longer term stroke adequate - bmj quality &...

Quality in Health Care 1997;6:125-130

Longer term quality of life and outcome in strokepatients: is the Barthel index alone an adequatemeasure of outcome?

Peter R Wilkinson, Charles D A Wolfe, Fiona G Warburton, Anthony G Rudd,Robin S Howard, RalphW Ross-Russell, Roger Beech

Division of PublicHealth SciencesPeter R Wilkinson,honorary lecturer inpublic health medicineCharles D A Wolfe,senior lecturer in publichealth medicineFiona G Warburton,assistant statisticianRoger Beech, seniorresearch fellow inoperation research

Department of Care ofthe ElderlyAnthony G Rudd,consultant physician

Department ofNeurology, StThomas's Hospital,LondonRobin S Howard,consultant neurologistRalphW Ross-Russell,consultant neurologist

Correspondence to:Dr C D A Wolfe, Division ofPublic Health Sciences,UMDS Guy's and StThomas's Medical andDental School, Block 8(South Wing), St Thomas'sHospital, Lambeth PalaceRoad, London SEI 7EH.

Accepted for publication11 April 1997

AbstractObjectives-To consider whether the Bar-thel Index alone provides sufficient infor-mation about the long term outcome ofstroke.Design-Cross sectional follow up studywith a structured interview questionnaireand measures of impairment, disability,handicap, and general health. The scalesused were the hospital anxiety and depres-sion scale, mini mental state examination,Barthel index, modified Rankin scale,London handicap scale, Frenchay activi-ties index, SF36, Nottingham health pro-file, life satisfaction index, and thecaregiver strain index.Setting-South east London.Subjects-People, and their identifiedcarers, resident in south east London in1989-90 when they had their first in a life-time stroke aged under 75 years.Interventions-Observational study.Main outcome measures-Comparisonand correlation of the individual Barthelindex scores with the scores on other out-come measures.Results-One hundred and twenty three(42%) people were known to be alive, ofwhom 106 (86%) were interviewed. Themedian age was 71 years (range 34-79).The mean interval between the stroke andfollow up was 4.9 years. The rank correla-tion coefficients between the Barthel andthe different dimensions of the SF36ranged from r=0.217 (with the role emo-tional dimension) to r=0.810 (with thephysical functioning dimension); with theNottingham health profile the range wasr-0.189 (with the sleep dimension, NS) tor=-0.840 (with the physical mobility di-mension); with the hospital and anxietyscale depression component the coef-ficient was r=-0.563, with the life satisfac-tion index r=0.361, with the Londonhandicap scale r=0.726 and with theFrenchay activities index r=0.826.Conclusions-The place of the Barthelindex as the standard outcome measurefor populations of stroke patients is stilljustified for long term follow up, and maybe a proxy for different outcome measuresintended for the assessment of otherdomains.(Quality in Health Care 1997;6:125-130)

Keywords: stroke; outcome; quality of life

IntroductionRecently there has been increasing interest inthe measurement of outcomes of health care.This is as true for stroke medicine as it is forany other specialty. The Key Area Handbook forcoronary heart disease and stroke produced bythe Department of Health' suggested action tostreamline the assessment of people withstrokes. These were the development of "com-mon assessment procedures throughout therehabilitation process" and for "a standardassessment procedure which could act as thebasis of outcome measures, against which per-formance could be monitored". The implica-tion of these statements is that providers ofservices to stroke patients, be they healthcarers, social services, or others, should worktogether to develop uniformly agreed policiesand outcome assessments.

Quality of life is difficult to define but deHaan et a2 suggest that over time a broad con-sensus has arisen as to which dimensionsshould be included in assessing the quality oflife of stroke patients; physical, functional, psy-chological, and social health. The availablemeasures are not always specifically aimed atone of these dimensions alone. An alternativeframework for measurements of stroke out-come is the classification of impairments,disabilities, and handicaps of the World HealthOrganisation (WHO).'The objective of the present study was to

consider whether the Barthel index alone pro-vides sufficient information about the longterm outcome of stroke. This should helpdecide which outcome measures are the mostpragmatic and appropriate for assessing thelong term outcome of stroke patients.

