ORIGINAL ARTICLE
Beck Depression Inventory (BDI):A Reliability and Validity Test in theMalaysian Urological Population
K F Quek, M.Phi1*, W Y Low,PhD*, A H Razack, FRCS**, C S Loh, FRCS (Urol)***, *HealthResearch Development Unit, **Department of Surgery, Faculty of Medicine, University ofMalaya, 50603 Kuala Lumpur, ***Sunway Medical Centre and Gleneagles Intan Medical Centre
Introduction
Benign prostatic hyperplasia (BPH) is a commondisease, which is rarely life-threatening andimpose a psychological impact on an individualsuch as depression' . It is no doubt that lowerurinary tract symptoms (LUTS) can be a hindranceto patients and interfere with their daily activities.The physical symptoms such as the frequency,nocturia, urgency, urge incontinence anddribbling can have a profound psychologicalaffect on the patient's quality of life which cancause tremendous emotional burden, namelyanxiety, stress and a high level of depression,
,2.
There are several measures of depression such as theBeck Depression Inventory (BDI), HamiltonDepression Rating Scale3• All these have been widely
This article was accepted: 11 June 2001
Med J Malaysia Vol 56 No 3 Sept 2001
used in other countries and therefore need to bevalidated for the local population. The BeckDepression Inventory (BDI) has thus become acommonly used instrument in multicentre,international clinical trials to assess psychiatricdisorders. The BDI, developed by Beck & Steer,(986) has been widely used in the studies ofdepreSSion in many countries both in the communitybased studies as well as the clinical studies4•
The study was conducted at the UniversityHospital Kuala Lumpur and was aimed to assessthe reliability and validity of the Beck DepressionInventory (BDI) among urological patients.
Materials and Methods
Study sample
The patients were selected based on the inclusionand exclusion criteria. For patients with LUTS, theinclusion criteria were patients who are stable,
285
ORIGINAL ARTICLE
whereas the exclusion criteria were patients whowere not treated with surgical and medicaltreatment for lower urinary tract symptoms priorto this study. Patients less than 40 years old whowere illiterate were excluded in the study, as werepatients with any chronic and acute diseases. Forthe control group, the inclusion criteria includepatients who were free from all major chronic andacute diseases while the exclusion criteria, thosewith urological problems except renal stone withminimal severity.
Instrument
The BDI is designed to assess the severity ofdepression among the psychiatric patients as wellas possible depression in normal population. Thisinventory measures cognitive, affective, somaticsymptoms, neurovegetative and endogenousaspects of depression4
• The self-reportquestionnaire is rated on a four-point scaleranging from 0 (no symptom) to 3 (severesymptom). The items were chosen to assess onlythe severity of depression, and were not selectedto reflect any particular theory of depression. The21 symptoms and attitudes assessed by the BDIinclude: (1) Mood; (2) Pessimism; (3) Sense offailure; (4) Self-dissatisfaction; (5) Guilt; (6)Punishment; (7) Self-dislike; (8) Self-accusation;(9) Suicidal ideas; (0) Crying; (1) Irritability;(12) Social withdrawal; (3) Indecisiveness; (4)Body image change; (5) Work difficulty; (6)Insomnia; (7) Fatigability; (8) Loss of appetite;(9) Weight loss; (20) Somatic preoccupation; and(21) Loss of libido.
The BDI score was obtained by summing theratings given by the interviewer for each of the 21items. The overall depression scores range from 0to 63 and normally divided into four categories.Scores of 0 to 9 were considered within thenormal range or asymptomatic, scores of 10 to 15indicate mild depression, scores of 16 to 23indicate moderate depression and scores of 24 to63 indicate extremely severe depression. Thequestionnaire is easily administered and takesabout 5 - 10 minutes to complete5•
286
The psychometric properties of the BDI wereassessed in three different samples. Validity andreliability were studied in a group of patients withLUTS (N=108) and control patients without LUTS(N=50) and responsiveness was assessed in a groupof patients admitted for transurethral resection ofthe prostate (TURP)(N=79). Management of casesbased on clinical criteria such as medical history,physical and rectal examinations was entirely doneby a Urologist (AH.R).
Procedure
The study protocol was approved by the EthicsCommittee, University Hospital Kuala Lumpur.Written informed consent was obtained from allparticipants after being explained the nature ofthe study. The patients were then required tocomplete the BD!. All questionnaires were selfadministered as well as assisted guidance ifnecessary by one of the authors (K.F.Q). Allpatients included in the validity study werescheduled for twelve weeks after the firstadministration of the BD!. In the sensitivity tochange study, patients completed thequestionnaires a week prior to surgical treatmentand were retested at three months follow up.
