1462

Improving Stroke RehabilitationA Controlled Study

Lalit Kalra, PhD, MRCP(UK); Penny Dale, MCSP, SRP; Peter Crome, MD, FRCP

Background and Purpose: Assessment of stroke rehabilitation is complicated by the heterogeneity ofpatients and settings and by difficulties in disentangling effects of organization from effects of types andamounts of treatment input.

Methods: A prospective controlled study was undertaken in 245 stroke patients stratified into threegroups according to prognosis and managed on a stroke rehabilitation unit (n= 124) or general medicalwards (n=121). Patients were randomly allocated to either setting 2 weeks after stroke and were

comparable for baseline characteristics.Results: Patients on general medical wards received more physiotherapy on average (16.2±7.2 versus

14.3±3.2 hours; P<.05) but similar amounts of occupational therapy (9.3±2.8 versus 9.5±3.2 hours)compared with stroke unit patients. More time was spent on individual rehabilitation on the stroke unitcompared with general wards (P<.001). Functional abilities at discharge, destination of discharge, andlength of hospital stay in patients with good prognosis were comparable in both settings. Patients withpoor prognosis managed on general wards showed higher mortality (P<.05) and longer hospital stay(123.2±48.2 versus 52.3±19.8 days; P<.OO1), but functional abilities at discharge in survivors werecomparable with those of stroke unit patients. Patients with intermediate prognosis had significantlybetter outcome on the stroke unit, with more patients being discharged home (75% versus 52%; P<.OO1),shorter average length of hospital stay (48.7± 17.2 versus 104.6±28.6 days; P<.001), and better functionalabilities at discharge (P<.05).

Conclusions: Stroke units improve outcome and reduce hospital stay without increasing therapy time.Their effectiveness may be enhanced by patient selection. (Stroke. 1993;24:1462-1467.)KEY WoRDs * hospitalization * prognosis * rehabilitation

O rganization of rehabilitation is a key consider-ation in stroke management. Treatment ofstroke on general wards has been criticized

because of poor coordination between disciplines, lackof planning consistent with patient needs or abilities,and breakdown of communication between profession-als, patients, and carers.' Dissatisfaction with standardsof provision on general medical wards has resulted inthe development of more specific strategies in strokemanagement during the last decade.2 Although inten-sive treatment of stroke patients may be beneficial, thebenefits of stroke intensive care units3-6 in reducingmortality and morbidity remain unproven.7.8 There issome evidence that stroke rehabilitation units mayreduce disability and long-term institutionalization, butdespite several studies to evaluate the benefits of suchunits, their effectiveness remains controversial.79 ~14

Several factors contribute to the difficulties in assess-ing effectiveness of stroke rehabilitation units. Measure-ment of differences in stroke rehabilitation is compli-cated because of (1) the heterogeneity of patientcharacteristics, (2) failure to stratify for severity of

Received February 4, 1993; final revision received April 21,1993; accepted April 23, 1993.From the Orpington Stroke Unit, Bromley Hospitals, Bromley

(L.K., P.D.), and the Department of Health Care for the Elderly,King's College School of Medicine, London (L.K., P.C.), UK.Correspondence to Dr L. Kalra, Orpington Hospital, Sevenoaks

Rd, Orpington BR6 9JU, UK.

stroke15 (which determines both prognosis as well as thelevel of resources needed), (3) the variety of settings inwhich stroke is treated, (4) differences in quantity andquality of treatment received by patients, (5) variationin resources allocated to stroke management and theorganization of services, (6) difficulties in disentanglingthe effects of differing service organizations from theeffects of different types and duration of treatmentreceived by patients, and (7) difficulties in assessingobjectively the impact of available services or newdevelopments because of the lack of baseline informa-tion and poor quality of data collected in this field.8The present study is a controlled prospective study

comparing therapy input and outcome in stroke pa-tients, stratified according to expected prognosis, whowere managed either on general wards or a strokerehabilitation unit.

