The pathways between female garment workers’ experience of violence and development of depressive symptoms

Background The prevalence of intimate partner violence (IPV) is high (54%) in Bangladesh. Moreover, female garment workers report higher rates of IPV and are also vulnerable to workplace violence (WPV). Experience of violence puts women at increased risk of developing depressive symptoms, which are related with low self-esteem, lower life satisfaction and lower productivity. To our knowledge, there has been no previous research on depression among female garment workers and its connections to IPV and WPV in Bangladesh. This paper aims to address this gap by studying the relationship of IPV, WPV and depression among female garment workers. Methods The data for this paper comes from a cross-sectional survey of female garment workers (n = 800) conducted as baseline survey of a quasi-experimental study known as HERrespect. Survey data were collected during September-December, 2016 among randomly selected female garment workers from eight garment factories in and around Dhaka city. Structural equation modelling was conducted to explore the relationship among IPV, WPV and depression. Results The findings show high rates of any IPV (69%); WPV (73%, experienced or witnessed) and depressive symptomatology (40%) among female garment workers. The analysis of pathways shows that IPV impacts a woman’s experience of WPV and work related stress leading to the development of depression; while WPV had direct and mediated pathways to depression. Experience of controlling by their husband leads to WPV and increased work related stress, and thus leads to depression. It also reveals that a worker’s ability to mobilize resources in emergency, however, increased self-esteem and reduced work related stress. Conclusion This study shows the pathways through which experience of IPV and WPV lead to development of depressive symptoms among female garment workers. The link between women’s ability to mobilize resources with self-esteem and work related stress indicates the need for socio-economic empowerment of women and may suggest that combined intervention to address IPV and women’s empowerment could be successful in dealing with WPV and mental health.


Methods
The data for this paper comes from a cross-sectional survey of female garment workers (n = 800) conducted as baseline survey of a quasi-experimental study known as HERrespect. Survey data were collected during September-December, 2016 among randomly selected female garment workers from eight garment factories in and around Dhaka city. Structural equation modelling was conducted to explore the relationship among IPV, WPV and depression.

Results
The findings show high rates of any IPV (69%); WPV (73%, experienced or witnessed) and depressive symptomatology (40%) among female garment workers. The analysis of pathways shows that IPV impacts a woman's experience of WPV and work related stress leading to the development of depression; while WPV had direct and mediated pathways to depression. Experience of controlling by their husband leads to WPV and increased work related stress, and thus leads to depression. It also reveals that a worker's ability to mobilize resources in emergency, however, increased self-esteem and reduced work related stress. PLOS

Introduction
Violence against women (VAW) is a well-recognized and serious social, human rights and public health problem [1,2]. It incurs substantial costs to nations [3]. Though there are numerous perpetrators of VAW, intimate partners remain the most common perpetrators [4]. Intimate partner violence (IPV) was ranked 31 st among the risk factors for the loss of global disability-adjusted life years in 2016 [5]. There is considerable regional variation in its prevalence. Globally between 15 and 71% of ever-partnered women experience physical and/or sexual IPV in their lifetime [1]. In South Asia, the estimated prevalence of lifetime physical and/or sexual IPV is 42% [6].
Overall, the level of IPV in Bangladesh is high with 54% of ever-married women reporting lifetime physical and/or sexual IPV and 27% reporting such violence during the last 12 months [7], even though the national legislation regarding VAW in Bangladesh is in place [8,9]. According to available statistics, women who earn an income report more IPV compared to their non-income earning peers, as do poorer women. Research with female garment workers suggests that there may be a higher level of IPV (53% in the past 12 months) compared to the overall income earning female population (33% in the past 12 months) [7,10]. The literature from Bangladesh and elsewhere shows that women's experiences of IPV are associated with a wide range of physical and mental health consequences for women [11,12]. IPV is also known to affect women's work and productivity [13,14].
Previous studies conducted amongst garment workers in Bangladesh indicate that female garment workers also experience high levels of IPV, and workplace violence (WPV). In workplaces workers are subject to verbal, physical and sexual abuse [15][16][17]. Studies suggest that around 60% of female garment workers report verbal or physical abuse [16] in the workplace. Sexual harassment in the workplace is also common [16,17]. There is evidence of adverse consequences of WPV on workers' physical and mental health [18,19] and job performence [20]. Thus, it would be expected that female garment workers experience greater vulnerability to mental health problems being exposed to both IPV and WPV.
Depression is one of the most common adverse mental health outcomes of VAW [12,21]. Women who experience IPV are twice as likely to develop depressive symptoms, than those who don't experience IPV [4,12]. Similarly, WPV has been shown to increase depressive symptoms [22]. Depressive disorders are ranked as the single largest contributor to non-fatal health loss [23]and is a precursor of other mental and chronic health problems [24,25]. Moreover, depression is related to low self-esteem, lower life satisfaction and lower productivity [26][27][28]. Despite the adverse impact of depression, it has not been studied rigorously; the estimated prevalence of depression among adults in Bangladesh is 4. 1% [23], and not a priority area for health care delivery systems as mental health problems are not considered health problems [29].
However, considering the needs of VAW survivors, the Government has implemented the Multi-Sectoral Programme on Violence Against Women under the Ministry of Women and Children Affairs, which is meant to provide medical, legal, psychological support and rehabilitation to the survivors of VAW. Unfortunately mental health services are inadequate [30].
To our knowledge there has been no previous research on depression among female garment workers and its connections to IPV and WPV, in the context of Bangladesh. Based on the literature cited above, we hypothesised that women who experience IPV and WPV are more likely to report depressive symptoms, and further that IPV and WPV are interrelated and their combined effect exacerbates workers' mental ill-health. To investigate this we present an analysis of data collected in research with female garment workers in factories in Bangladesh and use structural equation modelling to understand the pathways through which IPV and WPV impact the development of depressive symptomatology.

