SEARCH

SEARCH BY CITATION

Keywords:

  • Western Ontario and McMaster Universities Osteoarthritis Index function;
  • Short form;
  • Osteoarthritis;
  • Applicability;
  • Performance

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. ROLE OF THE STUDY SPONSOR
  9. REFERENCES

Objective

A short version of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) function scale has recently been developed to enhance the applicability of the scale in routine practice and clinical research for patients with hip and knee osteoarthritis. The goal of the present study was to validate this short form.

Methods

We conducted a prospective 4-week cohort study of 1,036 outpatients. Performance on the WOMAC function long form (LF) and short form (SF) was compared. Agreement between responses on the 2 forms was examined according to a Bland-Altman plot. Responsiveness to change (by standardized response mean [SRM]), reproducibility (intraclass correlation coefficient [ICC]), and internal consistency (Cronbach's alpha) were computed for both forms. Construct validity was assessed based on functional impairment as measured on a numerical rating scale.

Results

At baseline, 24% of patients who completed the WOMAC LF had missing data for at least 1 item as compared with only 6% of patients who completed the WOMAC SF. The mean WOMAC SF score was greater than the mean WOMAC LF score (mean ± SD difference −4.3 ± 4.8 on a 0–100 scale). SRMs were 0.61 and 0.73, ICCs were 0.76 and 0.68, and Cronbach's alphas were 0.93 and 0.85 for the WOMAC LF and SF, respectively. The 2 forms had comparable correlation with functional impairment.

Conclusion

The WOMAC function short form has a low rate of missing data and is a responsive, reproducible, and valid measure. The mean SF score was 4 points higher than the mean LF score.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. ROLE OF THE STUDY SPONSOR
  9. REFERENCES

The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) function subscale is widely used in clinical trials of hip and knee osteoarthritis (OA) (1). It is a valid, reliable, and responsive measure of function in hip and knee OA (2, 3). This index is self administered and the function subscale includes 17 items addressing the degree of difficulty in accomplishing 17 activities of daily life. The original WOMAC function subscale has been shortened recently to increase its applicability in epidemiologic studies, clinical trials, and daily clinical practice. Most procedures to shorten a scale usually rely on statistical approaches (4). The procedure used to derive the WOMAC short form was based mainly on the perspective of patients and rheumatologists (5). The WOMAC function short form (SF) consists of 8 of the 17 items of the long form (LF) and has psychometric properties equivalent to that of the LF. However, this SF has not been validated in an independent sample. The main objective of this study was to investigate the psychometric performance of the WOMAC SF in a large independent sample of patients with hip and knee OA.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. ROLE OF THE STUDY SPONSOR
  9. REFERENCES

Study population.

We conducted a prospective cohort study of 4 weeks' duration involving outpatients with hip or knee OA as defined by the American College of Rheumatology (6,7) included by 308 private rheumatologists in France. Each rheumatologist had to include 4 patients, 3 with knee OA and 1 with hip OA. To be included in the study, patients had to be receiving treatment with a nonsteroidal antiinflammatory drug (NSAID). Inclusion began with the onset of treatment or with a switch from one NSAID to another. Patients were excluded if they had a prosthesis on the assessed joint or if they had been treated with intraarticular injection in the 4 weeks before the study began. All patients initially visited the rheumatologist in charge of their case, and NSAID therapy was prescribed (the drug and its dose were chosen by the physician). A final visit to the same rheumatologist was scheduled 4 weeks later.

The study protocol was approved by the ethics committee of Cochin Hospital (Paris, France). The study was conducted in compliance with the Good Clinical Practice and the Declaration of Helsinki principles. In accordance with French national law, all patients gave their written informed consent.

Measurements.

At the baseline visit, demographic and disease data (particularly disease duration) were collected. At baseline and at final visit, all patients were asked to assess their pain on movement, global assessment of disease activity, and functional impairment on an 11-point numerical rating scale (NRS), the score ranging from 0 to 10 (high score indicates a high level of symptoms).