Subjects and MethodsSubjects for the study were recruited from 377 stroke

patients admitted to a general hospital during a periodof 18 months. Stroke was defined as acute onset ofneurological deficit lasting more than 24 hours or lead-ing to death, with no apparent cause other than cere-brovascular disease. Patients with first (83%) as well asrecurrent (17%) strokes were included in the study. Thediagnosis of stroke was based on history and clinicalexamination. Computed tomographic (CT) scanningwas not routinely undertaken except when indicated by

by guest on January 27, 2015http://stroke.ahajournals.org/Downloaded from

Kalra et al Patient Selection in Stroke Unit Rehabilitation

TABLE 1. Indications for ComputedTomographic Scan(1) Doubt about the clinical diagnosis

(a) Unclear history of focal neurological symptoms

(b) Atypical clinical features

(c) Atypical progression of stroke after onset

(2) Patient is <60 years of age with no vascular risk factors

(3) Cerebellar stroke suspected

(4) Subarachnoid hemorrhage suspected(5) Anticoagulation indicated or already being given

(6) Carotid endarterectomy being considered

defined criteria (Table 1). Eighty-two (22%) of the 377stroke patients had CT scans.Because the incidence of stroke increases with age,14

42 (11%) patients with dementia were included in thestudy since this frequently complicates stroke rehabilita-tion in the older age group.16 No attempts were made toclassify dementia patients into multi-infarct or the Alz-heimer type because of difficulties in accurately differ-entiating between the two conditions clinically and theirfrequent association. It was presumed that both groupswould be equally disadvantaged in rehabilitation becauseof diminished learning abilities. Patients with space-occupying lesions, cerebral metastatic disease, low-pres-sure hydrocephalus, congenital malformations, head in-jury, or central nervous system (CNS) infections onclinical or CT evidence were excluded. Although CTscanning was undertaken in all patients in whom theclinical diagnosis of stroke was equivocal, the possiblelimitation of not having 100% CT scanning isacknowledged.

Stroke patients were admitted to general medicalwards during the acute phase of their illness for initialmanagement and stabilization. Of the 377 patientsdiagnosed as having stroke, CT scans demonstratedsubdural hematomas in 2 and brain tumors in 7 patients.Seventy-nine (21%) patients died, and 37 (10%) pa-tients with mild deficits secondary to reversible ischemicneurological disease were discharged within 2 weeks ofadmission.The remaining 252 survivors at 2 weeks were entered

into the study. Details of age, sex, side of stroke, powerin the arm and leg on the affected side (MedicalResearch Council grading),17 hemianopia, dysphasia,dysphagia, sensory deficits, inattention (visual/sensory),continence, mobility,18 and Barthel activities of dailyliving (ADL) scores'9 were recorded. Cognitive statewas assessed using Hodkinson's abbreviated mental testscore, which is a well-validated 10-item test for memoryand orientation.20,21 The test was conducted on the wardwith the patient responding verbally to questions askedby the observer. In the presence of dysphasias, patientswere expected to respond appropriately by speech or

signs to spoken or written answers suggested by theobserver.The 252 patients were standardized for severity of

stroke and expected outcome by stratification intogroups according to prognostic criteria based on clinicalmeasures of impairment.22 Possible scores ranged from

TABLE 2. The Orpington Prognostic ScaleUsed to Categorize Patients in the ThreeGroups Included in the StudyClinical Features Score

Motor deficit in arm

MRC grade

0

0.4

0.8

1.2

1.6

5

4

3

1-2

0

Proprioception (eyes closed)

Locates affected thumb

Accurately

Slight difficulty

Finds thumb via arm

Unable to find thumb

Balance

Walks 10 feet without help

Maintains standing position

Maintains sitting position

No sitting balance

Cognition

Mental test score

10

8-9

5-7

0-4

0

0.4

0.8

1.2

0

0.4

0.8

1.2

0

0.4

0.8

1.2

Total score= 1.6+ motor+ proprioception+ balance+cognition

MRC indicates Medical Research Council.

1.6 (best prognosis) to 6.8 (worst prognosis) (Table 2).Three groups were identified: patients with mild tomoderate deficits showing the best prognosis (prognos-tic score less than 3), patients with moderate to severedeficits in whom prognosis was intermediate (prognosticscore 3 through 5), and patients with severe or verysevere deficits who had poor prognosis (prognostic scoregreater than 5). After stratification, patients were ran-domly allocated to a 13-bed stroke rehabilitation unit orcontinued to be managed on general medical wardsaccording to existing practices.

Despite different settings, all stroke patients receivednursing care, physiotherapy, and occupational therapyappropriate to their disability. Input was also providedby the speech therapists, social workers, and nursinghome placement officer for patients unable to returnhome. Progress, therapy, rehabilitation goals, and dis-charge plans of patients were monitored in multidis-ciplinary meetings in both settings.