Design
This study uses baseline (cross-sectional) data of the HERrespect trial, an evaluation of an intervention aiming to reduce IPV and WPV against female garment workers in Bangladesh. The details of the methodology are described elsewhere [31]. Briefly, the HERrespect trial employs a quasi-experimental design, involving four intervention and four control factories in and around Dhaka city. One out of eight factories was from Export Processing Zones (EPZ). The factories were recruited through buyers in the garment industry for inclusion in the research.
The total sample size was 800 female garment workers (100 from each factory). A worker was eligible to be a study participant if she was currently married and living with her husband, had been working in the current factory for at least 12 months and willing to participate in the study. The final sample was randomly selected from eligible female workers list obtained from the factories. Baseline data were collected between September and December 2016 using faceto-face interviews in private in a location convenient for participants outside the factory, with data captured using Personalized Digital Assistants (PDAs).

Measurement
Outcome. Depression in the past week was measured using Center for Epidemiologic Studies Depression (CES-D) scale. It has been used in many different settings (including Bangladesh) with high reliability and validity [32]. It is a 20-item 4-point ordinal scale. The responses are recorded as: rarely or none of the time (less than 1 day); some or a little of the time (1-2 days); occasionally or a moderate amount of the time (3-4 days); and most or all of the times (5-7 days). A summative score was derived based on the responses. We followed the cut-off point recommended by the scale to define presence of depressive symptomatology, which is 16 [32]. The scale shows high reliability among this population (Cronbach's Alpha = 0. 87). The continuous score was used in the path model as the outcome variable where higher scores represents higher depression.
Covariates. Intimate partner violence (i. e. physical, sexual, emotional, economic violence) was measured using a set of questions based on the WHO violence against women instrument [33]. These instruments are designed to minimize reporting biases that arise from subjective perceptions of abuse by asking only about specific behaviours perpetrated by a male partner (e. g., "Has your husband ever slapped you or threw something at you that could hurt you?). Questions captured behaviours that reflect minor and severe physical violence (e. g., slapped, pushed, hit with a fist, kicked, dragged, choked, burned, threatened with a weapon); sexual violence (e. g., forced sexual intercourse or other sexual acts); emotional violence (e. g., insulted, humiliated); economic violence (e. g., prohibited from earning, thrown out of house) during the last 12 months. We created a dummy variable "any IPV" if a worker had affirmative response to any item of the four forms(physical, sexual, economic and emotional) of IPV in the last 12 months.
A modified version of sexual relationship power scale [34] was used to measure controlling behaviours by the garment worker's husband. Items included: "He (husband) has more to say than you do about important decisions that affect both of you"; "He tells you who you can spend time with" etc. Responses were measured using 4-point Likert scale. A summative score was then used to assess control, with higher scores indicating more controlling behaviours by her husband. Cronbach's Alpha for the scale was 0. 66.
An adapted peer victimisation scale [35] was used to measure workplace violence (WPV). The adaptation was made to measure 'experiencing or witnessing' victimisation as this was deemed to be less sensitive with factory management (who had to approve the survey prior to use) than just actual reported victimisation. The adaptation of questions drew heavily on the findings of formative research on the types of violence, which occur in the workplace [15]. The questions have 4-category scaled response options (never, once, few times, many times) and a typical item was 'How often within the past 4 weeks have you experienced or witnessed a manager call an operator or helper names?" A dummy variable "any WPV" was then generated to report any workplace violence considering the items used. We used the continuous summative score of WPV (where a higher score was indicative of a higher level of WPV) in the path model. Cronbach's Alpha was 0. 83 for the scale. The Rosenberg Self-Esteem Scale (RSE) was used to measure self-esteem [36]. It is a 10-item scale that measures global self-worth by measuring both positive and negative feelings about self. The scale is believed to be uni-dimensional. All items are answered using a 4-point Likert scale format ranging from strongly agree to strongly disagree. A summative score was derived where higher score means higher self-esteem and the continuous score was included in the path model. For the self-esteem scale the Cronbach's Alpha was 0. 76.
