implemented in plemental Digital Content 3, http://links.lww.com/JOT/
a successful manner for distal femoral fractures treated with A672 for a figure that demonstrates how the SAFN was
intramedullary nails. Secondarily, we sought to determine measured): This is the intramedullary canal width minus
whether fractures that were treated with blocking screw supple- the intramedullary nail width, measured adjacent to the
mentation had similar postoperative alignment and stability fracture site in the sagittal and coronal plane at both prox-
compared with those that were not treated with blocking screws. imal and distal segments. All measurements were perpen-
dicular to the anatomical axis of the femur.
3. Distal segment length (see Figure 4, Supplemental Dig-
METHODS ital Content 4, http://links.lww.com/JOT/A673 for a figure
that demonstrates how the distal segment length was mea-
Patients and Data Collection sured): This represents the average length of each cortex in
This is a retrospective comparative study that was the distal fracture segment (medial, lateral, anterior, and
performed per STROBE guidelines (Strengthening the posterior). It was measured parallel to the anatomical axis
J Orthop Trauma Volume 33, Number 6, June 2019 Factors Predictive of Blocking Screw Placement
from the nearest intact cortex to the perpendicular of the

TABLE 1. Demographics of the Study Sample (n = 84)
distal femoral condyles. The distal segment length was
normalized as a ratio of the bicondylar femoral width BLOCK NO BLOCK
(n = 30) (n = 54) P
(BCW).
4. Alignment: This is the angle between the main proximal Mean age (y) 43 (618) 41 (619) 0.79
and distal fragment in the sagittal and coronal planes, Sex (M:F) 19:11 41:13 0.22
measured using the anatomical axis. Measurements were BMI (kg/m2) 29.0 (66.2) 25.7 (67.3) 0.03*
taken on the immediate postoperative radiographs and at Obesity (.30 kg/m2) 11 (37%) 9 (17%) 0.03*
final follow-up. Overall, malalignment was calculated as Diabetes or PVD 4 (13%) 6 (11%) 0.74
the sum of malalignment in both planes.13 Tobacco use 16 (59%) 28 (52%) 0.90
All radiographic measurements were performed using Mean follow-up (mo) 6 (65) 8 (610) 0.29
Fiji (National Institutes of Health, Bethesda, MD).14,15 Meas- OTA/AO type
urements were performed twice; the reliability of the meas- A1, A2, A3 47%, 27%, 63%, 17%, 0.57
13% 11%
urements was calculated using an equal-weighted kappa,16
B1, B2, B3 0%, 3%, 0% 0%, 6%, 0%
and the measurement error was calculated as the difference
C1, C2, C3 3%, 3%, 3% 2%, 0%, 2%
between the first and second sets of measurements (mean 6
Mechanism of injury (%)
SD).
MVA 18 (60) 25 (46) 0.71
Statistical Analysis GSW 6 (20) 16 (30)
Fall 4 (13) 8 (15)
Patients were analyzed based on whether or not they
Other 2 (7) 5 (9)
received blocking screws (BLOCK, NO BLOCK). Continu-
Open fracture (%) 11 (37) 21 (39) 0.84
ous variables were compared using Student t test, and cate-
Additional weight-bearing fx (%) 14 (47) 22 (41) 0.59
gorical variables were compared using x2 analysis or Fisher
Additional non–weight-bearing 8 (27) 14 (26) 0.94
exact test. Single and multivariable logistic regression was fx (%)
used to quantify the variance predicted by each factor. The Admitted as polytrauma (%) 10 (33) 19 (35) 0.86
R2 values were reported using Nagelkerke’s estimation17 and Traumatic organ injury (%) 8 (27) 8 (15) 0.18
interpreted per Cohen18 [0.02 (small), 0.15 (medium), 0.35
(large)]. The receiver-operating characteristics of each factor PVD, peripheral vascular disease; MVA, motor vehicle accident; GSW, gunshot
wound; fx, fracture.
were reviewed for optimal cutpoints that maximize *P , 0.05.
sensitivity/specificity. Statistical analysis was performed
using R version 3.4.0 (R Foundation for Statistical Comput-
ing, Vienna, Austria) with RStudio version 1.0.153 (RStudio
Inc, Boston, MA).19,20 The following R packages were used:
irr [Gamer, Lemon, Fellows, and Singh (version 0.84)],21 Patients treated with blocking screws had a higher
nortest [Gross and Ligges (version 1.0-4)],22 BaylorEdPsych body mass index (BMI) (BLOCK: 29.0, NO BLOCK 25.7,
[Beaujean (version 0.5)],17 and OptimalCutpoints [Lopez- P = 0.03), and the rate of obesity (.30 kg/m2) was twice as
Raton and Rodriguez-Alvarez (version 1.1-3)].23 high in the BLOCK group (BLOCK: 37%, NO BLOCK:
17%, P = 0.03). In addition, the BLOCK group had larger
radiographic apparent bone gaps (BLOCK: 8.2 mm, NO
RESULTS BLOCK: 3.6 mm, P = 0.02), more SAFN (BLOCK:
There were 30 patients in the BLOCK group and 54 51.4 mm, NO BLOCK: 39.8 mm, P = 0.02), and shorter distal
patients in the NO BLOCK group. In total, 36 blocking segment fractures (BLOCK: 1.7 · BCW, NO BLOCK: 2.0 ·
screws were placed as a reduction aid at an average 2.3 cm BCW, P = 0.01) (Table 2). In a multivariable logistic regres-
from the fracture site: 23 patients received 1 blocking screw, sion (Table 3), the combination of these factors was signifi-
5 patients received 2 blocking screws, and 1 patient received cantly predictive of blocking screw placement with a large
3 blocking screws. Of the 36 blocking screws, 28 were effect size (X2 (4) = 25, R2 = 0.36, P , 0.01). In descending
placed in the sagittal plane, and 8 were placed in the coronal order of predictive value, the significant factors were RABG
plane. Of the 23 patients who had 1 blocking screw, 16 were (R2 = 0.18, P = 0.01), length of distal segment (R2 = 0.12, P =
placed in the sagittal plane and 7 were placed in the coronal 0.01), SAFN (R2 = 0.10, P = 0.01), and BMI (R2 = 0.07, P =
plane. Of the 5 patients who had 2 blocking screws, all were 0.04). A distal segment length # ·2 BCW was 77% sensitive
placed in the sagittal plane. The patient who received 3 for blocking screw placement, and a BMI $25 kg/m2 was
blocking screws had 2 placed in the sagittal plane and 1 in 70% sensitive. Negative predictive values for blocking screw
the coronal plane. All blocking screws were placed free- placement were distal segment length . ·2 BCW (79%),
hand. There was no significant difference in age, sex, BMI ,25 kg/m2 (77%), RABG ,4 mm (76%), and SAFN
comorbid medical conditions (diabetes or peripheral vascu- ,50 mm (71%) (Table 4).
lar disease), tobacco use, duration of follow-up, OTA/AO The plane of screw placement and number of screws
classification, mechanism of injury, or concurrent injury placed were not associated (P values .0.10) with the RABG,
frequency between the BLOCK and NO BLOCK groups SAFN, length of distal segment, or BMI. Fractures that
(Table 1). received blocking screws had similar postoperative alignment
Schumaier et al J Orthop Trauma Volume 33, Number 6, June 2019
TABLE 2. Radiographic Analysis and Fracture Alignment (n = 84)

BLOCK (n = 30) NO BLOCK (n = 54) P
Radiographic apparent bone gap (mm) 8.2 (69.2) 3.6 (62.5) 0.01
Space available for nail (mm) 51.4 (623.3) 39.8 (617.2) 0.02
Distal segment length/bicondylar width 1.7 (60.39) 2.0 (60.45) 0.01
Postoperative alignment 2.9 degrees (62.3 degrees) 3.7 degrees (63.2 degrees) 0.20
Final alignment 3.1 degree (63.0 degrees) 3.2 degrees (63.0 degrees) 0.80
Change in alignment 0.16 degrees (61.7 degrees) 0.44 degrees (61.4 degrees) 0.72, 0.09
Values are reported as mean (SD). Measurement error = 2.1 mm (63.8 mm) and 1.0 degrees (61.4 degrees).
(P . 0.05) and stability (P . 0.05) compared with those to the screws, a finding similar to what has been previously
which did not require blocking screws. Reliability of the reported.4,6,24,26
measurements was substantial (kappa = 0.85), and the mean There are several reports of blocking screws being
difference between the first and second sets of measurements used to treat nonunions. In 2009, Gao et al detailed 5
was 2.1 mm (63.8 mm) and 1.0 degree (61.4). There were femoral and 7 tibial nonunions treated with exchange
12 reoperations in the NO BLOCK group (8 nonunions and nailing and blocking screws. Bony union was achieved in
4 painful hardware removals) and 5 reoperations in the all patients.5 Prospective studies of small cohorts have
BLOCK group (4 nonunions and 1 painful hardware shown blocking screws maintain reduction until healing,
removal) (P . 0.05). possibly due to a reduction in deforming forces.4,6,27 In
a cadaveric study by Krettek et al,13 blocking screws
reduced the deformation of proximal and distal tibial shaft
DISCUSSION fractures by 25% and 57%, respectively, following appli-
This retrospective comparative study identified several cation of a 150-newton load. The effect of blocking screws
factors that were predictive of successful blocking screw on union and time to union remains unclear. Interestingly,
utilization: a higher BMI, larger RABGs, more SAFN, and a retrospective comparative study of 116 infraisthmal
shorter distal segments. These factors could be useful for femur fractures by Van Dyke et al28 did not find any differ-
preoperative planning and are not too surprising when one ences in union, alignment, or stability between fractures
considers situations that can increase the difficulty of nail that did or did not receive blocking screws; although the
reduction. Control of an intramedullary nail may be groups in this series did not differ in OTA/AO type, it is not
challenging in shorter bone segments, and patient position- clear if there were differences in BMI, fracture location,
ing or starting point acquisition may be more difficult in bone loss, or medullary canal space.
those who are obese. Fractures that occur where the This study has several limitations. Operative notes
medullary canal is wide are not likely to be reduced by the stated that the screws were placed to improve alignment or
nail itself because there is less cortical fit; similarly, bone stability, but definitive thresholds for a malaligned or unstable
gaps could diminish construct stability and make it difficult fracture could not be provided due to the retrospective nature
to judge alignment.
Blocking screws can be placed in the anteroposterior
plane to correct coronal deformity or the mediolateral plane to
correct sagittal deformity. The technique has been used to TABLE 4. Diagnostic Value of Radiographic and Demographic
treat tibial,3 femoral,4 humeral,24 and periprosthetic frac- Factors
tures25 as well as various nonunions.5 They are most effec- Sensitivity Specificity PPV NPV
tively placed on the concave side of the fracture deformity (%) (%) (%) (%)
before or after reaming but can also be placed after nail inser- Radiographic apparent
tion. The screws should be at least 1 centimeter from the bone gap
fracture site to avoid fracture extension.4 In our series, there $4 mm 57 76 57 76
were no cases of fracture extension or complications related Length of distal
segment/bicondylar
width
TABLE 3. Logistic Regression #2 77 48 45 79
R2 P Effect Size Space available for nail
$50 mm 47 72 48 71
Radiographic apparent bone gap 0.18 0.01 Medium
Body mass index
Length of distal segment/bicondylar 0.12 0.01 Small
width $25 kg/m2 70 58 49 77
Space available for nail 0.10 0.01 Small Optimal cutpoints are provided for the radiographic parameters based on receiver-
Body mass index 0.07 0.04 Small operating characteristics.
PPV, positive predictive value; NPV, negative predictive value.
Combined model 0.36 ,0.01 Large
J Orthop Trauma Volume 33, Number 6, June 2019 Factors Predictive of Blocking Screw Placement
of the study, and the timing of screw placement was not 12. Sanders DW, Bhandari M, Guyatt G, et al. Critical-sized defect in the
determined. Furthermore, this study did not evaluate func- tibia: is it critical? Results from the sprint trial. J Orthop Trauma. 2014;
28:632–635.
tional outcomes, and it did not evaluate union rates. Finally, 13. Krettek C, Miclau T, Schandelmaier P, et al. The mechanical effect of
there are several factors that could not be evaluated in this blocking screws (“Poller screws”) in stabilizing tibia fractures with short
study, such as the starting point of the nail and the accuracy of proximal or distal fragments after insertion of small-diameter intramedul-
reaming. lary nails. J Orthop Trauma. 1999;13:550–553.
14. Schindelin J, Rueden CT, Hiner MC, et al. The ImageJ ecosystem: an
In conclusion, this retrospective study of distal femur open platform for biomedical image analysis. Mol Reprod Dev. 2015;82:
fractures treated with retrograde nails identified several 518–529.
factors that can be used to predict when blocking screw 15. Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25
placement may be useful for increasing stability and main- years of image analysis. Nat Methods. 2012;9:671–675.
taining alignment in distal third femur fractures treated with 16. Landis JR, Koch GG. The measurement of observer agreement for cat-
egorical data. Biometrics. 1977;33:159–174.
retrograde IM nails. Patients treated with blocking screws had 17. Beaujean AA. BaylorEdPsych: R Package for Baylor University Educa-
a higher BMI, greater cortical bone loss, more SAFN, and tional Psychology Quantitative Courses. 2012. Available at: https://cran.
shorter distal segments. There was no difference in post- r-project.org/web/packages/BaylorEdPsych/BaylorEdPsych.pdf. Accessed
operative alignment or stability between the 2 groups. April 26, 2018.
18. Cohen J. A power primer. Psychol Bull. 1992;112:155–159.
19. R Core Team. R: A Language and Environment for Statistical Comput-
REFERENCES ing. Vienna, Austria: R Foundation for Statistical Computing; 2017.
1. Bong MR, Koval KJ, Egol KA. The history of intramedullary nailing. Available at: www.R-project.org. Accessed April 26, 2018.
Bull NYU Hosp Joint Dis. 2006;64:94–97. 20. RStudio Team. RStudio: Integrated Development Environment for R.
2. Rockwood CA, Green DP, Bucholz RW, eds. Rockwood and Green’s Boston, MA: RStudio, Inc; 2016. Available at: http://www.rstudio.
Fractures in Adults. 6th ed. Philadelphia, PA: Lippincott Williams & com/. Accessed April 26, 2018.
Wilkins; 2006. 21. Gamer Matthias, Lemon Jim, Fellows Ian, et al. Irr: Various Coefficients
3. Krettek C, Stephan C, Schandelmaier P, et al. The use of Poller screws as of Interrater Reliability and Agreement. 2012. Available at: https://
blocking screws in stabilising tibial fractures treated with small diameter CRAN.R-project.org/package=irr. Accessed April 26, 2018.
intramedullary nails. J Bone Joint Surg Br. 1999;81B:963–968. 22. Gross J, Ligges U. Nortest: Tests for Normality. 2015. Available at:
4. Ostrum RF, Maurer JP. Distal third femur fractures treated with retrograde https://cran.r-project.org/web/packages/nortest/nortest.pdf. Accessed
femoral nailing and blocking screws. J Orthop Trauma. 2009;23:681–684. April 26, 2018.
5. Gao K, Huang J, Li F, et al. Treatment of aseptic diaphyseal nonunion of 23. Lopez-Raton M, Xose Rodriguez-Alvarez M. OptimalCutpoints: Com-
the lower extremities with exchange intramedullary nailing and blocking puting Optimal Cutpoints in Diagnostic Tests. 2014. Available at: https://
screws without open bone graft. Orthop Surg. 2009;1:264–268. cran.r-project.org/web/packages/OptimalCutpoints/OptimalCutpoints.
6. Seyhan M, Cakmak S, Donmez F, et al. Blocking screws for the treat- pdf. Accessed April 26, 2018.
ment of distal femur fractures. Orthopedics. 2013;36:e936–941. 24. Stedtfeld H-W, Mittlmeier T, Landgraf P, et al. The logic and clinical
7. Advanced Orthopaedic Solutions: Tibia Nail System. 2009. Available at: applications of blocking screws. J Bone Joint Surg Am. 2004;86A(suppl
http://www.aquilantscientific.com/assets/aquilantorthopaedics/products/ 2):17–25.
procedure/821024/AOS_Tibial_ST.pdf. Accessed April 26, 2018. 25. Tonogai I, Hamada D, Goto T, et al. Retrograde intramedullary nailing
8. Smith and Nephew: Trigen Meta-Nail System. Available at: https://www. with a blocking pin technique for reduction of periprosthetic supracon-
smith-nephew.com/global/assets/pdf/temp/71181112_meta-nail_tibia_ dylar femoral fracture after total knee arthroplasty: technical note with
st_low_res_(copy-1).pdf. Accessed April 26, 2018. a compatibility chart of the nail to femoral component. Case Rep Orthop.
9. von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of 2014;2014:856853.
Observational Studies in Epidemiology (STROBE) statement: guidelines 26. Seyhan M, Unay K, Sener N. Comparison of reduction methods in intra-
for reporting observational studies. J Clin Epidemiol. 2008;61:344–349. medullary nailing of subtrochanteric femoral fractures. Acta Orthop
10. Marsh JL, Slongo TF, Agel J, et al. Fracture and dislocation classification Traumatol Turc. 2012;46:113–119.
compendium—2007: Orthopaedic Trauma Association classification, 27. Gavaskar AS, Chowdary N. Blocking screws: an adjunct to retrograde
database and outcomes committee. J Orthop Trauma. 2007;21(10 nailing for distal femoral shaft fractures. J Orthop Surg Hong Kong.
suppl):S1–S133. 2013;21:158–162.
11. Haines NM, Lack WD, Seymour RB, et al. Defining the lower limit of 28. Van Dyke B, Colley R, Ottomeyer C, et al. Effect of blocking screws on
a “critical bone defect” in open diaphyseal tibial fractures. J Orthop union of infraisthmal femur fractures stabilized with a retrograde intra-
Trauma. 2016;30:e158–e163. medullary nail. J Orthop Trauma. 2018;32:251–255.
ORIGINAL ARTICLE
Risk Factors for Infection After Intramedullary Nailing of

