Threats and deprivations as predictors of IPV perpetration: a cross-sectional analysis of a nationally representative sample of Bangladeshi young adults Authors: Azharul Islam, Mahjabeen Haque, Mohammad Salim Chowdhury Published: 2026 ABSTRACT Abstract AbstractPurposeIntimate partner violence (IPV) is a major public health concern in Bangladesh, where approximately two-thirds of women report experiencing partner violence and over half of young men admit to perpetrating it. While adverse childhood experiences (ACEs) are among the strongest predictors of IPV, existing research has relied on cumulative ACE scores, overlooking how distinct dimensions of childhood adversity may differentially predict violence. This study examines the relationship between two dimensions of ACEs, threat (eg, exposure to violence) and deprivation (eg, lack of emotional or cognitive support) and IPV perpetration among Bangladeshi young adults.DesignCross-sectional survey.SettingBangladeshi youths aged 18–35 years across the eight divisions who participated in the national Gender Norms Attitude Study in 2022.ParticipantsOf the 2790 participants, 2016 (mean age 28.27, SD 5.11) were included in this study as they were either married or in a committed relationship.Primary outcome measuresThe primary outcome measures were IPV perpetration, assessed using the Hurt, Insult, Threaten, Scream scale. The primary exposures were two dimensions of childhood adversity (ACE): threat and deprivation, derived from the nine items commonly used in public health research.ResultsOverall, 59.88% of men and 41.53% of women reported having perpetrated IPV. Both threat (adjusted OR (aOR)=2.57; 95% CI 2.03 to 3.23) and deprivation (aOR=1.75; 95% CI 1.39 to 2.20) ACEs significantly associated higher odds of IPV perpetration. Women had lower odds of perpetration than men (aOR=0.41; 95% CI 0.32 to 0.52), as did those with higher socioeconomic status (aOR=0.83), whereas the odds varied by region (eg, highest in Barisal, aOR=4.40).ConclusionsThe findings highlight the importance of adopting a dimensional approach to ACEs and considering regional and socioeconomic factors in IPV prevention efforts. By integrating these findings into public health interventions and policy frameworks, Bangladesh and other low- and middle-income countries can develop more effective, culturally sensitive strategies to reduce IPV and its devastating consequences.Keywords: Adverse events, Family, GENERAL MEDICINE (see Internal Medicine), Cross-Sectional Studies STRENGTHS AND LIMITATIONS OF THIS STUDY Used multistage random sampling across all eight administrative divisions of Bangladesh, ensuring generalisability to young adults (aged 18–35) in low- and middle-income country contexts. Used multistage random sampling across all eight administrative divisions of Bangladesh, ensuring generalisability to young adults (aged 18–35) in low- and middle-income country contexts. Implemented the novel threat-deprivation adversity model instead of cumulative adverse childhood experience (ACE) scores. Implemented the novel threat-deprivation adversity model instead of cumulative adverse childhood experience (ACE) scores. Collected perpetrator data through gender-matched interviewers using validated Bengali translations of instruments (Hurt, Insult, Threaten, Scream scale α=0.71). Collected perpetrator data through gender-matched interviewers using validated Bengali translations of instruments (Hurt, Insult, Threaten, Scream scale α=0.71). Prevents causal inferences about ACE-intimate partner violence pathways; longitudinal data needed to establish temporality. Prevents causal inferences about ACE-intimate partner violence pathways; longitudinal data needed to establish temporality. Risk of recall bias in ACE exposure measurement; no corroborative data (eg, caregiver reports) available. Risk of recall bias in ACE exposure measurement; no corroborative data (eg, caregiver reports) available. Introduction Intimate partner violence (IPV) includes various harmful behaviours, such as physical, emotional and psychological abuse, as well as coercive control within intimate relationships.1 It is a challenging global public health issue due to its severe physical, mental and social consequences.2 In Bangladesh, approximately two-thirds of women report experiences of violence from an intimate partner3 4 and over 50% of young men admit to abusing their married partners.5 Interventions aiming to curb IPV are disproportionately based on evidence derived from a victim-centric