A validation study of the Cardiac Depression Scale(CDS) in a UK population

Yvonne Birks1 , Alun Roebuck1 and David R. Thompson2*1University of York, UK2Chinese University of Hong Kong, China

Objectives. This study was designed to validate the Cardiac Depression Scale (CDS)in a UK cardiac population.

Method. A battery of questionnaires (the Medical Outcomes Study Short-Form 36[SF-36] Health Survey, the Beck Depression Inventory [BDI], the Hospital Anxiety andDepression Scale [HADS] and the Cardiac Depression Scale [CDS]) was mailed to 487individuals with coronary heart disease (CHD) recruited from cardiac support groups.The process was repeated on a subsample of 80 participants foursix weeks later forthe purpose of testretest analysis.

Results. The response rate from the rst administration was 81% and from the testretest subsample 54%. Factor analysis revealed a one-factor solutionwith a high internalreliability (Cronbachs a= 0.93) and an acceptable testretest reliability (0.79). Con-current validation against the SF-36, BDI and HADS demonstrated strong correlations.

Conclusions. The CDS is both a reliable and sensitive instrument for measuringdepression in cardiac patients.

A growing body of research attests to high levels of depression being reported after

cardiac events. For example, depressive disorders have been reported in the region of

1319% of patients following myocardial infarction (MI) (Forrester, Lipsey, Teitelbaum,

DePaulo, & Andrzejewski, 1992; Frasure-Smith, Lesperance, & Talajic, 1993; Schlieferet al., 1989), and minor depression is present in a much larger proportion of patients.There is some evidence that depressed MI patients have a higher mortality rate than

their non-depressed counterparts. For example, Frasure-Smith et al. (1993, 1995)

reported a ve-fold increase in risk of mortality at six months after MI and an eight-fold

15

British Journal of Health Psychology (2004), 9, 1524 2004 The British Psychological Society

www.bps.org.uk

* Correspondence should be addressed to Professor David R. Thompson, Nethersole School of Nursing, Chinese University ofHong Kong, Esther Lee Building, Shatin, New Territories, Hong Kong, China (e-mail: davidthompson@cuhk.edu.hk).

increase at 18 months. This translates to approximately 20 000 cardiac patients per

annum in the UK alone (Creed, 1999). However, more recent studies have failed to

conrm the association between depression and increased mortality (Lane, Carroll,

Ring, Beevers, & Lip, 2001; Mayou et al., 2000).

Although much of the work looking at severity and duration of depression is

somewhat complicated by the use of a wide variety of assessment tools, there isgeneral agreement that depression in MI is independent of the severity of the

infarction (Creed, 1999; Frasure-Smith et al., 1993; Ladwig, Roll, Briethardt, Budde,

& Borggrefe, 1994) and is associated with poor health outcomes. Chronic depres-

sion (i.e., depression which was present before the cardiac event) is associated

with more severe depression, more social problems and a reduced chance ofsmoking cessation (Freedland, Carney, Lustman, Rich, & Jaffe, 1992; Lloyd &

Cawley, 1983).

Depression may result from social stress and social isolation, which have also been

linked with an increased risk of MI and subsequent poorer prognosis (Case, Moss, Case,

McDermott, & Eberly, 1992; Jenkinson, Madeley, Mitchell, & Turner, 1993; Ruberman,Weinblatt, Goldberg, & Chaudhary, 1984). In patients with existing depression, low

social support has also been implicated in increased mortality (Lesperance, Frasure-

Smith, & Talajic, 1995). Denollet and Brutsaert (1998) have described a personality

construct, Type D, which comprises depression and social inhibition, and is closely

related to an increased cardiac risk.

Several possible mechanisms for this link have been described and authors refer toincreased sympathetic tone resulting from depression (Cameron, 1996), increased

heart rate and decreased heart rate variability (Carney et al., 1988; Krittayaphonget al., 1997). However, Carney, Freedland, Veith, and Jaffe (1999) have argued that

autonomic tone often fails to normalize in patients treated for depression and that the

principal reason for treating depression should be to improve health-related quality oflife. Mayou et al. (2000) identied that depressed patients are less likely to adopt a

healthier lifestyle, utilize health care resources and have a reduced health-related

quality of life one year after an MI than their non-depressed counterparts. Hence, the

early recognition and treatment of depression is signicant in the treatment of

patients post-MI (Creed, 1999).There is good evidence that detecting and treating depression in MI and other

cardiac patients is worthwhile. Successful treatment is thought to improve health-

related quality of life and possibly reduce mortality. It is, therefore, surprising that little

work has been done to develop a specic tool for identifying and measuring depression

in cardiac patients. While generic measures of depression have the advantage of

allowing comparison of levels between groups of patients with different conditions,if the purpose of a study is to focus on depression in a cardiac population then there is an

argument for the development of a psychometrically sound disease-specic measure.

