Reliability and concurrent validity of the Expanded Timed Up-and-Go test in older people with impaired mobility

Transcription

1 Physiotherapy Research International Published online in Wiley InterScience ( Reliability and concurrent validity of the Expanded Timed Up-and-Go test in older people with impaired mobility PERNILLE BOTOLFSEN Faculty of Health Sciences, Oslo University College, Pilestredet 44, N-0167 Oslo, Norway JORUNN L. HELBOSTAD Department of Neuroscience, Norwegian University of Science and Technology (NTNU), Edward Griegs g., N-7030 Trondheim, Norway ROLF MOE-NILSSEN Department of Public Health and Primary Health Care, University of Bergen, P.O. 7804, N-5020, Bergen, Norway JAMES C. WALL Department of Physical Therapy, University of South Alabama, Mobile, AL, USA ABSTRACT Background and Purpose. Expanded Timed Up-and-Go (ETUG) was developed to assess each of the subtasks of the Timed Up-and-Go (TUG). The aim of the study was to test the intrarater, interrater, test retest reliability and internal consistency of the ETUG, and the concurrent validity with the TUG. Methods. The present study is a reliability and a validity study. Twenty-eight subjects (80 ± 4.1 years) with balance and gait problems were included. Three raters timed the ETUG subtasks from a video, using a computer-based scoring programme, and the total ETUG time was calculated. TUG was registered by a regular stopwatch. Results. The intrarater and interrater reliability (intraclass correlation [ICC][1,1]) ranged from 0.55 to The test retest reliability (ICC[1,1]) ranged from 0.54 to The absolute measurement error of the total time (1.96 S w ) was 2.8 seconds. The internal consistency (Cronbach s alpha) was The correlation (Pearson s r) between ETUG total time and TUG after correcting for attenuation caused by restricted reliability in each of the measures was Conclusion. The ETUG scored from a video shows a good reliability for experienced raters and acceptable internal consistency. The ETUG showed a higher reliability than TUG when tested on the same sample of older subjects with impaired mobility, and the high concurrent validity between ETUG and TUG suggests that the two tests may have similar properties. Since ETUG also adds new information compared with TUG, we suggest that ETUG is an interesting alternative to existing clinical tests of mobility.. Key words: functionally impaired elderly, mobility limitations, outcome measures, reproducibility of findings

2 Botolfsen et al. INTRODUCTION Balance is an integral part of almost any movement, and is often a limitation for function at high age (Gill-Body et al., 2000; Marsh et al., 2003). Since the 1990s, there has been an increased focus on and development of objective functional outcomes with - in rehabilitation and geriatric medicine (Huijbregts et al., 2002). These include measures of fall risk, balance, gait and activities of daily living (Podsiadlo and Richardson, 1991; Cole et al., 1994; Buchner et al., 1996; Finch et al., 2002). There is still a lack of objective balance measures that support clinical decision making. We need measures reflecting everyday balance function that are useful for diagnosis, for choosing specific interventions or evaluating the effects of interventions (Huijbregts et al., 2002). Hence, measurements must be reliable and valid. Reliability tells us if measurements are repeatable and free from measurement errors and can be measured within and between raters, and between two different measurements (Domholdt, 2000). Validity tells us how meaningful and useful the scores are (Domholdt, 2000). When developing a new outcome measure, one commonly looks at concurrent validity, which is assessed as the correlation with an instrument seen as a gold standard (Dunn, 1989). Additionally, internal consistency should be assessed when individual items of a test are used to make a summary score. Internal consistency (Cronbach s alpha) looks at the extent to which individual items correlates with the other items and add information to the summary score (Shrout and Fleiss, 1979). Reliability is a prerequisite, but not necessarily a sufficient condition for good validity of a test (Bork, 1993). When developing a new test, reliability and validity of the test need to be established relative to the clinical populations on which the test is going to be used. The Get Up and Go (GUG) developed by Mathias et al. (1986) aimed to screen fall risk in elders through the performance of a combined task: rise from a chair, walk 3 m, turn, walk back and sit down again. The performance was evaluated on a fivepoint ordinal scale (Mathias et al., 1986). Podsiadlo and Richardson modified the test by timing the combined task (Podsiadlo and Richardson, 1991), described as the Timed Up-and-Go (TUG). Several studies concluded that TUG is an indicator of risk of falling, balance and general locomotion (Podsiadlo and