perspective. Little is known about the standpoint of IPV perpetrators, particularly among young adults and those from low and middle-income countries (LMICs). Even this limited body of perpetration literature mainly focuses on male perpetrators, creating an incomplete understanding of IPV dynamics.6 Childhood adversities (aka adverse childhood experiences, ACEs), such as physical and emotional abuse and neglect that happened before 18 years of age, are among the strongest predictors of IPV perpetration.1 7 8 The cycle of violence and social learning theories provide a conceptual foundation for understanding how ACEs contribute to IPV perpetration. The cycle of violence hypothesis posits that individuals exposed to violence during childhood, either as victims or witnesses, are more likely to replicate violent behaviours in their adult relationships.9 Similarly, social learning theory suggests that children learn aggressive behaviours by observing violence in their familial environment.10 The 2007 Bangladesh Demographic Health Survey supports these theories. Using the survey data, one study revealed that men who witnessed their fathers being violent towards their mothers were over three times more likely to report perpetrating any form of physical or sexual IPV compared with those who did not witness such violence.11 This finding persisted even after controlling for various sociodemographic factors. Rahman et al,12 also using the same data, found a significant association between women’s reports of experiencing IPV and their husbands’ childhood experiences of witnessing father-to-mother violence. Much of the research on childhood adversity, however, has relied on a cumulative approach, treating various ACEs as a single construct without distinguishing their types and severity.7 Recently, scholars have advocated for a dimensional approach that categorises ACEs into two dimensions: threat and deprivation.13,15 This approach highlights the distinct pathways through which threat and deprivation impact psychological and behavioural outcomes, leading to IPV perpetration. Threat ACEs, such as physical or emotional abuse, are associated with heightened emotional reactivity and impaired emotion regulation due to the chronic activation of stress-response systems in childhood. This heightened arousal makes individuals more likely to respond to interpersonal conflicts with aggression or violence in adulthood. Studies also suggest that early exposure to threats is associated with changes in neural circuits responsible for emotional learning, resulting in maladaptive responses to perceived threats in relationships.16 17 In contrast, deprivation ACEs, characterised by neglect or the absence of emotional and cognitive support, disrupt the development of cognitive functioning and social skills. This disruption can lead to difficulty forming healthy relationships and resolving conflicts non-violently, contributing to IPV perpetration.18 By distinguishing between these two types of adversity, the dimensional approach provides deeper insights into how specific childhood experiences shape distinct pathways to violence, thereby guiding targeted interventions. The dimensional approach further helps elucidate gender differences in IPV perpetration. Threat experiences often lead to externalised aggression in men, influenced by societal norms that emphasise dominance, while women internalise stress, resulting in relational conflicts or self-directed harm.19 Conversely, deprivation impacts emotional regulation and attachment, with men exhibiting avoidant tendencies and IPV behaviours aimed at control, while women display anxious attachment and manipulative relational patterns.20 Neurological differences, such as heightened amygdala activity in men and hypothalamic–pituitary–adrenal axis dysregulation in women, further mediate these outcomes.16 Additionally, IPV expression aligns with gendered norms, as men are more likely to use physical violence and women psychological aggression.21 These findings emphasise the need for examining gender-stratified associations between childhood adversities and IPV perpetration to guide prevention and intervention strategies.22 Research on IPV perpetration, both globally and in Bangladesh, exhibits a bias towards male perpetrators.56 23,26 First, this oversight is particularly significant in patriarchal societies like Bangladesh, where rigid gender norms, in addition to the reasons mentioned above, may shape IPV differently for men and women. Second, most evidence