Literacy and stigma of suicide in Bangladesh: Scales validation and status assessment among university students

Abstract Background Increased suicide literacy and reduced stigma toward suicide facilitate the care‐seeking for suicidal behavior. However, no attempt has been identified to determine these two vital aspects in Bangladesh. Objectives We aimed to validate the literacy of suicide scale and stigma of suicide scale into Bangla along with the determination of the level of suicide literacy and stigma toward suicide. Methods We conducted this study between April and June 2021. Data were collected from medical school and university students by Google form. We used a questionnaire consisting of four segments (i.e., sociodemographic questionnaire, a questionnaire for suicidal behavior, Bangla literacy of suicide scale [LOSS‐B], and Bangla stigma of suicide scale [SOSS‐B]) for data collection. We tested the psychometric properties of the scales in a sample of 529 students and examined factors associated with suicide stigma and literacy. Results The mean age of the students was 22.61 ± 1.68 (range 18–27) years, 274 (51.8%) were males, 476 (89.9%) were graduate students, and 490 (92.6%) were unmarried. The mean score of LOSS was 4.27 ± 1.99 ranging from 0 to 10. Factor analysis revealed acceptable psychometric properties of SOSS‐B. The literacy was significantly higher in females, students of medicine, having a family history of suicidal attempts, and a history of student nonfatal attempts, while stigma was significantly lower among the females and a history of past attempts. Conclusions This study revealed the level of literacy and stigma and culturally tested the psychometric properties of the LOSS‐B and SOSS‐B among university students in Bangladesh.


INTRODUCTION
Suicide is an important cause of mortality across the globe. More than 8,00,000 people are dying every year by suicide, and it is the second leading cause of death among the 15-29 years old age group (World Health Organization [WHO], 2014). Suicide attempts are more than 20 times higher than the suicides that cause significant economic burdens (WHO, 2014). It happens as a result of a complex interaction between several risk factors (WHO, 2014;Zalsman et al., 2016). Studies revealed that 90% of the suicides have at least one psychiatric disorder (Zalsman et al., 2016). Therefore, psychiatric disorders have been identified as an important risk factor for suicide, and adequate psychiatric care is an important suicide prevention strategy (Zalsman et al., 2016). However, several factors such as a higher level of stigma toward suicidal behavior, inadequate suicide and/or mental health literacy, cultural representation, and criminal legal status hinder the care-seeking for suicidal behaviors (Aldalaykeh et al., 2020;Batterham et al., 2013aBatterham et al., , 2013bCalear et al., 2014;United for Global Mental Health, 2021).
The literacy of suicide scale-short form (LOSS-SF) was developed by Calear et al. in 2012, and the stigma of suicide scale-short form (SOSS-SF) was developed by Batterham et al. in 2013a in Australia (Aldalaykeh et al., 2020;Batterham et al. 2013aBatterham et al. , 2013bCalear et al., 2012cited in Batterham et al., 2013b. These instruments were validated and used in several languages such Arabic (Aldalaykeh et al., 2020), Chinese (Han et al., 2017), and Turkish (Oztürk et al., 2017) to assess the suicide literacy and stigma toward suicide. These instruments are short, focused, and easily administrable.
Bangladesh is a densely populated country in South Asia with more than 160 million populations (World Population Review, nd), while Bangla is the sixth most widely used language (Ethnologue, nd). The country lags far behind in formulating a national suicide prevention strategy and prioritizing suicide prevention (Arafat, 2021). Suicide has still been a criminal offense in the legal system that hinders the disclosure of suicidal behavior and help-seeking for that. There is no suicide surveillance system, and quality suicide data are scarce in the country (Arafat, 2019a;United for Global Mental Health, 2021). There are also sporadic and scanty prevention activities and only one psychiatric setup specialized for suicide prevention (Arafat, 2019a). None of the prevention initiatives is available 24/7. The country has 0.13 psychiatrists per 100,000 population, which is much lower to cope with the mental health burden (Mental Health ATLAS, 2017). The case-control psychological autopsy study revealed that 61% of suicides had at least one psychiatric disorder; among them, only 13% were under psychiatric care and diagnosed before their death . There have been strong negative cultural attitudes and stigma toward suicide resulting in the hindrance of suicidal behavior and delayed or lack of help-seeking from the mental health professionals (Arafat, 2019a).
However, there is no available instrument in Bangla to measure literacy and the stigma of suicide in the country. Against this background, we aimed to adapt and validate the literacy of suicide scale and stigma of suicide scale into Bangla. We also aimed to determine the level of sui-cide literacy and stigma toward suicide in Bangladesh. The study results would act as a baseline reference as well as facilitate further studies and policy-making while formulating the national suicide prevention strategy in the country.