Richardson, 1991; Cole et al., 1994; Samson et al., 2000; Trueblood et al., 2001; Boulgarides et al., 2003). Each of the subtasks of the TUG is relevant for daily life functioning. However, when only focusing on the one score based on the combined tasks, problems in performing separate subtasks may be camouflaged. Furthermore, such a score may not give sufficient information to guide the choice of intervention, even though it can be useful in assessing the effect of such treatment. Therefore, a test that includes the same subtasks as the TUG, but that scores each subtask separately, can possibly add information of relevance for clinical decision making in rehabilitation and geriatric medicine and in evaluating changes over time. The Expanded Timed Get Up and Go (ETGUG) was developed to answer these shortcomings of GUG and TUG (Wall et al., 2000). In the ETGUG, a combined task similar to the one used in GUG and TUG was applied, but each part of the test, such as turning, was timed separately. In addition, the total time was reported. As opposed to GUG and TUG, the ETGUG used a longer walkway (10 m) and an armless chair

3 Validity of the ETUG test (Mathias et al., 1986; Podsiadlo and Richardson, 1991; Wall et al., 2000). What is common to all three tests is that they were completed as one continuous task. This made it difficult to determine when certain subtasks had been completed. For example, the subject was instructed to stand up and walk, which posed the problem of determining when the subject was standing. This was particularly difficult when subjects started walking before becoming fully upright. Although the multimemory stopwatch used to time parts of the ETGUG is an inexpensive, reliable and valid instrument, it does require manual transcription of the data to a results sheet as well as certain calculations, such as walking speed, to be made. To overcome these problems, a modified version of the ETGUG is presented here (Expanded Timed Up-and-Go [ETUG]), where subjects perform each of the subtasks of TUG in a series, but with a stop and a new instruction before the next subtask is undertaken. To be more like the original TUG, a chair with armrests is used, and the timed walking distance is 3 m (Podsiadlo and Richardson, 1991). A computer programme was written that mimicked the use of the multimemory stopwatch but had the advantage of being able to collect, store and analyse the data. The first step in validating the ETUG is to establish reliability when scored from a video. A next step includes testing the reliability when the test is scored from the direct observation of the performed test. The aim of the present study was to test the intrarater, interrater and test retest reliability, and also the internal consistency of the ETUG from video recordings, as well as concurrent validity with the TUG. MATERIAL AND METHODS Design The study was designed as a reliability and validity study where all the subjects underwent the same test protocol, and repeated measurements were evaluated within the same test session. Subjects A convenience sample of 33 home-dwelling subjects (27 women, 6 men) of 75 years of age and older with impaired mobility were recruited from an ongoing randomized controlled exercise study (RCT) (Helbostad et al., 2004a, 2004b). The inclusion criteria for the RCT was that subjects 1) had fallen at least once during the last year or used some kind of walking aid daily; 2) were able to walk at least 10 m without any walking aid; and 3) had not experienced a cerebral vascular accident during the last six months. They were excluded from the RCT if they had a mini-mental status examination score below 22, were terminally ill or had diseases that made physical exercise contraindicated. An additional inclusion criterion for the present study was that the subjects had to be able to walk without an assistive device. Only one person from the RCT needed a walking aid during testing and data from this person were therefore excluded. Four additional persons, initially included in the present study, were excluded from the analyses be cause of incomplete or missing video data due to technical problems, resulting in a final sample size of 28 subjects (23 women, 5 men). The study was approved by the regional ethics committee for medical research and the Data Inspectorate in Norway. Three raters, two experienced physiotherapists (A and C) and one researcher (B)

