Pe d i a t r i c I m a g i n g • O r i g i n a l R e s e a r c h

Wu et al.
Radiography and MRI for Pediatric Knee Pain

Pediatric Imaging
Original Research
Downloaded from www.ajronline.org by 122.52.137.249 on 10/11/15 from IP address 122.52.137.249. Copyright ARRS. For personal use only; all rights reserved

Marginal Value of Radiographs in

the Interpretation of MR Images
Obtained for Pediatric Knee Pain
Yen-Ying Wu1 OBJECTIVE. Radiographs are often obtained before MRI in the evaluation of pediatric
Robert C. Orth knee pain, but the value of these radiographs is undefined. The purpose of this study was to
R. Paul Guillerman determine the marginal value of knee radiographs in the interpretation of knee MR images
Wei Zhang requested by pediatric sports medicine and orthopedic subspecialists.
J. Herman Kan MATERIALS AND METHODS. Knee MRI examinations of 194 pediatric patients (mean
age, 14 [SD, 3.1] years; range 4–18 years) performed over a 3-year period were reviewed retro-
Wu Y, Orth RC, Guillerman RP, Zhang W, Kan JH spectively. Patients were separated into groups based on MRI findings: normal, ligamentous in-
jury, osteochondral injury, or both ligamentous and osteochondral injury. Two pediatric radiolo-
gists blinded to the MRI findings reviewed the knee radiographs in consensus and categorized
the findings into the same groups. Radiographic and MRI findings were compared, and the influ-
ence of radiographic findings on MRI interpretation was designated as noncontributory if radio-
graphic findings did not aid MRI interpretation, erroneous for false-negative and false-positive
radiographic findings, or helpful if radiographs aided MRI interpretation.
RESULTS. Radiographic findings were normal in 166 of 194 cases (86%). Among the
166, MRI findings were normal in 73 (44%) cases and abnormal in 93 (56%). Twenty-five of
28 patients (89%) with abnormal radiographic findings had abnormal MRI findings. Radio-
graphs were deemed helpful in 14 of the 25 cases (56%) and noncontributory in 11 (44%).
Overall, radiographs were helpful in 14 of 194 cases (7%), noncontributory in 84 (43%), and
erroneous in 96 (50%).
CONCLUSION. In the interpretation of knee MRI studies requested by pediatric sports
medicine and orthopedic subspecialists for knee pain, radiographs provide limited marginal
value. Reliable clinical predictors are needed to identify which subset of pediatric patients
with knee pain referred for MRI will benefit from the acquisition of knee radiographs.

M
RI is highly accurate for identify-
ing the underlying causes of knee
pain in children [1]. The Ameri-
can College of Radiology (ACR)
Appropriateness Criteria [2, 3] recommenda-
tion for knee pain imaging is radiography be-
Keywords: children, knee, MRI, pediatric, radiography
fore MRI. However, the requirement for radi-
DOI:10.2214/AJR.12.9323 ologist review of knee radiographs before
MRI can cause logistical challenges when ra-
Received May 29, 2012; accepted after revision diographs have not been obtained or have
August 14, 2012. been obtained at a facility other than that at
1
which MRI is being performed and are not
All authors: E. B. Singleton Department of Pediatric
immediately available. Delays in performing
Radiology, Texas Children’s Hospital, 6701 Fannin St,
No. 1280, Houston, TX 77030. Address correspondence knee MRI can occur when radiography has to
to Y. Wu (YWu@IIu.edu). be repeated or MRI rescheduled until outside
radiographs become available. This can lead
AJR 2013; 200:891–894 to frustration among referring physicians and
0361–803X/13/2004–891
patient families, increased costs, and increased
radiation exposure. At our institution, orthope-
© American Roentgen Ray Society dic surgeons and sports medicine specialists
most commonly refer pediatric patients for
knee MRI. The purpose of this study was to
determine the marginal value of radiographs
in the interpretation of knee MRI studies re-
quested by pediatric sports medicine and or-
thopedic specialists in the evaluation of sus-
pected internal derangement.
Materials and Methods
Patient Selection
After the study received institutional review
board approval, all pediatric knee MRI exami-
nations performed between September 2008 and
September 2011 at our facility were identified. In-
clusion criteria for the study were referral from
sports medicine or orthopedic surgery and a knee
radiograph obtained within 14 days of MRI. Im-
aging examinations were excluded if there was a
clinical history of infection, inflammatory arthri-
tis, neoplasm, postsurgical follow-up, or Blount

AJR:200, April 2013 891

 Wu et al.

