561

An Inventoryfor Measuring

Depression

A. T. BECK, M.D.C. H. WARD, M.D.

M. MENDELSON, M.D.J. MOCK, M.D.

AND

J. ERBAUGH, M.D.PHILADELPHIA

The difficulties inherent in obtaining con-sistent and adequate diagnoses for the pur-poses of research and therapy have beenpointed out by a number of authors.Pasamanick12 in a recent article viewedthe low interclinician agreement on diagnosisas an indictment of the present state ofpsychiatry and called for "the developmentof objective, measurable and verifiablecriteria of classification based not on per-sonal or parochial considerations, but onbehavioral and other objectively measurablemanifestations."

Attempts by other investigators to subjectclinical observations and judgments to ob-jective measurement have resulted in awide variety of psychiatric rating scales.4,15These have been well summarized in a re-view article by Lorr11 on "Rating Scalesand Check Lists for the Evaluation ofPsychopathology." In the area of psy-chological testing, a variety of paper-and-pencil tests have been devised for the purposeof measuring specific personality traits; forexample, the Depression-Elation Test, de-vised by Jasper9 in 1930.

This report describes the developmentof an instrument designed to measure thebehavioral manifestations of depression. Inthe planning of the research design of aproject aimed at testing certain psychoanalyt-ic formulations of depression, the necessityfor establishing an appropriate system foridentifying depression was recognized. Be-cause of the reports on the low degree ofinterclinician agreement on diagnosis,13 wecould not depend on the clinical diagnosis,but had to formulate a method of definingdepression that would be reliable and valid.

The available instruments were not con-sidered adequate for our purposes. The Min-nesota Multiphasic Personality Inventory,for example, was not specifically designed

Submitted for publication Nov. 29, 1960.This investigation was supported by Research

Grant M3358 from the National Institute of Men-tal Health, U.S. Public Health Service.

From the Department of Psychiatry, Universityof Pennsylvania School of Medicine and the Phila-delphia General Hospital.

53

Downloaded From: http://archpsyc.jamanetwork.com/ by a Princeton University User on 08/19/2013

562 ARCHIVES OF GENERAL PSYCHIATRY

for the measurement of depression; its scalesare based on the old psychiatric nomenclature; and factor analytic studies revealthat the Depression Scale contains a number of heterogeneous factors only one ofwhich is consistent with the clinical conceptof depression.3 Jasper's Depression-Elationtest 9 was derived from a study of normalcollege students, and his report does notrefer to any studies with a psychiatricpopulation.

MethodA. Construction of the Inventory.The items in

this inventory were primarily clinically derived.In the course of the psychoanalytic psychotherapyof depressed patients, the senior author made systematic observations and records of the characteristic attitudes and symptoms of depressed patients.He selected a group of these attitudes and symptomsthat appeared to be specific for these depressedpatients and which were consistent with the descriptions of depression contained in the psychiatric literature.10 On the basis of this procedure,he constructed an inventory composed of 21 categories of symptoms and attitudes. Each categorydescribes a specific behavioral manifestation ofdepression and consists of a graded series of 4to 5 self-evaluative statements. The statementsare ranked to reflect the range of severity of thesymptom from neutral to maximal severity. Numerical values from 0-3 are assigned each statement to indicate the degree of severity. In manycategories, 2 alternative statements are presentedat a given level and are assigned the same weight ;these equivalent statements are labeled a and b(for example, 2a, 2b) to indicate that they areat the same level. The items were chosen on thebasis of their relationship to the overt behavioralmanifestations of depression and do not reflect anytheory regarding the etiology or the underlyingpsychological processes in depression.

The symptom-attitude categories are as follows :a. Mood k. Irritabilityb. Pessimism 1. Social Withdrawalc. Sense of Failure m. Indecisivenessd. Lack of Satis- n. Body Image

faction o. Work Inhibitione. Guilty Feeling p. Sleep Disturbancef. Sense of Punish- q. Fatigability

ment r. Loss of Appetiteg. Self-Hate s. Weight Lossh. Self Accusations t. Somatic Pre-i. Self Punitive occupation

Wishes u. Loss of Libidoj. Crying Spells

B. Administration of the Inventory.The inventory was administered by a trained interviewer(a clinical psychologist or a sociologist) who readaloud each statement in the category and askedthe patient to select the statement that seemedto fit him the best at the present time. In orderthat the instrument reflect the current status ofthe patient, the items were presented in such away as to elicit the patient's attitude at the timeof the interview. The patient also had a copyof the inventory so that he could read eachstatement to himself as the interviewer read eachstatement aloud. On the basis of the patient'sresponse, the interviewer circled the number adjacent to the appropriate statement.

In addition to administering the Depression Inventory, the interviewer collected relevant background data, administered a short intelligence test,and elicited dreams and other ideational materialrelevant to the psychoanalytic hypotheses beinginvestigated. These additional procedures were alladministered after the Depression Inventory.

