Validation of the suicidal ideation scale for Hijra (third gender) people in Bangladesh

Introduction

The term hijra represents a minority group of transgender females in the Indian subcontinent (Jayadeva, 2017). Although the term hijra refers to biological males who have a female gender and gender expression, their identities are created by a variety of variables other than sex and gender, such as religion, culture, and community (Jayadeva, 2017). Hijra people experience various mental health problems worldwide, including suicidal ideation. Evidence suggests that they experience perceived and internalized stigma, isolation, discrimination, and victimization leading them to develop mental health problems such as anxiety, depression, and substance use (Jayadeva, 2017). When compared to their heterosexual counterparts, they have a higher prevalence of mental health difficulties (Meyer, 2003; Russell & Fish, 2016). Various psychiatric disorders such as alcohol and substance use and reduced well-being are prevalent among hijra people (Kalra & Shah, 2013; Lin et al., 2021). Evidence suggests the rates of lifetime mental illness among hijra people was 38% (Sartaj et al., 2021). They experience somatoform disorder, eating disorder (e.g., bulimia nervosa) (Sartaj et al., 2021), and lower life satisfaction (Anderssen et al., 2020). Evidence also showed that hijra people have poor mental health than physiological problems (Bhattacharya & Ghosh, 2020) that include a higher susceptibility of HIV and sexually transmitted infections (STIs) (Jayadeva, 2017; Sha et al., 2021).

Higher rates of suicidal ideation, suicidal attempt, and self-harm behaviors were also found to be prevalent among hijra people (Anderssen et al., 2020; Lin et al., 2021). Estimates showed that suicide rate and suicidal attempts among the transgender community, including hijra people, have been reported to be higher compared to the general population (Cochran & Mays, 2000; Proctor & Groze, 1994; Russell & Joyner, 2001; Thoma et al., 2019). For example, it was estimated that the rate of suicide among transgender individuals in India is about 31%, and 50% of them have attempted suicide at least once before their 20^th birthday (Virupaksha et al., 2016). Similarly, in Canada’s Ontario the prevalence rate for suicide is 35.1% and, of them, 11.2% attempted suicide in the past year (Bauer et al., 2015). Risk factors for this disproportionately high rate include discrimination, rejection by family, friends, and community, bullying, internalized transphobia, gender-based victimization, policy and public, intimate partner violence, and being denied appropriate bathroom, proper treatment at health-care system, and lack of access to housing, unemployment, mental disorders such as depression, physical assault, desire for gender affirming surgery (Narang et al., 2018; Virupaksha et al., 2016; Zwickl et al., 2021). In addition, the persistent experience of minoritization can cause emotional distress and mental health repercussions among gender diverse people. Therefore, deprivation and minority stress have the potential to develop suicidal ideation among gender diverse people.

Transgender people are generally referred to as hijra in Bangladesh and possess a culture that is distinct and easily identifiable due to their unique way of living, such as feminine clothing with masculine characteristics. They usually live and move in groups (Mozumder, 2017). Their sources of income include claiming and collecting money from shops, which is known as cholla in the hijra community, singing and dancing, and claiming money when babies are born (Jebin, 2018; Mozumder, 2017). Some of them are involved in the sex-trade (Dietert & Dentice, 2009). Hijra people are subject to myriad societal issues ranging from changing official documents (e.g., birth certificates and social benefit cards, etc.) (Dietert & Dentice, 2009) to being treated viciously due to their atypical gender identity that does not conform to societal ideas of gender. Despite the Bangladesh governments’ recognition of hijra as ‘third gender’, this plight continues (Jebin, 2018). They are widely vulnerable and continue to be among the poorest of the poor and often are denied basic education, health care, and employment (Khan et al., 2009). They are also prone to ill-health including sexually transmitted diseases and HIV (Jebin, 2018). Parts of their reluctance to see doctors at hospitals were attributed to a feeling of discomfort due to socio-cultural perceptions toward them and little understanding of diverse sexual identities among health service providers (Chan & Khan, 2007; Joseph, 2013). Employment opportunities are restricted due to their lower education and feminine behavioral attributes (Jebin, 2018). Furthermore, hijra people experience disproportionate verbal harassment and there is little opportunity to seek legal support in response to this (Khan et al., 2008). This plight is even worse for older hijra people living alone with poverty and suffering from illness, who are without family members and restricted social connections as social outcasts (Khan et al., 2008).

