J CliaEphkmiol Vol. 44, No. 1, pp. 21-28,1991
0895-4356/91 $3.00 + 0.00 Copyright 0
Printed in Great Britain. All rights reserved
1991Pergamon Pressplc
STROKE REHABILITATION: AUSTRALIAN PATIENT PROFILE AND FUNCTIONAL OUTCOME SURYA
SHAH, FRANK VANCLAY and BETTY COOPER
OccupationalTherapy, University of Queensland,Qld 4072, Australia (Received in revised form
16 July 1990)
Abstract-A prospective, multi-institutional, population based study identified 1274 non-surgical stroke admissions to all hospitals in a major Australian city during 1984. The demographic and diagnostic profile and the nature of functional recovery of all 258 first stroke survivors who were referred for inpatient rehabilitation are presented. The median duration of rehabilitation stay was 49 days. The mean functional independence score, as measured on a modified Barthel Index at admission’was 44, compared with 78 on discharge, a mean improvement of 34. Stair climbing had the lowest mean value on admission (12), while bowel control had the lowest residual deficit on discharge (95). The stroke study group was representative of the unimpaired aged population in all respects except ethnic&y, where differences are attributed to age. The variables identified as significant are; side and severity of paralysis, age and sex, marital status and ethnicity. Stroke rehabilitation outcome was not influenced by etiology, site of lesion, arterial distribution, occupation or education. Stroke rehabilitation Functional recovery Cerebrovascular accident Australia
ADL
Outcome measure
To date, there has been no published information available on the characteristics of stroke patients in Australia who undergo inpatient rehabilitation. Thus, there is no standard against which one can compare patient outcome following comprehensive rehabilitation, nor does one know how these patients compare with older unimpaired persons in the community. This information is vital in developing guidelines for appropriate intervention. This prospective, multi-institutional study addresses many of the fundamental issues outlined in the previous studies [S-lo] and compares profiles of first stroke survivors who were referred for inpatient rehabilitation with the older unimpaired population. The demographic and diagnostic variables associated with functional status are identified by examining differences between function at commencement
INTRODUCTION To identify patients likely to benefit most from
inpatient rehabilitation, a profile of diagnostic and demographic variables which have a significant correlation with functional recovery after stroke is required to evaluate the potential for recovery [l-3]. While the aim of rehabilitation services is to achieve the best possible improvement, regardless of whether the individual patient’s episode was a first or recurrent one, multiple recurrence of stroke is the single best negative predictor of functional outcome [4,5]. The high degree of confounding potentially associated with multiple episodes, and the small number admitted to inpatient comprehensive rehabilitation suggests that in the analysis of the recovery patterns of patients with stroke, this biased group should be excluded [6,7]. 21
22
SURYA
!hAH
of and discharge from comprehensive rehabilitation. METHOD
The study methodology was a prospective pre-test/post-test design based on the population of all patients with first stroke referred for inpatient comprehensive rehabilitation during 1984. The patients were drawn from the population primarily resident in the Brisbane Statistical Division (BSD), Australia, which numbered slightly over 1 million in 1984. All seven public hospitals’ Ethics Committees approved the project and participated in our study. Informed consent to participate in the study was obtained from each patient, or his/her guardian. Since the information on cerebrovascular accidents is practically non-existent in Australia, as a first step towards rehabilitation of stroke study, we examined the acute hospital admissions for the cerebrovascular accidents. However, the overall figures for all hospital admissions that were related to cerebrovascular disease (International Classification of Diseases, ICD-9 codes 430-436) are presented briefly to give a sense of the total workload that was a result of cases of arterial disease that involved the brain. There were 2676 patients admitted with a provisional diagnosis of cerebrovascular disease (ICD-9 codes 430436). From these 2676 patients, 1342 patients identified as cerebrovascular accidents were: 201 intracerebral haemorrhages (ICD 431); 873 occlusions of the cerebral arteries by thrombus or embolus (ICD 434~this 873 group also includes precerebral artery occlusions which produced neurological deficits; and 268 acute but illdefined cerebrovascular accident cases (ICD 436). The remaining 1334 patients relate to the cerebrovascular diseases that were not classified as “common strokes” and excluded from this group of cases of cerebrovascular accident were: 112 subarachnoid haemorrhages (ICD 430); 646 transient ischaemic attacks (ICD 435); and 536 occlusions and stenoses of pre-cerebral arteries, including carotid, basilar and vertebral arteries (ICD 433). The 536 patients in this ICD 433 classification included only those precerebral artery occlusions which did not produce any neurological deficit, and all acute hospital admissions were pre-arranged for either endarterectomy, or, for exploratory neuroradiological investigations.
et al.
There were 21 incomplete or untraceable records, and 19 had a provisional but unsubstantiated cerebrovascular accident. Of the 1342 stroke patients, 265 (20%) died within the first 2 weeks, 98 (7%) with persistent altered conscious state were transferred directly to extended-care facilities and 359 (27%) who were independent in ADL were sent directly home. We did not consider the 68 (5%) with neuro-surgical intervention or the 291 (22%) with a second or multiple stroke in order to minimize potential confounding [14]. The remaining 261 (19%) who had survived their first stroke and who were referred for inpatient rehabilitation following their discharge from acute-care hospitals were the subjects for the study. Three patients died during rehabilitation, leaving a total of 258 survivors of first stroke. Discharge from rehabilitation was determined when the team considered the patient had achieved maximum benefit [l l-141. Rather than evaluate patients at some more remote point in time, patients were assessed on discharge, so as to allow outcome to be measured following medical rehabilitation potentially uncontaminated by social, cultural, family and other influences. Since documentation for rehabilitation evaluation and discharge was subjective and differed from hospital to hospital and from therapist to therapist, the patients were assessed independently, both on admission to rehabilitation and on discharge, on a standardized system. Functional recovery was measured as improvement in the ability to perform ADL and was operationalized by the Barthel Index (BI). The BI measures the individual’s performance on 10 ADL functions for a total of 100 points. This empirically derived scale has proven inter-rater reliability, test-retest reliability and validity. Kendall’s coefficient of concordance W is highly significant (p < 0.001) with an overall reliability of 0.93. A small variation in the BI to increase its sensitivity to change was incorporated for this study and gave a higher content reliability than the original BI of 0.90 [ll]. Occupation refers to the occupational status of the patient immediately prior to retirement. In the case of female patients, the occupation of the spouse and in case of retirement over 1 decade, a son’s or daughter’s occupation was also recorded. For this study previous occupation classification was based on the Australian Bureau of Statistics 1981 Census. However, this classification was changed for the
