10.1192/bjp.164.3.380Access the most recent version at doi: 1994 164: 380-386 The British Journal of Psychiatry

  M Sharpe, K Hawton, V Seagroatt, J Bamford, A House, A Molyneux, P Sandercock and C Warlow  

lesion volumewith demographic and social factors, functional status, and brain Depressive disorders in long-term survivors of stroke. Associations  

References

http://bjp.rcpsych.org/cgi/content/abstract/164/3/380#otherarticlesArticle cited in:  

permissionsReprints/

permissions@rcpsych.ac.ukto To obtain reprints or permission to reproduce material from this paper, please write

to this article atYou can respond http://bjp.rcpsych.org/cgi/eletter-submit/164/3/380

serviceEmail alerting

click heretop right corner of the article or Receive free email alerts when new articles cite this article - sign up in the box at the

fromDownloaded

The Royal College of PsychiatristsPublished by on July 12, 2011 bjp.rcpsych.org

 

http://bjp.rcpsych.org/subscriptions/ go to: The British Journal of PsychiatryTo subscribe to

British Journal of Psychiatry (1994), 164, 380—386

Depressive Disorders in Long-Term Survivors of StrokeAssociations with Demographic and Social Factors, Functional Status,

and Brain Lesion Volume

MICHAEL SHARPE, KEITH HAWTON, VALERIE SEAGROATT, JOHN BAMFORD, ALLAN HOUSE,ANDREW MOLYNEUX, PETER SANDERCOCK and CHARLES WARLOW

Sixty surviving patients from a community-based stroke register who had computerisedtomography (CT) scan evidence of a single brain lesion were interviewed three to five yearsafter their first ever stroke. Depression (DSM—IlI—Rmajor depression, partially resolved majordepression, and dysthymia) was present in 11 (18%) of the patients and was associated withimpaired physical and cognitive functioning, greater age, residence in an institution, absenceof a close personal relationship, and larger original brain lesion. Of these variables, onlyfunctional dependence (odds ratio 16.4; confidence interval 1.6—170), larger lesion volume(6.6; 1—50),and female sex (8; 1.1-56) remained significantly associated with depressionafter controlling for all other variables. We conclude that depression in long-term survivorsof stroke has many of the same associations as depression in non-stroke elderly populations.Depression in long-term stroke survivors may also be associated with larger original brainlesions, although this requires confirmation in a prospective study.

Recent publications have highlighted the importantissue of depression in survivors of stroke (reviewsby Starkstein & Robinson (1989), Johnson (1991),and House et a! (1991)). Studies to date haveexamined two groups of possible causes of depressionin this group: (a) those directly relating to the brainlesion, such as its location and size; and (b) thoseunrelated or only indirectly related to the lesion, suchas demographic characteristics, history of depression,cognitive and functional impairment, and thepatient's social circumstances.

Associations between depressive disorder andlesion location have been reported (Robinson et a!,1984a), but not replicated by other investigators(House eta!, 1991). The volume of the brain lesionresulting from the stroke has also been associatedwith depressive disorder in some studies (Eastwoodet a!, 1989; Schwartz et a!, 1990), but this findingwas not replicated by others (Starkstein et a!, 1988;Dam el a!, 1989). There has been a similarinconsistency in the findings regarding factors otherthan the brain lesion. Functional impairment, cognitive impairment, and previous history of depressionhave all been reported to be associated withdepressive disorder by some investigators, butnot by others (Johnson, 1991). Methodologicalproblems may account for this inconsistency (House,1987; House et a!, 1991; Johnson, 1991). Patientsamples have usually included only patients admitted to hospital. Measures of depression haveoften been limited to self-report, used modified

diagnostic criteria, or have combined symptomscores from several instruments. Finally, patientshave been studied at varying time intervals afterstroke.

Most investigators have concerned themselveswith the first six months, and there have beenno systematic studies of persons who have survivedfor two years or more after suffering a stroke.However, there are indications that depressionis an important problem for long-term (three to fiveyears) survivors, many of whom report a poorquality of life (Nieme et a!, 1988) and showincreasing use of psychiatric services (Mayou et a!,1991).