Open Tibial Shaft Fractures in Low- and Middle-Income
Countries
Paul S. Whiting, MD,* Daniel D. Galat, MD,† Lewis G. Zirkle, MD,‡ Michael K. Shaw, PhD,§
and Jeremiah D. Galatk
0.03) and from injury to wound closure (13.7 vs. 3.6 days, P ,
Objectives: (1) To determine the infection rate after fixation of 0.001). Distal fractures had a higher infection rate than midshaft
open tibial shaft fractures using the Surgical Implant Generation fractures (13.3% vs. 8.2%, P = 0.03). Infection rates were not asso-
Network (SIGN) intramedullary nail in low- and middle-income ciated with time from injury to initial debridement, time from injury
countries (LMICs) and (2) to identify risk factors for infection. to initial antibiotic administration, or total duration of antibiotics.
Design: Prospective cohort study using an international online Conclusions: Open tibia fractures can be managed effectively
database. using the SIGN intramedullary nail in LMICs with an overall
infection rate of 11.9%. Risk factors for infection identified include
Setting: Multiple hospitals in LMICs worldwide.
more severe soft-tissue injury, delayed nailing, delayed wound
Patients/Participants: A total of 1061 open tibia fractures treated closure, and distal fracture location.
with the SIGN nail in LMICs between March 2000 and February
Key Words: tibial shaft fracture, open fracture, complications, post-
2013.
operative infection, SIGN nail
Intervention: Intravenous antibiotic administration, surgical Level of Evidence: Therapeutic Level IV. See Instructions for
debridement, and definitive intramedullary nailing within 14 days
of injury.
Main Outcome Measurements: Deep or superficial infection at
follow-up, implant breakage/loosening, angular deformity .10 de-
INTRODUCTION
grees, repeat surgery, radiographic union, weight bearing, and ability
Tibial shaft fractures represent the most common long
to kneel.
bone fractures and follow an age- and sex-related bimodal
Results: The overall infection rate was 11.9%. Infection rates by distribution, with fractures commonly occurring in young
the Gustilo and Anderson classification were type 1: 5.1%, type II: males and elderly women.1–3 Road traffic accidents account
12.6%, type IIIa: 12.5%, type IIIb: 29.1%, and type IIIc: 16.7% (P = for more than half of all tibial shaft fractures; the remainder
0.001 between groups). Patients who developed infection had a lon- occur as the result of simple falls, sports-related injuries, or
ger mean time from injury to definitive surgery (4.7 vs. 3.9 days, P = direct trauma.2,3 Because of the subcutaneous location of the
anteromedial tibia, approximately 24% of tibia shaft fractures
are open, 60% of which are Gustilo and Anderson type III
Accepted for publication December 29, 2018. fractures.3,4 The classification system of Gustilo and Ander-
From the *Department of Orthopedics and Rehabilitation, University of Wis-
consin, Madison, WI; †Department of Orthopaedics, Kijabe Hospital, Ki- son,5 subsequently modified in 1984 by Gustilo et al,6 is used
jabe, Kenya, East Africa; ‡SIGN Fracture Care International, Richland, most commonly in classifying open fractures of the long bones.
WA; §Department of Medical Education, Ascension Health Michigan, In developed countries, intramedullary nail fixation has
Detroit, MI; and kArizona State University, Tempe, AZ. become the standard of care for open tibial shaft fractures,
L. G. Zirkle is President and Founder of SIGN Fracture Care International but
receives no financial compensation from the organization. The remaining largely replacing the use of casting and external fixation as
authors report no conflict of interest. There was no external source of definitive treatment methods.7,8 Intramedullary nailing of
funding for this study. long bone fractures offers several advantages to these tradi-
Presented in part at the Annual Meeting of the Orthopaedic Trauma tional techniques, including earlier mobilization and avoid-
Association, October 11, 2013, Phoenix, AZ. ance of complications associated with prolonged bed rest
appear in the printed text and are provided in the HTML and PDF such as pneumonia, venous thromboembolism, and decubitus
versions of this article on the journal’s Web site (www.jorthotrauma. ulcers.9–11 In developing countries, however, many barriers to
com). optimal orthopaedic trauma care exist, including the absence
Reprints: Paul S. Whiting, MD, Department of Orthopedics and Rehabilita- of trauma care delivery systems, a shortage of health care
tion, University of Wisconsin, 1685 Highland Ave, MFCB 6227,
Madison, WI 53705 (e-mail: whiting@ortho.wisc.edu).
workers, a lack of education and training, and inadequate
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. health care facilities/resources.12 Consequently, available
DOI: 10.1097/BOT.0000000000001441 equipment and techniques, such as prolonged traction or the
J Orthop Trauma Volume 33, Number 6, June 2019 SIGN Nail Open Tibia Fractures
“pin-in-plaster” method, are often used in the treatment of

open long bone fractures.13
The Surgical Implant Generation Network (SIGN)
nailing system was specifically designed for use in
resource-limited settings and does not require intraoperative
fluoroscopy, power instrumentation, or specialized operating
tables, equipment that is not universally available in the
developing world.14 This implant, designed for the tibia, has
been used to perform intramedullary fixation of femur, tibia,
and humerus fractures. The solid construction of the nail
makes it stronger than a cannulated nail of the same diameter
and results in less surface area for microbial adherence. Frac-
ture reduction is achieved by closed or open means, hand
reamers are used to prepare the canal if desired, and distal
interlocking is accomplished by means of an extended distal
targeting arm (see Figure, Supplemental Digital Content 1,
http://links.lww.com/JOT/A659). Further details of the SIGN
nail system and insertion technique have been published
previously.15,16
Several studies have documented the utility of the
SIGN nail for management of tibial shaft fractures in the
developing world,17–20 and a small case series of open tibial
shaft fractures treated with SIGN intramedullary nailing in
a resource-limited setting reported an acceptably low infec-
tion rate.21 To date, however, there has been no large-scale
study investigating outcomes after SIGN intramedullary nail-
ing of open tibial shaft fractures in the developing world.
The objectives of this investigation were (1) to determine
the infection rate after fixation of open tibia fractures using
the SIGN intramedullary nail in low- and middle-income
countries (LMICs) and (2) to identify risk factors for infection
in this patient population.
METHODS
SIGN Program and Database Characteristics
The SIGN Online Surgical Database (SOSD) in-
cludes prospectively collected data entered by surgeons at
all SIGN sites worldwide. Such data collection and
reporting is a requirement of each site to maintain a steady
supply of new implants. Data collected and entered into the
SOSD at the time of admission and initial operative
intervention include preoperative x-rays, Gustilo and
Anderson open fracture type (if applicable), fracture
location, time from injury to intravenous antibiotic admin-
istration, and time from injury to initial surgical debride-
ment. Data collected at the time of intramedullary nailing
include surgical reduction technique (open or closed), nail
insertion technique (antegrade or retrograde), C-arm utili-
zation (yes or no), time from injury to nailing, time from
injury to skin closure, and total duration of intravenous
antibiotics. Postoperative x-rays are obtained immediately
after surgery, and follow-up x-rays are generally obtained
at all postoperative outpatient visits. Clinical follow-up
data collected in the SOSD include weight-bearing status, FIGURE 1. Flowchart demonstrating application of the
ability to kneel, presence or absence of deep infection, and exclusion criteria to the patients in the open tibia fracture
need for repeat surgery. cohort.
Whiting et al J Orthop Trauma Volume 33, Number 6, June 2019
Data Extraction and Inclusion/Exclusion

