Copyright 1998 by The Journal of Bone and Joint Surgery, Incorporated

Development of a Patient-Reported Measure of Function of the Knee*

BY JAMES J. JRRGANG, M.S., P.T., A.T.Ct, PITTSBURGH, LYNN SNYDER-MACKLER, SC.D., P.T., A.T.C4, NEWARK, DELAWARE, ROBERT S. WAINNER, M.S., P.T., O.C.S., E.C.S.f, FREDDIE H. FU, M.D., AND CHRISTOPHER D. HARNER, M.D., PITTSBURGH, PENNSYLVANIA

Investigation coordinated through the Department of Physical Therapy, University of Pittsburgh School of Health and Rehabilitation Sciences, Pittsburgh

ABSTRACT: The purpose of the present study was to demonstrate the reliability, validity, and responsiveness of the Activities of Daily Living Scale of the Knee Outcome Survey, a patient-reported measure of functional limitations imposed by pathological disorders and impairments of the knee during activities of daily living. The study comprised 397 patients; 213 were male, 156 were female, and the gender was not recorded for the remaining twenty-eight. The mean age of the patients was 33.3 years (range, twelve to seventysix years). The patients were referred to physical therapy because of a wide variety of disorders of the knee, including ligamentous and meniscal injuries, patellofemoral pain, and osteoarthrosis. The Activities of Daily Living Scale was administered four times during an eight-week period: at the time of the initial evaluation and after one, four, and eight weeks of therapy. Concurrent measures of function included the Lysholm Knee Scale and several global measures of function. The subjects also provided an assessment of the change in function, with responses ranging from greatly worse to greatly better, at one, four, and eight weeks. The Activities of Daily Living Scale was administered to an additional sample of fiftytwo patients (thirty-two male and twenty female patients with a mean age of 31.6 years [range, fourteen to sixty-six years]) before and after treatment within a single day to establish test-retest reliability. Factor analysis revealed two dominant factors: one that reflected a combination of symptoms and functional limitations and the other, only symptoms. The internal
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. Funds were received in total or partial support of the research or clinical study presented in this article. The funding source was a grant from the Section on Research of the American Physical Therapy Association. tDepartment of Physical Therapy, University of Pittsburgh School of Health and Rehabilitation Sciences, Room 6058, Forbes Tower, Meyran Avenue, Pittsburgh, Pennsylvania 15260. E-mail address for Dr. Irrgang: irrgang@newton.isd.upmc.edu. ^Department of Physical Therapy, University of Delaware, 309 McKinly Laboratory, Newark, Delaware 19716. Department of Orthopaedic Surgery, University of Pittsburgh School of Medicine, 1010 Kaufman Building, 3471 5th Avenue, Pittsburgh, Pennsylvania 15213.

consistency of the Activities of Daily Living Scale was substantially higher than that of the Lysholm Knee Scale (coefficient alpha, 0.92 to 0.93 compared with 0.60 to 0.73), resulting in a smaller standard error of measurement for the former scale. Validity was demonstrated by moderately strong correlations with concurrent measures of function, including the Lysholm Knee Scale (r = 0.78 to 0.86) and the global assessment of function as measured on a scale ranging from 0 to 100 points (r = 0.66 to 0.75). Analysis of variance with repeated measures revealed significant improvements in the score on the Activities of Daily Living Scale during the eight weeks of physical therapy ( F ^ = 108.13; p < 0.0001); post hoc testing indicated that the change in the score at eight weeks was significantly greater than the change at four weeks and that the change at four weeks was significantly greater than that at one week (p < 0.0001 for both). As had been hypothesized, the patients in whom the knee had somewhat improved had a significantly smaller change in the score, both at four weeks (F wro = 33.50; p < 0.001) and at eight weeks (FM56 = 22.48; p < 0.001), compared with those in whom the knee had greatly improved. The test-retest reliability coefficient (intraclass correlation coefficient! 2,11) was 0.97. These results suggest that the Activities of Daily Living Scale is a reliable, valid, and responsive instrument for the assessment of functional limitations that result from a wide variety of pathological disorders and impairments of the knee. A complete system for the evaluation of clinical outcomes should include measures in each disability domain (active pathology, impairment, functional limitations, and disability) described by Nagi. The state of active pathology involves an "interruption of or interference with normal processes and the simultaneous efforts of the organism to regain a normal state." Such a disorder may arise from infection, trauma, metabolic imbalance, degenerative disease, or other etiologies'7. Clinical measures of active pathology in orthopaedics include information defining the extent of the injury, disease, or healing as determined from the interview with the patient, the medical history, the physical examination, the laboratory and imaging studies, and the
THE JOURNAL OF BONE AND JOINT SURGERY

1132

DEVELOPMENT OF A PATIENT-REPORTED MEASURE OF FUNCTION OF THE KNEE

1133

operation. Clinical measures of impairment of the musculoskeletal system include findings at the clinical examination (for example, the range of motion, muscle function, and joint stability). Measures of functional limitations and disability include performance-based clinical assessments, such as the one-leg-hop test and patient-reported assessment. The difficulties with use of performance-based measures of function in the clinical setting, and the dearth of normative data for interpretation, have led practitioners to consider alternatives such as patient-reported measures of function. Patient-reported measures of function include general and specific measures of health status. General measures of health status are applicable across a number of disease processes and across demographic and cultural subgroups815. General measures of health status that have been used for patients who have orthopaedic conditions include the Medical Outcomes Study Short Form-36 (SF-36)1W420 and the Sickness Impact Profile (SIP)26. Specific measures of health status are designed to focus on aspects of health that are specific to the primary condition of interest, with the intent of creating a more responsive measure. Specific patientreported measures of function of the knee include the Lysholm Knee Scale25, the Cincinnati Knee Scale19, and the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) 3 . Several instruments combine patient-reported measures of function with measures of physical impairment. These include the guidelines established by the International Knee Documentation Committee (IKDC) 9 for the evaluation of knee-ligament injuries and those developed by the Knee Society11 for the assessment of outcome following total knee arthroplasty. The available patient-reported measures of function of the knee31925 were developed for specific pathological conditions, such as osteoarthrosis and injuries of the ligaments. Their content is therefore limited to the common symptoms and functional limitations associated with the disorder for which they were developed. This limited content makes it difficult to select an instrument for the assessment of patients who have concomitant pathological conditions, such as a ligamentous injury combined with patellofemoral pain or osteoarthrosis. The content of the Cincinnati Knee Scale, for example, is limited to items that pertain to symptoms and functional limitations (such as pain, instability, and the inability to run, cut, pivot, or jump) that are associated with ligamentous or meniscal injury, or both. The usefulness of this scale is therefore limited for individuals who have concomitant ligamentous and patellofemoral symptoms or osteoarthrosis. The alternative is the use of multiple scales, but this may make comparisons across different disease states difficult. An additional problem with most existing measures of knee function is the lack of a rationale for the scaling systems used to establish a score. For example, pain and
VOL. 80-A, NO. 8, AUGUST 1998