At baseline and at final visit, half of the patients (the LF group) completed the original French Canadian version of the WOMAC function subscale (17 items, 5-point Likert scale version where 0 = none, 1 = slight, 2 = moderate, 3 = severe, and 4 = extreme), with the total score ranging from 0 to 68 and a high score indicating a high degree of functional impairment (8). In the LF group at baseline, the WOMAC SF (8 items, total score ranging from 0 to 32) was extracted from the LF. The other half of the patients (the SF group) completed the SF but not the LF (5). Items of the LF and SF are shown in Table 1. At the final visit only, all patients (LF and SF groups) assessed their response to therapy over 4 weeks in terms of functional impairment on a 3-point Likert scale (where 1 = improved function, 2 = no change, and 3 = worsened function).

Table 1. Items of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) function long form (LF) and short form (SF): missing data per item
 Proportion of missing data at baseline, %
WOMAC LF*WOMAC SF
  • *

    2.5% of the patients completed none of the items of the WOMAC LF.

  • 2.3% of the patients completed none of the items of the WOMAC SF.

Descending stairs3.32.7
Ascending stairs3.72.9
Rising from sitting3.53.3
Standing3.9 
Bending to floors4.9 
Walking on flat surface3.53.0
Getting in/out of a car3.73.0
Shopping5.54.0
Putting on sock4.52.9
Rising from bed4.1 
Taking off socks4.1 
Lying in bed4.7 
Getting in/out of bath9.8 
Sitting3.9 
Getting on/off toilet4.12.5
Heavy domestic duties9.6 
Light domestic duties8.4 

Statistical analysis.

Scores on the WOMAC LF, WOMAC SF, and NRS (pain, global assessment, and functional impairment) were linearly transformed to 0–100 scores by multiplying scores by 100 and dividing by the theoretical maximum of each score. Correlations and agreement between the 2 forms were investigated in the LF group, with the scores of the SF extracted from the LF. Correlation between the WOMAC LF and SF scores at baseline was evaluated by Spearman's correlation coefficient. The same method was applied to the change in score between baseline and final visits. Agreement between the WOMAC LF and SF was examined by a Bland-Altman approach (9). The difference between the WOMAC LF and SF scores was plotted against the average of the 2 scores, and limits of agreement involving the mean difference and the SD of the difference were computed and plotted. By definition, if differences are normally distributed, 95% of individual differences are within 2 SDs of the mean difference (i.e., within the limits of agreement). The Bland-Altman plot is useful in searching for any systematic bias, assessing random error, and revealing whether the difference between scores depends on the level of scores.

We computed the internal consistency of the LF in the LF group and the SF in the SF group. Internal consistency was assessed by Cronbach's alpha to examine the degree to which the items in a scale measure the same concept (10). A value ≥0.70 is generally regarded as acceptable (11).

The responsiveness to change and the reproducibility of each tool were assessed in the subsample in which the tool was used (i.e., LF group for the WOMAC LF, SF group for the WOMAC SF). The responsiveness to change measured over 4 weeks was computed using the standardized response mean (SRM) defined as the ratio of the mean change (pretreatment minus posttreatment) to the SD of the change scores. Calculation of 95% confidence intervals (95% CIs) involved use of a bootstrap method. SRM can be considered large (>0.80), moderate (0.5–0.8), or small (<0.5) (12). As recommended (13), the SRM was also calculated for the 3 groups of patients classified according to the direction of change in function (improved function, no change, worsened function), which was considered an external indicator of change. Intraclass correlation coefficients (ICCs) indicating reproducibility were computed in the subgroup of patients who reported no change in function between baseline visit and visit at week 4 (14). This computation relies on the strong assumption that no change in the outcomes is equivalent to no perceived change in functional status of the patient measured between baseline and final visit. However, this method reduces the influence of information recall associated with shorter periods of retest (15). ICC values range from 0 (totally unreproducible) to 1 (perfectly reproducible); an ICC ≥0.75 is considered excellent reproducibility (16).