4 Botolfsen et al. who had developed the ETGUG and ETUG tests were enrolled. Raters A and B were more experienced with the rating procedures than rater C, who only had one and a half hours of assessment practice. An experienced research physiotherapist (D) timed the TUG for the concurrent validity testing against the ETUG. Test procedures The subjects were tested and retested with TUG and ETUG during the same test session, with a few minutes rest between the two trials of each test. The testing order was the same for all the subjects. TUG was performed following the test protocol of Podsiadlo and Richardson. While Podsiadlo only use the second trial as counting, we included both the first and the second trial in the analyses (Podsiadlo and Richardson, 1991). A regular stopwatch was used to register the time (in 1/10 of a second). The ETUG was recorded on a digital video camera. 1 The test was videotaped from a side view and the recording included sound. The stationary camera was placed half a metre from the edge of the walkway and operated by a person. The placement of the camera gave a maximal angle of 50 for recordings at each end of the walkway relative to a side view of the person being tested. The field of view was limited to the hips and the lower extremities for the walkway; however, the whole person was captured during the sit-tostand and turn subtasks. The videotapes were copied to analogue VHS video cassettes for further analyses. Using a customdesigned computer programme developed for the ETUG and written in Metacard, 2 each sub task was timed by clicking the left mouse button. The programme allowed the raters to time each subtask repeatedly before deciding on the value to be saved. The first test series was used to test the intra- and interrater reliability. In the intrarater reliability, study raters A, B and C analysed the first test series twice, at least one month apart. The test retest reliability study was based on the data from the two test series collected by rater A. Scoring the ETUG Each subtask was instructed in the same way: After I have counted three, two, one, I want you to start. Are you ready? Three, two, one, start. Subtask 1: sit-to-stand The subjects sat on a 46-cm high chair with armrests with their back against the back of the chair. The instruction was to rise to an upright position and stand still. There was no instruction on the use of armrests, but whenever used, it was recorded. The time was started at the instruction start and stopped when the subject was standing upright and still. Subtask 2: 3-m walk at preferred speed The subjects were asked to walk a distance of 6 m at their preferred gait speed and then stop without turning. Time was recorded for the middle 3 m. Start and stop times were registered when the subject s hips/body 1 Sony DCR-PC100E. Manufacturer: Sony Corporation, Tokyo , Japan. 2 Metacard has been acquired by Runtime Revolution, Runtime Revolution Ltd York Place Edinburgh Scotland, UK EH1 3EB.

5 Validity of the ETUG test passed two lines on the floor, one at the beginning and the other at the end of the central 3-m region of the walkway. Subtask 3: 180 turn At the start of the 180 turn, the subjects stood with their back against the walkway. Time was taken from the instruction start until the subjects had turned 180 and were standing still, facing the walkway and the chair. Subtask 4: 3-m walk at fast speed The subjects were asked to walk a distance of 6 m at their fast but safe speed and then stop without turning. Time was recorded for the middle 3 m. Start and stop times were registered when the subject s hips passed two lines on the floor, one at the beginning and the other at the end of the central 3-m region of the walkway. Subtask 5: turn and sit down The subjects stood in front of and facing the chair and were instructed to turn and sit down. Time was registered from the instruction start until the subjects were sitting on the chair. Total time The ETUG total time was calculated by adding up the time for all five subtasks: sitto-stand, walking at preferred speed, 180 turn, walking at fast speed, and turn and sit down. Data analysis Statistical analyses were performed in SPSS A significance level of 0.05 was chosen. Intraclass correlation (ICC) was computed to express measurement variability relative to total variability (Shrout and Fleiss, 1979) and within-subject standard deviation (S w ) calculated as an expression of absolute variability (Bland and Altman, 1996a). TUG and ETUG were both performed twice, giving two values per test. These values were all used in the calculations. ICC(1,1) and ICC(3,1) were used to assess reliability. ICC(1,1) is based on a one-way analysis of variance (Shrout and Fleiss, 1979) and sees all within-subject variability as measurement error, while ICC(3,1) does not include systematic difference between the repeated measurements as measurement error (Shrout and Fleiss, 1979; Bland and Altman, 1996a). If ICC(1,1) and ICC(3,1) are identical, no systematic error is present (Bland and Altman, 1996c; Moe-Nilssen, 1998). ICC is a measure of agreement between the measurements, is not influenced by the order of the measurements (Bland and Altman, 1996b) and is independent of the measurement unit. S w is the measurement error for repeated measurements reported in absolute units (Bland and Altman, 1996c; Moe-Nilssen, 1998) and is calculated as the square root of the mean between the subjects variance (Bland and Altman, 1996a). The difference between a subject s measurement and the true value is expected to be less than 1.96 S w for 95% of the observations (Bland and Altman, 1996a) S w is reported for intraand interrater reliability as well as for the test retest reliability. Within-scale between-items correlation and Cronbach s alpha were reported for the first trial series for rater A to test the internal consistency of the timed subtasks of the ETUG. To test the concurrent validity between the two trial series of the ETUG and the two