C. Description of Patient Population.The patients were drawn from the routine admissions tothe psychiatric outpatient department of a university hospital (Hospital of the University ofPennsylvania) and to the psychiatric outpatientdepartment and psychiatric inpatient service of ametropolitan hospital (Philadelphia General Hospital). The outpatients were seen either on thesame day of their first visit to the outpatientdepartment or a specific appointment was madefor them to come back a few days later for thecomplete work-up. Hospitalized patients were allseen the day following their admission to thehospital, i.e., during their first full day in thehospital. The demographic features of the population are listed in Table 1. It will be notedthat there are 2 patient samples, one the originalgroup (226 patients), the other the replicationgroup (183 patients). The original sample (StudyI) was taken over a 7-month period starting inJune, 1959, the second (Study II) over a 5-monthperiod. The completion of the first study coincided with the introduction of some new projective tests not relevant to this report.

The most salient aspects of this table are thepredominance of white patients over Negro patients, the age concentration between 15 and 44,and the high frequency of patients in the lowersocioeconomic groups (IV and V). The socialposition was derived from Hollingshead's TwoFactor Index of Social Position,7 which uses thefactors of education and occupation in the classlevel determination.

The distribution of diagnoses was similar forStudies I and II. Patients with organic braindamage and mental deficiency were automaticallyexcluded from the study. The proportions amongthe major diagnostic categories were psychotic

54- Vol. 4, June, 1961

Downloaded From: http://archpsyc.jamanetwork.com/ by a Princeton University User on 08/19/2013

DEPTH OF DEPRESSION 563

s

00 t-i2 S 2S^ F5 CO

M H H

+

-i G

disorder 41%, psychoneurotic disorder 43%, personality disorder 16%. The distributions amongthe subgroups were in order of frequency as follows :

Per CentSchizophrenic reaction 28.2Psychoneurotic depressive reaction 25.3Anxiety reaction 15.5Involutional reaction 5.5Psychotic depressive reaction 4.7Personality trait disturbance 4.5Sociopathic personality 4.5Psychophysiological disorder 3.4Manic-depressive, depressed 1.8Personality pattern disturbance 1.8All other diagnoses 4.8

100.0D. External Criterion.The patient was seen

either directly before or after the administrationof the Depression Inventory by an experiencedpsychiatrist who interviewed him and rated himon a 4-point scale for the Depth of Depression.The psychiatrist also rendered a psychiatric diagnosis and filled out a comprehensive form designed for the study. In approximately half thecases, the psychiatrist saw the patient first ; inthe remainder, the Depression Inventory was administered first.

Four experienced psychiatrists participated inthe diagnostic study.* They may be characterizedas a group as follows : approximately 12 yearsexperience in psychiatry, holding responsibleteaching and training positions, certified by theAmerican Board of Psychiatry, interested in research, and analytically oriented.

The psychiatrists had several preliminary meetings during which they reached a consensus regarding the criteria for each of the nosologicalentities and focused special attention on the varioustypes of depression. In every case, the Diagnosticand Statistical Manual of Mental Disorders of theAmerican Psychiatric Association1 was used, butit was found that considerable amplification of thediagnostic descriptions was necessary. After theyhad reached complete agreement on the criteriato be used in making their clinical judgments, thepsychiatrists composed a detailed instruction manual to serve as a guide in their diagnostic evaluations.

The psychiatrists then participated in a seriesof interviews, during which two of them jointlyinterviewed a patient while the other two observed through a one-way screen. This served as

* the initial group of 226 patients, some ofthe diagnostic evaluations were made by a "non-standard diagnostician," that is, a psychiatrist otherthan the 4 regular psychiatrists. In all, 40 patientswere seen by these psychiatrists.

Beck et al. 55

Downloaded From: http://archpsyc.jamanetwork.com/ by a Princeton University User on 08/19/2013

564 ARCHIVES OF GENERAL PSYCHIATRYa practical testing ground for the application ofthe agreed-on instructions and principles and allowed further discussion of interview techniques,the logic of diagnosis, and the pinpointing ofspecific diagnoses.

Since the main focus of the research was to beon depression, the diagnosticians also establishedspecific indices to be used in making a clinicalestimation of the Depth of Depression. Theseindices represented the pooled experience of the4 clinicians and were arrived at independentlyof the Depression Inventory. For each of thespecified signs and symptoms the psychiatristsmade a rating on a 4-point scale of none, mild,moderate, and severe. The purpose of specifyingthese indices was to facilitate uniformity amongthe psychiatrists; however, in making the over-allrating of the Depth of Depression, they made aglobal judgment and were not bound by the ratings in each index.f They also concentrated onthe intensity of depression at the time of theinterview; hence, the past history was not as important as the mental status examination.

The indices of depression which were devisedand used by the psychiatrists were as follows :

I. Apperance II. Thought ContentFacies Reported MoodGait HelplessnessPosture PessimismCrying Feelings of In-Speech adequacy and

Volume InferiorityKey Somatic pre-Speed occupationAmount Conscious guilt

Suicidal contentIII. Vegetative Signs IV. Psychosocial

Sleep PerformanceAppetite IndecisivenessConstipation Loss of drive

Loss of interestFatigability

The diagnosticians also rated the patient on thedegree of agitation and overt anxiety and filledout a check list to indicate the presence of otherspecific psychiatric and psychosomatic symptomsand disturbances in concentration, memory, recall,judgment, and reality testing. He also made a rating of the severity of the present illness on a4-point scale.