on IPV perpetration comes from high-income countries, with limited evidence from LMICs, where socio-cultural contexts differ significantly.27 Third, despite high rates of victimisation in Bangladesh, there is a dearth of IPV perpetration research.28,35 We found only one study, although with men, that systematically examined childhood adversities linked with IPV perpetration, in which about 50% of the participants (n=570, 18–34 years) reported having committed physical IPV, with 64% having experienced violence in childhood.5 Fourth, this and other Bangladeshi studies on IPV rely on small, regionally focused samples, limiting their generalisability.23 36 37 For example, the United Nations multicountry study on IPV and various outcome measures includes a sample from only two of the eight divisions of Bangladesh (eg, Dhaka, Chattogram).36 Finally, none of the above studies considered a dimensional approach to ACEs, leaving a significant aspect unexplored. Objectives Addressing the above gaps, we aimed to investigate the association between distinct types of ACEs, specifically threat and deprivation, and IPV perpetration among young adults in Bangladesh. We recruited a nationally representative sample comprising men and women from all divisions of Bangladesh. We considered socioeconomic status (SES), educational attainment, religiosity and gender-norm attitudes as potential covariates for the association of interest. Economic stressors can exacerbate relationship conflicts, leading to violence as a maladaptive coping mechanism.27 Educational attainment has been linked to lower risks of IPV, particularly among women.8 Education equips individuals with better conflict resolution skills, increased autonomy and greater access to resources, which may buffer against IPV perpetration. Gender-norm attitudes were included, as the lone Bangladeshi study noted that the impact of childhood adversity may be reduced if the community holds gender-equal attitudes.5 Methods Sample and sampling Data for this study were collected as part of a large, national cross-sectional study of gender-norm attitudes among Bangladeshi youth aged 18–35.38 The age range aligns with the Bangladesh National Youth Policy (2017), which defines youth as individuals aged 18–35 years.39 A total of 2790 young adults from all divisions were recruited following a multistage random sampling method. Of this sample, 2016 comprising 800 men (M=29.63 years, SD=4.13 years) and 1216 women (M=27.37 years, SD=5.48 years) were married or in a committed relationship; hence, they were chosen for the present study. Multistage random sampling involves selecting areas, households and individuals within them. The administrative divisions in Bangladesh were treated as individual strata, and districts were selected by systematic random sampling. Two to five districts were chosen as the primary sampling units from each division. Two upazilas were then selected from each district as the secondary sampling units. A smaller subdivision was randomly selected as the tertiary sampling unit from each upazila. The tertiary sampling unit was divided into 3–4 starting points. The first house was selected using the dating method. The succeeding houses were selected using the right-hand method with a fixed gap (five fixed gaps in urban areas and two in rural areas). Eligible respondents were identified within selected households and listed in descending order by age, and one respondent from each household was randomly chosen using a random number generation technique. Non-response participants were managed through systematic selection attempts and follow-up calls. Procedure The explanatory statement explicitly stated that participation in the study was voluntary and that participants could withdraw at any time. Written consent was obtained from each participant. To avoid biased responses or nonresponses to gender-sensitive questions, we matched participants’ genders with those of the interviewers. Trained research assistants (RAs) administered the questionnaire orally, one-on-one, in a private room at participants’ homes or local community centres. No third parties were present, and participants were assured their responses would remain confidential. The survey took approximately 20–30 min. At the end of each interview, participants received contact details for local counselling centres and the national IPV helpline. Interviewers were trained to pause interviews if distress was observed and provide psychological first aid. The research team members