instability account for 50 per cent of the total score on the Lysholm Knee Scale, making this instrument very dependent on these symptoms. Changes in pain and instability therefore result in greater changes in the total score than do changes in other items in the scale. There are no objective data supporting the use of this scaling system. Conversely, the overall rating provided by the IKDC guidelines is determined by the worst rating for any item. The overall rating therefore will not change unless there is improvement in the item that received the worst rating, even if there is improvement in all other items. The Knee Outcome Survey was developed at the University of Pittsburgh as a patient-reported instrument for the measurement of functional limitations commonly experienced by individuals who have various pathological disorders of the knee, including ligamentous and meniscal injuries, patellofemoral pain, and osteoarthrosis. It also was designed to address the limitations inherent in patient-reported measures of function. The Knee Outcome Survey consists of two separate scales: the Activities of Daily Living Scale and the Sports Activity Scale. The Activities of Daily Living Scale includes items related to symptoms and functional limitations experienced during activities of daily living, while the Sports Activity Scale consists of items related to symptoms and functional limitations experienced during sports activities. The purpose of the current study was to assess the psychometric properties of the Activities of Daily Living Scale and, specifically, to determine the test-retest reliability, internal consistency, concurrent and construct validity, and responsiveness of this instrument. Materials and Methods The Activities of Daily Living Scale The Activities of Daily Living Scale of the Knee Outcome Survey is an instrument for the assessment of the full spectrum of symptoms and functional limitations that may occur as a result of a wide variety of pathological disorders and impairments of the knee (Appendix). The scale was developed on the basis of a review of existing instruments, including the Cincinnati Knee Scale'9, the Lysholm Knee Scale25, the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) 3 , and the guidelines developed by the International Knee Documentation Committee (IKDC) 9 . The items were selected to reflect problems reported by patients who have ligamentous or meniscal injuries, patellofemoral pain, or osteoarthrosis. The initial list of items was reviewed by twelve physical therapists who specialized in rehabilitation therapy for musculoskeletal conditions and who had extensive experience in the evaluation and treatment of the knee. This review resulted in the addition or modification of several items. The authors of some instruments have argued that the content of an instrument should be defined on the

1134

J. J. IRRGANG ET AL.

basis of interviews with patients who have the condition of interest216. We relied not only on a review of existing instruments but also on the opinion of those who had expertise in the treatment of a variety of conditions involving the knee. This methodology has been used by others to develop a patient-reported measure of disability of the upper extremity10. While it is possible that the method may have resulted in underrepresentation by the construct, we believe that the current instrument is an adequate representation of the functional limitations and disability that may result from many different pathological conditions and impairments affecting the knee. The symptoms that were included in the Activities of Daily Living Scale were pain, crepitus, stiffness, swelling, instability, and weakness. The responses to each item were graduated in terms of the functional limitations that the symptom imposed on the individual during activities of daily living. The responses ranged from absence of the symptom to complete loss of function due to the symptom. The functional limitations that were included in the scale were difficulty with regard to walking on level surfaces and stairs, standing, kneeling, squatting, sitting, and rising from a sitting position. The response to each item ranged from no limitation in performing the activity to an inability to perform it. An ordinal scoring system was used to assign a value to the responses, with a lower level of function resulting in a lower score. Only four logical responses could be developed for Item 17, which is related to the ability to rise from a chair. In order to avoid placing less weight on this item as a result of the smaller number of responses, the highest and lowest responses were assigned the same value as those for the other items and the scores for the intermediate responses were assigned in a manner to create roughly equal intervals. This resulted in a variance for the item that was similar to that for the other items. Concurrent Measures of Function Related to Injury of the Knee Two global measures of function as well as the Lysholm Knee Scale25 were used as alternative measures of the construct measured by the Activities of Daily Living Scale. To provide a global rating of function, the patient was asked to assess the level of function on a scale ranging from 0 to 100 points, with 100 points representing the level of function before the injury and 0 points indicating complete loss of function due to the injury of the knee. The patient's subjective assessment of function according to the IKDC guidelines also was used as a global measure of function, by having the patient respond to the questions "How does your knee function?" and "How does your knee affect your level of activity?" The responses to these questions were limited to "normal," "nearly normal," "abnormal," and "severely abnormal." The Lysholm Knee Scale is a patient self-report consisting of eight items related to the ability to walk

without assistive devices or a limp; the presence of locking, pain, instability, or swelling; and the ability to squat and to climb stairs. This instrument was chosen as a concurrent measure of the construct because it is limited primarily to an assessment of symptoms and function associated with activities of daily living. Assessment of Internal Consistency, Validity, and Responsiveness The data for the present study were collected at nine physical-therapy centers: Berkshire Institute, Wyomissing, Pennsylvania; University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania; CORE Network, McKeesport, Pennsylvania; East Suburban Sports Medicine, Monroeville, Pennsylvania; Health South Birmingham, Birmingham, Alabama; North Hills Orthopedics and Sports Physical Therapy, Wexford,Pennsylvania; St. Margaret's Memorial Hospital, Pittsburgh, Pennsylvania; Western Wisconsin Sports Medicine, La Crosse, Wisconsin; and University of Delaware Physical Therapy Clinic, Newark, Delaware. All patients who were referred to physical therapy for evaluation and treatment of the knee were eligible for participation. A variety of pathological disorders and types of impairment were included so that the usefulness of the instrument could be assessed for the entire target population for which it was developed and in order to enhance generalizability of the results. Individuals were excluded from participation in the study if they had a pathological disorder or impairment involving both knees or if they had another condition that might affect the function of the lower extremity. Individuals who met the criteria for inclusion were recruited for participation in the study, and informed consent was obtained. Baseline demographic data, including the date of the initial visit, date of birth, date of the injury, gender, diagnosis, and type of operative treatment, were collected at the time of the initial visit. The Activities of Daily Living Scale, the Lysholm Knee Scale, and the global measures of function also were administered at that time. After collection of the baseline information, the patient was evaluated and treated by the physical therapist. No attempt was made to standardize the treatment as this study was not designed to assess the effectiveness of any particular type of physical therapy. The Activities of Daily Living Scale, the Lysholm Knee Scale, and the global measures of function were readministered at one, four, and eight weeks after the initial visit. At each follow-up interval, the patient also provided an assessment of the change in function that had occurred since the initiation of treatment. The responses included "greatly worse," "somewhat worse," "no change," "somewhat better," and "greatly better." These ratings were used to assess the reliability and responsiveness of the instrument. Throughout the period of followup, the patient continued to receive treatment as needed. The patient-reported measures of function were sent by
THE JOURNAL OF BONE AND JOINT SURGERY