TABLE 1. Infection Rates by Gustilo and Anderson Open
Criteria Fracture Type
Access to the SOSD was granted by SIGN Fracture Care Gustilo/ Patients Patients
International. On February 1, 2013, an export of the SOSD Anderson Open With Without Infection
without patient-identifying information was performed. All Fracture Type Infection Infection Rate, % P
open tibia fractures treated with the SIGN intramedullary nail Type I 20 373 5.1 0.001*
in World Bank–defined low- and middle-income countries Type II 50 348 12.6 ,0.001
from March 2000 (initiation of SOSD) through February 1, Type IIIa 26 182 12.5 0.002
2013, were eligible for inclusion in the study. Cases with no Type IIIb 16 39 29.1 ,0.001
recorded follow-up visit, fractures treated more than 14 days Type IIIc 1 5 16.7 0.25
from injury, duplicate cases, and cases with erroneous dates or
incomplete data for antibiotic administration were excluded Statistically significant p-values (,0.05) are indicated in bold.
*An ANOVA test was used to compare the infection rate in type I fractures with
from the analysis. A database export was again performed on rates in type II and III fractures. Infection rates in type II and each type III fracture
August 1, 2013, allowing a minimum of 6 months for the most subtype were then compared pairwise with the infection rate in type I fractures.
recent fractures to return for follow-up.
Primary and Secondary Outcome Measures (P , 0.01 for each comparison, P = 0.001 analysis of vari-
and Data Analysis ance (ANOVA) for entire group) but not type IIIc fractures (P
The primary outcome measure was infection reported at = 0.25). Infection rates by Gustilo and Anderson open frac-
any follow-up visit. Infection rates by Gustilo and Anderson ture type are displayed in Table 1.
fracture type were also calculated. We considered any infection
reported at follow-up to be clinically significant and therefore did Secondary Outcome Measures
not differentiate between deep and superficial infections. Sec- Patients who developed infection were significantly
ondary outcome measures included implant breakage/loosening, more likely to develop loosening of the nail (3.5% vs. 0.1%,
angular deformity .10 degrees, repeat surgery, radiographic P = 0.001) or screw (9.7% vs. 0.8%, P , 0.001) and were 15
union, partial or painless weight bearing, and ability to kneel. times more likely to require repeat surgery (25.7% vs. 1.7%,
Using a univariate analysis, variables recorded in the P , 0.001) than those without infection (Table 2). Patients
SOSD were analyzed to identify risk factors for infection. with infection were less likely to achieve painless weight
Time from injury to initial antibiotic administration was bearing at most recent follow-up (37.2% vs. 63.9%, P ,
divided into the following time intervals: ,6 hours, 6–24 0.001). There were no differences in rates of screw or nail
hours, 24–48 hours, 2–10 days, and .10 days. Chi-square breakage, angular deformity .10 degrees, radiographic
or Fisher exact tests were performed where appropriate for union, percentage achieving partial weight bearing, and abil-
categorical variables; t-tests were used for continuous varia- ity to kneel between patients with and without infection.
bles. All tests were 2 tailed, and significance was set at P ,
0.05. A multivariable logistic regression analysis, controlling
for age and sex, was then performed for all risk factors iden- TABLE 2. Secondary Outcome Measures and Intraoperative
tified in the univariate analysis. Variables Among Patients With and Without Infection
With Infection, Without
% Infection, % P
RESULTS % Screw breakage 0.0 0.7 0.71
During the study period, a total of 68,212 fractures % Screw loosening 9.7 0.8 ,0.001
treated with the SIGN nail in LMICs were entered into the % Nail breakage 0.0 0.1 1.0
database. Forty-one percent of these patients returned for at % Nail loosening 3.5 0.1 0.001
least 1 follow-up visit. As shown in Figure 1, the database % Deformity .10 3.5 0.1 0.21
contained a total of 7285 open tibia fractures treated with the degrees
SIGN nail, and 2887 of these patients (39.6%) returned for at % Repeat surgery 25.7 1.7 ,0.001
least 1 follow-up visit. After excluding patients with no re- Reaming method 93.6 88.5 0.78
corded follow-up visit, fractures that underwent IM nailing (% hand)
more than 14 days after injury, duplicate cases, and cases with Fracture reduction 87.6 73.4 0.24
(% open)
erroneous dates or incomplete data for antibiotic administra-
% Partial weight 73.5 61.4 0.28
tion, 1061 patients were included in the final analysis. The bearing
average duration of follow-up was 19.9 weeks. % Painless weight 37.2 63.9 ,0.001
bearing
Primary Outcome Measure % Union on 63.7 74.9 0.35
The overall infection rate for open tibia fractures treated radiographs
with the SIGN nail in LMICs was 11.9%. When compared % Can kneel 87.6 93.3 0.72
with type I fractures, infection rates were significantly higher Statistically significant p-values (,0.05) are indicated in bold.
for Gustilo and Anderson type II, IIIa, and IIIb fractures
TABLE 3. Timing From Injury to Debridement, Nailing, and TABLE 5. Infection Rates by Fracture Location
Wound Closure and Duration of Antibiotics Among Patients Without With Infection P P (vs.
With and Without Infection Infection Infection Rate (ANOVA) Distal)
With Without Proximal 123 17 12.1% 0.34 0.88
Infection Infection P fracture
n 113 948 Shaft 493 44 8.2% 0.03
Age 35.5 6 14.0 33.4 6 13.1 0.11 fracture
% Male 80.5% 82.3% 0.94 Distal 312 48 13.3%
fracture
Mean time to surgery (d) 4.7 6 3.9 3.9 6 3.7 0.03
% With antibiotics 100% 99.4% 1.0 Statistically significant p-values (,0.05) are indicated in bold.
Mean duration of antibiotic 10.3 6 12.1 9.7 6 13.2 0.64
administration (d)
Injury to debridement time 25.8 6 54.9 20.6 6 45.7 0.27 DISCUSSION
(hours)
Infection rates reported in the literature after reamed
Injury to closure (d) 13.7 6 9.4 3.6 6 4.5 ,0.001
intramedullary nailing of open tibial shaft fractures in the
Time to follow-up (d) 165.4 6 135.8 6 196.9 0.13
226.6 developed world range from 4.3% to 31%.22–29 Keating
et al22 reported only 2 deep infections in 47 open tibia frac-
Statistically significant p-values (,0.05) are indicated in bold. tures (4.3%) treated with immediate reamed nailing, whereas
Wiss and Stetson23 reported an infection rate of 24% among
33 type I and II fractures, and Craig et al24 reported infection
As shown in Table 3, patients who developed infection rates of 14.4% for type IIIa and 31% for type IIIb and IIIc
had a significantly longer mean time from injury to surgery (4.7 fractures. In a large case series of over 1100 tibial shaft frac-
vs. 3.9 days, P = 0.03) and mean time from injury to wound tures treated with reamed intramedullary nailing, Court-
closure (13.7 vs. 3.6 days, P , 0.001) compared with those who Brown25 reported 19 infections among the 247 open fractures
did not develop infection. Infection rates were not associated treated, representing an overall infection rate of 7.7%. Infec-
with mean time from injury to initial debridement or total duration rates by Gustilo open fracture type were 6.9% for type 1,
tion of antibiotic treatment. Data on antibiotic administration are 6.6% for type II, 0% for type IIIa, and 16.4% for type IIIb.
displayed in Table 4. Time from injury to initial antibiotic Petrisor et al26 investigated the 19 cases of infection reported
administration was not associated with infection rates. As shown in the previous study and found that 10 of them (52.5%) were
in Table 5, distal fractures had a significantly higher infection attributable to late complications or failures of soft-tissue
rate than midshaft fractures (13.3% vs. 8.2%, P = 0.03) but not coverage for the open fracture wounds.
proximal fractures (13.3% vs. 12.1%, P = 0.88). Kakar and Tornetta8 developed a protocol for the treat-
Results of the multivariable logistic regression analysis ment of type I-IIIb open tibia fractures consisting of immedi-
are displayed in Table 6. After controlling for age and sex, ate irrigation, debridement, and unreamed tibial nailing,
increased time from injury to surgery was associated with followed by repeat irrigation and debridement every 2–3 days
a significantly higher infection rate [odds ratio (OR) 1.06, until delayed closure or flap coverage could be accomplished.
95% CI (1.001–1.13), P = 0.048]. Similarly, increasing Gus- Use of this protocol resulted in an overall infection rate of
tilo and Anderson fracture type was also associated with an 2.8% among 143 open tibial shaft fractures. Although one
increased risk of infection (OR 1.466, 95% CI (1.15–1.87), P hypothesis driving the development of this protocol was that
= 0.002]. Although there remained a trend toward higher reamed intramedullary nailing was associated with higher
infection rates with increased time from injury to wound clo- rates of infection than unreamed nailing in the treatment of
sure and distal fracture location, these differences did not open fractures, results of the Study to Prospectively Evaluate
meet statistical significance. Reamed Intramedullary Nails in Patients with Tibial Fractures
TABLE 4. Infection Rates Stratified by Time From Injury to

Initial Antibiotic Administration TABLE 6. Multivariable Logistic Regression Analysis
With Without Infection Rate, 95% Confidence
Infection Infection % P OR Interval P
,6 h 51 380 11.8 0.74* Male sex 1.006 0.56–1.80 0.98

