Original Article
Performance measures were necessary to obtain a complete picture of osteoarthritic patients

https://doi.org/10.1016/j.jclinepi.2005.07.012Get rights and content

Abstract

Objectives

Self-report questionnaires and performance measures represent two methods for assessing physical function. A recurring theme is that self-report measures are superior to performance measures. This study investigated the association between three performance test outcomes of four activities (pain, exertion, and time or distance; for self-paced walk, stair test, timed up-and-go, 6-minute walk) with self-reports of physical function (WOMAC Physical Function subscale and LEFS) and the association between the change scores of the performance tests and those of the self-report measures.

Study Design and Setting

Performance and self-report measures were administered three times (presurgery and at ∼1 week and ∼8 weeks post arthroplasty) to 85 patients who underwent total hip or knee arthroplasty. Components of the performance tests were pooled within each domain across the four measures. Multiple regression analyses were applied. Independent variables were performance tests components; dependent variables were self-report measures. Standardized regression coefficients described the cross-sectional and longitudinal associations.

Results

Pain was the principal determinant of WOMAC Physical Function subscale scores. Pain, exertion, and time or distance were strongly associated with the LEFS at the first, second, and third assessments, respectively. Change in pain was most strongly associated with change in self-reported physical function.

Conclusion

Our findings caution against the isolated use of self-report assessments of physical function.

Introduction

Self-report questionnaires and performance measures represent two methods for assessing the physical function of patients with osteoarthritis of the hip or knee, and of patients who undergo arthroplasty. A recurring theme is that self-report measures are to be preferred over performance measures [1], [2], [3]. Here we examine the rationale for and validity of this preference.

The goal of many interventions for patients with osteoarthritis of the hip or knee and those undergoing arthroplasty is to decrease pain and improve physical function [1]. Although the meaning of pain is self-evident, the interpretation of physical function is less clear. For example, without providing operational definitions or distinguishing between methods of assessment, the OMERACT III group specified that the assessment of physical function is essential for Phase III clinical trials, but the evaluation of performance is optional [1]. Many investigators, however, use performance measures as a patient-centered assessment of physical function [3], [4], [5], [6], [7], [8]. Adding to the confusion is that investigators do not typically clarify the intended meaning of the term difficulty, which appears on many self-report measures, as applied to physical function [9], [10], [11], [12]. A notable exception is the WOMAC Physical Function subscale [13], which provides the following statement: “by this [difficulty with physical function] we mean your ability to move around and to look after yourself.” We suspect that this statement is consistent with the interpretation afforded the term physical function by many health care professionals; however, in the absence of operational definitions or a common citation, it is difficult to know the extent to which clinicians, researchers, and measure developers agree on the term's intended meaning.

Several arguments have been advanced in support of the position that the application of self-report measures of physical function is mandatory and performance measures are optional. Correlations between self-report questionnaires and performance measures on the order of 0.4 to 0.6 are typical and used in support of the notion that the measures are assessing a common attribute [4], [14], [15]. In addition, studies comparing the ability of the two assessment methods to detect change cite larger responsiveness coefficients—most often the standardized response mean or effect size—in favor of self-report measures [2], [7], [16]. Such findings have led investigators to conclude that little is to be gained by supplementing efficient self-report measures with potentially time-consuming and resource-intensive performance measures [2].

Not everyone agrees with the interpretation of the data used to support the questionnaire-alone advocates. For example, a correlation of .6 indicates that one measure accounts for 36% of the variance in the other measure. Others have shown that correlations as high as .90 may not be sufficient for one measure to act as a surrogate for another, particularly if decisions concerning individuals are of interest [17]. A second argument challenges the interpretation of data from responsiveness studies. Typically, responsiveness studies examine change over two times, and a standardized index is used to compare measures. For this approach to make sense, one must believe that the measures assess the same attribute and nothing else. To the extent that different or compound attributes are being assessed, one would expect different change trajectories. In such cases, the size of the responsiveness coefficients is dependent on where the study's sampling interval lies with respect to the clinical course of the attributes being assessed. Studies examining more than one change interval have reported different measures being more responsive over different intervals [6], [8], [18].

A growing body of published data provides dramatic evidence that the time course of change for self-report and performance measures of physical function in patients post arthroplasty is dissimilar [6], [8], [19]. A striking example is found in the work of Parent and Moffet [6]. These investigators studied 65 patients with osteoarthritis of the knee post arthroplasty. The WOMAC, SF-36, 6-minute walk distance, timed stair test, and gait speed over 10 m were assessed preoperatively and at 2 and 4 months post arthroplasty. The WOMAC Physical Function subscale and SF-36 Physical Function domain showed significant improvement at the 2-month postoperative assessment compared to preoperative values; gait speed and the stair test showed values comparable to the preoperative values; and the 6-minute walk distance was significantly less than the preoperative value. This phenomenon is not restricted to the work of Parent and Moffet [6], [8], [19]. Thus, depending on the outcome measure used, a clinician, researcher, or health care policy maker could declare patients better or worse than their preoperative state. That such a discrepancy exists argues against the isolated application of self-report measures.