Stroke Rehabilitation Profile
1986 Census and as the 1981 occupational classification for the over 55 referent group was unavailable, the comparison was made with the 1986 data. Since the majority of epidemiological and health studies in Australia are based at city or state level, the socioeconomic status for the analysis of health and disease correlates was compared with the total workforce occupational prestige data presented by Nixon and Pearn [17J This data uses a 4-point scale of occupational status and prestige in Australia. The diagnosis of stroke was based on neuroradiologic investigations where available. Of the 1342 patients with strokes who were admitted to the hospitals, reports of neuro-radiological investigations were available for 543 (40%) patients. Of the first stroke admissions to inpatient rehabilitation, 52% of patients had received neuro-radiological investigations, nearly 60% of these patients underwent a CT scan, while EEG and/or angiogram were used for the remaining cases. When these investigations were not recommended, specific clinical attribution to an exact diagnostic type was obtained from the notes and an interview with the admitting/ treating physician, who were MDs, neurologists and/or geriatricians. The data was coded and entered on a DEC-10 mainframe computer using the 1022 data base system. Analysis was performed using SPSS-X on a VAX 8550 mainframe. Descriptive statistics and initial correlations were obtained, conventional x2, t and F tests were performed to assess the independence of two variables. Comparison of differences between the study group and the population were tested by Kendall’s goodness of fit test. Efficiency was measured as the mean improvement in Barthel
23
score per day, while effectiveness was measured as actual improvement over potential improvement and the result percentaged to reflect the proportion of potential improvement actually obtained during rehabilitation [12. 15, 161. RESULTS Length
of stays
Onset of stroke to acute hospitalization ranged from 0 to 68 days. Seventy-two percent of individuals were admitted immediately following onset, but the skewness of the data resulted in a mean of 6.5 days being recorded. The median was 0 days. Rehabilitation followed an acute hospitalization period which ranged from 1-59 days, with a median of 8 days. Thus, the time from onset of stroke to commencement of rehabilitation ranged from 1 to 79 (mean + SD 17 + 14.2; median 13) days. Inpatient comprehensive rehabilitation time ranged from 6 to 276 (mean & SD 61 f 45.1; median 49) days [ 1l-l 31. Functional
improvement
The mean Barthel score at admission to rehabilitation was 44 compared with 78 on discharge, while the mean level of improvement was 34. The specific task performance data are presented in Table 1 and were calculated by removing the weighing associated with each item, and expressing, the value out of 100. Stair climbing had the lowest mean value at admission (12) and remained the item that had the greatest remaining deficit [58). The item with the lowest residual deficit was bowel control (95), although this had the highest initial value (87). The proportional improvement on each activity,
Table 1. Mean improvement on the individual Barthel items Initial score
Bathing (5) Feeding (10) On-Off Toilet (10) Bladder Control (10) Bowel Control (10) Dressing (10) Stair Climbing (10) Ambulation (15) Chair/Bed Transfer (15)
Improvement
(n = 256)*
Item
Hygiene(5)$
Discharge score
60 31 64 43 70 87 27 10 21 42
87 74 88 79 89 95 70 58 67 79
27 43 24 36 19 8 43 46 46 37
Effectiveness W)t 67.5 62.2 66.7 63.2 63.3 61.5 58.9 52.2 58.2 63.8
*Two patients excluded because of an initial score. of 100. tNumber in parentheses represents the weighting of that item in the total Barthel score. .jEffectiveness is expressed as actual improvement over potential improvement. The result obtained is percentaged to reflect the proportion of potential improvement actually obtained during rehabilitation.
24
SURYA
SHAHet a!.
(N=258) +
Maximum
Function
Score
Poaalble
Indlvldual Item* liyplsneo*
4
~&&-J--I
Dress (‘O)
ti
Stairs
0
10
20
30
40
Functional
50
SO
70
SO
so
(10)
100
SCOW at referral.
Improvement
Ez2a
Residual
0 ??
loss of function
Amount
In bracketa
o, that
item In the total
at discharge
represents Barthal
tha
walghtlne
*core
Fig. 1. Mean improvements in the BI tasks.
i.e. the effectiveness of rehabilitation for that item, was, relatively consistent across all items, although stair climbing was the lowest (52%), with hygiene having the highest improvement (68%). A diagram of total functional capability highlighting the changes in functional independence due to the stroke episode is presented by representing the level of disability on admission to rehabilitation, the level of improvement achieved and the residual disability as part of an area representing total functional independence (Fig. 1). The total effectiveness score, indicating the mean percentage achievement of maximum possible improvement in functional recovery, was 66% (median 76%), with 22% of the patients attaining a score of 100, or full recovery. The mean efficiency was 0.845, indicating that patients increased their Barthel score by an average of 0.845 Barthel units per day during rehabilitation. Efficiency ranged from -0.88 to 4.57 units, and had a median value of 0.60. Negative scores were recorded by patients who regressed.
Demographic variables Age and sex. The youngest patient with stroke referred for comprehensive rehabilitation was 27, although those under 50 years comprised only 6% of the study group. The mean of 68 years for male survivors is significantly lower than 71.6 years for females (p < 0.01). Males comprise a greater proportion (56%) of stroke rehabilitation candidates. Of stroke admissions 75 years and older, 53% were female, with 39% of females for the under 75 years group. Female patients had significantly lower initial Bar-the1 scores (41 vs 47, p < 0.05) as well as significantly lower discharge scores (73 vs 81, p < 0.01). However, females gained just as much as males in treatment. Age was significantly correlated with initial Barthel score at -0.18 (p < 0.005) and with discharge Barthel score at -0.28 (p c O.OOl), indicating the higher the age, the lower the Barthel score. Occupation. Due to the age group of the stroke population, 87% were pensioners, while 12% were in full- or part-time employment. Two individuals regarded themselves as being unemployed. The study group has a much higher proportion in receipt of pensions than the same age group for the state of Queensland (Table 2). Initial and discharge Barthel scores were significantly different, however 3 of the 5 categories had only 2 cases each, making comparison invalid. As far as can be compared on the basis of the information available in Australia, stroke patients approximated a representative sample of occupational classification for that age referent group (Table 3). On the basis of the state-specific occupational prestige data there is no significant difference (p > 0.05) between the stroke study group’s socioeconomic status and the Queensland distribution [ 171.
Table 2. Proportion of pensioners Stroke study group % Age group (years) 50-59 6064 6569 70-74 75+
Males (n = 145)
Females (n = 113)
Total Queensland (%)*
31 81 96 100 100
64 90 100 96 100
15 31 49 65 82
*Proportion of pensioners in Queensland, based on Department of Social Security figures for April 1988, and Australian Bureau of Statistics population estimates for March 1988.
