Abstract
Knee osteoarthritis (OA) is a serious disease and has no cure to date. Knee OA is a leading cause of functional limitation (e.g., difficulty walking). Walking speed is 1 method of quantifying difficulty with walking and should be assessed in clinical practice for adults with knee OA because it has prognostic value and is modifiable. Specifically, slow walking speed is associated with increased risk of adverse health outcomes, including all-cause mortality in adults with knee OA and can be modified by engaging in physical activity or exercise. However, at present, there is little consensus on the distance and instructions used to conduct the walk test. Distance is often selected based on space availability, and instruction varies, from asking the participants to walk at a comfortable pace versus as fast as possible. Therefore, the purpose of this narrative review is to summarize the measurement properties, strengths, and limitations of a fixed-distance walk test ≤ 40 meters in adults with knee OA. Good measurement properties in terms of reliability and validity were observed across the different testing protocols for fixed-distance walk test (i.e., any distance ≤ 40 m and fast- or self-paced). Therefore, clinicians and researchers can select a testing protocol that can safely and consistently be performed over time, as well as provide a practice trial to acclimatize the patients to the fixed-distance walk test.
Knee osteoarthritis (OA) is a leading cause of functional limitation, such as difficulty walking1,2,3,4, and it affects over 250 million people worldwide5. Walking ability serves as an important indicator of overall health in adults with knee OA, given that adults who report difficulty walking have a 51% higher risk for all-cause mortality compared to those with no difficulty6. Walking speed is one method of quantifying difficulty with walking. Slow walking speed is associated with a variety of health outcomes, including disability, prolonged hospitalization, and all-cause mortality in well-functioning older adults7,8,9,10,11 and other patient populations, such as those with stroke and spinal cord injury12,13,14. For this reason, walking speed is considered a “functional” vital sign of overall health10,15. Additionally, engaging in physical activity or exercise improves walking speed and minimizes the risk of developing walking difficulty16,17,18; these are important for the knee OA population given that no cure has been found to date. Therefore, walking speed should be assessed in clinical practice for adults with knee OA because it is modifiable and has prognostic value.
There are different protocols for walk tests, specifically a fixed-duration [6-min walk test (6MWT)], fixed-distance (8-foot, 10-meter, or 20-meter walk test), or walk tests with incremental speed (shuttle walk test). In this review, we will focus only on a fixed-distance walk test ≤ 40 meters because it measures the ability to walk over short distances, an activity commonly restricted in adults with knee OA. Fixed-distance walk tests measure the time needed to ambulate over a specific distance. Participants are instructed to walk either at a comfortable/usual pace [self-paced walk test (SPWT)] or as fast as possible [fast-paced walk test (FPWT)]. Walking speed can then be calculated by dividing the total distance by the total time taken to complete the test. Slower walking speed indicates worse physical function11. Both self-paced and fast-paced walk tests have been used in various patient populations, including those with and without musculoskeletal pain, and those with neurological diseases such as stroke12,13,14.
To facilitate the use of walk tests, understanding its measurement properties is critical for both clinicians and researchers. From a clinical perspective, knowledge about psychometric properties will guide healthcare professionals to screen patients and assess their expected health, as well as describe whether patients have improved or worsened over time. For instance, it will help healthcare professionals to identify whether observed change is meaningful or considered as a measurement error. Comparing scores to normative values will help quantify walking impairments. From a research perspective, understanding the psychometric properties of walk tests will assist researchers in choosing the most reliable and responsive walk tests when selecting outcomes for clinical trials or observational studies.
The Osteoarthritis Research Society International (OARSI) recommends the 40-meter FPWT (40mFPWT) to assess physical function in adults with knee OA19. However, the distances and instructions utilized for conducting walk tests (i.e., participants are told to walk at comfortable pace vs as fast as possible) are variable. Distance is usually selected based on the space available within the clinical and research settings. To better facilitate the use of the walk test in research and clinical practice, there is a need to understand the psychometric properties and predictive elements of walking speed measured using fixed-distance walk tests. A systematic review on walk tests in knee OA was conducted in 201220. Therefore, the purpose of this narrative review is to provide an up-to-date description of measurement properties of fixed-distance walk tests ≤ 40 meters in knee OA and discuss their strengths and limitations. A goal of this review is to encourage clinicians and researchers to make use of walk tests, given that they are quick and easy tests to utilize. Another goal is to provide insights on how to implement walk tests in the clinical and research settings based on the latest evidence. Additionally, an infographic has been created to provide a visual representation for the narrative review (Figure 1).