are national-level trainers in psychological first aid, who provided training to RAs prior to data collection. Data were collected from 1 June 2022 to 29 January 2023. The explanatory statement and consent form are provided as online supplemental file 1 in Bangla, with its English translation. Measures Threat and deprivation Threat and deprivation were derived from the nine items frequently used in public health research to assess the ACE.40 Participants were asked if they had experienced those events before the age of 18 years. The response options were in binary format (yes/no). Following previous studies,13,15 we divided ACE items into threat and deprivation. The threat dimension consisted of five ACE items (ie, witnessed domestic violence, was a victim/witnessed neighbourhood violence, lived with someone who was mentally ill or suicidal, lived with someone with an alcohol or drug problem and experienced discrimination or unfair treatment due to their race or ethnicity). The remaining four ACE items (ie, experiencing the divorce/separation of a parent, parent’s incarceration, death of a parent and experiencing socioeconomic hardship) were used to create the deprivation dimension. We dichotomised the responses (none vs any). IPV perpetration The four-item Hurt, Insult, Threaten, Scream (HITS) scale was used to measure IPV perpetration in romantic relationships.41 42 We asked the participants if they had HITS at their partners. The items were presented in Bangla, translated by bilingual experts, and vetted by the research team (Cronbach’s α=0.71). Participants responded to each item with a 5-point frequency format: never, rarely, sometimes, fairly often and frequently. We dichotomised the responses (never vs any). Attitudes towards gender norms The Multidomain Gender Norm Attitude Scale (M-GNAS) was used to assess attitudes towards gender norms.43 The M-GNAS comprised 13 items organised into four domains. The domains were gender-appropriate behaviour (example item: ‘Girls should not speak loudly’), family financial decisions (eg, ‘A man should have the final words for large family expenses’), family responsibility (eg, ‘If the husband earns sufficiently, the wife needs not to earn’), and career choice (eg, ‘A girl’s higher education can be considered only if she is meritorious’). A five-point Likert scale was used to record the responses. Aligning with gender egalitarianism, the highest value was assigned when a person disagreed with a gender-inequitable item (ie, totally disagree=5; totally agree=1). The total score is the sum of all the item scores. A higher score indicates a more equitable attitude towards gender norms. The reliability scores of the four domains were 0.72, 0.76, 0.71 and 0.68, respectively. The measurement invariance of the M-GNAS across genders was established in the original scale development study,43 using a multigroup confirmatory factor analysis with progressively restrictive models (configural, metric and scalar invariance). All models demonstrated acceptable fit (comparative fit index (CFI) and Tucker–Lewis index>0.95, root mean square error of approximation≤0.06, standardised root mean square residual<0.06), with changes in CFI (ΔCFI) not exceeding 0.01 between successive models, confirming that the scale functions equivalently across genders. Sociodemographic covariates We recorded participants’ ages, genders and geographic locations. Subjective SES was assessed using MacArthur’s SES ladder,44 ranging from 10 (highest) to 1 (lowest). Participants reported their religion and adherence to religious practices (eg, religiosity). Regarding adherence to religious practices, participants reported one of three categories: very little, moderate (weekly for one or more days) or regular (those who perform religious rituals every day and adhere to religious values). We collected educational qualifications for both participants and their parents, ranging from no schooling, primary (1–5 years of schooling), secondary (10 years of schooling), higher secondary (12 years of schooling), to bachelor’s or above (more than 16 years of schooling). The full survey instrument is available in online supplemental file 2). Patient and public involvement Patients/the public were not involved due to the sensitive nature of IPV perpetration research in a high-stigma setting. Prioritising participant safety and confidentiality, we excluded involvement in the design, conduct or outcome measures (HITS/ACE tools), or in recruitment, to prevent re-traumatisation. Findings