6 Botolfsen et al. trials of the TUG, Pearson s product moment correlation coefficients (r) were calculated using the ETUG test retest scores from rater A. When the concurrent validity is reported by Pearson s r, the correlation coefficient is attenuated by the restricted reliability of each of the measures. Thus, a measure cannot be closer associated to another measure obtained on a separate occasion than to itself obtained on separate occasions. Spearman suggested a procedure for correcting Pearson s r for the attenuation caused by restricted reliability based upon the measured correlation between the two measures and the correlation obtained for repeated measurements of each of the two measures (Spearman, 1987; Zimmerman and Williams, 1997). To correct for the unreliability, we therefore applied Spearman s design to the concurrent validity correlation coefficients for the ETUG and TUG. systematic differences. Differences between ICC(1,1) and ICC(3,1) interrater reliability between raters A and B was <0.04, between raters A and C was <0.13, and raters B and C was <0.22. The subtasks sit-to-stand and walking at preferred speed showed the highest differences. The test retest reliability also demonstrated small differences (<0.06) between ICC(1,1) and ICC(3,1), suggesting no learning effect between test and retest. Thus, only ICC(1,1) is reported. The intrarater reliability for the three raters, expressed as ICC(1,1), varied between 0.91 and 0.97 for the ETUG total time, and between 0.75 and 0.97 for the subtasks. The absolute variability of the total time (1.96 S w ) was between 1.3 and 2.2 seconds for the three raters (Table 3). The interrater reliability (ICC[1,1]) between raters A and B, between raters A TABLE 1: Background variables (N = 28) RESULTS Mean SD The sample characteristics are presented in Table 1. The subjects were 23 women and 5 men. Eighteen subjects (64%) reported the use of a walking aid sometimes or always, and 19 subjects (68%) reported to have fallen the year before study onset. Fifteen subjects (54%) used one or two armrests during the sit-to-stand subtask. Twenty-two subjects (79%) used walking shoes and six subjects (21%) used dress shoes during testing. Reliability There were small differences in the mean absolute test values between the two measurements for each rater, as demonstrated in Table 2. There were small differences in the intrarater reliability (<0.07) between ICC (1,1) and ICC(3,1), indicating very low Age (years) 80.0 (4.14) Height (m) 1.63 (0.08) Weight (kg) 69.8 (14.4) BMI (kg/m²) 25.9 (4.0) No. of prescribed medications 3.5 (2.4) No. of medical diagnoses 4.5 (1.6) Number Percent Previous medical diagnosis Stroke 5 (17.9) Heart disease 15 (53.6) Hypertension 10 (35.7) Respiratory disease 4 (14.3) Diabetes mellitus 3 (10.7) Muscular/skeletal disease 21 (75.0) Cognitive impairment 4 (14.3) Depression 6 (21.4) Incontinence 8 (28.6) Visual impairment 17 (60.7) Hearing impairment 8 (28.6) Syncope 5 (17.9) Other diagnoses 23 (82.1) SD = standard deviation, BMI = Body Mass Index.

7 Validity of the ETUG test TABLE 2: Expanded Timed Up-and-Go s means and standard deviations (SDs) for test series 1 for raters A, B and C, and test series 2 for rater A Subtask Rater A test series 1 Rater B test series 1 Rater C test series 1 Rater A test series 2 1. Rating 2. Rating 1. Rating 2. Rating 1. Rating 2. Rating Re-test rating x SD x SD x SD x SD x SD x SD x SD Total time (seconds) Sit-to-stand (seconds) Walking, preferred speed (seconds) Turn 180 (seconds) Walking, fast speed (seconds) Turn and sit down (seconds) Data are based on the results of 28 subjects. x = mean. TABLE 3: Expanded Timed Up-and-Go s relative and absolute intrarater reliability for raters A, B and C: intraclass correlation (ICC)(1,1) with 95% confidence interval (CI) and 1.96 S w Subtask Intrarater reliability rater A Intrarater reliability rater B Intrarater reliability rater C ICC(1,1) (95% CI) 1.96 S w ICC(1,1) (95% CI) 1.96 S w ICC(1,1) (95% CI) 1.96 S w Total time (seconds) ± ± ±2.21 Sit-to-stand (seconds) ± ± ±0.81 Walking, preferred speed (seconds) ± ± ±0.91 Turn 180 (seconds) ± ± ±0.86 Walking, fast speed (seconds) ± ± ±0.55 Turn and sit down (seconds) ± ± ±0.72 Data are based on the results of 28 subjects. S w = within subject standard deviation.

