PLACEBO-CONTROLLED TREATMENT TRIALOF DEPRESSION IN ELDERLY PHYSICALLY

ILL PATIENTSMAVIS EVANS1�, MARGARET HAMMOND2, KENNETH WILSON3, MICHAEL LYE4 AND JOHN COPELAND5

1Consultant, Psychiatry of Old Age, Wirral and W Cheshire Community Healthcare NHS Trust, Bebington, Wirral, UK2Research Associate, University of Liverpool, Liverpool, UK

3Professor, Psychiatry of Old Age, University of Liverpool, Liverpool, UK4Professor, Geriatric Medicine, University of Liverpool, Liverpool, UK

5Professor, Psychiatry, Institute of Human Ageing, University of Liverpool, Liverpool, UK

ABSTRACT

Objectives. To determine the response of physically ill elderly depressed patients to treatment.Design. Acute geriatric medical inpatients with depression, randomly assigned to an 8-week double-blind placebo-

controlled trial of ¯uoxetine.Main outcome measure. Response rate as de®ned by the 17-item Hamilton Depression Rating Scale.Results. Eighty-two patients entered the trial; 62 patients (all those who had completed at least 3 weeks of

treatment) were included in the e�cacy analysis. Forty-two completed the full 8 weeks (21 in each group) withresponse rates of 67% in the ¯uoxetine group and 38% in the placebo group. No signi®cant di�erence was foundbetween the responses of the two groups ( p � 0:12). There was a trend for results in the ¯uoxetine group to continueto improve with time. On secondary analysis those patients with serious physical illness who completed 5 or moreweeks (N � 37) showed a signi®cant improvement in mood if treated with ¯uoxetine ( p � 0:02).

Conclusions. The main bene®t of antidepressants is to approximately double the chances of recovery. Thistrial showed the response rate of the ¯uoxetine treated group was increased by a factor of 1.8 over the placebo groupin an 8-week period. The presence of physical illness, often severe and/or multiple, did not reduce the e�ectivenessof the medication, which was well tolerated overall. Those with serious physical disease responded signi®cantlybetter to drug treatment; this will require further work. Psychological support was also considered to be important.# 1997 John Wiley & Sons, Ltd.

Int. J. Geriat. Psychiatry, 12, 817±824 (1997)No. of Figures: 0. No. of Tables: 4. No. of References: 45.

KEY WORDSÐold age; depression; physically ill; treatment; liaison psychiatry; ¯uoxetine

Depression in elderly people is strongly correlatedwith physical illness, especially chronic progressivedisease (Gurland et al., 1983; Murphy, 1983) andthere is evidence that the over-75s are especiallyvulnerable (Koenig et al., 1988; Harper et al.,1990). The prevalence of depression in elderlymedical inpatients is over 20% (O'Riordan et al.,1989; Burn et al., 1993; Hammond et al., 1993),approximately twice that found in the community

(Copeland et al., 1987; Lindesay and Murphy,1989). The prevalence of physical problems ishighÐover 80% of people aged 70 years or moresu�er from a signi®cant (i.e. in need of treatment)physical illness (Rossman, 1979) and over 50%have a disorder which interferes with independentfunctioning (Jarvik and Perl, 1981). Murphy'sstudy (1982) of community and hospital mentallyill patients found that approximately 40% hadchronic physical illnesses.

Outcome studies of treatment of depression inthe elderly have shown response to be generallygood but have emphasized the association betweenpoor response to treatment and serious physicalillness (Murphy, 1983; Baldwin and Jolley, 1986;

�Correspondence to: Mavis Evans, Wirral and West CheshireCommunity NHS Trust, Clatterbridge Site, Bebington, WirralL63 4JY, UK.

Contract grant sponsor: Lilly Industries Limited; Contractgrant number: B1Y BP HC55.

CCC 0885±6230/97/080817±08$17.50 Received 22 October 1996# 1997 John Wiley & Sons, Ltd. Accepted 27 January 1997

INTERNATIONAL JOURNAL OF GERIATRIC PSYCHIATRY, VOL. 12: 817±824 (1997)

Meats et al., 1991). Mortality among the depressedis at least double that expected, even when con-trolled for physical illness and age (Cole, 1983;Murphy et al., 1988; Evans, 1993a). Physically illdepressed patients are more likely to be admitted tohospital than those not depressed (Kay et al., 1964)and they have increased inpatient mortality andsigni®cantly more days in hospital, both during theindex and subsequent admissions (Koenig et al.,1989b). Psychiatric intervention has been shown toincrease recovery rate, reduce duration of hospitalstay and the need for residential care afterdischarge and therefore also reduce costs (Levitanand Kornfeld, 1981; Verbosky et al., 1993).