will be disseminated to policymakers post-publication. Analyses We presented descriptive statistics separately for men and women, along with the distribution of threat and deprivation ACEs across sample characteristics. Next, we examined whether the association between ACE exposures varied by gender. Preliminary analysis indicated that gender did not moderate the relationship (data not shown); therefore, we combined the data for men and women to produce multivariable binary logistic models. The first model was unadjusted, while the second was adjusted for age, gender, division, place of upbringing, religiosity, participants’ and their parents’ education levels and gender norm domains (table 1). We conducted gender-stratified logistic regressions to explore gender-specific associations (see online supplemental file 3). Table 1Adjusted associations of ACEs with IPV perpetration for the combined sample (N=2016)PredictorModel 1Model 2cOR95% CIaOR95% CILowerUpperLowerUpperPrimary exposuresThreatNo (ref) Yes 3.38 2.814.06 2.57 2.033.23DeprivationNo (ref) Yes 2.12 1.782.54 1.75 1.392.20CovariatesGenderMen (ref) Women 0.48 0.400.57 0.41 0.320.52 Age 1.04 1.021.061.021.001.04DivisionDhaka (ref) Chattogram 0.26 0.180.36 0.26 0.180.38 Barisal 4.04 2.646.18 4.40 2.747.06 Khulna 1.36 1.001.830.880.601.27 Rajshahi 2.12 1.562.861.430.992.07 Sylhet 1.62 1.112.36 1.91 1.272.89 Rangpur 1.43 1.041.981.470.982.21 Mymensingh0.780.521.17 0.64 0.410.99Raised before 16 yearsUrban areas (ref) Rural areas 1.22 1.011.491.120.881.42ReligiosityVery little (ref) Moderate1.230.951.591.290.951.76 Regular0.820.631.061.080.791.48 SES0.860.820.90 0.83 0.770.88Participant’s educationNo schooling (ref) Primary or equivalent (≥5 years)0.680.451.02 0.62 0.391.00 SSC or equivalent (≥10 years)0.660.450.97 0.59 0.370.95 HSC or equivalent (≥12 years)0.620.410.94 0.53 0.310.90 Bachelor or equivalent and above (≥14 years)0.650.411.020.550.301.01Father’s educationNo schooling (ref) Primary or equivalent (≥5 years)1.010.821.240.960.691.32 SSC or equivalent (≥10 years)0.870.681.120.890.591.35 HSC or equivalent (≥12 years) and above0.570.410.780.670.371.21Mother’s educationNo schooling (ref) Primary or equivalent (≥5 years)0.880.721.070.770.561.06 SSC or equivalent (≥10 years)0.710.540.930.690.451.07 HSC or equivalent (≥12 years) and above0.650.411.04 0.48 0.231.00M-GNAS factors Gender appropriate behaviour0.970.950.991.030.991.08 Family financial decision0.950.930.980.970.931.02 Family responsibility0.970.941.00 1.05 1.001.10 Career choice0.930.900.96 0.93 0.890.97Note: ACEs (threat and deprivation) are the primary predictors of interest. Other variables serve as control covariates to adjust for demographic and psychosocial influences. The ORs in bold indicate statistical significance.Model 1 is unadjusted, and model 2 is adjusted for age, gender, division, raising place, religiosity, participants’ and their parents’ education and gender norm domains.ACEs, adverse childhood experiences; aOR, adjusted OR; cOR, crude OR; HSC, higher school certificate; IPV, intimate partner violence; M-GNAS, Multidomain Gender Norm Attitude Scale; SES, socioeconomic status; SSC, secondary school certificate. Adjusted associations of ACEs with IPV perpetration for the combined sample (N=2016) Note: ACEs (threat and deprivation) are the primary predictors of interest. Other variables serve as control covariates to adjust for demographic and psychosocial influences. The ORs in bold indicate statistical significance. Model 1 is unadjusted, and model 2 is adjusted for age, gender, division, raising place, religiosity, participants’ and their parents’ education and gender norm domains. ACEs, adverse childhood experiences; aOR, adjusted OR; cOR, crude OR; HSC, higher school certificate; IPV, intimate partner violence; M-GNAS, Multidomain Gender Norm Attitude Scale; SES, socioeconomic status; SSC, secondary school certificate. Results Over 50% (59.88%) of men and 41.53% of women reported IPV perpetration. Online supplemental file 3 presents the distributions of social and psychological characteristics, as well as the prevalence of threat-related and deprivation-related ACEs, separately for men and women. Distribution of threat and deprivation ACEs across sample characteristics Among men, 59.92% of IPV perpetrators experienced threat ACEs compared with 21.50% of non-perpetrators (p<0.001), and 54.70% experienced deprivation ACEs in contrast to 33.96% of non-perpetrators (p<0.001). Similarly, among women, 56.04% of IPV perpetrators reported threat