Age (by decade) 1.010 0.86–1.19 0.90
6 to ,24 h 36 336 9.7 0.42
24 to ,48 5 50 9.1 0.82 Injury to surgery (d) 1.061 1.001–1.13 0.048
h Fracture location (middle vs. all 0.666 0.43–1.04 0.072
other locations)
48 h–10 d 19 158 10.7 0.78
Injury to closure (d) 1.029 0.99–1.07 0.15
.10 d 2 20 9.1 0.78
Gustilo–Anderson open frac- 1.466 1.15–1.87 0.002
*An ANOVA test was used to compare the infection rate in the ,6-hour group with ture type
rates in all other groups. Infection rates in the other groups were then compared pairwise
with the infection rate in the ,6-hour group. Statistically significant p-values (,0.05) are indicated in bold.
Whiting et al J Orthop Trauma Volume 33, Number 6, June 2019
(SPRINT trial) demonstrated no significant difference in com- to a threshold of 5%, but further increases in the overall follow-
posite outcome between reamed and unreamed nails for open up rate did not lead to higher reported rates of infection. The
fractures.30 Interpreted together, these data suggest that early authors concluded that the reported infection rates in the SOSD
bony stabilization and restoration of a healthy soft-tissue are reliable as long as follow-up exceeded the 5% threshold. Our
envelope are critically important for avoiding infections after average duration of follow-up (19.9 weeks) is in keeping with
open tibial shaft fractures. the average duration of follow-up reported in the largest pub-
Based on our database-wide analysis of open tibia lished study of tibia fractures treated with the SIGN nail in
fractures treated with the SIGN intramedullary nail in LMICs, LMICs (19.1 weeks).20 Although some patients may have devel-
we report an overall infection rate of 11.9%. Only two other oped complications (including infection) at a later date and
published studies from LMICs report infection rates after SIGN therefore would not have been counted in our overall infection
nailing of open tibia fractures. Shah et al21 reported an overall rate, having returned for at least one follow-up appointment
infection rate of 8.3% in 36 open tibia fractures treated with the demonstrates that all patients in our final cohort were at least
SIGN nail, and Stephens et al20 reported an 18.5% infection rate capable of returning to the index hospital if needed.
among the 65 open fractures in their multicenter study of 162 A formidable barrier to better rates of follow-up in
distal tibial metaphyseal fractures treated with the SIGN nail. developing countries is the significant cost associated with
When calculating our overall infection rate, we included patients routine follow-up visits for most patients.37 The ubiquity of
who developed superficial infections in addition to those who fee-for-service models in the developing world, coupled with
had deep infections. Although this may have overestimated the the lack of health insurance coverage for outpatient services,
true rate of clinically significant surgical site infection, given the translates into a significant financial burden for patients, who
absence of a universal or consistent definition of deep infection often must pay for their own clinic visit fees, x-rays, and any
among the many treating surgeons, we felt this was the most costs associated with travel. Consequently, patients without
reliable method of capturing all pertinent infections. complications have a disincentive to return for scheduled
In keeping with previous literature,31,32 our study identified follow-up visits. Therefore, because many patients with favor-
higher infection rates with increasing severity of soft-tissue injury able outcomes never return for follow-up, the true infection rate
as represented by the Gustilo and Anderson open fracture clas- for our cohort may be lower than the reported rate of 11.9%.
sification system. This risk factor remained statistically significant Our study also identified areas of potential improvement
on our multivariable analysis. Distal fracture location (compared in international database design. The inclusion of additional
with midshaft) was associated with higher rates of infection in outcome variables in the SOSD, along with modifications to
univariate analysis, a finding corroborated by Stephens et al.20 In the existing variables, represent opportunities for improvement
our multivariable regression analysis, distal fracture location in the quality and accuracy of data collection and analysis in
showed a trend toward higher risk of infection, but this no longer international orthopaedic research. More than 200,000 SIGN
met statistical significance (P = 0.072, Table 6). nails have been used for fixation of long bone fractures in the
Other risk factors for infection identified in our univar- developing world to date. The SOSD holds tremendous
iate analysis included time from injury to definitive intra- potential for further large-scale analyses of long bone fracture
medullary nailing and to wound closure. Time from injury to treatment. It also has the potential to serve as a model for the
nailing remained a statistically significant risk factor for development of a prospective, multicenter, international data-
infection in our multivariable analysis. The fact that the base within orthopaedic trauma. Furthermore, the SIGN
association between infection rates and increased time from network represents a strategic resource for international
injury to wound closure no longer met statistical significance in collaboration aimed at improving the management of muscu-
our multivariable analysis is likely related to confounding with loskeletal injuries in the developing world through education,
the “Gustilo and Anderson fracture type” variable. Type IIIb training, and research.38
fractures, which have the highest infection rates, are also most
likely to result in delayed wound closure. In a recent study by
Jenkinson et al,33 delayed wound closure was found to be a risk ACKNOWLEDGMENTS
factor for infection in open fractures. Using a propensity score- The authors thank the surgeons at SIGN sites world-
matching algorithm, the authors reported infection in 17.8% of wide who work tirelessly to treat musculoskeletal trauma and
fractures treated with delayed closure versus 4.1% in fractures carefully record data in the SIGN Online Surgical Database,
treated with immediate primary closure, representing a relative making it a useful tool for improving orthopaedic trauma
risk of 11.0 for delayed closure. In our study, timing from care through research.
injury to IV antibiotic administration and surgical debridement
were not associated with infection rates, although the impor-
tance of these variables has been demonstrated previously.34,35 REFERENCES
1. Court-Brown CM, Caesar B. Epidemiology of adult fractures: a review.
One limitation of the current study is the low follow-up Injury. 2006;37:691–697.
rate (39.6% overall). However, there is evidence to suggest that 2. Court-Brown CM, McBirnie J. The epidemiology of tibial fractures. J
rates of reported infection in the SOSD are reliable even with Bone Joint Surg Br. 1995;77:417–421.
limited follow-up. Young et al36 reviewed over 34,000 long 3. Court-Brown CM, Rimmer S, Prakash U, et al. The epidemiology of
open long bone fractures. Injury. 1998;29:529–534.
bone fractures treated in 55 countries with the SIGN nail. Using 4. Melvin JS, Dombroski DG, Torbert JT, et al. Open tibial shaft fractures:
a generalized additive regression model, the authors found that I. Evaluation and initial wound management. J Am Acad Orthop Surg.
reported rates of infection increased as follow-up rates increased 2010;18:10–19.
5. Gustilo RB, Anderson JT. Prevention of infection in the treatment of one 22. Keating J, O’Brien P, Blachut P, et al. Locking intramedullary nailing
thousand and twenty five open fractures of long bones. J Bone Joint Surg with and without reaming for open fractures of the tibial shaft: a pro-
Am. 1976;58-A:453–458. spective, randomized study. J Bone Joint Surg Am. 1997;79:334–341.
6. Gustilo RB, Mendoza RM, Williams DN. Problems in the management 23. Wiss D, Stetson W. Unstable fractures of the tibia treated with a reamed
of type III (severe) open fractures: a new classification of type III open intramedullary interlocking nail. Clin Orthop. 1995;315:56–63.
fractures. J Trauma. 1984;24:742–746. 24. Craig J, Fuchs T, Jenks M, et al. Systematic review and meta-analysis of
7. Melvin JS, Dombroski DG, Torbert JT, et al. Open tibial shaft fractures: the additional benefit of local prophylactic antibiotic therapy for infection
II. Definitive management and limb salvage. J Am Acad Orthop Surg. rates in open tibia fractures treated with intramedullary nailing. Int Or-
2010;18:108–117. thop. 2014;38:1025–1030.
8. Kakar S, Tornetta P III. Open fractures of the tibia treated by immediate 25. Court-Brown CM. Reamed intramedullary tibial nailing: an overview
intramedullary tibial nail insertion without reaming: a prospective study. and analysis of 1106 Cases. J Orthop Trauma. 2004;18:96–101.
J Orthop Trauma. 2007;21:153–157. 26. Petrisor B, Anderson S, Court-Brown CM. Infection after reamed intra-
9. Giannoudis PV, Papakostidis C, Roberts C. A review of the management medullary nailing of the tibia: a case series review. J Orthop Trauma.
of open fractures of the tibia and femur. J Bone Joint Surg Br. 2006;88: 2005;19:437–441.
281–289. 27. Bone LB, Johnson KD. Treatment of tibial fractures by reaming and
10. Seibel R, LaDuca J, Hassett JM, et al. Blunt multiple trauma (ISS 36), intramedullary nailing. J Bone Joint Surg Am. 1986;68:877–887.
femur traction, and the pulmonary failureseptic state. Ann Surg. 1985; 28. Court-Brown C, McQueen M, Quaba A, et al. Locked intramedullary
202:283–295. nailing of open tibial fractures. J Bone J Surg Br. 1991;73:959–964.
11. Bone LB, Johnson KD, Weigelt J, et al. Early versus delayed stabiliza- 29. Jenny J, Jenny G, Gaudias J, et al. Risk of infection in centro-medullary
tion of femoral fractures. A prospective randomized study. J Bone Joint locking nailing of open fractures in the femur and tibia. Acta Orthop
Surg Am. 1989;71:336–340. Belg. 1995;61:212–215.
12. Spiegel DA, Gosselin RA, Coughlin RR, et al. The burden of 30. Study to Prospectively Evaluate Reamed Intramedullary Nails in Patients
musculoskeletal injury in low and middle-income countries: with Tibial Fractures (SPRINT) Investigators, Bhandari M, Guyatt G,
challenges and opportunities. J Bone Joint Surg Am. 2008;90: et al. Randomized trial of reamed and unreamed intramedullary nailing
915–923. of tibial shaft fractures. J Bone Joint Surg Am. 2008;90:2567–2578.
13. Bassey LO. Open fractures of the femur treated by the pin-in-plaster 31. Papakostidis C, Kanakarisn NK, Pretel J, et al. Prevalence of complica-
technique. Contribution to the art and practice of trauma surgery in the tions of open tibia shaft fractures stratified as per the Gustilo-Anderson
Third World. Arch Orthop Trauma Surg. 1990;109:139–143. classification. Injury. 2011;42:1408–1415.
14. Phillips J, Zirkle LG, Gosselin RA. Achieving locked intramedullary 32. Thakore RV, Francois EL, Nwosu SK, et al. The Gustilo-Anderson
fixation of long bone fractures: technology for the developing world. classification system as predictor of nonunion and infection in open tibia
Int Orthop. 2012;36:2007–2013. fractures. Eur J Trauma Emerg Surg. 2017;43:651–656.
15. Feibel RJ, Zirkle LG. Use of interlocking intramedullary tibial nails in 33. Jenkinson RJ, Kiss A, Johnson S, et al. Delayed wound closure increases
developing countries. Tech Orthop. 2009;24:239. deep-infection rate associated with lower-grade open fractures: a propen-
16. Zirkle LG, Shahab F, Shahabuddin. Interlocked intramedullary nail with- sity-matched cohort study. J Bone Joint Surg Am. 2014;96:380–386.
out fluoroscopy. Orthop Clin North Am. 2016;47:57–66. 34. Penn-Barwell JG, Murray CK, Wenke JC. Early antibiotics and debride-
17. Ikpeme I, Ngim N, Udosen A, et al. External jig-aided intramedullary ment independently reduce infection in an open fracture model. J Bone
interlocking nailing of diaphyseal fractures: experience from a tropical Joint Surg Br. 2012;94:107–112.
developing centre. Int Orthop. 2011;35:107–111. 35. Hull PD, Johnson SC, Stephen DJ, et al. Delayed debridement of severe
18. Khan I, Javed S, Khan GN, et al. Outcome of intramedullary interlocking open fractures is associated with a higher rate of deep infection. Bone
SIGN nail in tibial diaphyseal fracture. J Coll Physicians Surg Pak. 2013; Joint J. 2014;96-B:379–384.
23:203–207. 36. Young S, Lie SA, Hallan G, et al. Low infection rates after 34,361
19. Zain-ur-Rehman M, Khan RDA, Saeed UB, et al. Clinical outcome of intramedullary nail operations in 55 low- and middle-income countries:
patients with isolated tibial shaft fractures treated with SIGN interlock validation of the Surgical Implant Generation Network (SIGN) online
nails in tems of surgical site infection and radiological bone healing on surgical database. Acta Orthop. 2011;82:737–743.
follow up. J Pak Med Assoc. 2015;65:S175–S178. 37. Shearer D, Cunningham B, Zirkle L. Population characteristics and clin-
20. Stephens KR, Shahab F, Galat D, et al. Management of distal tibial ical outcomes from the SIGN online surgical database. Tech Orthop.
metaphyseal fractures with the SIGN intramedullary nail in 3 developing 2009;24:273–276.
countries. J Orthop Trauma. 2015;29:e469–e475. 38. Whiting PS, Anderson DR, Galat DD, et al. State of pelvic and acetabular
21. Shah RK, Moehring HD, Singh RP, et al. Surgical implant generation surgery in the developing world: a global survey of orthopaedic surgeons
network (SIGN) intramedullary nailing of open fractures of the tibia. Int at surgical implant generation network (SIGN) hospitals. J Orthop
Orthop. 2004;28:163–166. Trauma. 2017;31:e217–e223.
ORIGINAL ARTICLE
Biomechanical Comparison of Tension Band Fixation of

Patella Transverse Fracture: Headless Screws Versus Headed
Screws
Jill M. Martin, MD,* Dylan T. Applin, BS,* Linda M. McGrady, BS,*†
Mei Wang, PhD,*† and Gregory J. Schmeling, MD*†
fixation with higher construct rigidity, smaller interfragmentary

Objective: This study aimed to investigate the stability and motion, and greater fixation strength.
strength of tension band wire fixation using headless compression
screws versus headed screws for transverse patella fractures. Key Words: patella fracture, tension band wiring, headless screws,
interfragmentary movement, biomechanics
Methods: Six matched pairs of fresh-frozen cadaveric knees with
transverse osteotomies created at the midpoint of the patella were (J Orthop Trauma 2019;33:e240–e245)
surgically fixed, with one knee randomly receiving fixation with
headless screws (Acumed Acutrak 4/5) and the other with headed INTRODUCTION
screws (Synthes 4.0 partially threaded cannulated screws). The Fractures of the patella constitute approximately 1% of
specimens were mounted onto a servohydraulic load frame in a 45- all skeletal fractures, among which transverse fractures are the
degree flexed position and loaded through the quadriceps tendon. most common.1 The patella is a sesamoid bone embedded
Interfragmentary movement was recorded with a motion analysis within the tendon of the knee extensor and acts as a guide
system. The initial fixation stiffness, range of interfragmentary that centralizes the divergent forces from the 4 quadriceps
motion, and strength of the headless screw construct were compared muscle.2,3 Mechanically, it functions as a fulcrum that in-
with the headed screw construct. Failure was defined as either creases the moment arm of the quadriceps muscle with
a sudden drop in applied tendon force or 2 mm of separation on the respect to the center of rotation of the knee and, therefore,
anterior surface of the patella (ie, clinical failure), whichever increases the efficiency of the extensor mechanism.4,5 In vitro
occurred first. cadaver studies found that pulling the quadriceps at 100–200
N was enough to extend the knee against gravity in most
Results: Mean primary interfragmentary motion was 0.31 6 0.28 cases, and the resulting compressive patellofemoral force
degrees for the headed screws and 0.10 6 0.06 degrees for headless
was less than 50% of bodyweight.6–8 However, daily activi-
screws under 150 N load (P = 0.03). Mean construct stiffness was
ties, especially ones with the center of gravity placed posterior
277 6 243 N/degrees for the headed screws and 510 6 362 N/
to the knee, such as squatting, impose high patellofemoral
degrees for the headless screws (P = 0.03). None of the constructs
compressive forces up to 9 times bodyweight.2,9 Transverse
from either group displayed structural failure before reaching the
patella fractures can cause complete disruption of the extensor
clinical failure gap of 2 mm. The mean clinical failure strength
mechanism. Surgical fixation is recommended when frag-
was 808 6 183 N for the headless screws construct and
ments are displaced over 2 mm, with step-off involving the
520 6 241 N for the headed screws construct (P = 0.03).
joint surface, or with a disrupted extensor mechanism.1,10
Conclusions: Headless screw tension band fixation demonstrated Because of the high-loading environment of the patella, ten-
superior biomechanical behaviors over standard headed screw sion band wiring with K-wires became the standard fixation
to protect against displacement. More recent literature sup-
ports the use of tension band wiring passed through cannu-
Accepted for publication January 8, 2019. lated screws, as this construct has been shown to improve
From the *Department of Orthopaedic Surgery, Medical College of Wisconsin,
Milwaukee, WI; and †Orthopaedic and Rehabilitation Engineering Center, fixation strength and reduce the clinical failure rate compared
Marquette University, Milwaukee, WI. with the traditional modified tension band wiring with
This study was funded in part by an intramural research grant from the K-wires.11–14
Medical College of Wisconsin Department of Orthopaedic Surgery. Despite being a widely performed technique, modified
Implants were donated by Synthes and by Froedtert Memorial Lutheran tension band fixation has been associated with high rates of
Hospital.
The authors report no conflict of interest. postoperative complications, many related to implant prom-
Presented as a poster at the Annual Meeting of Orthopaedic Research Society, inence and soft-tissue irritation. Symptomatic implant rates
March 19–22, 2017, San Diego, CA. have been reported at 20%–60% and often require a second
Reprints: Mei Wang, PhD, Department of Orthopaedic Surgery, Medical procedure for implant removal.5,15–18 Although the tension
College of Wisconsin, 9200 West Wisconsin Avenue, Milwaukee, WI
53226 (e-mail: meiwang@mcw.edu).
band wire is the typical culprit of pain, prominent screw heads
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. are another potential source of postoperative irritation. Newer
DOI: 10.1097/BOT.0000000000001447 headless compression screws, which provide interfragmentary
J Orthop Trauma Volume 33, Number 6, June 2019 Headless Compression Screws in Patella Fracture Fixation
compression through a variable-pitch thread and tapered core, This flexed knee testing position was chosen to replicate the
may be a viable alternative to reduce painful implant compli- peak patella loading condition during extension. Previous
cation rates if they are shown to produce similar fixation as studies demonstrated that the midanterior longitudinal patellar
conventional screws. strain reaches its peak value at 45–60 degrees of knee flex-
The objective of this study was to determine whether ion19 and the patellofemoral joint reaction force reaches its
tension band fixation using headless compression screws maximum between 50 and 60 degrees of flexion.6,7 The
produce similar levels of stability and resistance to failure flexed knee position was reinforced with a pair of length-
compared with traditional cannulated screws in a noncommin- adjustable steel bars secured to the medial and lateral aspects
uted transverse patella fracture model. We hypothesized that of tibial and femoral ends of the specimen.
patellar fractures fixed with headless compression screws A tensile force was applied through a 6-hole stainless
would demonstrate similar interfragmentary motion and steel plate that was sutured to the tip of the quadriceps tendon,
maximum force before clinical failure during activation of
the knee extensor mechanism when compared with traditional
cannulated screws.
METHODS
Surgical Preparation
Six matched pairs of fresh-frozen cadaveric knees with
at least 10 cm of distal femur and proximal tibia attached were
procured from musculoskeletally normal donors (3 men and 3
women, age range: 42–75 years, mean: 62 years, body mass
index: 21–25) and stored at 2208C until the day of fixation or
testing. Surgical preparations were performed in batches by
the same orthopaedic surgeon to ensure consistency, while
mechanical testing was conducted on a separate day. The
specimens were kept in the freezer after the surgical proce-
dure and were removed to thaw overnight at room tempera-
ture before testing.
Each specimen was prepared with a midline incision
through the skin to expose the quadriceps tendon, patella,
retinaculum, and patellar tendon. The length of the patella
was measured with a caliper to identify the midpoint, where
a transverse osteotomy was subsequently created using a 1-
mm blade band saw. The osteotomy site was then manually
reduced and held in place with reduction clamps. Two parallel
k-wires were then drilled through the patella. A depth gauge
was used to measure the length of each k-wire. The k-wires
were overdrilled, and the appropriate length cannulated
headless or headed screws were inserted over the k-wires,
after which both k-wires were removed. One knee of each
pair randomly received fixation with conventional headed
screws (Synthes 4.0 partially threaded cannulated screws;
DePuy Synthes, West Chester, PA) and the other with
headless compression screws (Acutrak 4/5; Acumed, Hills-
boro, OR). An 18-gauge wire was passed through one screw,
brought over the anterior surface of the patella, and passed
through the second screw in a figure-of-eight pattern (Fig. 1).
The wire was twisted at 2 points to create a standard tension
band construct.
Biomechanical Testing
On testing day, the specimens were skeletonized,
sparing the joint capsule, the quadriceps tendon, and the
patellar tendon. The specimen was then mounted onto
a servohydraulic load frame (Landmark 370; MTS Systems,
Eden Prairie, MN) through an angle bracket attached to the FIGURE 1. Surgical fixation using headless screws with a ten-
tibial end for testing in a 45-degree flexed position (Fig. 2). sion band.
Martin et al J Orthop Trauma Volume 33, Number 6, June 2019
band wiring.13,20 Three cycles of loading and unloading were