Subjects were followed up from entry to the studyuntil discharge from the hospital. Objective assessmentsfor neurological deficit, cognitive function, continence,mobility, and ADL were undertaken at weekly intervals

1463

by guest on January 27, 2015http://stroke.ahajournals.org/Downloaded from

1464 Stroke Vol 24, No 10 October 1993

in both groups. The duration and type of therapy givento individual stroke patients in both groups was re-corded by the physiotherapists and occupational thera-pists working with the patients. The duration of therapywas measured in 30-minute time units of face-to-facecontact with the patient. The amount of time devoted todifferent types of therapy work within each disciplinewas also recorded. Professionals involved in the assess-ment and day-to-day management of these patientswere unaware of their prognostic scores or expectedoutcome.

Primary outcome measures included mortality duringhospital stay, the percentage of patients dischargedhome, the percentage of patients discharged to long-term institutional care, and the length of hospital stay.Because stroke is predominantly a disease of the elderlyin whom destination on discharge may depend onfactors other than stroke,'1123 the Barthel ADL score atdischarge, the change in Barthel ADL score frominclusion in the study to discharge, and the proportionof patients with a Barthel ADL score of greater than 11in each group were also recorded.The sample size was calculated using a comparison

nomogram24 to include the minimum number of pa-tients in each prognostic group to give the study a 90%power at 5% significance level for primary outcomemeasures (destination of discharge, median dischargeBarthel ADL scores, length of hospital stay) based onobservations in previous studies.22,25 Group homogene-ity was analyzed with a x2 test for sex, neurologicaldeficits, dementia, recurrent strokes, and prognosticclassification in each group. Age on admission, motorpower on affected side, and Barthel ADL scores oninitial assessment were analyzed by the Mann-Whitneytest. Mortality, destination of discharge, differences inthe type of therapy received, and the proportion ofpatients with a Barthel ADL score greater than 11 atdischarge in each group were analyzed using the X2 test.Statistical analysis was not undertaken if any cell had 0value or if a value of less than 5 was present in morethan 20% of cells. The length of hospital stay, amount oftherapy received, discharge Barthel scores, and changein Barthel score during rehabilitation were analyzed bythe Mann-Whitney test. The study was approved by theBromley Ethics Committee.

ResultsOf the 252 patients in the study, 126 were treated on

the stroke rehabilitation unit and 126 on general med-ical wards. Seven patients (2 from the stroke rehabili-tation unit and 5 from general medical wards) weretransferred to other hospitals (residing out of district orto be closer to relatives) and hence did not complete thestudy.The baseline characteristics of the 124 patients man-

aged on the stroke unit were comparable to the 121patients treated on general medical wards (Table 3).Patients with an intermediate prognosis formed thelargest group, accounting for nearly 60% of patients inboth settings. The extent of neurological deficit, mobil-ity, and functional abilities at the initial assessment inpatients treated on the stroke rehabilitation unit werecomparable to those treated on general wards.

Patients treated on general medical wards received

on the stroke rehabilitation unit during their hospital stay(Table 4). A greater proportion of physiotherapy input onstroke rehabilitation wards was directed toward individualneeds of patients compared with general medical wards(P<.001). There were no differences in the averageamount of occupational therapy received by patients ineither setting. However, a significantly greater proportion(P<.001) of time on the stroke rehabilitation unit wasspent on specific needs of individual patients comparedwith general medical wards (Table 4).

There were no significant differences in functionalabilities at discharge, destination of discharge, or length ofhospital stay in stroke patients with a good prognosis(prognostic score less than 3) managed in either setting(Table 5). One patient managed on general medical wardsrequired long-term care because of social circumstancesrather than disability (Barthel ADL score of 16).Although a high mortality was seen in severely dis-

abled patients with poor prognosis (prognostic scoregreater than 5) in both settings, a significantly greaternumber of patients died on the general medical wardcompared with the stroke rehabilitation unit (Table 5).Clinical causes of deaths were aspiration pneumonia,pulmonary embolism, recurrent stroke, and unrelatedmyocardial infarction. The functional abilities and thedischarge destination of survivors were, however, com-parable between the two groups (Table 5).The greatest differences between the stroke rehabil-

itation unit and general medical wards were seen instroke patients with intermediate prognosis (prognosticscore 3 through 5). A significantly greater proportion ofpatients managed on the stroke rehabilitation unit weredischarged home compared with those on general med-ical wards. In addition, patients managed on the strokeunit had significantly better functional abilities at dis-charge and a shorter length of hospital stay (Table 5).