Confusion over the cause of somatic symptomssuch as anorexia, insomnia and anergia mayconfound the diagnosis. It is often assumed thatdepression in the presence of physical illness isreactive and therefore less amenable to medication,and will resolve with treatment of the medicalcondition. Concerns about multiple drug therapyand potential side-e�ects in a frail populationmay persuade many physicians against attemptingantidepressant therapy. Anticholinergic andcardiotoxic side-e�ects may be a contraindicationfor the use of tricyclic antidepressants or may causeunacceptable side-e�ects (Koenig, 1990). Koeniget al. (1992) concluded that the disinclination ofphysicians to treat was due to their worries over therisk±bene®t ratio of known therapies.

Treatment trials which do not exclude peoplewith acute or unstable chronic physical conditionsare rare, and treatment trials in speci®c physicalillnesses have had di�ering results. Doxepin wasfound to be no better than placebo in a small groupof depressed patients with chronic obstructiveairways disease (Light et al., 1986). House et al.(1991) concluded that persistent depression requir-ing intervention was rare in the ®rst year afterstroke, although a report of two placebo-controlledstudies (Fedoro� and Robinson, 1989) showedsigni®cant enhancement of both physical andcognitive recovery as well as of emotional statewith active treatment. Koenig et al. (1989a), in anattempted placebo-controlled trial with nortrip-tyline, found that 90% of elderly acutely physicallyill patients meeting the criteria for depression hadmajor or minor medical contraindications to theuse of tricyclic antidepressants; this trial was dis-continued after 10 months' duration as only sevenpatients had been randomized. Evans (1993a) con-ducted an open treatment study with the selectiveserotonin reuptake inhibitor (SSRI) ¯uoxetine of

patients diagnosed as depressed during inpatienttreatment for physical illness, and found no prob-lems with tolerability or side-e�ects. Remission orimprovement in depression was noted on follow-upin all treated patients.

Elderly patients with mild to moderate cognitiveimpairment are usually excluded from antidepres-sant trials (Burvill et al., 1991), although there is asigni®cant correlation between cognitive impair-ment and depression in elderly ill patients (Sadavoyet al., 1990). Cognitive impairment has beenimplicated, along with active physical illness, inpromoting chronicity in depression (Alexopoulosand Chester, 1992).

The exclusion of both the physically ill and thecognitively impaired from studies of depressionmeans that drug treatment in the very people whomprimary care workers and general physicians mostoften meet has not been adequately assessed.To assess the e�ciency of drug treatment fordepression in elderly inpatients, most of whomhad at least one acute and one or more chronichealth problems including cognitive impairment, arandomized placebo-controlled trial of SSRI treat-ment was undertaken.

METHOD

Men and women aged 65 years and over admittedunder the care of a geriatrician or general physicianto the Royal Liverpool University Hospitalbetween July 1992 and December 1993 werede®ned as eligible for the study. Hospital casenotes were examined for exclusion criteria resultingin approximately 35% of patients being eliminatedat this stage. Exclusion criteria were as follows:suicidal intent or severe depression requiring ECT,other serious mental illness or already receivingpsychotropic medication other than hypnotics,and unstable epilepsy. Patients had to be able tounderstand the purpose of the study and giveinformed consent, thus patients with severe cogni-tive impairment (MMSE510; Folstein et al., 1975)were excluded.

The method chosen was a double-blind, random-ized parallel-group study. Patients meeting entrycriteria were randomly allocated within blocks offour to receive either placebo or ¯uoxetine 20 mgin the morning for a target period of 8 weeks.The primary outcome variable was the severity ofdepression measured using the 17-item HamiltonDepression Rating Scale (HAMD), with a good

INT. J. GERIAT. PSYCHIATRY, VOL. 12: 817±824 (1997) # 1997 John Wiley & Sons, Ltd.

818 M. EVANS ET AL.

response de®ned as a reduction in score of50% and/or a ®nal score of 10 or less (Hamilton,1960).