Study place and procedure
We conducted this cross-sectional study among the university students of Bangladesh between April and June 2021. We selected three items (Aldalaykeh et al., 2020;Batterham et al., 2013aBatterham et al., , 2013bCalear et al., 2012). The 12 statements have three responding options, namely yes, no, and do not know. The total score is calculated based on the correct answer. Therefore, the total score ranges from 0 to 12. It assesses literacy about suicide in four dimensions: signs and symptoms (three items), nature of suicide (four items), risk factors (three items), and preventive measures (two items) (Calear et al., 2012;Chan et al., 2014).
Due to the response pattern (yes, no, and do not know), internal consistency form of reliability and factor analysis are not justified for LOSS.
Our adapted LOSS-B has true statements in 2, 4, 6, and 8 numbers, and rests are false.

Bangla stigma of suicide scale
It was adapted from the original instrument that was developed by Batterham et al. (2013aBatterham et al. ( , 2013b and consists of 16 items. The scale contains a common descriptor, that is, "people who die by suicide are:" The respondents would mention their amount of agreement with each descriptor by indicating one from the five mentioned options, that is, strongly disagree (1), disagree (2), neutral (3), agree (4), and strongly agree (5). The scale has three subscales: stigmatization (eight items), isolation/depression (four items), and normalization/glorification (four items). The score of each subscale score is determined by calculating the mean of responses to the items within the subscale (Aldalaykeh et al., 2020). The subscales revealed an acceptable internal consistency measured by Cronbach's alpha greater than 0.78 (Batterham et al., 2013a(Batterham et al., , 2013b).

Adaptation of LOSS and SOSS into Bangla
The adaptation of the instruments was performed following standard recommendations forward-backward translation Beaton et al., 2000). One medical graduate who was informed about the study and one university graduate who was disguised about the study were involved in forward translations. The forward translated versions were compared, contrasted, and a single forward translated version was compiled after addressing the discrepancies. Then, the compiled forward version was translated back into English by another medical graduate and another university graduate. All the translators are native speakers of Bangla and fluent in English. As per the recommendation, we created a complied back-translated version following the same methods while compiling the forward translations. Subsequently, all four versions were assessed by the expert committee, which was formed for this study. The expert committee reviewed and suggested the final adaptation of the instruments. With that version, pretesting was done among the 34 persons and adaptation was finalized. During adaptation, several items were modified. We linguistically modified item numbers 2 and 4 and the scoring of item 11 on the LOSS scale. We included, in item 2 ("seeing a psychiatrist or psychologist can help prevent someone from suicide"), mental health professional as defined in the mental health act of Bangladesh instead of a psychiatrist to broaden the services settings as the country has an inadequate number of psychiatrists (Mental Health Act, 2018). We included, in item 4 ("there is a strong relationship between alcoholism and suicide"), substance addiction instead of alcoholism as studies revealed that other forms of substances are major risk factors for suicide . Moreover, as a Muslim majority country, alcoholism is not a major public health problem in the country. We had to change the scoring of item 11 ("men are more likely to commit suicide than women") due to the reverse gender pattern of suicide in Bangladesh (Arafat, 2019b;WHO, 2018).

Data analysis
We collected 529 responses and analyzed them by Statistical Pack- following standard recommendations . The internal consistency form of reliability was measured by Cronbach's alpha coefficient, and a cut-off of ≥0.70 was considered acceptable. Face and content validities were assessed while performing the adaptation of the instruments . We tested the construct validity by exploratory factor analysis. Factor rotation was performed to identify the factors of the construct which is preloaded in the statistical package. Items with loading value <0.5 was considered to drop from the construct as considered in previous studies (Aldalaykeh et al., 2020;Batterham et al., 2013aBatterham et al., , 2013b. Independent t-test was performed to assess the differences between the groups.

Ethical aspects
We conducted the study complying with the declaration of Helsinki (1964). Before starting the study, we took formal permission from the instrument developing authors (Batterham et al., 2013a(Batterham et al., , 2013b  subscale, and the lowest agreement was reported in the glorification subscale (Table 3).
The correlation assessment revealed that stigma and isolation subscales were positively correlated (p = <.001) ( Table 4). The suicide literacy was significantly higher in females, students of medicine, having a family history of suicidal attempts, and a history of student nonfatal attempts, while stigma was also significantly lower among the females and a history of past attempts (Table 5). It is interesting to note a significantly higher score in the glorification subscale among the students with past attempts (Table 5).