We have studied depressive disorders in long-termsurvivors of stroke in a community sample, usinga standardised interview-based assessment ofdepressive disorder and computerised tomography(CT) assessment of brain lesions. We have previouslyreported our findings of the association of lesioncharacteristics and depressive disorder in this sample(Sharpe et a!, 1990). No significant association wasfound between the location of the brain lesion anddepression, although there was an associationbetween the volume of the brain lesion and depressivedisorder.

In this paper, we report the associations ofdepressive disorder with non-lesion variables. Therelative contributions of lesion volume and nonlesion factors to the occurrence of depressive disorderare also examined.

380

381DEPRESSIONAFTER STROKE

Method

From 1 November 1981to 31 October 1986, a communitystrokeregisterinOxfordshirerecordedall first-evercasesofstrokefrom50generalpractitioners(in tengrouppractices).The register included both urban and rural practices in asocially varied population of approximately 105 000 people.Care was taken to achieve complete case ascertainment, andtodefineaccuratelythepopulationfromwhichthereferralsweremade.Generalpractitionersnotifiedthe projectofany patient they thought to have suffered a stroke ortransient ischaemic attack . All patients were assessed bya project neurologist soon after notification. The finaldiagnosis of stroke was made at a weekly clinical consensusmeeting by a modification of the World Health Organization(WHO)definitionof stroke (Ahoet a!, 1980).All patientswith a first-ever stroke were recorded by the stroke project.CT brain scanning was undertaken whenever possible.Patients were followed up at least annually by OxfordshireCommunity Stroke Project (OCSP) research nurses; if anynew neurological events were detected, the patient wasreassessed by a study neurologist. Full details of the OCSPmethodshavebeenreportedelsewhere(Bamfordetal,1988).

Patients

The records of survivors from the complete cohort of 515confirmed first-ever strokes registered by the OCSP betweenNovember 1981 and October 1984 were reviewed. Thesepatients (and others seen subsequently by the OCSP) havebeen fully described elsewhere (Bamford eta!, 1988). Oneof the aims of the current study was to examine theassociation of lesion characteristics with depression. Forthat reason, patients were included only if they had a CTscan performed shortly after the initial stroke which showeda neurologically appropriate single lesion (infarct orhaemorrhage). Patients with subarachnoid haemorrhagewere excluded. All patients who met these criteria and whowere alive at the time of interview (June/July 1987) were

included.Of the 515 patients, 399 (77°lo)had undergone a CT scan,

and approximately half of the scans had revealed a brainlesion in keeping with the clinical fmdings. Of those patientswith positive scans, 75 had survived to June 1987. Fromthis group, 14 patients were excluded: original scan filmswere not availablein fivecases,and thelesionwas notclassifiable (by study criteria) in nine (three posterior fossa,one midline, five bilateral). Only one patient refusedinterview. Data are presented on the remaining 60 patients.

Patientinterviews

Allpatientswereinterviewedattheirhome, theinterviewer(MS) being unaware of the initial CT scan findings. Theinterviews took place at a median of 44 months (range31—64)after the stroke.

In each case, an informant who knew the patient well(usually a close relative) was also interviewed. Clinical,demographic, and social information was obtainedfor all patients. Family history and history of depressionwere consideredto be positive only if there was a

clear history of treatment for psychiatric problems.Current social circumstances and place of residencewere also recorded, and the presence of a close personalrelationship (a person whom the subject saw at leastweekly and with whom worries and concerns couldbe shared) was determined. The information obtained byinterview was combined with that already collected by theOCSP, which included clinical details of the stroke. Inaddition, mood disorder, functional and cognitive impairment, and dysphasia were assessed by standardisedmeasures.

