Chronicity of depressive symptoms and intimate partner violence experiences over time among sexual minority men: Substance use as a pathway

Abstract

Objective. –

Past research found depression, substance use, and intimate partner violence (IPV) to be prevalent among sexual minority men (SMM) and noted on their synergistic relationship, based on the syndemic framework. In exploration of potential mechanisms, we examined depression as a predictor of future IPV experiences, with substance use as a mediator.

Method. –

We used three timepoints of longitudinal data (baseline, 6-month, and 12-month follow-up) of 500 SMM from the US. Directionality between IPV experiences and depressive symptoms was assessed using cross-lagged panel modeling across three timepoints. Then we used linear and generalized structural equation modeling (SEM) to examine the associations between depressive symptoms from baseline to 6-month follow-up and IPV experiences at 12-month follow-up with mediation via substance use at 6-month follow-up. For generalized SEM models, depressive symptoms were categorized as no symptoms, intermittent, and chronic. All SEM models adjusted for corresponding baseline IPV experiences and sociodemographic characteristics.

Results. –

Prevalence of any experiences of IPV at 12-month was 25.5% for victimization and 17.9% for perpetration. Chronic depressive symptoms were directly associated with any experiences of both IPV victimization and perpetration, with no evidence of mediation through substance use. Conversely, substance use fully mediated the association between intermittent depressive symptoms and IPV victimization.

Conclusion. –

These findings suggest that IPV prevention programs for SMM should incorporate strategies to address mental health and its coping mechanisms, including substance use. There is a need for greater investment towards raising awareness and development of syndemic-focused interventions to address multiple, co-occurring health challenges among SMM.

Intimate partner violence (IPV) is an act of violence often in the forms of physical, emotional, and sexual abuse, and coercive control (World Health Organization & London School of Hygiene and Tropical Medicine, 2010) that is committed to assert power and gain control over intimate partners (De Coster & Heimer, 2021). Experiences of IPV transcends gender and sexual orientation and have been shown to be prevalent within same-sex partnerships between men. A meta-analysis reported lifetime prevalence rates as high as 33% for IPV victimization and 29% for IPV perpetration among sexual minority men (SMM) (Liu et al., 2021). Given the notable prevalence of IPV among SMM, research has explored how stigmatization and marginalization rooted in heterosexism contribute to heightened vulnerability and its broader public health impact (Callan et al., 2021; Hong et al., 2025).

Syndemic theory offers an important lens for understanding IPV and its relation to health and well-being among SMM (Singer et al., 2017). Syndemic refers to clustering of social and health problems that interact synergistically within socially marginalized populations resulting in heightened health burdens (Singer et al., 2017). Prior work has identified IPV as a key component of the syndemic specific to SMM, particularly in relation to substance use and human immunodeficiency virus (HIV) risk among SMM (Hart et al., 2017; Singer et al., 2017; Stall et al., 2003; Walters et al., 2020). Studies have further documented on the co-occurrence between IPV and poor mental health, especially depression (Hong et al., 2025), as well as with substance use (Duncan et al., 2018; Metheny et al., 2024; Scheer & Mereish, 2021; Wu et al., 2015) among SMM.

While syndemic theory does not specify a directionality in the relationships between IPV and other syndemic conditions, a majority of past literature has positioned IPV as a risk factor for depressive symptoms (Hong et al., 2025) and substance use (Duncan et al., 2018; Metheny et al., 2025; Metheny et al., 2024; Scheer & Mereish, 2021; Wu et al., 2015). A pooled analysis of 18 studies reported that experiences of IPV were associated with nearly doubling the odds of depressive symptoms among SMM (Hong et al., 2025). However, all included studies employed cross-sectional designs, with recall periods for IPV ranging from the past 30 days to lifetime, thereby limiting the capacity to infer a causal relationship between IPV and depressive symptoms (Hong et al., 2025). Moreover, a longitudinal study of young SMM found that depression was associated with both IPV victimization and perpetration after accounting for within-subject correlations (Stults et al., 2021). However, a temporal precedence of depression was not ascertained.