Methods
Relevant literature for articles investigating the measurement properties of walk tests in adults with knee OA were identified using targeted search in PubMed and Google Scholar. Broad searches were initially performed using “walk test” and “knee osteoarthritis” alone and in combination with several different terms, including “performance-based measures,” “psychometric property,” “reliability,” “validity,” “construct validity,” “predictive validity,” “health outcomes,” and “measurement property.” Only English-language articles indexed in PubMed or Google Scholar through August 2020 were included. Titles, abstracts, and full reports of the identified articles were screened for relevance. The articles were included in this review if they examined at least 1 measurement property of fixed-distance walk test ≤ 40 meters in adults with knee OA.
In this review, we examined the following measurement properties of walk tests of varying distances in adults with knee OA: reliability, validity, measurement error, and responsiveness.
Reliability refers to the consistency of the walk test. Specifically, test-retest reliability examines the reproducibility of the test results across different (interrater) and/or same (intrarater) examiners over multiple sessions, which can be evaluated using intraclass correlation coefficients (ICC)21.
Construct validity examines the relationship of the walk test to another test of physical function or purported constructs, which can be evaluated using a correlation coefficient. Predictive validity provides information regarding the ability of the test to predict future health outcomes, including mortality21.
The standard error of measurement (SEM) is the amount of error that reflects the measurement error of the walk test. SEM is the dispersion around the true value for the walk test. Minimal detectable change (MDC) is the minimum amount of change, beyond measurement error, necessary to ensure that the change was not due to random variability21.
Responsiveness is the ability of the test to identify clinically relevant or meaningful change following an intervention, or over time. This can be determined using test scores anchored to patient-reported or provider-reported thresholds, and interpreted using the minimum clinically important difference (MCID)22.
Quality review of the literature
The methodological quality of the studies included in the review was investigated using the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) checklist23. This appraisal was performed only for studies that investigated the following measurement properties: reliability, construct validity, measurement error, and responsiveness. We acknowledged that the COSMIN checklist was originally developed to investigate the methodological quality of each measurement property for patient-reported outcome measures. However, since its inception, the checklist has been used to evaluate methodological quality for performance-based measures including walk tests20,24. We used the checklist to evaluate reliability, construct validity, measurement error, and responsiveness when reported by the included studies. Each measurement property contains items related to study design and statistical methods that can be used to assess whether a study on a specific measurement property meets the standard for good methodological quality. Each item is rated as excellent, good, fair, or poor. We utilized a similar scoring approach that was taken by Dobson, et al20. Specifically, in agreement with the COSMIN developers, the authors chose to use the “second worst score counts” method, as several of the studies would be appraised as “poor” based on small sample size. We reported the scoring as “+” for positive, “?” for indeterminate, and “–” for poor ratings. The detailed methodology for determining positive, indeterminate, and poor rating for each psychometric properties of the walk tests has been published in a previous study by Dobson, et al20. “NR” (not reported) was used if the specific measurement property was not reported or investigated in the study.
Results
An electronic broad search yielded 1012 abstracts, which were reviewed (by titles and abstracts) for relevance according to our inclusion criteria and 972 articles were subsequently excluded. This left a total of 40 articles for full review. After full review, including manual search of reference lists, 22 articles were subsequently excluded, given that they used fixed-time walk tests (e.g., 6MWT) or other performance-based measures (e.g., sit-to-stand test), or did not include the knee OA population). Thus, a total of 18 articles were included in this review, of which 12 articles described reliability, construct validity, and/or measurement error, and 6 articles described predictive validity of the fixed-distance walk test up to 40 meters in adults with, or at risk of, knee OA. The majority of the included studies only assessed 2 or fewer of the psychometric properties (17/18, 94.4%), while Tolk, et al25 was the only study to assess 4 properties (reliability, construct validity, measurement error, and responsiveness). Gill, et al assessed 3 properties (reliability, construct validity, and measurement error) from the same sample (i.e., adults awaiting for hip/knee replacement as well), but these were reported in 2 separate articles26,27.
