Validation of the Bangla version of Beck Depression Inventory‐II

Abstract Background Beck Depression Inventory (BDI‐II) is a widely used valid instrument to assess the severity of depression in clinical and normal settings. To meet the necessity of a standard scale for measuring depression among above 265 million Bangla speaking population around the world, this scale was translated and validated. Methods Two translations of BDI‐II into Bangla were prepared, and then, two back translations were done by medical and language experts in parallel. Thereafter, sentence revision followed by pretest on 20 respondents was done to finalize the Bangla version of BDI‐II (BDI‐II BV). Afterward, a cross‐sectional, comparative, and descriptive study was conducted to validate the scale by purposive sampling technique consisting of 111 persons (both clinical and normal) in three tertiary‐level hospitals in Bangladesh. Everyone was given to fill up BDI‐II BV at first. Then, they were given to fill up BDI‐II BV (n = 49), Bangla version of Depression Anxiety Stress Scales 21‐item (DASS21‐BV, n = 47) and BDI‐II (n = 25) 3–7 days later. The diagnosis of depressive disorder was made according to DSM‐5. Correlation study and factor analysis were completed. Results The mean age was 28.83(±8.70) years. The male–female ratio was 1:0.82. Correlation of scores for BDI‐II BV with the DASS21‐BV depression subscale was .920; BDI‐II BV with BDI‐II was .985 (Cronbach's α .993; t test not significant) and BDI‐II BV applied first and the second time was .960 (Cronbach's α .979; z test not significant). The interitem correlation for all the items was found highly significant (p < .01). Patients having depressive disorder or episodes had significantly higher BDI‐II BV scores than normal (M + SD 30.18 + 10.127 than 8.34 + 5.910; p < .001). Partial confirmatory factor analysis demonstrated two‐factor loading comprising Cognitive and Somatic‐affective symptoms. Conclusions Through the translation and validation process, a validated Bangla version of BDI‐II was produced to measure depression and its severity among the Bengali population.

As there is no crucial change has been made in the diagnostic criteria of Depression & Depressive Disorders in DSM 5, BDI-II can be well fitted with this edition. It is found that the BDI-II has been translated into several languages, including Arabic, Chinese, Dutch, Finnish, French, German, Icelandic, Italian, Japanese, Persian, Spanish, Swedish, Turkish, and Xhosa (Smarr & Keefer, 2011). It has also been translated and validated into the Malay language in Malaysia where there is similarity in religious and cultural ambiance with Bangladesh (Mahmud, Awang, Herman, & Mohamed, 2004). Again, one of our SARK countries Sri Lanka has also validated this scale in Sinhalese (Rodrigo, Kuruppuarachchi & Pathmeswaran, 2015).
In recent years, researchers of Bangladesh and West Bengal of India have given scientific endeavors to translate and validate few internationally accepted scales to measure depression into Bangla, for example, Bangla version of the Depression Anxiety Stress Scales 21-item (DASS21-BV) (Alim et al., 2014), Bangla Montgomery Asberg Depression Rating Scale (MADRSB) (Soron, 2017), and Bengali adaptation of Edinburgh Postnatal Depression Scale (EPDS-B) (Gausia et al., 2007;Maity, Saha, Sanyal, & Biswas, 2015). As DASS21-BV is a multidimensional scale that was validated only among medical students, its use among the patient is uncertain. On the other hand, the clinician-rated Montgomery Asberg Depression Rating Scale (MADRS) was developed in the late 1970s to measure the severity of depression in clinical settings (Montgomery & Asberg, 1979). Because this scale was never updated or modified, it does not target reverse neurovegetative symptoms (Cusin et al., 2010). Moreover, Self-rating scales, such as the BDI, offer some advantages over clinician-rated scales, as they may take less time, do not require trained personnel, and their administration and scoring process appear more standardized (Biggs, Wylie, & Ziegler, 1978).
As written by Soron (2017); researchers also tried to translate BDI-II and validate the Bangla version; but the attempted Bangla version of the BDI lacked sound methodology as the question raised for its permission from the author, copyright and ethical approval related pitfalls. However, no journal article was found regarding the validation of the Bangla version of BDI-II to date.
As there are a few scales for measuring depression among the Bangla speaking population especially in Bangladesh, research regarding this issue would be helpful to reduce physical in addition to mental symptoms and improve quality of life as well as an overall economic burden to patients having depressive disorder or comorbid depression. This study was carried out for adaptation and validation of the Bangla version of BDI-II for measuring the severity of depression among the Bangladeshi population.