8 Botolfsen et al. and C, and between raters B and C for the ETUG total time was 0.96, 0.87 and 0.87 respectively, and varied between 0.55 and 0.96 for the subtasks. The absolute error (1.96 S w ) for total time was smaller between raters A and B (1.5 seconds) than between raters A and C, and between raters B and C (both 2.7 seconds) (Table 4). The test retest reliability ICC(1,1) for the ETUG total time for rater A was 0.84 and between 0.54 and 0.85 for the subtasks (Table 5). For comparison, the test retest reliability for TUG was calculated, returning ICC(1,1) = The internal consistency of the ETUG measured by Cronbach s alpha was The correlation between timed subtasks varied from r = 0.12 between the sit-to-stand and the 180 turn to r = 0.77 between normal and fast walking (Table 6). Validity The concurrent validity was investigated by comparing the total time for rater A s test and retest ETUG rating with rater D s test and retest TUG rating. A measured correlation coefficient (r) of 0.65 was found for the second of two trials. Based upon repeated measures, Pearson s r of 0.87 for ETUG and 0.68 for TUG, Spearman s correction for attenuation was applied, resulting in a corrected correlation between the two measures of DISCUSSION This is the first study to examine the reliability and validity of the ETUG. In order to separate the rater reliability, where measurement error is caused by variability in scoring, from the test retest reliability, where measurement error also includes the variability in subject performance, we chose to perform TABLE 4: Expanded Timed Up-and-Go s relative and absolute interrater reliability between raters A and B, A and C, and B and C: intraclass correlation (ICC)(1,1) with 95% confidence interval (CI) and 1.96 S w Subtask Interrater reliability raters A and B Interrater reliability raters A and C Interrater reliability raters B and C ICC(1,1) 95% CI 1.96 S w ICC(1,1) 95% CI 1.96 S w ICC(1,1) (95% CI) 1.96 S w Total time (seconds) ± ± ±2.69 Sit-to-stand (seconds) ± ± ± ± ± ±1.04 Walking, preferred speed (seconds) Turn 180 (seconds) ± ± ±0.94 Walking, fast speed (seconds) ± ± ±0.47 Turn and sit down (seconds) ± ± ±0.70 Data are based on the results of 28 subjects. S w = within subject standard deviation.

9 Validity of the ETUG test TABLE 5: Expanded Timed Up-and-Go s relative and absolute for test retest reliability for rater A: intraclass correlation (ICC)(1,1) with 95% confidence interval (CI) and 1.96 S w Subtask Test retest reliability rater A ICC(1,1) (95% CI) 1.96 S w Total time (seconds) ±2.76 Sit-to-stand (seconds) ±1.13 Walking, preferred speed (seconds) ±0.74 Turn 180 (seconds) ±1.39 Walking, fast speed (seconds) ±0.54 Turn and sit down (seconds) ±1.32 Data are based on the results of 28 subjects. S w = within subject standard deviation. TABLE 6: Expanded Timed Up-and-Go s inter-item correlation, test series 1 from rater A Sit-to-stand (seconds) Walking, preferred speed (seconds) Turn 180 (seconds) Walking, fast speed (seconds) Walking, preferred speed (seconds) 0.28 Turn 180 (seconds) Walking, fast speed (seconds) Turn and sit down (seconds) Data are based on the results of 28 subjects. all scoring from videotapes. We found an acceptable intrarater and interrater reliability. As expected, the test retest reliability was somewhat lower than the intra- and interrater reliability. The internal consistency was acceptable, and there was a high correlation with the TUG after having controlled for restriction in reliability in each of the measures. The ETUG showed an overall good reliability within and between raters (Fleiss, 1986; McCall, 2001; Munro, 2001). The intrarater reliability for the total time measured by ICC was slightly higher for the more experienced raters than for the less experienced rater. The absolute intrarater reliability of the total score reported as 1.96 S w varied between 1.3 and 2.2 seconds for the three raters, suggesting 95% confidence intervals for the difference from a true value for each tester (Table 3). The interrater reliability confirmed the importance of experience (Table 4). This issue has not always been approached in previous studies (Podsiadlo and Richardson, 1991; Shumway-Cook et al., 2000; Wall et al., 2000), or it was stated that raters were experienced therapists, but not whether they were trained in the specific scoring procedures of interest (DiFabio and Seay, 1997; Siggeirsdottir et al., 2002). Our results indicate that this may be an important aspect to report, and we believe this is a factor that applies to all outcome measures used in physiotherapy practice. We therefore suggest that therapists also familiarize themselves