Patients were assessed by the same researchworker throughout the trial. They were screenedfor depression using the Evans Liverpool Depres-sion Rating Scale (ELDRS) (Evans, 1993b), a15-item interview and observer-rated screeningscale which has been validated for use with thehospitalized physically ill elderly. Patients witha score of 5 or more were interviewed usingthe Geriatric Mental State (GMS)/AGECAT(Copeland et al., 1976, 1986), a semi-structuredcomputerized interview and diagnostic package.The criterion variable for entry to the study was aGMS±AGECAT case level of depression with aduration of at least 4 weeks.

Details of family and personal history of depres-sion, duration of current depression, days inhospital, concomitant medication, use of servicesand social circumstances were recorded. Physicalhistory and examination was extracted from allavailable medical case notes, with results oflaboratory tests and other investigations. Physicalillnesses were rated as mild, moderate or severe,acute or chronic (Burvill, 1990). Ratings were doneon entry to the study, relapse or new illness wasrecorded as it occurred as an adverse event.

Informed consent was obtained from all patientsin the study. General practitioners were informedby letter when the patient was entered into the trial,and were informed of outcome and treatmentrecommendations at the end of 8 weeks. Ethicalapproval for the study was given by the RoyalLiverpool University Hospital Ethical Committee.

Patients were assessed at weeks 1, 3, 5 and 8either in hospital or at home after discharge.Depressive symptoms were evaluated and scoredusing the HAMD; pulse and blood pressure,medication and contact with medical professionalswere also recorded; details of any adverse experi-ences were sought from the patient using non-leading questions. At the 8-week visit, patients werereinterviewed using the GMS to obtain anAGECAT diagnosis and the MMSE was repeated.Patients were examined a ®nal time by thepsychogeriatrician after all trial ratings had beencompleted, to assess the need for further treatmentor follow-up of their depressive illness.

A minimum (alpha � 0:05, power � 0:8) sam-ple size of 58 patients completing the study pergroup with a placebo response of 45% and atreatment response of 70% was calculated before

the trial began. Parametric tests were used fornormally distributed interval scale data (paired andunpaired t-tests) and non-parametric tests forcategorical or ordinal scale and non-normallydistributed data (Fisher's exact test, Wilcoxonsigned-rank test within groups, Wilcoxon ranksum test between groups). All tests were two-tailedand signi®cance was de®ned as p < 0:05.

Patients were included in the `intention to treat'e�cacy analysis if they had completed at least3 weeks of treatment (ie visit 3) (N � 62), as it wasconsidered that any di�erence in e�cacy between¯uoxetine and placebo after only 1 week of treat-ment would not be valid. A subgroup of patientswith serious physical illness was also analysed.Serious physical illness was de®ned from theBurvill ratings as cardiac or respiratory diseaserated as moderate or severe on entry to the trial ordiagnosis of neoplastic disease, i.e. illness involving`life threat' to the patient.

RESULTS

During the recruitment period approximately 3000patients were admitted to the target wards. Due torapidity of throughput and logistics, 530 wereprescreened for willingness to take part andexcluded if terminally ill or with a communicationproblem; 323 were screened and 161 were given theGMS interview. One hundred and forty patientswere diagnosed as su�ering from depression byGMS±AGECAT, of whom 58 (mean age 79 years,SD 7.3) were not included in the trial for thefollowing reasons: refusal (25), physician's decisionto prescribe open treatment (15), medical contra-indication (5), concomitant medication exclusion(5), depression considered to be resolving or due todigoxin toxicity or hypothyroidism (5), and logist-ical problems such as discharge before consent wasobtained (3). There was no signi®cant di�erence inseverity of depression in those who were notincluded in the trial.

Eighty-two patients (20 men and 62 women,mean age 80.4 years, SD 6.6) were randomized toplacebo or treatment groups; 76 had completebaseline data. Patients in the study were predomin-antly inner city residents on low income; 12% weremarried, 72% lived alone; 16% lived in shelteredhousing, residential or nursing homes. There wasno signi®cant di�erence in age, sex or severity ofillness between the two groups (Table 1), andthere were also no signi®cant di�erences in these

# 1997 John Wiley & Sons, Ltd. INT. J. GERIAT. PSYCHIATRY, VOL. 12: 817±824 (1997)

RESPONSE TO TREATMENT OF PHYSICALLY ILL DEPRESSED PATIENTS 819

variables between those patients who withdrewearly from the study and those who were enteredfor analysis. Patients had on average two acutephysical problems on admission and three chronichealth problems (Table 2).