CHOICE OF OUTCOME MEASURES USED IN THIS

STUDYWade4 suggested that the ideal stroke outcomemeasure should be simple to administer, validand reliable, sensitive to clinically relevantchange, relevant to both user and patient, andunderstood by a non-specialist. Very few of theavailable scales satisfy these requirements.The Barthel index5 is a leading contender as

an outcome measure for populations of strokepatients, and most closely satisfies thesecriteria.4 It is valid and reliable, simple toadminister, and easily understood and conse-quently has become an accepted outcomemeasure in stroke research. The scale assesses10 activities of self care and daily activity. Themaximum score of 20 suggests that the subject

125

on 25 July 2018 by guest. Protected by copyright.

http://qualitysafety.bmj.com

/Q

ual Health C

are: first published as 10.1136/qshc.6.3.125 on 1 Septem

ber 1997. Dow

nloaded from

http://qualitysafety.bmj.com/

Wilkinson, Wolfe, Warburton, Rudd, Howard, Ross-Russell, et al

is functionally independent. The functionalbasis of the Barthel index allows an estimate ofthe level of disability and gives some insightinto the likely need for services. However, theBarthel index lacks any assessment of the abil-ity to carry out tasks in the community or anymeasures of communication, mental and emo-tional factors, or of family involvement.The mini mental state examination' assesses

cognitive function. It has been widely used andvalidated. The maximum score is 30. Peoplewith a normal mental state have high scores.The hospital anxiety and depression scale7 is avalid measure of anxiety and depression.Scores between 8 and 10 were considered bor-derline and scores of 11 or more as indicatingdepression or anxiety.The Frenchay activities index was included

as a measure of social function and ofhandicap.8 It was designed specifically forstroke patients. Although there was no pre-stroke assessment to allow for changes overtime the index provided a framework of activi-ties to ask the patients about. A scoring systemwith a minimum score of 15 and a maximum of60 was used. The higher the score the greaterthe number of activities done. The modifiedRankin scale9 is used as a handicap measure. Ithas many limitations and combines impair-ment, disability, and dependency into onescore. Scores range from 0 to 5. A score of 0 to2 indicates functional independence and ascore of 5 indicates severe handicap. The Lon-don handicap scale is a recently developedscale'0 which uses a questionnaire to assess sixdifferent dimensions of handicap. Each dimen-sion has its own scale but a formula with scaleweights calculates an overall handicap score ofbetween 0 (maximum handicap) and 1 (mini-mum). Initial assessments of validity andreliability with stroke patients have beensatisfactory." The scale is intended for grouprather than individual assessment.There are many different measures of self

perceived health. The short form 36 (SF36)allows the domains of physical and social func-tioning and mental health to be assessed as wellas the effects of physical and mental health onthe daily activities of stroke patients.'2 Eachdimension is scored from 0 to 100. A high scoresuggests good health. Assessment of the valid-ity and reliability of the SF36 is ongoing, butsatisfactory to date.

Suggestions have been made about how toimprove the SF36 for use with the elderlypeople. 13 Both the original and modifiedversions of the SF36 were used in this study.The Nottingham health profile (NHP)'4 hasbeen used with stroke patients.'5 Only the firstpart of the Nottingham health profile was usedin this study as there are reservations about thereliability of the second part.'6 The Notting-ham health profile assessments include thedomains of energy, emotion, social isolation,and physical mobility. Each dimension isscored between 0 and 100. A high scoresuggests poor health. The life satisfactionindex"' was included as an example of a singleglobal self assessment of outcome. A totaloverall score for life satisfaction is calculated

from eight questions. Four of the questions areinterpreted as measuring "acceptance-contentment" and the remainder as"achievement-contentment".For carers the first part of the Nottingham

health profile and the hospital anxiety anddepression scale were used. The caregiverstrain index'8 requires further work on itsvalidity and reliability, but was included as itprovides information not just about strain oncarers but also some of the particular problemscarers face. This could help identify thesupport and services carers require.A wide range of scales have been selected for