ACEs as opposed to 32.35% of non-perpetrators (p<0.001), and 60.99% experienced deprivation ACEs compared with 41.77% of non-perpetrators (p<0.001). There were no significant differences in mean age between participants with and without threat or deprivation ACEs for either gender. The prevalence of both threat and deprivation ACEs varied significantly across different divisions for both genders. In men, Rajshahi exhibited the highest prevalence of threat ACEs (80.31%), while Khulna recorded the highest prevalence of deprivation ACEs (80.17%). Among women, Khulna and Mymensingh displayed the highest prevalence of threat ACEs (69.43%) and deprivation ACEs (46.27%), whereas Sylhet and Rangpur had the lowest prevalence of threat ACEs (22.47%) and deprivation ACEs (20.29%), respectively. The type of residence was significantly associated with deprivation-related ACEs among women (55.82% in urban areas vs 48.13% in rural areas, p=0.017), but not among men. Threat ACEs did not significantly differ by residence type for either gender. Religiosity levels were significantly associated with ACEs, displaying different patterns across genders. Among men, those reporting ‘very little’ religiosity had a higher prevalence of threat ACEs (60.83%) than those with ‘moderate’ (38.94%) or ‘regular’ (46.34%) religiosity (p<0.001). However, religiosity was not significantly associated with deprivation ACEs in men (p=0.107). Conversely, among women, participants with ‘moderate’ religiosity exhibited higher prevalence rates of both threat ACEs (51.19%) and deprivation ACEs (57.62%) compared with those with ‘very little’ (36.6% for threat and 53.61% for deprivation) and ‘regular’ (37.71% for threat and 43.02% for deprivation) religiosity levels (p<0.001 for both ACE types). Lower SES scores were significantly associated with deprivation ACEs across genders (men: M=6.55 vs 5.89, p<0.001; women: M=6.67 vs 6.48, p=0.005). Additionally, among men, threat ACEs were associated with lower SES scores (M=6.03 vs 6.32, p=0.045), whereas no significant association was observed among women (p=0.680). The educational attainment of both participants and their parents was significantly related to deprivation ACEs in both genders. Participants with lower education levels reported a higher prevalence of deprivation ACEs. For instance, among men, 53.55% of those with primary education experienced deprivation ACEs compared with 21.26% of those with a bachelor’s degree or higher. Similarly, 62.20% of women with no schooling experienced deprivation ACEs, whereas 36.25% of those with a bachelor’s degree or above reported the same (p<0.001). Parental education also demonstrated strong associations; men with fathers lacking schooling had a higher prevalence of deprivation ACEs (53.54%) compared with those with fathers holding a higher school certificate (HSC) or higher education (15.79%) (p<0.001). Women exhibited similar trends, with 55.21% of those whose fathers had no schooling experiencing deprivation ACEs vs 26.14% among those with fathers holding HSC or higher education (p<0.001). Maternal education was also significantly associated with deprivation ACEs, with men whose mothers had no schooling reporting 53.26% compared with 18.87% among those with mothers holding HSC or higher education (p<0.001). Among women, 55.23% of those with mothers having no schooling experienced deprivation ACEs compared with 40.74% among those with higher maternal education (p=0.001). Deprivation ACEs were associated with specific gender norm attitudes in both genders. Among men, those with deprivation ACEs had significantly lower total M-GNAS scores (M=30.33 vs 27.57, p<0.001) and scored lower on the ‘family responsibility’ (p<0.001) and ‘career choice’ (p<0.001) factors. In women, deprivation ACEs were associated with lower scores in ‘family financial decision’ (p=0.003) and ‘career choice’ (p=0.039), although overall M-GNAS scores did not differ significantly (p=0.070). Association of ACEs with IPV perpetration Multivariable regression models (table 1) confirmed childhood adversities as independent predictors of IPV perpetration after comprehensive adjustment for other variables. Threat ACEs were associated with approximately 2.6 times higher odds of perpetration (adjusted OR (aOR)=2.57; 95% CI 2.03 to 3.23), while deprivation ACEs increased the odds by 75% (aOR=1.75; 95% CI 1.39 to 2.20). The female gender was strongly protective (aOR=0.41; 95% CI 0.32 to 0.52), corroborating the descriptive findings of lower female perpetration