Fourteen patients (seven ¯uoxetine, seven place-bo) had previously been prescribed antidepressantmedication; two (both placebo) felt they hadrecovered with the treatment.

Sixty-two patients completed 3 weeks of the trialand entered e�cacy analysis. They were takingamong them 166 di�erent types of medicationon a regular basis (median 6.0 ¯uoxetine group,5.0 placebo group). There was no change in thenumber of drugs taken in the ¯uoxetine group(N � 29, p � 0:98), but there was a signi®cantrise in the placebo group (N � 33, p � 0:001)(Wilcoxon signed-rank test). The di�erencebetween groups with respect to the change wasalso signi®cant (Wilcoxon test p � 0:005).

Forty-two patients completed the trial. Fourpatients in each group died, two in the placebogroup were withdrawn due to lack of e�cacy atweek 5. Fifty per cent of all withdrawals occurredin the ®rst 3 weeks, 30% in week 1. Patients werede®ned as responders or non-responders accordingto their HAMD score at 8 weeks (those whocompleted) or at the last recorded visit after 3 weeksof treatment (Table 3). In the e�cacy analysis the¯uoxetine group showed 55% response comparedwith 36% of the placebo group. In those complet-ing 8 weeks, the ®gures were 67% and 38% respect-ively. There was no signi®cant di�erence betweengroups although there was a trend towards a betterresponse in the ¯uoxetine group at 8 weeks, withthe number of responders in the ¯uoxetine groupincreasing with time. Percentage of responders inthe placebo group remained static after 5 weeks.

A secondary analysis of patients with seriousphysical illness showed that of those completing5 weeks or more of treatment (N � 37), thosereceiving ¯uoxetine were signi®cantly more likelyto recover from their depression: 63% on ¯uoxetinecompared with 24% on placebo, p � 0:023.

The Mini Mental State was also recorded atbaseline and at the last assessment. There wasno signi®cant di�erence between the two groupsof those who completed (N � 42) at baseline(¯uoxetine group median 23, interquartile range21±24; placebo group median 23, interquartile

Table 1. Patients with complete baseline data, N � 76

Fluoxetine (N � 38) Placebo (N � 38)

Interquartile Interquartile

Median range Median range

Age (yr) 82 75±85 81 76±85

Sex 31 female 28 female

Hospital stay 8 6±14 7 5±12

(days)

HAMD 20.5 18±24 21 20±24

(completers)

Length of current episode (months)

56 9 14

6±12 9 6

12� 20 18

Table 2. Acute and chronic illnesses in the patientsentered into the trial (most patients had multiplepathologies)

Fluoxetine Placebo

(N � 39) (N � 43)

Cardiovascular 34 35

Respiratory 16 22

Cerebrovascular/neurological 14 14

Musculoskeletal 19 21

Malignancies 6 6

Urogenital 11 15

Endocrine/haematological 12 10

Gastrointestinal 16 22

Table 3. Percentage response (number in group) as de®ned by HAMD total

Illness Whole group �Serious physicalFluoxetine Placebo Probability� Fluoxetine Placebo Probability*

3 weeks 48 (29) 33 (33) 0.30 53 (19) 29 (24) 0.21

5 weeks 43 (23) 44 (27) 1.00 53 (15) 38 (21) 0.50

8 weeks 67 (21) 38 (21) 0.12 64 (14) 24 (17) 0.033

LV 3±8 wk 55 (29) 36 (33) 0.20 58 (19) 25 (24) 0.058

LV 5±8 wk 63 (24) 37 (27) 0.095 63 (16) 24 (21) 0.023

LV, last visit brought forward, ie survival analysis.�Fisher's exact test, 2-tailed.

INT. J. GERIAT. PSYCHIATRY, VOL. 12: 817±824 (1997) # 1997 John Wiley & Sons, Ltd.

820 M. EVANS ET AL.

range 19±27) or at 8 weeks (¯uoxetine groupmedian 26, interquartile range 22±27; placebogroup median 23, interquartile range 19±27), buta trend to a rise in score in the ¯uoxetine group wasseen ( p � 0:06).

Two hundred and sixty-four adverse events wererecorded for all patients, 130 in the ¯uoxetinegroup and 134 in the placebo group (Table 4). Inpatients with gastrointestinal diagnoses at baseline,there were signi®cantly more adverse events inthe ¯uoxetine group for all events at all levels ofseverity (32 ¯uoxetine, 19 placebo, p � 0:049).There were no other signi®cant di�erences betweenthe two drug groups.