The number of hijra people in Bangladesh is not accurately determined, with some estimates suggesting being between 4,504 and 8,882 (Mozumder, 2017) while other estimates suggest that more than 10,000 hijras live in Bangladesh (Sifat, 2020). It is of note that hijra people in Bangladesh have been granted a social safety net (e.g., allowance for older hijra people) (Jebin, 2018). While the social benefits and legal recognition seem very progressive, the reality is grim with various unexpected accounts of discrimination based on their sexual orientation and gender identity (Chaney et al., 2020). These experiences may synergistically contribute to greater mental health problems including suicidal ideation. Evidence suggests that sexual minority groups including transgender individuals are more susceptible to stressful life events such as physical, sexual, and psychological abuse, financial and legal difficulties (Mozumder, 2017). But little research exists in Bangladesh which documents their mental health experiences. Limited scholarly work suggests that exclusion from greater society has diminished their self-esteem and sense of social responsibility (Khan et al., 2009). They also experienced physical, verbal, and sexual abuse (Khan et al., 2009). These accounts of abuses and social exclusions can contribute to the development of suicidal ideation. Currently there are no known standardized instruments to measure the suicidal ideation for this sexually and gender diverse community (i.e., hijra community) thus our purpose was to assess the psychometric properties of a widely used suicidal ideation scale (SIS) (Luxton et al., 2011) in the context of Bangladesh.

Methods

Results

Factor analysis

Exploratory factor analysis (EFA)

The structure of the Bangla SIS was evaluated by EFA using principal component analysis (PCA) with direct oblimin rotation. Prior to performing PCA, the suitability of data for factor analysis was assessed. Inspection of inter-item correlation matrix revealed the presence of substantial number (42.22%) of coefficients 0.30 and above. The Kaiser-Myer-Oklin value was 0.863, exceeding the recommended value of 0.6, and Bartlett’s Test of Sphericity reached statistical significance (χ² = 622.607, p < 0.0001), supporting the factorability of the data. PCA revealed two components with eigen value >1 explaining 46.69% of the total variance. An inspection of the scree plot also revealed a clear break after the second component. However, the result of parallel analysis revealed only one component with eigenvalues exceeding the corresponding criterion values for a randomly generated data matrix of the same size (10 variables × 314 respondents). Nonetheless, we went for two factor solutions because parallel analysis is not the only criteria to determine the number of factors (Lim & Jahng, 2019).

The two factors solution of the present study addresses two constructs including suicidal desire (ongoing thoughts or desires) and resolved plans and preparation (intense thoughts, plans, courage, and capability to commit suicide). Table 4 presents the rotated factor loadings.

Table 4. Factor loading using principal component analysis and oblique rotation.

Item No.	Suicidal ideation scale item	Factor 2	Factor 1
		Suicidal desire	Resolved plans/preparation
10	I feel like I am going to take my own life.	0.851	-0.217
7	I just wish my life would end.	0.713	-0.054
6	Life is so hard I feel like giving up.	0.647	0.131
9	No solution can save me from the desire of taking my own life.	0.473	0.206
8	It would be better for everyone involved if I were to die.	0.472	0.250
4	I feel like I will make attempts to kill myself.	0.404	0.270
1	I have been thinking of ways to kill myself.	-0.121	0.828
2	I have told someone I want to kill myself.	0.004	0.679
3	I believe my life will end in suicide.	0.096	0.675
5	I feel life just isn’t worth living.	0.247	0.480

Confirmatory factor analysis (CFA)

For the solution of two factors of SIS, CFA was conducted with maximum likelihood estimation. The following fit indices were used to estimate the appropriateness of the model fit: (1) chi-square to df ratio (χ²/df), wherein a value of no more than 3 indicates a good fit (Carmines, 1981); (2) the comparative fit index (CFI); (3) the normed fit index (NFI); generally, values of the CFI and NFI exceeding 0.90 indicate a good fit (Hau et al., 2004); and (4) the root-mean-square error of approximation (RMSEA), in which the criterion for a good model fit is <0.06 (Marsh et al., 2004). In the present study, the result of the two factor CFA showed a good fit to the data (χ²/df = 52.31/34 = 1.54, CFI = 0.97, NFI = 0.92, RMSEA = 0.041). The first factor explained 36.06% of the variable and the second factor explained 10.64% of the variable. The correlation between the two factors was 0.79. The CFA factor loadings and squared multiple correlations are displayed in Table 5. Overall, the results of the CFAs support the two factors model of the SIS. Figure 1 represents the two-factor structure of the Bangla SIS scale.

Table 5. Confirmatory factor analysis (CFA) factor loadings of Bangla suicidal ideation scale.

Suicidal ideation scale (SIS) item	Factor loading
SIS 1	0.598
SIS 2	0.654
SIS 3	0.531
SIS 4	0.517
SIS 5	0.569
SIS 6	0.656
SIS 7	0.544
SIS 8	0.569
SIS 9	0.527
SIS 10	0.537

Figure 1. The two-factor structure of the Bangla Suicidal Ideation Scale.

Discussion

The present study was conducted amid the absence of an instrument measuring suicidal ideation specific for hijra people in Bangladesh. The cross-sectional study recruited 314 hijra people by means of purposive and snowball sampling. The scale underwent stages of translation, judge evaluation, and pilot testing prior to the field testing. Both exploratory and confirmatory factors analysis were employed. Construct validity was assessed using convergent and divergent methods.