performed with the first 2 cycles serving as preconditioning.
Range of interfragmentary movement in the sagittal plane was
identified.
Load-to-failure testing was then conducted to test
fixation strength. Each specimen was loaded to a maximum
of 1000 N following the same protocol while monitoring for
structural or clinical failure. Structural failure was defined as
a sudden drop in the applied tendon force. Clinical failure was
determined from the loading portion of the force-deformation
plot, where the force corresponds to 2 mm of separation
between fragments on the anterior surface of the patella. This
threshold was selected based on reported risks of inhibiting
bony healing associated with a fracture gap of more than
2 mm.21,22 Fixation stiffness was derived from the linear
portion of the load-displacement plots before failure. Each
specimen was visually examined to identify the mode of fail-
ure at the end of the experiment.
Data Analysis
The primary interfragmentary motion was plotted
against the applied tendon force. Two variables were used
to evaluate fixation stability: the range of movement from
nondestructive testing and the initial fixation stiffness deter-
mined through linear regression from the plot of the
destructive tests. Fixation strength was obtained from the
destructive test plots using force at either the point of clinical
failure or structural failure, which ever occurred first in the
test. Comparison between the headless and headed screw
groups was conducted using the Wilcoxon signed-rank test
FIGURE 2. Testing configuration: The 45-degree flexed knee due to small sample size. The level of significance was set at
position was reinforced with a pair of length-adjustable steel 0.05. The sample size of 6 was determined based on the
bars secured at the tibial and femoral ends of specimen. estimation made using the SD reported in a previous study by
Carpenter et al13 and achieved 0.8 statistical power in
a matched comparison with an SD of 150 N and an expected
rolled up, and secured in place with no. 5 sutures. A winch mean difference of 250 N. After the data collection was com-
rope, which has the strength of 5400 lbs, was looped around plete, a post hoc power analysis was conducted to estimate the
the tendon just proximal to the plate and tied to the base of the observed statistical power. All statistical analyses were per-
MTS frame. Tensile force was applied at a rate of 2 mm/s by formed using JMP statistical software (v.13, SAS Institute,
the actuator of the MTS to a predetermined force level and Inc, Cary, NC).
then unloaded at the same rate. An active marker triad from
an optical motion tracking system (Optotrak Certus; NDI,
Waterloo, ON, Canada, accuracy of 0.1 mm) was mounted on RESULTS
the center of each fragment to capture movement between In all cases, the primary interfragmentary motion was
patellar fragments. The custom-made triad was constructed the sagittal plane rotation of the fracture gapping on anterior
from 2 brass square tubes perpendicularly crossing each surface of the patella. Under 150 N of quadriceps tendon
other. The tube was tightly fitted over a 1/8-inch diameter force, the observed interfragmentary movement was small for
brass rod that was first drilled into the patella fragment both groups. The mean magnitude for tension band fixation
without penetrating the second cortex. Relative movement with headless screws was 0.10 6 0.06 degrees and 0.31 6 0.28
between the 2 fragments was derived based on rigid body degrees for headed screws (P = 0.03). During the destructive
kinematics. An extra marker was placed on the MTS actuator testing, structural failure was observed in all 6 knees from the
to synchronize kinematic data obtained from the Optotrak headed screw fixation group and 4 of the 6 knees from the
with the force data from the MTS system. Both the MTS force headless screw group. However, in these 10 cases, structural
data and markers’ location data were collected at 100 Hz. failure occurred after fracture separation reached the 2-mm
Specimens were first tested nondestructively to a max- predefined clinical failure gap. Two knees from the headless
imum tendon load of 150 N, as a previous study found that screw group withstood the 1000 N maximum applied force
most specimens achieved full extension under similar tendon without either structural or clinical failure. The peak interfrag-
force.6 This loading level represents 20%–25% of the re- mentary separation of these 2 specimens was 1.7 mm and
ported mean failure load with cannulated screws and tension 0.6 mm, respectively. Their matching headed screw
counterparts failed at 735 N and 799 N, respectively. The they are commonly used and readily available. This study
maximum applied force of 1000 N was considered as a con- shows that constructs with headless screws result in less in-
servative estimate of the survivors’ strength. terfragmentary motion under low-level tendon loading of
A representative pair of force-displacement plots is 150 N, higher construct rigidity, and greater fixation strength
shown in Figure 3. The mean stiffness derived from the linear against a 2-mm clinical threshold of fragment displacement
portion of the destructive test was 510 6 362 N/degree for the over constructs with headed screws.
construct with headless screws and 277 6 243 N/degree for A likely explanation for the findings lies in the design
the headed screws (P = 0.03), indicating a significantly more of the screws. The headed screws are only threaded along the
stable fixation construct by the headless screws within the distal half of the screw length. By contrast, headless
physiologic loading range (Fig. 4A). The mean clinical failure compression screws have a fully threaded tapered shaft with
strength was 520 6 241 N for the headed screws and a slightly larger proximal diameter. The headless screws also
808 6 183 N for the headless screws. This difference was have a variable pitch, meaning the distance between the
also statistically significant (P = 0.03) (Fig. 4B). The mean threads becomes coarser toward distal end of the screw,
structural failure strength was 664 6 246 N for the headed which provides compression along the shaft of the screw.
screws and 895 6 122 N for the headless screws (P = 0.02). This affords the headless screws greater thread-to-bone
The mode of structural failure for all 10 specimens was contact area, and thus greater holding power. In this study,
stripping of the screw thread within the patella. None of the stiffness was derived from the linear portion of the load-
tension band wires broke. Post hoc analysis estimated that the displacement plot before clinical failure and, therefore, is
observed power was 94% for failure strength, 80% for related to construct stability within the physiologic loading
construct stiffness, and 58% for interfragmentary movement. range. The greater stiffness afforded by the headless screws
provides a more stable condition for bone healing during
activities of daily living.
DISCUSSION Although no studies of headless compression screws
There have been many studies on various modifications have previously been done in patellar fractures, headless
of the tension band wiring fixation developed to reduce the screws performed more than adequately when applied to the
incidence of implant-related complications without sacrificing fixation of talar neck fractures.28,29 Capelle et al28 compared
fixation strength. Other techniques include locked plate the shearing strength and maximum talar head displacement
fixation, suture buttons, replacing wires with FiberWire between Acutrak 4/5 screws and conventional 4.0 cannulated
suture, and stapling techniques.23–27 To the best of our knowl- screws. They found headless screws had significantly less
edge, no study has examined the potential of replacing con- fracture site displacement and higher, although not statistically
ventional cannulated screws with headless compression significant, fixation strength (772 6 389 N vs. 639 6 423 N)
screws. This study set out to determine whether headless than the conventional headed screws. In the study by Karakasli
screws are a biomechanically viable option in the modified et al,29 the same headless screws demonstrated a mean ultimate
tension band fixation construct of transverse patella fractures. load of 839 6 441 N under simulated talar head loading and
A matched cadaveric comparison was performed by loading a fixation strength that was 10% higher than a medially applied
the quadriceps tendon with the knee locked in 45 degrees of 2.7-mm locking plate. The fixation strength of the headless and
flexion. A tension band construct using either Acutrak 4/5 headed screws obtained in this study is in general agreement
headless compression screws or Synthes partially threaded with these 2 studies, taking into account the differences in
4.0 cannulated screws was selected for the comparison, as fracture site anatomy and resultant stresses across the
FIGURE 3. A representative pair of load-dis-

placement plots obtained from the load-to-
failure tests.
Martin et al J Orthop Trauma Volume 33, Number 6, June 2019
FIGURE 4. A, Box plot diagram with individual