Convergent validity was assessed by examining the correlation of the WOMAC LF and SF with the measure of functional impairment (on an NRS) by Spearman's correlation coefficient in the LF group. We also assessed the correlation of the WOMAC LF and SF scores with that of other clinical measures of OA severity such as patient global assessment and pain (on an NRS). These variables were expected to correlate less with the WOMAC LF than the WOMAC SF. All statistical analyses involved use of SAS software, version 9.1 (SAS Institute, Cary, NC).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. ROLE OF THE STUDY SPONSOR
  9. REFERENCES

In all, 1,036 patients were enrolled in the study. The validation process was based on 878 patients (84.7%): 661 (75.3%) with knee OA and 217 (24.7%) with hip OA. Characteristics of these 878 patients are described in Table 2. Among the 878 patients, 821 (93.5%) completed the visit at week 4. Excluded patients (n = 158 [15.3%]) did not differ from included patients (Table 2). Reasons for exclusion were missing data for OA location (hip or knee) and incomplete WOMAC data at baseline (i.e., missing data for at least 1 item). The proportion of patients with incomplete WOMAC data at baseline was 23.9% for the LF group (with missing values per item varying between 3.3% and 9.8%) and 6.1% for the SF group (with missing values per item varying between 2.7% and 4.0%) (Table 1).

Table 2. Characteristics of the patients*
 WOMAC LF group (n = 512)WOMAC SF group (n = 524)
Knee OA (n = 291)Hip OA (n = 98)Patients excluded from the analysis (n = 123)Knee OA (n = 370)Hip OA (n = 119)Patients excluded from the analysis (n = 35)
  • *

    Values are the mean ± SD unless otherwise indicated. WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; LF = long form; SF = short form; OA = osteoarthritis; NSAID = nonsteroidal antiinflammatory drug; NRS = numerical rating scale.

  • Patients in the LF group with missing data at baseline for the WOMAC function LF (n = 122) or without arthritis location (n = 1).

  • Patients in the SF group with missing data at baseline for the WOMAC function SF (n = 32) or without arthritis location (n=3).

  • §

    Other than NSAID.

Age, years66.5 ± 10.764.8 ± 12.368.7 ± 10.666.7 ± 11.164.7 ± 11.969.5 ± 10.5
Female sex, no. (%)211 (72.5)63 (64.3)76 (61.8)262 (70.8)72 (60.5)24 (68.6)
Body mass index, kg/m228.1 ± 4.625.9 ± 4.028.0 ± 5.327.9 ± 4.426.4 ± 4.127.8 ± 3.2
Disease duration, years3.8 ± 4.23.1 ± 4.35.6 ± 8.24.3 ± 5.62.5 ± 3.52.4 ± 3.4
Kellgren/Lawrence grade, no. (%)      
 II 64 (22.0)23 (23.7)22 (18.0)105 (28.6)27 (22.9)10 (31.3)
 III141 (48.5)41 (42.3)50 (41.0)139 (37.9)63 (53.4)13 (40.6)
 IV 86 (29.6)33 (34.0)50 (41.0)123 (33.5)28 (23.7) 9 (28.1)
NSAID intake during the previous 4 weeks, no. (%)104 (36.0)37 (37.8)48 (39.3)136 (36.8)45 (37.8)16 (45.7)
Analgesic intake during the previous 4 weeks, no. (%)173 (60.1)54 (55.7)71 (58.7)209 (56.6)69 (59.0)19 (54.3)
Symptomatic slow-acting drug intake during the previous 4 weeks, no. (%)§109 (38.8)32 (33.7)39 (32.8)111 (31.1)37 (33.6) 6 (18.8)
Pain (0–100 NRS)      
 Baseline63.3 ± 17.960.8 ± 20.360.5 ± 19.061.8 ± 18.759.3 ± 18.968.1 ± 20.6
 Week 0 to week 4−21.0 ± 25.8−15.8 ± 20.6−12.7 ± 24.8−22.5 ± 24.5−19.0 ± 24.7−22.3 ± 25.6
Global assessment (0–100 NRS)      
 Baseline62.6 ± 17.360.1 ± 18.760.8 ± 18.659.2 ± 17.657.8 ± 17.761.0 ± 20.4
 Week 0 to week 4−19.3 ± 24.3−15.1 ± 24.1−13.4 ± 22.1−19.1 ± 21.8−16.9 ± 23.7−14.7 ± 27.0
Functional impairment (0–100 NRS)      
 Baseline61.4 ± 18.658.4 ± 21.159.5 ± 18.357.8 ± 19.755.6 ± 19.758.3 ± 25.0
 Week 0 to week 4−18.9 ± 23.0−15.8 ± 22.4−13.0 ± 21.0−17.6 ± 24.1−15.7 ± 24.4−14.5 ± 22.2
WOMAC function LF (0–100)      
 Baseline46.0 ± 16.447.0 ± 16.3
 Week 0 to week 4−10.9 ± 16.8−7.4 ± 15.2
WOMAC function SF (0–100)      
 Baseline50.6 ± 15.550.4 ± 16.249.8 ± 16.648.6 ± 14.4
 Week 0 to week 4−11.2 ± 18.1−8.3 ± 15.6−12.5 ± 16.5−11.3 ± 17.0