DiscussionThis study, undertaken in stroke patients stratified for

neurological deficit and prognosis, demonstrates thatorganized and directed stroke management does lead toa more rapid recovery of function and more rapiddischarge from the hospital without any major increasein time allocated by the therapist. Patients with moder-ately severe deficit and intermediate prognosis appearto benefit most by stroke unit rehabilitation comparedwith those with mild or very severe deficits.

Patients with dementia and recurrent strokes werenot excluded from the study to make the sample morerepresentative of the stroke population. Mortalitywithin the first 2 weeks was high in these groups, withless than 40% of patients in these groups completing thestudy and being included in the analysis (Table 3). Thedistribution of these patients between the stroke unitand general medical wards was comparable (Table 3).The prognostic disadvantage due to dementia or recur-rent strokes is reflected by higher Orpington PrognosticScale scores,22 and inclusion of such patients in thestudy does not compromise its value.Computed tomography was not undertaken routinely

in this study. The value of CT scanning has beeninvestigated previously, with the conclusion that it didnot predict or influence functional outcome in strokepatients.26 28 It may not be possible for every stroke

significantly more physiotherapy on average than patients patient to undergo CT scanning, even in health care

by guest on January 27, 2015http://stroke.ahajournals.org/Downloaded from

Kalra et al Patient Selection in Stroke Unit Rehabilitation 1465

TABLE 3. Baseline Characteristics of Stroke Patients on StrokeRehabilitation Unit and General Medical Wards at Time of InclusionInto StudyClinical Features SU GMW P

No. of patients 124 121 NS

Age, y (mean±SD) 77.8±11.4 78.6±12.2 NS

Sex, % female 56 59 NS

Recurrent strokes 14 1 1 NS

Dementia 7 8 NS

Recurrent stroke and dementia 3 4 NS

Left hemiplegia 59 58 NS

Right hemiplegia 55 51 NS

Brain stem/cerebellar 10 12 NS

Prognostic groups

OPS score <3 31 32 NS

OPS score 3-5 75 71 NS

OPS score >5 18 18 NSMean power in affected arm (triceps)* 2.3±1.7 2.6±1.2 NS

Mean power in affected leg (quadriceps)* 3.1±1.4 3.0±1.6 NS

Sensory loss 16 14 NS

Hemianopia 42 39 NS

Sensory/visual inattention 54 57 NS

Dysphasia 26 23 NS

Dysphagia 11 9 NS

Median FAC score (range) 1 (0-3) 1 (0-3)

Median Barthel ADL score (range) 5 (0-12) 5 (0-14) NS

CT scans 32 29 NS

SU indicates stroke unit; GMW, general medical wards; OPS, Orpington Prognostic Scale (seeTable 2); FAC, Functional Ambulation Categories; ADL, activities of daily living; CT, computedtomography; and NS, not significant.

*Medical Research Council grading for power.

systems such as the British National Health Service. Ithas been suggested that rapid access to expert clinicalevaluation of neurological deficit with urgent access toCT scanning and neurosurgical facilities if required maybe a more appropriate strategy in stroke management.8This approach has been followed in this study withwell-defined criteria for CT scanning after comprehen-sive clinical appraisal of patients.A double-blind study was not possible because of the

logistics of separate wards and the nature of interven-tion. The possibility of bias introduced by observerpreference and staff as well as patient motivation by theobserved positive, or even negative, discrimination byallocation to the stroke unit must be recognized. Thesepitfalls were reduced by "blinding" the nursing and thetherapy staffs to prognostic scores and outcome mea-sures. The broad categories of therapy input weredecided among professionals in advance. Therapists orclinicians involved in management on the stroke unit didnot provide input to general medical wards and hencewere unable to influence therapy input or outcome inthese settings. Finally, ensuring an even mix of patientswith good as well as poor prognostic expectations inboth settings prevented nihilistic or negative attitudes

among staff or patients in either setting, reinforcing thevalidity of the findings of this study.The conflicting results of previous studies on the

benefits of stroke units using similar outcome measuresmay have been due to the type of patients recruited intothese studies. Results of the present study show thatpatients with mild deficits achieve independence inpersonal ADL regardless of their setting, whereas thosewith very severe deficits and poor prognosis do notregain significant basic functional abilities irrespectiveof management on a stroke unit. Although most studiesare controlled for the severity of deficit, there is littleinformation about the actual proportion of patients withdifferent levels of disability.7'9-12 Inclusion of a largeproportion of patients at either end of the spectrumwould minimize differences between stroke units andgeneral wards and may have been responsible for thenegative results in some controlled studies.9"10"12The difference in mortality among patients with a