investigation. Some such as the hospital anxietyand depression scale, and Barthel andFrenchay activities index aim to assess impair-ment, disability, and handicap respectively.Others such as the Nottingham health profileand SF36 cannot be classified so easily. Not allthe domains of all the scales can be translatedinto service provision for stroke patients.Purchasers will be particularly interested inthose instruments which are valid, reliable,acceptable both in terms of cost and ease ofcompletion, and are easily interpreted asoutcomes or health needs in terms of serviceprovision. Table 1 shows the final choice ofscales used in this study.As well as the scales two simple stroke ques-

tions were included in the interviewquestionnaire.'9 The questions were designedto be used in the follow up of large groups ofstroke patients. The questions were "in the lasttwo weeks did you require help from anotherperson for everyday activities?" and "do youfeel that you have made a complete recoveryfrom your stroke?".

ESTABLISHING THE COMMUNITY BASED STROKEREGISTERIn 1989-90 a community based stroke registerwas established in south east England20 whichprovided a cohort for this follow up study in1994/1995. The methodology used to establishthe register in 1989 has been previouslydescribed.20 The World Health Organisation(WHO) definition of stroke was used.2' Peoplewho had had their first ever stroke aged under75 years were entered including those who didnot survive the initial event. Cases werenotified by general practitioners, districtnurses, the rehabilitation services, and hospitalconsultants. Field workers reviewed accidentand emergency registers and visited hospitalwards. Hospital and coroners' postmortemreports and the local district death certificateswere also reviewed. Information about thepatients was collected about their demography,their premorbid level of disability and handi-cap, their maximum clinical impairment, andthe therapy services they received. All thepatients were reviewed at the time of theirstroke and after three months.

MethodsASSESSMENTSIn the present study only those survivors whohad lived in London at the time of their strokewere traced and interviewed. Subjects who had

126



/Q

ual Health C


ber 1997. Dow

nloaded from


Longer term quality of life and outcome in stroke patients. Is the Barthel index alone an adequate measure of outcome?

Table 1 Outcome scales used in the study

Scale What it measures Reason for inclusion

SF36 General health Broad outcome measureNottingham health profile General health Broad measure. Used with

stroke patientsLife satisfaction index Quality of life Single global measureHospital anxiety and depression Psychological impairment Assesses both depression and

scale anxietyMini mental state examination Cognitive impairment Standard measureBarthel index Disability Standard stroke outcome

measureModified Rankin scale Handicap Widely used in stroke studiesFrenchay activities index Handicap Stroke specific measure

Details some aspects ofcommunity involvement

London handicap scale Handicap A new scaleCaregiver strain index General health Carer specific scale. Indicates

some causes of strain and stress

moved from London, but who lived anywherein the United Kingdom, were included. Thesubjects completed a structured interviewquestionnaire and the various scales wereassessed face to face at the same interview. Allthe interviews were completed in 1994-5 byone investigator (PW). The scales were com-pleted in the same order, and most were readout to the subjects. The original and modifiedversions of the SF36 were allocated randomlywith a block technique. The project receivedlocal ethical approval and the subjects gavetheir informed consent.

ANALYSIS OF THE DATAThe data were analysed primarily at theindividual level, but some analyses were basedon groups of subjects categorised according tothe disability categories based on the Barthelindex as suggested by Wade andLangton-Hewer." The nature of the scalesused meant that the subjects' scores were notnormally distributed and therefore non-parametric tests were used for most of theanalyses.

PSYCHOLOGICAL IMPAIRMENTThe subjects' scores on the mini mental stateexamination were described by disability cat-egory. The correlation between the subjects'individual scores on the hospital anxiety anddepression scale with their Barthel score wereexplored with the Spearman rank correlationcoefficient. The x' test was used to assess theassociation of disability with anxiety anddepression.

HANDICAPRank correlation was used to assess the corre-lation between the individual subjects' scoreson the London handicap scale and theFrenchay activities index with their Barthelindex scores. The same test was used to assessthe correlation between the modified Rankinscale and the London handicap scale.