10 Botolfsen et al. in using the ETUG scoring system before applying this tool in the clinic and for research purposes. The measurement stability over time was reported as the test retest reliability, which includes the true subject variability in addition to rater error. As expected, the test retest reliability of the total ETUG score was somewhat lower than the intrarater reliability (Tables 3 and 5), indicating that shortterm variability in the behaviour of the participating subjects did affect the results. This source of variability should not be viewed as a measurement error, but rather, as a characteristic of the sample that should be identified by a reliable measure. A very small difference between ICC(1,1) and ICC(3,1) indicates no learning effect between the test and retest in this study. Interestingly, the reliability of the ETUG total score was higher than that of the TUG, possibly because considerable random error is eliminated when adding together scores from a number of subtasks. This has similarities to the improvement in reliability seen when using the average of a number of repeated tests as test parameter instead of only one execution of a test. The sit-to-stand subtask demonstrated low reliability (Tables 4 and 5). There are several reasons for this. Firstly, the time to complete the subtask is very short, around three seconds, meaning that the measurement error may be relatively large. Secondly, it is difficult to define when a person has finalized the manoeuvre (Schenkman et al., 1990; Ashford and De Souza, 2000). However, we still consider this a relevant subtask to measure since it is a common everyday function, associated with functional limitations in the elderly population. However, used alone with only one repetition, one has to be aware of the restricted reliability. The internal consistency expressed as Cronbach s alpha is a way to look at the extent to which individual items in a scale go together to form a sum score. Cronbach s alpha is a model of ICC (Shrout and Fleiss, 1979), based on the average inter-item correlation. The obtained score of 0.74 in our study is in accordance with recommendations for acceptable internal consistency in multi-item batteries representing a single construct (Bland and Altman, 1997; Fayers, 2000; Munro, 2001). Our results also show acceptable correlations between subtasks. The inter-item correlation matrix (Table 5) shows that the sit-to-stand subtask has the lowest correlation with any of the other subtasks. It is to be expected that an item demonstrating low reliability, will also demonstrate low association with other items. Since the ETUG showed a good overall intra- and interrater reliability, it was appropriate to test the concurrent validity. TUG was chosen as the gold standard since TUG is well established (Domholdt, 2000) and was the original test from which the ETUG was developed. When corrected for attenuation caused by restricted reliability, we found a correlation of 0.85 for the test and retest between the TUG and ETUG. This indicates that the two tests, to a large degree, are testing the same phenomenon. This is not surprising, since the subtasks are quite similar, only that in the ETUG, subtasks are performed separately, while in the TUG, they are performed as a continuous task. A rationale for developing a new test is to add new information compared to older tests. Although the TUG and ETUG comprise similar motor tasks, only ETUG can identify component subtasks, such as sit-tostand, that should be the focus of treatment. Further studies are warranted to investigate whether this new test also demonstrates