DEVELOPMENT OF A PATIENT-REPORTED MEASURE OF FUNCTION OF THE KNEE

1135

mail to patients who had been discharged from physical therapy before the eight-week follow-up interval. All data were returned to the coordinating center for subsequent analysis. A computerized database was developed to generate letters reminding the physical therapist that the survey should be readministered at the fourth and eighth weeks. These letters were sent one week before the date for readministration. A second letter was automatically sent if a response was not received within one week of the due date. Three hundred and ninety-seven patients were enrolled in the study during a one-year period. There were 213 male and 156 female patients; the gender of the remaining twenty-eight was not recorded. The patients ranged in age from twelve to seventy-six years (mean age, 33.3 years; median age, 30.3 years). The median interval between the injury and the referral was sixtyfive days (range, one day to 26.5 years). The follow-up instrument was completed by 67 per cent (266) of the patients at one week, by 55 per cent (218) of the patients at four weeks, and by 45 per cent (179) of the patients at eight weeks. The primary diagnoses for the patients participating in this study were ligamentous or meniscal injury, patellofemoral pain, and degenerative osteoarthrosis (Table I). Two hundred and twenty-five patients (57 per cent) were referred for physical therapy after operative treatment. The interval between the operation and the referral ranged from zero to 739 days (mean, 29.1 days; median, eight days). The operative procedures included ligamentous reconstruction, meniscectomy, patellofemoral realignment, arthroscopy, high tibial osteotomy, and total knee arthroplasty (Table I). Test-Retest Reliability A critical premise for valid assessment of test-retest reliability is that the underlying condition measured by the instrument remains stable between repeated measures12. Our original plan was to assess test-retest reliability as the degree of concordance between the scores on the Activities of Daily Living Scale at the time of the initial visit and at one week. After one week of treatment, however, two patients (1 per cent) reported that the knee was somewhat worse, 167 (65 per cent) reported some improvement, and eighty-nine (34 per cent) reported great improvement. (Eight patients did not provide a global rating of the change in function at one week.) No patient reported that the status of the knee was unchanged. Because the underlying construct measured by the Activities of Daily Living Scale did not remain stable over a one-week period, the original data set could not be used to estimate test-retest reliability. Consequently, a second sample of fifty-two patients was recruited, from two of the original nine physical-therapy centers, for the assessment of testretest reliability before and after treatment within the
VOL. 80-A, NO. 8, AUGUST 1998

TABLE I
CHARACTERISTICS OF THE SAMPLES*

Sample 1

Sample 2 52 50% (26) 19% 15% 4% 4% 0% 4% 4% 39 44% 13% 13% 8% 5% 5% 13% (10) (8) (2) (2) (0) (2) (2)

Diagnosis
No. of patients Ligamentous or meniscal injury Patellofemoral pain Osteoarthrosis Tendinitis Fracture Plica syndrome Other Not recorded Operative procedure No. of patients Reconstruction of ligament Meniscectomy Patellofemoral realignment Arthroscopy High tibial osteotomy Total knee arthroplasty Other 197 57% (227) 20% 9% 2% 1% 1% 7% 3% (78) (34) (7) (5) (4) (29) (13)

>25 51% (115) 19% (43) 4% (10) 22% (50) 1% (3) 0.4% (1) 1% (3)

(17) (5) (5) (3) (2) (2) (5)

T h e numbers in parentheses after the percentages represent the number of patients. Sample 1 included patients who participated in the assessment of internal consistency, validity, and responsiveness, and Sample 2 included those who participated in the assessment of testretest reliability.

same day. All patients receiving physical therapy for the knee were eligible. The criteria for inclusion and exclusion were identical to those used for the original sample. All fifty-two subjects gave informed consent before participation. As was done for the sample of 397 patients, demographic data (the date of the initial visit, of birth, and of the injury; gender; diagnosis; and type of operative treatment) were collected, and the baseline Activities of Daily Living Scale was administered. The patient then received treatment as determined by the physical therapist. Again, no attempt was made to standardize treatment. Immediately after treatment, the Activities of Daily Living Scale was readministered. The data on the fifty-two patients were collected over a two-week period. There were thirty-two male and twenty female patients, who ranged in age from fourteen to sixty-six years (mean age, 31.6 years; median age, 28.2 years). The median interval between the injury and the referral was eighteen days (range, thirteen days to one year). The diagnoses and operative procedures for the patients participating in the reliability study were similar to those for the patients who participated in the assessment of internal consistency, validity, and responsiveness (Table I). Thirty-nine patients (75 per cent) were referred for physical therapy after operative treatment. The interval between the operation and the referral ranged from six to 368 days (mean, 83.6 days; median, seventy days). Data Management and Analysis The score on the Activities of Daily Living Scale was calculated by summing the point values for the re-

1136
100 90

J. J. I R R G A N G ET AL.

H ADLS Lysholm Global Rating

lected at the initial visit (baseline) as well as at one, four, and eight weeks. We hypothesized that one dominant component or factor would be identifiable and would represent the construct of functional limitations during activities of daily living that were imposed by the pathological condition and impairment of the knee. Reliability The internal consistency of the Activities of Daily Living Scale and the Lysholm Knee Scale was determined for each administration of the instruments with use of the coefficient alpha. The test-retest reliability was estimated as the degree of concordance between repeat administrations of the Activities of Daily Living Scale before and after treatment within the same day with an intraclass correlation coefficient (formula 2,1)22. This formula is appropriate for the assessment of testretest reliability when occasions are considered to be randomly selected from a larger population of occasions, and it allows for generalization of the results to other, similar occasions. Validity Construct validity was assessed by establishing the relationship between an individual's global rating of function on a scale ranging from 0 to 100 points and the score on the Activities of Daily Living Scale at each administration with use of a Pearson product-moment correlation coefficient. It was hypothesized that there would be moderately strong (for example, r > 0.60) positive correlations between the score on the Activities of Daily Living Scale and the global rating of function. Additionally, it was hypothesized that the correlation between the score on the Activities of Daily Living Scale and the global rating of function would be stronger than that between the score on the Lysholm Knee Scale and the global rating of function. Differences in the strength of the correlation coefficients at each point in time were tested with the Fisher r to z transformation. The a priori level of significance for each of these tests was set at 0.05. A one-way analysis of variance was performed to determine the relationship between the individual's subjective assessment of function according to the IKDC guidelines and the score on the Activities of Daily Living Scale. Specifically, it was hypothesized that the score on the scale would be associated with the responses to the questions "How does your knee function?" and "How does your knee affect your level of activity?" The responses ("normal," "nearly normal," "abnormal," and "severely abnormal") were used as the grouping variable for the analysis of variance to determine the relationship between the score on the Activities of Daily Living Scale and the global rating of function. Post hoc testing was performed with use of the Scheffe procedure to detect differences among patients whose responses had been "normal," "nearly normal," "abnormal," or "severely abnormal."
THE JOURNAL OF BONE AND JOINT SURGERY

Initial Visit

1 Week

4 Weeks

8 Weeks

FIG. 1

Graph showing the mean scores on the Activities of Daily Living Scale (ADLS) and the Lysholm Knee Scale as well as the mean global rating of function at the initial visit and at one, four, and eight weeks.

