Performance-based tests in subjects with stroke: outcome scores, reliability and measurement errors

Transcription

1 Article Performance-based tests in subjects with stroke: outcome scores, reliability and measurement errors Clinical Rehabilitation 0(0) 1 10! The Author(s) 2011 Reprints and permissions: sagepub.co.uk/journalspermissions.nav DOI: / cre.sagepub.com Christina DCM Faria, Luci F Teixeira-Salmela, Mansueto Gomes Neto and Fátima Rodrigues-de-Paula Abstract Objectives: To assess the intra- and inter-rater reliabilities and measurement errors of seven widely applied performance-based tests for stroke subjects (comfortable/maximal gait speeds and both stair ascending/descending cadences, as well as the Timed Up and Go test) and to verify whether the use of different types of outcome scores (one trial, the means of two and three trials, and the best and the worst values of the three trials) affected the score values, as well as their reliability and measurement errors. Design: Intra- and inter-rater reliability study. Setting: Research laboratory. Subjects: Sixteen stroke subjects with a mean age of years. Main measures: Seven performance-based tests, over two sessions, seven days apart, evaluated by two independent examiners. A third examiner recorded all data. One-way ANOVAs, intra-class correlation coefficients (ICCs) and percentages of the standard errors of measurement (SEM%) were used for analyses. Results: For all tests, similar results were found for all types of outcome scores (0.01 F 0.56; 0.34 p 0.99). For instance, at the comfortable gait speed, the means (SD) values for the first trial, the means of two and three trials and the best and worst of three trials were, respectively, 1.04 (0.25), 1.04 (0.24), 1.05 (0.24), 1.10 (0.26), 1.02 (0.24) seconds. Significant and adequate values of intra- (0.75 ICC 0.96; p 0.002) and inter-rater (0.75 ICC 0.97; p 0.001) reliabilities were found for all tests and outcome scores. Measurement errors were considered low (5.01 SEM% 14.78) and were also similar between all outcome scores. Conclusions: For the seven tests, only one trial was necessary to provide consistent and reliable results regarding the functional performances of stroke subjects. Department of Physical Therapy, Universidade Federal de Minas Gerais, Belo Horizonte, Minas Gerais, Brazil Corresponding author: Professor Luci Fuscaldi Teixeira-Salmela, Department of Physical Therapy, Universidade Federal de Minas Gerais, Avenida Antônio Carlos, 6627, Campus Pampulha, , Belo Horizonte, MG, Brazil lfts@ufmg.br

2 2 Clinical Rehabilitation 0(0) Keywords Stroke, rehabilitation, reliability, outcome measures, mobility Received: 15 June 2011; accepted: 27 August 2011 Introduction Performance-based tests to assess mobility of subjects with stroke are widely applied, since reduced mobility is one of their most common disabilities. 1,2 Among all available tests to assess mobility in stroke subjects, maximal and comfortable gait speeds and stair ascending and descending cadences, as well as the Timed Up and Go are most often applied, given their objectivity, feasibility and adequacy values of validity, intra-rater reliability, and responsiveness. 3 6 In addition, these tests provide information regarding walking performance, an activity which individuals with stroke rate as being the most important 7 and which is the best contributor to functional disabilities post-stroke. 8 An important question related to those tests, which has not been completely investigated is regarding the number of trials necessary to obtain reliable and consistent measures and clinicians and researchers could use different outcome scores for their purposes. 4,5,9 17 Tyson 18 compared the means of two and three trials for gait speed measures with stroke subjects and demonstrated that the average of two trials provided the most accurate measure. However, no study was found regarding the most adequate outcome scores for the other above cited tests, nor for other types of outcome scores. In performance-based tests, human observers are part of the measurement system. Therefore, the data obtained from this source cannot be interpreted with confidence unless those who collect and record the data are reliable. Thus, the intra- and inter-rater reliabilities should be investigated for all performance-based tests. 10 Despite the fact that information regarding the intrarater reliability was previously provided, 4,14 19 no studies were found regarding the inter-rater reliability of those tests with stroke subjects. Studies related to measurement properties, besides investigating reproducibility, should also determine the minimal changes which indicate real improvements. 4,20,21 Information regarding the magnitudes of the differences required to detect real changes for stroke subjects was found only for gait speed measures. 