Authors

  1. Chui, Kevin PT, DPT, PhD, GCS, OCS
  2. Hood, Ethan PT, DPT, MBA, GCS
  3. Klima, Dennis PT, MS, PhD, GCS, NCS

Abstract

Evidence-based practitioners need to consider the sensitivity to change or "responsiveness" of outcome measures such as walking speed. The responsiveness of an outcome measure refers to its ability to accurately detect a change or difference when it has occurred. In this review, we first describe distribution-based and anchor-based methods and the most commonly reported indexes of responsiveness (ie, effect size, standard error of the measurement, minimal detectable change, standardized response mean, and minimal clinically important difference) for walking speed. We then summarize and synthesize the recent literature on the responsiveness of walking speed in different populations of older adults, patients with neurologic conditions (ie, stroke, Parkinson's disease, and Alzheimer's disease), and patients with orthopedic conditions (ie, hip fracture and knee osteoarthritis). In all of the studies cited, walking speed was sensitive to change over time and, when looking across studies, there is considerable agreement that meaningful change in walking speed is approximately 0.1 m/s.

 

Article Content

In addition to considering clinometric properties such as reliability, validity, and feasibility, the evidence-based practitioner must also consider the sensitivity to change or "responsiveness" of an outcome measure.1 The responsiveness of an outcome measure refers to its ability to accurately detect a change or difference when it has occurred, and the importance of calculating and reporting this clinometric property in physical therapy research has been emphasized.2,3 There is, unfortunately, no universally accepted definition of responsiveness or measure of responsiveness. In fact, a recent review by Terwee and colleagues4 found 25 different definitions and 31 different measures of responsiveness. Debate also exists over which of the 2 broad categories of analysis methods, distribution-based or anchor-based, is most appropriate. In recent studies, researchers have triangulated distribution-based methods with anchor-based methods or combined both methods to improve our understanding of meaningful change.3,5 In the next section, we describe distribution-based and anchor-based methods and the most commonly reported indexes of responsiveness for walking speed.

 

DISTRIBUTION-BASED METHODS

Distribution-based methods depend on statistical (eg, variability) and clinometric (eg, reliability) properties of a measure from a sample (or population) to calculate meaningful change scores. An effect size (ES) measures the magnitude of a treatment effect and is calculated by dividing the difference in means at baseline and follow-up by the standard deviation (SD) at baseline [ES = (Meanbaseline - Meanfollow-up)/SDbaseline]. Effect sizes of 0.2, 0.5, and 0.8 are considered small, medium, and large, respectively.6 The standard error of the measurement (SEM) is a measure or response stability and is related to measurement error. The SEM is calculated using the formula (SEM = SDbaseline x [square root]1 - ICC3,1), where ICC3,1 is the test-retest reliability coefficient (note that the more conservative ICC2,1 may also be used). The minimal detectable change (MDC) is the smallest amount of change that is not likely to be due to error in the measurement. The SEM is used to calculate MDC, using the formula (MDC90 = z x SEM x [square root]2), where z is equal to 1.65 when using a 90% confidence interval (CI).7 In several of the studies cited in this review, the SEM was reported by the authors but not the MDC. We, therefore, calculated the MDC90 whenever possible because it seems to be reported most often in the literature.3 The standardized response mean (SRM) is the ratio between the mean change score and the SD (variability) of that change score. The formula for SRM is [(Meandischarge - Meanbaseline)/SD[DELTA]], where SD[DELTA] is the SD of (Meandischarge - Meanbaseline). The values of SRM have often been interpreted in the same way as ES, however, this may lead to over- or underestimation of effect.8

 