No patient reported the emergence of suicidalideation during the trial, although one patient(placebo group) was withdrawn by the supervisingpsychogeriatrician (ME) due to worsening depres-sion.

Two possible drug interactions were recorded;both (one placebo, one ¯uoxetine) were patients onwarfarin whose anticoagulant control altered. Fourpatients in each group were on warfarin; no othershad any di�culties in anticoagulant control.

DISCUSSION

The patients in this study were identi®ed while theywere inpatients on a medical ward. The physicallyill sample and the inclusion of those with moderatecognitive impairment make this trial directlyrelevant to clinical practice. The number complet-ing was below our required target. Estimatedrecruitment rate (35 every 3 months) was basedupon a prior open treatment study with a similarpatient population using the same drug (Evans,1993a) and failed to take into account pre- andpost-randomization losses involved in clinical trialsusing placebos (physician and patient refusal,protocol requirements) which necessitate identify-ing at least twice the number needed (Charlesonand Horwitz, 1984). The 50% attrition rate seenin this trial appears to be normal for 8-weekantidepressant trials with elderly subjects (Feighneret al., 1990; Tiller et al., 1990).

A placebo group was important for this study,both to answer the question of necessity fortreatment, but also to assess adverse e�ects in thisalready ill population: `it is particularly di�cult toassess adverse e�ects without a placebo group sincemany symptoms that are rated as side e�ects maysimply be symptoms of depression' (Rockwell et al.,

1988). The adverse events recorded appear to showthe expected high rate of gastrointestinal symptomsfrom an SSRI until the similar ®gure in the placebogroup is noted. The only signi®cant di�erence wasin the number of infections recorded between thegroups (chi-squared � Yates correction p < 0:05)but this was felt to be a chance ®nding.

The response rate to ¯uoxetine, measured by theHAMD, during the 8-week trial period is similar tothat found in most antidepressant trials (Georgotasand McCue, 1989) despite the high physicalmorbidity of this group. Placebo response washigh but follow-up was very close (at least six visitsin 8 weeks), and some psychological support andcounselling was inevitable. Response rate in thetreatment group increased with time throughoutthe trial; this delay in response to antidepressanttreatment in the elderly has already been described(Georgotas andMcCue, 1989); physical illness mayalso be a factor in delaying response. This may be afactor in the therapeutic nihilism prevailing amongphysicians treating this age group.

The group of patients entered into this trial washeterogeneous, with di�ering physical conditionsrepresenting the reality of clinical practice. Acuteillnesses recorded as reason for admission variedfrom myocardial infarction or cardiac failure toconstipation or falls. Our results have shown thatthere are subgroups within the whole groupwith di�ering response to treatment. All patientsreceived similar visits and support from theresearch team, but those with serious physicalillness as de®ned earlier appeared to bene®t fromthe addition of active antidepressant medication,unlike other outcome studies (Murphy, 1983;

Table 4. Adverse events reported

Fluoxetine Placebo

(N � 39) (N � 43)

Anxiety/agitation 9 6

Gastrointestinal 20 18

Anorexia 3 0

Headaches 7 4

Myalgia 4 0

Arrhythmias 6 2

Dyspnoea 4 5

Pruritis 1 3

Dizziness/vertigo 4 4

Accidents/falls 7 5

Infections 7 21

Other 58 66

Total 130 134

# 1997 John Wiley & Sons, Ltd. INT. J. GERIAT. PSYCHIATRY, VOL. 12: 817±824 (1997)

RESPONSE TO TREATMENT OF PHYSICALLY ILL DEPRESSED PATIENTS 821

Baldwin and Jolley, 1986; Meats et al., 1991).However, these studies are not directly compar-able: Murphy's study was of community patients,Baldwin and Jolly's a retrospective review ofpsychiatric patients and Meats et al.'s a prospectivereview of psychiatric admissions.The elderly have a tendency not to admit to

feelings of depression and relatives may be unawareof the condition (Hanley and Baikie, 1984). Theyaccept low levels of psychological well-being as`normal' and have a disturbingly low expectation ofquality of life. All illnesses except the very trivialinvolve an element of psychological adjustment.Serious medical illness is likely to be a potentpsychological stressor a�ecting body image, self-esteem, sense of identity, capacity to live inde-pendently and to maintain social and familyrelationships (Rodin and Voshart, 1986), thereforetreatment needs to involve psychological support aswell as medication. This study design with frequentfollow-up visits by the same investigator allowedpsychological support to occur. However, thosewith serious medical illness on ¯uoxetine showed asigni®cantly improved response over those onplacebo. Psychological support did not appear ase�ective in the severely ill without active anti-depressant treatment.