(º) and mean (·) values for construct stiffness
from headed and headless screw fixations. B,
Box plot diagram with individual (º) and mean
(·) values for clinical failure force of the headed
and headless screw fixations. The headless fix-
ation in pairs 5 and 6 withstood clinical failure
beyond the maximum applied force of 1000 N
but are included in data analysis using the
maximum value of 1000 N.
fixation (ie, combined bending and shear stress in talar neck Specimens from both study groups were able to withstand this
fracture and combined bending and tensile stress in patella force, with the maximum interfragmentary displacement well
fractures). under the 2-mm clinical critical range. The mean values for
The patella is subjected to complex loading modes the headed screw fixation group (0.31 6 0.28 mm) were in
including bending moments that results in tension at its similar range as the 0.18 6 0.33 mm obtained by Banks et al.6
anterior surface. The magnitude of the bending moment Although this applied force is much smaller than those devel-
varies over the range of knee flexion and extension, so the oped in vivo during weightbearing activities, it has been
relative contribution of the tension band to the fixation shown to be a good approximation of the forces the patella
construct depends on the position of the knee joint. The knee encounters during controlled postoperative rehabilitation of
was tested at a fixed 45-degree flexed position in this study. the knee.31
This testing configuration was selected over extension of the This study had some limitations. The fixtures designed
knee to represent a more rigorous state of patellar loading, as for the study were able to apply a continuous loading of the
it corresponds to peak patellofemoral joint reaction forces. quadriceps tendon up to 1000 N. This was not enough force
One study showed that the failure load of modified tension to produce clinical failure in 2 specimens in the headless
band fixation obtained at 45 degrees of knee flexion was only screws fixation group. In these 2 cases, the maximum applied
55% of the failure load at full extension.30 force was used to represent fixation strength, which resulted
Our results of tension band constructs with conven- in a conservative estimate of the headless screw strength. The
tional screws are in good accordance with previous studies study involved a small number of specimens, and the
using similar test configurations.13 Before testing the fixation experiments were limited to quasi-static loading without
strength of the patella construct, range of interfragmentary prolonged cyclic or fatigue testing. This study also did not
movement under low-level loading of 150 N was conducted. include a bone density measurement of the specimens.
Although the use of matched pairs should theoretically reduce patella fractures: a strobe-compliant retrospective observational study.
potential bias from interspecimen variations, the effect of Medicine (Baltimore). 2016;95:e4992.
12. Miller MA, Liu W, Zurakowski D, et al. Factors predicting failure of
bone density on the differences in construct performance patella fixation. J Trauma Acute Care Surg. 2012;72:1051–1055.
could not be evaluated. Finally, a mismatch in the core 13. Carpenter JE, Kasman RA, Patel N, et al. Biomechanical evaluation of
diameter of Acutrak 4/5 and Synthes 4.0 cannulated screws current patella fracture fixation techniques. J Orthop Trauma. 1997;11:
might have given an advantage to the headless screws in 351–356.
14. Berg EE. Open reduction internal fixation of displaced transverse patella
terms of holding power. fractures with figure-eight wiring through parallel cannulated compres-
In conclusion, findings from this study provide bio- sion screws. J Orthop Trauma. 1997;11:573–576.
mechanical evidence that supports the use of headless 15. Kumar G, Mereddy PK, Hakkalamani S, et al. Implant removal following
compression screws in tension band wiring of transverse surgical stabilization of patella fracture. Orthopedics. 2010;33:301.
patella fractures. The improved strength from using headless 16. Dy CJ, Little MT, Berkes MB, et al. Meta-analysis of re-operation, non-
union, and infection after open reduction and internal fixation of patella
screws will generate greater interfragmentary compression fractures. J Trauma Acute Care Surg. 2012;73:928–932.
that may lead to more predictable healing. It also encourages 17. Smith ST, Cramer KE, Karges DE, et al. Early complications in the
future exploration of constructing tension bands with more operative treatment of patella fractures. J Orthop Trauma. 1997;11:
compliant material, such as high-strength suture, to further 183–187.
18. Catalano JB, Iannacone WM, Marczyk S, et al. Open fractures of the
reduce the chance of implant irritation. Future investigations patella: long-term functional outcome. J Trauma. 1995;39:439–444.
will focus on clinical evidence to determine if headless 19. Goldstein SA, Coale E, Weiss AP, et al. Patellar surface strain. J Orthop
compression screw fixation is biomechanically sound Res. 1986;4:372–377.
in vivo and whether it reduces the rate of symptomatic 20. Wurm S, Augat P, Bühren V. Biomechanical assessment of locked plat-
implant complications. Although the headless compression ing for the fixation of patella fractures. J Orthop Trauma. 2015;29:e305–
8.
screws are more expensive than headed screws, lowering the 21. Hughes SC, Stott PM, Hearnden AJ, et al. A new and effective tension-
rate of reoperation due to symptomatic implants could offset band braided polyester suture technique for transverse patellar fracture
the increased cost. fixation. Injury. 2007;38:212–222.
22. Gaston MS, Simpson AHRW. Inhibition of fracture healing. J Bone Joint
Surg Br. 2007;89-B:1553–1560.
REFERENCES 23. Han F, Pearce CJ, Ng DQ, et al. A double button adjustable loop device
1. Boström A. Fracture of the patella. A study of 422 patellar fractures. Acta is biomechanically equivalent to tension band wire in the fixation of
Orthop Scand Suppl. 1972;143:1–80. transverse patellar fractures-a cadaveric study. Injury. 2017;48:270–276.
2. Schindler OS, Scott WN. Basic kinematics and biomechanics of the 24. Wild M, Fischer K, Hilsenbeck F, et al. Treating patella fractures with
patello-femoral joint. Part 1: the native patella. Acta Orthop Belg. a fixed-angle patella plate-A prospective observational study. Injury.
2011;77:421–431. 2016;47:1737–1743.
3. Waligora AC, Johanson NA, Hirsch BE. Clinical anatomy of the quad- 25. Schnabel B, Scharf M, Schwieger K, et al. Biomechanical comparison of
riceps femoris and extensor apparatus of the knee. Clin Orthop Relat Res. a new staple technique with tension band wiring for transverse patella
2009;467:3297–3306. fractures. Clin Biomech (Bristol, Avon). 2009;24:855–859.
4. Kaufer H. Patellar biomechanics. Clin Orthop Relat Res. 1979:51–54. 26. Thelen S, Schneppendahl J, Jopen E, et al. Biomechanical cadaver testing
5. Carpenter JE, Kasman R, Matthews LS. Fractures of the patella. Instr of a fixed-angle plate in comparison to tension wiring and screw fixation
Course Lect. 1994;43:97–108. in transverse patella fractures. Injury. 2012;43:1290–1295.
6. Banks KE, Ambrose CG, Wheeless JS, et al. An alternative patellar 27. Wright PB, Kosmopoulos V, Coté RE, et al. FiberWire is superior in
fracture fixation: a biomechanical study. J Orthop Trauma. 2013;27: strength to stainless steel wire for tension band fixation of transverse
345–351. patellar fractures. Injury. 2009;40:1200–1203.
7. Hungerford DS, Barry M. Biomechanics of the patellofemoral joint. Clin 28. Capelle JH, Couch CG, Wells KM, et al. Fixation strength of anteriorly
Orthop Relat Res. 1979:9–15. inserted headless screws for talar neck fractures. Foot Ankle Int. 2013;34:
8. Buff HU, Jones LC, Hungerford DS. Experimental determination of 1012–1016.
forces transmitted through the patello-femoral joint. J Biomech. 1988; 29. Karakasli A, Hapa O, Erduran M, et al. Mechanical comparison of head-
21:17–23. less screw fixation and locking plate fixation for talar neck fractures. J
9. Reilly DT, Martens M. Experimental analysis of the quadriceps muscle Foot Ankle Surg. 2015;54:905–909.
force and patello-femoral joint reaction force for various activities. Acta 30. Dargel J, Gick S, Mader K, et al. Biomechanical comparison of tension
Orthop Scand. 1972;43:126–137. band- and interfragmentary screw fixation with a new implant in trans-
10. Melvin JS, Mehta S. Patellar fractures in adults. J Am Acad Orthop Surg. verse patella fractures. Injury. 2010;41:156–160.
2011;19:198–207. 31. Benjamin J, Bried J, Dohm M, et al. Biomechanical evaluation of various
11. Tan H, Dai P, Yuan Y. Clinical results of treatment using a modified forms of fixation of transverse patellar fractures. J Orthop Trauma. 1987;
K-wire tension band versus a cannulated screw tension band in transverse 1:219–222.
10. Bahl V, Hu HM, Henke PK, et al. A validation study of a retrospective 19. Pelet S, Roger ME, Belzile EL, et al. The incidence of thromboembolic
venous thromboembolism risk scoring method. Ann Surg. 2010;251: events in surgically treated ankle fracture. J Bone Joint Surg Am. 2012;
344–350. 94:502–506.
11. Caprini JA. Risk assessment as a guide for the prevention of the many 20. Selby R. Symptomatic venous thromboembolism uncommon without
faces of venous thromboembolism. Am J Surg. 2010;199:S3–S10. thromboprophylaxis After isolated lower-limb fracture. J Bone Joint Surg
12. Pannucci CJ, Bailey SH, Dreszer G, et al. Validation of the Caprini risk Am. 2014;96:e83.
assessment model in plastic and reconstructive surgery patients. J Am 21. Geerts WH, Code KI, Jay RM, et al. A prospective study of venous throm-
Coll Surg. 2011;212:105–112. boembolism after major trauma. N Engl J Med. 1994;331:1601–1606.
13. Pannucci CJ, Dreszer G, Wachtman CF, et al. Postoperative enoxaparin 22. Greenfield LJ, Proctor MC, Rodriguez JL, et al. Post trauma thrombo-
prevents symptomatic venous thromboembolism in high-risk plastic sur- embolism prophylaxis. J Trauma. 1997;42:100–103.
gery patients. Plast Reconstr Surg. 2011;128:1093–1103. 23. Parkland PE Risk Assessment. Available at: http://www.parklandhospital.
14. Shuman AG, Hu HM, Pannucci CJ, et al. Stratifying the risk of venous com/phhs/orthopaedic-research.aspx. Accessed April 2016.
thromboembolism in otolaryngology. Otolaryngol Head Neck Surg. 24. Geerts WH, Jay RM, Code KI, et al. A comparison of low-dose heparin
2012;146:719–724. with low-molecular-weight heparin as prophylaxis against venous throm-

15. Zhou HX, Peng LQ, Yan Y, et al. Validation of the Caprini risk assess- boembolism after major trauma. N Engl J Med. 1996;335:701–707.
ment model in Chinese hospitalized patients with venous thromboembo- 25. Al-Sallami H, Jordan S, Ferguson R, et al. Current enoxaparin dosing
lism. Thromb Res. 2012;130:735–740. guidelines have dubious credibility. N Z Med J. 2010;123:62–67.
16. Luksameearunothai K, Sa-ngasoongsong P, Kulachote N, et al. 26. Riha GM, Van PY, Differding JA, et al. Incidence of deep vein throm-
Usefulness of clinical predictors for preoperative screening of deep bosis is increased with 30 mg twice daily dosing of enoxaparin compared
vein thrombosis in hip fractures. BMC Musculoskelet Disord. 2017; with 40 mg daily. Am J Surg. 2012;203:598–602.
18:208. 27. Rutherford EJ, Schooler WG, Sredzienski E, et al. Optimal dose of
17. Saragas NP, Ferrao PN, Saragas E, et al. The impact of risk assessment enoxaparin in critically ill trauma and surgical patients. J Trauma.
on the implementation of venous thromboembolism prophylaxis in foot 2005;58:1167–1170.
and ankle surgery. Foot Ankle Surg. 2014;20:85–89. 28. Bush S, LeClaire A, Hampp C, et al. Review of a large clinical series:
18. Lassen MR, Borris LC, Nakov RL. Use of the low-molecular-weight once- versus twice-daily enoxaparin for venous thromboembolism pro-
heparin reviparin to prevent deep-vein thrombosis after leg injury requir- phylaxis in high-risk trauma patients. J Intensive Care Med. 2011;26:
ing immobilization. N Engl J Med. 2002;347:726–730. 111–115.
Invited Commentary
Risk Assessment After Orthopaedic Trauma: Coming of Age
A fatal pulmonary embolus is the number 1 preventable

cause of death after surgery.1 It has been shown that using
traditional anticoagulants prevent almost all these deaths and
associated with a much lower incidence of thrombosis than
those involving the proximal leg and/or pelvis.
The key element in applying the Caprini score to a
sublethal thrombotic events which on occasion may be life- given population is to identify the cutoff between low- and
changing. To preserve good outcomes, surgeons must care- high-risk individuals using the score rather than the type of
fully weigh the thrombotic versus bleeding risk of each fracture. This enables selection of the appropriate prophylaxis
patient. Traditional anticoagulation may cause bleeding carefully balancing the risks of bleeding versus thrombosis.
that jeopardizes a well-performed procedure. The current study was performed to explore the validity of the
The incidence of thrombotic events is related to the Caprini score in the orthopaedic fracture population. They
presence and number of thrombosis risk factors in addition to also compared the score between 2 groups of fracture patients
the type and extent of injury. The Caprini score has been believed to be high versus low risk of thrombosis.
validated in multiple specialties, and when the data are One of the features of the Caprini score is to look at the
properly collected, every study shows a statistically signifi- entire risk profile of the patient independent of the type of
cant correlation between the score and the incidence of surgery or fracture. A low-risk fracture may be associated
venous thromboembolism (VTE). This is most obvious when with a high risk of thrombosis when additional risk factors are
the group tested does not receive thrombosis prophylaxis.2 present. Patients with high-risk fractures but no additional risk
The authors report that the Caprini score represents a factors may have a lower risk than average for that type of
thorough history and physical. This score, in fact, provides a injury. The bottom line is to look at all the risks associated
profile of the patient and allows for a selection of the type, with an individual patient not just the type of fracture or
strength, and duration of prophylaxis. One of the main procedure.
criticisms of the Caprini score after Orthopaedic trauma is The authors found that the low-risk fracture group had
that all fractures are scored the same. The authors correctly fewer deep vein thrombosis (DVT) events than seen in those
point out that there is a disparity between different types of with high-risk fractures. The Caprini score was not statisti-
fractures depending on location. Ankle fractures overall are cally different between these groups when looking at fracture
Caprini J Orthop Trauma Volume 33, Number 6, June 2019
type alone. This validates the concept that one has to look In conclusion, this study suggests that the Caprini score
beyond the type of fracture and track all the patients’ risk has value classifying the degree of risk in traumatic fracture
factors to calculate an individuals’ thrombotic risk. When patients. The next step is to pursue a prospective multi-
complete evaluation was performed, the Caprini score was institutional randomized clinical trial to validate these pre-
successful in identifying those with thrombosis regardless liminary findings to better balance the risks of thrombosis
of fracture type. According to the authors—“In this study, it versus bleeding after major orthopaedic trauma.
was determined that a Caprini Score of 11-12 correlated Boston University has been a nationwide leader in
with a higher rate of VTEs in both the high and low risk applying the Caprini risk score to a variety of surgical specialties,
populations, which suggests that this screening tool may have and the current authors have continued this tradition.5
utility in stratifying patients for VTE risk.”
Furthermore, the authors report a statistically signifi- Joseph A. Caprini, MD, FACS*,†
cant cutoff score of 12, above which patients were at *Emeritus, NorthShore University Health System
increased risk of developing VTE. This cutoff was also the Evanston, IL
same for patients in the high-risk category according to type †University of Chicago Pritzker School of Medicine
of fracture compared with the low-risk category. They found Chicago, IL
that a score of less than 11 was associated with a lower risk
of VTE. Finally, based on their studies, they recommend that REFERENCES
the score cutoff should be greater than 10 for these fracture 1. Mahan CE, Borrego ME, Woersching AL, et al. Venous thromboembo-
patients. lism: annualised United States models for total, hospital-acquired and
The authors state further regarding the cutoff point of preventable costs utilising long-term attack rates. Thromb Haemost.
2012;108:291–302.
the score that—“Our study’s findings are consistent with prior 2. Pannucci CJ, Swistun L, MacDonald JK, et al. Individualized venous
literature that reported a Caprini Score of 12 or greater had a thromboembolism risk stratification using the 2005 Caprini score to iden-
significantly higher relative risk of pre-operative DVT versus tify the benefits and harms of chemoprophylaxis in surgical patients:
no DVT in hip fracture patients.3” Those authors advocate a meta-analysis. Ann Surg. 2017;265:1094–1103.
preoperative screening in hip fracture patients with these high 3. Luksameearunothai K, Sa-Ngasoongsong P, Kulachote N, et al. Useful-
ness of clinical predictors for preoperative screening of deep vein throm-
Caprini scores. bosis in hip fractures. BMC Musculoskelet Disord. 2017;18:208.
In another study, it was found that a score of 10 or 4. Krauss ES, Segal A, Cronin M, et al. Implementation and validation of the
above correctly identified all but one of the patients who 2013 Caprini score for risk stratification of arthroplasty patients in the
developed thrombosis postoperatively after total joint replace- prevention of venous thrombosis. Clin Appl Thromb Hemost. 2019;25:
1076029619838066.
ment. The authors speculated that the higher scores may 5. Cassidy MR, Rosenkranz P, McAneny D. Reducing postoperative venous
dictate using traditional anticoagulation, while scores of less thromboembolism complications with a standardized risk-stratified prophylaxis
than 10 aspirin was sufficient.4 protocol and mobilization program. J Am Coll Surg. 2014;218:1095–1104.
TECHNICAL TRICK
Orthogonal Plating With a 95-Degree Blade Plate for