Stroke Rehabilitation Profile Table 3. Previous occupational classification Stroke study group* 14 3 10 26 16 t; lOO$
1986 RSDt (> 54 years)
1986 Census classitications Managers Professional Para professionals Tradespeople Clerical sales Plant machine Labower
182 176 34 I 14
(n = 252)
100x
*In 6 cases occupation was not available. ?I986 Census data on previous occupational~ ci,assification of persons aged 55 years and older, residing in the Brisbane Statistical Division (BSD). / $Totals do not necessarily add up to 100 due to accumulated rounding errors.
Marital status. The marital status of the stroke study group was representative of the unimpaired population aged 55 years and older, indicating that marital status was not associated with referral for rehabilitation after stroke (Table 4). There was no significant difference between the various categories of marital status for the initial Barthel score, but there was a significant difference for discharge score (p -c 0.05), with single people having a mean discharge Barthel score of 95, a mean for married people of 78, 79 for those divorced or separated and 75 for widowed people. Marital status was highly related to age and it is apparent that the relationship between marital status and the discharge score is confounded by age. Ethnicity. Of the stroke study group, 84% were Australian, 10% were British, with the remaining 6% from a variety of backgrounds.
25
The 1986 Census results for those aged 55 years and over reveals 74% Australian born, 13% born in the U.K. and 14% born in other countries. Although this represents a significant dif?erence (p < 0.05), it is likely that this is due to age differences in the two groups. There was no significant difference between the Australians and the overseas-born new Australians on initial Barthel score, however, the overseas-born new Australians had a significantly lower discharge Barthel score (70 vs 80, p < 0.05) (Table 5). data for Education. While comparative educational status for people 55 years and over is not available for Australia, there is no reason to suggest that the stroke study group was not representative of the aged population. There was no significant difference for initial and discharge Barthel scores across the seven educational groups. Handedness. Right-handed people, as selfstated, comprised 96% of the study group. This is not regarded as high for this age group and reduces confounding in analysis. Diagnostic variables Etiology. Of the 1342 in the original stroke group, 725 (54%) were thrombotic in origin, 148 (11%) embolic, 201 (15%) haemorrhagic and 268 (20%) were acute but ill-defined ICD code 436. Of the 261 patients that comprised the study group, 148 (57%) were thrombotic, 41 (16%) embolic, 41 (16%) haemorrhagic and 3 1 (12%) were acute but ill-defined in origin. The distribution of specific etiological types was not significantly different (p > 0.05) between the
Table 4. Marital status Stroke study group Single
1986 BSD* (> 54 years)
Initial Barthel score (I) (mean)
Discharge Barthel score (D) (mean)
7
(6:796
95 56193
29/t(Vt
6
Married
56
61
[5$90
4;;1
38/69]
Divorced
6
I
[363;6
44:977
321771
[?;-
4&
Widowed Total
32
25
100
:%I) (n
F 1.7 df 3,254 Signif. 0.164
421721 2.7 3,254 0.046
Kendall’s goodness of fit test statistic (x2, 3 df) = 6.47, p > O.OS$ *Brisbane Statistical Division 1986 Census. t[ ] Figures in brackets refer to initial and discharge Barthel scores for the marital status subgroups for x55, 55-14 and 215 years. $Based on the actual observed frequencies and expected frequencies using the distribution for the Brisbane Statistical Division 1986.
26
SURYA
SHAHet al.
Table 5. Ethnicitv and Barthel scores Australian citizens, I/D* scores
British migrants, I/D scores
Other migrants, I/D scores
45180 53195 47181 40173
38166 -41;70 35162
42177 47180 43183 38171
All age groups 0.25). Arterial distribution. With 77% of the patients being classified as having middle cerebral artery lesions, there was little differentiation in the arterial distribution of stroke. Side and severity of paralysis. Side of paralysis/side of site of lesion and severity of paralysis are both strongly associated with initial and discharge Barthel scores (Table 8). Table 9 provides the comparison of the site of
and migrant
lesion-and the side and severity of paralysis. It is difficult to provide a significance value as 16 of the 28 cells have a value of < 5. Associated illnesses. Four percent of patients did not have any associated illness. Hypertension was the illness most commonly associated with hemiplegia (70%), followed by myocardial infarction (29%), chronic heart disease (29%), congestive cardiac failure (14%), peripheral vascular disease (8%) and fractured femur (4%). Those with hypertension had a significantly lower initial Barthel score (42 vs 52, p < 0.05), however it was not associated with the discharge Barthel score. Discharge disposition
Following comprehensive rehabilitation, 67% of patients were discharged to their pre-admission residence with a mean rehabilitation stay of 55 days, and mean discharge Barthel score of 88. Twenty-five percent were transferred to nursing home, with 74 rehabilitation days and a mean Barthel score of 51. The remaining 8% were discharged to the care of relatives after a rehabilitation stay of 69 days, with a mean discharge Barthel score of 82. DISCUSSION
Patient profile and functional recovery patterns of all 258 first stroke survivors referred Table 7. Site of lesion
Table 6. Etiology
% Thrombosis Embolism Haemorrhage Unclassified Total
57 16 16 12 100* F (n = 258) df Signif.
Initial Discharge Barthel Barthel score (I) score (D) (mean) (mean) 43 50 39 50
78 81 76 79
2.0 3,254 0.118
0.3 3,253 0.857
*Total does not necessarily add up to 100 due to accumulated rounding errors.
Initial Discharge Barthel Barthel score (I) score (D) (mean) (mean)
% Cortex Internal capsule Brain stem Diffuse Basal ganglia Other Unclassified
:: 9
1 4 1 12
Total (n :58)
$ Signif.
