001 screening detection and management of depression in elderly primary care attenders part 1

8/3/2019 001 Screening Detection and Management of Depression in Elderly Primary Care Attenders Part 1

1/7

Family Practice Oxford University Press 1994

Vol . 11 . No . 3Printed in Great Britain

Screening, De tection andM a n a g e m e n t of Depression inElderly Primary Care Attenders.I: The Acceptability and Performanceof the 15 Item Geriatric DepressionScale (GDS15) and the Developmentof Short VersionsPENNY D'ATH ,* PHILIPPA KA TO N A ," ELEANOR MUL LA N,** SANDRA EVANS* ANDCORNELIUS KATONA* tD'Ath P, Katona P, Mullan E, Evans S and Katona C. Screening, detection and management of depres-sion inelderly primary care attenders. I: The acceptability and performance of the 15 item G eriatric Depres-sion Scale (GDS15) and the development of short versions. Fam ily Practice 1994; 11 : 260-266.One-hundred and ninety-eight elderly subjects attending their general practitioners (GPs) were asked tocomplete the 15 item Geriatric Depression Scale (GDS15). Analysable results were obtained from 194(98%). Of these, 67 (34%) scored above the GDS15 cut-off (4/5) for significant depressive symp-tom atolog y. 8 7.6% found the questionnaire to be acceptable and only 3.6% found it very difficult orverystressful. The GDS1 5 had a high level of internal consistency (Cronbach's alpha = 0.80). All the individualitems of the GDS15 associated significantly (P < 0.01) with total score and 'caseness'. A single question"d o you feel that your life is empty?" identified 84% of 'cases'. In an attempt to devise short scales toscreen elderly primary care patients fordepression, the data were subjected to logistic regression analysis.Ten (GDS10), four (GDS4) and one (GDS1) item versions were generated. Agreement between thesesho rt scales and the GDS1 5 in the original sample was 95, 91 and 79% respectively. C ronbach's alpha was0.72 for the GDS1 0 and 0.55 for the GD S4. The sho rt scales were then validated in an independent sampleof 120 patients in whom both GDS data and the results of a detailed psychiatric interview (the GeriatricMen tal Status Schedule, GMS) w ere available. The sensitivity and specificity of the GDS10 against GMScaseness were 87 and 77% (cut-off 3/4); those of the GDS4 were 89 and 65% (cut-off 0/1) and 61 and8 1 % (cut-off 1/2). Sensitivity and specificity for the GDS1 were 59 and 75% . It is concluded that theseshort scales may be useful in helping GPs and practice staff to identify elderly patients with significantdepressive sympto ms.

INTRODUCTIONDepression in old age carries a poor prognosis' with in-creased use of health and social service facilities2 and Psychiatric W ing, Princess Alexandra Hosp ital, Hamstel Ro ad,Harlow, Essex, Lower Clapton Health Centre, Lower ClaptonRoad, London E5, * Department of Psychiatry, University CollegeLondon Medical School, Wolfson Building, Middlesex Hospital, Lon-don W1N 8AA and J St Bartholomew's Hospital M edical Schoo l, UK.Correspondence to C.K. at UCLMS.

excess mortality3 as well as frequent chronicity of thedepressive symptoms.4 However, while depressivesymptoms are common in elderly patients attendingtheir general practitioners (GPs),5'6 depression isseldom recorded in the GP case notes even where aclear diagnosis can be made.7 It is also clear thatdepression in old age, as identified through communitysurveys, is seldom treated.4'8. These findings suggestthat depression in old age may be both underdetectedand undertreated in primary care.260