Salvage of Unsuccessful Cephalomedullary Nailing of
Atypic Femur Fractures: A Technical Trick
Natalie C. Rollick, MD, FRCSC, Jessica Bear, MD, Owen Diamond, MD, FRCSEd(T&O),
David S. Wellman, MD, and David L. Helfet, MD
through the use of an angled blade plate. An additional

Summary: Atypical femur fractures secondary to bisphosphonate anterior reconstruction plate is used to optimize reduction
use are often complicated by nonunion, which often results in and provide an additional strut against varus collapse
progressive varus collapse, femoral head and neck bone loss, and through the nonunion. We also present a case series of 9
eventual implant failure. We present a technique that uses orthog- patients treated by our orthogonal plating technique as
onal, dual plating of the proximal femur in addition to biologic a salvage surgery for subtrochanteric atypical femur frac-
augmentation for the management of these challenging fractures. tures, which failed to unite with initial cephalomedullary
Key Words: atypical femur fracture, bisphosphonate, cephalome- nail fixation (Figs. 2, 3).
dullary nail, revision
SURGICAL TECHNIQUE
All patients undergo a preoperative evaluation includ-
INTRODUCTION ing clinical and laboratory assessment for infection. Patients
Atypical femur fractures, often associated with long- are uniformly referred for evaluation by the metabolic bone
term bisphosphonate use, are gaining increasing recognition service for assessment and treatment of bone endocrine
as a challenging clinical problem. The patients affected are anomalies and consideration for perioperative teriparatide
often elderly, polycomorbid, and have poor tolerance for therapy. Preoperative radiographs are carefully evaluated for
prolonged immobility. Several surgical options exist, but all angular and rotational deformity. It is crucial to template
are well documented to be complicated by high rates of both blade insertion angle to adequately correct varus angulation.
delayed union and nonunion. The rate of union at 6 months This often means inserting the blade at less than 95 degrees to
has been documented at 50%–70%, and the time to union the shaft. Patients are also consented preoperatively for the
ranges between 7 and 18 months.1–4 potential use of off-label BMP-2.
Recent evidence has shown a reduction in nonunion The patient is positioned supine on a radiolucent flat top
rates when atypical femoral fractures are treated with intra- table with the operative side brought to the edge of the table. A
medullary fixation when compared with extramedullary bump or bean bag can be used to tilt the patient into a “sloppy
fixation5,6; therefore, cephalomedullary nailing has become lateral” position to facilitate the approach or fluoroscopic imag-
the current standard of fixation for atypical femoral fractures. ing per surgeon preference. This allows the surgeon to ade-
Despite improvements with surgical techniques, atypical fem- quately position his or her hand during preparation and blade
oral fractures require revision surgery in 12%–25% of cases, insertion. A traction table could also be used; however,
4 times more frequently than standard femoral fractures.1,7,8 although it may facilitate nail extraction and lateral fluoro-
When used to treat subtrochanteric atypical femoral fractures, scopic imaging, the traction table impedes the surgeon’s ability
cephalomedullary nails tend fail into varus and thus result in to manipulate the leg to reduce the fracture. An anatomical
medial calcar and femoral head bone loss, further complicat- reduction is paramount to the success of this procedure, and
ing revision surgeries (Fig. 1). thus, we prioritize this when selecting an operative table.
We present a technique to address nonunions of Previous incisions are extended to accommodate a lat-
subtrochanteric, atypical femur fractures, which mitigates eral subvastus approach to the proximal femur and facilitate
issues relating to varus collapse and femoral head bone loss implant removal. Distal locking screws are removed first.
Broken screws that will not impede nail extraction are left in
Accepted for publication December 9, 2018.
From the Orthopaedic Trauma Service, Hospital for Special Surgery and New situ. The proximal aspect of the intramedullary nail is
York Presbyterian Hospital, Weill Cornell Medicine, New York, NY. identified. This can be facilitated by using the 2.5-mm guide
The authors report no conflict of interest. wire on fluoroscopy to cannulate the top of the nail and
Reprints: David L. Helfet, MD, Orthopaedic Trauma Service, Hospital for Special adducting the extremity. A 10-mm acorn reamer is passed
Surgery and New York Presbyterian Hospital, Weill Cornell Medicine, 535
East 70th St, New York City, NY 10021 (e-mail: helfetd@hss.edu).
over the guide wire to clear overgrown bone and allow for
Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. attachment of the extraction handle. After the extraction
DOI: 10.1097/BOT.0000000000001426 handle is firmly in place, the cephalomedullary screw is
J Orthop Trauma Volume 33, Number 6, June 2019 Orthogonal Plating With a 95-Degree Blade Plate
FIGURE 1. Cephalomedullary nail failure at 10

months postoperative.
identified and removed. Osteotomes and curettes are fre- The nonunion is then identified and debrided back to
quently required to clear the lateral exit point of the bleeding bone. All fibrous tissue and callous are meticulously
cephalomedullary screw, as it is also often covered with removed from the nonunion site, and cultures are sent to
heterotopic bone. The cephalomedullary screw and nail are rule out indolent infection. The sclerotic boney edge of the
then removed. Nail removal often contributes significantly to nonunion is then removed with osteotomes to maximize
the challenging nature of this procedure and can contribute the potential healing response. Drill bits are used to open the
significantly to blood loss, and it should not be under- medullary canal of the femur proximally and distally to allow
estimated. The manufacturer of the cephalomedullary nail for inflow of healthy marrow contents to the nonunion site.
should be contacted preoperatively to ensure all necessary The guide wire for a 95-degree–angled blade plate
removal equipment is available. (Synthes, Paoli, PA) is placed at the previously templated
FIGURE 2. Intraoperative fluoroscopic evaluation

of alignment.
Rollick et al J Orthop Trauma Volume 33, Number 6, June 2019
FIGURE 3. A, A 69-year-old woman initially presents 14 months after intramedullary nailing after an atraumatic femur fracture. B,
Immediate postoperative films and (C) final follow-up films 15 months postoperative.
angle relative to the femoral shaft in the coronal plane and as the patient heals. The plate is contoured to accommodate
centrally through the femoral neck in the sagittal plane to for the presence of any callous, but care is taken to under-
restore appropriate proximal femoral valgus angulation. The contour the plate to counteract sagittal plane forces and to
seating chisel is then inserted and used to gain control of the allow the anterior plate to be used as a reduction tool. Minor
proximal fragment. It is crucial to meticulously position both adjustments in alignment can be completed with the plate
the guide wire and seating chisel because the chisel position fixed to the bone because the 3.5mm reconstruction plate is
will ultimately significantly contribute to the reduction. The malleable. At this point, the femur is reduced and provision-
position should be rigorously evaluated on fluoroscopy. ally stabilized, and the chisel is removed for insertion of the
Attention is then turned to reduction of the nonunion. blade plate. If desired, the defect within the head and neck
The surgeon can easily visualize the lateral and anterior secondary to the removed cephalomedullary screw is filled
aspects of the nonunion after the subvastus approach and with allograft chips or demineralized bone matrix before
debridement have been completed. Standard reduction clamps blade insertion. The blade plate is then further fixed to the
are used in both the sagittal and coronal planes to reduce the proximal segment through cortical screws. The distal aspect
fracture. Care is taken to obtain anatomical alignment in both of the plate is left unfixed to allow for usage of the articulated
the coronal and sagittal plane on fluoroscopy. A Verbrugge compression device.
clamp can also be useful for this task. The reduction clamps An articulated tensioning device is then used to
are typically placed with the tines medial and lateral to allow compress the fracture through the distal aspect of the blade
for placement of the reconstruction plate with the clamps plate. The distal screws in the anterior plate are loosened
undisturbed. With the clamps in position, a 3.5-mm locking before blade plate compression. The loosened screws along
reconstruction plate (Synthes) is then applied to the anterior with the malleability of the reconstruction plate do not impede
femur to control deforming forces. Once the anterior plate is the compression of the articulated tensioning device. Both
applied, the large reduction clamps used to hold the pro- plates are then completely secured to the femur with cortical
visional reduction can be removed to allow for placement of screw fixation. Compression plating technique is used
the blade plate, making the anterior reconstruction plate through the 3.5-mm plate to prevent gapping through the
a useful provisional fixation “clamp” and an orthogonal fix- medial cortex. When possible, interfragmentary fixation
ation device later in the case. It is important to ensure that the (either through the blade or through the 3.5-mm plate) is
anterior plate is medialized as far as possible to allow for placed after the fracture has been appropriately compressed in
compression across the medial cortex later in the procedure. both planes. Although atypical femoral fracture patients
Medial placement also provides a strut against varus collapse generally suffer from osteoporosis, the bone density
J Orthop Trauma Volume 33, Number 6, June 2019 Orthogonal Plating With a 95-Degree Blade Plate
TABLE 1. Patient Demographics