45 38 58 32 37 48 48
79 74 84 70 61 92 79
2.4 6,251 0.027
1.3 6,25 1 0.255
Stroke Rehabilitation Profile Table 8. Side and severity of paralysis
% Right plegia Right paresis Left plegia Left paresis
(n ?56)*r;f
67 89 73 89
30 63 37 56
E 18
Total
The high male rehabilitation rate is reflected in the crude cerebrovascular disease admission rate of 2.05 for males and 1.67 for females per 1000 head of population. Analysis of age and sex reveals that females have higher rates of admission at lower age groups, while males have higher rates of admission at older age groups [14]. However, the greater proportion of females in the population for older groups means that overall the number of admissions is higher for females, with 53% being 75 years and over. Although females have significantly lower initial Barthel scores, sex was not a predictor of discharge Barthel score [13], nor did it influence efficiency, effectiveness or duration of rehabilitation [12]. Age was significantly correlated with the initial and discharge Barthel scores and was a predictor of the discharge Barthel score, with each 10 years of age lowering the discharge score by 6 Barthel units [13]. Age also significantly reduced efficiency by 0.1 Barthel units per day, for every 10 years of age. Age reduced effectiveness by 9 Barthel units for every 10 years. Despite reducing efficiency, age reduced duration of rehabilitation by 7 days for every 10 years of age, because of the lower effectiveness that is likely to be obtained [13]. This study considered demographic and diagnostic variables comprehensively, as compared to other studies [5,8,9, 15,181. Educational level, occupational status and classification, socioeconomic status, etiology, arterial distribution and site of lesion were not related to the initial and discharge Barthel scores, nor to efficiency, effectiveness or duration of rehabilitation. These also did not influence the prediction of outcome [12, 131. Despite significance in Barthel scores, marital status and hypertension did not influence efficiency, effectiveness or initial impairment. While other associated illnesses did not influence initial and discharge Barthel scores, the illnesses with significant effect on stroke outcome were myocardial
Initial Discharge Barthel Barthel score (I) score (D) (mean) (mean)
25
33.8 3,252 Signif. 0.000
21
12.0 3,252 0.000
1
A Newman_Keuls posthocanalysis showed that!+&pite the F value, severity is important but side is not. . :I *In 2 cases data was not available.
for inpatient comprehensive rehabilitation are reported. The functional outcome findings are presented on discharge from inpatient rehabilitation so as to minimize social, cultural and other factors which may influence outcome score; and for first stroke patients only, as opposed to other studies [3,8, 15, 19-211, which have assessed outcome findings at a fixed interval, and have included second and multiple stroke episodes. Unlike other studies, this study recorded onset to acute admission and acute admission to commencement of rehabilitation time separately as delay at both stages contribute to inferior rehabilitation outcome [8,9,13]. Similar to other studies, the mean functional improvement in Barthel score was 34, with significant improvement in all items, Stair climbing, ambulation, dressing and bathing had lower mean values initially and remained the areas of greatest residual deficit [3,7, l&16, 181.We feel that the small variation in the BI incorporated for this study to increase sensitivity on individual items will assist therapist to target intervention. Although many factors influence discharge disposition, total functional independence scores were valuable in determining this, with patients achieving a mean discharge score of 88 returning home, while those with a mean of 51 requiring nursing home or extended care facilities.
Table 9. Lesion site vs side and severity of paralysis Right plegia
Right paresis
Left plegia
Cortex Internal capsule Brain stem Diffuse Basal ganglia Other Unclassified
29 19 5 1 4 5
27 8 9 :
53 23 1 2 1
8
9
9
52 21 9 1 4 1 12
%
25
22
35
18
100
Site
Left paresis 23 s5 1
%
,
Sfmv.4 SHAHet al.
28
infarction and diabetes, both reducing the likely outcome by 6 points on the Barthel scale. Diabetes reduced rehabilitation stay by 16 days, while peripheral vascular disease increased rehabilitation stay by 22 days [12. 131. CONCLUSION
This study identified important demographic, diagnostic and attribute variables which do and do not correlate with stroke rehabilitation functional outcome. In the light of current literature, the significant variables which contribute to efficiency, effectiveness, duration of rehabilitation and prediction of functional outcome were further highlighted. Comparison of stroke patients with the unimpaired aged population was made to facilitate discharge planning and community integration. Acknowledgemenf-This project was supported by a grant from the Mayne Bequest Fund.
REFERENCES 1. Christie D. Prevalence of stroke and its sequelae. Med J Aust 1981; 2: 182-184. 2. Partridge CJ, Johnston M. Edwards S. Recovery from physical disability after stroke: normal patterns as a basis for evaluation. Lancet 1987; 1: 373-375. 3. Wade DT, Skilbeck CE, Hewer RL. Predicting Barthel ADL score at 6 months after an acute stroke. Arch Phys Med Rehabil 1983; 64: 24-28. 4. Gerston G. Stroke rehabilitation. In: Licht S, Ed. Rehabilitation Potential. New Haven, Conn.: Licht; 1975. 5. Boureston NV. Predictors of long-term recovery in cerebrovascular disease. Arch Phys Med Rehabil 1967; 48: 418419.
6. Novack TA, Satterfield WT, Lyons K, Kolski G. Hackmeyer L, Connor M. Stroke onset and rehabilitation: time lag as a factor in treatment outcome. Arch Phys Mad Rebabil 1984; 65: 316319. 7. Mills VM, DiGenio M. Functional differences in patients with left or right cerebrovascular accidents. Pbys Ther 1983; 63: 481-488. 8. Dombovy ML, Sandok BA, Basford JR. Rehabilitation for stroke: a review. Stroke 1986; 17: 363-369. 9. Jongbloed L. Prediction of function after stroke: a critical review. Stroke 1986; 17: 765-776. 10. Gresham GE. Stroke outcome research. Stroke 1986; 17: 358-360. 11. Shah S, Vanclay F, Cooper B. Improving the sensitivity of the Barthel Index. J Clin Epidemiol 1989; 42: 703-709. 12. Shah S, Vanclay F, Cooper B. Efficiency, effectiveness and duration of stroke rehabilitation. Stroke 1990; 21: 241-246. 13. Shah S, Vanclay F, Cooper B. Predicting discharge status at commencement of stroke rehabilitation. Stroke 1989; 20: 766-769. 14. Shah S, Bain C. Admissions, patterns of utilization and disposition of acute strokes in Brisbane hospitals. Med J Aast 1989; 150: 256-260. 15. Heinemann AW. Roth EJ. Cichowski K. Betts HB. Multivariate analysis of improvement in outcome following stroke rehabilitation. Arch Neural 1987; 44: 1167-1172. 16. Granger CV, Hamilton BB, Gresham GE. The stroke rehabilitation outcome study-Part 1: general description. Arch Phys Med Rehabil 1988; 69: 506509. 17. Nixon J, Pearn J. Australian state specific reference data by socioeconomic status for the analysis of health and disease correlates. Med J Aust 1984; 140: 51-54. 18. Granger CV, Dewis LS, Peters NC, Sherwood CC, Barrett JE. Stroke rehabilitation: analysis of repeated Barthel Index measures. Arch Phys Med Rehabil 1979; 60: 14-17. 19. Lehman JF, DeLateur BJ, Fowler RS. Stroke rehabilitation: outcome and prediction. Arch Phys Med Rehabil 1975; 56: 383-389. 20. Waylonis GW, Keith MW, Aseff JN. Stroke rehabilitation in midwestem countrv. Arch Phvs Med Rehabil 1973; 54: 151-155. 2 1. Lind K. A synthesis of studies on stroke rehabilitation. J Chron Dis 1982; 35: 133-149.