prevalence (41.53% vs 59.88%). Contextual factors significantly moderated the odds of perpetration. For example, participants from Barisal faced 4.4 times higher odds (95% CI 2.74 to 7.06) compared with those from Dhaka. A one-unit increase in SES was associated with 17% lower odds of IPV perpetration (aOR=0.83; 95% CI 0.77 to 0.88). Secondary education decreased odds by 41%–47% compared with no schooling, reinforcing education’s buffering role observed in descriptive analyses. Significantly, traditional gender attitudes correlated with ACEs exposure in descriptive tables (online supplemental file 3) but showed limited predictive power in adjusted models. Only progressive career attitudes reduced the odds of IPV (aOR=0.93; 95% CI 0.89 to 0.97), contrasting with the descriptive pattern among men, where threat exposure was strongly associated with restrictive career attitudes. Regional effects remained robust: Chattogram division showed 74% lower odds (aOR=0.26; 95% CI 0.18 to 0.38), while Mymensingh’s protective effect (aOR=0.64; 95% CI 0.41 to 0.99) contradicted its high ACEs prevalence, suggesting resilient community factors that merit qualitative investigation. Urban–rural upbringing, religiosity and parental education showed no significant associations despite descriptive variations. Discussion The present study examined the association between threat and deprivation types of ACEs and IPV perpetration among young adults in Bangladesh. Consistent with a previous study on men’s perpetration,5 more than half of the young men in Bangladesh reported committing IPV. Uniquely, we found that over 40% of women also do the same. One of the most important findings of this study is the robust association between experiencing threats in childhood and IPV perpetration. This is consistent with previous research highlighting the role of early exposure to violence and other forms of harm in shaping aggressive and violent behaviours in adulthood.9 The link between threat-type ACEs and IPV perpetration points to the essential role of interventions aimed at addressing emotional dysregulation and trauma-related responses that result from childhood violence and neglect. These findings underline the need for trauma-informed approaches in both preventing IPV and treating individuals with a history of childhood adversity. Deprivation-type ACEs, on the other hand, were also significantly associated with IPV perpetration. Deprivation, characterised by the absence of cognitive and social stimulation during childhood, can impair interpersonal communication and social skills, which are crucial for conflict resolution and emotional expression.16 45 46 These deficits might contribute to IPV perpetration for both genders, especially for women, who face additional social and economic stressors. Therefore, it is essential to incorporate social skill training and emotional regulation strategies into IPV prevention programmes, particularly for women who have experienced childhood deprivation. For instance, programmes could focus on enhancing communication skills and conflict resolution techniques, informed by the socioeconomic challenges women face in Bangladesh. While this study did not find a moderating effect of gender, supplementary gender-stratified analyses yield several descriptive insights (online supplemental file 3). While both men and women were affected by childhood threats, women appeared more sensitive to the effects of deprivation. This suggests that women’s IPV perpetration may be driven by different psychosocial mechanisms than men’s, emphasising the importance of gender-sensitive interventions. The literature often focuses on men as primary perpetrators of IPV, with women considered predominantly as victims.6 The present findings challenge this dichotomy by illustrating that women can also perpetrate violence, although through different pathways. Future research should continue to explore these gendered dynamics, particularly in culturally specific contexts like Bangladesh, where rigid gender roles may influence how violence is expressed. Another important finding is significant regional variation in childhood adversities across divisions in Bangladesh, resulting in diverse IPV perpetration (eg, individuals from Barisal and Rajshahi exhibit higher odds of perpetrating IPV). These findings suggest that local cultural, economic, and social conditions may play a crucial role in shaping IPV dynamics. For example, regional differences in economic development, social support systems, and