What could account for such differences? Previous work has examined the factorial complexity of the Lower Extremity Functional Scale (LEFS), a 20-item region specific self-report measure of physical function, with respect to three performance measures [20]. The subjects were 93 patients with osteoarthritis of the hip or knee awaiting arthroplasty. The performance measures were the timed up-and-go test, a 40-m self-paced walk test, and a timed stair test [21]. Immediately following each performance test, patients reported the pain associated with the test on a 10-cm visual analog scale and their perceived exertion on the modified Borg scale [22]. A factor analysis with varimax and oblique rotations yielded three factors: time, pain, and exertion. The LEFS showed factorial complexity with similar loadings across the factors (time: .35; pain: .44; exertion: .41). These findings prompt the hypothesis that self-report measures of physical function are influenced by more than a patient's ability to move around. It seems that patients are not simply reporting their ability to move around, but that their response also includes what they are experiencing.

Our objectives were to investigate the following on a sample of patients undergoing total hip (THA) or knee arthroplasty (TKA): (i) the association between three performance specific test outcomes (pain, exertion, time [distance] assessed for the self-paced walk, stair test, timed up-and-go, and 6-minute walk) with self-reports of physical function (WOMAC Physical Function subscale and LEFS); (ii) the stability of the association between the three components of the performance measures and two self-report measures at three times (presurgery, within 16 days of surgery, and at least 20 days following the first postoperative assessment); and (iii) the association between change in the three components of the performance tests and change in the self-report measures.

Most performance measures use time or distance to quantify physical function [4], [5], [6], [14], [15], [16], [23]. Accordingly, we expected to find the following: (i) the association between the self-report measures and the time or distance component of the performance measures would be the strongest; (ii) the association between the self-report measures and the components of the performance measures would be reasonably stable over time; and (iii) the association between change in the self-report measures and change in the time or distance component of the performance measures would be the strongest.

Section snippets

Sample

The study sample consisted of 85 consecutive patients with osteoarthritis of the hip (n = 43) or knee (n = 42) who underwent THA or TKA and fulfilled the remaining eligibility criteria. Because no information was available on which to base sample size, our choice of 85 subjects was arbitrary. Of the 36 female patients, 19 received TKA. Median age was 65 years (1st, 3rd quartiles: 55, 71) and median body mass index was 28.3 kg/m2 (1st, 3rd quartiles: 26.2, 33.4). Patients were eligible for this

Results

Table 1, Table 2 summarize the scores and change scores. All patients completed the measures at the preoperative and second postoperative assessments; however, fewer patients were able to complete the measures at the first postoperative assessment (sample sizes are given in the tables). In Table 2, positive change values represent improvement (e.g., a decrease in pain, exertion, and time, and an increase in distance), negative change values depict deterioration, and CIs that include zero

Discussion

Our goals were to describe the association between pain, exertion and time or distance of performance measures, with two self-reports measures of physical function, and to determine whether the association was consistent across three time points. We had expected (see Introduction) that the association between self-report measures and the time or distance component of the performance measures would be strongest, and that the association between self-report and performance component measures

Conclusion

We examined the association between performance rated assessments of pain, exertion, and time or distance with self-reported ratings of physical function. Compared to the preoperative assessment values, the ∼2-month post arthroplasty assessment showed a significant improvement in self-reported physical function and a significant deterioration in the time (or distance) to complete the performance task. Change in pain rather than time (or distance) was found to be the principal determinant of

References (32)

  • J.M. Binkley et al.

    The North American Orthopaedic Rehabilitation Research Network. The Lower Extremity Functional Scale (LEFS): scale development, measurement, properties, and clinical application

    Phys Ther

    (1999)
  • J.A. Kopec et al.

    The Quebec Back Pain Disability Scale: measurement properties

    Spine

    (1995)
  • K.E. Roach et al.

    Development of a shoulder pain and disability index

    Arthritis Care Res

    (1991)
  • S. Solway et al.

    The DASH outcome measure user's manual

    (2002)
  • N. Bellamy et al.

    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)
  • Y.C. Lin et al.

    Tests for physical function of the elderly with knee and hip osteoarthritis

    Scand J Med Sci Sports

    (2001)
  • Cited by (241)

    View all citing articles on Scopus
    View full text