0895-4356/91 $3.00 + 0.00 Copyright 0
Printed in Great Britain. All rights reserved
1991Pergamon Pressplc
STROKE REHABILITATION: AUSTRALIAN PATIENT PROFILE AND FUNCTIONAL OUTCOME SURYA
SHAH, FRANK VANCLAY and BETTY COOPER
OccupationalTherapy, University of Queensland,Qld 4072, Australia (Received in revised form
16 July 1990)
Abstract-A prospective, multi-institutional, population based study identified 1274 non-surgical stroke admissions to all hospitals in a major Australian city during 1984. The demographic and diagnostic profile and the nature of functional recovery of all 258 first stroke survivors who were referred for inpatient rehabilitation are presented. The median duration of rehabilitation stay was 49 days. The mean functional independence score, as measured on a modified Barthel Index at admission’was 44, compared with 78 on discharge, a mean improvement of 34. Stair climbing had the lowest mean value on admission (12), while bowel control had the lowest residual deficit on discharge (95). The stroke study group was representative of the unimpaired aged population in all respects except ethnic&y, where differences are attributed to age. The variables identified as significant are; side and severity of paralysis, age and sex, marital status and ethnicity. Stroke rehabilitation outcome was not influenced by etiology, site of lesion, arterial distribution, occupation or education. Stroke rehabilitation Functional recovery Cerebrovascular accident Australia
ADL
Outcome measure
To date, there has been no published information available on the characteristics of stroke patients in Australia who undergo inpatient rehabilitation. Thus, there is no standard against which one can compare patient outcome following comprehensive rehabilitation, nor does one know how these patients compare with older unimpaired persons in the community. This information is vital in developing guidelines for appropriate intervention. This prospective, multi-institutional study addresses many of the fundamental issues outlined in the previous studies [S-lo] and compares profiles of first stroke survivors who were referred for inpatient rehabilitation with the older unimpaired population. The demographic and diagnostic variables associated with functional status are identified by examining differences between function at commencement
INTRODUCTION To identify patients likely to benefit most from
inpatient rehabilitation, a profile of diagnostic and demographic variables which have a significant correlation with functional recovery after stroke is required to evaluate the potential for recovery [l-3]. While the aim of rehabilitation services is to achieve the best possible improvement, regardless of whether the individual patient’s episode was a first or recurrent one, multiple recurrence of stroke is the single best negative predictor of functional outcome [4,5]. The high degree of confounding potentially associated with multiple episodes, and the small number admitted to inpatient comprehensive rehabilitation suggests that in the analysis of the recovery patterns of patients with stroke, this biased group should be excluded [6,7]. 21
22
SURYA
!hAH
of and discharge from comprehensive rehabilitation. METHOD
The study methodology was a prospective pre-test/post-test design based on the population of all patients with first stroke referred for inpatient comprehensive rehabilitation during 1984. The patients were drawn from the population primarily resident in the Brisbane Statistical Division (BSD), Australia, which numbered slightly over 1 million in 1984. All seven public hospitals’ Ethics Committees approved the project and participated in our study. Informed consent to participate in the study was obtained from each patient, or his/her guardian. Since the information on cerebrovascular accidents is practically non-existent in Australia, as a first step towards rehabilitation of stroke study, we examined the acute hospital admissions for the cerebrovascular accidents. However, the overall figures for all hospital admissions that were related to cerebrovascular disease (International Classification of Diseases, ICD-9 codes 430-436) are presented briefly to give a sense of the total workload that was a result of cases of arterial disease that involved the brain. There were 2676 patients admitted with a provisional diagnosis of cerebrovascular disease (ICD-9 codes 430436). From these 2676 patients, 1342 patients identified as cerebrovascular accidents were: 201 intracerebral haemorrhages (ICD 431); 873 occlusions of the cerebral arteries by thrombus or embolus (ICD 434~this 873 group also includes precerebral artery occlusions which produced neurological deficits; and 268 acute but illdefined cerebrovascular accident cases (ICD 436). The remaining 1334 patients relate to the cerebrovascular diseases that were not classified as “common strokes” and excluded from this group of cases of cerebrovascular accident were: 112 subarachnoid haemorrhages (ICD 430); 646 transient ischaemic attacks (ICD 435); and 536 occlusions and stenoses of pre-cerebral arteries, including carotid, basilar and vertebral arteries (ICD 433). The 536 patients in this ICD 433 classification included only those precerebral artery occlusions which did not produce any neurological deficit, and all acute hospital admissions were pre-arranged for either endarterectomy, or, for exploratory neuroradiological investigations.
et al.
There were 21 incomplete or untraceable records, and 19 had a provisional but unsubstantiated cerebrovascular accident. Of the 1342 stroke patients, 265 (20%) died within the first 2 weeks, 98 (7%) with persistent altered conscious state were transferred directly to extended-care facilities and 359 (27%) who were independent in ADL were sent directly home. We did not consider the 68 (5%) with neuro-surgical intervention or the 291 (22%) with a second or multiple stroke in order to minimize potential confounding [14]. The remaining 261 (19%) who had survived their first stroke and who were referred for inpatient rehabilitation following their discharge from acute-care hospitals were the subjects for the study. Three patients died during rehabilitation, leaving a total of 258 survivors of first stroke. Discharge from rehabilitation was determined when the team considered the patient had achieved maximum benefit [l l-141. Rather than evaluate patients at some more remote point in time, patients were assessed on discharge, so as to allow outcome to be measured following medical rehabilitation potentially uncontaminated by social, cultural, family and other influences. Since documentation for rehabilitation evaluation and discharge was subjective and differed from hospital to hospital and from therapist to therapist, the patients were assessed independently, both on admission to rehabilitation and on discharge, on a standardized system. Functional recovery was measured as improvement in the ability to perform ADL and was operationalized by the Barthel Index (BI). The BI measures the individual’s performance on 10 ADL functions for a total of 100 points. This empirically derived scale has proven inter-rater reliability, test-retest reliability and validity. Kendall’s coefficient of concordance W is highly significant (p < 0.001) with an overall reliability of 0.93. A small variation in the BI to increase its sensitivity to change was incorporated for this study and gave a higher content reliability than the original BI of 0.90 [ll]. Occupation refers to the occupational status of the patient immediately prior to retirement. In the case of female patients, the occupation of the spouse and in case of retirement over 1 decade, a son’s or daughter’s occupation was also recorded. For this study previous occupation classification was based on the Australian Bureau of Statistics 1981 Census. However, this classification was changed for the