SF36 AND THE NOTTINGHAM HEALTH PROFILEThe differences between the two versions of theSF36 were compared with the Mann-WhitneyU test. Rank correlation was used to assess thecorrelation between the subjects' individualscores on the various dimensions of the SF36and Nottingham health profile with their indi-

vidual Barthel scores. The Wilcoxon matchedpairs test was used to compare the scores onthe SF36 and Nottingham health profile withthe predicted norms.

LIFE SATISFACTION INDEXRank correlation was used to assess the corre-lation between the individual person's scoresand their Barthel scores.

TWO SIMPLE STROKE QUESTIONSThe sensitivity and specificity of the two ques-tions were assessed with a modified Rankinscore of 3-5 as suggesting help would berequired, and a score of 0-2 as indicating that itwould not. A modified Rankin score of 0 wastaken as indicating a complete recovery. For the"any help" question a Barthel score of 0-19was taken as indicating the need for help with20 indicating functional independence.

CARERSThe results from the carers were analysed withthe same tests as the subjects.

ResultsOf the 291 people aged under 75 yearsoriginally registered in 1989-90, 154 (53%)had died and 14 (5%) were lost to follow up. Ofthe remaining 123,106 (86%) were inter-viewed in person (the remaining 17 eitherdeclined or their general practitioner did so ontheir behalf). In six cases the informant waseither the main carer or a healthcare profes-sional and the subjective quality of life scaleswere not completed.The median age of the people interviewed at

the time of their stroke was 66 years (range28-74), and at follow up 71 years (range34-79). The mean interval between the strokeand the assessments was 4.9 years with a rangeof 4.3 to 5.5 years. Fifty seven (54%) subjectswere male, 89 (84%) were white, and 13 (12%)West Indian or African. Fifty six (53%)subjects described themselves as married, and77 (73%) as retired. Only five (5%) peoplewere in full time employment. Fifty (47%)subjects identified a main carer, usually aspouse (72%), son, or daughter (20%). Twentynine carers (58%) were women. The medianage of the carers was 65 years with a range of 26to 87.With the disability categories based on the

Barthel index score 36 (34%) people werefunctionally independent (a score of 20), 39(37%) mildly disabled (a score of 15-19), 17(16%) moderately disabled (a score of 10-14),and 14 (13%) severely disabled (a score of0-9). Information was available about the Bar-thel score at three months after their stroke for103 subjects. Fifty six (54%) subjects were inthe same disability category at five years as theywere at three months, but 40 (39%) had dete-riorated by at least one category. The remain-ing seven (7%) had improved.Twelve people did not complete all the qual-

ity of life scales, either because they werephysically unable to complete the forms andunable to make their responses understood, orthe interview was ended at their request. Six

127



/Q

ual Health C


ber 1997. Dow

nloaded from



were severely disabled, three moderately disa-bled, and two mildly disabled. The remainingperson was functionally independent and com-pleted all except one question on the Londonhandicap scale. Those people who did notcomplete the scales were generally moredisabled than those who did. In interpretingthe following results it should be rememberedthat the distribution, if those who did not com-plete the questionnaires had been included,would probably have moved towards a poorerquality of life.

PSYCHOLOGICAL IMPAIRMENT

Ninety eight (92%) people completed the minimental state examination. Of these 27 (28%)scored s 24, suggesting a possible cognitivedisorder.23 The importance of this finding iscomplicated by some of the subjects beingunable to complete some sections because ofarm weakness. Rather than exclude over a

quarter ofthe subjects it was decided to includethose who scored < 24 in the overall analysis.The median scores and ranges for the minimental state examination for the disability cat-egories based on the Barthel index were severe23 (14-29), moderate 26 (6-29), mild 28(1930), and functionally independent 28 (20-30).The hospital anxiety and depression scale

was completed by 96 (91 %) subjects. Thenumber ofpeople identified as being anxious or

depressed for each disability category was

small, but only six of the 36 functionally inde-pendent subjects had scores in the depressed or

borderline depressed categories compared with29 of the 60 with some disability (x2=9.739,df=l, P=0.0018). For the anxiety scale thenumbers were seven and 23 respectively(X2=3.737, df=l, P=0.0532). The correlationcoefficient between the subjects' Barthel score

and their hospital anxiety and depression scalescore was much stronger for the depressionscale (r=-0.563 p<0.001 df=94) than for theanxiety scale (r=-0.187, P>0.05, df=94).