Assessment

Affective disorder. This was assessed by the StructuredClinical Interview for DSM—III—R(SCID), a structuredinterview (Spitzer et al, 1986) for making diagnosesaccording to DSM-III-R(American Psychiatric Association,1987). The affective disorders section was administered toall patients. Where appropriate, informants' accounts wereused to supplement patient interviews. DSM—III—Rdiagnoses were made by strict and unmodified time criteria.Dysthymic depression was diagnosed for those patients whomet the criteria, irrespective of whether there was a historyof majordepressionprecedingitsonset.For thepurposeof assessing associations with other variables, majordepression and dysthymia (including partially resolvedmajor depression) have been combined and are referred toas ‘¿�DSMdepression'. The category of organic mooddisorder was not used.

The reliability of assignment of DSM—III—Rdiagnoses,including all the diagnostic categories used in the study,was assessed in a subsample of 12 patients. Another psychiatrist (AH) assigned patients to DSM—III—Rcategorieson the basis of the interview data. There was agreementin 11cases; in one case AH diagnosed anxiety disorder whenMS had recorded no disorder present.

BarthelADL Index (Barthel). This 10-item observer-ratedscaleassessesthepatient'sabilitytoperformbasicactivitiessuch as washingand eating(Mahoney & Barthel,1965).It was administered to all patients and scored from acombination of patients' and informants' accounts.The scale was used without weighting items to producea score of 0—20,where 20 indicates full functionalindependence.

Frenchay Activities Index (FAI). This 15-item scale(Holbrook & Skilbeck, 1983)was designed for use in strokepatients to assess a broader range of activities (e.g.gardening and reading) than the basic activities of lifecovered by the Barthel. It produces a score of 15—60.Thereis no recommended cut-off score to indicate impairment.We employed a cut-off score of 25 or less to indicate severeimpairment of activities after inspection of the total scoresrevealed a bimodal distribution.

Mini-Mental State Examination (MMSE). This 11-itemscale was used to assess cognitive function (Folstein eta!, 1975). It produces scores ranging from 0 to 30. Thestandard cut-off of 23 or less was employed to indicatesignificant cognitive impairment.

Frenchay Aphasia Screening Test (FAST). This testprovides an assessment of aphasia, with separate scores

382 SHARPE El AL

(each out of 10) for comprehension and expression. A cutoff of 13 or less in the total score was used to indicatesignificant dysphasia (Enderby et a!, 1987).

CT scans

Cl scans had been performed shortly after the originalstrokes. The interval between scan and stroke was 30 daysor less in 90°loof patients (range 0—77days). The volumeof the stroke lesion was estimated from the hard copy,according to the method of Pullicino et al(1980), as follows.

A correction factor c was used to adjust measurementson the reduced scan image to life size, and the scan slicethickness t was noted. All scan slices on which the lesionwas visible were identified, and the lesion was measuredon the slice on which it was largest. The longest diametera, the orthogonal diameter b, and the number of slices non which the lesion appeared were recorded. The lesionvolume was then calculated by approximation to a sphereby the following formula:

Volume in cubic centimetres (cc) = 1/2 [(a x c) x (b xc) x (n x t)]

Lesion volume was then categorised into small (10 cc orless) or large (more than 10cc). The method of lesionassessment and its interrater reliability are fully describedin an earlier paper (Sharpe et a!, 1990).

Data analysis

To determine the association of each of the variables withDSM depression, we fitted logistic models to the data, usingthe Generalised Linear Interactive Modelling System(GLIM) (Baker & Nelder, 1978). The dependent variablechosen was the logit of the proportion of patients with DSMdepressive disorder. The independent non-lesion variableswere as follows: (a) sex; (b) age (less than 75 years or 75and older); (c) cognitive impairment; (d) close relationship;(e) functional impairment; (f) level of activities; and (g)place of residence (an institution or at home). The lesionvolume variable was also included.

Each variable was fitted separately, and after adjustmentfor all the other variables to allow for possible confounding.The significance of these associations was found from thereduction in deviance when the individual variable wasincluded in the fitted model. This reduction in devianceapproximates the@ statistic. The odds ratios of depressionwere calculated as the exponentials of the parameters ofthe fitted model.