Stroke Rehabilitation Profile
1986 Census and as the 1981 occupational classification for the over 55 referent group was unavailable, the comparison was made with the 1986 data. Since the majority of epidemiological and health studies in Australia are based at city or state level, the socioeconomic status for the analysis of health and disease correlates was compared with the total workforce occupational prestige data presented by Nixon and Pearn [17J This data uses a 4-point scale of occupational status and prestige in Australia. The diagnosis of stroke was based on neuroradiologic investigations where available. Of the 1342 patients with strokes who were admitted to the hospitals, reports of neuro-radiological investigations were available for 543 (40%) patients. Of the first stroke admissions to inpatient rehabilitation, 52% of patients had received neuro-radiological investigations, nearly 60% of these patients underwent a CT scan, while EEG and/or angiogram were used for the remaining cases. When these investigations were not recommended, specific clinical attribution to an exact diagnostic type was obtained from the notes and an interview with the admitting/ treating physician, who were MDs, neurologists and/or geriatricians. The data was coded and entered on a DEC-10 mainframe computer using the 1022 data base system. Analysis was performed using SPSS-X on a VAX 8550 mainframe. Descriptive statistics and initial correlations were obtained, conventional x2, t and F tests were performed to assess the independence of two variables. Comparison of differences between the study group and the population were tested by Kendall’s goodness of fit test. Efficiency was measured as the mean improvement in Barthel
23
score per day, while effectiveness was measured as actual improvement over potential improvement and the result percentaged to reflect the proportion of potential improvement actually obtained during rehabilitation [12. 15, 161. RESULTS Length
of stays
Onset of stroke to acute hospitalization ranged from 0 to 68 days. Seventy-two percent of individuals were admitted immediately following onset, but the skewness of the data resulted in a mean of 6.5 days being recorded. The median was 0 days. Rehabilitation followed an acute hospitalization period which ranged from 1-59 days, with a median of 8 days. Thus, the time from onset of stroke to commencement of rehabilitation ranged from 1 to 79 (mean + SD 17 + 14.2; median 13) days. Inpatient comprehensive rehabilitation time ranged from 6 to 276 (mean & SD 61 f 45.1; median 49) days [ 1l-l 31. Functional
improvement
The mean Barthel score at admission to rehabilitation was 44 compared with 78 on discharge, while the mean level of improvement was 34. The specific task performance data are presented in Table 1 and were calculated by removing the weighing associated with each item, and expressing, the value out of 100. Stair climbing had the lowest mean value at admission (12) and remained the item that had the greatest remaining deficit [58). The item with the lowest residual deficit was bowel control (95), although this had the highest initial value (87). The proportional improvement on each activity,
Table 1. Mean improvement on the individual Barthel items Initial score
Bathing (5) Feeding (10) On-Off Toilet (10) Bladder Control (10) Bowel Control (10) Dressing (10) Stair Climbing (10) Ambulation (15) Chair/Bed Transfer (15)
Improvement
(n = 256)*
Item
Hygiene(5)$
Discharge score
60 31 64 43 70 87 27 10 21 42
87 74 88 79 89 95 70 58 67 79
27 43 24 36 19 8 43 46 46 37
Effectiveness W)t 67.5 62.2 66.7 63.2 63.3 61.5 58.9 52.2 58.2 63.8
*Two patients excluded because of an initial score. of 100. tNumber in parentheses represents the weighting of that item in the total Barthel score. .jEffectiveness is expressed as actual improvement over potential improvement. The result obtained is percentaged to reflect the proportion of potential improvement actually obtained during rehabilitation.
24
SURYA
SHAHet a!.
(N=258) +
Maximum
Function
Score
Poaalble
Indlvldual Item* liyplsneo*
4
~&&-J--I
Dress (‘O)
ti
Stairs
0
10
20
30
40
Functional
50
SO
70
SO
so
(10)
100
SCOW at referral.
Improvement
Ez2a
Residual
0 ??
loss of function
Amount
In bracketa
o, that
item In the total
at discharge
represents Barthal
tha
walghtlne
*core
Fig. 1. Mean improvements in the BI tasks.
i.e. the effectiveness of rehabilitation for that item, was, relatively consistent across all items, although stair climbing was the lowest (52%), with hygiene having the highest improvement (68%). A diagram of total functional capability highlighting the changes in functional independence due to the stroke episode is presented by representing the level of disability on admission to rehabilitation, the level of improvement achieved and the residual disability as part of an area representing total functional independence (Fig. 1). The total effectiveness score, indicating the mean percentage achievement of maximum possible improvement in functional recovery, was 66% (median 76%), with 22% of the patients attaining a score of 100, or full recovery. The mean efficiency was 0.845, indicating that patients increased their Barthel score by an average of 0.845 Barthel units per day during rehabilitation. Efficiency ranged from -0.88 to 4.57 units, and had a median value of 0.60. Negative scores were recorded by patients who regressed.
Demographic variables Age and sex. The youngest patient with stroke referred for comprehensive rehabilitation was 27, although those under 50 years comprised only 6% of the study group. The mean of 68 years for male survivors is significantly lower than 71.6 years for females (p < 0.01). Males comprise a greater proportion (56%) of stroke rehabilitation candidates. Of stroke admissions 75 years and older, 53% were female, with 39% of females for the under 75 years group. Female patients had significantly lower initial Bar-the1 scores (41 vs 47, p < 0.05) as well as significantly lower discharge scores (73 vs 81, p < 0.01). However, females gained just as much as males in treatment. Age was significantly correlated with initial Barthel score at -0.18 (p < 0.005) and with discharge Barthel score at -0.28 (p c O.OOl), indicating the higher the age, the lower the Barthel score. Occupation. Due to the age group of the stroke population, 87% were pensioners, while 12% were in full- or part-time employment. Two individuals regarded themselves as being unemployed. The study group has a much higher proportion in receipt of pensions than the same age group for the state of Queensland (Table 2). Initial and discharge Barthel scores were significantly different, however 3 of the 5 categories had only 2 cases each, making comparison invalid. As far as can be compared on the basis of the information available in Australia, stroke patients approximated a representative sample of occupational classification for that age referent group (Table 3). On the basis of the state-specific occupational prestige data there is no significant difference (p > 0.05) between the stroke study group’s socioeconomic status and the Queensland distribution [ 171.
Table 2. Proportion of pensioners Stroke study group % Age group (years) 50-59 6064 6569 70-74 75+
Males (n = 145)
Females (n = 113)
Total Queensland (%)*
31 81 96 100 100
64 90 100 96 100
15 31 49 65 82
*Proportion of pensioners in Queensland, based on Department of Social Security figures for April 1988, and Australian Bureau of Statistics population estimates for March 1988.