To establish the degree of agreement, thepsychiatrists interviewed 100 patients and made independent judgments of the diagnosis and the Depthof Depression. All 4 diagnosticians participatedin the double assessment and were randomly pairedwith one another so that each of the patients wasseen by 2 diagnosticians. The procedure was tohave one psychiatrist interview the patient andthen after a resting period of a few minutes, theother psychiatrist would interview the patient.After the second interview was conplete, theclinicians generally would meet and discuss thecases seen concurrently to ascertain the reasonsfor disagreement (if any).

ResultsA. Reliability of Psychiatrists' Ratings.

The agreement among the psychiatrists regarding the major diagnostic categories ofpsychotic disorder, psychoneurotic disorder,and personality disorder was 73% in the100 cases that were seen by 2 psychiatrists.^This level of agreement, while higher thanthat reported in many investigations, wasconsidered too low for the purposes of ourstudy.

The degree of agreement, however, in therating of "Depth of Depression" was muchhigher. Using the 4-point scale (none, mild,moderate, and severe) to designate the intensity of depression, the diagnosticiansshowed the following degree of agreement:Complete agreement 56%One degree of disparity 41%

Two degrees of disparity 2%Three degrees of disparity 1%This indicates that there was agreement

within one degree on the 4-point scale in97% of the cases.B. Reliability of Depression Inventory.Two methods for evaluating the internal

consistency of the instrument were used.First, the protocols of 200 consecutive caseswere analyzed. The score for each of the21 categories was compared with the totalscore on the Depression Inventory for eachindividual. With the use of the Kruskal-Wallis Non-Parametric Analysis of Vari-

t A number of problems arose in assessing therelative degree of depression of patients with con-trasting clinical pictures. For example, would apatient who is regressed and will not eat be ratedas more depressed than a patient who is not re-gressed but has made a genuine suicidal attempt?Such problems involved complex clinical judgmentsand will be the subject of a later report.

\s=dd\AA detailed description of the reliability studieswill be reported in a separate article.3 The typesof disagreement regarding the nosological cate-gories and the reasons for disagreement are beingsystematically investigated in another study.

56 Vol. 4, June, 1961

Downloaded From: http://archpsyc.jamanetwork.com/ by a Princeton University User on 08/19/2013

DEPTH OF DEPRESSION 565

ance by Ranks,14 it was found that allcategories showed a significant relationshipto the total score for the inventory. Significance was beyond the 0.001 level for allcategories except category S (Weight-losscategory), which was significant at the 0.01level.

The second evaluation of internal consistency was the determination of the split-half reliability. Ninety-seven cases in thefirst sample were selected for this analysis.The Pearson r between the odd and evencategories was computed and yielded a reliability coefficient of 0.86; with a Spearman-Brown correction, this coefficient roseto 0.93.5

Certain traditional methods of assessingthe stability and consistency of inventoriesand questionnaires, such as the test-retestmethod and the inter-rater reliability method,were not appropriate for the appraisal ofthe Depression Inventory for the followingreasons: If the inventory were readmin-istered after a short period of time, thecorrelation between the 2 sets of scores couldbe spuriously inflated because of a memoryfactor. If a long interval was provided,the consistency would be lowered becauseof the fluctuations in the intensity of depression that occur in psychiatric patients.The same factors precluded the successiveadministration of the test by different interviewers.

Two indirect methods of estimating thestability of the instrument were available.The first was a variation of the test-retestmethod. The inventory was administered toa group of 38 patients at two differenttimes. At the time of each administrationof the test, a clinical estimate of the Depth

of Depression was made by one of thepsychiatrists. The interval between the 2tests varied from 2 to 6 weeks. It was foundthat changes in the score on the inventorytended to parallel changes in the clinicalDepth of Depression, thus indicating a consistent relationship of the instrument to thepatient's clinical state. (These findings arediscussed more fully in the section on validation studies.)

An indirect measure of inter-rater reliability was achieved as follows : Each of thescores obtained by each of the 3 interviewerswas plotted against the clinical ratings. Avery high degree of consistency among theinterviewers was observed for the meanscores respectively obtained at each level ofdepression. Curves of the distribution ofthe Depression Inventory scores plottedagainst the Depth of Depression werenotably similar, again indicating a high degree of correspondence among those whoadministered the inventory.

C. Validation of the Depression Inventory.-The means and standard deviationsfor each of the Depth of Depression categories are presented in Table 2. It can beseen from inspection that the differencesamong the means are as expected; that is,with each increment in the magnitude ofdepression, there is a progressively highermean score. The Kruskal-Wallis One-wayAnalysis of Variance by Ranks was usedto evaluate the statistical significance of thesedifferences; for both the original group(Study I) and the replication group(Study II), the /-value of these differences is

566 ARCHIVES OF GENERAL PSYCHIATRY

00 00

n