ResultsAll 60 patients were white, and most were married and livingwith a spouse. Fifty-one (85°lo)were living in their ownhomes, either alone or with relatives. Fifty-six (93°lo)wereright-handed (Table 1).

Thirty-one patients had left hemisphere lesions, and 29right. There was a very wide variation in lesion volume(median 6cc; range 1—170).Thirty-seven patients had smalllesions (10cc or less). Approximately half of the lesions

involved cortical tissue, and most (81 %) were infarcts ratherthan haemorrhages. A detailed description of the lesionscan be found elsewhere (Sharpe et a!, 1990).

During follow-up by the OCSP, eight patients (13%) werediagnosed by a project neurologist as having suffered arecurrent stroke before interview. None had undergonerepeat CT scan, and most recurrent strokes were associatedwith only minor changes in physical signs.

There was a wide range of severity of functionalimpairment: 26 patients (43%) were functionally impaired(Barthel score < 20), while 15 patients (25°lo)wereconsidered to be severely impaired (Barthel < 15).

Thirteen patients (22°lo)scored 23 or less on the MMSEand were, therefore, considered to be cognitively impaired.Five had significant dysphasia (FAST < 14). Four weresufficiently impaired to make the psychiatric interviewpotentially unreliable (see below).

Depressive disorder

No patient was completely excluded from the study becauseof dysphasia or cognitive impairment. By using theinformants' reports to supplement interview findings, it waspossible to complete the SCID for all 60 patients.

The DSM—III—Rdiagnoses were either anxiety ordepressive disorders; no cases of mania were found. Sixteen(27%) of the sample of 60 patients fulfilled the criteria fora DSM—lIl—Raxis one diagnosis. The total prevalence ofDSM depressive disorders was 11/60 (18°lo).Five patientsmet DSM—III—Rcriteria for major depression, and the othersix patients met criteria for the less severe but chronic formof depression termed ‘¿�dysthymia'.Of the six patients withdysthymia at the time of the interview, four had previouslymet criteria for major depression. All cases of depressionhad onset after the stroke.

None of the patients had been referred to the psychiatricservices. Five patients had received antidepressant drugs,of whom three were still receiving them. None of thepatients had taken these drugs either consistently, or in whatwould usually be regarded as therapeutic dosages. The fivepatients treated with antidepressant drugs included two withmajor depression at the time of interview, one withdysthymia, and two with no diagnosis. Of the three withouta current diagnosis of major depression, only one had aretrospective diagnosis of major depression. It is, therefore,unlikely that efforts at treatment significantly influencedthe prevalence of depression.

Associations with depressive disorder: univariate analysis

The association of each of the variables with depressivedisorder was calculated by the x2 statistic and odds ratio(Table 1). The association between DSM—III—Rdepressivedisorders and the variables studied is shown in Table I.

Depression was slightly more common in female patients,but this difference failed to reach statistical significance.However, depression was significantly more common amongthe older patients. Neither a family history of depressionnor a personal history of depression predicted depression afterstroke; only one patient with depression had a family history,and no patient had a history of depression before the stroke.

Variable n % depressed Oddsratio 95%confidenceinterval

383DEPRESSION AFTER STROKE

Table 1Association of individual explanatory variables with depression: odds ratios for depression with 95% confidence interval