Work related stress was measured using three statements related to work and income, specifically, "you are frequently stressed or depressed because of not having enough income"; "you are frequently stressed or depressed because you are not proud of what you do to get money"; and "you are frequently stressed or depressed because you want or have to help your family with money". Responses for all items were measured using a 4-point Likert scale ranging from strongly disagree to strongly agree. A continuous summative score, with higher scores indicating higher stress, was used in the path model.
To measure life satisfaction we used four items from of the Diener [37] scale on life satisfaction. The items are "In most ways your life is close to your ideal"; "The conditions of your life are excellent"; "You are satisfied with your life"; and "So far you have got the important things you wanted in life". A 5-pointLikert scale was offered with options ranging from strongly disagree to strongly agree was used to record respondent's answer. A continuous score was used, where higher scores reflect higher life satisfaction. Cronbach's Alpha for this scale was 0. 88.
Disability, as a measure of worker's health status, was defined by asking questions adapted from the Washington Group Short Set of Questions on Disability [38]. There are six questions related to having difficulties in doing certain activities such as difficulties in hearing, in walking or climbing steps, in washing all over or dressing, in remembering or concentrating etc. Responses ranged from 'cannot do at all' to 'no difficulties'. A summative score was used in the path model, where higher score means less disability.
Ability to mobilize resources was determined by asking a single question "If you had an emergency at home and needed 50,000 Tk, (USD 599) how easy would you say it would be to find the money?". The response options are very difficult, somewhat difficult, fairly easy and very easy.
In addition, workers age, level of education, age at marriage, number of children, NGO membership and income were also included in the analysis.

Statistical analyses
All analyses were performed using STATA version 13. Descriptive analyses were performed to describe the characteristics of the sample, to state the rates of the outcome variable and the covariates. Range and standard deviations were reported where applicable. Bivariate analyses were also performed to describe associations between the main variable (depressive symptomology) and covariates (Chi-square and F-test tests for categorical variables; t-tests for continuous variables). Multivariate logistic regression analysis was performed to determine the factors associated with depression. For all statistical tests the p-value was set at 0. 05.
Structural Equation Modelling (SEM) was conducted to assess the inter-relationship between variables and to explore the path between IPV, WPV and depression. The variables included in the model were based on bivariate analyses, regression analysis, theory and conceptual framework (Fig 1). We fitted a path model using maximum likelihood missing values (mlmv) estimation to model all available data. Insignificant (p value >0. 05) paths from the exogenous variables to the outcome variable and mediation paths were removed from the model using backward elimination. With each path removed we tested the model again. The final model was built based on theory and statistically meaningful modifications. To asses model fit of the observed data, we used the comparative fit index (CFI >0. 95); Tucker-Lewis Index (TLI >0. 95) [39] and root mean square error of approximation (RMSEA �0. 05) [40].

Ethics
The study received approval from Institutional Review Board of icddr,b (PR#16036)and the South African Medical Research Council (SAMRC) ethics committee (PR# EC013-5/2016). The study followed the WHO guidelines on researching VAW [41]. All the interviews were conducted only upon receiving verbal consent from the participants considering low readability and high concerns about confidentiality in Bangladesh. Participants were free to stop the interview at any time and refuse to answer any question that they did not want to. No identifying information was recorded in the questionnaire and kept in a separate file to which only the researchers have access. The research findings will be presented in sufficiently aggregated form to ensure that no participating worker can be identified.