4,18,19,22,23 Therefore, the aims of the present study were: (1) to compare the performance of stroke subjects during maximal and comfortable gait speeds and for both stair ascending and descending cadences, as well as the Timed Up and Go, considering different sources of outcome scores (one trial, the means of two and three trials, and the best and the worst scores of the three trials); (2) to assess their intra- and inter-rater reliabilities, considering the different sources of outcome scores; and (3) to determine their measurement errors and, therefore, the minimal changes which indicate real improvements for subjects with stroke. Methods Subjects Subjects with stroke were recruited from the general community by screening outpatient clinics in university hospitals, based upon the following criteria: stroke diagnosed by a physician, be over 20 years of age, had no receptive aphasia, and demonstrated the ability to perform all of the tests with or without assistive devices. According to Ng et al. 17 a sample size of 10 subjects would be needed if two observations per subject achieved 80% of power to detect intra-class correlation coefficients (ICCs) of

3 Faria et al with a significance level of No significant effects between groups were found for the walking speed values assessed using their preferred assistive devices. 24 Therefore, the subjects were allowed to use their assistive devices or orthoses, if needed. 4 Before data collection, eligible participants were informed about the objectives of the study and provided consent, based upon approval from the university ethical review board. Demographic and clinical data were collected by trained researchers in the university laboratory, shortly after recruitment. Performance-based tests All subjects performed the following seven widely applied performance-based tests: Maximal and comfortable gait speeds, 5,25 29 maximal and comfortable stair ascent and descent cadences, 4 and the Timed Up and Go. 3,30,31 Gait speed was assessed while the subjects walked at both their maximal and most comfortable speeds along a 14 m hallway wearing their normal shoes. The time to cover the central 10 m was recorded with a digital stopwatch and the speed (m/s) was calculated. The instructions to walk at maximum speeds were standardized as follows: When I say go, walk safely in a straight line, as fast as you can. For the comfortable speeds, the commands were: When I say go, walk safely in a straight line, as you usually do, in a comfortable pace. 5 For the stair ascent, subjects were asked to climb a set of 11 stairs of standardized dimensions (110 cm wide, 15 cm high and 30 cm deep) with rails on both sides. They were instructed to climb the stairs preferably in a step-through pattern. Before starting the test, the subjects decided whether or not to use the handrails and nine used them. The subjects climbed one flight of stairs and after a 1-minute rest interval, they were asked to descend with the same commands. The time from the moment the first foot left the ground until the second foot touched the ground on the last step was recorded for both for ascending and descending and the cadences (steps/s) were calculated. For the Timed Up and Go test, subjects sat on a chair (depth of 45 cm, width of 49 cm and armrest height of 20 cm), 4 whose height was adjusted to 100% of their leg length, determined as the distance from the lateral femoral condyle to the ground, 32 and the backrest was adjusted to a trunk position at approximately 90 degrees to horizontal, 33 since these variables might alter the test performances. 32,33 Subjects were instructed to sit comfortably with their backs against the chair, and on the word go, stand up, walk at their self-selected comfortable speed for three meters, turn around, return, and sit down. To prevent risks of falling, the examiner followed the subjects a half-step behind, so as not to set their walking pace. 4 The time, in seconds, was recorded. Procedures The subjects were evaluated three times over two visit sessions, seven days apart, at the same time of the day and in the same research environment. The first and second assessments were carried out during the first visit and the third during the second visit. Three previously trained physical therapists, who had at least two years of experience with the test applications with stroke subjects, collected the data. The first examiner assessed all tests twice over the two visits, whereas, the second rated the tests once during the first visit. The third physical therapist was responsible for reading and recording all data, so both raters were blinded regarding the test results. No feedback nor further discussions were allowed between the examiners. All subjects performed one familiarization trial and, after a 1-minute rest, they performed three trials of each of the seven investigated tests. One- and three-minute rest intervals were allowed between the trials and tests, respectively. The order of the raters and the tests were randomly determined during both visits.