Mean ± SD WOMAC LF and SF scores were 46.3 ± 16.4 (median 47.1) and 49.5 ± 16.1 (median 50.0), respectively. The overall distribution of WOMAC LF and SF scores was similar at baseline (Figure 1), and the WOMAC LF and SF scores were highly correlated (r = 0.95, P < 0.0001). The WOMAC LF and SF were also strongly correlated in change in score measured between baseline and week 4 (r = 0.95, P < 0.0001). Agreement between the WOMAC LF and SF evaluated by the Bland-Altman approach is presented in Figure 2. The variability was random and uniform throughout the range of values. The mean WOMAC SF score was greater than the mean WOMAC LF score (mean ± SD difference −4.3 ± 4.8). The limits of agreement showed that 95% of the differences between the 2 scales can be expected to lie between −13.7 and 5.2 points (2 SDs of the mean difference). Cronbach's alpha coefficients at baseline demonstrated high internal consistency for the WOMAC LF and SF (0.93 and 0.85, respectively).

thumbnail image

Figure 1. Distribution of baseline Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores (on a 0–100 numerical rating scale) for the long form and short form.

Download figure to PowerPoint

thumbnail image

Figure 2. Bland-Altman plot of difference between scores on the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) long form (LF) and short form (SF) by mean WOMAC LF and SF (d) in patients who completed the full WOMAC function LF (n = 389). The 3 horizontal lines indicate mean individual difference (d) ± 2 SD. The mean WOMAC SF score was greater than the mean WOMAC LF score (mean difference −4.3). The limits of agreement show that 95% of the differences between the 2 scales can be expected to lie between –13.7 and 5.2 points. The agreement is not dependent on the level of score.

Download figure to PowerPoint

The 4-week responsiveness to change of the LF was lower than that of the SF (SRM 0.61; 95% CI 0.51, 0.71 versus SRM 0.73; 95% CI 0.64, 0.85). When considering the direction of change in function, the SRM was similar for the LF and SF for patients who reported no change or worsened function, but not for patients who reported improved function (Table 3). ICCs were 0.76 (95% CI 0.66, 0.83) for the WOMAC LF and 0.68 (95% CI 0.57, 0.77) for the WOMAC SF. Both the WOMAC SF and the WOMAC LF correlated with each of the following 3 criteria: functional impairment as measured on an NRS (which indicates similar convergent validity) and pain and global assessment as measured on an NRS (correlations with pain and global assessment were less than with functional impairment) (Table 4).

Table 3. Responsiveness to change of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) function long form (LF) and short form (SF) according to the patient's assessment of perceived change in function after 4 weeks*
ScaleImproved functionNo changeWorsened function
SRM95% CISRM95% CISRM95% CI
  • *

    SRM = standardized response mean; 95% CI = 95% confidence interval.

  • Improved function: n = 203, no change: n = 97, worsened function: n = 17.

  • Improved function: n = 295, no change: n = 119, worsened function: n = 13.

WOMAC LF0.920.77, 1.080.200.00, 0.39−0.65−1.36, −0.20
WOMAC SF1.090.95, 1.250.180.00, 0.36−0.63−1.35, −0.15
Table 4. Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) function long form (LF) and short form (SF) Spearman's correlations with scores on a numerical rating scale of functional impairment, pain, and global assessment by patients who completed the full WOMAC function LF (n = 389)
 WOMAC SFWOMAC LFFunctional impairment*Pain*Global assessment*
  • *

    Measured on a 0–10 numerical rating scale.