ANCHOR-BASED METHODS

Anchor-based methods examine the relationship between outcome measure scores (eg, walking speed) and an independent measure (an independent standard used to ground the meaning of clinical importance), which is the anchor. Receiver operating characteristic (ROC) curve analysis is used to identify which change score (ie, cut score) in the outcome measure of interest (ie, walking speed) results in the best balance of sensitivity and specificity for distinguishing between those who report or demonstrate an important change on the anchor from those who do not.9 Anchors used to establish the minimal clinically important difference (MCID) of walking speed include the modified Rankin Scale,10 the Timed Up and Go Test,11 and the Global Rating of Change Scale.12

 

What follows is a summary and synthesis of recent literature on walking speed in different populations of older adults, patients with neurologic conditions (ie, stroke, Parkinson's disease, and Alzheimer's disease), and patients with orthopedic conditions (ie, hip fracture and knee osteoarthritis). Whenever possible, we attempted to also calculate the MDC90 to better enable the reader to compare and contrast the responsiveness of walking speed from different studies. For each cited reference, the population, the method and distance used to measure walking speed, and the index of responsiveness are summarized in the Table.

  
TABLE 1-a Meaningful... - Click to enlarge in new windowTABLE 1-a Meaningful Change in Walking Speed

OLDER ADULTS

In a study (secondary analysis) by Perera and colleagues,13 the meaningful change of common performance measures was examined for 3 groups of older adults: (1) community-dwelling older adults; (2) individuals with mobility disabilities; and (3) individuals who have sustained a subacute stroke. The data for community-dwelling older adults (n = 492; 43.7% women; mean age of 74.1 +/- 5.7 years; ~18.5% were African Americans) were obtained from a study by Studenski and colleagues,25 who measured walking speed over 4 m with a stopwatch. From this data set, Perera and colleagues then calculated small (an ES of 0.2 = 0.05 m/s and the SEM = 0.06 m/s) and substantial (an ES of 0.5 = 0.12 m/s) meaningful change. Using their SEM value of 0.06 m/s, we calculated the MDC90 of 0.139 m/s, using the formula MDC90 = 1.65 x SEM x [square root]2. Their recommended criteria for meaningful change when measuring the walking speed of community dwellers over 4 m is 0.05 m/s for small meaningful change and 0.10 m/s for substantial meaningful change.

 

The data for those with mobility disabilities (n = 100; 50.0% women; mean age of 77.6 +/- 7.6 years; 34% were African Americans) were obtained from a study by Chandler and colleagues.26 As part of the inclusion criteria, all participants were community dwellers and met a prespecified criterion for frailty. Walking speed was measured over 2 distances, 10 m and 10 ft (as part of part of a mobility skills protocol), with a stopwatch. From this data set, Perera and colleagues then calculated small (an ES of 0.2 = 0.05 m/s and the SEM = 0.06 m/s) and substantial (an ES of 0.5 = 0.13 m/s) meaningful change for the 10-m distance. A similar small (an ES of 0.2 = 0.04 m/s and the SEM = 0.10 m/s) and substantial (an ES of 0.5 = 0.10 m/s) meaningful change was calculated for the 10-ft distance. Using their SEM values, we then calculated the MDC90 of 0.139 m/s and 0.231 m/s for the 10-m and 10-ft distances, respectively. Their recommended criteria for meaningful change when measuring the walking speed of frail community dwellers over a 10-m or 10-ft course is 0.05 m/s for small meaningful change and 0.10 m/s for substantial meaningful change.

 

In another study of community-dwelling older adults (n = 118; 70.3% women; mean age of 84.8 +/- 5.3 years; 2.5% were African American), Chui and Lusardi14 measured walking speed over 3.66 m with the GAITRite (CIR Systems, Inc., Havertown, PA), an instrumented walkway system. Walking speed data were first disaggregated by gender and decade before SEM and MDC (reported as MDD) values were calculated. The values of SEM and MDC varied between decades for both genders as a function of sample size. For all women, the SEM = 0.0345 m/s and MDC90 = 0.0797 m/s and for all men the SEM = 0.0473 m/s and MDC90 = 0.0109 m/s. For all subjects (both genders), the SEM = 0.0298 m/s and MDC90 = 0.0688 m/s.