4 4 Clinical Rehabilitation 0(0) During the tests, no verbal encouragements were given. The same digital stopwatch was used to measure the time for both examiners in both sessions. Prior to data collection, the subjects provided information regarding their histories of falls, other health conditions which could influence the test results. For all subjects, no adverse health issues were reported. Statistical analyses Descriptive statistics and tests for normality (Shapiro Wilk) and homegeneity of variance (Levene) were performed for all variables. Oneway ANOVAs were used to compare the results of each test between the different outcome scores (one trial, the means of the two, three, and the best and the worst performances of the three trials), considering the values obtained by the first rater assessments. Intra-class correlation coefficients (ICCs) were employed to assess the intra- and inter-rater reliabilities and all analyses were performed with SPSS for Windows (SPSS Inc., Chicago, IL, USA) and the significance level was set at Measurement errors were determined by the percentages of the standard errors of the measurements (SEM%), which are independent units. These values represented the limits for the minimal changes, which indicated real improvements for the group of subjects. To perform these analyses, the following calculations proposed by Beckerman et al. 20 and Flansbjer et al. 4 were applied: SEM ¼ square root of the within-subjects error variance, and SEM% ¼ (SEM/mean) 100. Results Sixteen subjects with stroke were assessed (11 men and five women) with a mean age of years (ranging from 26 to 81). Demographic and clinical characteristics of the subjects are reported in Table 1. Table 2 shows the descriptive statistics, as well as the ANOVA results of the comparisons between the different outcome scores for all seven performance-based tests. As can be observed for the mean and standard deviation (SD) values and the ANOVAs for all tests, the Table 1. Subjects demographic and clinical characteristics (n ¼ 16) Variable Result Age (years): mean (SD); range [min max] 52.0 (17.1); [26 81] Time since the onset of stroke (years): mean (SD); range [min max] 4.9 (4.5); [1 12.9] Body mass index (kg/m 2 ): mean (SD); range [min max] 23.9 (5.6); [ ] Gender Men: number (percentage) 11 (68.8%) Women: number (percentage) 5 (31.2%) Paretic side Right: number (percentage) 11 (68.8%) Left: number (percentage) 5 (31.2%) Type of stroke Ischaemic: number (percentage) 10 (62.5 %) Haemorrhagic: number (percentage) 6 (37.5 %) Use of assistive devices No walking aids: number (percentage) 14 (87.5%) Walking aids: number (percentage) 2 (12.5%) No orthoses: number (percentage) 14 (87.5%) Orthoses: number (percentage) 2 (12.5%)

5 Faria et al. 5 Table 2. Descriptive statistics (means SD) and ANOVA results of the comparisons between different outcome scores for the investigated performance-based tests (n ¼ 16) Test First trial Mean of two trials Mean of three trials Best of three trials Worst of three trials ANOVA (F; p-value) Comfortable gait speed (m/s) Maximal gait speed (m/s) Comfortable stair ascent cadence (steps/s) Maximal stair ascent cadence (steps/s) Comfortable stair descent cadence (steps/s) Maximal stair descent cadence (steps/s) F p F p F p F p F p F p 0.98 Timed Up and Go test (s) F p 0.96 values provided by all types of outcome scores were similar (0.01 F 1.03; 0.34 p 0.99). Table 3 gives the ICC values for the intra- and inter-rater reliabilities for all of the investigated tests, considering the different outcome scores. As can be observed, for all tests and outcome scores, significant and adequate values of intra- (0.75 ICC 0.96; p 0.002) and inter-rater (0.75 ICC 0.97; p 0.001) reliabilities were obtained. Table 4 gives the results of the SEM for all tests, considering the different types of outcome scores. As can be observed, the SEM scores, as well as the SEM% values, were quite similar between the different types of outcome scores for all of the investigated tests. All the SEM% values were low, ranging from 5.0% for the