Such a measure would most usefully be evaluated against a range of concurrently

administered responsive instruments to determine its validity. The Cardiac Depression

Scale (CDS), developed with an Australian population as a disease-specic measure ofdepression for cardiac patients, has been proven to be both reliable and sensitive in this

population (Hare & Davis, 1996). This is particularly important since many of these

patients only have relatively mild depression. While this scale shows promise, it has

been little used outside Australia and thus requires further validation if it is to be widely

adopted. This study reports the results of a validation study of the CDS in a UKpopulation.

16 Yvonne Birks et al.

Method

SampleA total 487 participants were recruited through two cardiac support group networks,

the British Cardiac Patients Association and the British Heart Foundation. Individuals

were invited to contact the investigators if they were interested in participating in the

study. All participants had coronary heart disease (CHD) (i.e., MI, cardiac surgery/

angioplasty, heart failure or angina). Those who had a solely valvular pathology or hadreceived a cardiac transplant were not included in the study.

Instruments

Cardiac Depression Scale (CDS)The CDS is a 26-item, self-administered, disease-specic depression scale for use in

patients with CHD (Hare & Davis, 1996). It was originally developed in Australia on a

sample of 248 CHD patients with various disease pathologies (arrhythmias, MI, cardiac

surgery, heart failure) attending an out-patient cardiology clinic. Evidence for its validity

and reliability are cited as reliability coefcients of 0.9 and correlations with the BeckDepression Inventory of 0.73 and with clinical assessment of 0.67. The scale is reported

to be made up of two dimensions and seven subscales (sleep, anhedonia, uncertainty,

mood, cognition, hopelessness, and inactivity) and is scored using a Likert-scale of 17,

a higher score indicating more severe depression.

Beck Depression Inventory (BDI)The BDI is a 21-item, self-administered, generic measure of depression that deals with

sadness, pessimism, sense of failure, dissatisfaction, guilt, expectation of punishment,

self-dislike, irritability, social withdrawal, indecisiveness, body image, work retardation,insomnia, fatigue, anorexia, weight loss, somatic preoccupation, and loss of libido

(Beck, Ward, Mendelson, Mock, & Erbaugh, 1961). The respondent selects a response of

how he or she has been feeling over the past week and scores this on a scale of 03. It is

both a valid and reliable measure of depression and is signicantly correlated with

clinical diagnosis (Shaver & Brennan, 1991). Although developed for a psychiatric

population, the BDI is used extensively in the USA for assessing depression in CHDpopulations, but is less popular in the UK.

Hospital Anxiety and Depression Scale (HADS)The HADS is a widely used self-report instrument which consists of 14 items (two

subscales of seven items) designed to measure anxiety and depression (Zigmond &

Snaith, 1983). It is quick and easy to administer and as a result has been used extensively

in a variety of clinical situations. It is the most widely used instrument for assessing

depression in CHD populations in the UK (Lewin, Ingleton, Newens, & Thompson,

1998).

Medical Outcomes Study Short Form 36 (SF-36) Health SurveyThe SF-36 is a 36-item, generic, health-related, quality of life instrument (Ware &

Sherbourne, 1992) that measures eight dimensions of health: physical functioning,

social functioning, role limitations due to physical problems, role limitations due to

emotional problems, mental health, energy/vitality, pain, and general health perception.

Validation study of the CDS 17

It is widely used and is a valid and reliable instrument in a CHD population (Failde &

Ramos, 2000).

ProcedureParticipants were mailed the battery of questionnaires (CDS, SF-36, HADS and BDI) andasked to complete and return them in a prepaid envelope to the study centre. The

process was repeated on a subsample of 80 participants foursix weeks later.