poor prognosis between the stroke rehabilitation unitand general medical wards was an unexpected finding inthis study (Table 5). The possibility of type I error dueto small numbers cannot be excluded because mortalitywas not taken into consideration in determination of

by guest on January 27, 2015http://stroke.ahajournals.org/Downloaded from

1466 Stroke Vol 24, No 10 October 1993

TABLE 4. Comparison of Therapy Input in Stroke Patients Managed onStroke Rehabilitation Unit With Those Managed on General Medical Wards

SU GMWTherapy Input and Type (n=124) (n=121) P

Physiotherapy

Mean duration per patient, h* 14.3+3.2 16.2-+7.2 <.05

No. of half-hour sessions spent on

Sitting balance (%) 486 (13.7) 580 (14.8)

Standing balance (%) 720 (20.3) 855 (21.8)

Transfers (%) 571 (16.1) 596 (15.2)

Ambulation (%) 734 (20.7) 858 (21.9)

Individual rehabilitationt (%) 1036 (29.2) 1031 (26.3) <.02tOccupational therapy

Mean duration per patient, h* 9.5±3.2 9.3±2.8 NS

No. of half-hour sessions spent on

Personal ADL (%) 1430 (60.7) 1476 (65.6)

Kitchen activities (%) 245 (10.4) 216 (9.6)

Home visits (%) 304 (12.9) 333 (14.8)

Postdischarge follow-up (%) 106 (4.5) 58 (2.6)

Individual rehabilitationt (%) 271 (11.5) 166 (7.4) <.001t

SU indicates stroke unit; GMW, general medical wards; ADL, activities of daily living; and NS, notsignificant.*Time spent in face-to-face activities with patients, excluding administrative time.tindividual rehabilitation: time spent on activities aimed at addressing specific needs of individual

patients (eg, specific transfer/washing/dressing techniques, use of aids) identified by therapist orpatient as contributing significantly to discharge to chosen environment.

fLargest determinant of a significant two-sided x 2 test for independent proportions.

sample size. However, reduction in medium-term mor-

tality on stroke units has been observed elsewhere,7'1achieving statistical significance in only one otherstudy.7 This may possibly be due to better managementof swallowing problems and awareness of deep veinthrombi on stroke units, although no definite conclu-sions can be drawn from this study.

Several factors may have contributed to the improvedoutcome on the stroke rehabilitation unit, and it is difficult

to isolate specific components important to overall results.Despite the general belief that patients on stroke units dobetter because of increased therapy input, this was not thecase in the study. In keeping with another study," resultsshowed that the average duration of therapy input on thestroke unit was less than that on general medical wards.The type of treatment, however, differed in that it wasspecifically matched to individual patient needs. This mayhave contributed significantly to the observed differences

TABLE 5. Outcome According to Prognostic Groups in Patients Treated on Stroke Unit and GeneralMedical Wards

Prognostic Score <3 Prognostic Score 3-5 Prognostic Score >5

SU GMW GMW GMW(n=31) (n=32) SU (n=75) (n=71) SU (n=18) (n=18)

Measure No. % No. % P No. % No. % P No. % No. % PMortality 0 ... 0 ... ... 2 3 3 4 NS 7 39 12 67 <.05

Discharge home 31 100 31 97 NS 56 75 37 52 <.001 3 16 1 6 NSt

Long-term care 0 ... 1 3 ... 17 22 31 44 <.001 8 45 5 23 NS

Discharge BADL >11 31 100 32 100 NS 61 81 42 60 <.05 1 6 0 0 ...