2/7

DEPRESSION IN ELDERLY PRIMARY CARE ATTENDERS, I 261In a younger population, recognition of depressionby the GP improves outcome even when the patientdoes not comply with treatment.' Validating apatient's feelings by making a formal diagnosis mayitself have therapeutic potential. The possible con-tribution of a short questionnaire, acceptable to bothGPs and their patients, that would improve detectionrates and facilitate a higher rate of effective inter-

vention, is thus considerable. Screening for suchdepressive symptoms would be a legitimate part of themandatory screening of patients aged 75 and over,which now forms part of the GP contract.10The Geriatric Depression Scale (GDS)" is a 30 itemyes/no questionnaire devised specifically to detectdepression in elderly subjects. It has been extensivelyvalidated in hospital samples.12 Evans and Katona6have shown it to have acceptable sensitivity (85%) andspecificity (68%) against diagnosis based on structuredpsychiatric interview (the Geriatric Mental StatusSchedule (GMS)13 in a British primary care sampleaged 65 and over. It also agreed with GMS diagnosis

significantly more often than GPs' own diagnoses(76% versus 65% ). A shorter, 15 item version of theGeriatric Depression Scale (GDS 15) has been devisedby Sheikh and Yesavage.14 The GDS15 has beenrecommended for use within the over-75 health checkby the Royal College of General P ractitio ners."The first aim of the present study was to assess thefeasibility and acceptability of using the GDS 15 todetect depression in elderly GP surgery attenders.Since the GDS15 takes up to 10 minutes to complete, itremains relatively cumbersome to be an -ideal part ofthe routine screening of elderly patients (within a briefconsultation with the GP, or in the context of a briefdomiciliary health screen administered by non-medicalpractice staff). A second aim of the study was to deviseshorter and more acceptable versions of the GDS.

METHODSThe study was performed at the Lower Clapton HealthCentre, a seven partner general practice that serves apopu lation of approximately 10 500 patients. Thepractice has a computerized appointment systemlinked to its age-sex register, enabling easy identifica-tion of all elderly surgery attenders. All patients aged65 or over attending their GPs during weekday morn-ing, afternoon or evening surgeries over an 8 weekperiod between January and March 1993 were ap-proached in the waiting room by a researcher (PD)and asked if they minded participating in a study ex-amining whether a short questionnaire was useful inhelping doctors to know how their patients are feelingin terms of mood.

If there were no objections to participating, PDthen asked the 15 questions of the GDS15. For 10 ofthe questions the answer 'yes' gives a positive score (in-dicating depression); in the remaining five the answer' no ' scores positively. The scores are then summed to

give a total of 0-15, with a score of 5 or moreindicating probable depression.In addition, as a measure of acceptability of thequestionnaire, patients were asked three questions:"did you find the questionnaire difficult?", "did youfind the questionnaire stressful?" and "did you findthe questionnaire acceptable?". These were scored as"not at all", "quite" or "very".Patients who required home visits or who attendedweekend surgeries were excluded from data collectionbut weekday emergency surgery attenders were in-cluded. The weekday emergency clinic runs parallel tothe other clinics and is for patients who need to see adoctor that day and are unable to make an appoint-ment due to all clinic lists being full.Student's /-test and the Wilcoxon test were usedto examine the relationship between individual itemscores and the GDS 15 total and the internal consistencyof the scale examined using Cronbach's alpha. In addi-tion, subjects were divided into cases and non-cases onthe basis of a GDS 15 cut-off of 4/ 5 . The relationsh ipbetween individual item scores and caseness/non-caseness was examined using the chi-square test, as wasthe relationship between caseness and test accepta bility.Subjects' scores on the individual GDS 15 items weresubjected to logistic regression analysis, with GDS15'caseness' as the dependent variable. At each stepin the analysis, the most discriminating item waseliminated and the remainder reanalysed. The pro-cedure was continued until no further items signifi-cantly improved the goodness of fit of the logisticregression model. Ten, four and one item short ver-sions (GDS10, GDS4 and GDS1) were derived; these

are described in the Results below. The internal con-sistencies of the GDS 10 and GDS4 were assessed usingCronbach's alpha, and correlations (Pearson's andKendall's) were calculated between scores on theGDS15 and each of the short GDSs.The sh ort versions of the GDS were then tested in anindependent validation sample. This consisted of thefirst 120 patients from the dataset collected by Evansand Katona6 in whom GMS diagnoses were availableand individual GDS30 (and thus GDS 15) items hadbeen entered. Their demographic characteristics,derived GDS 15 scores and proportion identified asbeing 'case s' of depression on the GMS are summarizedin Table 1. Scores on the GDS10, GDS4 and GDS1were calculated for each subject in the validationsample. Correlations, between-group comparisons andchi-square tests were calculated as for the internalvalidation. In addition, sensitivity (percentage GMScases detected by each short GDS), specificity (percen-tage of short GDS detected cases confirmed by GMS)and negative predictive value (percentage of short GDSnon-cases confirmed as non-cases by GM S) were calcu-lated. All data analysis was carried out usingSPSS/PC+, version 3.1.16