| MATERIAL S AND ME THODS
The research was conducted in two phases, translation phase and validation phase. As prior permission from the copyright holder NCS Pearson, INC was obtained after paying licensure fees for translation and validation, no additional approval was taken from the author according to the patent rules. Ethical clearance was obtained from the Ethical Committee of Mental Hospital, Pabna; permission was taken from appropriate authorities of particular hospitals. Copies of permission letter from NCS Pearson, Inc. and ethical clearance were sent to the editor of this journal.

| Translation phase
This segment was carried out from March to October 2015. The translation process was according to guidelines stipulated in the US Census Bureau Guideline (Pan & Puente, 2005) and standards of the American Psychological Association (2014).

| Forward translation and backward translation
Two forward translations into Bangla were done by a psychologist having sound knowledge regarding depression along with its sociocultural variation in Bangladesh and another by a professor of English expert in translation. Then, first translation was given to a psychiatrist and second to another professor of English for backward translations who were blind to the original English version of BDI-II. All persons involved in the translation process were Bangladeshi having sound knowledge in both Bangla and English language.

| Expert review
Sentence revision was done by experts involved in the translation process in a panel discussion after the reconcilement of the forward and backward translations. Then, a multidisciplinary expert committee comprised of psychiatrists, psychologists, a psychosocial worker, language experts, and all the translators reviewed original version and translated materials of BDI-II in meetings. At the end of these processes, a preliminary Bangla version of BDI-II (BDI-II PBV) was produced.

| Pretest
Beck Depression Inventory-II PBV was pretested (tryout) among 20 adult literate individuals in Dhaka (both patients and healthy persons) of either sex, that is, 11 males and 9 females, having different sociodemographic characteristics (eight service holders, three housewives, three students, and six others) to identify any flaws in BDI-II PBV. The respondents were requested in writing whether they could understand the items and comment on each of the items for improvement. Their suggestions were discussed with the experts and necessary modifications were done. At the end of the pretest, a harmonized Bangla version of BDI-II (BDI-II HBV) was obtained.

| Face and content validity
Beck Depression Inventory-II HBV was given to a panel of reviewers consisting of two Professors of Psychiatry, a Professor of English and an Assistant Professor of Clinical Psychology at different universities.
After reviewing BDI-II HBV, the BDI-II Bangla version (BDI-II BV) was finalized.

| Study description
This was a cross-sectional, comparative, and descriptive study con- Hospital. Informed consent was taken from the respondents.

| Participants
The researchers reviewed the literature extensively but no single agreement regarding sample size determination of a validation study was found. According to Gorsuch (1983), the minimum ratios of participants to items should be 5:1 or 10:1 and it has been cited in different psychological research (Soron, 2017). Although the greater sample size could produce better analysis, due to constraints of funding and time, participants to items ratio were decided to keep about 5:1.
A heterogeneous group of samples consisting of outdoor or admitted patients, their caregivers, hospital staffs of different categories was taken from the above-mentioned hospitals to validate the BDI-II BV among both the clinical and nonclinical individuals. The sampling technique was purposive. Bengali speaking and by born Bangladeshi from age 14 to 70 years of both sexes having at least primary level education (fifth-grade) were included. Disoriented, agitated, mute, stuporous, psychotic and having significant neurocognitive impairment were excluded. Although 126 persons were approached, a total of 115 persons gave consent for the study.