Median discharge BADL* 18 18 NS 15 13 <.05 6 6 NS

Median change in BADL* 12 12 NS 12 8 <.05 4 4 NS

Length of stay, d (mean+SD) 13.2+6.7 14.6±4.2 NS 48.7±+17.2 104.6+28.6 <.001 52.3±+19.8 123.2±48.2 <.001

SU indicates stroke unit; GMW, general medical wards; BADL, Barthel activities of daily living score; and NS, not significant.*Barthel score measured in survivors.tFisher's exact test.

by guest on January 27, 2015http://stroke.ahajournals.org/Downloaded from

Kalra et al Patient Selection in Stroke Unit Rehabilitation 1467

in outcome. Better multidisciplinary coordination withpatients and carers, a positive attitude among nurses,29and their involvement as informal therapists may beadditionally responsible for improved outcome. The psy-chological impact of being on the stroke unit may havealso contributed by boosting patients' morale and motiva-tion to achieve greater functional independence. Theseinteractive effects are highly complex and have not beenassessed in this study.Because stroke is predominantly a disease of advanc-

ing years, assessing outcome of rehabilitation, even inthe short term, presents problems. The number ofpatients discharged home is a simplistic measure anddoes not take into account other factors that mayinfluence discharge. To enable a more accurate evalu-ation of rehabilitation outcome, measures of functionalability have also been included in this study (Table 5).Previous experience has shown that patients with Bar-thel ADL scores greater than 11 require supervision orintermittent help for walking and self-care and can bemaintained at home.30 The percentage of patientsachieving this functional level was greater than thosedischarged home in both settings and appears to be abetter measure of stroke rehabilitation. However, theeffectiveness of stroke rehabilitation cannot be assessedby one measure in isolation, and it would be moreappropriate to use a combination of measures whenevaluating the effectiveness of strategies in strokemanagement.

This study has demonstrated that patient selection cansignificantly influence the effectiveness of stroke units.While it is inconceivable to deny any stroke patientadequate treatment solely on the basis of severity ofdisability, there may be advantages both for the patientand the hospital service in directing stroke unit resourcestoward patients most likely to benefit from such input.31With the exception of a small group of patients with poorprognosis in whom mortality may be reduced, rehabilita-tion on stroke units would be of little benefit to strokesurvivors who would do well or those who would do badlywhatever their setting or therapy input. It appears that themost appropriate patients for stroke unit rehabilitationare those with moderately severe deficits and an interme-diate prognosis. Identification of this subgroup of patientscan be facilitated by incorporating major determinants ofoutcome into a well-defined set of simple but objectiveclinical criteria that can be applied in day-to-day hospitalwork and by professionals who may not be medicallytrained. Several sets of criteria have been suggested.22'32-35The criteria used in this study incorporate measures ofpower, balance, proprioception, and cognitive functionand are recommended for wider use.22230

Well-defined criteria of patient selection can improvethe effectiveness of stroke rehabilitation units, but it willnever be possible to design a single simple mathematicalmodel that can predict outcome in every single strokepatient. Hence, selection of patients for stroke unitrehabilitation needs to remain flexible, depending onmultidisciplinary assessment of patients' needs.

References1. King's Fund Consensus Statement. The Treatment of Stroke.

London, England: King's Fund Forum; 1988.2. Langton Hewer R. Rehabilitation after stroke. Q JMed. 1990;279:

659-674.

3. Kennedy FB, Pozen TJ, Gabelman EH. Stroke intensive care: anappraisal. Am Heart J. 1970;80:188-196.

4. Drake WE, Hamilton MJ, Carlsson M, Blumenkrantz J. Acutestroke management and patient outcome: the value of neu-rovascular care units. Stroke. 1973;4:933-945.

5. Norris JW, Hachinski V. Intensive care management of strokepatients. Stroke. 1976;7:573-575.

6. Millikan CH. Stroke intensive care units. Stroke. 1979;10:235-237.7. Indredavik B, Bakke F, Solberg R, Rokseth R, Haaheim LL,

Holme I. Benefit of a stroke unit: a randomized controlled trial.Stroke. 1991;22:1026-1031.

8. Wade DT. Epidemiologically Based Needs Assessment: Stroke.London, England: DHA Research Programme, NHS ManagementExecutive; 1992.

9. Hamrin E. Early activation in stroke: does it make a difference?Scand J Rehabil Med. 1982;14:101-109.

10. Stevens RS, Ambler NR, Warren MD. A randomised controlledtrial of a stroke rehabilitation ward. Age Ageing. 1984;13:65-75.

11. Garraway WM, Akhtar AJ, Prescott RJ, Hockey L. Managementof acute stroke in the elderly: preliminary results of a controlledtrial. Br Med J. 1980;280:1040-1043.

12. Edmans JA, Towle D. Comparison of stroke unit and non-strokeunit in patients on independence in ADL. Br J Occ Ther. 1990;53:415-418.