Charlston Time to Neck–Shaft
Patient Number Age Comorbid Index BMI Smoking Status Fracture Cause Union (Days) Angle Complications
1 76 9 25.6 Remote Radiation 120 123 None
2 69 4 24.4 Nonsmoker Bisphosphonates 219 124 Revision grafting
3 65 3 25.0 Remote Bisphosphonates 97 120 None
4 59 3 21.6 Nonsmoker Bisphosphonates 111 128 None
5 70 3 25.7 Nonsmoker Bisphosphonates 57 140 Superficial infection
6 97 5 24.2 Smoker Bisphosphonates 90 127 None
8 69 2 29.3 Nonsmoker Bisphosphonates 92 127 None
surrounding completed subtrochanteric stress fractures is patients lived independently before injury, with 2 patients reporting
often paradoxically increased. As a result, having multiple previous cane use. One patient had an atypical fracture treated pre-
replacement drill bits available is beneficial. viously on the contralateral leg. At the time of treatment, 3 patients
The wounds are thoroughly irrigated, and biologic had hypertrophic nonunion, 1 was oligotrophic, and 5 were atrophic.
Three patients had a previous failed revision attempt with isolated
augmentation is added as per surgeon preference. Biologic
blade plating before conversion to orthogonal plating. Of these, 2
augmentation consisted of autogenous bone grafting derived converted from atrophic nonunion before blade plating to hypertro-
from morcelized hypertrophic callous with or without added phic nonunion.
BMP-2 augmentation or with demineralized bone matrix and At the time of surgery, 7 patients received augmentation with
BMP-2 if insufficient bone was present. The wounds are demineralized bone matrix and BMP-2, 1 patient had local
closed in a layered fashion over a drain. autogenous bone grafting with added BMP-2, and a single patient
had isolated local autogenous bone grafting. Eight of 9 patients
Postoperative Rehabilitation received teriparatide therapy postoperatively. The patient who had
previously been treated for breast cancer was contraindicated for
A compressive dressing is applied for the first 24 to 48
both BMP-2 and teriparatide therapy.
hours. Antibiotics are prophylactically continued until culture Fracture union occurred in all patients at an average of 15
results are finalized. Postoperative venous thromboprophy- weeks (8–31 weeks). One patient experienced a delayed union
laxis is directed as per surgeon protocol. The patient is requiring a secondary procedure for Hernigou style bone grafting10
restricted to toe-touch weightbearing for a minimum of 6 at 6 months. This patient went on to subsequent union 6 weeks later;
weeks. All patients were evaluated for indications and no implants were augmented or changed during the revision grafting
contraindications for teriparatide therapy postoperatively by procedure. There were no cases of implant failure within our series.
the hospital metabolic bone service. The average neck–shaft angle at union was 129 degrees (124–140
degrees). One patient experienced a superficial wound infection,
which resolved with a single course of oral antibiotics. There were
no other documented complications.
CASE SERIES
Surgical logs of 2 fellowship-trained trauma surgeons
between January 2010 and December 2017 were reviewed to
identify patients who underwent revision surgery for proximal femur DISCUSSION
fractures. Clinical charts and radiographs were reviewed to identify Patients with atypical femur fracture are at a biologic
patients with atypical femur fractures initially treated with cepha- and mechanical disadvantage; a challenge that is magnified as
lomedullary nail fixation. Radiographic identification of atypical
these fractures frequently occurs in the subtrochanteric
femur fracture was performed using 2013 American Society for
Bone and Mineral Research Task Force Revised Case Definition.9 region. Cephalomedullary nails have been documented to
Patients were included if they underwent revision surgery using our have the highest union rates5,6 in this setting but unfortunately
dual plating technique. Patients were excluded if they were treated still go on to delayed union or nonunion in up to 40% of
with a different fixation construct or were lost to follow-up before cases.1–4 Several authors have described successful techni-
fracture union or revision surgery. Radiographic union was defined ques for management of these fractures after failed cephalo-
as bridging callous on 3 corticies on orthogonal radiographs. Mea- medullary nails with small case reports including
surement of the final neck–shaft angle was completed at time of compression plating,11 revision cephalomedullary nails4 and
union. Charts were reviewed for age, sex, body mass index (BMI), proximal femoral replacement.12 Recent improvements in
comorbidities, smoking history, and ambulatory status (Table 1). nailing technique and advancements in nail systems, includ-
Nine patients were identified that met inclusion criteria. All
ing multiple cephalomedullary screw options, have expanded
the patients were female. The mean age of the cohort was 73.1
(59–97) years, and the mean BMI was 24.5. One patient was an revision indications. However, medial calcar bone loss with
active smoker. The average Charleston comorbidity index was 4.1 a varus fracture malreduction remains challenging to combat
6 2.0. Eight patients had a confirmed history of long-term bi- with nail technology.
sphosphonate usage, and 1 patient had an atypical fracture after We present the largest case series to date with
radiation treatment for metastatic breast cancer of the pelvis. All a technique that resulted in well-aligned, revision-free union
Rollick et al J Orthop Trauma Volume 33, Number 6, June 2019
in 88% of patients in an average time of 3.5 months. CONCLUSION

Orthogonal plating using a blade plate allows for stable Atypical femur fractures remain a challenging problem
fixation within a postsurgical femoral head and controlled with high nonunion rates. Although technically difficult,
compression across a grafted fracture site. The anterior orthogonal plating in combination with biologic optimization
reconstruction plate acts as a valuable reduction tool and is a successful salvage option for the treatment of subtrochan-
provides a stable medial column to minimize varus strain teric atypical femur fractures after failed cephalomedullary
through the blade plate during healing and the addition of nails. Further research is needed into the optimal form of
a medial strut further resists varus collapse. The additional primary atypical fracture treatment and the ideal form of
plate allows for orthogonal fixation in a slow-healing biologic optimization.
environment while offering additional screw holes for
interfragmentary fixation after tensioning to improve con- REFERENCES
struct stability and longevity. 1. Koh A, Guerado E, Giannoudis PV. Atypical femoral fractures related to
Cho et al13 identified varus angulation of the femoral bisphosphonate treatment: issues and controversies related to their surgi-
cal management. Bone Joint J 2017;6:144–153.
head, defined as a neck shaft angle less than 125.6 degrees or 2. Egol KA, Park JH, Rosenberg ZS, et al. Healing delayed but generally
a difference of greater than 4 degrees compared with the reliable after bisphosphonate-associated complete femur fractures treated
contralateral side, and sagittal malalignment greater than 5 with IM nails. Clin Orthop Relat Res 2014;472:2728–2734.
degrees, as significant predictors of nonunion. There is an 3. Lee KJ, Yoo JJ, Oh KJ, et al. Surgical outcome of intramedullary nailing
extremely low tolerance for malreduction in this region, and in patients with complete atypical femoral fractures. Injury 2017;48:941–
945.
it is imperative to restore anatomical alignment of the fracture 4. Weil YA, Rivkin G, Safran O, et al. The outcome of surgically treated
for optimal healing and functional outcomes. It is important to femur fractures associated with long-term bisphosphonates use. J
select a construct, which can maintain an adequate reduction Trauma 2011;71:186–190.
while providing compression across the fracture. We believe 5. Teo BJ, Koh JS, Goh SK. Post-operative outcomes of atypical femoral
subtrochanteric fractures in patients on bisphosphonate therapy. Bone
that orthogonal plating of the femur has greater potential to Joint J 2014;96-B:658–664.
achieve these goals than revision intramedullary nailing, espe- 6. Prasarn ML, Ahn J, Helfet DL, et al. Bisphosphonate-associated femur
cially in the setting of a varus malreduction. fractures have high complication rates with operative fixation. Clin Or-
A thorough surgical plan involves optimization of thop Relat Res 2012;470:2295–2301.
7. Bogdan Y, Tornetta P III, EInhorn TA, et al. Healing time and compli-
metabolic bone health with consideration of supplementation cations in operatively treated atypical femur fractures associated with
of vitamin D, calcium, and potentially teriparatide when bisphosphonate use: a multicenter retrospective cohort. J Orthop Trauma
managing atypical femur fractures. This has also been 2016;30:177–181.
emphasized by Giannoudis et al14 who coined the term “dia- 8. Schlicher J. High revision rate and good healing capacity of atypical
mond concept,” which entails biologic supplementation of femoral fractures. A comparison with common shaft fractures. Injury
2015;45:2468–2473.
nonunions with growth factors, mesenchymal cells and 9. Shane E, Burr D, Abrahamsen B, et al. Atypical subtrochanteric and
scaffolding. diaphyseal femoral fractures: second report of a task force of the Amer-
There are several limitations to this report. Most ican Society for Bone and Mineral Research. J Bone Miner Res 2014;29:
notable, the number of patients included in the study is small; 1–23.
10. Hernigou P, Poignard A, Beaujean F, et al. Percutaneous autologous
however, to the best of our knowledge, this is the largest case bone-marrow grafting for nonunions. Influence of the number and con-
series of its type reported to date. Second, as this was centration of progenitor cells. J Bone Joint Surg Am 2005;87:1430–1437.
a retrospective study, there was no standardization in terms of 11. Grady MK, Watson JT, Cannada LK. Treatment of femoral fracture
intraoperative biologic augmentation used. In addition, sev- nonunion after long-term bisphosphonate use. Orthopedics 2012;35:
eral patients have already had an attempted revision limiting e991–5.
12. Jethanandani RG, Nwankwo C, Wolinsky PR, et al. Proximal femoral
the comparisons that can be made to the general population. reconstruction for failed internal fixation of a bisphosphonate-related
Finally, determination of union was completed on the basis of femur fractures. Arthroplast Today 2016;2:153–156.
clinical symptoms and plain radiographs. Therefore, it is 13. Cho JW, Oh CW, Leung F, et al. Healing of atypical subtrochanteric
possible that more advanced imaging would reveal a more femur fractures after cephalomedullary nailing: which factors predict
union? J Orthop Trauma 2017;31:138–145.
delayed healing course. Despite the limitations, we believe 14. Giannoudis PV, Admad MA, Mineo GV, et al. Subtrochanteric fracture
this technique remains a valuable tool in the management of non-unions with implant failure managed with the “Diamond” concept.
these challenging fractures. Injury 2013;44(suppl 1):S76–S81.

JOT June 2019 PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

JOT June 2019 PDF

Încărcat de

Drepturi de autor:

Formate disponibile

IN MEMORIAM

In Memoriam: AONA Tribute to Jeffrey

regenerative surgical treatment of hip disease, which

J Orthop Trauma  Volume 33, Number 6, June 2019 www.jorthotrauma.com | 267

Is the Caprini Score Predictive of Venothromboembolism

J Orthop Trauma  Volume 33, Number 6, June 2019 www.jorthotrauma.com | 269

TABLE 1. Caprini Score Risk Factors

heparin, or dalteparin (87%). Seventy-two percent of the low- RESULTS

Funding Caprini Score Characteristics

FIGURE 1. Patient cohort

TABLE 2. VTE Rates in Study Population

TABLE 3. VTE Characteristics

Risk Assessment After Orthopaedic Trauma: Coming of Age

A fatal pulmonary embolus is the number 1 preventable

Radiographic Healing of Far Cortical Locking

J Orthop Trauma  Volume 33, Number 6, June 2019 www.jorthotrauma.com | 277

FIGURE 1. Blinded radiographs shown to in-

Union Rate and Revision Surgery

Hypoalbuminemia Is an Independent Risk Factor

Key Words: malnutrition, orthopaedic trauma, complications, albu-

284 | www.jorthotrauma.com J Orthop Trauma  Volume 33, Number 6, June 2019

RESULTS extremity fracture were identiﬁed from the NSQIP database.

(continued on next page )

TABLE 3. Hypoalbuminemia and Morbidity After Lower Extremity Orthopaedic Trauma

Loss of Independence After Operative Management of

Methods: We conducted a descriptive analysis to quantify patients’

292 | www.jorthotrauma.com J Orthop Trauma  Volume 33, Number 6, June 2019

Changes in Living Status

TABLE 1. Changes in Living Status for Patients

TABLE 2. Changes in the Use of Aids for Patients

TABLE 3. Breakdown of Aids Used by Nonindependent Ambulators

before their fracture, and an additional 232 participants were

in our analyses. Our results were found to be fairly consistent

APPENDIX 1. FAITH Investigators

Inter-Rater Reliability of the Modified Radiographic Union

Main Outcome Measurements: Inter-rater reliability was

J Orthop Trauma  Volume 33, Number 6, June 2019 www.jorthotrauma.com | 301

PATIENTS AND METHODS

Cortical scoring has been shown to correlate both with REFERENCES

APPENDIX 1. Major Extremity Trauma and Rehabilitation Consortium Corporate Authors

Knee Disarticulations Versus Transfemoral Amputations:

Conclusions: We detected no functional differences measured on

Intervention: Knee disarticulation versus transfemoral amputation. INTRODUCTION

308 | www.jorthotrauma.com J Orthop Trauma  Volume 33, Number 6, June 2019

study, which demonstrated that in patients who underwent

Ability of a Risk Prediction Tool to Stratify Quality and Cost

Key Words: ankle fracture, risk stratiﬁcation, triage, geriatric, cost

312 | www.jorthotrauma.com J Orthop Trauma  Volume 33, Number 6, June 2019

Effect of Mental Health Conditions on Complications,

Conclusions: Comorbid MH conditions are associated with higher

e210 | www.jorthotrauma.com J Orthop Trauma  Volume 33, Number 6, June 2019

exact-match basis for CCI and gender; age was matched by

FIGURE 1. A, Forest plot of odds ratios for

Fewer Hospitals Provide Operative Fracture Care to

Conclusions: Fewer hospitals provide operative fracture care to

J Orthop Trauma  Volume 33, Number 6, June 2019 www.jorthotrauma.com | e215

TABLE 1. Demographic Characteristics of Study Cohort

Standardized Hospital-Based Care Programs Improve

Key Words: hip fractures, standardized care program, organized

J Orthop Trauma  Volume 33, Number 6, June 2019 www.jorthotrauma.com | e223

inpatient facility instead of home [77.3% vs. 81.5%, OR 0.72 DISCUSSION

Factors Predictive of Blocking Screw Placement in

J Orthop Trauma  Volume 33, Number 6, June 2019 www.jorthotrauma.com | e229

J Orthop Trauma Volume 33, Number 6, June 2019 www.jorthotrauma.com | 267

J Orthop Trauma Volume 33, Number 6, June 2019 www.jorthotrauma.com | 269

J Orthop Trauma Volume 33, Number 6, June 2019 www.jorthotrauma.com | 277

284 | www.jorthotrauma.com J Orthop Trauma Volume 33, Number 6, June 2019

292 | www.jorthotrauma.com J Orthop Trauma Volume 33, Number 6, June 2019

J Orthop Trauma Volume 33, Number 6, June 2019 www.jorthotrauma.com | 301

308 | www.jorthotrauma.com J Orthop Trauma Volume 33, Number 6, June 2019

312 | www.jorthotrauma.com J Orthop Trauma Volume 33, Number 6, June 2019

e210 | www.jorthotrauma.com J Orthop Trauma Volume 33, Number 6, June 2019

J Orthop Trauma Volume 33, Number 6, June 2019 www.jorthotrauma.com | e215

J Orthop Trauma Volume 33, Number 6, June 2019 www.jorthotrauma.com | e223

J Orthop Trauma Volume 33, Number 6, June 2019 www.jorthotrauma.com | e229

e240 | www.jorthotrauma.com J Orthop Trauma Volume 33, Number 6, June 2019

e246 | www.jorthotrauma.com J Orthop Trauma Volume 33, Number 6, June 2019