Data Analysis
The Cronbach's alpha coefficient was used toassess the internal consistency of the BDI6
• Testretest reliability was assessed. using the intraclasscorrelation coefficient (ICC), which is derivedfrom the analysis of variance (ANOVA) model.Values of ICC varies from 1 (perfectly reliable) too (totally unreliableY. Responsiveness wasanalyzed by calculating the mean differencebetween BD! before and after TURP and dividingit by the mean standard deviation of the scoresbefore TURP (effect size)8.
Mean differences in BDI scores before and afterTURP was also calculated for each individual itemby means of a paired t test or by dividing it by themean standard deviation of stable patients(Guyatt statistic)9.
Med J Malaysia Vol 56 No 3 Sept 2001
Sensitivity of the BDI was assessed by comparingbetween the means of pre-treatment and posttreatment item scores of patients who haveundergone TURF whereas specificity was assessedby comparing the pre-treatment and post-treatmentitem scores in subjects rated as control.
Results
237 patients consisting of medical, surgical andcontrol group participated in this study. The meanage for the medical group was 63.67 years old(SD=8.57), surgical group was 70.01 years old(SD=8.17) and control group was 50.04 years old(SD=12.29). By ethnicity, the Chinese formed thelargest ethnic groups in all the three groups.Patients with LUTS in the surgical group weresignificantly more depressed as compared tocontrol patients (p<O.OOOl). Of the 237, 108 LUTSpatients has the total BDI mean scores of 8.07(SD=5.58) compared to the 79 patientsundergoing TURP (~=14.43, SD=7.87)(p<0.000l).This showed that prior to treatment, the surgicalgroup was significantly more depressed than themedical and control patients.
Internal consistency for the BDI was high for allthe items of BDI indicating a high level ofhomogeneity among items in the scale. Theinternal consistency showed the resulting valuesof Cronbach's alpha for the scale when individualitem were excluded from the analysis. Test-retestreliability was assessed in 108 patients after 12weeks interval and the total scores of BDI had anICC of 0.85 (p<O.OOl) (Table 1). The total scoresof BDI before (8.07) and after (7.67) showed mostof the patients were not depressed according tothe classification of the severity of BDI.
Responsiveness was assessed in patientsundergoing TURP. Table II showed the pre andpost scores, mean difference, effect size, and theGuyatt statistic for individual items, and for totalscores. Before treatment, patients in the surgicalgroup were depressed and subsequently improvedafter treatment. The mean pre intervention score onthe total BDI (~= 14.43, sd=7.87) was significantlyhigher (p<O.OOOl) than the mean post intervention
Med J Malaysia Vol 56 No 3 Sept 2001
BECK DEPRESSION INVENTORY (BDI)
score (~=10.04, sd=5.4l), glvmg an averageimprovement after TURP on the depression level of4.39. Overall effect size and responsiveness werefound to be high indicating improvement in TURPinduced urinary symptoms in these patients.
In terms of treatment responsiveness, sensitivityand specificity of the instrument was evaluated bycomparing the change between baseline and endpoint scores following treatment. All items of theBDI demonstrated a high degree of sensitivity andspecificity to the effects of treatment (Table III).Significant changes were observed across morethan half of the BDI items in the LUTS group. Thelowest magnitude of change was noted in item21. In contrast, except for item 19 (p<O.OOl), noneof the comparison in the treatment of the controlsubjects approached significance.
The result of the discriminant validity between thesurgical and the control group is shown in TableIV. Significant differences between the twogroups were noted in the total scores of BDI. Thesurgical group tend to be more depressed (meanBDI=14.43) than the control group (meanBDI=7.76) who were asymptomatic.
Discussion
The psychometric properties of the BDI validatedhere and other countries showed that they arevirtually identical with respect to theirmeasurement properties4
• Although test-retestexhibited statistical significant at p<0.05 at someitems of BDI, this is not surprising because thetest-retest was done at 12 weeks interval, thusallowing the urinary symptoms and depression toimprove or worsen. In contrast, the minimalchanges of symptoms and depression wouldoccur if test-retest were done at one week, twoweek or the latest one month after the initialassessment. The reasons for the test-retest to becarried out at twelve weeks interval because mostpatients in the TURP group would then haveachieved the maximum benefit or total symptomimprovement and this would make comparisonsbetween the patients with LUTS and patientsundergoing TURP much easier.