HANDICAPWith the modified Rankin scale 38 (36%) peo-ple were functionally independent (scores of0-2). The London handicap scale scores were

correlated with the individual subjects' modi-fied Rankin category (r=0.784, P<0.001, df =94) and with their Barthel scores (r0.726,P<0.001, df =94).Most people with a moderate or more severe

disability carried out few of the activities in theFrenchay activities index. Fourteen (13%)people scored 15, 37 (35%) 16-30, 39 (37%)31-45, and 16 (15%) 46-55. The correlationcoefficient between the individual scores on theFrenchay activities index and the Barthel indexwas 0.826 (P<0.001, df= 104).

SHORT FORM 36 AND THE NOTTINGHAM HEALTH

PROFILEThe SF36 was completed by 97 subjects.Although the scores tended to be higher for themodified version there was no significantdifference between the two scores on any of thedimensions and for the analysis the results have

been combined. All the various dimensions ofthe SF36 showed a significant correlationbetween the individual subject's scores andtheir Barthel score (table 2). The Nottinghamhealth profile was completed by 96 respond-ents. All the dimensions, apart from that forsleep, had a significant correlation with theBarthel score (table 2).The only significant correlations between the

various dimensions and the patient's age atinterview were with "role-emotional" on theSF36 (P<0.02), and with sleep (P<0.05), onthe Nottingham health profile.

Tables 3 and 4 give the median differencebetween the scores of the subjects on thedimensions of the SF36 and Nottinghamhealth profile and the expected scores from ageand sex matched interviewed normal values forthe SF3624 and from age, sex, and social classmatched norms for the Nottingham healthprofile.25

LIFE SATISFACTION INDEX

The life satisfaction index was completed by 95subjects. The results suggest a trend of increas-ing overall satisfaction, and for both subscales,with decreasing disability. The correlationcoefficient between the individual Barthelscores and the scores on the life satisfactionindex were 0.361 for the overall total(P<0.001, df 93), 0.357 (P<0.001) for theacceptance total, and 0.307 (P<0.01) for theachievement total. The life satisfaction indexwas originally intended for elderly subjects,and similar results were obtained for patientsaged 65 and over.

TWO SIMPLE STROKE QUESTIONSThese were completed for all 106 subjects. Thesensitivity of the "any help in the last twoweeks" question was 74% (50/68) (95% CI 63to 84) and the specificity 87% (33/38) (95% CI76 to 98) against the modified Rankin scaleand 74% (52/70) (95% CI 64 to 85) and 92%(33/36) (95% CI 83 to 100) respectivelyagainst the Barthel index. For the "completerecovery question" everyone with a modifiedRankin score of 0 answered yes (9/9). The spe-cificity was 75% (73/97) (95% CI 67 to 84).Table 2 The rank correlation coefficients between theindividual Barthel index score atfive years and thecorresponding scores on the various dimensions of the SF36and Nottingham health profile

Rank correlationcoefficient with Barthelindex

Quality of life scale r P value

SF36 (n=97):Physical functioning 0.810 <0.001Social functioning 0.481 <0.001Role: physical 0.415 <0.001Role: emotional 0.217 <0.05Mental health 0.332 <0.01Vitality 0.500 <0.001Bodily pain 0.356 <0.001General health 0.438 <0.001

Nottingham health profile (n=96):Energy -0.605 <0.001

Pain -0.499 <0.001Emotion -0.423 <0.001Sleep -0.189 >0.05Social interaction -0.460 <0.001Physical mobility -0.840 <0.001

128



/Q

ual Health C


ber 1997. Dow

nloaded from


Longer term quality of life and outcome in stroke patients. Is the Barthel index alone an adequate measure ofoutcome?