(before any adjustment) and x2 statistic for test of association

Demographicvariablessex

male1female

age: years<751

>75follow-up:months

<441

>44marriedyes1no

social class1 +2134+5

History of depressive disorderpersonal

no1yes

familyno1yes

Socialcircumstancesinstitutionno1yes

close relationshipyes1no

Personalfunctioningfunctionallyimpaired

no1yes

cognitively impairedno1yes

activitylevelhigh1low

Lesion volumevolume

small1large

372310.8 30.413.60.9-14.53.633

279.1 29.614.21.0-18.44.2'30

3026.7 10.010.30.1-1.72.940

2015 2511.90.5-7.40.910

321810.0

21.916.71

2.51.80.3—24.50.2-21.00.853

720.8 010—3.056

417.8 25.011.50.1-17.10.151

913.7 44.415.01.0-24.24.0'52

813.4 5016.41.3-32.85.0'34

262.9 38.5120.62.3-182.913.5―47

1312.8 38.514.31.0-18.03.9'43

1711.6 35.314.11.0-16.64.2'37

238.1 34.81 6.01.4-26.86.6'

x2 statistic has one degreeof freedom for all casesexcept social class where it has two degreesof freedom.1. Reference category.‘¿�Significantat 5% level.

‘¿�Significantat 1% level.

Both the absence of a close personal relationship andresidence in an institution were associated with DSMdepressive disorder. DSM depressive disorder was alsomore common in the functionally impaired and wasassociated with lower scores on the FAI, reflecting reducedactivity. While cognitive impairment was significantly

more prevalent in depressedpatients, dysphasia was not. Twoof the five patients (40°le)who were dysphasic (FAST <14)were depressed (one major depression; one dysthymia), ascompared with 9/55 (16%) of the remaining patients(x@=1.7). Finally, large initial brain lesions (> 10cc) weresignificantly associated with depressive disorder.

VariableOdds ratio95%confidenceintervalSexmale11female8.01

.1—¿�56Volumesmall11large6.91.0-50Functional

impairmentno11yes16.41.6-170

384 SHARPE El AL

survivors because small lesions which are likely tocause less disability may not be detectable on a scan(Bamford et a!, 1990). Furthermore, five patients wereknown to have suffered recurrent strokes, and despitecareful follow-up it is conceivable that other survivingpatients also suffered recurrences which may havecomplicated the initial assessments of brain injury.

The group available for interview represented lessthan half of the initial sample selected from thosewith positive CT scans. We did not assess the lesionvolume in those who had died, and we do not knowwhether they suffered depression. Furthermore,survival may be influenced by both lesion size anddepressive disorder. The conclusions of this study arerelevant only to survivors, and we cannot draw firmconclusions concerning the ability of the initial CTscan to predict future depression.

The prevalence of DSM depressive disorder in theselong-term survivors of stroke was 18°lo(95¾ confidence intervals 10-28%). This may be comparedwith a prevalence of 8.5% (3.5—13.5%) in a randomsample of 109 elderly patients drawn from age—sexregisters of four of the general practices collaboratingin the OCSP (House et a!, 1991). Thus, it appearsthat there may be an increased long-term risk ofdepression in survivors of stroke. Why might this be?

The associations between depressive disorder andthe non-lesion factors studied here have all beendemonstrated in the general elderly population.Although older people have a lower overall prevalenceof depressive disorder than younger persons (Blazer,1989), within the older age group the prevalence ofdepressive disorder may increase with age (Koeniget a!, 1988). Similarly, lack of social support(Murphy, 1982), residence in an institution (Blazer,1989), and both functional (Murphy, 1982) andcognitive impairment (Reifler et a!, 1982) have beenfound to increase the risk of depressive disorder inthe elderly. Thus, the factors associated withdepressive disorder in the sample of long-termsurvivors are similar to those associated withdepressive disorder in the general population.

The association found between the volume of theinitial brain lesion in patients who have sufferedstroke, and later depression has been previouslyreported for short-term survivors (Sinyor et a!, 1986;Eastwood et a!, 1989; Schwartz eta!, 1990). This riskfactor may therefore be specific to those elderlypersons who have suffered stroke.