It may be postulated that depression in physicallyill patients is reactive. Where the illness is minor ortreatable, the response to antidepressants is similarto that to placebo as treatment of the physicalproblem will have removed the maintaining factorfor the depression. Where the illness is serious, theuse of antidepressants shows a signi®cant responseover placebo, treatment of the physical illness beingable to modify rather than cure.There were eight deaths in the 82 patients

randomized to the trial, four in the ®rst week.The average death rate for a population of this agewithout physical or psychiatric illness is approx-imately 5% per year (Murphy, 1983). However, theincrease in mortality in patients with both depres-sion and physical illness is well documented(Murphy, 1983; Koenig et al., 1989b; Evans,1993a). Depression may a�ect the course andoutcome of, or the reaction to, the physical illnessas much as it a�ects the quality of life (Koenig,1990). Although the follow-up period in this studywas too short to demonstrate any di�erences inmortality between the groups, the e�ect of success-ful treatment with SSRIs on mortality over a12-month period has been shown in open treatmentof a similar population (Evans, 1993a) to reduce

the rate from the signi®cantly higher levels of thedepressed and physically ill back to the rate of thenon-depressed physically ill. Treatment with SSRIscan be considered safer than leaving the conditionuntreated.

Seventy-two per cent of patients entering thestudy dated their onset of symptoms to over3 months and a third to over 1 year. As fewpatients had previously considered their symptomsto be due to a treatable condition, the lack ofprevious psychiatric treatment is not surprising.

The improvement in MMSE score is moreobvious in the ¯uoxetine group. Fluoxetine is anSSRI, thus it increases the level of this particularmonoamine in the synaptic cleft. Serotonin canhave an alerting e�ect and thus may improvescoring through increased attention and concen-tration, there may be a direct e�ect on cognition, orthe greater percentage of patients recovering fromtheir depression on active treatment may have hada direct e�ect. Future research in this area will behelpful.

Series (1992) considers the main bene®t of drugtreatment is to approximately double the chances ofrecovery from depression with possible associatedimprovements in the physical condition. This trialshows the response rate of the ¯uoxetine treatedgroup was increased by a factor of 1.8 over theplacebo group. The presence of physical illness,often severe or multiple, did not reduce thee�ectiveness of the medication, which was welltolerated overall; indeed, it appeared to increase theresponse in those who completed 8 weeks. A trial ofnortriptyline in a similar population (Koenig et al.,1989a) had extreme di�culties with recruitment dueto medical contraindications to the tricyclic anti-depressant which were not a problem in this study.

This study has shown that drug treatment ofdepression in the physically ill with ¯uoxetine iswell tolerated. Psychological treatment and sup-port is also of bene®t. The placebo group appearedto continue to complain of somatic symptoms,showing a signi®cant rise in the number of con-comitant medications taken overall. It is possiblethe `placebo e�ect' may only be seen in the psycho-logical symptoms of a depressive illness, not thesomatic ones. This also deserves further study.

ACKNOWLEDGEMENTS

Thanks to Jackie Turner at Pharmakopius Inter-national, Reading, for statistical support; and to

INT. J. GERIAT. PSYCHIATRY, VOL. 12: 817±824 (1997) # 1997 John Wiley & Sons, Ltd.

822 M. EVANS ET AL.

Lilly Industries Limited for funding the study,number B1Y BP HC55.

REFERENCES

Alexopoulos, G. S. and Chester, J. G. (1992) Outcomesof geriatric depression. Clin. Geriatr. Med. 8, 363±376.

Baldwin, R. C. and Jolley, D. J. (1986) The prognosis ofdepression in old age. Brit. J. Psychiat. 149, 574±583.

Burn, W. K., Davies, K. N., McKenzie, F. R. andBrothwell, J. A. (1993) The prevalence of psychiatricillness in acute geriatric admissions. Int. J. Geriatr.Psychiat. 8, 171±174.

Burvill, P. W. (1990) Quanti®cation of physical illness inpsychiatric research in the elderly. Int. J. Geriatr.Psychiat. 5, 161±170.