sponses to all seventeen items on the scale; dividing by 80, which is the total possible number of points for all of the items on the scale; and multiplying by 100, with the final total score expressed as a percentage. Only individuals who responded to all items on the scale were included in the analysis of the data. The score on the Lysholm Knee Scale was calculated by summing the point values for the responses to all eight items. A score of 100 per cent on the Activities of Daily Living Scale and the Lysholm Knee Scale indicated that the individual had no symptoms or functional limitations related to the knee. All data were entered into a computerized database for analysis. Statistical analyses were performed with version 7.5 of SPSS for a personal computer (SPSS, Chicago, Illinois) and version 7.1 of BMDP (BMDP Statistical Software, Los Angeles, California) for a VAX/VMS mainframe system. The initial step in the data analysis was to perform a factor analysis to determine the dimensions of the Activities of Daily Living Scale. The underlying construct that this scale is intended to measure is the level of function that the pathological condition or impairment of the knee imposes during activities of daily living. If the scale is able to measure this construct, the factor analysis should identify one dominant component or factor that represents this ability. This should result in one factor that has an eigenvalue (the amount of variance that is accounted for by a factor) that is much larger than all of the others. It is recommended that factors that have an eigenvalue of greater than 1.0 be retained. The rationale for this is that, for standardized data, the amount of variance extracted by a factor should at least be equal to the variance of at least one variable21. The factor analysis was performed with use of a maximum-likelihood procedure. Separate exploratory factor analyses were performed on data that were col-

DEVELOPMENT OF A PATIENT-REPORTED MEASURE OF FUNCTION OF THE KNEE TABLE II

DESCRIPTIVE DATA ON THE ACTIVITIES OF DAILY LIVING SCALE

1137

Measure Baseline 1 week 4 weeks 8 weeks

Mean (per cent) 55.2 61.2 71.7 77.4

Standard Deviation (per cent) 20.8 20.0 18.2 17.2

Median (per cent) 53.8 61.3 72.5 82.5

Skewness 0.059 -0.072 -0.488 -0.834

Standard Error 0.131 0.149 0.165 0.182

Kurtosis -0.753 -0.740 -0.412 0.040

Standard Error 0.262 0.298 0.328 0.361

Floor Effect* (per cent) 0 0 0 0

Ceiling Effectt (per cent) 0 1.5 0.9 5.6

*The percentage of patients who had the minimum score (0 points). tThe percentage of patients who had the maximum score (100 points).

Concurrent validity was evaluated by determining the relationship between the score on the Activities of Daily Living Scale and that on the Lysholm Knee Scale with use of the Pearson product-moment correlation coefficient. It was hypothesized that there would be moderately strong (for example, r > 0.60) positive relationships between the scores on the Activities of Daily Living Scale and this concurrent measure of function during activities of daily living. The a priori level of significance was set at 0.001 for the Pearson productmoment correlations and at 0.05 for the one-way analysis of variance and the Scheffe post hoc procedure. Responsiveness Responsiveness was evaluated by determining the change in the score on the Activities of Daily Living Scale between the time of the initial administration and the one, four, and eight-week administrations. These changes in the score were calculated by subtracting the initial score from the follow-up score. A positive change in the score reflected an improved level of function. It was hypothesized that the level of function would improve over the eight-week course of treatment. Specifically, it was hypothesized that the magnitude of improvement at eight weeks would be greater than that at four weeks and that the magnitude of improvement at four weeks would be greater than that at one week. This hypothesis was assessed with a one-way repeatedmeasures analysis of variance of the changes in the score at one, four, and eight weeks, with the a priori level of significance set at 0.05. To illustrate the magnitude of the change in the score, effect sizes were calculated as the mean change in the score from the time of the initiation of treatment to the one, four, and eight-week follow-up intervals, divided by the standard deviation of the initial score23. To assess responsiveness further, it was hypothesized that the magnitude of the change in the score at one, four, and eight weeks would be dependent on the change reported by the individual since the initiation of treatment. To assess this hypothesis, a one-way analysis of variance was performed to determine the relationship between the change in the score and the individual's rating of such change since the initiation of treatment. It was hypothesized that patients who reported that the knee had greatly improved would have a larger change
VOL. 80-A, NO. 8, AUGUST 1998

in the score than those who reported that the knee was somewhat better and that those who indicated that the knee was somewhat better would have a larger change in the score than those who reported no change or worsening of the status of the knee. After a significant F test, these hypothesized differences were tested with use of the Scheffe post hoc procedure. The a priori level of significance for the one-way analysis of variance and the Scheffe post hoc test was set at 0.05. Results In general, the mean scores on the Activities of Daily Living and Lysholm Knee Scales and the global rating of function improved over the course of the eightweek program (Fig. 1). Descriptive statistics, including the mean and standard deviation, median, skewness, and kurtosis, were recorded for the Activities of Daily Living Scale for each point in time (Table II). The mean scores (and standard deviation) at baseline and at one, four, and eight weeks were 55.2 20.8,61.2 20.0,71.7 +18.2, and 77.4 17.2 per cent, respectively. The median values at baseline and at one, four, and eight weeks were 53.8, 61.3, 72.5, and 82.5 per cent, respectively. The scores at baseline and at one week were not skewed but were slightly platykurtic (that is, their distribution was flatter than a normal distribution), whereas the scores at four and eight weeks were negatively skewed but were not kurtotic (that is, they clustered around the central point of distribution in a manner similar to that seen in normal distribution). There were no floor effects at any point in time. There were minimum ceiling effects at one, four, and eight weeks. Factor Analysis Factor analysis with use of a maximum-likelihood procedure for each of the four administrations of the test revealed two factors that had eigenvalues of greater than 1.0, which, when combined, accounted for 61 to 63 per cent of the total variance in the score on the Activities of Daily Living Scale. The eigenvalues for factor one ranged from 8.2 to 8.6 and accounted for 48 to 51 per cent of the total variance in the score. The eigenvalues for factor two ranged from 1.8 to 2.3 and accounted for 11 to 13 per cent of the total

1138

J. J. I R R G A N G E T AL.

TABLE III
INTERNAL CONSISTENCY AND STANDARD ERROR OF MEASUREMENT FOR THE ACTIVITIES OF DAILY LIVING SCALE*

ra Activities of Daily Living Scale Lysholm Knee Scale 0.92 0.60

Baseline Standard Error of Measurement 5.9 12.5

Weekl Standard Error of Measurement 5.3 11.5

Week 4 Standard Error of Measurement 4.8 10.8

cc

Week 8 Standard Error of Measurement 4.6 9.7

m

0.93 0.64

0.93 0.68

0.93 0.73

*r = coefficient alpha correlation coefficient. The standard error of measurement was calculated as SEM = SD(1 - ra) where SD is the standard deviation of observed scores for the Activities of Daily Living Scale and the Lysholm Knee Scale.