Results
The female garment workers were 27 years old on average, with a mean of five years of education and a monthly average income of BDT 8,505(USD 109). One-third (34%) of workers reported physical IPV, 43% sexual IPV, half emotional IPV, and a third (35%) of them reported economic IPV during the last 12 months. Just under three quarters of the workers (69%) reported any IPV in the past year. Three in every four workers (74%) experienced or witnessed WPV. Around 40% of the workers reported potentially clinically relevant depressive symptomatology ( Table 1).
The mean scores of the continuous variables used in the pathway analysis are presented in Table 1 along with the ranges and standard deviations. The mean score of controlling by husband was 3, self-esteem was 31 and life satisfaction was 14. While, the mean scores of WPV, work related stress and depression were 6, 7 and 15 respectively. Table 2 presents a comparison of background characteristics and life experiences of the women by the presence of depressive symptoms. Women with depressive symptoms were significantly older, and had more children, less ability to mobilize resources in emergency, lower self-esteem, lower life satisfaction, disability and high work stress. They were also more highly controlled by husbands, and a higher proportion of them reported their husband having extra marital sex. A higher proportion of women with depressive symptoms also reported IPV and WPV compared to the women who did not have any symptoms of depression.
The SEM shows the pathways (Fig 2) through which IPV and WPV leads to the development of depressive symptoms among female garment workers. Experience of any IPV did not have any direct effect on depression rather the paths between any IPV and depression were mediated through different life experiences. Life satisfaction (the higher IPV, the lower life satisfaction) mediated a path between IPV and depression (the lower life satisfaction, the higher depression). Self-esteem (the higher IPV, the lower self-esteem) mediated a second path between these variables and another path was mediated by both self-esteem and life satisfaction (The lower life satisfaction and self-esteem, the higher depression). WPV mediated the third path between IPV and depression (the higher IPV, the higher WPV and higher depression) and also by both WPV and work related stress. While the fourth path was mediated by work related stress (the higher IPV, the higher work related stress), which then mediate the path through life satisfaction (the higher work related stress, the lower life satisfaction) and disability (the higher work related stress, the higher disability). Worker's disability also mediated a path between IPV and depression (the higher IPV, the higher disability and the higher depression).
WPV had a direct positive path to depression (whereby the higher WPV, the higher depression). The effect of WPV on depression is also mediated through various paths. The path is mediated by work related stress (the higher work related stress, the higher depression). Work related stress also mediated the paths through life satisfaction (the higher work related stress, the lower life satisfaction) and disability (the higher work related stress, the higher disability). Life satisfaction mediated a path between WPV and depression (the higher WPV, the lower life satisfaction and the higher depression). Another path is mediated by disability of the workers (the higher WPV, the higher disability and the higher depression).
A workers experience of controlling by her husband and her experience of WPV, work related stress and disability lead to depression. While there are additional paths between a worker's self-esteem and work related stress were also highly affected by her ability to mobilize resources in emergency.
The standardized coefficients, standard errors, p-values, 95% confidence intervals, and variance of the disturbance and equation-level goodness of fit statistics, for the path model are presented in Table 4. The model fit was very good as root mean square error of approximation (RMSEA) = 0. 040, comparative fit index (CFI) = 0. 993 and Tucker-Lewis Index (TLI) = 0. 973.