287
I'V
Tabl
eI
0<X
>;>
0<X
>V
alid
ityan
dR
elia
bilit
y:G
)
Mea
nTe
st-R
etes
tSco
re,I
ntra
elas
sC
orre
latio
nC
oeH
icie
ntan
dz ;!:
:In
tern
alCo
nsis
tenc
yfo
rIn
divi
dual
BDIi
tem
s»
Intra
elas
sIn
tern
al95
%C
onfid
ence
Inte
rval
;>0
BDI
Mea
nSD
Mea
nSD
Mea
nSD
-I n
Item
sC
orre
latio
nC
onsi
sten
cy**
Test
Rete
stD
iHer
ence
Low
erU
pper
r-
rn
Coe
ffici
ent*
(ICC)
Scor
eSc
ore
10.
670.
670.
320.
470.
300.
500.
028
0.48
-0.1
20.
062
0.87
0.87
0.32
0.47
0.45
0.85
0.13
0.88
-0.0
40.
303
0.82
0.82
0.14
0.44
0.18
0.47
0.03
70.
36-0
.03
0.11
40.
730.
730.
540.
600.
560.
590.
019
0.55
-0.0
80.
125
0.69
0.69
0.35
0.50
0.34
0.53
·0.0
093
0.50
-0.1
00.
096
0.75
0.75
0.28
0.53
0.21
0.63
0.06
50.
52-0
.16
0.03
70.
660.
660.
270.
440.
210.
410.
056
0.43
-0.1
40.
038
0.73
0.73
0.46
0.62
0.39
0.56
0.07
40.
54-0
.18
0.03
90.
560.
560.
028
0.16
0.01
90.
130.
0093
0.17
-0.0
40.
0210
0.69
0.69
0.34
0.95
0.33
0.95
0.00
930.
92-0
.18
0.17
110.
700.
700.
721.
050.
780.
980.
056
0.97
-0.1
30.
2412
0.68
0.68
0.21
0.41
0.29
0.45
0.07
40.
43-0
.007
0.15
130.
720.
720.
290.
530.
350.
620.
065
0.53
-0.0
40.
1714
0.79
0.79
0.16
0.44
0.18
0.43
0.01
90.
36-0
.05
0.09
150.
840.
840.
470.
630.
560.
630.
093
0.46
0.00
40.
1816
0.77
0.77
0.59
0.71
0.55
0.67
0.04
60.
60-0
.16
0.07
~17
0.84
0.84
0.43
0.60
0.51
0.66
0.07
40.
47-0
.010.
16CD C
L'-
180.
600.
610.
240.
450.
160.
370.
083
0.43
-0.1
7-0
.000
3~
190.
830.
840.
200.
510.
130.
430.
074
0.35
-0.1
4-0
.006
Q 0"
'<20
0.76
0.77
0.53
0.70
0.39
0.61
0.14
***
0.57
-0.2
5-0
.03
V> c' Q:
210.
820.
821.
020.
960.
860.
950.
16**
*0.
74-0
.30
-0.0
2tJ
1To
talsc
ores
0.85
0.85
8.07
5.58
7.67
5.69
0.41
***
4.08
-0.3
7-0
.37
0- z
p<O.
OOI
foraJ
lICCs
I*)
0 wCr
onba
ch's
alph
a:no
teth
atCr
onba
ch's
alph
ava
lue
give
nfor
each
item
repr
esen
tthe
effe
ctof
rem
ovin
gth
atite
mfro
mth
eca
lcula
tion
ofth
eal
pha
valu
eleg
ifC
/)
item
1is
omitt
ed,t
here
sulti
ngva
lue
forth
esc
ale
is0.