With respect to substance use and its relationship with IPV among SMM, longitudinal studies have examined associations in both directions (Metheny et al., 2025; Metheny et al., 2024; Stults et al., 2021). Two studies reported that experiences of IPV victimization were associated with subsequent increases in risky drinking and use of other substances among sexual and gender minorities (Metheny et al., 2025; Metheny et al., 2024). Stults et al. (2021) presented evidence that the use of substances other than alcohol was associated with both IPV victimization and perpetration; however, their results did not establish temporal ordering among these experiences. Given the emphasis of syndemic theory on complex synergistic relationships between health and health-related conditions (Singer et al., 2017), the existing literature underscores a need to examine directionalities in the pathways between depression, substance use, and IPV among SMM, with attention to depression and substance use as potential antecedents of future experiences of IPV.

Depression and substance use have been noted to predict IPV victimization and perpetration among studies with men and women in different-sex relationships (Bacchus et al., 2018; Kim & Lee, 2013; Saunders et al., 2023; Yu et al., 2019). Several longitudinal studies of women have reported on associations between depression and subsequent IPV experiences (Bacchus et al., 2018; Iverson et al., 2022; Kim & Lee, 2013). Studies with married women in Korea (Kim & Lee, 2013) and US women veterans (Iverson et al., 2022) found that depressive symptoms predicted future experiences of IPV victimization, after adjusting for past experiences of IPV victimization. Depression has also been noted as a factor of IPV perpetration among the general population (Saunders et al., 2023), with specific attention to its comorbidity with substance use for men in different-sex relationships (Yu et al., 2019). Men with comorbid depression with either alcohol use disorder or substance use disorder were found to have an increased rate of perpetration of IPV against women (Yu et al., 2019). These studies assessing the temporal pathways between depression, substance use, and IPV among individuals in different-sex relationships have yet to be replicated among SMM.

Previous research noted that IPV victimization and perpetration are potentially linked to the diminished cognitive and affective functioning characteristic of depression (Birkley & Eckhardt, 2015; Dutton & Karakanta, 2013; Iverson et al., 2011; Iverson et al., 2022; Maloney et al., 2023; Yu et al., 2019). Depression among IPV survivors may manifest in the forms of internalization of negative affect such as guilt, worthlessness, and helplessness and diminished physical and emotional capacity to assert themselves in violent situations and also to leave the relationship, thereby resulting in greater vulnerabilities towards IPV (Iverson et al., 2011; Iverson et al., 2022; Yu et al., 2019). Iverson et al. (2011) found that cognitive behavioral therapy targeting symptoms of posttraumatic stress disorder (PTSD) and depression and the resulting reductions in the symptoms were associated with decreased reports of IPV victimization among women with PTSD. Simultaneously, dysregulation of negative affect due to depression is linked to increased anger and hostility (Dutton & Karakanta, 2013), which may escalate into violence against intimate partners (Barros-Gomes et al., 2019; Birkley & Eckhardt, 2015; Maloney et al., 2023). Use of alcohol and substances in conjunction to depressive symptoms would further exacerbate impairments in judgment, emotion regulation, and impulse control (Jakubczyk et al., 2018; Weiss et al., 2022), thereby reinforcing the existing pathways between depression and IPV experiences. Furthermore, SMM with experiences of IPV may be at greater risk due to the stressors attributed to their sexual minority status, such as anti-gay discrimination, internalized homophobia, and outness (Finneran & Stephenson, 2014; Pless & Gibbs, 2025; Stephenson & Finneran, 2017). These stressors not only are detrimental to depressive symptoms and substance use (Hatzenbuehler et al., 2008; Hoy-Ellis, 2023; Parent et al., 2019), but also have been shown to be associated with increased vulnerability to IPV victimization among SMM (Finneran & Stephenson, 2014; Pless & Gibbs, 2025; Stephenson & Finneran, 2017), thereby contributing to a syndemic context in which sexual minority stress, mental health concerns, substance use, and IPV mutually reinforce one another.