comfortable gait speed to 14.8% for the Timed Up and Go test. Discussion The results of the present study demonstrated that, for the seven investigated performancebased tests (maximal and comfortable gait speeds and both stair ascending and descending, as well as the Timed Up and Go ), similar results were found between the scores of one trial, the means of the two, three and the best and the worst scores of the three trials, for subjects with stroke. In addition, similar and adequate values of intra- and inter-rater reliabilities, as well as acceptable measurement errors were found for all types of outcome scores for all of the seven tests. Therefore, only one trial was required to provide consistent and reliable results, with similar and acceptable values of measurement errors obtained with the other types of outcome scores. Gait speeds, stair ascending and descending cadences, and the Timed Up and Go are the most applied and recommended tests to assess mobility in subjects with stroke in both community and institutional settings, given that they are easy to administer, require no specialized training and equipment, and provide adequate measures of subjects mobility, when performing important and common everyday activities. Changes in gait speed are the greatest contributors to functional disabilities in subjects with stroke. 8 Considering that stroke subjects are able to increase their walking speeds above their comfortable levels, 25,28 and this ability

6 6 Clinical Rehabilitation 0(0) Table 3. Intra-class correlation coefficient (ICC) values for the intra- and inter-rater reliabilities for the investigated performance-based tests, considering different outcome scores First trial Mean of two trials Mean of three trials Best of three trials Worst of three trials Test Intra Inter Intra Inter Intra Inter Intra Inter Intra Inter Comfortable gait speed (m/s) 0.94* 0.96* 0.87* 0.97* 0.95* 0.97* 0.92* 0.93* 0.88* 0.93* Maximal gait speed (m/s) 0.86* 0.91* 0.92* 0.96* 0.92* 0.97* 0.84* 0.92* 0.83* 0.91* Comfortable stair ascent cadence (steps/s) 0.89* 0.91* 0.95* 0.97* 0.95* 0.97* 0.86* 0.92* 0.89* 0.93* Maximal stair ascent cadence (steps/s) 0.86* 0.85* 0.94* 0.93* 0.95* 0.94* 0.92* 0.75* 0.85* 0.85* Comfortable stair descent cadence (steps/s) 0.86* 0.84* 0.93* 0.95* 0.93* 0.96* 0.86* 0.92* 0.84* 0.90* Maximal stair descent cadence (steps/s) 0.91* 0.87* 0.96* 0.95* 0.96* 0.97* 0.94* 0.94* 0.88* 0.91* Timed Up and Go test (s) 0.75* 0.91* 0.81* 0.96* 0.85* 0.96* 0.82* 0.91* 0.84* 0.93* *p Table 4. Standard error of measurement (SEM) values for the investigated performance-based tests, considering different outcome scores First trial Mean of two trials Mean of three trials Best of three trials Worst of three trials Test SEM SEM% SEM SEM% SEM SEM% SEM SEM% SEM SEM% Comfortable gait speed (m/s) Maximal gait speed (m/s) Comfortable stair ascent (steps/s) Maximal stair ascent (steps/s) Comfortable stair descent (steps/s) Maximal stair descent (steps/s) Timed Up and Go (s)

7 Faria et al. 7 is important for acceptable mobility levels, 25,28 performance-based tests were proposed to be assessed at both comfortable and maximal speeds. The ICCs are indicated as the preferred and most adequate indices for reliability analyses of interval ratio data, since they represent both correlations and levels of agreement. 10,34 According to Portney and Watkins, 10 ICC values of 0.75 are indicative of acceptable reliability and those below 0.75 are considered poor to moderate. As found in the present results, for both intra- and inter-rater reliabilities for all seven performancebased tests and outcome scores, the ICC values were classified as acceptable. The lowest ICC value was 0.75 for the intra-rater reliability of the Timed Up and Go test, and all of the other values were above As stated by Flansbjer et al., 4 few studies have actually employed ICCs to evaluate the reliability of performance-based tests in subjects with stroke. The ICC values provided in previous studies 4,15,17,35 were similar to those of the present study. Previous studies investigated only the intrarater reliability of these performance-based tests with stroke subjects. According to current knowledge, this is the first study which also investigated the inter-rater reliability of these measures. The SEM% represents the limits for the smallest changes which indicate real improvements for a group of individuals following an intervention. In other words, measurement values following an intervention should be outside the range of the SEM% to indicate real improvements. Therefore, lower values of SEM% are better than higher ones and SEM% values up to 15% are acceptable. 