Stroke Rehabilitation Profile Table 3. Previous occupational classification Stroke study group* 14 3 10 26 16 t; lOO$
1986 RSDt (> 54 years)
1986 Census classitications Managers Professional Para professionals Tradespeople Clerical sales Plant machine Labower
182 176 34 I 14
(n = 252)
100x
*In 6 cases occupation was not available. ?I986 Census data on previous occupational~ ci,assification of persons aged 55 years and older, residing in the Brisbane Statistical Division (BSD). / $Totals do not necessarily add up to 100 due to accumulated rounding errors.
Marital status. The marital status of the stroke study group was representative of the unimpaired population aged 55 years and older, indicating that marital status was not associated with referral for rehabilitation after stroke (Table 4). There was no significant difference between the various categories of marital status for the initial Barthel score, but there was a significant difference for discharge score (p -c 0.05), with single people having a mean discharge Barthel score of 95, a mean for married people of 78, 79 for those divorced or separated and 75 for widowed people. Marital status was highly related to age and it is apparent that the relationship between marital status and the discharge score is confounded by age. Ethnicity. Of the stroke study group, 84% were Australian, 10% were British, with the remaining 6% from a variety of backgrounds.
25
The 1986 Census results for those aged 55 years and over reveals 74% Australian born, 13% born in the U.K. and 14% born in other countries. Although this represents a significant dif?erence (p < 0.05), it is likely that this is due to age differences in the two groups. There was no significant difference between the Australians and the overseas-born new Australians on initial Barthel score, however, the overseas-born new Australians had a significantly lower discharge Barthel score (70 vs 80, p < 0.05) (Table 5). data for Education. While comparative educational status for people 55 years and over is not available for Australia, there is no reason to suggest that the stroke study group was not representative of the aged population. There was no significant difference for initial and discharge Barthel scores across the seven educational groups. Handedness. Right-handed people, as selfstated, comprised 96% of the study group. This is not regarded as high for this age group and reduces confounding in analysis. Diagnostic variables Etiology. Of the 1342 in the original stroke group, 725 (54%) were thrombotic in origin, 148 (11%) embolic, 201 (15%) haemorrhagic and 268 (20%) were acute but ill-defined ICD code 436. Of the 261 patients that comprised the study group, 148 (57%) were thrombotic, 41 (16%) embolic, 41 (16%) haemorrhagic and 3 1 (12%) were acute but ill-defined in origin. The distribution of specific etiological types was not significantly different (p > 0.05) between the
Table 4. Marital status Stroke study group Single
1986 BSD* (> 54 years)
Initial Barthel score (I) (mean)
Discharge Barthel score (D) (mean)
7
(6:796
95 56193
29/t(Vt
6
Married
56
61
[5$90
4;;1
38/69]
Divorced
6
I
[363;6
44:977
321771
[?;-
4&
Widowed Total
32
25
100
:%I) (n
F 1.7 df 3,254 Signif. 0.164
421721 2.7 3,254 0.046
Kendall’s goodness of fit test statistic (x2, 3 df) = 6.47, p > O.OS$ *Brisbane Statistical Division 1986 Census. t[ ] Figures in brackets refer to initial and discharge Barthel scores for the marital status subgroups for x55, 55-14 and 215 years. $Based on the actual observed frequencies and expected frequencies using the distribution for the Brisbane Statistical Division 1986.
26
SURYA
SHAHet al.
Table 5. Ethnicitv and Barthel scores Australian citizens, I/D* scores
British migrants, I/D scores
Other migrants, I/D scores
45180 53195 47181 40173
38166 -41;70 35162
42177 47180 43183 38171
All age groups 0.25). Arterial distribution. With 77% of the patients being classified as having middle cerebral artery lesions, there was little differentiation in the arterial distribution of stroke. Side and severity of paralysis. Side of paralysis/side of site of lesion and severity of paralysis are both strongly associated with initial and discharge Barthel scores (Table 8). Table 9 provides the comparison of the site of
and migrant
lesion-and the side and severity of paralysis. It is difficult to provide a significance value as 16 of the 28 cells have a value of < 5. Associated illnesses. Four percent of patients did not have any associated illness. Hypertension was the illness most commonly associated with hemiplegia (70%), followed by myocardial infarction (29%), chronic heart disease (29%), congestive cardiac failure (14%), peripheral vascular disease (8%) and fractured femur (4%). Those with hypertension had a significantly lower initial Barthel score (42 vs 52, p < 0.05), however it was not associated with the discharge Barthel score. Discharge disposition
Following comprehensive rehabilitation, 67% of patients were discharged to their pre-admission residence with a mean rehabilitation stay of 55 days, and mean discharge Barthel score of 88. Twenty-five percent were transferred to nursing home, with 74 rehabilitation days and a mean Barthel score of 51. The remaining 8% were discharged to the care of relatives after a rehabilitation stay of 69 days, with a mean discharge Barthel score of 82. DISCUSSION
Patient profile and functional recovery patterns of all 258 first stroke survivors referred Table 7. Site of lesion
Table 6. Etiology
% Thrombosis Embolism Haemorrhage Unclassified Total
57 16 16 12 100* F (n = 258) df Signif.
Initial Discharge Barthel Barthel score (I) score (D) (mean) (mean) 43 50 39 50
78 81 76 79
2.0 3,254 0.118
0.3 3,253 0.857
*Total does not necessarily add up to 100 due to accumulated rounding errors.
Initial Discharge Barthel Barthel score (I) score (D) (mean) (mean)
% Cortex Internal capsule Brain stem Diffuse Basal ganglia Other Unclassified
:: 9
1 4 1 12
Total (n :58)
$ Signif.