variance. No other factor had eigenvalues of greater than 1.0. Factor one represented a combination of symptoms and functional limitations commonly experienced during activities of daily living by an individual who has a pathological condition or impairment of the knee. Items that loaded onto factor one were related to pain, stiffness, swelling, weakness, walking, use of crutches, limping, ascending and descending stairs, standing, kneeling on the front of the knee, squatting, sitting with the knee bent, and rising from a chair. Factor two represented only symptoms commonly experienced by an individual who has a pathological condition or impairment of the knee. Items that loaded onto factor two included pain, crepitus, stiffness, swelling, partial and full giving-way, and weakness. Items related to pain, stiffness, swelling, and weakness loaded onto both factors, which accounts for the correlation between factors that ranged from 0.28 to 0.49. Internal Consistency Internal consistency was calculated separately with coefficient alpha for each administration of the ActiviB Normal Nearly Normal Abnormal

ties of Daily Living Scale and the Lysholm Knee Scale (Table III). Coefficient alpha was generally higher for the Activities of Daily Living Scale than it was for the Lysholm Knee Scale at each administration. This resulted in a substantially smaller standard error of measurement for the score on the Activities of Daily Living Scale (Table III). The individual items that make up the Activities of Daily Living Scale were correlated with the total score on the scale without the item being included for each administration of the scale. The lowest correlations between the items and the total score over all administrations of the scale were those for crepitus (r = 0.19 to 0.40), slipping (r = 0.28 to 0.45), buckling (r = 0.34 to 0.50), use of assistive devices (r = 0.35 to 0.58), and kneeling on the front of the knee (r = 0.55 to 0.62). Higher correlations were found for pain (r = 0.58 to 0.78), stiffness (r = 0.70 to 0.74), swelling (r = 0.67 to 0.74), weakness (r = 0.66 to 0.76), limping (r = 0.70 to 0.73), walking (r = 0.74 to 0.81), ascending and descending stairs (r = 0.77 to 0.81), standing (r = 0.71 to 0.77), squatting (r = 0.71 to 0.74), sitting with the knee bent (r = 0.61 to 0.65), and rising from a chair (r = 0.66 to 0.74).

G Severely Abnormal

FIG. 2 Graph comparing the mean scores on the Activities of Daily Living Scale (ADLS) with the ratings of knee function at the initial visit and at one, four, and eight weeks. Post hoc testing revealed that all pairwise tests were significant (p < 0.01) except for the comparison between normal and nearly normal at the initial visit and at one week and that between abnormal and severely abnormal at eight weeks.

THE JOURNAL OF BONE AND JOINT SURGERY

D E V E L O P M E N T OF A PATIENT-REPORTED M E A S U R E O F F U N C T I O N O F T H E KNEE

1139

TABLE IV
INTERCORRELATIONS BETWEEN THE SCORES ON THE ACTIVITIES OF DAILY LIVING SCALE, THE LYSHOLM KNEE SCALE, AND THE GLOBAL RATING OF FUNCTION*

Correlation Score on Activities of Daily Living Scale and global rating Score on Lysholm Knee Scale and global rating Scores on Activities of Daily Living and Lysholm Knee Scales

Baseline 0.75

Week 1 0.68

Week 4 0.69

Week 8 0.66

0.54

0.52

0.57

0.53

0.78

0.81

0.84

0.86

"Correlations represent Pearson correlation coefficients.

Validity The correlations between the scores on the Activities of Daily Living Scale, the Lysholm Knee Scale, and the global rating of function were summarized at each point in time (Table IV). The relationships between the scores on the Activities of Daily Living Scale and the global rating of function were significantly higher (p < 0.05) than those between the scores on the Lysholm Knee Scale and the global rating of function at baseline and at one and four weeks; however, the difference between the correlation coefficients at eight weeks was not found to be significant. There were moderately strong (for example, r > 0.60) relationships between the scores on the Activities of Daily Living Scale and the Lysholm Knee Scale at each time-period. The mean scores on the Activities of Daily Living Scale were compared with the responses to the ques-

tions "How does your knee function?" (Fig. 2) and "How does your knee affect your activity level?" (Fig. 3). The results for each one-way analysis of variance were significant (p < 0.001). Post hoc testing with the Scheffe procedure revealed that all pairwise comparisons were significant (p < 0.01), with the following exceptions. For the question "How does your knee function?" the comparisons between normal and nearly normal at baseline (p = 0.998) and at one week (p = 1.00) and the comparison between abnormal and severely abnormal at eight weeks (p = 0.268) were not found to be significant, with the numbers available (Fig. 2). Similarly, for the question "How does your knee affect your level of activity?" the comparison between normal and nearly normal at baseline (p = 0.111) and that between abnormal and severely abnormal at eight weeks (p = 0.648) were not significant, with the numbers available (Fig. 3). As expected, higher ratings for function and level of activity were associated with higher scores on the Activities of Daily Living Scale. Responsiveness The mean change in the score on the Activities of Daily Living Scale after one, four, and eight weeks of physical therapy was recorded (Fig. 4). The one-way repeated-measures analysis of variance revealed significant improvement (F2236 = 108.13, p < 0.0001); post hoc testing with the Scheffe procedure indicated that the mean change in the score at eight weeks was significantly greater than that at four weeks and that the mean change in the score at four weeks was significantly greater than that at one week (p < 0.0001 for both). The effect size for the change in the score on the Activities

H Normal Nearly Normal

Abnormal Q Severely Abnormal

FIG.

Graph comparing the mean scores on the Activities of Daily Living Scale (ADLS) with the rating of overall level of activity at the initial visit and at one, four, and eight weeks. Post hoc testing revealed that all pairwise tests were significant except for the comparison between normal and nearly normal at the initial visit and that between abnormal and severely abnormal at eight weeks.
VOL. 80-A, NO. 8, AUGUST 1998

1140

J. J. IRRGANG ET AL.

1 Week

4 Weeks

8 Weeks

FIG. 4 Graph showing the mean change in the scores on the Activities of Daily Living Scale (ADLS) after one, four, and eight weeks of physical therapy. The change in the score was calculated as the follow-up score minus the baseline score. Positive values represent an improvement in function.

4 Weeks

8 Weeks

5 Graph comparing the mean change in the scores on the Activities of Daily Living Scale (ADLS) and the global rating of change after four and eight weeks of physical therapy. The change in the score was calculated as the follow-up score minus the baseline score. Positive values represent an improvement in function.
FIG.

of Daily Living Scale was 0.44,0.94, and 1.26 at one, four, and eight weeks, respectively. The mean change in the score on the Activities of Daily Living Scale was compared with the global rating of change at four and eight weeks (Fig. 5). At four and eight weeks, no patient reported that the knee was greatly worse or unchanged; therefore, these categories were not included in the analysis. Additionally, only six patients reported that the knee was somewhat worse (four, at four weeks, and two, at eight weeks). These patients were eliminated from the analysis because of the disparity in the sample size and the unequal variances with respect to those who reported some or great improvement. As hypothesized, the analysis of variance indicated that the patients who said that the knee was somewhat improved had a significantly smaller change in the score compared with those who said that it was greatly improved, both at four weeks (Flil89 = 33.50, p < 0.001) and at eight weeks (F, m = 22.48, p < 0.001). Test-Retest Reliability As noted, within-day test-retest reliability was assessed by administering the Activities of Daily Living