Informed by the syndemic framework and the cognitive and emotional regulation models that explain the role of depression and substance use in IPV experiences, the present study aims to address the limited research on the temporal pathways linking depressive symptoms, substance use, and IPV experiences among SMM. Using data collected with a cohort of partnered SMM collected across 3 timepoints over a 12-month period, we examined longitudinal associations between depressive symptoms and both IPV victimization and perpetration to assess for directionality of influence between depressive symptoms and IPV experiences. Based on the analyses, we examined the role of substance use as a potential mediator in their relationships, with specific focus on severity and chronicity of depressive symptoms over time. It is hypothesized that 1) there would be temporal relationships between depressive symptoms and IPV experiences among SMM. Also, it is hypothesized that 2) substance use would mediate the association between depressive symptoms and future IPV experiences for both victimization and perpetration, where 3) a more severe and chronic depression would predict increased substance use, in turn predicting higher rates of IPV victimization and perpetration.

Methods

Participants and Procedures

The data came from the EROS (Empowering Relationships and Opportunities for Safety) study, which is an ongoing observational cohort study aimed to assess IPV experiences among SMM residing in the US and examine their relationship to HIV prevention (Storholm et al., 2022). The current analyses used three timepoints of survey data (baseline, 6-month follow-up, and 12-month follow-up). The study protocol was approved by the San Diego State University Institutional Review Board (HS-2022–0094), and the details of the study protocol are available elsewhere (Storholm et al., 2022).

The parent study collected its baseline data between September 2022 and December 2023. Participants were recruited through advertisements in online and offline settings, including social networking sites, dating hook-up apps, LGBTQ venues, and events. A link and quick response (QR) code to a screener, administered by Qualtrics survey software, were included in the advertisement materials. A total of 9,010 people completed the screener and 1,042 were eligible for the study. To be eligible, one had to be a) a cisgender man, b) in a relationship with a cisgender man for longer than 3 months at the time of screening, c) between 18 and 45 years old, d) HIV negative or unknown HIV status, e) living in a CDC-identified Ending the HIV Epidemic priority jurisdictions in the US (Fauci et al., 2019), and f) willing to complete at-home HIV and sexually transmitted infection test kits. Partners of enrolled participants were excluded to ensure safety of participants in the enrollment of study that required disclosure of information on IPV and other information regarding their relationship. In addition, the online survey program was developed to incorporate IPV safety protocols, which included quick exit buttons in every page, tips on clearing browser history, resources such as number to the National Domestic Violence Hotline, and contact information to the study team for further IPV-related resources and assistance with getting linked to local services. Study personnel contacted and virtually onboarded eligible screener respondents to verify identity and obtain verbal agreement and electronic signature of informed consent.

A total of 690 completed the baseline survey and 522 returned their baseline test kit. After excluding 22 individuals with positive HIV test at baseline, a cohort of 500 SMM was established to be followed for 24 months, with a survey and test-kit follow-up conducted every 6 months. The study is currently ongoing, and this study includes survey responses from baseline, 6-month, and 12-month follow-up. At each of the follow-up timepoints, 474 (6-month, follow-up rate; 94.8%) and 448 (12-months, follow-up rate: 89.6%) completed the survey. Participants were compensated $25 for each of the surveys they completed.

Measures

Depressive symptoms.

Patient Health Questionnaire-8 (PHQ-8) was used to measure depressive symptoms. This 8-item scale asked about the frequency of depressive symptoms in the past two weeks using a 4-point Likert scale (0: not at all – 3: nearly every day)(Kroenke et al., 2009). Depressive symptoms were collected at every timepoint. Cronbach’s alpha of PHQ-8 scales across three time ranged between 0.90 and 0.93, demonstrating high internal consistency.

PHQ-8 scale was used in two different ways in the current study to indicate depressive symptoms. First, individual items were modeled as indicators of a latent construct of depressive symptoms for use with analyses using structural equation models (SEM). Secondly, summed PHQ-8 scores were dichotomized to examine chronicity of depressive symptoms from baseline to 6-month follow-up. PHQ-8 measured at baseline and 6-month follow-up were each summed to give a score ranging from 0 to 24, which were dichotomized using a cutoff of 10 or more (Kroenke et al., 2009), which was validated to be the cutoff with maximized combined sensitivity and specificity (Wu et al., 2020). The dichotomous variables created at each timepoint were merged to give a three-category variable of chronic depressive symptoms: none (0 in both baseline and 6-month follow-up), intermittent (1 in either baseline or 6-month follow-up), and chronic (1 in both baseline and 6-month follow-up). This allowed for comparison of substance use and IPV prevalence between SMM whose depressive symptoms remained below the cutoff across six months and those with depressive symptoms above the threshold in the past 6 months, with specific focus on those in the chronic category, which would not be captured in continuous-score analyses of depressive symptoms.