Originally developed by Rudd (1989), the SIS assesses the severity of suicidal ideation among college students. The original scale demonstrated a high internal consistency (Cronbach’s alpha = 0.86) and adequate item-total correlations (r = 0.45 to 0.74). The concurrent validity assessed with the Centre for Epidemiologic Studies - Depression scale was found to be moderate (r = 0.55). Moderate validity was also found with the Beck Hopelessness Scale (r = 0.49) (Rudd, 1989). The scale has been translated into other languages such as Malay (Ibrahim et al., 2021) and Indonesian (Fitriana et al., 2022). The scale has been used to assess suicidal ideation among a variety of target population and settings [see (Bhargav & Swords, 2022; Che Din et al., n.d.; Ibrahim et al., 2019; Luxton et al., 2011; Morris et al., 2021; Nkwuda et al., 2020)]. The scale has not been validated in Bangladesh among gender and sexually diverse people, especially those who identified themselves as hijra.

The Bangla version of the scale underwent a series of statistical analyses to assess psychometric properties.

The scale demonstrated excellent test-retest reliability (r = 0.89) based on the recommended criteria (Cicchetti, 1994). This implies that the scale remained consistent when administered twice with a gap of two weeks. The paired sample t-test showed no significant differences between the two time points which further indicates the stability of the scale.

Construct validity was determined with regards to convergent and divergent validity. The scale demonstrated a weak correlation with PSS and WHO-5 Well-being Index (Akoglu, 2018) for divergent and convergent validity, respectively. The weak correlations may be interpreted as the measure not adequately reflecting the construct under study. However, this is not always the case, as validity coefficients are sensitive to sample size (Nunnally & Bernstein, 1994) where collecting data from a more diverse group of participants can improve the validity of the measure. In addition, the absence of multiple measures may also result in poor validity coefficients. Evidence suggests that using multiple measures to assess a construct increases the validity of a measure (Havron, 2022; Nassaji, 2020) as multiple measures provide different perspectives on the construct under investigation, eventually leading to an increase of validity. The present study employed PSS and WHO-5 Well-being Index as these measures are short and easy to understand and take relatively a short period of time to administer. Considering the transient nature of hijra peoples’ residences and demand for increased income on a daily basis, using short measures was thought to be a feasible idea. Besides, it is often argued that long measures can lead to response fatigue and vice versa (Tourangeau et al., 2000). Therefore, it is generally suggested that while conducting research, especially survey research, it is important to consider the length of the measures to reduce the likelihood of response fatigue (Tourangeau et al., 2000). The present study recognizes these weak validity coefficients as limitations and recommends using multiple measures to improve the validity in future studies.

The SIS is a unidimensional scale and yielded a two-factor structure namely suicidal intention and resolved plan. However, not all validation studies yielded the same factor-structure. For example, the Malay version of the scale demonstrated a single factor structure where all of the items loaded into the single factor (Ibrahim et al., 2021). The present study demonstrated the two-factor structure with item no. 5 (I feel life just isn’t worth living) loaded in the resolved plans/preparations as opposed to the original scale where the item was retained in the factor named suicidal desire. This variation can be attributed to the cross-cultural validation of a study where the central construct (in this case resolved plans for suicide) may be subject to indigenous ideas about suicide, sociocultural aspects, and individual factors. The idea of life not worth living may be different across cultures and gender and sexually diverse groups. Giving up life when it is not perceived worthy may lead to a final resolution implying a resolved plan. The other reasons for this discrepancy may include a relatively small sample size (314 in the present study) as opposed to the original study. Factor structures may be susceptible to the small sample size that can lead to the difficulties in identifying the underlying factors of an instrument (Nunnally & Bernstein, 1994). Others have argued that sample size can be problematic in interpreting and replicating the results of a study (Hair, 2009). Future studies are required to investigate whether the discrepancy is related to the small sample size or the inherent cultural differences in construing resolved plans for suicide.

The same data set was used to carry out CFA as evidence suggests that cross-validation study using different data sets may lead to a lack of correspondence between CFA and EFA (van Prooijen & van der Kloot, 2001). In addition, using the same different data sets can lead to overfitting that occurs when the model fits the data but fails to generalize the new data leading to unreliable factor structures. CFA results demonstrated acceptable fit indices suggesting a good model fit of the two-factor structures of the Bangla SIS scale.

Conclusions

The study is the first of its kind in Bangladesh that aimed to validate a suicidal ideation scale for the hijra population- a group of people that drew relatively little attention, especially for their suicidal ideation. The scale is easy to understand and takes only 5 minutes to administer. The scale can be used to measure the severity of suicidal ideation of hijra population. The scale is likely to produce more work into sexual orientation research. Finally, the scale can be used as an outcome measure in clinical settings.