20 All of the SEM% values obtained in the present study were below 15%, which implied that these tests were sensitive and could be used to detect small changes in subjects with stroke. In addition, these values were similar to those reported by Flansbjer et al., 4 who investigated the SEM% for comfortable and maximal gait speeds, comfortable and maximal stair ascending and descending cadences, and the Timed Up and Go, using the means of two trials. It is important to point out that although the SEM values do not represent clinically relevant changes, they could be used as reference values to indicate if a measurement is able to assess such differences. The clinically relevant changes are arbitrarily chosen by the levels of changes which clinicians and researchers could minimally expect or judge to be important after an intervention, but they must exceed the SEM values to be a valid measure for their purposes. 36 Therefore, the present SEM or SEM% values could be used by clinicians as a guide to determine the clinically relevant changes for the seven investigated performance-based tests. Tyson 18 developed the only study which was found regarding the effects of outcome scores for the results of one performance-based test with stroke subjects. The 5-metre gait speed scores, obtained by the mean of two and three trials, were comparable. In the present study, comparable results were also found between the scores of the first trial and those obtained by the means of two, three, and the best and the worst values of the three trials for all of the investigated tests. In addition, the outcome scores provided by only one trial after familiarization did not reduce the intra- and inter-rater reliability coefficients, nor increase the measurement errors. Tyson 18 also investigated the measurement errors considering different outcome scores for the 5-metre gait speed (means of two and three trials) and reported that the average of two trials provided a more accurate measure than the average of three trials. In the present study, the measurement errors were very similar between the scores of one trial, the means of two and three trials, and the best and the worst values of the three trials, for all investigated tests. These results justify the recommendations of using only one trial after a trial of familiarization, as outcome scores for the seven performance-based tests for subjects with stroke. Besides saving valuable clinical time, while reducing the assessment burden placed on the subjects, which is important in daily clinical practice, the outcome score of only one trial did

8 8 Clinical Rehabilitation 0(0) not alter the reliability and measurement error values. In addition, using the scores provided by only one trial may improve the applicability or the feasibility of the performance-based tests. Requiring subjects to perform multiple trials may exclude the least able subjects, who may be able to just perform a test once, but could not do it multiple times. 18 This means that the seven investigated performance-based tests are not only convenient for the researchers and clinicians (e.g. simple to use), but also convenient for the subjects (e.g. comfortable, painless, and easy). 37 Mathematically, more trials would be expected to reduce the measurement errors. 