access to education and resources may contribute to these disparities. Policymakers and practitioners should consider these regional differences when designing IPV prevention and intervention programmes, ensuring that strategies are tailored to the unique socio-cultural contexts of each region. Additionally, the protective effect of higher SES on IPV perpetration aligns with previous research suggesting that economic empowerment can reduce the risk of violence.27 Higher SES may provide individuals with greater access to resources, social support, and education, which can buffer against the stresses that contribute to violent behaviour. This finding advocates the need for multi-level interventions that address both individual-level factors, such as childhood trauma, and structural determinants, such as poverty, to reduce IPV perpetration. To our knowledge, this is the first study that utilises a nationally representative youth sample comprising both genders to demonstrate the association between childhood adversity and IPV perpetration in adults, particularly in the context of LMICs. The finding that both men’s and women’s experiences of childhood threat and deprivation increase the odds of IPV differentially in adulthood underscores the need for gender-sensitive childcare programmes to address persistent gender disparities. These findings highlight the importance of tailored IPV prevention strategies, such as anger management and aggression-focused therapies for men and interventions addressing emotional regulation and relational insecurities for women.22 Several limitations must be acknowledged. Although this study used a cross-sectional design, the temporal ordering between ACEs (experienced before age 18) and adult IPV perpetration provides a degree of implicit temporality that strengthens the plausibility of the observed associations. Nonetheless, the cross-sectional design precludes definitive causal claims, as unmeasured confounders (eg, mental health conditions, substance use) and recall bias in retrospective ACE reporting may have influenced the findings. Longitudinal studies are needed to further clarify the causal pathways between childhood adversity dimensions and IPV perpetration. Our use of the lifetime HITS screen constrains understanding of IPV type, frequency, duration and recency. Future work should use multi-item and time-bound measures. Furthermore, the definition of partner violence has cultural implications. For example, shouting and screaming (considered IPV perpetration in line with many Western, Educated, Industrialised, Rich and Democratic countries in this study) might be regarded as normal behaviour, especially in the LMIC context. In addition to addressing the aforementioned limitations, future studies should investigate the meaning and patterns of perpetration rooted in the Bangladeshi cultural context and their association with IPV behaviour. Conclusions This study provides important insights into the relationship between the threat and deprivation dimensions of ACE and IPV perpetration among young adults in Bangladesh. The findings highlight the importance of adopting a dimensional approach to ACEs, addressing gender differences in IPV dynamics and considering regional and socioeconomic factors in IPV prevention efforts. By integrating these findings into public health interventions and policy frameworks, Bangladesh and other LMICs can develop more effective, culturally sensitive strategies to reduce IPV and its devastating consequences. Supplementary material Footnotes Funding: The authors acknowledge financial support from the BRAC Advocacy Programme for this study (Grant Number: NA). Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2025-106731). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: The Institutional Review Board of the Institute of Health Economics at the University of Dhaka granted ethical clearance (IHE/IRB/DU/01/2022/Final). Participants gave informed consent to participate in the study before taking part. Data availability free text: The data generated during and/or analysed during the current study are available at https://doi.org/10.17632/z4g7xj4r3h.1. Collaborators: Not applicable. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Data availability statement The data generated during and/or analysed during the current study are available at (Dataset).47 References Supplementary Materials Data Availability Statement The data generated during and/or analysed during the current study are available at (Dataset).47