Scale to fifty-two subjects before and after a single treatment session. The mean score (and standard deviation) on the Activities of Daily Living Scale was 64.3 18.4 per cent (range, 33.8 to 98.8 per cent) before treatment and 65.8 19.0 per cent (range, 31.3 to 98.8 per cent) after it. The intraclass correlation coefficient(2,l) was 0.97. The standard error of measurement with use of the test-retest reliability correlation coefficient was 3.2. Discussion The results of the current study indicate that the Activities of Daily Living Scale can be used to assess functional limitations resulting from a wide range of conditions affecting the knee. The scale appears to be reliable, valid, and responsive for the measurement of function related to pathological disorders and impairments of the knee. The results of the factor analysis yielded two factors. Items that loaded onto factor one included a combination of functional limitations and symptoms (pain, stiffness, swelling, weakness, walking, use of crutches, limping, ascending and descending stairs, standing, kneeling on the front of the knee, squatting, sitting with
THE JOURNAL OF BONE AND JOINT SURGERY

D E V E L O P M E N T OF A PATIENT-REPORTED M E A S U R E OF FUNCTION OF T H E KNEE

1141

the knee bent, and rising from a chair), while those that loaded onto factor two included only symptoms (pain, crepitus, stiffness, swelling, partial and full giving-way, and weakness). Symptoms that loaded onto both factors included pain, stiffness, swelling, and weakness, accounting for a correlation between factors that ranged from 0.28 to 0.49. Symptoms that loaded only onto factor two included crepitus and partial and full giving-way of the knee. In the current sample of patients, pain, stiffness, swelling, and weakness had a greater influence on function than did crepitus and instability, as evidenced by the loading of pain, stiffness, swelling, and weakness onto factor one. Crepitus by itself is not disabling; despite a recent report 1 suggesting analysis of recorded knee sounds to identify damage to the articular cartilage, we are not aware of any study that has demonstrated a relationship between crepitus and function. In the current study, there were relatively weak correlations between crepitus and items that reflected function (r = -0.17 to 0.41). Instability often is associated with functional limitations. Paradoxically, in the current study, correlations between instability and items that represented functional limitations were low (r = -0.10 to 0.45). The functional activities represented on the Activities of Daily Living Scale are low-level activities that are unlikely to be strongly affected by instability. A stronger relationship would be expected between instability and items that reflect higher levels of function, such as running, jumping, cutting, and pivoting. In order to maintain the content validity of the Activities of Daily Living Scale, which was designed to assess functional limitations related to a wide spectrum of pathological conditions and impairments of the knee, it was decided to include the items that loaded only onto factor two (that is, crepitus and instability) in the total score of the scale. The inclusion of both factors captures the full range of symptoms and functional limitations that may occur as a result of a wide variety of pathological disorders and impairments of the knee, including ligamentous and meniscal injuries, patellofemoral conditions, and osteoarthrosis. The high values for coefficient alpha for the Activities of Daily Living Scale indicate that the items consistently measure the underlying construct of functional limitations that occur during activities of daily living as a result of pathological disorders and impairments of the knee. In contrast, the values for coefficient alpha for the Lysholm Knee Scale were substantially lower. The underlying concept for internal consistency is that the consistency with which an individual responds from one item to the next can be used to provide an estimate of reliability for the total test score18. High internal consistency implies that the items are homogeneous with regard to the measurement of the underlying trait or attribute12.
VOL. 80-A, NO. 8, AUGUST 1998

The difference in internal consistency between the two scales has implications for the precision of measurement as evidenced by the differences in the standard error of measurement for the two instruments. The score on the Activities of Daily Living Scale appears to give a more precise estimate of the true level of function. For example, if an individual has an observed score of 60 per cent on the Activities of Daily Living Scale and the Lysholm Knee Scale, the 95 per cent confidence interval for the true score on the Activities of Daily Living Scale ranges from 48.4 to 71.7 per cent, while that on the Lysholm Knee Scale ranges from 35.5 to 84.5 per cent. The greater precision of measurement demonstrated by the Activities of Daily Living Scale is primarily a function of higher internal consistency, since the standard deviations for the two scales are of the same magnitude. The greater precision of measurement demonstrated by the Activities of Daily Living Scale has implications when the sample sizes for clinical trials involving use of the scale as a dependent variable are estimated. The Activities of Daily Living Scale demonstrated acceptable concurrent and construct validity as evidenced by the hypothesized relationships with the global measures of function and the Lysholm Knee Scale. The positive correlations between the scores on the Activities of Daily Living Scale and the other measures of the construct indicate that higher scores on the Activities of Daily Living Scale are associated with higher levels of function after a knee injury. The correlations between the global rating of function and the scores on the Activities of Daily Living Scale were substantially higher than those between the global rating of function and the scores on the Lysholm Knee Scale. This indicates that the Activities of Daily Living Scale is a better measure of an individual's perceived level of function than is the Lysholm Knee Scale. The ability to demonstrate the validity of this instrument is hampered by lack of a criterion (gold-standard) measure of functional limitations that occur as a result of pathological disorders and impairments of the knee. Therefore, the global rating of function and the Lysholm Knee Scale were used as proxies. Global ratings by patients have been used to validate other patient-reported measures of outcome4'6724. The Lysholm Knee Scale was selected as a proxy for the construct because it has been widely used in previous studies and because the content of the scale is similar to that of the Activities of Daily Living Scale. Both scales contain items related to symptoms and functional limitations experienced during activities of daily living. Responsiveness refers to an instrument's ability to detect real changes in the construct that it is intended to measure. Stratford et al.23 described several designs that can be used to assess responsiveness. In the current study, two approaches were used. It was hypothesized that, in general, an individual's level of function would

1142

J. J. IRRGANG ET AL.

improve over an eight-week course of physical therapy and that this would be accompanied by improvements in the score on the Activities of Daily Living Scale. Additionally, it was hypothesized that the magnitude of the change in the score on the Activities of Daily Living Scale would be related to the individual's own rating of change since the initiation of treatment. The results of this study indicate that the magnitude of the change in the score since the initiation of treatment increases over time and that it is related to the individual's impression of change since the initiation of treatment. These findings lend support to the claim that the scale is responsive. The mean change in the score on the Activities of Daily Living Scale was 19.5 at four weeks and 26.2 at eight weeks. From a practical point of view, this represents an improvement of approximately 0.9 for each item (almost one response level per item) at four weeks and an improvement of 1.2 for each item at eight weeks. The effect size22, calculated as the mean change in the score divided by the standard deviation of the initial scores, was 0.94 at four weeks and 1.26 at eight weeks. Effect sizes of greater than 0.80 are considered large5. In contrast, the effect size for the Lysholm Knee Scale was 0.82 at four weeks and 1.13 at eight weeks. Initially, this study was designed to estimate the test-retest reliability of the Activities of Daily Living Scale when the instrument was readministered after a one-week interval. The underlying tenet for test-retest reliability is that the test score should remain stable when the underlying construct being measured by the scale remains stable8. After one week of treatment, no patient reported that the condition of the knee was unchanged. Therefore, the Activities of Daily Living Scale was administered before and after a single treatment session in order to estimate test-retest reliability. The resulting test-retest reliability coefficient was 0.97. Similar values have been reported for the Lysholm Knee Scale when it was administered twice within the same day25. Patients who receive physical therapy have a rapid increase in function, as demonstrated by the improvement in the global rating of change and the change in the score on the Activities of Daily Living Scale over a one-week period. As a result of this rapid change in function, it was necessary to estimate test-retest reliability by administering the Activities of Daily Living Scale twice within a very short period. While it was not possible to determine if an individual's level of function truly changed after a single session of physical therapy, such a change would result in a lower estimate of test-retest reliability. Thus, the high level of test-retest reliability that was observed in this study implies that the level of function remained relatively stable within a single session of physical therapy. The estimation of test-retest reliability within the same day may be confounded by the individual's recall of responses from