10,18 As pointed out by Portney and Watkins: 10 Theoretically, the most representative scores should be achieved through the mean or average values, because the sum of the error components over an infinite number of trials would be zero. Therefore, it was surprising to find that taking the scores of only the first trial provided error values similar to the mean of three trials for all investigated tests. However, performance-based tests are influenced by the subjects motivations. As pointed out by Tyson, 18 fatigue, boredom, frustration, or loss of attention may become a problem with multiple repetition tests for subjects with stroke. In addition, only two or three trials may not be enough to modify previous assumptions related to infinite number of trials. Therefore, there is no ready answer to the question regarding which type of outcome scores should be used. Researchers should investigate which outcome score provides the most reliable basis for data analyses, 10 asperformedinthe present study. Although the sample size was planned to achieve 80% of power to detect ICCs of 0.95 with a significant level of 0.05, the relatively small number of subjects could not cover the whole range of possible patients, which is one of the main limitations of the present study. The inclusion of a non-randomized convenience sample of subjects with stroke is another limitation. Reliability studies should have variability in subjects characteristics and because of this, the included subjects were of different ages and times since onset of stroke. However, subjects with high levels of disabilities could not be included because of the number of assessment sessions (i.e. three sessions with three trials of seven performance-based tests). Clinical messages. Similar and adequate indices of intra- and inter-rater reliabilities and low measurement errors were found for all investigated outcome scores.. Just one trial of the seven investigated performance-based tests was shown to be adequate to provide consistent and reliable results. Acknowledgements The authors acknowledge the Brazilian Government Agencies (CAPES, CNPq and FAPEMIG) for funding support and Dr John Henry Salmela for his precious contribution in copy-editing the manuscript. Funding This work was supported by the Brazilian National Funding Agencies: CAPES (grant no. BEX0344/07-0), CNPq (grant no /2008-3), and FAPEMIG (grant no ). References 1. Chae J, Johnston M, Kim H and Zorowitz R. Admission motor impairment as a predictor of physical disability after stroke rehabilitation. Am J Phys Med Rehabil 1995; 74: LeBrausser NK, Sayers SP, Ouellette MM and Fielding RA. Muscle impairments and behavioral factors mediate functional limitations and disability following stroke. Phys Ther 2006; 86: Barak S and Duncan PW. Issues in selecting outcome measures to assess functional recovery after stroke. NeuroRX 2006; 3: Flansbjer U, Holmback AM, Downham D, Patten C and Lexell J. Reliability of gait performance tests in men and women with hemiparesis after stroke. J Rehabil Med 2005; 37:

9 Faria et al Salbach NM, Mayo NE, Higgins J, Ahmed S, Finch LE and Richards CL. Responsiveness and predictability of gait speed and other disability measures in acute stroke. Arch Phys Med Rehabil 2001; 82: Salter K, Jutai JW, Teasell R, Foley NC, Bitensky J and Bayley M. Issues for selection of outcome measures in stroke rehabilitation: ICF activity. Disabil Rehabil 2005; 27: Bohannon RW, Andrews AW and Smith BM. Rehabilitation goals of patients with hemiplegia. Int J Rehabil Res 1988; 11: Jorgensen H, Nakayama H, Raaschou H and Olsen T. Recovery of walking function in stroke patients: The Copenhagen study. Arch Phys Med Rehabil 1995; 76: Jansen CWS, Simper VK, Stuart Jr. HG and Pinkerton HM. Measurement of maximum voluntary pinch strength: effects of forearm position and outcome score. J Hand Ther 2011; 16: Portney LG and Watkins MP. Foundations of clinical research: applications to practice, second edition. New Jersey: Prentice-Hall, Patterson KK, Gage WH, Brooks D, Black SE and McIlroy WE. Changes in gait symmetry and velocity after stroke: a cross-sectional study from weeks to years after stroke. Neurorehabil Neural Repair 2010; 24: Plummer P, Behrman AL, Duncan PW, et al. Effects of stroke severity and