 

Limited data exist with respect to gait performance in select ethnic groups. Mangione and colleagues15 examined physical performance measures in older adults (n = 52; 87% women; mean age of 78 +/- 8 years; 100% were African American). Their sample consisted of a wide range of functional and cognitive abilities. Walking speed was measured over 3.87 m with the Gait Mat II (E.Q., Inc., Chalfont, PA), another type of instrumented walkway system. The values for SEM and MDC were reported as 0.08 and 0.19 m/s, respectively.

 

Small differences in SEM and MDC values between the studies by Perera and colleagues,13 Chui and Lusardi,14 and Mangione and colleagues15 may be attributed to differences in gender, age, ethnicity, health status, and methods used to measure walking speed. Each of these studies report indexes of responsiveness to assist with clinical interpretation of walking speed for a specific population of older adults. Data findings can also be used to develop goals and determine readiness for discharge for patients intending to return to the community.

 

STROKE

Numerous studies have examined meaningful change in self-selected or comfortable walking speed after stroke. Salbach and colleagues16 examined meaningful change in patients after their first stroke (n = 50; 38.0% women; mean age of 68 +/- 13 years). In this study, walking speed was measured over 2 distances, 5 and 10 m, with a stopwatch. The ES for walking speed over 5- and 10-m distances was reported as 0.83 and 0.74, respectively. In addition, the SRM for walking speed over 5- and 10-m distances was reported as 1.22 (95% CI = 0.93-1.50) and 0.92 (95% CI = 0.64-1.18), respectively. Of the 2 distances, 5 m was more responsive. Therefore, the researchers recommended using the 5-m distance for measuring change in walking ability in the first 5 weeks after stroke.

 

Flansbjer and colleagues17 examined meaningful change in a battery of tests that have a walking component in a group of patients with chronic mild to moderate hemiparesis after stroke (n = 50; 24% women; mean age of 58 +/- 6.4 years). All patients were a minimum of 6 months and maximum of 48 months after stroke. Walking speed was measured over 10 m with a stopwatch, and the SEM was reported as 0.07 m/s. Using their SEM of 0.07 m/s, we then calculated the MDC90 of 0.162 m/s. On the basis of results, the authors recommended using walking speed to evaluate improvements in patients with chronic mild to moderate hemiparesis after stroke.

 

English and colleagues18 examined patients receiving rehabilitation following their first or recurrent stroke with commonly used outcome measures (n = 61; percentage of women not reported; mean age of 65.2 +/- 13.1 years). Walking speed was measured over 5 m with a stopwatch and showed a large ES of 0.81 between admission and discharge measurements. The authors conclude that walking speed is sensitive to change over time in this population.

 

In the same secondary analysis by Perera and colleagues previously discussed, meaningful change was calculated from a study by Duncan and colleagues27 on individuals who have sustained a subacute stroke (n = 100; 44.0% women; mean age of 69.8 +/- 10.3 years) that measured walking speed over a 10-m distance with a stopwatch. From this data set, Perera and colleagues13 then calculated small (an ES of 0.2 = 0.06 m/s and the SEM = 0.04 m/s) and substantial (an ES of 0.5 = 0.14 m/s) meaningful change. Using their SEM of 0.04 m/s, we then calculated the MDC90 of 0.092 m/s. Their recommended criteria for meaningful change when measuring walking speed over 10 m is 0.05 m/s for small meaningful change and 0.10 m/s for substantial meaningful change.

 

Fulk and Echternach19 examined meaningful change in walking speed in patients after stroke (n = 35; percentage of women not reported; mean age 67.4 +/- 13.8 years) who required different levels of assistance to walk during rehabilitation. In this study, walking speed was measured over 5 m with a stopwatch. The results indicated that MDC90 varied as a function of the level of assistance required to walk. Walking speed was more sensitive to change for those who required physical assistance (MDC90 = 0.07 m/s) or an assistive device (MDC90 = 0.18 m/s) than those who could walk without physical assistance (MDC90 = 0.36 m/s). For all patients, regardless of level of assistance required, the MDC90 was 0.30 m/s.

 

In contrast to the previously discussed studies of patients who have sustained a stroke that used distribution-based methods (ie, ES, SEM, MDC, and SRM), Tilson and colleagues10 and Fulk and colleagues12 used an anchor-based approach to calculate the responsiveness of walking speed. In the study by Tilson and colleagues, the walking speed of patients who have sustained a first-time stroke (n = 283; 48.1% women; mean age of 63.5 +/- 12.5 years) was measured over 10 m with a stopwatch.10 Walking speed was measured at 20 and 60 days after stroke, and improvements at 60 days was anchored to the modified Rankin Scale. A change of 1 or more on the modified Rankin Scale, as scored by the physical therapist (assessor), was used to calculate the MCID with an ROC curve. The ROC curve indicated a change in walking speed of 0.16 m/s (ie, the MCID) and resulted in the optimal combination of sensitivity and specificity for detecting improvements in the modified Rankin Scale. Therefore, patients who have sustained a subacute stroke with improvement in walking speed of 0.16 m/s or more are more likely to have less disability (as measured by the modified Rankin Scale) than those who do not.

 

In a similar study, Fulk and colleagues12 measured the walking speed of patients who have sustained a first-time stroke (n = 44; 36.4% women; mean age of 61.8 +/- 14.7 years) over 5.2 m with the GAITRite. Walking speed was measured at admission (mean time from stroke to admission = 55.9 +/- 45.8 days) to and at discharge (mean time from stroke to discharge = 138.6 +/- 74.5 days) from outpatient physical therapy. Improvement at discharge was anchored to a 15-point Global Rating of Change Scale and was rated by the patients and physical therapists. Anchor-based methods using ROC curves indicated that the patients' perception of improvement yielded a better estimate of meaningful change in walking speed than the physical therapists' perception (as measured by the areas under each ROC curve). On the basis of the ROC curve using the patients' perception of improvement, the MCID for walking speed is 0.175 m/s.

 

In all of the previous cited studies on patients who have sustained a stroke, despite the differences in inclusion criteria (eg, first vs recurrent stroke and acute, subacute, and chronic), distances used, level of assistance required, or index or responsiveness reported, the authors concluded that walking speed can be used to measure meaningful change in patients after stroke. In fact, Salbach and colleagues16 concluded that measuring walking speed over 5 m was more sensitive to change than other commonly used measures such as the Barthel Index, the Berg Balance Scale, STREAM, and the Timed Up and Go Test. In contrast, English and colleagues18 reported measuring walking speed over 5 m, which was comparable with that assessed with the Berg Balance Scale, but that both had better sensitivity to change and larger ESs than most items on the Motor Assessment Scale. When measuring walking speed over 10 m, Flansbjer and colleagues17 reported comparable SEM and MDC percentages for walking speed, Timed Up and Go Test, performance, stair-climbing ascend, stair-climbing descend, and the 6-Minute Walk Test. Given that walking speed is comparable, if not better, at measuring meaningful change in patients after stroke than other commonly used outcome measures, we contend that walking speed should be measured for all ambulatory patients after stroke.

 

PARKINSON'S DISEASE

In a study of community-dwelling older adults with Parkinson's disease, Steffen and Seney20 examined the responsiveness of a battery of tests and measures (n = 37; 29.7% women; mean age of 71 +/- 12 years). For the sample, the average Unified Parkinson's Disease Rating Scale score was 33/176, the average Hoehn and Yahr Scale score was 2, and the average disease duration was 14 +/- 6 years. Walking speed was measured over 6 m with a stopwatch.