Discussion
Our study extends the existing literature on IPV, WPV and depression by confirming the relationship between these experiences of violence (IPV and WPV) and depression. It also demonstrates that experiences of IPV and WPV increased depressive symptoms. Amongst female garment workers, IPV was common (69%), as was witnessing or experiencing WPV (74%), and clinically relevant depressive symptomology (40%). In the adjusted regression model, experience of IPV and witnessing or experiencing of WPV was associated with increased depressive symptomology. These findings confirm previous research showing that these forms of VAW are independently associated with increased depressive symptomology amongst women [4,11,22]. Moreover, our findings also supports the inverse and direct relationship of work stress, life satisfaction, and self-esteem with depression [26,27,42]. Analysis using SEM showed that the association between IPV and depressive symptomology was mediated by a number of pathways. Thus, we find that experience of IPV impacts a woman's experience of WPV and work related stress leading to the development of depressive symptomatology. These findings reinforce the fact that women's experience of one form of violence increases the likelihood of other forms of abuse. The relationship between IPV and work related stress may be explained by job interference such as absenteeism, lower Experience of violence and development of depressive symptoms productivity and low concentration at work [43] and the fact that IPV affects work performance [11,43]. Experience of IPV compromises a woman's ability to concentrate at work [44] and also leads to loss of working hours [45] increasing work related stress. IPV also leads to depression through reduced self-esteem, life satisfaction and increased disability. These findings support and confirm the existing relationship between IPV and selfesteem [46], life satisfaction [47] and health [12].
WPV had direct and mediated pathways to depression. The direct pathway showed that increased WPV was associated with increased depressive symptoms. In addition, the relationship between WPV and depression was mediated by work related stress whereby WPV increases work related stress increasing depressive symptomatology. This is in line with the literature showing that workers who experience WPV are most likely to suffer from work related stress and are more likely to have physical and mental health problems [48]. Studies show that experience of WPV contributes to low productivity [20] and low job satisfaction [49], which in turn may lead to higher WPV and stress. Our findings extend the existing literature by showing that depression is heavily linked with life satisfaction and health [24,25,27] by showing the path and direction of these relationships.
Female garment workers' experience of control by their husband leads to WPV and increased work related stress. The association between husband's controlling behaviour and work related stress may be explained by a husband's control over the worker's earnings, which may exert extra strain on the worker [15]. All of these eventually contribute to depression. On top of these indirect links, controlling by husband also directly contributes to the development of depressive symptoms.
Ability to mobilize resources is fundamental to well-being [50] and it lowered the effect of work stress [51]. The SEM shows that ability to mobilise resources in an emergency is protective of depression, through mediated pathways showing it reduces work related stress, and in turn depression, and also increases life satisfaction reducing depression. The ability to mobilize resources may suggest that the worker has wider or stronger network/connections. Thus she may be relatively more empowered, capable of decision making and negotiating. All of which may make her less vulnerable to violence and depression.
Self-esteem, life satisfaction, work related stress and disability are the largest contributors to depressive symptoms among female garment workers. All these important predictors of depression were heavily affected by a worker's experience of IPV and WPV.
The analysis showed that addressing both IPV and WPV is important for improving female garment worker's mental health. The link between women's ability to mobilize resources with selfesteem and work related stress indicates the need for socio-economic empowerment of women. All these may suggest that combined intervention to address IPV and women's empowerment could be successful also in dealing with WPV and mental health. It is also important to change attitudes related to gender norms and gender-role beliefs that influence IPV. To achieve the full benefit of women's economic empowerment addressing gender norms is important [52]. Previous literature suggests the relationship between negative gender attitudes and women's experience of violence [53,54]. Thus, it is also important to work with men, to address gender norms as it is linked with controlling behaviour and IPV. Successful workplace based intervention to promote gender equity and address gender based violence [31,55] may be tested and replicated form other places. Mental health services must be an integral part of interventions to address IPV and WPV.

Limitation
This study has some limitations. First, although path models are often stated as providing causal paths, any causality cannot be drawn from this analysis as data are cross-sectional. However, the reference periods for IPV and WPV reporting were longer than that of depression, suggesting a temporal ordering. Another limitation of the study is that prevalence of WPV cannot be estimated as we asked questions regarding experience of WPV and witnessing of such events at workplace in one item. It might be possible that the same event was witnessed and reported by several workers. However, this was the only possible way to gather information on WPV from workers in this setting, given that manager approval was required for the questionnaire. These findings are based on data from 800 female garment workers from eight selected factories and thus are not generalizable broadly in the garment sector or outside it. The women interviewed were from eight factories and whilst randomly selected, and so hopefully representative of the workers in their eligibility band per factory they may not represent all workers in their factories (including those with short periods of work there) nor those from other Dhaka factories.

Conclusion
This study showed the complex relationship and the pathways through which experience of IPV and WPV leads to development of depressive symptoms among female garment workers. Women's experience of IPV impacted on violence and stress at work resulting in development of depressive symptoms. This pathway analysis is expected to guide researchers, practitioners and policy makers to design and implement appropriate comprehensive intervention strategies to address IPV, workplace violence and to improve mental health of the female garment workers. This study clearly shows the stake of the garment industry in preventing violence and thus improving productivity. Addressing violence against women will benefit not only the individuals, families, the garment industry, but also contribute to development of the country as a whole.