Main findings of the study
We aimed to adapt and assess the psychometric properties of LOSS-B and SOSS-B and determine the level of suicide literacy and stigma

TA B L E 5
Association between demography and Bangla literacy of suicide scale (LOSS-B) and Bangla stigma of suicide scale (SOSS-B) score measured by independent t-test  (Batterham et al., 2013a). The Arabic validation was performed among 160 university students (Aldalaykeh et al., 2020). The Turkish validation study was done among 1100 university students (Oztürk et al., 2017). The Chinese validation study was done among 224 university students (Han et al., 2017). Among the students, only 12.1% had a suicidal attempt, 9.6% had a family history of suicide attempts, and 5.6% had a family history of suicide (Table 1). It was reported 11.3% each in the Arabic study (Aldalaykeh et al., 2020), 12.6% (attempt) and 8.5% (family history of suicidal attempt) in Turkey (Oztürk & Akin, 2018), 10.7%
The current study revealed a low literacy in suicide as the mean value of the LOSS-B was 4.27, and only 43.3% of the students scored more than 4 ( Table 2). The students had extremely low knowledge of depression and suicidality. On the other hand, they had good knowledge regarding the role of mental health professionals and suicide prevention. The mean LOSS score and rate of passing the mean was 5.63 and 55%, respectively, in the Arabic study (Aldalaykeh et al., 2020), 5.83 and 53% in the Chinese study (Han et al., 2017), a bit lower in the Turkish study (36.9%) (Oztürkand & Akin, 2018), and higher in Australian community (>60%) (Batterhamet al., 2013a(Batterhamet al., , 2013b. The existing stigma, criminality as legal status, culture, and lack of attention in educating the general population regarding suicide could be the responsible factors for this low level of literacy . This low literacy is supposed to hinder the help-seeking for suicidal behavior in Bangladesh. Universal strategies should be targeted to raise awareness and improve suicide literacy. Additionally, psychoeducation could improve the literacy status (Batterham et al., 2013a(Batterham et al., , 2013b. The current study revealed an acceptable KMO (0.83; p = <.0001) as a value >0.5 has been considered as the criteria .
The internal consistency of the SOSS-B was measured by Cronbach's alpha which was acceptable (>0.70) in the isolation and stigma subscale . It was close to an acceptable value for the glorification subscale (0.68) ( Table 3). A similar picture was revealed in the Arabic study (stigma, 0.81; isolation, 0.71; and glorification, 0.68) which could be attributed by translating words into another language that may not produce the exact meaning (Aldalaykeh et al., 2020). The Australian study (Batterham et al., 2013a), Chinese study (Han et al., 2017), and Turkish study (Oztürk et al., 2017) revealed acceptable values of internal consistencies Due to low factor loading (<0.5), of three items (embarrassment, pathetic, and shallow) of stigma subscale we dropped these three items from the analysis, and the final SOSS-B contains 13 items and three subscales, that is, stigma (five items; item 2, 6, 6, 15, and 16), isolation (four items; item 4, 8, 9, and 10), and glorification (four items; item 1, 3, 11, and 14). The same procedure was followed in the Chinese validation study where four items were dropped due to the poor loading in the stigma subscale and cross-loading with the isolation subscale (Han et al., 2017). All other studies, that is, Australia (Batterham et al., 2013a(Batterham et al., , 2013b, Jordan (Aldalaykeh et al., 2020), China (Han et al., 2017), and Turkey (Oztürk et al., 2017), revealed three subscales. The observation that three items, namely embarrassment, pathetic, and shallow, had low factor loadings may indicate that either these items are not part of the same construct or that they were not clearly understood by respondents. The three factors covered 61.7% (35.3%, 17.3%, and 9.1%) of variance which was 50% (21.69%, 14.35%, and 13.96%) in Arabic validation (Aldalaykeh et al., 2020), about 60% in the primary validation study (28.5%, 18.2%, and 12.7% (Batterham et al., 2013a), and 61% The isolation subscale had the highest approval rate among the three subscales of SOSS-B, followed by the stigma subscale and glorification had the lowest approval (Table 3). Broadly similar results were noted in the studies conducted in China and Turkey (Han et al., 2017;Oztürket al., 2017;Oztürk & Akin, 2018). These results suggest that students may ascribe suicide to isolation or loneliness more than pro-viding stigmatizing or glorifying explanations. The correlation assessment revealed a similar structure to the Arabic validation (Aldalaykeh et al., 2020). The overall reliability statistics and the correlation revealed a similar structure to the Arabic validation. We postulate that this might be explained by the same religion of Jordan and Bangladesh.
Interestingly, the approval rate for all stigmatization items was higher among Bangladeshi students compared to Australian students (Batterham et al., 2013b). In contrast, the approval rates for the glorification items were similar between cultures. These findings, from a sample of university students, suggest that suicide is stigmatized among the student community in Bangladesh and that there may be a role for targeted stigma reduction efforts, similar to recommendations from the depression stigma literature (Griffiths et al., 2008).
Such interventions may focus on reducing stigmatizing attitudes and increase understanding of why suicides occur. Our findings that males had lower suicide literacy while also endorsing higher stigmatizing attitudes to suicide suggest that interventions targeting suicide-related stigma and awareness must focus on this group.
We noticed that the suicide literacy was significantly higher in females, students of medicine, having a family history of suicidal attempts, and a history of student nonfatal attempts, while stigma was also significantly lower among the females and a history of past attempts ( Table 5). The results indicate that destigmatization programs or education programs for suicide prevention might have the greatest impact if they are targeted to males and to people in the wider community without direct experience of suicide. Less stigmatization in females and among the students with psychology degrees was revealed in the primary validation study (Batterham et al., 2013a).
Although the current undergraduates have a negligible focus on suicide, the clinical and academic environment could be attributable to this less stigmatization among the medicine faculty students. It is noted that there was a significantly higher score in the glorification subscale among the students with past attempts (Table 5). Also, there was a similar high score in glorification among the community people of Australia with suicidal ideation; however, no change was identified in the past attempters (Batterham et al., 2013b). Another study from Australia identified that the presence of suicidal ideation was negatively associated with help-seeking behavior (Calear et al., 2014).