11 Validity of the ETUG test sensitivity to change and thereby can be used to assess the outcome of clinical interventions. Although the ETUG and TUG tests were designed so that they were recorded in real time, in this study, timing was done from videotape recordings so that the intrarater reliability could be determined. In future studies, timing should be performed in real time to ensure that it meets the clinical preference for the measurement of functional mobility. The test retest was done with only a few minutes rest. The literature shows varied time between the test and retest in studies with older subjects (Roorda et al., 1996; Franchignoni et al., 1998; Rockwood et al., 2000; Helbostad et al., 2004c). Deitz (1989) discussed the criteria for length of time between test and retest and suggested that one should decide the time between tests according to age, health and type of test (Deitz, 1989). The subjects in the present study had impaired mobility and this may cause variability in function from one day to the other (Bortz, 2002). To eliminate this source of error, we chose to do the test and retest during the same session. The present study was performed on a sample of older subjects with gait or balance problems and at risk of falling, similar to the population for whom the TUG was developed (Podsiadlo and Richardson, 1991). Our results on reliability may therefore be generalized to such a population. However, reliability may vary between populations (Helbostad et al., 2004c), and care should be taken in generalizing the results of this study to other clinical populations. CONCLUSION In conclusion, ETUG scored from a video showed a good reliability within and between experienced raters. The test retest reliability showed a somewhat larger measurement error than the rater reliability, which was based upon repeated video scoring of the same events. The tests of the internal consistency of ETUG demonstrated that all the timed subtasks add information to the total time score. ETUG showed a higher reliability than the TUG when tested on the same sample of older subjects with impaired mobility, and the high concurrent validity between the ETUG and TUG suggests that the two tests may have similar properties. Since ETUG, as opposed to TUG, adds new information on performance of subtasks, we suggest that ETUG is an interesting alternative to existing clinical tests of mobility. CLINICAL IMPLICATIONS ETUG is a new mobility test that can assess each of the subtasks of the TUG. ETUG is a reliable and valid clinical test that can be used for older persons with impaired mobility. ACKNOWLEDGEMENTS This project is partly financed by the Norwegian Foundation for Research in Physiotherapy and University of Bergen. The authors also acknowledge Karen Clausen, Physical Therapist, for her part in data registration. REFERENCES Ashford S, De Souza L. A comparison of the timing of muscle activity during sitting down compared to standing up. Physiotherapy Research International 2000; 5: Bland JM, Altman DG. Statistics notes: measurement error. British Medical Journal 1996a; 313: 744. Bland JM, Altman DG. Statistics notes: measurement error and correlation coefficients. British Medical Journal 1996b; 313:

12 Botolfsen et al. Bland JM, Altman DG. Statistics notes: measurement error proportional to the mean. British Medical Journal 1996c; 313: 106. Bland JM, Altman DG. Cronbach s alpha. British Medical Journal 1997; 314: 572. Bork CE. Research in Physical Therapy. Philadelphia, PA: Lippincott, Bortz WM. A conceptual framework of frailty: a review. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences 2002; 57: M283 M288. Boulgarides LK, McGinty SM, Willett JA, Barnes CW. Use of clinical and, impairment-based tests to predict falls by community-dwelling older adults. Physical Therapy 2003; 83: Buchner DM, Cress ME, Esselman PC, Margherita AJ, de Lateur BJ, Campbell AJ, Wagner EH. Factors associated with changes in gait speed in older adults. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences 1996; 51: M297 M302. Cole B, Finch E, Gowland C, Mayo N. Physical Rehabilitation Outcome Measures. Toronto, Canada: Canadian Physiotherapy Association, Deitz JC. Reliability. Physical & Occupational Therapy in Pediatrics 1989; 9: DiFabio RP, Seay R. Use of the fast evaluation of mobility, balance, and fear in elderly community dwellers: validity and reliability. Physical Therapy 1997; 77: Domholdt E. Physical Therapy Research: Principles and Applications. Philadelphia, PA: Saunders, Dunn W. Validity. Physical & Occupational Therapy in Pediatrics 1989; 9: Fayers PM. Quality of Life: Assessment, Analysis and Interpretation. Chichester: John Wiley, Finch E, Brooks D, Statford PW, Mayo NE. Physical Rehabilitation Outcome Measures: A Guide to Enhanced Clinical Decision-Making. Baltimore, MD: Lippincott