45 38 58 32 37 48 48
79 74 84 70 61 92 79
2.4 6,251 0.027
1.3 6,25 1 0.255
Stroke Rehabilitation Profile Table 8. Side and severity of paralysis
% Right plegia Right paresis Left plegia Left paresis
(n ?56)*r;f
67 89 73 89
30 63 37 56
E 18
Total
The high male rehabilitation rate is reflected in the crude cerebrovascular disease admission rate of 2.05 for males and 1.67 for females per 1000 head of population. Analysis of age and sex reveals that females have higher rates of admission at lower age groups, while males have higher rates of admission at older age groups [14]. However, the greater proportion of females in the population for older groups means that overall the number of admissions is higher for females, with 53% being 75 years and over. Although females have significantly lower initial Barthel scores, sex was not a predictor of discharge Barthel score [13], nor did it influence efficiency, effectiveness or duration of rehabilitation [12]. Age was significantly correlated with the initial and discharge Barthel scores and was a predictor of the discharge Barthel score, with each 10 years of age lowering the discharge score by 6 Barthel units [13]. Age also significantly reduced efficiency by 0.1 Barthel units per day, for every 10 years of age. Age reduced effectiveness by 9 Barthel units for every 10 years. Despite reducing efficiency, age reduced duration of rehabilitation by 7 days for every 10 years of age, because of the lower effectiveness that is likely to be obtained [13]. This study considered demographic and diagnostic variables comprehensively, as compared to other studies [5,8,9, 15,181. Educational level, occupational status and classification, socioeconomic status, etiology, arterial distribution and site of lesion were not related to the initial and discharge Barthel scores, nor to efficiency, effectiveness or duration of rehabilitation. These also did not influence the prediction of outcome [12, 131. Despite significance in Barthel scores, marital status and hypertension did not influence efficiency, effectiveness or initial impairment. While other associated illnesses did not influence initial and discharge Barthel scores, the illnesses with significant effect on stroke outcome were myocardial
Initial Discharge Barthel Barthel score (I) score (D) (mean) (mean)
25
33.8 3,252 Signif. 0.000
21
12.0 3,252 0.000
1
A Newman_Keuls posthocanalysis showed that!+&pite the F value, severity is important but side is not. . :I *In 2 cases data was not available.
for inpatient comprehensive rehabilitation are reported. The functional outcome findings are presented on discharge from inpatient rehabilitation so as to minimize social, cultural and other factors which may influence outcome score; and for first stroke patients only, as opposed to other studies [3,8, 15, 19-211, which have assessed outcome findings at a fixed interval, and have included second and multiple stroke episodes. Unlike other studies, this study recorded onset to acute admission and acute admission to commencement of rehabilitation time separately as delay at both stages contribute to inferior rehabilitation outcome [8,9,13]. Similar to other studies, the mean functional improvement in Barthel score was 34, with significant improvement in all items, Stair climbing, ambulation, dressing and bathing had lower mean values initially and remained the areas of greatest residual deficit [3,7, l&16, 181.We feel that the small variation in the BI incorporated for this study to increase sensitivity on individual items will assist therapist to target intervention. Although many factors influence discharge disposition, total functional independence scores were valuable in determining this, with patients achieving a mean discharge score of 88 returning home, while those with a mean of 51 requiring nursing home or extended care facilities.
Table 9. Lesion site vs side and severity of paralysis Right plegia
Right paresis
Left plegia
Cortex Internal capsule Brain stem Diffuse Basal ganglia Other Unclassified
29 19 5 1 4 5
27 8 9 :
53 23 1 2 1
8
9
9
52 21 9 1 4 1 12
%
25
22
35
18
100
Site
Left paresis 23 s5 1
%
,
Sfmv.4 SHAHet al.
28
infarction and diabetes, both reducing the likely outcome by 6 points on the Barthel scale. Diabetes reduced rehabilitation stay by 16 days, while peripheral vascular disease increased rehabilitation stay by 22 days [12. 131. CONCLUSION
This study identified important demographic, diagnostic and attribute variables which do and do not correlate with stroke rehabilitation functional outcome. In the light of current literature, the significant variables which contribute to efficiency, effectiveness, duration of rehabilitation and prediction of functional outcome were further highlighted. Comparison of stroke patients with the unimpaired aged population was made to facilitate discharge planning and community integration. Acknowledgemenf-This project was supported by a grant from the Mayne Bequest Fund.
REFERENCES 1. Christie D. Prevalence of stroke and its sequelae. Med J Aust 1981; 2: 182-184. 2. Partridge CJ, Johnston M. Edwards S. Recovery from physical disability after stroke: normal patterns as a basis for evaluation. Lancet 1987; 1: 373-375. 3. Wade DT, Skilbeck CE, Hewer RL. Predicting Barthel ADL score at 6 months after an acute stroke. Arch Phys Med Rehabil 1983; 64: 24-28. 4. Gerston G. Stroke rehabilitation. In: Licht S, Ed. Rehabilitation Potential. New Haven, Conn.: Licht; 1975. 5. Boureston NV. Predictors of long-term recovery in cerebrovascular disease. Arch Phys Med Rehabil 1967; 48: 418419.
6. Novack TA, Satterfield WT, Lyons K, Kolski G. Hackmeyer L, Connor M. Stroke onset and rehabilitation: time lag as a factor in treatment outcome. Arch Phys Mad Rebabil 1984; 65: 316319. 7. Mills VM, DiGenio M. Functional differences in patients with left or right cerebrovascular accidents. Pbys Ther 1983; 63: 481-488. 8. Dombovy ML, Sandok BA, Basford JR. Rehabilitation for stroke: a review. Stroke 1986; 17: 363-369. 9. Jongbloed L. Prediction of function after stroke: a critical review. Stroke 1986; 17: 765-776. 10. Gresham GE. Stroke outcome research. Stroke 1986; 17: 358-360. 11. Shah S, Vanclay F, Cooper B. Improving the sensitivity of the Barthel Index. J Clin Epidemiol 1989; 42: 703-709. 12. Shah S, Vanclay F, Cooper B. Efficiency, effectiveness and duration of stroke rehabilitation. Stroke 1990; 21: 241-246. 13. Shah S, Vanclay F, Cooper B. Predicting discharge status at commencement of stroke rehabilitation. Stroke 1989; 20: 766-769. 14. Shah S, Bain C. Admissions, patterns of utilization and disposition of acute strokes in Brisbane hospitals. Med J Aast 1989; 150: 256-260. 15. Heinemann AW. Roth EJ. Cichowski K. Betts HB. Multivariate analysis of improvement in outcome following stroke rehabilitation. Arch Neural 1987; 44: 1167-1172. 16. Granger CV, Hamilton BB, Gresham GE. The stroke rehabilitation outcome study-Part 1: general description. Arch Phys Med Rehabil 1988; 69: 506509. 17. Nixon J, Pearn J. Australian state specific reference data by socioeconomic status for the analysis of health and disease correlates. Med J Aust 1984; 140: 51-54. 18. Granger CV, Dewis LS, Peters NC, Sherwood CC, Barrett JE. Stroke rehabilitation: analysis of repeated Barthel Index measures. Arch Phys Med Rehabil 1979; 60: 14-17. 19. Lehman JF, DeLateur BJ, Fowler RS. Stroke rehabilitation: outcome and prediction. Arch Phys Med Rehabil 1975; 56: 383-389. 20. Waylonis GW, Keith MW, Aseff JN. Stroke rehabilitation in midwestem countrv. Arch Phvs Med Rehabil 1973; 54: 151-155. 2 1. Lind K. A synthesis of studies on stroke rehabilitation. J Chron Dis 1982; 35: 133-149.