TABLE 1: Comparison of Knee Radiographic and MRI Findings

Radiographs (%) MRI Finding (%)
Finding % No. Normal Ligamentous Osteochondral Ligamentous and Osteochondral
Normal 86 166 44 (73) 33 (55) 10 (17) 13 (21)
Ligamentous 1 2 50 (1) 50 (1)
Downloaded from www.ajronline.org by 122.52.137.249 on 10/11/15 from IP address 122.52.137.249. Copyright ARRS. For personal use only; all rights reserved

Osteochondral 13 26 8 (2) 38 (10) 54 (14)

Note—Values in parentheses are numbers of cases.

disease. A total of 264 knee MRI examinations
of 264 patients were excluded from the provision-
al study population of 458, leaving a final sam-
ple of 194 knee MRI examinations of 194 patients
(87 boys, 107 girls; mean age, 14 [SD, 3.1] years;
range, 4–18 years). Thirty radiographic examina-
tions randomly selected from those of the 194 pa-
tients yielded an average of 2.6 views for each ex-
amination (all examinations included the frontal
and lateral views; 53% also included the sunrise
view, oblique view, or both).
Data Review
In consensus, two certificate of added qualifi-
cation pediatric radiologists blinded to the MRI
findings reviewed the knee radiographs and cat-
egorized them into the following groups: normal,
suspected ligamentous injury, osteochondral inju-
ry (osteochondral lesion, fracture, or avulsion), or
both ligamentous and osteochondral injury. The
original report from each MRI examination was
reviewed, and cases were similarly grouped on
the basis of MRI findings as normal, ligamentous
injury, osteochondral injury, or both ligamentous
and osteochondral injuries.
Analysis
The correlation of radiographic findings with
the knee MRI interpretation was retrospectively
classified as helpful, noncontributory, or errone-
ous. Radiographs were considered helpful when
the radiographic findings were subjectively con-
sidered to have assisted in MRI interpretation. Ra-
diographs were designated noncontributory when
the findings were deemed not to aid in MRI inter-
pretation and erroneous when false-negative (nor-
mal radiographic and abnormal MRI) or false-
positive (abnormal radiographic and normal MRI)
findings were made.
Results
Of the 194 knee radiographs, 166 (86%) had
normal findings; findings on two radiographs
(1%) suggested ligamentous injury and on 26
(13%), osteochondral injury. Table 1 details
interpretation of the radiographic and MRI
examinations. There were 25 true-positive,
three false-positive, 73 true-negative, and 93
false-negative radiographic examinations, so
radiography had a sensitivity of 21%, speci-
ficity of 96%, and accuracy of 51% compared
with MRI as the reference standard. The neg-
ative predictive value of normal knee radio-
graphic findings was 44%, and the positive
predictive value of abnormal knee radiograph-
ic findings was 89%.
Twenty-five of 28 patients with abnormal
radiographic findings had abnormal MRI find-
ings (Table 2). Fourteen of the 25 true-posi-
tive radiographic findings (56%) were deemed
helpful in MRI interpretation. Twelve of the 14
patients had small fracture fragments or loose
bodies that were better visualized on radio-
graphs than on MR images (Fig. 1), one pa-
tient had physeal widening better seen on radio-
graphs than on MR images, and one patient had
tibial tuberosity fraying and fragmentation
better seen on radiographs than on MR images.
Five of the 11 cases (44%) in which true-
positive radiographic findings were consid-
ered noncontributory for MRI interpretation
involved osteochondral lesions, and six cases
involved fractures that were readily apparent
at MRI (Fig. 2). Three of the osteochondral
lesions did not exhibit additional findings at
MRI. One osteochondral lesion was associ-
ated with ligamentous injury at MRI, and the
other was associated with bone contusions at
MRI. Five of the patients with fractures had
MRI findings of ligamentous injury. One of
the fracture patients did not have additional
findings at MRI.
Approximately 50% (96/194) of all radio-
graphic examinations had false-negative or
false-positive findings on the basis of the MRI
interpretations. True-negative and true-positive
radiographic findings that were considered non-
contributory to MRI interpretation accounted for
43% (84/194) of all radiographic examinations.
True-positive radiographic findings helpful in
MRI interpretation accounted for approximate-
ly 7% (14/194) of all radiographic examinations.
Of these 14 radiographic examinations found
helpful in the interpretation of MRI examina-
tions, the diagnosis could have been made at
MRI without the radiographs, but the radio-
graphs increased diagnostic confidence.
Discussion
The ACR Appropriateness Criteria are evi-
dence-based guidelines for assisting referring
physicians in selecting the most appropriate
imaging modality for a specific clinical pre-
sentation; they are not recommendations for
imaging study interpretation. The ACR Ap-

TABLE 2: Breakdown of Abnormal Radiographic Findings Compared With MRI Findings

Finding Percentage of Cases No. of Cases Details
False-positive 11 (3) 2 Osteochondral injury on radiographs found to be ligamentous injuries at MRI
1 Ligamentous injury on radiographs, normal findings at MRI
True-positive but noncontributory 39 (11) 6 Fractures
5 Osteochondral lesions
True-positive and helpful 50 (14) 12 Small fracture fragments or loose bodies
1 Physeal widening
1 Tibial tuberosity fraying and fragmentation
Note—Values in parentheses are numbers of cases.