Williams & Wilkins, Fleiss JL. The Design and Analysis of Clinical Experiments. New York, NY: Wiley, Franchignoni F, Tesio L, Martino MT, Ricupero C. Reliability of four simple, quantitative tests of balance and mobility in healthy elderly females. Aging Clinical and Experimental Research 1998; 10: Gill-Body KM, Beninato M, Krebs DE. Relationship among balance impairments, functional performance, and disability in people with peripheral vestibular hypofunction. Physical Therapy 2000; 80: Helbostad JL, Sletvold O, Moe-Nilssen R. Effects of home exercises and group training on functional abilities in home-dwelling older persons with mobility and balance problems. A randomized study. Aging Clinical and Experimental Research 2004a; 16: Helbostad JL, Sletvold O, Moe-Nilssen R. Home training with and without additional group training in physically frail old people living at home: effect on health-related quality of life and ambulation. Clinical Rehabilitation 2004b; 18: Helbostad JL, Askim T, Moe-Nilssen R. Short-term repeatability of body sway during quiet standing in people with hemiparesis and in frail older adults. Archives of Physical Medicine and Rehabilitation 2004c; 85: Huijbregts MPJ, Myers AM, Kay TM, Gavin TS. Systematic outcome measurement in clinical practice: challenges experienced by physiotherapists. Physiotherapy Research International 2002; 54: Marsh AP, Rejeski WJ, Lang W, Miller ME, Messier SP. Baseline balance and functional decline in older adults with knee pain: the Observational Arthritis Study in Seniors. Journal of the American Geriatrics Society 2003; 51: Mathias S, Nayak US, Isaacs B. Balance in elderly patients: the get-up and go test. Archives of Physical Medicine and Rehabilitation 1986; 67: McCall RB. Fundamental Statistics for Behavioral Sciences. Belmont, CA: Wadsworth Publishing Company, Moe-Nilssen R. Test-retest reliability of trunk accelerometry during standing and walking. Archives of Physical Medicine and Rehabilitation 1998; 79: Munro BH. Statistical Methods for Health Care Research. Philadelphia, PA: Lippincott, Podsiadlo D, Richardson S. The Timed Up and Go a test of basic functional mobility for frail elderly persons. Journal of the American Geriatrics Society 1991; 39: Rockwood K, Awalt E, Carver D, MacKnight C. Feasibility and measurement properties of the functional reach and the timed up and go tests in the Canadian study of health and aging. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences 2000; 55: M70 M73.

13 Validity of the ETUG test Roorda LD, Roebroeck ME, Lankhorst GJ, van Tilburg T, Bouter LM. Measuring functional limitations in rising and sitting down: development of a questionnaire. Archives of Physical Medicine and Rehabilitation 1996; 77: Samson MM, Meeuwsen IBAE, Crowe A, Dessens JAG, Duursma SA, Verhaar HJJ. Relationships between physical performance measures, age, height and body weight in healthy adults. Age and Ageing 2000; 29: Schenkman M, Berger RA, Riley PO, Mann RW, Hodge WA. Whole-body movements during rising to standing from sitting. Physical Therapy 1990; 70: Shrout PE, Fleiss JL. Interclass correlations: uses in assessing rater reliability. Psychological Bulletin 1979; 86: Shumway-Cook A, Brauer S, Woollacott M. Predicting the probability for falls in communitydwelling older adults using the Timed Up & Go Test. Physical Therapy 2000; 80: Siggeirsdottir K, Jonsson BY, Jonsson H Jr, Iwarsson S. The timed Up & Go is dependent on chair type. Clinical Rehabilitation 2002; 16: Spearman C. The proof and measurement of association between two things. By C. Spearman, The American Journal of Psychology 1987; 100: Trueblood PR, Hodson-Chennault N, McCubbin A, Youngclarke D. Performance and impairmentbased assessment among community dwelling elderly; sensitivity and specificity. Issues on Aging 2001; 24: 2 6. Wall JC, Bell C, Campbell S, Davis J. The timed get-up-and-go test revisited: measurement of the component tasks. Journal of Rehabilitation Research and Development 2000; 37: Zimmerman DW, Williams RH. Properties of the spearman correction for attenuation for normal and realistic non-normal distributions. Applied Psychological Measurement 1997; 21: Address correspondence to: Pernille Botolfsen MSc, Faculty of Health Sciences, Oslo University College, Pilestredet 44, N-0167 Oslo, Norway. Pernille.botolfsen@hf.hio.no (Submitted July 2007; accepted December 2007)