Burvill, P. W., Hall, W. D., Stampfer, H. G. andEmmerson, J. P. (1991) The prognosis of depressionin old age. Brit. J. Psychiat. 158, 64±71.

Charleson, M. E. and Horwitz, R. I. (1984) Applyingresults of randomised trials to clinical practice: Impactof losses before randomisation. Brit. Med. J. 289,1281±1284.

Cole, M. G. (1983) Age, age of onset, and course ofprimary depressive illness in the elderly. Can.J. Psychiat. 28, 102±104.

Copeland, J. R. M., Dewey, M. E. and Gri�ths-Jones,H. M. (1986) A computerised diagnostic system andcase nomenclature for elderly subjects: GMS andAGECAT. Psychol. Med. 16, 89±99.

Copeland, J. R. M., Dewey, M. E., Wood, N., Searle, R.,Davidson, I. A. and McWilliam, C. (1987) Range ofmental illness among the elderly in the community:Prevalence in Liverpool using the GMS±AGECATpackage. Brit. J. Psychiat. 150, 815±823.

Copeland, J. R. M., Kelleher, M. J., Kellett, J. M.,Gourlay, A. J., Gurland, B. J., Fleiss, J. L. and Sharpe,L. (1976) A semi-structured clinical interview for theassessment of diagnosis and mental state in the elderly.The Geriatric Mental State Schedule I: Developmentand reliability. Psychol. Med. 6, 439±449.

Evans, M. E. (1993a) Depression in elderly physicallyill inpatients: A 12-month prospective study. Int.J. Geriatr. Psychiat. 8, 587±592.

Evans, M. E. (1993b) Development and validation of ascreening test for depression in the elderly physicallyill. Int. Clin. Psychopharmacol. 8(4), 329±331.

Fedoro�, J. P. and Robinson, R. G. (1989) Tricyclicantidepressants in the treatment of post-stroke depres-sion. J. Clin. Psychiat. 50(7), 18±23.

Feighner, J. P., Boyer, W. F., Hendrickson, G. G.,Pambekian, R. A. and Doroski, V. S. (1990) Acontrolled trial of adinazolam vs desipramine ingeriatric depression. Int. J. Clin. Psychopharmacol. 5,227±232.

Folstein, M. F., Folstein, S. E. andMcHugh, P. R. (1975)`Mini Mental State'. A practical method for grading

the cognitive state of patients for the clinician.J. Psychiat. Res. 12, 189±198.

Georgotas, A. and McCue, R. E. (1989) The additionalbene®t of extending an antidepressant trial past7 weeks in the depressed elderly. Int. J. Geriatr.Psychiat. 4, 191±195.

Gurland, B. J., Copeland, J. R. M., Kelleher, M. J.,Kuriansky, J., Sharpe, L. andDean, L. (1983)TheMindandMood of Ageing: TheMental Health Problems of theCommunity Elderly in New York and London. HaworthPress, New York; Croom Helm, London.

Hamilton, M. (1960) A rating scale for depression.J. Neurol. Neurosurg. Psychiat. 23, 56±62.

Hammond, M., Evans, M. E. and Lye, M. (1993)Depression in medical wards. Int. J. Geriatr. Psychiat.8, 957±958.

Hanley, I. and Baikie, E. (1984) Understanding andtreating depression in the elderly. In PsychologicalApproaches to the Care of the Elderly (I. Hanley andJ. Hodge, Eds). Methuen, New York.

Harper, R. G., Kotik-Harper, D. and Kirby, H. (1990)Psychometric assessment of depression in an elderlygeneral medical population. J. Nerv. Ment. Dis. 178,113±119.

House, A., Dennis, M., Mogridge, L., Warlow, C.,Hawton, K. and Jones, L. (1991) Mood disorders inthe year after ®rst stroke. Brit. J. Psychiat. 158, 83±92.

Jarvik, L. and Perl, M. (1981) Overview of physiologicdysfunction related to psychiatric problems in theelderly. In Neuropsychiatric Manifestations of PhysicalDisease in the Elderly (A. J. Levenson and R. C. W.Mill, Eds). Raven Press, New York.

Kay, D.W. K., Beamish, R. and Roth, M. (1964) Old agemental disorders in Newcastle upon Tyne. Part I.A study of prevalence. Brit. J. Psychiat. 110, 146±158.