The multivariate analysis of our sample of longterm survivors revealed that functional impairmentwas the factor most strongly associated with depressivedisorder. These results confirm findings from studiesof short-term stroke survivors (Robinson et a!, 1983;Robinson et a!, 1984b; Wade et a!, 1987; Parikh eta!,

Table 2Odds ratios estimated from logistic model including sex,

lesion volume, and functional impairment

1. Reference category.

Associations with depressive disorder: multivanate analysis

Because several variables were significantly associated withDSM depressive disorder, a logistic regression analysis wasperformed. The significance of the association of individualvariables with depression was determined after adjustmentfor the other variables. Only functional dependence, lesionvolume, and sex remained significant associations (x2=11.1, 9.9, and 3.9 respectively, d.f. = 1, P< 0.05). The oddsratios estimated from the logistic model consisting only ofthese variables are shown in Table 2.

Discussion

This is the first systematic study of the pathogenesisof depression in long-term stroke survivors. It hasadvantages over previously reported studies in thatthe study sample was derived from a complete cohortof first-ever stroke patients obtained from a definedcommunity population, and was, therefore, unbiasedby hospital admission. Depressive disorder wasassessed by a standardised interview with unmodifieddiagnostic criteria, and initial brain lesions were allassessed by CT scan in conjunction with a fullneurological assessment.

However, the current study also has methodologicalshortcomings. The principal shortcomings arise fromthe size of the sample interviewed, and possible biasin the sample selection. The effect of the relativelysmall sample size can be seen in the wide confidencelimits around the odds ratios for the variablesassociated with depressive disorder. Possiblesampling biases may have occurred as a result ofinitial patient selection, or because of factorsinfluencing survival from the time of stroke tothe time of interview. Initial patient selectionrequired that the patient have a visible lesion on CTscan. This requirement may have biased the sampletoward a greater degree of functional impairmentthan that of the whole population of long-term stroke

385

1988). While functional impairment may causedepression, it has also been found that early depression predicts later functional impairment (Parikhet a!, 1990). The relationship between these variablesin long-term survivors is therefore likely to becomplex.

Functional impairment correlated with lesionvolume in our sample, an association also found inshort-term stroke survivors (Robinson et a!, 1983).We therefore considered it most likely that theassociation between initial lesion volume and depression was mediated by functional impairment.However, the association between brain lesionvolume and depression remained significant afteradjusting for functional impairment and sex. Thisfmding raises the possibility that the amount of initialbrain damage may play a direct role in the origin ofdepression up to 5 years after the stroke.

We conclude that depression in long-term survivorsof stroke is more likely to occur in the old, the lonely,the functionally and cognitively impaired, and thoseliving in institutions. The strongest association is withfunctional dependence. The size of the original brainlesion is associated with depression in those whosurvive and may, therefore, be a risk factor fordepressive disorder that is specific to the strokepopulation. Although this association is probablymediated principally through functional dependence,there may also be a direct causal link between braindamage and later depression. However, this possibilityrequires further examination in a prospective study.

Acknowledgements

This work was supported by grants from the Wellcome Trust (forDr M. Sharpe) and the Medical Research Council. The followingdepartments have collaborated in the OCSP: Oxford UniversityDepartment of Clinical Neurology (Dr P. Sandercock, Dr J.Bamford, Dr M. Dennis, Dr C. Warlow); Oxford UniversityDepartment of Neuro-Radiology (Dr A. Molyneux); OxfordUniversity Department of Community Medicine and GeneralPractice (Professor M. Vessey, Dr K. McPherson, Dr 0. Fowler,Ms L. Jones); Oxford University Department of Psychiatry(Dr K.Hawton, Dr A. House, Dr M. Sharpe). General practitionerscollaborating (name of liaison partner for each practice only): Dr A.McPherson, Oxford; Dr A. Markus, Thame; Dr D. Leggau, Oxford;Dr M. Agass, Berinsfield;Dr D. Otterburn,Abingdon; Dr S. Street,Kidlington; Dr V. Drury, Wantage; Dr R. Pinches, Abingdon;Dr N. Crossley, Abingdon; Dr H. O'Donnell, Deddington.Ms Hadine Joffe made helpfulcommentson the manuscript.

References

AH0, K., HARMSEN,P., HATANO,S., et a! (1980) Cerebrovasculardisease in the community; results of a WHO collaborative study.WHO Bulletin, 58, 113—130.