 

The authors calculated an MDC95 of 0.18 m/s, using the formula MDC95 = z x SD x [square root]2(1 - ICC3,1). We used their baseline SD data (SD = 0.34) and test-retest coefficient (ICC3,1 = 0.96) to calculate an SEM of 0.068 m/s, using the formula SEM = SDbaseline x [square root]1 - ICC3,1. This SEM value was used to calculate an MDC90 of 0.157 m/s. In addition, Steffen and Seney provide MDC values for other ambulation tests, several measures of balance, and quality-of-life and disease severity rating outcome measures to assist with interpreting meaningful change in patients with Parkinson's disease.

 

ALZHEIMER'S DISEASE

Ries and colleagues21 examined the responsiveness of a battery of performance-based outcome measures for patients with Alzheimer's disease (n = 51; 66.7% women; mean age of 80.71 +/- 8.77 years). On the basis of the Functional Assessment Staging Test scale, patients were dichotomized into 2 groups: (1) mild to moderate and (2) moderately severe to severe Alzheimer's disease. In this study, walking speed was measured over 4.57 m with the GAITRite. The SEM values for the mild to moderate, moderately severe to severe, and all patient groups were similar and reported as 0.0607, 0.0548, and 0.0572 m/s, respectively. The MDC90 for all patients was reported as 0.0944 m/s. In addition, Ries and colleagues provide MDC values for the Timed Up and Go Test and the 6-Minute Walk Test to assist with interpreting meaningful change in patients with Alzheimer's disease.

 

HIP FRACTURE

Palombaro and colleagues11 combined data from 3 previous studies of patients after hip fracture (n = 92, ~70.6% were women, mean age of 78.65 +/- 7.5 years). For all 3 studies, walking speed was measured over 3.87 m with the Gait Mat II. All patients were living at home, and the average time since fracture was 9.24 +/- 16.97 months (the median was 6 months since hip fracture). The SEM was calculated as 0.04 m/s. Using a slightly different formula (MDC95 = 1.96 x SEM), the authors calculated an MDC95 value of 0.08 m/s. Using their SEM value, we calculated the MDC90 of 0.092 m/s. The authors also used 2 different methods to estimate the MCID: a clinical expert panel and a statistical (anchor-based) calculation (an ROC curve using a Timed Up and Go change score of >=2.5 seconds as the anchor). The median estimated MCID value from the expert panel was 0.10 m/s and the calculated MCID was 0.10 m/s. This study reported comparable indexes of responsiveness for walking speed after hip fracture using 3 different method.

 

In another study of hip fracture, Hollman and colleagues22 studied this patient population during the acute rehabilitation stage (n = 16; percentage of women not reported; mean age not reported). All patients were older than 65 years and used a rolling walker with weight-bearing restrictions.22 The average time since surgical fixation was 4.7 +/- 2.0 days. The authors calculated an SEM of 0.029 m/s and MDC95 of 0.082 m/s, using the formula MDC95 = z x SD x [square root]2(1 - ICC3,1). We used their SEM value and calculated an MDC90 of 0.067 m/s.

 

Latham and colleagues23 examined the responsiveness of a battery of self-reported and performance-based measures of function on patients after hip fracture (n = 108; 73.2% women; mean age of 78.9 +/- 8.1 years). Walking speed was measured over 4 m with a stopwatch at baseline and 12 weeks after enrollment. Several distribution-based measures of responsiveness were reported (ES = 0.85; SEM = 0.075; MDC90 = 0.17; SRM = 1.04). An anchor-based method was also used in which the Global Assessment of Improvement was the external criterion measure. Receiver operating characteristic curve analyses were used for the patient and physician Global Assessment of Improvement scores, and both scores had acceptable and comparable responsiveness as measured by the area under the curve (~60%-70%; exactly values were not provided). This study also provides useful meaningful change values for 3 self-reported and 3 other performance-based measures of function commonly used for patients after hip fracture.