Intimate partner violence.

Participants responded to two sets of 62 yes/no items on IPV experiences in the past 6 months, the first set asking about experiences of victimization and the second set asking about experiences of perpetration. Of the 62 items, 21 were sourced from the Intimate Partner Violence among Gay and Bisexual Men (IPV-GBM) scale, a validated measure of IPV among SMM, after consultations with the author of the scale (Stephenson & Finneran, 2013). An additional 39 items were developed by the research team based on findings from initial qualitative interviews with SMM who reported recent IPV experiences and with IPV-related service providers for SMM IPV survivors and perpetrators conducted during a previous phase of the parent study (Jessup et al., 2024; Storholm et al., 2023). Finally, 3 items on economic abuse were added based on items developed by Postmus et al. (2016). In total, the IPV questionnaire measured the following eight domains: physical abuse, sexual abuse, coercive control, sexual health, emotional abuse, stalking, identity-based abuse, and economic abuse.

Two parallel sets of IPV items were asked, first on IPV victimization participants experienced from a male partner, and second on IPV perpetrated by participants against a male partner. These same items were administered again at every timepoint. The responses were both summed and dichotomized as any versus none to compose variables on number of IPV experiences and any experiences of IPV, respectively.

Substance use.

Substance use was measured at baseline and 6-month follow-up using a modified version of question 2 from National Institute on Drug Abuse-Modified Alcohol, Smoking and Substance Involvement Screening Test (NIDA-Modified ASSIST) V2.0 (National Institute on Drug Abuse), which was originally adapted from the test of the same name developed by World Health Organization (Humeniuk et al., 2010). Question 2 of the NIDA-Modified ASSIST V2.0 was intended to ask about the frequency of use of the following 10 substances in the past 3 months, with corresponding street names for each substance: cannabis, cocaine, prescription stimulants, methamphetamines, inhalants, sedatives or sleeping pills, hallucinogens, street opioids, prescription opioids, and other. The following modifications were made in this study: binge drinking, defined as five or more drinks within a couple of hours for men (Wechsler & Nelson, 2001), was added as an item; and reference period was changed to past 6 months to match the survey intervals. In total, 11 items (i.e., 10 substances and binge drinking) composed our substance use scale.

The frequency of substance use was measured using the following scale: 0, never; 2, once or twice; 3, monthly; 4, weekly; and 6, daily or almost daily (National Institute on Drug Abuse). The responses were summed up to create a continuous variable (range: 0–66); in our study the scores ranged between 0 and 58 at baseline and between 0 and 54 at 6-month follow-up.

Covariates.

To adjust for potential confounding by sociodemographic characteristics, we included the following variables from baseline as covariates: age (continuous), race and ethnicity (Latino/Hispanic, Black/African American, Asian American and Pacific Islander, White, and Multiracial/other), highest attained education level (high school graduate (including General Educational Development [GED]), some college, college graduate, and graduate school degree), residential situation (living at my own place, a rented place, and someone else’s place at no cost), and employment status (employed, self-employed, home maker or student, and out of work).

Analytic Plan

Descriptive Analyses

Distribution of any experiences of IPV victimization and perpetration at 12-month follow-up across key variables of interest and covariates measured at baseline. Descriptive analyses included 424 participants, after excluding loss to follow-up (n = 52) and participants with missing responses (n = 24). For continuous variables such as age and substance use scores, mean and standard deviation were calculated for the total sample and for each category of any IPV victimization and perpetration at 12-month follow-up. Chi-square tests were conducted for categorical variables and analyses of variance were conducted for continuous variables to examine bivariate associations with any IPV victimization and perpetration. All statistical analyses were conducted using STATA/MP 16.1 (StataCorp, 2019).