WOMAC LF1.000.950.640.520.53
WOMAC SF 1.000.650.540.56
Functional impairment*  1.000.720.74
Pain*   1.000.79
Global assessment*    1.00

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. ROLE OF THE STUDY SPONSOR
  9. REFERENCES

In this study, we compared the performance of the complete WOMAC function LF with the WOMAC SF in a large sample of patients with OA (hip and knee) and with a broad range of disease severity. The LF and SF have equivalent properties, except for rate of missing data, which was lower with the WOMAC SF.

Authors of the original version of the WOMAC recommend imputing data for ≥1 missing items by using the mean of the completed items. The major disadvantage of this technique is the underestimation of the variance. Shortening the WOMAC function subscale from 17 to 8 items greatly influences the level of missing data. When considering baseline data, only 6% of the WOMAC SF items had incomplete data, as compared with 24% of the WOMAC LF items. This simple result could strengthen the data quality and appeal of the SF. Although no consensual method of handling missing data exists (17, 18), this concern should not be neglected. When comparing data from the WOMAC SF extracted from the WOMAC LF in the LF group and that from the WOMAC SF in the SF group, the rate of missing data for each item was higher for the former data. This result could mean that supplementary items affect the global data quality.

All of the characteristics of the WOMAC SF were expected to be similar to those of the WOMAC LF. The validity of the WOMAC SF was demonstrated not only by its similar correlation with the WOMAC LF, but also with the Bland-Altman plot. However, we found that the WOMAC SF overestimated the WOMAC LF by 4 points on average (on a 0–100 scale). In randomized controlled trials, because the same instrument (WOMAC LF or SF) is used in all groups, this overestimation will equally affect the 2 groups and will not affect the difference between groups. This concern should be taken into account when performing meta-analyses of pooled data from trials using either the WOMAC LF or SF. Ninety-five percent of the difference between the 2 scales was between –13.7 and 5.2 points (on a 0–100 scale). Our Cronbach's alpha values of the LF and SF are identical to those observed for both forms in the study from which the WOMAC SF was derived (5). The WOMAC SF is an internally consistent instrument for patients. The high coefficient alpha value of the WOMAC LF may reflect redundancy among items. We found sensitivity to change values of the 2 instruments to be less similar than those in the study from which the WOMAC SF was derived. One possible explanation is that in the present study the WOMAC SF was not directly derived from the WOMAC LF. When considering perceived change in function as an external indicator, the SRM was higher for the WOMAC SF in the subgroup of patients who reported improvement. This finding reinforces the usefulness of this scale because what we are looking for in trials is improvement. The reduction of the scale length from 17 to 8 items did not seriously affect the excellent reproducibility of the LF. In our study, we evaluated reproducibility in patients who reported no change in functional impairment at 4 weeks. To provide a more exact measure, reproducibility should be evaluated in an independent test–retest study with a reduced interval of time (e.g., 2 days). Finally, this study provides some evidence of convergent validity of the 2 scales.

A limitation of this study is the use of the data provided by the LF to validate some of the properties of the SF in the LF group. This use of data may constitute a framing bias, and our results could lead to an overestimation of the similarity between the 2 forms (19). This problem is inherent to many validation studies. In the present study, whenever possible, we performed analysis on separate samples (Cronbach's alpha, SRM, ICC).

In conclusion, our study indicates that the WOMAC SF, which was initially developed with participation of a large panel of patients and rheumatologists, is a good alternative to the WOMAC LF for the assessment of function in patients with hip and knee OA. The WOMAC SF retained properties of the parent WOMAC LF (responsiveness, reproducibility, and validity). Because of its simplicity, ease of application, and low rate of missing data, the SF may be of interest in routine practice and in clinical research.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. ROLE OF THE STUDY SPONSOR
  9. REFERENCES

Dr. Baron had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Tubach, Ravaud, Logeart, Dougados.

Acquisition of data. Logeart.

Analysis and interpretation of data. Baron, Tubach, Ravaud, Logeart, Dougados.