AMERICANPSYCHIATRICAssociATioN (1987) Diagnostic andStatistical Manual of Mental Disorders (3rd edn) (DSM—III).Washington, DC: APA.

BAKER, R. J. & NELDER, J. A. (1978) The Generahred LinearInteractive Modelling System (Release 3). Oxford: OxfordNumerical Algorithms Group.

B@swoan, J., SANDERcOCK,P., Der@rr.ns,M., eta!(1988)A prospectivestudy of acute cerebrovascular disease in the community: theOxfordshire Community Stroke Project 1981-86. I. MethOdOlOgy,demography and incident cases of first-ever stroke. Journal ofNeurology, Neurosurgery and Psychiatry, 51, 1373—1380.

, —¿� , et a! (1990) A prospective study of acute

cerebrovascular disease in the community. Journal of Neurology,Neurosurgery and Psychiatry, 53, 16-22.

BLAZER, D. (1989) Depression in the elderly. New England JournalofMedicine, 320, 164-166.

D@si,H., PEDERSON,H. E. & AHLOREN,P. (1989)Depressionamong patients with stroke. Ada Psychiatrica Scandinavica, 80,118—124.

EASTWOOD,M. R., RIFAT, S. L., Noses, H., ci a! (1989) Mooddisorder following cerebrovascular accident. British Journal ofPsychiatry, 154, 195—200.

ENDERBY, P. M., WOOD, V. A., WADE, D. T., ci a! (1987) TheFrenchayAphasia ScreeningTest: A short simpletest for aphasiasuitable for non-specialists. International Journal ofRehabilitation Medicine, 8, 166—170.

ForsmlN, M. F., FOLSTEIN,S. E. & MCHUGH,P. R. (1975) MiniMental State. Journal of Psychiatric Research, 12, 189—198.

HolsRooic, M. & SKILBECK, C. E. (1983) An activities index for use

with stroke patients. Age and Aging, 12, 166-170.HOUSE, A. (1987) Mood disorders after stroke: a review of the

evidence. International Journal of Geriatric Psychiatry, 2,211—221.

, DEr.muS, M., WARLOW, C., ci a! (1990) The relationship

between intellectual impairment and mood disorder in the firstyear after stroke. Psychological Medicine, 20, 805—814.

, —¿�, MOGRIDGE, L., et al(l99l) Mood disorders in the

year after first stroke. British Journa!ofPsychiatry, 158, 83—92.JOHNsoN, 0. A. (1991) Research into psychiatric disorder after

stroke: the need for further studies. Australian and New ZealandJournalofPsychiatry,25,358—370.

KOENIG,H. 0., MEADOR,K. 0., Comsi@i,H. J., ci al (1988)Depressionin elderlyhospitalizedpatientswithmedicalillness.Archives of Internal Medicine, 142, 1929—1936.

MAHONEY, F. & BARTHEL, D. (1965) Functional evaluation: theBarthel Index. Maryland State Medical Journal, 14, 61—65.

MAY0U, R. A., SEAGR0Arr, V. & GOLDACRE,M. (1991) Use ofpsychiatric services by patients in a general hospital. BritishMedical Journal, 303, 1029—1032.

MURPHY,E. (1982) Social origins of depression in old age. BritishJournal of Psychiatry, 141, 135—142.

NIEME, M., LAAKSONEN, R., KOTILA, M., ci a! (1988) Quality oflife 4 years after stroke. Stroke, 19, 1101—1107.

PARIKH, R. M., LIPsEY, J. R., ROBINsON, R. 0., et a! (1988) A twoyear longitudinal study of poststroke mood disorders: prognosticfactors related to one and two year outcomes. InternationalJournalofPsychiatryinMedicine,18,45—56.

ROBINSON, R. 0., LIPSEY, J. R., ci a! (1990) The impactof poststroke depression on recovery in activities of daily livingover a 2-year follow-up. Archives of Neurology, 47, 785-789.