Implications of study findings
The major implications of this study are threefold, First, it provides preliminary evidence of the reliability and validity of the LOSS and SOSS scales in a different religious and cultural setting. This demonstrates the applicability of these instruments to diverse populations and contexts. Next, it has identified setting specific knowledge gaps in suicide literacy that may be used to inform suicide awareness programs. Finally, data from the study will assist in developing strategies for dealing with stigmatizing attitudes to commit suicide and enhance suicide-related awareness of society. We suggest that the validity of these scales be examined in the community-related samples with different demographic attributes as suicide-related stigma may vary as a function of age and educational attainment (Griffiths et al., 2008

What is already known
Suicide is a neglected public mental health problem in Bangladesh where no attempt to determine the suicide literacy and stigma toward suicide was documented.

What this study adds
This study validated the two vital instruments LOSS-B and SOSS-B into Bangla those could be utilized in several settings in Bangladesh. It also revealed the level of suicide literacy and stigma among the university students of the country.

Future directions
Appropriate strategies should be designed to improve the literacy of suicide and reduce the stigma in Bangladesh. Future community-based studies assessing the relationship of help-seeking and level of literacy and stigma should be aimed.

Strengths and limitations
This is the first attempt to test the psychometric properties of the LOSS-B and SOSS-B as well as to determine the level of suicide literacy and stigma in Bangladesh. However, the study has several limitations.
First, only internal consistency form of reliability was assessed without determining other forms such as test, retest, and inter-rater. Second, we did not assess the detailed psychometric properties of the LOSS-B as it is an edumetric instead of psychometric tool where there are correct answers rather than opinions/attitudes/behaviors. Third, data were collected from the university students that may restrict the generalization of study results. Fourth, samples were collected conveniently that might be a source of selection and response bias that hinders the generalization of study results.

CONCLUSION
This study culturally adapted the literacy of suicide scales and stigma of suicide scales into Bangla and determined the psychometric properties of the latter among university students in Bangladesh. The SOSS Bangla contains 13 items in three subscales (stigma five, isolation four, and glorification four). The instrument showed acceptable psychometric properties in Bangla. The suicide literacy was significantly higher in females, students of medicine, having a family history of suicidal attempts and a history of student nonfatal attempts while stigma was also significantly lower among the females and a history of past attempts. These instruments could be utilized in further research and academic and clinical settings.