Manuscript preparation. Baron, Tubach, Ravaud, Logeart, Dougados.

Statistical analysis. Baron, Tubach, Ravaud, Dougados.

ROLE OF THE STUDY SPONSOR

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. ROLE OF THE STUDY SPONSOR
  9. REFERENCES

Merck Sharp & Dohme-Chibret Laboratories participated in the study design, data collection, agreement to submit the manuscript, and approval of the content of the submitted manuscript. They did not participate in the data analysis or in the writing of the manuscript.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. AUTHOR CONTRIBUTIONS
  8. ROLE OF THE STUDY SPONSOR
  9. REFERENCES
  • 1
    Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. 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. J Rheumatol 1988; 15: 183340.
  • 2
    Bellamy N, Kean WF, Buchanan WW, Gerecz-Simon E, Campbell J. Double blind randomized controlled trial of sodium meclofenamate (Meclomen) and diclofenac sodium (Voltaren): post validation reapplication of the WOMAC Osteoarthritis Index. J Rheumatol 1992; 19: 1539.
  • 3
    McConnell S, Kolopack P, Davis AM. The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC): a review of its utility and measurement properties. Arthritis Rheum 2001; 45: 45361.
  • 4
    Whitehouse SL, Lingard EA, Katz JN, Learmonth ID. Development and testing of a reduced WOMAC function scale. J Bone Joint Surg Br 2003; 85: 70611.
  • 5
    Tubach F, Baron G, Falissard B, Logeart I, Dougados M, Bellamy N, et al. Using patients' and rheumatologists' opinions to specify a short form of the WOMAC function subscale. Ann Rheum Dis 2005; 64: 759.
  • 6
    Altman R, Alarcon G, Appelrouth D, Bloch D, Borenstein D, Brandt K, et al. The American College of Rheumatology criteria for the classification and reporting of osteoarthritis of the hip. Arthritis Rheum 1991; 34: 50514.
  • 7
    Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al. Development of criteria for the classification and reporting of osteoarthritis: classification of osteoarthritis of the knee. Arthritis Rheum 1986; 29: 103949.
  • 8
    Choquette D, Bellamy N, Raynauld JP. A French-Canadian version of the WOMAC osteoarthritis index [abstract]. Arthritis Rheum 1994; 37: S226.
  • 9
    Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1: 30710.
  • 10
    Bland JM, Altman DG. Cronbach's alpha. BMJ 1997; 314: 572.
  • 11
    Nunally JC. Pshychometric theory. New York: McGraw-Hill; 1978.
  • 12
    Husted JA, Cook RJ, Farewell VT, Gladman DD. Methods for assessing responsiveness: a critical review and recommendations [review]. J Clin Epidemiol 2000; 53: 45968.
  • 13
    Fortin PR, Stucki G, Katz JN. Measuring relevant change: an emerging challenge in rheumatologic clinical trials. Arthritis Rheum 1995; 38: 102730.
  • 14
    Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull 1979; 86: 4208.
  • 15
    Streiner DL, Norman GR. Health measurement scales: a practical guide to their development and use. 2nd ed. New York: Oxford Medical Publications; 1995.
  • 16
    Marx RG, Menezes A, Horovitz L, Jones EC, Warren RF. A comparison of two time intervals for test-retest reliability of health status instruments. J Clin Epidemiol 2003; 56: 7305.
  • 17
    Committee for Proprietary Medicinal Products. ICH topic E 9: statistical principles for clinical trials (CPMP/ICH/363/96). URL: http://www.emea.europa.eu/pdfs/human/ich/036396en.pdf.
  • 18
    Committee for Proprietary Medicinal Products. Points to consider on missing data (CPMP/EWP/1776/99). URL: http://www.emea.europa.eu/pdfs/human/ewp/177699EN.pdf.
  • 19
    Haavardsholm EA, Kvien TK, Uhlig T, Smedstad LM, Guillemin F. A comparison of agreement and sensitivity to change between AIMS2 and a short form of AIMS2 (AIMS2-SF) in more than 1,000 rheumatoid arthritis patients. J Rheumatol 2000; 27: 28106.