 

Collectively, the results from these studies are similar despite differences in methodology, distances used to measure walking speed, and stages of rehabilitation and recovery. In each of these studies, walking speed was responsive to change after hip fracture.

 

KNEE OSTEOARTHRITIS

Borjesson and colleagues24 examined meaningful change in patients with moderate knee osteoarthritis (n = 54; 56% women; mean age of 63 +/- 5 years). Walking speeds were measured over 5 m with forceplates before surgery and 1 year after surgery. The ES and SRM for self-selected walking speed was 0.35 and 0.52, respectively. These values were comparable with slow and fast walking speeds; however, slow walking speed was slightly more responsive than self-selected walking speed. Given the energy efficiency and frequency of use associated with self-selected walking speed, the authors suggest that self-selected walking speed should be the primary measure of walking in this patient population.

 

SUMMARY

The indices of responsiveness reported in the studies cited include ES, SEM, MDC, SRM, and MCID, most of which are calculated using distribution-based methods. Whenever possible, we calculated the MDC90 to better enable the reader to compare and contrast the responsiveness of walking speed from different studies. In all of the studies cited, walking speed was sensitive to change over time. This outcomes measure is an important part of the examination that can be used to measure improvement or deterioration in functional status. When looking across studies of different populations and considering differences in age, gender, and methods used to both measure walking speed and calculate responsiveness, there is considerable agreement that meaningful change in walking speed is approximately 0.1 m/s.28 Interestingly, a change in walking speed of 0.1 m/s is predictive of and is associated with survival in older adults.29,30

 

References

 

1. Beattie P. Measurement of health outcomes in the clinical setting: applications to physiotherapy. Physiother Theory Pract. 2001;17:173-185. [Context Link]

 

2. Beaton DE, Bombardier C, Katz JN, et al. Looking for important changes/differences in studies of responsiveness. J Rheumatol. 2001;28:400-405. [Context Link]

 

3. Haley SM, Fragala-Pinkham MA. Interpreting change scores of tests and measures used in physical therapy. Phys Ther. 2006;86:735-743. [Context Link]

 

4. Terwee CB, Dekker FW, Wiersinga WM, et al. On assessing responsiveness on health-related quality of life instruments: guidelines for instrument evaluation. Qual Life Res. 2003;12:349-362. [Context Link]

 

5. Eton DT, Cella D, Yost KJ, et al. A combination of distribution- and anchor-based approaches determined minimally important differences (MIDs) for four endpoints in a breast cancer scale. J Clin Epidemiol. 2004;57:898-910. [Context Link]

 

6. Cohen J. Statistical Power Analysis for the Behavioral Sciences. New York, NY: Academic Press; 1977. [Context Link]

 

7. Stratford P. Getting more from the literature: estimating the standard error of measurement from reliability studies. Physiother Can. 2004;56:27-30. [Context Link]

 

8. Middel B, van Sonderen E. Statistical significant change versus relevant or important change in (quasi) experimental design: some conceptual and methodological problems in estimating magnitude of intervention-related change in health services research. Int J Integr Care. 2002;2(17):1-18. [Context Link]

 

9. Stratford PW, Spadoni G, Kennedy D, et al. Seven points to consider when investigating a measure's ability to detect change. Physiother Can. 2002;54:16-24. [Context Link]

 

10. Tilson JK, Sullivan KJ, Cen SY, et al. Meaningful gait speed improvement during the first 60 days poststroke: minimal clinically important differences. Phys Ther. 2010;90:196-208. [Context Link]

 

11. Palombaro KM, Craik RL, Mangione KK, et al. Determining meaningful changes in gait speed after hip fracture. Phys Ther. 2006;86:809-816. [Context Link]

 

12. Fulk GD, Ludwig M, Dunning K, et al. Estimating clinically important change in gait speed in people with stroke undergoing outpatient rehabilitation. J Neurol Phys Ther. 2011;35:82-89. [Context Link]