Table 3 Median differences for SF36 scores for patientswith stroke compared with normal values matchedfor ageand sex

Median difference (range) P value

Physical functioning -28.80 (-91.6 to +46.2) <0.0001Social functioning -20.80 (-87.5 to +29.2) <0.0001Mental health - 11.60 (-57.9 to +26.4) <0.0001Vitality - 11.10 (-68.9 to +52.4) 0.001Pain -9.30 (-64.1 to +39.7) 0.23General health 0.00 (-70.2 to +53.4) 0.35Role physical -27.00 (-87.1 to +52.6) <0.0001Role emotional -44.20 (-92.9 to +22.5) <0.0001

A negative value indicates that patients have worse health.

Table 4 Median differences for Nottingham health profilescores for patients with stroke compared with normal valuesmatchedfor age, sex, and social class

Median difference (range) P value

Energy 23.30 (-35.3 to +93.4) <0.0001Pain -2.00 (-19.0 to +90.1) 0.041Emotion 9.00 (-15.7 to +89.5) <0.0001Sleep 0.95 (-27.3 to +85.9) 0.002Social isolation 14.80 (-13.4 to +95.3) <0.0001Physical mobility 31.24 (-22.1 to +88.9) <0.0001

A positive value indicates that patients have worse health.

CARERSOf the 50 carers, 36 (72%) completed the Not-tingham health profile, 37 (74%) the hospitalanxiety and depression scale, and 38 (76%) thecaregiver strain index. Of the 12 who did notcomplete any of the scales, one was caring for a

patient with a Barthel score of 0-9, three forpatients with a score of 10-14, and eight forpatients with a score of 15-19.There was no significant correlation between

the carers' scores on any of the Nottinghamhealth profile dimensions and the patients'Barthel scores. With the Wilcoxon matchedpairs test the only significant difference be-tween the carers' scores on the Nottinghamhealth profile and the age and sex matchednorms was for the emotion scale where themedian difference was 6.22 (range -16.6 to+89.4, P=0.017).Twelve (32%) of the 37 carers who com-

pleted the hospital anxiety and depression scalehad borderline or clinically anxious scores. Six(16%) had borderline or depressed scores. Onthe caregiver strain index eight (21 %) carers

had scores of seven or more, suggesting thatthey were stressed.

DiscussionThis study has described the long term qualityof life of a community based population ofstroke patients. There are no agreed definitionsor scales for measuring quality of life. In an

attempt to describe it with one or more scales,it is acknowledged that the subjects may feelthe scales do not accurately reflect their own

quality of life in terms of their individualcircumstances and requirements.

Also, the study has considered whether it isjustifiable to use the Barthel index as the onlyoutcome measure because of its correlationwith measures focusing on other domains. Inassessing this through correlations it is ac-

knowledged that those measures with dimen-sions or components which are strongly basedon physical functioning are likely to have the

strongest correlation. It should also be remem-bered that a significant correlation coefficientdoes not necessarily mean a very strong linearrelation.From a clinical viewpoint the measures of

interest in the long term outcome of strokehave focused mainly on survival, recurrence,physical recovery, and the occurrence ofcomplications. But interest in more generaloutcomes, such as quality of life, is beingencouraged. The drawback of general meas-ures, such as the SF36 and Nottingham healthprofile, is that it may be more difficult to equatea particular finding with either the structure orprocess involved in the care. A search of themedical literature for quality of life scalesshows many possible instruments. In this areaof work investigators tend to devise new meas-ures rather than use those already available.Many of the instruments have not beenadequately assessed for validity and reliability.

In 1987, a review of outcome measures instroke rehabilitation research criticised the lackof use of broader measures. The authors drewattention to the limitations of assessing onlyphysical recovery and self care and suggestedthat subjective measures of health status andquality of life should be included inassessments.26 Increasing a patient's quality oflife is an important target of rehabilitation.Ebrahim"7 suggested that this may be equatedwith reducing handicap and that assessmentsof quality of life and of handicap may be morerelevant than changes in impairment or disabil-ity.