892 AJR:200, April 2013

 Radiography and MRI for Pediatric Knee Pain

have been validated in children [4]. Knee radi-

ography can be avoided when these rules are not
satisfied and the principal clinical concern of
the clinician is suspected internal derangement.
Many radiologists ritualistically review
radiographs before an MRI report is issued
Downloaded from www.ajronline.org by 122.52.137.249 on 10/11/15 from IP address 122.52.137.249. Copyright ARRS. For personal use only; all rights reserved

because radiographs are traditionally avail-

able. Though the diagnosis of a tumor or tu-
morlike lesion of bone greatly relies on the
radiographic findings [5], review of radio-
graphs is rarely required for diagnosis in our
specific patient subset because the value of
radiography over MRI is small, and radiogra-
phy can be safely deferred if deemed unnec-
essary by the referring clinician. Certainly,
the imaging workup for suspected neoplas-
tic or inflammatory conditions and uncertain
incidental MRI findings ought to include ra-
diographs. For this reason, our study popu-
A B lation excluded patients with a history of in-
fection, inflammatory arthritis, or neoplasm.
Although relatively inexpensive, knee ra-
diography is the most commonly performed
study for knee pain and is estimated to cost
more than $1 billion annually [6]. Previous
studies of the value of knee radiography for
knee pain have shown that radiographs have
low yield (less than 12%) in the diagnosis of
clinically significant fractures [ 6–8]. In con-
trast, history and physical examination have
been reported to have a high yield (greater
than 70%) for the evaluation of internal de-
rangement [9, 10]. The delayed diagnosis of
a fracture and subsequent delayed treatment
can lead to chronic morbidity [11]. If a frac-
ture or osteochondral lesion is suspected at
C D the time of initial injury or visit, the radio-
Fig. 1—14-year-old boy referred because of knee pain. Radiographs were deemed helpful for MRI graphic examination is not only fast and easy
interpretation. but also warranted. Our study provides ev-
A, Frontal radiograph shows medial femoral condylar chondroosseous fracture (arrow) with superior idence to give radiologists more flexibility
resolution.
B, Coronal proton density–weighted fat-suppressed MR image shows only marrow edema (arrow). at the time of MRI in clinical situations in
C, Sunrise view radiograph shows subtle medial patellar sleeve fracture (arrow). which patients have been evaluated but the
D, Axial proton density–weighted fat-suppressed MR image shows only marrow edema (arrow). radiograph is either unavailable or deemed
unnecessary by the referring clinician, par-
propriateness Criteria for nontraumatic knee ings are abnormal, only a small fraction (7% ticularly when the clinical concern is sus-
pain call for an initial imaging examination in this study) of knee radiographs are helpful pected internal derangement.
with anteroposterior and lateral radiographs. in MRI interpretation. Abnormal radiograph- This study had several limitations. First,
MRI is not indicated before physical exam- ic findings were most frequently helpful for the knee radiographic examinations consist-
ination or routine conventional radiography increasing diagnostic confidence when small ed largely of anteroposterior and lateral views.
[3]. The small marginal value of radiography fracture fragments or small osteochondral le- The marginal value of alternative radiograph-
revealed in our study in the care of pediat- sions were identified. Most of the radiograph- ic projections such as tunnel views and axi-
ric patients referred for MRI from pediatric ic findings were either noncontributory to the al patellar views was not defined in this study.
sports medicine and orthopedic surgery spe- MRI interpretation or erroneous. Second, we excluded patients imaged for sus-
cialists challenges the ACR criteria for this For patients with acute trauma to the knee, pected infection, inflammatory arthropathy, or
subset of patients with suspected internal de- numerous decision rules based on clinical neoplasm; patients with a history of previous
rangement. Our study showed low sensitivity predictors have been devised to determine the knee surgery; and patients with a chronic de-
and low negative predictive value of knee ra- appropriateness of knee radiography. The most velopmental abnormality such as Blount dis-
diographs in this setting. Even when the find- established are the Ottawa knee rules, which ease. Therefore, our results are limited to only a

AJR:200, April 2013 893

 Wu et al.

dren with knee pain referred from subspecial-

ty pediatric sports medicine and orthopedic
surgeons were often noncontributory, occa-
sionally misleading, and almost never altered
the interpretation of the knee MRI examina-
tion. Our data do not justify delay in interpre-
Downloaded from www.ajronline.org by 122.52.137.249 on 10/11/15 from IP address 122.52.137.249. Copyright ARRS. For personal use only; all rights reserved

tation of MRI examinations for this patient

population while outside radiographs are ob-
tained. We also recommend modification of
the ACR Appropriateness Criteria to allow
knee MRI as the first-line study in the care of
this patient population.