Koenig, H. G. (1990) Reply to: depressed or just sick.Arch. Intern. Med. 150, 1349±1350.

Koenig, H. G. and Breitner, J. C. S. (1990) Use ofantidepressants in medically ill older patients. Psycho-somatics 31(1), 22±32.

Koenig, H. G., Goli, V., Shelp, F., Kudler, H. S., Cohen,H. J. and Blazer, D. G. (1992) Major depression inhospitalised medically ill older men: Documentation,management and outcome. Int. J. Geriatr. Psychiat. 7,25±34.

Koenig, H. G., Goli, V., Shelp, F., Kudler, H. S., Cohen,H. J., Meador, K. G. and Blazer, D. G. (1989a)Antidepressant use in elderly medical inpatients:Lessons from an attempted clinical trial. J. Gen. Int.Med. 4, 498±505.

Koenig, H. G., Meador, K. G., Cohen, H. J. and Blazer,D. G. (1988) Self rated depression scales and screeningfor major depression in the older hospitalised patientwith medical illness. J. Am. Geriatr. Soc. 36, 699±706.

Koenig, H. G., Shelp, F., Goli, V., Cohen, H. J. andBlazer, D. G. (1989b) Survival and health careutilisation in elderly medical inpatients with majordepression. J. Am. Geriatr. Soc. 37, 599±606.

# 1997 John Wiley & Sons, Ltd. INT. J. GERIAT. PSYCHIATRY, VOL. 12: 817±824 (1997)

RESPONSE TO TREATMENT OF PHYSICALLY ILL DEPRESSED PATIENTS 823

Levitan, S. J. and Kornfeld, D. S. (1981) Clinical andcost bene®ts of liaison psychiatry. Am. J. Psychiat.138(6), 790±793.

Light, R. W., Merrill, E. J., Despers, J., Gordon, G. H.and Mutalpassi, L. R. (1986) Doxepin treatment ofdepressed patients with chronic obstructive pulmonarydisease. Arch. Intern. Med. 146, 1377±1380.

Lindesay, J. and Murphy, E. (1989) Dementia, depres-sion and subsequent institutionalisationÐthe e�ect ofhome support. Int. J. Geriat. Psychiatr. 4, 3±9.

Meats, P., Timol, M. and Jolley, D. (1991) Prognosis ofdepression in the elderly. Brit. J. Psychiat. 159,659±663.

Murphy, E. (1982) Social origins of depression in oldage. Brit. J. Psychiat. 141, 135±142.

Murphy, E. (1983) The prognosis of depression in oldage. Brit. J. Psychiat. 142, 111±119.

Murphy, E., Smith, R., Lindesay, J. and Slattery, J.(1988) Increased mortality rates in late-life depression.Brit. J. Psychiat. 152, 347±353.

O'Riordan, T. G., Hayes, J. P., O'Neill, D., Shelley, R.,Walsh, J. B. and Coakley, D. (1989) The e�ect of mildto moderate dementia on the Geriatric DepressionScale and on the General Health Questionnaire. AgeAgeing 19, 57±61.

Rockwell, E., Lam, R. and Zisook, S. (1988) Anti-depressant drug studies in the elderly. Psychiat. Clin.N. Am. 11, 215.

Rodin, G. and Voshart, K. (1986) Depression in themedically ill, an overview. Am. J. Psychiat. 143,696±705.

Rossman, I. (1979) Mortality and morbidity overview.In Clinical Geriatrics (I. Rossman, Ed.). Lippincott,Philadelphia.

Sadavoy, J., Smith, I., Conn, D. K. and Richards, B.(1990) Depression in geriatric patients with chronicmedical illness. Int. J. Geriatr. Psychiat. 5, 187±192.

Series, H. G. (1992) Drug treatment of depression inelderly medically ill patients. J. Psychosom. Res. 36,1±16.

Tiller, J., Maguire, K. and Davies, B. (1990) A sequentialdouble-blind controlled study of moclobemide andmianserin in elderly depressed patients. Int. J. Geriatr.Psychiat. 5, 199±204.

Verbosky, L. A., Franco, K. N. and Zrull, J. P. (1993)The relationship between depression and length of stayin the general hospital patient. J. Clin. Psychiat. 54,177±181.

INT. J. GERIAT. PSYCHIATRY, VOL. 12: 817±824 (1997) # 1997 John Wiley & Sons, Ltd.

824 M. EVANS ET AL.