Reliability. To determine the reliability of the walk test, 1 study used ICC1,1 (1-way random single measures)26, 3 studies used ICC2,1 (2-way random single measures)28,29,30, 1 study used ICC3,1 (2-way mixed single measures)31, 3 studies used ICC (no information on type available)25,32,33, and 1 study used Spearman correlation34. Good reliability has been seen in the 50-foot, 8-meter, 13-meter, 20-meter, and 40-meter walk tests. Specifically, the test-retest and/or intra-/interrater reliability for the 50-foot, 8-meter, and 40-meter FPWT were excellent, with ICC being > 0.925,26,29,30,31. Two studies have shown good test-retest reliability for the 13-meter SPWT and 20-meter SPWT (20mSPWT), ICC > 0.9, or Spearman rho between the sessions being > 0.932,34. A few studies reported improvement in test-retest reliability when the first trial was excluded from the analysis26,29,34 (Table 1). Fransen, et al29 investigated reliability for both SPWT and FPWT in the study. In this study, 1 test was conducted prior to each session where participants were instructed to walk at an unspecified pace. Regardless of whether it was fast-paced or self-paced, the reliability of a fixed-distance walk test ≤ 40 meters was excellent in adults with varying severity of knee OA (mild OA to those waiting for knee replacement surgery)26,28,34. Of the 9 studies that assessed reliability, 5 studies25,26,28,31,35 had participants who used assistive devices/walking aids. However, the effects of the use of an assistive device on reliability were not fully explored.
Construct validity. Regardless of test distances, FPWT have shown good construct validity with other known measures of physical function [i.e., 30-second chair stand test, 36-item Short Form Health Survey (SF-36) physical function scale, quadriceps strength, SF-36 physical component summary scale, Western Ontario and McMaster Universities Arthritis Index (WOMAC) function, and Lower Extremity Functional Scale27,32,35,36; Table 1]. One study by Tolk, et al25 found that the 40mFPWT had limited construct validity (Spearman rho of 40mFPWT with Knee injury and Osteoarthritis Outcome Score-Physical Function Short Form < 0.3). However, the same study found a moderate to strong correlation between the 40mFPWT and quadriceps strength (Spearman rho > 0.6)25. The evidence for SPWT was limited, although a study conducted by Bacon, et al37 showed that quadriceps strength and 20mSPWT may have nonlinear relation in adults with symptomatic knee OA.
Predictive validity. Slow walking speed can predict incident symptomatic and radiographic knee OA in community-dwelling participants38. Further, slow walking speed was associated with increased mortality risk39 irrespective of the history of decline over the past 1 year in adults with knee OA40, and indicates physical inability to engage in physical activity (i.e., walking fewer steps per day41; Table 2). However, evidence on predictive validity for FPWT is limited.
Walking slower than 1.2 m/s on a 20mSPWT and walking slower than 0.5 m/s on an 8-foot SPWT was predictive of all-cause mortality in adults with radiographic knee OA39. The risk of developing radiographic knee OA increases by 8% for every 0.1 m/s decline in walking speed over 1 year after accounting for age, knee injury, BMI, and Physical Activity Scale for the Elderly score42. Decline in walking speed over 1 year was associated with a 104% increase in risk of knee replacement in the following year compared to those without any change in walking speed43.
Measurement error. SEM and/or MDC with a 90% CI (MDC90) were reported for both SPWT and FPWT in adults with knee OA. Table 3 shows the values for the SEM and MDC90 for walk tests that were found in adults with knee OA. Specifically, the SEM and MDC90 for the 40mSPWT was 0.14 m/s and 0.32 m/s, respectively, in adults with endstage hip and knee OA28. SEM and MDC90 for the 10-meter FPWT were 0.10 m/s and 0.28 m/s, respectively, in adults with hip or knee OA or following joint replacement30.
Responsiveness. Tolk, et al25 reported that the 40mFPWT was responsive, given that at least 75% of the hypothesis on the correlation between the change scores in the walk test and anchor question were confirmed. However, we did not find the evidence for responsiveness for SPWT in adults with knee OA.
Appraisal of methodological quality of studies included in this review. We used the COSMIN checklist23 to evaluate the methodological quality of the included articles and determined that all studies met either an indeterminate or positive rating if the property was assessed. Table 4 displays the rating for each specific property reported by the articles. Twelve individual articles were appraised for methodological quality for the following measurement properties: reliability, construct validity, measurement error, and responsiveness only. For reliability, we rated 8 articles as positive (66.7%), 1 as indeterminate (8.3%), and 3 could not be rated as they did not include the appropriate metric (25%). For construct validity, 4 articles were appraised as positive (36.4%) and 1 article was appraised as indeterminate (9.1%), while the remaining 7 articles did not report on construct validity (54.5%). For measurement error, 6 articles were rated as positive (50.0%) and 6 articles did not receive a rating (50.0%). Only 1 article, which was rated positive (100%), assessed responsiveness in adults with knee OA.