PuLuCIN0, P., NELSON, R. F., KENDALL, B. E., ci a! (1980) Small

deep infarcts diagnoses on computed tomography. Neurology,30, 1090-1096.

REWLER, B. V., LARSON, E. & HANLEY, R. (1982) Coexistence ofcognitiveimpairmentand depressionin geriatricoutpatients.AmericanJournalofPsychiatry,139,623—626.

ROBINSON, R. 0., KUBO5, K. L., STARR, L. B., et a! (1983) Moodchanges in stroke patients: relationship to lesion location.Comprehensive Psychiatry, 24, 556—566.

STARR,L. B., KUBOs,K. L., et a! (1983) A two yearevaluation of post-stroke mood disorders: findings during theinitial evaluation. Stroke, 14, 736—741.

DEPRESSION AFTER STROKE

386 SHARPE ET AL

KUBOS, K. L., STARR, L. B., ci a! (1984a) Mood disorders

in stroke patients: importance of location of lesion. Brain, 107,

81—93.STARR, L. B. & PRICE, T. R. (1984b) A two year

longitudinal study of mood disorders following stroke: prevalenceand duration at six months follow-up. British Journal ofPsychiatry, 144, 256—262.

SCHWARTZ, J. A., SPEED, N. M., MOUNTZ, J. M., ci a! (1990)

@“¿�Tc-hexamethylpropyleneamineoxime single photon emissionCT in poststroke depression. American Journal of Psychiatry,147, 242—244.

SHARPE, M., HAwroN, K., HOUSE, A., et a! (1990) Mood disordersin long-term survivors of stroke: associations with brainlesion location and volume. Psychological Medicine, 20,815—828.

SINYOR, D., JACQUES, P., KALOUPEK, D. 0., ci a! (1986) Poststroke

depression and lesion location. An attempted replication. Brain,109, 537—546.

SPITZER, R. L., WILLIAMS, J. B. & GIBBON, M. (1986) Instruction

Manual for the Structured Clinical Interview for DSM—III-R.New York: Biometrics Research Department, New York StatePsychiatric Institute.

STARKSTEIN, S. E., RoBINsoN, R. 0., BERTHIER, M. L., ci a! (1988)

Depressivedisordersfollowingposteriorcirculationas compared

with middle cerebral artery infarcts. Brain, 111, 375—387.& (1989) Affective disorders and cerebral vascular

disease. British Journal of Psychiatry, 154, 170—182.WADE, D. T., LEGH SMITH, J. & HEWER, R. A. (1987) Depressed

mood after stroke. A community studyof its frequency.BritishJournal of Psychiatry, 151, 200—205.

* Michael Sharpe, MRCP, MRCPsych, Clinical Tutor, University Department of Psychiatry, Oxford; Keith

Hawton, DM, FRCPsych, Consultant Psychiatrist and Clinical Lecturer, Warneford Hospital and UniversityDepartment of Psychiatry, Oxford; Valerie Seagroatt, MSc, Statistician, Unit of Clinical Epidemiology,Department of Clinical Epidemiology and Primary Care, University of Oxford, Oxford; Allan House,DM, MRCP, MRCPsych, Consultant Psychiatrist, Department of Psychological Medicine, Leeds GeneralInfirmary, Leeds; Andrew Molyneux, FRCR, Consultant Neuroradiologist, Department of Radiology,Radcliffe Infirmary, Oxford; Peter Sandercock, DM, MRCP, Senior Lecturer, Department of ClinicalNeurosciences, Western General Hospital, Edinburgh; John Bamford, MD, MRCP, Senior NeurologicalRegistrar, St James University Hospital, Beckett Street, Leeds; Charlow Warlow, MD, FRCP, Professorof Medical Neurology, Department of Clinical Neurosciences, Western General Hospital, Edinburgh

*Correspondence: University Department of Psychiatry, Warneford Hospital, Oxford 0X3 7JX

(First received August 1992, final revision April 1993, accepted June 1993)