 

13. Perera S, Mody SH, Woodman RC, et al. Meaningful change and responsiveness in common physical performance measures in older adults. J Am Geriatr Soc. 2006;54:743-749. [Context Link]

 

14. Chui KK, Lusardi MM. Spatial and temporal parameters of self-selected and fast walking speeds in healthy community-living adults aged 72-98. J Geriatr Phys Ther. 2010;33:173-183. [Context Link]

 

15. Mangione KK, Craik RL, McCormick AA, et al. Detectable changes in physical performance measures in elderly African Americans. Phys Ther. 2010;90:921-927. [Context Link]

 

16. Salbach NM, Mayo NE, Higgins J, et al. Responsiveness and predictability of gait speed and other disability measures in acute stroke. Arch Phys Med Rehabil. 2001;82:1204-1212. [Context Link]

 

17. Flansbjer UB, Holmback AM, Downham D, et al. Reliability of gait performance tests in men and women with hemiparesis after stroke. J Rehabil Med. 2005;37:75-82. [Context Link]

 

18. English CK, Hillier SL, Stiller K, et al. The sensitivity of three commonly used outcome measures to detect change amongst patients receiving inpatient rehabilitation following stroke. Clin Rehabil. 2006;20:52-55. [Context Link]

 

19. Fulk GD, Echternach JL. Test-retest reliability and minimal detectable change of gait speed in individuals undergoing rehabilitation after stroke. J Neurol Phys Ther. 2008;32:8-13. [Context Link]

 

20. Steffen T, Seney M. Test-retest reliability and minimal detectable change on balance and ambulation tests, the 36-Item Short-Form Health Survey, and the Unified Parkinson Disease Rating Scale in people with Parkinsonism. Phys Ther. 2008;88:733-746. [Context Link]

 

21. Ries JD, Echternach JL, Nof L, et al. Test-retest reliability and minimal detectable change scores for the Timed "Up & Go" Test, the Six-Minute Walk Test, and gait speed in people with Alzheimer disease. Phys Ther. 2009;89:569-579. [Context Link]

 

22. Hollman JH, Beckman BA, Brandt RA, et al. Minimum detectable change in gait velocity during acute rehabilitation following hip fracture. J Geriatr Phys Ther. 2008;31:53-56. [Context Link]

 

23. Latham NK, Mehta V, Nguyen AM, et al. Performance-based or self-reported measures of physical function: which should be used in clinical trials of hip fracture patients? Arch Phys Med Rehabil. 2008;89:2146-2155. [Context Link]

 

24. Borjesson M, Weidenhielm L, Elfving B, et al. Tests of walking ability at different speeds in patients with knee osteoarthritis. Physiother Res Int. 2007;12:115-121. [Context Link]

 

25. Studenski S, Perera S, Wallace D, et al. Physical performance measures in the clinical setting. J Am Geriatr Soc. 2003;51:314-322. [Context Link]

 

26. Chandler JM, Duncan PW, Kochersberger G, et al. Is lower extremity strength gain associated with improvement in physical performance and disability in frail, community-dwelling elders? Arch Phys Med Rehabil. 1998;79:24-30. [Context Link]

 

27. Duncan P, Studenski S, Richards L, et al. Randomized clinical trial of therapeutic exercise in subacute stroke. Stroke. 2003;34:2173-2180. [Context Link]

 

28. Fritz S, Lusardi M. White paper. "Walking speed: the sixth vital sign." J Geriatr Phys Ther. 2009;32:2-5. [Context Link]

 

29. Hardy SE, Perera S, Roumani YF, et al. Improvements in usual gait speed predicts better survival in older adults. J Am Geriatr Soc. 2007;55:1727-1734. [Context Link]

 

30. Studenski S, Perera S, Patel K, et al. Gait speed and survival in older adults. JAMA. 2011;305:50-58. [Context Link]

 

gait; responsiveness; walking speed