Both English and Bangla versions of Beck Depression Inventory-II (BDI-II and BDI-II BV) and Bangla version of Depression Anxiety
Stress Scales 21-item (DASS21-BV) were used in this study (Alim et al., 2014). Diagnosis of depressive disorders and other psychiatric illness were done according to DSM-5 Criteria (American Psychiatric Association, 2013).
The original English version of BDI-II consists of 21 items to measure the severity of depression. Each item is a list of four statements arranged in ascending order of severity about a particular symptom of depression which could be rated from 0 (symptom not present) to 3 (symptom strongly present), with resulting summary scores ranging from 0 to 63. The time reference for the response set has 2 weeks. The severity rating guidelines and cutoff score suggested by the authors for total scores of patients diagnosed with major depression are 0-13 minimal; 14-19 mild; 20-28 moderate; and 29-63 severe (Beck et al., 1996). It can be scored manually or by software. The severity rating

| Procedure
Respondents were divided into three groups purposively. All of them were given BDI-II BV at the start. The first group consisted of 52 persons who were given the Depression subscale of DASS21-BV Other than the second group, all the respondents were offered to fill up BDI-II BV again 3-7 days apart for reliability assessment. The participants were diagnosed by a psychiatrist according to DSM 5 criteria whether they were suffering from depressive disorder or not in the broad sense; depressive episode with other disorders was also assigned.

| Statistical analysis
The scoring of all the questionnaires was done manually.
Pearson's correlation was measured, and Cronbach's alpha was  The disease profile is presented in Figure 1.

| Internal consistency
Cronbach's Alpha Based on Standardized items for BDI-II BV was .926, and intraclass correlation coefficients (ICC) using an absolute agreement definition was .92. The interitem correlation for all the items was highly significant at the .01 level (two-tailed). These discoveries are portrayed in Tables 2 and 4, respectively.

| Convergent validity
Pearson's correlation for BDI-II BV and DASS21-BV depression subscale was .920 (Findings are depicted in Table 2). A comparison of severity shows an almost similar result for BDI-II BV with DASS21-BV Depression subscale. As BDI-II BV does not have an extremely severe option in severity rating, that option of the DASS21-BV Depression subscale is merged with the severe rating. These findings are given in Table 3.

| Criterion (concurrent) validity
As BDI-II was designed to measure the severity of depression, it was expected that the BDI-II BV score would be higher in patients with depressive disorder or episode. Cross tabulation among DSM 5 diagnosis and BDI-II BV severity measures were done assuming minimal Again, the correlation between the BDI-II BV score and the BDI-II score was .985; Cronbach's alpha (based on standardized Items) was .993. Here, t test showed no significant difference in means (Table 2). While comparing the severity, it was found that 100% of the sample with minimal, moderate, and severe severity in BDI-II BV matched with BDI-II, whereas 75.0% matched with mild. A comparison can be seen in Table 3.

| Reliability
Correlation for BDI-II BV applied 1st and 2nd time was .960; Cronbach's alpha was .979. Z test shows no significant difference in means in this group. Pearson's correlation and ICC of all the groups were significant (p < .001). The outcomes can be seen in Table 2. Test-retest of BDI-II BV shows the highest correlation for the item "Loss of interest in sex" and lowest for the item "Past failure." Again, test-retest of BDI-II BV applied at 1st and BDI-II 2nd time demonstrates the highest correlation for the item "Loss of interest in sex" but lowest for the item "Tiredness or fatigue." Reliability measurements for individual items are presented in Table 4. Note: r tot = corrected item-total correlation. Findings are revealed in Tables 5 and 6. were involved in this process. This method was done to ensure the translated version would be grammatically sound and the terms used were correct. At the same time, meanings and contents of original BDI-II were well preserved. Good translations were reflected by the production of two English back translations, which were almost similar to the original English version. Minor discrepancies were resolved by a panel discussion. Great care was taken to ensure that the translation was culturally sensitive.

| D ISCUSS I ON
Internal consistency of this version was very high which produced similar result with original English where internal consistency on student sample was α .93 and among psychiatric sample was α .92 (Beck et al., 1996) also in Spanish and Xhosa version where the coefficients were α .91 (Wiebe & Penley, 2005) and α .93 (Steele & Edwards, 2008).
In the study, correlation and ICC between BDI-II BV with BDI-II and DASS21-BV Depression subscale were also very good in addition to