AnalysisStatistical analysis was performed using SPSS. The factor structure of the CDS was

investigated using principal axis factoring and, in the case of the two-factor extraction,

an orthogonal rotation. Pearsons correlation coefcient was used to establish con-

current validity and testretest reliability.

ResultsOf the participants, 396 (81%; 318 men, 78 women) returned completed question-

naires. The mean age of the participants was 67 (SD= 9, range= 3790) years. Asubsample of 80 participants was randomly selected for the testretest analysis. Of

these, 43 (54%) returned completed questionnaires. The mean age of these participants

(33 men and 10 women) was 63 (SD= 9, range= 4084) years.

Factor analysisSince two dimensions were described in the original study, the data were scored for

these two dimensions. Internal reliability for the two dimensions was good with

coefcient alphas of .91 and .86. However, the two dimensions were highly correlated

(.649). A scree plot indicating a two- or one-factor extraction is appropriate for the data(see Fig. 1).

A two-factor solution yielded factors that were very similar to the two dimensions

described by Hare and Davis (1996). However, there was a great deal of shared variance

with items loading on both factors in the majority of cases. A one-factor solutionproduced a factor on which all 26 items loaded above .3, accounting for 36% of the

variance.

The highest loading item was `I feel frustrated (.75) and the lowest loading item was

`My memory is as good as it always was (.38). The factor had high internal reliability

(Cronbachs a= .93).

RetestThe retest coefcient was .79 (N= 43), indicating satisfactory reliability over time.

Concurrent validationThe scores on the CDS were correlated with the scores on the SF-36, BDI and HADS to

determine the validity of the scale alongside currently used generic measures of

depression and health-related quality of life. Table 1 presents the intercorrelations

between the CDS and the validation tools.

18 Yvonne Birks et al.

The CDS correlates in the expected directions with all of the validation instruments

used in this study. The strongest correlationwaswith the BDI (.788) and both the anxiety

and depression subscales of the HADS (.751 and .770 respectively). However, signicant

relationships were also found between the CDS and all domains of the SF-36. While many

correlations of small magnitude reached signicance in this study, due to the sample sizethese correlations were mostly of a reasonable magnitude as well as signicance.

Since many patients may not meet the criteria for a major depressive illness but may

still experience similar morbidity, it is important to determine the sensitivity of the CDS

in comparison to the more generic BDI. The distribution of scores found in this sample

is presented below in Figures 2 and 3.

The CDS clearly shows a more normal distribution of scores that would indicategreater sensitivity to both extremes of scoring. Since the BDI scores were clearly skewed

towards the lower scores, an examination of the distribution of CDS scores was made

within these lower scores (i.e. below 11). Figure 4 demonstrates that within this lower

group of BDI scorers there is still useful variance in the CDS. The distribution of the

HADS scores was very similar to that of the BDI.

DiscussionThe results of this study indicate that the CDS will prove to be a useful disease-specic

measure of depression and, by association, of health-related quality of life in CHDpatients. While there may some debate over the factor structure which would be best

settled by using structural equation modelling to determine whether the scale is

unidimensional or not, the one-factor CDS described in this study shows signicant

relationships with the most frequently employed concurrent tools. The initial promise

Validation study of the CDS 19

Figure 1. Scree plots for CDS data.

20 Yvonne Birks et al.

Table

1 .Inter-co

rrelations

betw

eenthemaininstrumen

tsused

inthestud

y

HADS

SF-36

BDI

Anx

Dep

CIH

EVHP

MH

PFg

PRLM

RLP

SFg

CDS

.788

.751

.770

.197

.353

.670

.522

.476

.476

.596

.586

.570

BDI

.669

.751

.103

.264

.537

.530

.394

.395

.552

.462

.517

HADSAnx

.729

.126

.333

.548

.596

.344

.399

.527

.469

.518

HADSDep

.212

.266

.610

.511

.464

.494

.614

.587

.637

SF-C

IH.044

*.283

.096

*.229

.288

.155

**.193

.185

SF-EV

.289

.346

.221

.244

.263

.302

.249

SF-H

P.413

.618

.637

.549

.680

.601

SF-M

H.311

.350

.512

.383

.479

SF-PF

.636

.604

.759

.566

SF-P

.496

.644

.580

SF-RLM

.750

.659

SF-RLP

.640

Allps