This study considered the long term out-come of stroke in terms of the relation betweenmeasures of quality of life, impairment, andhandicap, with disability as measured by theBarthel index. Overall the subjects' scores onthe various scales were correlated with theirdisability. The major component of thisdisability is likely to have been related to theoriginal stroke. The strongest correlation coef-ficients were between the Barthel index scoresand the hospital anxiety and depression scaledepression score, Frenchay activities indextotal, physical functioning and vitality dimen-sions on the SF36, physical mobility andenergy dimensions on the Nottingham healthprofile, and the London handicap scale.The weakest correlation coefficients were

between the Barthel index scores and the"role-emotional" and "mental health" dimen-sions on the SF36 and the "sleep" and"emotion" dimensions on the Nottinghamhealth profile. The correlation coefficientsbetween the Barthel and the anxiety compo-nent of the hospital anxiety and depressionscale as well as the life satisfaction index werealso poor.Not surprisingly the strongest correlations

were between the Barthel index and those out-come measures most closely related to physicaldisability. The weaker correlations with themental health and emotion components of theSF36 and Nottingham health profile reflectthat the Barthel indexdoes not measure anypsychological components. Despite this areasonable correlation with the depression

129



/Q

ual Health C


ber 1997. Dow

nloaded from



scale of the hospital anxiety and depressionscale was found.

Therefore to achieve a truly comprehensiveassessment of the long term outcome of strokeit is likely that outcome measures aimed atoverall wellbeing and mental health would berequired as well as the Barthel index. This willincrease the burden on the patients and requireadditional resources. How easily the extrainformation obtained could be used to changeprovision of services is questionable.What do these findings mean in terms of

outcome measures for stroke patients? It seemsthat the place of the Barthel index as the stand-ard outcome measure is still justified for longterm follow up. This is particularly relevant forthe provision of services. The Barthel indexalso has the advantages of being simple andquick to complete. However, the most suitablemeasures for research studies will depend onthe question being asked. The additional infor-mation from further outcome measures is lim-ited by their correlation with the Barthelscores. This could be interpreted as suggestingthat a broad general measure, or handicapmeasure, could be used on its own as it wouldreflect the underlying disability. However, thedisadvantage of the broader outcome measureis that it is not possible to relate scores to theprocess of care either to identify potentialproblems or to estimate the need for particularservices. This is not as true of the Barthelindex, which could also be used to estimatesome of the service need-for example, for helpwith bathing, feeding, and toileting. A reason-able aim for rehabilitation is to help the patientreturn to their premorbid lifestyle. For this tobe used as an outcome measure requires anassessment of the patient's premorbid state.This can be achieved with the Barthel indexand the Frenchay activities index, but mightnot be possible for the general health measures.As a result of the need to follow up many

patients in a trial, Lindley et all9 investigated byphone or by post the validity of two simplequestions related to stroke, and concluded thatsuch simple questions could be used to assessstroke outcomes. In this study the two simplestroke questions had comparable sensitivityand specificity to the original work for the helpquestion, but a worse specificity for the recov-ery question. The confidence intervals for thevarious values do overlap. These differencesmay be due to the questions being asked inperson and not by phone or post as in the studyby Lindley et al.'9 Further work is clearlyneeded but these findings are promising andmay lead on to the use of reliable and validsimple patient self assessed outcome measures.The study by Lindley et al"9 also used a postaland telephone version of the Barthel index, butthe sensitivities of the questions were poorwhen judged against the assessment by the vis-iting nurse. Despite the potential problems of

developing a valid and reliable telephone orpostal version, the use of the Barthel index asthe standard outcome measure may give someinsight into the population's quality of life aswell as their disability.

We thank all the subjects and their carers for their help with thestudy. We also thank local healthcare professionals for their helpin establishing the original register and for their help inapproaching the subjects. This study was funded by Researchand Development, Northern and Yorkshire Regional HealthAuthority and the Stroke Association.

1 Department of Health. Health of the Nation key areahandbook: coronary heart disease and stroke. London:Department of Health, 1993.

2 de Haan R, Aaronson N, Limburg M, Langton-Hewer R,Crevel H. Measuring quality of life in stroke. Stroke 1993;24:320-7.