67if
item
2is
omitt
ed,
itis
0.87
,an
dso
forth)
(**)
CD ~tt
estf
orpa
ired
com
paris
ons
signi
fican
t1**
*)I'
V 0 0
BECK DEPRESSION INVENTORY (BDI)
Table IIResponsiveness: Mean Scores Before and After TURP, Effect Size and GuyattStatistic
PostTURPBDI PreTURPItems Mean SD Mean SD Mean Differences* SD Effect Size Guyatt statistic1 0.63 0.482 0.77 0.953 0.27 0.554 1.01 0.615 0.68 0.546 0.49 0.577 0.59 0.498 0.67 0.639 0.025 0.1610 0.63 1.1811 0.81 1.0312 0.82 0.6913 0.80 0.6714 0.28 0.4815 1.02 0.7216 1.15 0.7517 0.61 0.5418 0.46 0.5919 0.51 0.7720 0.77 0.7521 1.51 1.01Total scores 14.43 7.87
0.320.570.0760.660.380.230.210.37o
0.390.510.630.670.210.900.770.470.290.230.331.54
10.04
0.470.890.310.570.490.500.410.56o
0.970.840.600.590.410.650.680.570.460.420.550.945.41
0.320.20*0.190.350.300.270.380.300.025*0.24*0.300.19*0.13*0.063*0.13*0.380.14*0.16*0.280.440.038*4.39
0.67 0.671.36 0.210.64 0.350.89 0.570.79 0.560.73 0.470.67 0.780.79 0.480.16 0.161.59 0.201.36 0.290.95 0.280.95 0.190.63 0.130.94 0.180.94 0.560.84 0.260.77 0.270.90 0.360.84 0.591.44 0.0385.90 0.56
0.680.430.430.580.600.510.860.480.160.250.290.460.250.140.210.540.230.360.550.630.040.79
Effect size=Mean difference/SD PreTURPGuyatt statistics=Mean difference/SD of stable LUTS patients (medication group)* t test for paired comparisons not significant
BDI with its ability to discriminate betweenpatients with LUTS and those without showed highlevels of sensitivity and specificity likewise with thediscriminant validity. The reasonably large effectsize obtained indicated a high degree of sensitivityto change, likewise with other studies8
•
This study showed that the LUTS significantlycontributed to the increased of thedepression level especially those with severeLUTS. The improvement of the symptomsfollowing treatment resulted in theimprovement of the depression.
Med J Malaysia Vol 56 No 3 Sept 2001
These findings provide substantial assurance thatscores obtained using the BDI are reliable. Thevalidity and reliability scores were consistent withscores of BDI validated in various countries5•
Conclusion
This study showed that the intraclasscorrelation coefficient for total scores of theitems of the BDI indicates high intraclassreliability and the high Cronbach's alphacoefficient suggest that BDI exhibits goodinternal consistency. BDI seems to be a valid,
289
ORIGINAL ARTICLE
Table IIIBDI Items Characteristics of Patients UndergoinPi TURP
and the Control Group: Sensitivity and Sped icitySensitivity N Mean Changes SEM t statistics p value
Items1 79 0.32 0.075 4.19 0.00012 79 0.20 0.15 1.32 0.193 79 0.19 0.072 2.63 0.014 79 0.35 0.10 3.53 0.0015 79 0.30 0.089 3.42 0.0016 79 0.27 0.082 3.24 0.0027 79 0.38 0.075 5.07 0.00018 79 0.30 0.089 3.42 0.0019 79 0.025 0.018 1.42 0.159
10 79 0.24 0.18 1.35 0.18211 79 0.30 0.15 1.98 0.0512 79 0.19 0.11 1.78 0.07913 79 0.13 0.11 1.18 0.24114 79 0.063 0.071 0.90 0.37315 79 0.13 0.11 1.20 0.23416 79 0.38 0.11 3.60 0.0117 79 0.14 0.095 1.47 0.14618 79 0.16 0.087 1.89 0.06319 79 0.28 0.10 2.74 0.00820 79 0.44 0.095 4.67 0.00121 79 0.038 0.16 0.235 0.815
Total scores 79 4.39 0.66 6.61 0.0001
1 50 0.02 0.073 0.275 0.7852 50 0.12 0.84 1.43 0.1593 50 0 0.081 0 14 50 0.04 0.075 0.531 0.5985 50 0.04 0.049 0.814 0.426 50 0.14 0.099 1.41 0.1647 50 0.04 0.057 0.704 0.4858 50 0.16 0.072 2.22 0.0319 50 0.02 0.02 1 0.322