3/7

262 FAMILY PRACTICEAN INTERNATIONAL JOURNALRESULTSDemographic Characteristics of Patients StudiedAs shown in Table 1, the mean age of the subjects was73.5 years. Eighty-six subjects (43%) were aged 75 orover. The female:male ratio was approximately 2:1.GDS Total ScoresThe mean total score for the 15 questions was 3.7(median 3, range 0-15). The distribution of GDS15total scores is shown in Figure 1. Table 1 summarizesthe proportion of patients scoring above the 4/5 cut-off. There were no statistically significant relationshipsbetween G DS15 score and age (caseness rate of 33% inthose aged 65-74 and 35% in those aged 75 and over).Similarly, there was no relationship between theGDS15 score and gender, with 35% of the men and33% of the women scoring in the depressed range.Relationship Between Individual Item Scores andGDS15For each of the GDS 15 questions, there was astatistically significant relationship (P < 0.01) between

TABLE 1 Demographic and psychiatric characteristics

CharacteristicnSex (F/M)Mean age (range)Mean GDS 15 score(range)% GDS 15 cases% GMS cases

Derivationsample198126/7273.5 (65-92)

3.7 (0-15)34N /A

Validationsample12082/3874.1 (65-90)

4.7 (0-13)4838

20 r

15

10

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15GDS15 total score

FIGURE 1 The distribution of CDS15 total scores

a positive score and the GDS 15 total. Similarly signifi-cant associations were found between each item andcaseness/non-caseness. The prop ortion of subjects giv-ing 'depre ssed' responses to each GDS question (in thetotal group and after division into GDS 15 'cases' and'non-cases') is summarized in Table 2. Question 3 (doyou feel that your life is empty?) was found to be thebest predictor of caseness with an 84% level of agree-ment (chi-square = 76.2, P < 0.0001) while question 9(do you prefer to stay at home, rather than going outand doing new things?), had the most modest statisti-cal relationship (chi-square = 8.6, P < 0.01) withcaseness. Cronbach's alpha for the GDS15 was 0.80.Acceptabil i ty of the GDS15Of 198 patients app roach ed, only four refused to par-ticipate, leaving 194 (98%) for subsequent analysis. Atotal of 88% found the questionnaire to be 'very' ac-ceptable, 89% found it to be 'not at all' stressful and88% found it 'not at all' difficult (Table 3). There wereno statistically significant relationships between genderor age and any of the parameters of acceptability. Ascan be seen from Table 3, there was a statisticallysignificant relationship between GDS15 'caseness' andfinding the questionnaire 'difficult' or 'stressful', witha similar trend linking 'caseness' and the experience ofthe GDS 15 as unacceptable.Logistic Regression AnalysisThe results of the pooled within-groups correlationsbetween individual items of the GDS 15 are summarizedin Table 4, showing the order of elimination andresidual correlation coefficients. A 10 item version,GDS10, was derived by totalling the scores of all theitems contributing significantly to the logistic regres-sion model. Similarly, a four item version, GDS4,was derived from the totals for the four bestdiscriminating items. The questions forming GDS10are shown in Table 5, with the GDS4 items in bold.Internal Validation of the Short GDSsThe GDS10 and GDS4 scores both correlated veryhighly with the GDS 15 (r = 0.97 and r = 0.89 respec-tively, both P < 0.0001). Student's /-test on GDS 15scores in those scoring yes versus no on the single mostdiscriminating question (GDS1) gave / = 12.6, P


4/7

DEPRESSION IN ELDERLY PRIMARY CARE ATTENDERS, ITABLE 2 Relationships between individual item scores and GDSI5

263

Item Cases(67)4955707843

Per cent scoring+ veNon-cases(127)

5199

228

Total(194)2031304120

Per centagreement7972847875

Are you basically satisfied with your life?Have you dropped many of your activities and interests?Do you feel that your life is empty?Do you often get bored?Are you in good spirits most of thetime?Are you afraid that something bad is going to happento you?Do you feel happy most of the time?Do you often feel helpless?Do you prefer to stay at home, rather than going out anddoing new things?Do you feel you have more problems with memorythan most?Do you think it is wonderful to be alive now?Do you feel pretty worthless the way you are now?Do you feel full of energy?Do you feel that your situation is hopeless?Do you think that most people are better off thanyou are?