A B
C MRI is used as an adjunct to clinical ex-
subset, albeit a large subset, of patients referred
by pediatric sports medicine physicians and or-
thopedic surgeons. It is probably not appropri-
ate to generalize the results of this study to the
broader clinical setting of patients referred by
physicians who are not subspecialists. Last, the
retrospective design compared consensus in-
terpretations of the knee radiographs and knee
MR images rather than consisting of a blind-
ed consensus review of both the knee radio-
graphs and the MR images and review of the
history and physical examination findings.
Thus inherent biases related to clinical infor-
mation available to the radiologist who inter-
preted the MR images cannot be measured.
References
Fig. 2—17-year-old boy referred for knee pain.
Radiographs were considered noncontributory to
MRI.
A, Radiograph shows Segond fracture (arrow).
B, Coronal T1-weighted MR image delineates fracture
(arrow) equally as well as A.
C, Coronal proton density–weighted fat-suppressed
MR image delineates fracture (arrow) equally as well
as A.
The review of clinical predictors that can po-
tentially yield a higher proportion of help-
ful radiographs was beyond the scope of this
retrospective study. This question would be
better answered by a prospective study that
includes close collaboration with sports med-
icine and orthopedic colleagues. Because the
history and physical examination findings
were not reviewed, the frequency of adher-
ence of the referring clinicians to the Ottawa
and Pittsburgh decision rules is not known.
amination to establish a suspected diagnosis,
rule out alternative conditions, and acquire
additional information that will guide treat-
ment. Our study was prompted in part by the
frustration of our referring physicians with
logistic problems arising from the prerequi-
site for radiography before MRI when they
do not believe radiography will aid in a pa-
tient’s diagnosis. Should the MRI findings be
uncertain or abnormal, review of radiographs
or acquisition of radiographs after MRI be-
fore issue of the MRI report is standard.
Conclusion
We found that findings on radiographs ob-
tained before knee MRI examinations of chil-
1. Major NM, Beard LN Jr, Helms CA. Accuracy of
MR imaging of the knee in adolescents. AJR
2003; 180:17–19
2. American College of Radiology website. ACR Ap-
propriateness Criteria: acute trauma to knee. www.
acr.org/~/media/ACR/Documents/AppCriteria/
Diagnostic/AcuteTraumaKnee.pdf. Accessed De-
cember 6, 2012
3. American College of Radiology website. ACR Ap-
propriateness Criteria: nontraumatic knee pain. www.
acr.org/~/media/ACR/Documents/AppCriteria/
Diagnostic/NontraumaticKneePain.pdf. Accessed
December 6, 2012
4. Bulloch B, Neto G, Plint A, et al. Validation of the
Ottawa knee rule in children: a multicenter study.
Ann Emerg Med 2003; 42:48–55
5. Miller TT. Bone tumors and tumorlike condi-
tions: analysis with conventional radiography. Ra-
diology 2008; 246:662–674
6. Stiell IG, Greenberg GH, Wells GA, et al. Deriva-
tion of a decision rule for the use of radiography in
acute knee injuries. Ann Emerg Med 1995; 26:
405–413
7. Stiell IG, Wells GA, McDowell I, et al. Use of
radiography in acute knee injuries: need for clini-
cal decision rules. Acad Emerg Med 1995; 2:
966–973
8. Weber JE, Jackson RE, Peacock WF, et al. Clini-
cal decision rules discriminate between fractures
and nonfractures in acute isolated knee trauma.
Ann Emerg Med 1995; 26:429–433
9. Kocher MS, DiCanzio J, Zurakowski D, et al. Diag-
nostic performance of clinical examination and se-
lective magnetic resonance imaging in the evalua-
tion of intraarticular knee disorders in children and
adolescents. Am J Sports Med 2001; 29:292–296
10. Stanitski CL. Correlation of arthroscopic and
clinical examinations with MRI findings of in-
jured knees in children and adolescents. Am J
Sports Med 1998; 26:2–6
11. Gottsegen CJ, Eyer BA, White EA, Learch TJ, For-
rester D. Avulsion fractures of the knee: imaging
findings and clinical significance. RadioGraphics
2008; 28:1755–1770

894 AJR:200, April 2013