10 50 0.18 0.12 1.46 0.15111 50 0 0.16 0 112 50 0.10 0.059 1.70 0.09613 50 0.06 0.053 1.14 0.26114 50 0 0.064 0 115 50 0.10 0.082 1.22 0.22916 50 0.04 0.075 0.53 0.59817 50 0 0.076 0 118 50 0.02 0.053 0.375 0.70919 50 0.22 0.082 2.67 0.0120 50 0.06 0.088 0.685 0.49721 50 0.08 0.075 1.07 0.29
Total scores 50 0.68 0.73 0.925 0.36
290 Med JMalaysia Vol 56 No 3 Sept 2001
BECK DEPRESSION INVENTORY (BDI)
Table IVBDI Items Characteristics: Discriminant Validity
Items Pre TURP Patients ControlMean SEM Mean SEM Mean SEM 95% Confidence Interval Pvalue
Difference Lower Higher1 0.63 0.055 0.46 0.082 0.17 0.099 0.022 0.37 0.082 0.77 0.11 0.42 0.099 0.35 0.15 0.064 0.64 0.053 0.27 0.062 0.22 0.077 0.046 0.099 0.15 0.24 0.644 1.01 0.069 0.54 0.087 0.47 0.11 0.25 0.69 0.00015 0.68 0.061 0.30 0.065 0.38 0.093 0.20 0.57 0,00016 0.49 0.065 0.32 0.097 0.17 0.11 0.048 0.39 0.127 0.59 0.056 0.28 0.064 0.32 0.085 0.15 0.48 0.00018 0.67 0.071 0.50 0.096 0.17 0.12 0.062 0.40 0.159 0.025 0.018 0 0 0.025 0.018 0.01 0.061 0.1610 0.63 0.13 0.26 0.11 0.37 0.17 0.037 0.71 0.0511 0.81 0.11 0.70 0.15 0.11 0.19 0.26 0.48 0.5612 0.82 0.078 0.30 0.065 0.52 0.11 0.30 0.74 0.000113 0.80 0.075 0.30 0.66 0.50 0.11 0.28 0.71 0.000114 0.28 0.054 0.10 0.051 0.18 0.074 0.031 0.33 0.0515 1.02 0.08 0.48 0.077 0.54 0.12 0.31 0.78 0.000116 1.15 0.085 0.52 0.087 0.63 0.13 0.38 0.88 0.000117 0.61 0.061 0.48 0.077 0.13 0.098 0.066 0.32 0.1918 0.46 0.067 0.16 0.052 0.30 0.085 0.13 0.46 0.00119 0.51 0.086 0.36 0.086 0.15 0.13 0.11 0.40 0.2620 0.77 0.084 0.62 0.11 0.15 0.14 0.12 0.42 0.2621 1.51 0.11 0.50 0.10 1.01 0.15 0.70 1.31 0.0001Total scores 14.43 0.88 7.76 0.74 6.67 1.15 4.39 8.95 0.0001
useful and a reliable instrument for assessingthe severity of LUTS in the Malaysianurological population.
Med J Malaysia Vol 56 No 3 Sept 2001
Acknowledgements
The authors would like to express their gratitudeto the University Malaya's Research andDevelopment Management Unit for providing theresearch grant for this study.
291
ORIGINAL ARTiClE
1. Quek KF, Low WY, Razack AH, Loh CS.Psychological effects of treatment for lower urinarytract symptoms. British Journal UrologyInternational 2000; 86: 630-33.
2. Lukacs B, Leplege A, McCarthy C, Comet D.Construction and validation of a BPH specifichealth related quality of life scale, for medicaloutcome research studies. In Proceedings of theSecond International Consultation on BenignProstatic Hyperplasia (BPH) (eds ATK. Cockett , S.Khoury, Y. Aso et at.), Paris.: SCI. 1994; 139-43.
3. Hamilton M. A rating scale for depression.Journal of Neurology, Neurosurgery andPsychiatry; 1960; 23: 56-61.
4. Steer RA, Beck AT and Garrison B. Applications ofthe Beck Depression Inventory; in Sartorius N, BanTA (eds): Assessment of Depression. Berlin,Springer. 1986; 121-42.
5. Beck AT and Steer RA. Beck Depression Inventory:Manual. The Psychological Corporation, HarcourtBrace Jovanich Inc., San Diego. 1986.
6. Cronbach L]. Coefficient alpha and the internal.structure of test. Psychometrika 1951;16: 297.
7. Deyo RA, Dichr P, Patrick DL. Reproducibility andresponsiveness of health status measures. ControlClinical Trials 1991; 12 (supp 1): 142S-158S.
8. Cohen]. Statistical Power analysis for theBehavioral Sciences. New York: Academic Press.1977; 75-105.
9. Guyatt G, Walter S, Norman G. Measuringchanges overtime: assessing the usefulness ofevaluative instruments. Journal Chronic Disease1987; 40: 171-78.
292 Med J Malaysia Vol 56 No 3 Sept 2001