48

30

13 25

454934

38

15

182222

15

3643734249

94

39614

1818511826

73

797768

71

7278657673

TABLE 3 Acceptability of the GDSI5

DifficultNot at allQuiteVeryStressfulNot at allQuiteVeryAcceptableNot at allQuiteVery

GDS cases( )

5296

49126

3955

G D Snon-casesin )

11881

12331

111115

n

170177

172157

420170

TotalPer cent

87.68.83.6

88.77.73.6

2.110.387.6

Independent Sample ValidationThe product-moment correlation between the GDS10and the full GDS30 was 0.91 (P < 0.0001); for the 4item GDS it was 0.76 (P < 0.0001). For the GDS1,comparison between GDS30 scores for subjects scor-ing yes and no gave / = 7.3, P < 0.0001. The cor-responding non-parametric tests gave very similarresults, in all cases also significant at the 0.0001 level.Results for agreement against the GMS interview forthe GDS10 and GDS4 (using both possible cut-offpoints in each case) and for GDS1 are summarized inTable 6. The performance of the GDS15 at cut-offpoints of 4/5 and 5/6 (also derived from GDS30 data)in the validation sample is also shown for comparison.Since the GDS4 had higher sensitivity at the lower(0/1) and higher specificity at the higher (1/2) cut-off,we also examined the discriminating ability of theGDS10 in the 36 subjects with an intermediate score (1)on the GDS4. The GDS10 showed 74% sensitivity and58% specificity in this subsample (chi-square = 4.4;P < 0.05).


5/7

264 FAMILY PRACTICEAN INTERNATIONAL JOURNALTABLE 4 Logistic regression analysis

CDS question Order ofelimination Correlation1. Are you basically satisfied with your life?2 . Have you dropped many of your activities and interests?3 . Do you feel that your life is empty?4 . Do you often get bored?5 . Are you in good spirits most of the time?6. Are you afraid that something bad is going to happen to you?7 . Do you feel happy most of the time?8. Do you often feel helpless?9. Do you prefer to stay at home, rather than going out and doing new things?

1 0 . Do you feel you have more problems with memory than most?1 1 . Do you think it is wonderful to be alive now?1 2 . Do you feel pretty worthless the way you are now?1 3 . Do you feel full of energy?1 4 . Do you feel that your situation is hopeless?1 5 . Do you think that most people are better off than you are?

381N / AN / A4210N / A7N / AN / A659

0.290.250.54NSNS0.280.380.21NS0.23NSNS0.190.220.17

TABLE 5 10-ltem geriatric depression scale (4-item version in bold)Are you basically satisfied with your life?Have you dropped many of your activities and interests?Do you feel that your life is empty?Are you afraid that something bad is going to happen lo you?Do you feel happy most of the time?Do you often feel helpless?Do you feel you have more problems with memory than most?Do you feel full of energy?Do you feel that your situation is hopeless?Do you think that most people are better off than you are?

Yes/NOYES/NoYES/NoYES/NoYes/NOYES/NoYES/NoYes/NOYES/NoYES/No

TABLE 6 Performance of CDS 1 0 , GDS4 an d G D S 1 against CMS

GDSI5 (cut-off 4/5)GDS15 (cut-off 5/6)GDS10 (cut-off 3/4)GDS10 (cut-off 2/3)GDS4 (cut-off 1/2)GDS4 (cut-off 0/1)GDS1

iitivity%)

91788993619359

Specificity

72827763886375

Negativepredictive value

94879394809477

Chi-square

55.155.661.241.244.641.229.3

P

001 screening detection and management of depression in elderly primary care attenders part 1

Documents