Discussion
In this narrative review, we found that there were inconsistencies in the testing protocol for fixed-distance walk tests up to 40 meters. However, regardless of the variability in distance and instructions used to conduct the test, they were reliable in the knee OA population. FPWT have good construct validity and were responsive, whereas SPWT have good predictive validity. Specifically, slow walking speed (measured using self-paced fixed-distance walk tests) was associated with an increased risk of mortality in knee OA and increased the risk of developing radiographic and symptomatic knee OA. Good measurement properties, in terms of reliability and validity, were observed across the different testing protocols. Hence, clinicians and researchers can likely select a testing protocol that can safely and consistently be performed in a clinical or research setting, and be assured that it will likely perform well. We have summarized methods, interpretation, pros/cons, and the application of fixed-distance walk tests in an infographic (Figure 1).
We found that the reliability of the walk test was good in adults with varying severity of knee OA26,28,34, regardless of the distance and instructions used to conduct the test. However, the reliability may be affected by the number of trials. For both SPWT and FPWT, reliability improved when the first trial was removed26,29,34. The first trial may serve as a practice trial and may facilitate the adults with knee OA to get acclimated to the test; this may explain the improvement in the reliability. Therefore, it is recommended to administer practice trials in clinical and research settings to obtain reliable values for both SPWT and FPWT.
Walk tests are relatively easy for clinicians and researchers to administer and can be conducted in most clinical/research settings. The equipment needed includes a marked walkway, stopwatch, and tape measure. Further, scoring on most tests allows the use of assistive devices by patients if needed. However, it is important to note that the protocol used to conduct the walk test and information regarding the scoring needs to be documented and kept consistent for repeated measurements. The respondent and administrative burden is minimal, given that tests can be completed in less than 5 minutes, and no specialized training is needed to conduct the test. Further, they can be administered and adapted to any language. Normative comfortable walking speed values based on age range from 20–79 years are published44. A recent study showed reference values (by sex, age, Kellgren–Lawrence grade, or BMI) for walking speed measured using the 20-minute walk test45. Given walk tests are a performance-based measure, they are not subject to the same limitations as using patient-reported physical function measures, such as recall bias46. Though the evidence on the MCID for the walk test is not established in the knee OA population, previous studies in older adults have shown that the SPWT has the ability to detect clinically meaningful change. Specifically, Kwon, et al47 found the change of 0.08 m/s on the 4-meter SPWT as clinically meaningful change in sedentary adults aged 70–89 years.
Strengths and limitations. The major strengths of walk tests are that they are valid, have good test-retest reliability in adults with knee OA, and have good predictive validity for health outcomes in older adults. This test has shown to predict health outcomes, including the ability to be physically active, and mortality in adults with knee OA. Further, it is easy to administer and interpret, requires little equipment and/or training, and is thus inexpensive. However, there are several limitations. First, there is little consensus regarding the distance to be used for the test and whether participants were instructed to turn around while testing. This, in turn, can affect the acceleration and deceleration phases needed to complete the walk test. Prior studies caution against using walking speed values interchangeably for short- versus standard-distance walk tests (i.e., 4-m vs 20-m or 10-m)48,49,50. Thus, we caution against generalizing the evidence regarding walking speed measured using different test distances and instructions. Second, prior studies have shown that age, race, psychological factors (e.g., depression), and disease severity are associated with slower walking speed51,52,53,54. Therefore, patient-specific factors should be accounted for when interpreting walking speed values. Third, there are limited construct validity studies for SPWT; however, the predictive validity of the SPWT is strong. On the contrary, construct validity studies for the FPWT are extensive, but there is limited evidence on predictive validity in the knee OA population. Therefore, further research is needed to investigate the psychometrics for both SPWT and FPWT. Last, the focus of this review was on a fixed-distance walk test ≤ 40 meters. Therefore, future research or review on measurement properties on fixed-duration walk tests (e.g., 6MWT) or fixed-distance walk test > 40 meters in knee OA population is needed.
Summary
Regardless of test distances, walk tests are recommended to objectively assess walking difficulty in adults with varying degrees of knee OA (mild to endstage knee OA) for clinical and research purposes. However, there is a need to highlight the distance used to measure walking speed as well as whether the walk test was conducted at a comfortable or fast pace in research studies so the thresholds indicative of poor health outcomes can be applied appropriately. Healthcare providers should stick with the testing protocol (best suited as per the space availability) and use it over time to ensure the walk test’s reliability and ability to interpret change. Further, when possible, a practice trial may be considered to acclimatize the patients to the fixed-distance walk test.
Footnotes
The authors have no disclosures or conflicts of interest.
Full Release Article. For details see Reprints and Permissions at jrheum.org
RSH is supported by a National Health & Medical Research Council Senior Research Fellowship (#1154217).
- Accepted for publication October 2, 2020.
- Copyright © 2021 by the Journal of Rheumatology
Free online via JRheum Full Release option