the first administration of the test. Thus, the test-retest reliability coefficient must be interpreted cautiously. Additional research is needed to estimate test-retest reliability for patients who have a disorder that is truly stable over a longer period of time, such as those who have a chronic, slowly progressive condition (for example, osteoarthrosis) or those who are not actively receiving treatment. On the basis of the results of this study, the Activities of Daily Living Scale appears to be a useful instrument for the measurement of functional limitations during activities of daily living experienced by individuals who have a wide variety of pathological conditions of the knee, including ligamentous and meniscal injuries, patellofemoral pain, and osteoarthrosis. The instrument may be used to assess the level of function at a given point in time and to assess changes in function over time. Because the scale was designed to assess limitations during activities of daily living, a ceiling effect may occur when it is used for athletes and others who are involved in vigorous physical activities. The Knee Outcome Survey also includes a Sports Activity Scale, which will be the subject of a future study. In conclusion, the testing of a large multicenter population demonstrated that the Activities of Daily Living Scale is internally consistent, valid, and responsive to changes in an individual's level of function and that the score remains stable when the level of function is unchanged. Additional testing is needed to demonstrate test-retest reliability over a longer period of time and to determine the usefulness of the scale for other populations. Appendix Knee Outcome Survey Activities of Daily Living Scale Instructions: The following questionnaire is designed to determine the symptoms and limitations that you experience because of your knee while you perform your usual daily activities. Please answer each question by checking the statement that best describes you over the last 1 to 2 days. For a given question, more than one of the statements may describe you, but please mark ONLY the statement that best describes you during your usual daily activities. Symptoms 1. To what degree does pain in your knee affect your daily activity level? _5_ I never have pain in my knee. _4_ I have pain in my knee, but it does not affect my daily activity. _3_ Pain affects my activity slightly. _2_ Pain affects my activity moderately. _ 1 _ Pain affects my activity severely. _0_ Pain in my knee prevents me from performing all daily activities.
THE JOURNAL OF BONE AND JOINT SURGERY

D E V E L O P M E N T O F A PATIENT-REPORTED M E A S U R E O F FUNCTION O F T H E K N E E

1143

2. To what degree does grinding or grating of your knee affect your daily activity level? _5_ I never have grinding or grating in my knee. _4_ I have grinding or grating in my knee, but it does not affect my daily activity. _3_ Grinding or grating affects my activity slightly. _2_ Grinding or grating affects my activity moderately. _ 1 _ Grinding or grating affects my activity severely. _0_ Grinding or grating in my knee prevents me from performing all daily activities. 3. To what degree does stiffness in your knee affect your daily activity level? _5_ I never have stiffness in my knee. _4_ I have stiffness in my knee, but it does not affect my daily activity. _3_ Stiffness affects my activity slightly. _2_ Stiffness affects my activity moderately. _ 1 _ Stiffness affects my activity severely. _0_ Stiffness in my knee prevents me from performing all daily activities. 4. To what degree does swelling in your knee affect your daily activity level? _5_ I never have swelling in my knee. _4_ I have swelling in my knee, but it does not affect my daily activity. _3_ Swelling affects my activity slightly. _2_ Swelling affects my activity moderately. _ 1 _ Swelling affects my activity severely. _0_ Swelling in my knee prevents me from performing all daily activities. 5. To what degree does slipping of your knee affect your daily activity level? _5_ I never have slipping of my knee. _4_ I have slipping of my knee, but it does not affect my daily activity. _3_ Slipping affects my activity slightly. _2_ Slipping affects my activity moderately. _ 1 _ Slipping affects my activity severely. _0_ Slipping of my knee prevents me from performing all daily activities. 6. To what degree does buckling of your knee affect your daily activity level? _5_ I never have buckling of my knee. _4_ 1 have buckling of my knee, but it does not affect my daily activity level. _3_ Buckling affects my activity slightly. _2_ Buckling affects my activity moderately. _ 1 _ Buckling affects my activity severely. _0_ Buckling of my knee prevents me from performing all daily activities. 7. To what degree does weakness or lack of strength of your leg affect your daily activity level? _5_ My leg never feels weak. _4_ My leg feels weak, but it does not affect my daily activity.
VOL. 80-A, NO. 8, AUGUST 1998

_3_ _2_ _1_ _0_

Weakness affects my activity slightly. Weakness affects my activity moderately. Weakness affects my activity severely. Weakness of my leg prevents me from performing all daily activities.

Functional Disability with Activities of Daily Living 8. How does your knee affect your ability to walk? _5_ My knee does not affect my ability to walk. _4_ I have pain in my knee when walking, but it does not affect my ability to walk. _3_ My knee prevents me from walking more than 1 mile. _2_ My knee prevents me from walking more than 1/2 mile. _ 1 _ My knee prevents me from walking more than 1 block. _0_ My knee prevents me from walking. 9. Because of your knee, do you walk with crutches or a cane? _3_ I can walk without crutches or a cane. _2_ My knee causes me to walk with 1 crutch or a cane. _ 1 _ My knee causes me to walk with 2 crutches. _0_ Because of my knee, I cannot walk even with crutches. 10. Does your knee cause you to limp when you walk? _2_ I can walk without a limp. _ 1 _ Sometimes my knee causes me to walk with a limp. _0_ Because of my knee, I cannot walk without a limp. 11. How does your knee affect your ability to go up stairs? _5_My knee does not affect my ability to go up stairs. _4_ I have pain in my knee when going up stairs, but it does not limit my ability to go up stairs. _3_ I am able to go up stairs normally, but I need to rely on use of a railing. _2_ I am able to go up stairs one step at a time with use of a railing. _ 1 _ I have to use crutches or a cane to go up stairs. _0_ I cannot go up stairs. 12. How does your knee affect your ability to go down stairs? _5_ My knee does not affect my ability to go down stairs. _4_ I have pain in my knee when going down stairs, but it does not limit my ability to go down stairs. _3_ I am able to go down stairs normally, but I need to rely on use of a railing. _2_ 1 am able to go down stairs one step at time with use of a railing. _ 1 _ I have to use crutches or a cane to go down stairs. _0_ I cannot go down stairs.