training duration on locomotor recovery after stroke: a pilot study. Neurorehabil Neural Repair 2007; 21: Rose D, Paris T, Crews E, et al. Feasibility and effectiveness of circuit training in acute stroke rehabilitation. Neurorehabil Neural Repair 2011; 25: Collen FM, Wade DT and Bradshaw CM. Mobility after stroke: Reliability of measures of impairment and disability. Int Disabil Stud 1990; 12: Green J, Forster A and Young J. Reliability of gait speed measured by a timed walking test in patients one year after stroke. Clin Rehabil 2002; 16: Maeda A, Yuasa T, Nakamura K, Higuchi S and Motohashi Y. Physical performance tests after stroke: reliability and validity. Am J Phys Med Rehabil 2000; 79: Ng SS and Hui-Chan CW. The timed up & go test: its reliability and association with lower-limb impairments and locomotor capacities in people with chronic stroke. Arch Phys Med Rehabil 2005; 86: Tyson SF. Measurement error in functional balance and mobility tests for people with stroke: what are the sources of error and what is the best way to minimize error? Neurorehabil Neural Repair 2007; 21: Stephens J and Goldie P. Walking speed on parquetry and carpet after stroke: Effect of surface and retest reliability. Clin Rehabil 1999; 13: Beckerman H, Roebroeck ME, Lankhorst GJ, Becher JG, Bezemer PD and Verbeek AL. Smallest real difference, a link between reproducibility and responsiveness. Qual Life Res 2001; 10: Guyatt G, Walter S and Norma G. Measuring change over time: assessing the usefulness of evaluative instruments. J Chronic Dis 1987; 40: Evans M, Goldie P and Hill K. Systematic and random error in repeated measurements of temporal and distance parameters of gait after stroke. Arch Phys Med Rehabil 1997; 78: Goldie PA, Matyas TA and Evans OM. Deficit and change in gait velocity during rehabilitation after stroke. Arch Phys Med Rehabil 1996; 77: Dean CM, Richards CL and Malouin F. Task-related circuit training improves performance of locomotor tasks in chronic stroke: a randomized, controlled pilot trial. Arch Phys Med Rehabil 2000; 81: Bohannon RW. Walking after stroke: comfortable versus maximum safe speed. Int J Rehabil Res 1992; 15: Nadeau S, Arsenault AB, Gravel D and Bourbonnais D. Analysis of the clinical factors determining natural and maximal gait speeds in adults with a stroke. Arch Phys Med Rehabil 1999; 78: Olney SJ, Griffin MP and McBride ID. Temporal, kinematic, and kinetic variables related to gait speed in subjects with hemiplegia: a regression approach. Phys Ther 1994; 74: Olney SJ and Richards C. Hemiparetic gait following stroke. Part I: characteristics. Gait Posture 1996; 4: Cohen JJ, Sveen JD, Walker JM and Brummel-Smith K. Established criteria for community ambulation. Top Geriatr Rehabil 1987; 3: Bohannon RW and Walsh S. Association of paretic lower extremity muscle strength and balance with stair climbing ability in patients with stroke. J Stroke Cerebrovasc Dis 1991; 1: Podsiadlo D and Richardson S. The Timed Up & Go : a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc 1991; 39: Roy G, Nadeau S, Gravel D, Malouin F, McFadyen BJ and Piotte F. The effect of foot position and chair height on the asymmetry of vertical forces during sit-to-stand and stand-to-sit tasks in individuals with hemiparesis. Clin Biomech 2006; 21: Monger C, Carr JH and Fowler V. Evaluation of a home-based exercise and training programme to improve sit-to-stand in patients with chronic stroke. Clin Rehabil 2002; 16: Lexell JE and Downham DY. How to assess the reliability of measurements in rehabilitation. Am J Phys Med Rehabil 2005; 84:

10 10 Clinical Rehabilitation 0(0) 35. Eng JJ, Chu KS, Dawson AS, Kim CM and Hepburn KE. Functional walk tests in individuals with stroke: relation to perceived exertion and myocardial exertion. Stroke 2002; 33: Hébert R, Spiegelhalter DJ and Brayne C. Setting the minimal metrically detectable change on disability rating scales. Arch Phys Med Rehabil 1997; 78: Bussmann JBJ and Stam HJ. Techniques for measurement and assessment of mobility in rehabilitation: a theorical approach. Clin Rehabil 1998; 12: