Measurement_of_mobility_following_stroke.pdf

Physiotherapy 90 (2004) 132–138

Measurement of mobility following stroke: a comparison of the Modified Rivermead Mobility Index and the Motor Assessment Scale夽 Louise Johnson a,∗ , James Selfe b a

Division of Physiotherapy, School of Health Studies, University of Bradford, Unity Building, 25 Trinity Road, Bradford, West Yorkshire BD5 0BB, UK b Allied Health Professions Unit, University of Central Lancaster, Lancaster, UK

Abstract Background and purpose The Modified Rivermead Mobility Index (MRMI) is a newly developed outcome measure that aims to evaluate the effectiveness of physiotherapy on mobility following stroke. Any new measurement tool requires extensive testing of its validity and reliability before it can be recommended for use in clinical practice or research. The purpose of this study was to investigate the concurrent validity of the MRMI when measuring mobility in patients who have had a stroke. The internal consistency and test administration times of the MRMI and Motor Assessment Scale (MAS) were also investigated. Methods Twenty-six hospitalised acute/sub-acute stroke patients from the Medical and Elderly wards of a General Hospital in West Yorkshire were assessed independently with the MRMI and MAS. Test administration time was also recorded. Results Limits of agreement indicated that on average subjects scored three points higher on the MRMI than the MAS (mobility-related items). Ninety-five percent of subjects scored between one point lower and seven points higher on the MRMI than the MAS. Both scales possessed high internal consistency (MRMI α = 0.949 and MAS α = 0.953). Individual items also possessed high internal consistency (MRMI α = 0.743–0.959, MAS = 0.854–0.893) except the sitting balance items (MRMI α = 0.304 and MAS α = 0.545). Both scales took an average of 17 min to administer. Conclusions The mean difference between scores on the MRMI and MAS was small enough to allow clinicians to use either scale to measure mobility in stroke patients. Both scales possessed high internal consistency except the sitting balance items that may be measuring a different construct to mobility. The MRMI and MAS are sufficiently quick to administer to advocate use in routine clinical practice. © 2004 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved. Keywords: Validity; Measurement; Stroke; Outcomes; Mobility

Introduction In today’s climate of evidence-based practice the need to use valid, reliable and clinically sensitive outcome measures to evaluate effectiveness of interventions is acknowledged [1]. The Rivermead Mobility Index (RMI) is considered to be a valid and reliable tool that assesses mobility in stroke patients [2–5]. The RMI measures 15 mobility-related items using a two-point ordinal scale but 夽 This research was undertaken by L. Johnson as the dissertation towards the award of M.Sc. in Physiotherapy. Dr. J. Selfe assisted in data analysis and writing of this paper. ∗ Corresponding author. Tel.:+01274-236587. E-mail address: [email protected] (L. Johnson).

concern has been expressed that it may not be responsive to small clinically significant changes in mobility status [2,6]. The Modified Rivermead Mobility Index was subsequently developed from the RMI to provide a more responsive measure of mobility [7]. The MRMI uses a six-point ordinal scoring system to record whether activities can be achieved with the help of two people, one person, supervision, an aid or independently. The items include turning over, moving from a lying to sitting position, sitting balance, standing up from sitting, standing, transfers, walking and stair-climbing. In a previous study of hospitalised stroke patients the MRMI demonstrated high inter-rater reliability (ICC = 0.98) and was more responsive to change in mobility status

0031-9406/$ – see front matter © 2004 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.physio.2004.01.004

L. Johnson, J. Selfe / Physiotherapy 90 (2004) 132–138

133

than the original RMI. The MRMI also demonstrated high internal consistency (Cronbach’s alpha = 0.93) [8]. This indicated that items within the scale were closely related and therefore appeared to measure a similar construct; mobility [9]. The internal consistency of individual items was not reported. The Motor Assessment Scale (MAS) is a measure of motor impairment and mobility in stroke patients [10]. The MAS uses a seven-point ordinal scale to measure five mobility-related activities that are similar to the MRMI activities; rolling from supine to side-lying, rising from supine to a sitting position, balanced sitting, standing up from a sitting position and walking. Three additional items measure upper limb impairment and function. The MAS has documented validity and reliability [10–13]. Its internal consistency has not been reported. One limitation of the MAS is that it only measures stair-mobility in the highest ranked sub-category of the walking item. In order to score, patients have to be able to go up and down four steps three times with no rail and within 35 s. Consequently the MAS lacks clinical sensitivity when measuring stair-mobility in patients following stroke. Reported times for administering the MAS vary between 15 and 36 min [10,13]. Before the MRMI can be recommended for use in clinical practice or research the degree to which it measures characteristics associated with mobility should be established [14]. Scores on the MRMI should therefore relate to those of a previously validated measure of the same construct (concurrent validity) [15,16]. This is important if accurate conclusions are to be drawn from clinical interventions [17]. It is also important for a scale to be relatively quick to administer if it is to gain widespread clinical use [18]. The main purpose of this study was to establish whether the MRMI is a valid measure of mobility in the acute/sub-acute stroke phase. This was achieved by investigating the concurrent validity of the MRMI with the Motor Assessment Scale (MAS). This aimed to establish whether the MRMI could be used as an alternative to the MAS when measuring mobility. The MAS was selected as the ‘gold standard’ because of its documented validity and reliability and because it assesses similar components of mobility to the MRMI. The study also investigated the internal consistency and test administration time of the two scales.

simple instructions. Subjects meeting these criteria were excluded if they had severe cognitive problems, were registered blind or suffered musculo-skeletal pain that limited mobility. All subjects gave written informed consent. The Local Research Ethics Committee approved the study.

Method

Analysis

Subjects

Data were analysed using SPSS version 11.00 for Windows (SPSS Inc, Chicago, IL) Excel ’98 Microsoft Office (Microsoft Corporation, Redmond, WA). Only the mobility-related items of the MAS were compared in the study of concurrent validity because the MRMI does not include upper limb items. Total scores on the MRMI and MAS (mobility-related items) were 40 and 30 points,

Suitable patients were over 21 years old and had suffered a stroke resulting in residual hemiplegia within the past 3 months. Diagnosis was confirmed by computerised tomography. Patients also needed to be undergoing physiotherapy, medically stable and able to follow

Raters A researcher (L.J.) scored the MRMI and two physiotherapists scored the MAS. Different test administrators were used in order to eliminate the possibility of knowledge of scores on one scale biasing scores on the second scale. The researcher had 6 years experience in stroke rehabilitation, one therapist had 4 years and the other 6 months experience. The two more experienced raters were familiar with the MAS the third was not. All had regularly used the MRMI for at least 6 months. Procedure All stroke patients admitted to the medical and elderly wards of a General Hospital in West Yorkshire between November 2001 and January 2002 were screened for eligibility to participate in the study by the researcher (L.J.). Prior to the main investigation five subjects with varied levels of mobility gave written informed consent to participate in a pilot study. This aimed to standardise testing procedures and maximise inter-rater reliability between test administrators. The test administrators studied the MRMI and MAS guidelines and practised scoring the MAS with between two and four patients prior to the pilot study. The researcher administered the MAS and MRMI whilst all three raters independently scored subjects’ performance. In the main investigation each subject was assessed using the two scales. The MRMI was administered by one of the two physiotherapists and the MAS by the researcher. Assessors were blind to each others scores and were not involved in the treatment of subjects. For each subject the tests were administered separately but on the same day and in the same location before any physiotherapy intervention was carried out. Order of test administration was alternated between subjects to control for practice and fatigue effects. At least 1 hour elapsed between tests. Test administration was timed.

134

L. Johnson, J. Selfe / Physiotherapy 90 (2004) 132–138

respectively. To establish whether there was a consistent bias between the two scales the raw scores from each subject on each scale were plotted. Since the scales use different ranges consistency of differences between scores in the lower half of the scales and scores in the upper half of the scales were then visually analysed. Further comparisons were made by transforming raw scores into percentages of their maximum possible scores. Percentage scores on the two scales were then analysed using the Bland–Altman method of agreement [19]. In this method the difference between paired percentage scores is plotted against the mean of paired measurements. This determined the limits of agreement between total scores on the MRMI and MAS. In the absence of a ‘true measure’ of mobility the mean of the paired measurements indicated the best estimate of the true value [19]. All items within a multi-item unidimensional scale should measure the same construct (internal consistency). Internal consistency of the MRMI and the MAS was analysed by computing Cronbach’s alpha [9]. For each scale the item-total correlation and alpha if each item was deleted were calculated. This indicated whether each scale as a whole and each item within each scale measured a similar construct [9,20]. Cronbach’s alpha values between 0.80 and 0.90 are considered very good [9,21].

Results Pilot study There was never more than two-point disagreement between total scores assigned by the raters to each subject on the MRMI or one point on the MAS (mobility-related items). Complete agreement occurred on 90 (75%) of 120 comparisons between individual items on the MRMI and 58 (77%) of 75 comparisons on the MAS. Inter-rater reliability was considered sufficiently high to progress to the main study.

Table 1 Reasons for exclusion from study Reason for exclusion

Frequency

Died between consenting & assessment Discharged before test administered Refused Unable to follow instructions No longer receiving physiotherapy Registered blind Musculo-skeletal pain limited mobility Medically unwell Bilateral symptoms Total

2 3 2 8 2 1 1 2 2 23

The difference in raw scores on the MRMI and the MAS are shown in Fig. 1. All subjects scored higher on the MRMI than the MAS. The differences in scores were similar for those subjects scoring in the lower half of the range (approximately five points) and similar for those subjects scoring in the upper half of the range (approximately 10 points). When the scores were transformed into percentage of maximum possible score 23 subjects (90%) scored higher and two (7%) scored lower on the MRMI compared to the MAS. One subject scored equally on both scales. The differences between percentages of maximum possible scores on the MAS and MRMI for the study sample were normally distributed. It was therefore appropriate to analyse the limits of agreement between percentage scores on the two scales as interval data [22,23]. Fig. 2 displays the limits of agreement between percentages of total scores on the MAS and MRMI. The mean difference between scores on the MAS and MRMI was 7% (S.D. 4.8%). This difference equates to a mean of 3 points more on the MRMI (S.D. 2 points). Ninety-five percent of differences in scores lie within two standard deviations of the mean. Ninety-five percent of MRMI raw scores were therefore between seven points above and one point below the MAS scores. The magnitude of the difference between percentages of paired measurements did not appear to systematically vary between low, middle or high scores.

Main study Internal consistency Forty-nine stroke patients were admitted to the hospital during the study period. Twenty-three patients were ineligible. The remaining 26 subjects participated in the study. Table 1 displays the reasons for exclusion from the study. Thirteen men and 13 women participated. Their mean age was 77 years (S.D. 9, range 45–88). Subjects were on average 29 days post-stroke (S.D. 18, range 7–83). Twenty-four subjects had suffered a cerebral infarct, two a cerebral haemorrhage. Concurrent validity Table 2 summarises the MAS and MRMI scores of the study sample.

Cronbach’s alpha for the MRMI was 0.949 and for the MAS (items 1–8) 0.953. Table 3 displays corrected item-total correlations and Cronbach’s alpha if each Table 2 MAS (mobility items) and MRMI scores Score

MAS (items 1–5)

MRMI

Median Semi-interquartile range Range Mean score (%)

16.5 9.1

26.5 11.8

2–26 48 (S.D. 29, range 7–87)

4–36 55 (S.D. 28, range 10–90)

L. Johnson, J. Selfe / Physiotherapy 90 (2004) 132–138

135

40

Score on MRMI and MAS

30

20

10 MRMI total score (Upper point) MAS total score (Lower point) 0 1

3

5

7

9

11 13 15 17 19 21 23 25

Case Number Fig. 1. Comparison of MRMI and MAS (mobility items) scores.

item was removed from the analysis. On each scale the lowest item-total correlation was for the sitting balance item (MRMI α = 0.304, MAS α = 0.545). If these items were deleted α increased to 0.966 and 0.961, respectively.

Test administration time On average the MAS took 16 min and 39 s to administer (S.D. 3 min and 30 s, range 6–20 min). The MRMI took 16 min and 32 s (S.D. 4 min and 8 s, range 5–20 min).

Difference between percent of paired measurements (MAS minus MRMI)

10

5 + 2 SD

0 0

10

20

30

40

50

60

70

-5

-10

-15 – 2 SD

-20

Mean of paired measurements Fig. 2. Limits of agreement between percentages of total scores on the MAS (items 1–5) and MRMI.

80

90

100

136

L. Johnson, J. Selfe / Physiotherapy 90 (2004) 132–138

Table 3 Cronbach’s alpha coefficients for MRMI and MAS scales MRMI scale item

Corrected item-total correlation

Alpha if item deleted

MAS scale item

Corrected item-total correlation

Alpha if item deleted

Turning over Lying to sitting Sitting balance Sitting to standing Standing Transfers Walking indoors Stairs

0.872 0.809 0.304 0.943 0.912 0.959 0.936 0.743

0.940 0.942 0.966 0.933 0.936 0.931 0.934 0.946

Supine to side-lying Supine to sitting Balanced sitting Sitting to standing Walking Upper arm function Hand movements Advanced hand activities

0.892 0.870 0.545 0.878 0.881 0.893 0.889 0.854

0.942 0.943 0.961 0.943 0.949 0.942 0.942 0.945

Discussion Concurrent validity The main purpose of the study was to establish whether the MRMI demonstrated concurrent validity so that it could be used as an alternative measure of mobility to the MAS. Exact agreement between paired scores was not anticipated because the scales did not utilise the same scoring range nor did they measure exactly the same mobility-related activities. It was therefore necessary to establish how closely scores on the two scales related to each other and whether there was a consistent bias between scores. A mean difference of three points (out of 40) may be small enough to be clinically acceptable and allow therapists to select either of these outcome measures when measuring the average effectiveness of physiotherapeutic interventions in a group of patients. The decision on which scale to use may be based on a number of factors such as patient population clinician preference and clinical environment. The higher average percentage scores on the MRMI compared to the MAS could indicate that subjects had greater levels of independent mobility despite motor impairments. This is because the MRMI is a measure of a patient’s ability to perform an activity but does not measure quality of performance [8]. Higher scores were awarded if an activity was performed independently in any manner. In contrast in order to score highly on the MAS subjects needed to perform tasks in a specified time and use normal movement patterns [10]. This possible explanation of the average higher scores on the MRMI challenges the assumption commonly held by physiotherapists that quality of movement translates directly into greater functional ability [24]. If this were the case then scores on the two scales might be expected to approximate more closely. An alternative explanation of the higher average scores on the MRMI is that the MRMI items were easier to perform than the corresponding MAS items. For example the MRMI sit to stand item only required the subject to stand up once. The MAS required subjects to stand up and sit down three times in 10 s. Subjects therefore needed to have a higher level of mobility to gain a comparable numerical score on the corresponding MAS item. Additionally stair-climbing is

a separate item within the MRMI whilst on the MAS stairs are only included as the highest level of mobility within the walking item where patients are required to negotiate stairs without a rail within a set time. The MRMI items may therefore have a lower ‘ceiling effect’ than the MAS whereby subjects gain a maximum score and the MRMI item would be unable to detect further improvement over time [20]. This supports the use of the MRMI in the earlier stages of rehabilitation and the use of the MAS in the later stages or with subjects with milder impairments. Clinicians may select the MRMI in the acute phase when a ‘ceiling effect’ is less likely. It could also provide objective information to other members of the inter-disciplinary team regarding for example the amount of assistance required when a patient is getting in/out of bed or walking. In contrast during the later stages of rehabilitation or in mildly impaired stroke patients the MAS might be preferable to measure higher levels of mobility or if objective measurement of ‘quality’ of movement is required. Internal consistency A range of items needs to be assessed within a scale in order to acquire a comprehensive picture of the construct of interest [25]. Internal consistency is therefore important in multiple-item scales because it indicates whether all items are measuring a similar construct [9,20]. The MRMI and the MAS possessed high internal consistency overall. Individual items on both scales also demonstrated high internal consistency (except the sitting balance items). Sitting balance may therefore measure a different construct or be an imperfect measure of the mobility. Test administration time The MRMI and the MAS both took an average of 17 min to administer. This included administration of the upper limb items on the MAS. Both scales were sufficiently quick to administer to have clinically utility. The results of the present study question the opinion that the MAS might be too time-consuming for routine clinical use [7]. The MRMI took slightly longer than the 10–15 min stated in an earlier study [8]. The MAS administration time was similar to that

L. Johnson, J. Selfe / Physiotherapy 90 (2004) 132–138

reported by Carr et al. in 1985 [10] (15 min) but less than the 36 min reported by Malouin et al. [13]. The discrepancy in times between this and the present study is not easily explained. In the present study the median score on the MAS was 16.5. It is possible that the MAS may take longer to administer with patients who score higher on the scale because more sub-items may need to be tested. Limitations The researcher (L.J.) was experienced in stroke rehabilitation and familiar with the MAS. Less experienced physiotherapists might have taken longer to administer the MAS. Different therapists scoring the MRMI and the MAS could have introduced inter-rater variability however utilising different raters eliminated any possibility of knowledge of scores on one scale biasing scores on the second scale. This highlights some of the conflicts when attempting to minimise confounding variables in research. Also there are some limitations to using a percentage of the maximal possible score to compare limits of agreement and investigate whether there was a consistent bias between two scales that had different ranges of raw scores. Whilst percentage differences were clinically less meaningful they did provide an additional method of comparing the MRMI and MAS. Raw scores were also analysed for transparency. Another limitation of the study was the small convenience sample used that might not reflect the stroke population from which the sample was selected. Only two physiotherapists scored the MRMI. A largest sample of therapists with varied experience may have produced different results. Patients with chronic stroke and moderate/severe aphasia or cognitive impairment were excluded from the study. The results are therefore not generalisable to these sub-sets of stroke patients. Further investigation of the concurrent validity of the MRMI in other settings and with different stroke populations is required. The responsiveness of the MRMI versus the original RMI and MAS in measuring change in mobility status also warrants investigation. Conclusions The study established that scores on the MRMI were on average three points higher than scores on the MAS. The mean difference may be considered small enough to allow clinicians to select either of these outcome measures. This decision may be based on a number of factors such as severity of impairment, time since stroke, clinical environment and clinician preference. The MRMI and MAS possess high internal consistency. Individual items within each scale therefore appear to measure a similar construct (except sitting balance). The MRMI and MAS allow therapists to document mobility status relatively quickly. This is an important factor in whether an outcome measure will gain widespread use.

137

Key messages • The mean difference of three points higher on the MRMI than the MAS may be considered small enough to allow clinicians to select either of these outcome measures. • Internal consistency of the scales was high (except sitting balance). • The MRMI and MAS both took an average of 17 min to administer. Acknowledgements Thanks are due to Dr. Kerry Kirk who acted as supervisor during the dissertation and the physiotherapists at Airedale General Hospital who assisted in data collection. This research was partly funded by the Hospital Savings Association and the clinical interest group AGILE. References [1] Chartered Society Of Physiotherapy. Core standards of physiotherapy practice. London: CSP; 2000. [2] Collen FM, Wade GF, Robb, Bradshaw CM. The Rivermead Mobility Index: a further development of the Rivermead Motor Assessment. Int Disabilities Studies 1991;13:50–4. [3] Hsieh CL, Hsueh IP, Mao HF. Validity and responsiveness of the Rivermead Mobility Index in stroke patients. Scand J Rehabil Med 2000;32:140–2. [4] Rossier P, Wade DT. Validity and reliability comparison of 4 mobility measures in patients presenting with neurologic impairment. Arch Phys Med Rehabil 2001;82:9–13. [5] Vaney C, Blaurodk H, Meisels C. Assessing mobility in multiple sclerosis using the Rivermead Mobility Index and gait speed. Clin Rehabil 1996;10:216–26. [6] Wright J, Cross J, Lamb S. Physiotherapy outcome measures for rehabilitation of elderly people: responsiveness to change of the Rivermead Mobility Index and Barthel Index. Physiotherapy 1998;84(5):216–21. [7] Lennon S, Hastings M. Key physiotherapy indicators for quality of stroke care. Physiotherapy 1996;82(12):655–64. [8] Lennon S, Johnson L. The Modified Rivermead Mobility Index: validity and reliability. Disability Rehabil 2000;22(18):833–9. [9] DeVellis RF. Scale development: theory and applications. London: Sage Publications; 1991. p. 22–41, 82–8. [10] Carr R, Shepherd R, Nordholm L, Lynne D. Investigation of a new motor assessment scale for stroke patients. Phys Therapy 1985;65(2):175–80. [11] Loewen SC, Anderson BA. Reliability of the Modified Motor Assessment Scale and the Barthel Index. Phys Therapy 1988;68(7):1077– 81. [12] Poole JL, Whitney SL. Motor Assessment Scale for Stroke Patients: concurrent validity and inter rater reliability. Arch Phys Med Rehabil 1988;69:197–295. [13] Malouin F, Pichare L, Bonneau C, Durand A, Corriveau D. Evaluating motor recovery early after stroke: Comparison of the Fugl–Meyer Assessment and the Motor Assessment Scale. Arch Phys Med Rehabil 1994;75:1206–12. [14] Means KM, Rodell DE, O’Sullinan PS. Use of an obstacle course to assess balance and mobility in the elderly, a validation study. Am J Phys Med Rehabil 1996;75(2):88–95.

138

L. Johnson, J. Selfe / Physiotherapy 90 (2004) 132–138

[15] Durward B, Baer G, Rowe P. Functional human movement, measurement and analysis. Oxford: Butterworth Heinmann; 1999. [16] Hsieh CL, Hsueh IP, Chiang F, Lin P. Inter-rater reliability and validity of the Action Research arm test in stroke patients. Age Ageing 1998;27:107–13. [17] Greenfield ML, Kuhn JE, Wostys EM. A statistics primer, validity and reliability. Am J Sports Med 1998;26(3):483–5. [18] Lord SE, Halligan PW, Wade DT. Visual gait analysis: the development of a clinical assessment and scale. Clin Rehabil 1998;12:107– 19. [19] Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1(8476): 307–10. [20] Sim J, Wright C. Research in health care, concepts, designs and methods. Stanley Thornes Gloster; 2000.

[21] Daley K, Mayo N, Danys I, Cabot R, Wood-Dauphinee S. The stroke rehabilitation assessment of movement (STREAM): refining and validating the content. Physiotherapy Can 1997;49(4):269– 78. [22] Gaito J, Yokubynas R. An empirical basis for the statement that measurement scale properties are irrelevant in statistical analysis. Bull Psychlnomic Soc 1986;6:449–50. [23] Davison ML, Sharma AR. ANOVA and ANCOVA of pre- and post-test ordinal data. Psychometrika 1994;59:593–600. [24] Davidson I, Waters K. Physiotherapists working with stroke patients: a national survey. Physiotherapy 2000;86:69–80. [25] Biemans MA, Dekker J, Van Der Woude LH. The internal consistency and validity of the self-assessment parkinson’s disease disability scale. Clin Rehabil 2001;15:221–8.