1144

J. J. IRRGANG ET AL.

13. How does your knee affect your ability to stand? _5_ My knee does not affect my ability to stand. I can stand for unlimited amounts of time. _4_ I have pain in my knee when standing, but it does not limit my ability to stand. _3_ Because of my knee I cannot stand for more than 1 hour. _2_ Because of my knee I cannot stand for more than 1/2 hour. _ 1 _ Because of my knee 1 cannot stand for more than 10 minutes. _0_ I cannot stand because of my knee. 14. How does your knee affect your ability to kneel on the front of your knee? _5_ My knee does not affect my ability to kneel on the front of my knee. I can kneel for unlimited amounts of time. _4_I have pain when kneeling on the front of my knee, but it does not limit my ability to kneel. _3_ I cannot kneel on the front of my knee for more than 1 hour. _2_ I cannot kneel on the front of my knee for more than 1/2 hour. _ 1 _ I cannot kneel on the front of my knee for more than 10 minutes. _0_ I cannot kneel on the front of my knee. 15. How does your knee affect your ability to squat? _5_ My knee does not affect my ability to squat. I can squat all the way down.

_4_ I have pain when squatting, but I can still squat all the way down. _3_ I cannot squat more than 3/4 of the way down. _2_ 1 cannot squat more than 1/2 of the way down. _ 1 _ I cannot squat more than 1/4 of the way down. _0_ I cannot squat at all. 16. How does your knee affect your ability to sit with your knee bent? _5_ My knee does not affect my ability to sit with my knee bent. I can sit for unlimited amounts of time. _4_ I have pain when sitting with my knee bent, but it does not limit my ability to sit.. _3_ I cannot sit with my knee bent for more than 1 hour. _2_ I cannot sit with my knee bent for more than 1/2 hour. _ 1 _ I cannot sit with my knee bent for more than 10 minutes. _0_ I cannot sit with my knee bent. 17. How does your knee affect your ability to rise from a chair? _5_ My knee does not affect my ability to rise from a chair. _4_ I have pain when rising from the seated position, but it does not affect my ability to rise from the seated position. _2_ Because of my knee I can only rise from a chair if I use my hands and arms to assist. _0_ Because of my knee I cannot rise from a chair.

References
1. Bassiouni, M.; Bassiouni, H.; and el-Fcki, M.: Sensitivity versus specificity of phonoarthrography as an indicator for cartilage degeneration. Clin. Rheumatol., 14:135-142,1995. 2. Bellamy, N., and Buchanan, W. W.: A preliminary evaluation of the dimensionality and clinical importance of pain and disability in osteoarthritis of the hip and knee. Clin. Rheumatol., 5: 231-241,1986. 3. Bellamy, N.; Buchanan, W. W.; Goldsmith, C. H.; Campbell, J.; and Stitt, L. W.: Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. / Rheumatol., 15:1833-1840,1988. 4. Bergner, M.; Bobbitt, R. A.; Carter, W. B.; and Gilson, B. S.: The Sickness Impact Profile: development and final revision of a health status measure. Med. Care, 19:787-805,1981. 5. Cohen, J.: Statistical Power Analysis for the Behavioral Sciences, p. 26. New York, Academy Press, 1977. 6. Deyo, R. A.: Comparative validity of the Sickness Impact Profile and shorter scales for functional assessment in low-back pain. Spine, 11:951-954,1986. 7. Deyo, R. A., and Diehl, A. K.: Measuring physical and psychosocial function in patients with low-back pain. Spine, 8:635-642,1983. 8. Guyatt, G. H.; Feeny, D. H.; and Patrick, D. L.: Measuring health-related quality of life. Ann. Intern. Med., 118:622-629,1993. 9. Hefti, F.; Miiller, W.; Jakob, R. P.; and Staubli, H. U.: Evaluation of knee ligament injuries with the IKDC form. Knee Surg., Sports Traumat., Arthrosc, 1:226-234,1993. 10. Hudak, P. "L.; Amadio, P. C; Bombardier, C; and the Upper Extremity Collaborative Group (UECG): Development of an upper extremity outcome measure: the DASH (disabilities of the arm, shoulder, and hand). Am. J. Indust. Med., 29:602-608,1996. 11. Insall, J. N.; Dorr, L. D.; Scott, R. D.; and Scott, W. N.: Rationale of the Knee Society clinical rating system. Clin. Orthop., 248: 13i4,1989. 12. Kirshner, B., and Guyatt, G.: A methodological framework for assessing health indices. J. Chronic Dis., 38:27-36,1985. 13. McHorney, C. A.; Ware, J. E., Jr.; and Raczek, A.: The MOS 36-Item Short-Form Health Survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med. Care, 31:247-263,1993. 14. McHorney, C. A.; Ware, J. E., Jr.; Lu, J. E; and Sherbourne, C. D.: The MOS 36-Item Short-Form Health Survey (SF-36): III. Tests of data quality, scaling assumptions, and reliability across diverse patient groups. Med. Care, 32:40-66,1994. 15. McSweeny, A. J., and Creer, T. L.: Health-related quality-of-life assessment in medical care. Disease-a-Month, 41:1-71,1995. 16. Martin, D. P.; Engelberg, R.; Agel, J.; Snapp, D.; and Swiontkowski, M. F.: Development of a musculoskeletal extremity health status instrument: the Musculoskeletal Function Assessment Instrument./ Orthop. Res., 14:173-181,1996. 17. Nagi, S. Z.: Disability concepts revisited: implications for prevention. In Disability in America: Toward a National Agenda for Prevention, edited by A. M. Pope and A. R. Tarlov. Appendix A, pp. 309-327. Washington, D.C., National Academy Press, 1991. 18. Nitko, A. J.: Educational Tests and Measurement: An Introduction, p. 396. New York, Harcourt Brace Jovanovich, 1983.

THE JOURNAL OF BONE AND JOINT SURGERY

DEVELOPMENT OF A PATIENT-REPORTED MEASURE OF FUNCTION OF THE KNEE

1145

19. Noyes, F. R.; McGinniss, G. H.; and Mooar, L. A.: Functional disability in the anterior cruciate insufficient knee syndrome. Review of knee rating systems and projected risk factors in determining treatment. Sports Med., 1:278-302,1984. 20. Shapiro, E. T.; Richmond, J. C ; Rockett, S. E.; McGrath, M. M.; and Donaldson, W. R.: The use of a generic, patient-based health assessment (SF-36) for evaluation of patients with anterior cruciate ligament injuries. Am. J. Sports Med., 24:196-200,1996. 21. Sharma, S.: Applied Multivariate Techniques, p. 76. New York, John Wiley, 1996. 22. Shrout, P. E., and Fleiss, J. L.: Intraclass correlations: uses in assessing rater reliability. Psych. Bull, 86:420-428,1979. 23. Stratford, P. W.; Binkley, J. M.; and Riddle, D. L.: Health status measures: strategies and analytic methods for assessing change scores. Phys. Ther., 76:1109-1123,1996. 24. Stratford, P. W.; Binkley, J.; Solomon, P.; Gill, C ; and Finch, E.: Assessing change over time in patients with low back pain. Phys. Ther., 74:528-533,1994. 25. Tegner, Y., and Lysholm, J.: Rating systems in the evaluation of knee ligament injuries. Clin. Orthop., 198:43-49,1985. 26. Ware, J. E., Jr., and Sherbourne, C. D.: The MOS 36-Item Short-Form Health Survey (SF-36) I. Conceptual framework and item selection. Med. Care, 30:473-483,1992.

VOL. 80-A, NO. 8, AUGUST 1998