13. Friedman PJ. Stroke rehabilitation in the elderly: a new patientmanagement system. N Z Med J. 1990;103:234-236.

14. Ebrahim S. Clinical Epidemiology of Stroke. Oxford, England:Oxford University Press; 1990.

15. Spence DJ, Donner A. Problems in design of stroke treatmenttrials. Stroke. 1982;13:94-99.

16. Ebrahim S, Nouri F, Barer D. Cognitive impairment after stroke.AgeAgeing. 1985;14:345-350.

17. Macleod J. The Nervous System in Clinical Examination. 6th ed.Edinburgh, Scotland: Churchill Livingstone, Inc; 1983.

18. Holden MK, Gill KM, Magliozzi MR, Nathan J, Piehl-Baker L.Clinical gait assessment in the neurologically impaired: reliabilityand meaningfulness. Phys Ther. 1984;64:35-40.

19. Wade DT. Assessing disability after acute stroke. In: Wade DT,ed. Stroke: Epidemiological, Therapeutic and SocioeconomicAspects. London, England: Royal Society of Medicine; 1986:101-114.

20. Qureshi KN, Hodkinson HM. Evaluation of a ten-question mentaltest in the institutionalised elderly. Age Ageing. 1974;3:152-157.

21. Jitapunkul S, Pillay I, Ebrahim S. The abbreviated mental test: itsuse and validity. Age Ageing. 1991;20:332-336.

22. Kalra L, Crome P. The role of prognostic scores in targeting strokerehabilitation in elderly patients. J Am Geriatr Soc. 1993;41:396-400.

23. Issacs B, Marks R. Determinants of outcome of stroke rehabili-tation. Age Ageing. 1973;2:139-149.

24. Altman DG. How large a sample? Br Med J. 1980;281:1336-1338.25. Kalra L, Smith D, Crome P. Stroke in patients aged over 75 years:

outcome and predictors. Postgrad Med J. 1993;69:33-36.26. Sandercock P, Molyneaux A, Warlow C. Value of computed

tomography in patients with stroke: Oxfordshire CommunityStroke Project. Br Med J. 1985;290:193-197.

27. Sotaniemi KA, Phytinen J, Myllyla VV. Correlation of clinical andcomputed tomographic findings in stroke patients. Stroke. 1990;21:1562-1566.

28. Ricci S, Celani MG, LaRosa F, Vitali R, Duca E, Ferraguzzi R,Paolotti M, Seppoloni D, Caputo N, Chiurulla C. SEPIVAC: acommunity based study of stroke incidence in Umbria, Italy.J Neurol Neurosurg Psychiatry. 1991;54:695-698.

29. A reassessment of nurses' attitudes towards stroke patients ingeneral medical wards. JAdv Nurs. 1991;16:1336-1342.

30. Kalra L, Dale P, Randall G, Crome P. Evaluation of a clinicalprognostic scale for elderly stroke patients. Age Ageing. In press.

31. Young A. Assessment for rehabilitation after stroke. Ann AcadMed Singapore. 1988;17:267-274.

32. Prescott RJ, Garraway WM, Akhtar AJ. Predicting functionaloutcome following acute stroke using a standard clinical exami-nation. Stroke. 1982;13:641-647.

33. Allen CMC. Predicting recovery after acute stroke. Br J Hosp Med.1984;31:428-434.

34. Barer DH, Mitchell JRA. Predicting the outcome of acute stroke:do multivariate models help? Q J Med. 1989;261:27-39.

35. Shah S, Vanclay F, Cooper B. Stroke rehabilitation: Australianpatient profile and functional outcome. J Clin Epidemiol. 1991;44:21-28.

by guest on January 27, 2015http://stroke.ahajournals.org/Downloaded from

L Kalra, P Dale and P CromeImproving stroke rehabilitation. A controlled study.

Print ISSN: 0039-2499. Online ISSN: 1524-4628 Copyright © 1993 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Stroke doi: 10.1161/01.STR.24.10.1462

1993;24:1462-1467Stroke. 

http://stroke.ahajournals.org/content/24/10/1462World Wide Web at:

The online version of this article, along with updated information and services, is located on the

  http://stroke.ahajournals.org//subscriptions/

is online at: Stroke Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer process is available in the

Request Permissions in the middle column of the Web page under Services. Further information about thisOnce the online version of the published article for which permission is being requested is located, click

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.Strokein Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:

by guest on January 27, 2015http://stroke.ahajournals.org/Downloaded from