3 World Health Organisation. The international classification ofimpairments, disabilities and handicaps. Geneva: WHO,1980.

4 Wade DT. Assessing disability after acute stroke. In: F Clif-ford Rose, ed. Stroke: epidemiological, therapeutic andsocio-economic aspects. London: Royal Society of Medicine,1986:101-14.

5 Mahoney F, Barthel D. Functional evaluation: the Barthelindex. MdMedJ 1965;14:61-5.

6 Folstein M, Folstein S, McHugh P. "Mini-mental state": apractical method for grading the cognitive state of patientsfor the clinician. JPsychiatr Res 1975;12:189-98.

7 Zigmond A, Snaith R. The hospital anxiety and depressionscale. Acta Psychiatr Scand 1983;67:361-70.

8 Holbrook M, Skilbeck C. An activities index for use withstroke patients. Age Ageing 1983;12: 166-70.

9 Bamford J, Sandercock P, Warlow C, Slattery J. Interob-server agreement for the assessment of handicap in strokepatients. Stroke 1989;20:828.

10 Harwood R, Rogers A, Dickinson E, Ebrahim S. Measuringhandicap: the London handicap scale, a new outcomemeasure for chronic disease. Quality in Health Care 1994;3:11-5.

11 Harwood R, Gompertz P, Ebrahim S. Handicap one yearafter a stroke: validity of a new scale. J Neurol NeurosurgPsychiatry 1994;57:825-9.

12 Brazier J, Harper R, Jones N. O'Cathain A, Thomas K,Usherwood T, et al. Validating the SF36 health surveyquestionnaire: new outcome measure for primary care.BMJ 1992;305: 160-4.

13 Hayes V, Morris J, Wolfe C, Morgan M. The SF36 healthsurvey questionnaire: is it suitable for use with older adults?Age Ageing 1995;24: 120-5.

14 Hunt S, McKenna S, McEwen J, Backett E, Williams J,Papp E. A quantitative approach to perceived health status:a validation study. J Epidemiol Community Health 1980;34:281-6.

15 EbrahimS, Barer D, Nouri F. Use of the Nottingham healthprofile with patients after stroke. J7 Epidemiol CommunityHealth 1986;40:166-9.

16 Bowling A. Measuring health. Buckingham: Open UniversityPress, 1991;4:56-92.

17 Bigot A. The relevance of American life satisfaction indicesfor research on British subjects before and after retirement.Age Ageing 1974;3:1 13-21.

18 Robinson B. Validation of a caregiver strain index. .7 Geron-tol 1983;38:344-8.

19 Lindley R, Waddell F, Livingstone M, Sandercock P, DennisM, Slattery J, et al. Can simple questions assess outcomeafter stroke? Cerebrovasc Dis 1994;4:314-24.

20 Wolfe C, Taub N, Woodrow J, Richardson E, Warburton F,Burney P. Does the incidence, severity, or case fatality ofstroke vary in southern England? .7 Epidemiol CommunityHealth 1993;47:139-43.

21 Hatano S. Experience from a multicentre stroke register: apreliminary report. Bull World Health Organ 1976;54:541-53.

22 Wade D, Langton-Hewer. Functional abilities after stroke:measurement, natural history and prognosis. _7 Neurol Neu-rosurg Psychiatry 1987;50:177-82.

23 De Paulo J, Folstein M, Gordon B. Psychiatric screening ona neurological ward. Psychol Med 1980;10:125-32.

24 Lyons R, Fielder H, Littlepage B. Measuring health statuswith the SF36: the need for regional norms. _7 Public HealthMed 1995;17:46-50.

25 The Nottingham health profile user's manual. Manchester:Galen Research, 1994.

26 Seale C, Davies P. Outcome measurement in strokerehabilitation research. International Disability Studies 1987;9:155-60.

27 Ebrahim S. Clinical epidemiology of stroke. Oxford: OxfordUniversity Press, 1990.

130



/Q

ual Health C


ber 1997. Dow

nloaded from


longer term stroke adequate - bmj quality &...

Documents