Cross-lagged panel model between depressive symptoms and IPV experiences

To test hypothesis 1, we conducted cross-lagged panel model analyses to examine the directional pathways between IPV experiences and depressive symptoms (Figure 1), based on prior research examining directionality between IPV and depressive symptoms in US female veterans (Iverson et al., 2022). Depressive symptoms were modeled as a latent construct at each timpoint. Each of the PHQ-8 items were specified to load onto a single latent factor indicating depressive symptoms at each timepoints given the unidimensional structure of the PHQ-8 scale (de la Torre et al., 2023). Continuous measures of number of IPV experiences were used from all three timepoints. Models were separately tested for IPV victimization and perpetration.

Figure 1.

Conceptual model for cross-lagged panel model analysis of depressive symptoms and experiences of IPV across three timepoints

Note: All pathways included the following covariates measured at baseline: age (continuous), race and ethnicity, highest attained education level, place of residence, and employment. Within-wave associations between IPV and depression were modeled by estimating a covariance at baseline and residual covariances at subsequent waves.

IPV, intimate partner violence

The models incorporated autoregressive paths for depressive symptoms and number of IPV experiences across adjacent timepoints as well as direct autoregressive paths from baseline to 12-month follow-up. Cross-lagged paths between depressive symptoms and number of IPV experiences were also specified across adjacent timepoints and from baseline to 12-month follow-up. Within-wave associations between IPV experiences and depressive symptoms were modeled by estimating a covariance at baseline and residual covariances at subsequent timepoints. All pathways simultaneously controlled for covariates measured at baseline. Missing data were handled using full information maximum likelihood (Enders, 2001; Lang & Little, 2018), and standard errors were obtained using 500 bootstrap samples.

Based on the results from the cross-lagged panel models, two sets of mediation analyses were conducted to examine the role of substance use as a mediator in the longitudinal pathway from depressive symptoms to future IPV experiences and test for hypotheses 2 and 3. Figure 2 shows the conceptual design of the mediation analyses.

Figure 2.

Conceptual model for mediation analysis of depressive symptoms over time and experiences of IPV, mediated by substance use

Note: All pathways included the following covariates measured at baseline (age (continuous), race and ethnicity, highest attained education level, place of residence, and employment) and corresponding measure of baseline IPV experiences.

IPV, intimate partner violence.

Depressive symptoms over time as latent variable and IPV experiences and mediation by substance use (SEM)

First, we estimated SEM to examine mediation via substance use at 6-month follow-up in the longitudinal association between depressive symptoms over time (baseline and 6-month follow-up) and number of IPV experiences at 12-month follow-up to test for hypothesis 2 using linear models. Depressive symptoms over time were modeled as a single latent construct using all PHQ-8 items across baseline and 6-month follow-up to represent overall severity of depressive symptoms. Models were separately tested for IPV victimization and perpetration. All pathways simultaneously controlled for covariates measured at baseline along with corresponding baseline IPV experiences (e.g., for model including IPV victimization, number of IPV victimization at baseline was adjusted). Full information maximum likelihood was used to handle missing data (Enders, 2001; Lang & Little, 2018) and standard errors were obtained using 500 bootstrap samples. To decompose the effects, nlcom command was used to generate estimates of direct effects, indirect effects via substance use, and total effects in the association between depressive symptoms and IPV experiences at 12-month.

Chronic and intermittent depressive symptoms and IPV experiences and mediation by substance use (generalized SEM)

We further conducted generalized SEM using categorical variable of chronic depressive symptoms and any IPV experiences to examine the chronicity of depressive symptoms as a predictor using the gsem code and test for hypotheses 2 and 3. We estimated direct and indirect effects of chronic and intermittent depressive symptoms (baseline to 6-month follow-up) on IPV experiences at 12-month follow-up, with substance use at 6-month follow-up as a mediator. Models were separately tested for IPV victimization and perpetration. To assess the pathway between depressive symptoms and substance use, a linear model with robust standard errors was used to correct for heteroskedasticity. For the pathways leading to any IPV experiences, Poisson regression with robust standard errors were used to account for high prevalence of both any IPV victimization (25.5%) and perpetration (17.9%) at 12-month (Zou, 2004). All covariates and corresponding baseline measure of any IPV experiences were simultaneously included in all analyzed pathways.

Missing data were handled using multiple imputation by chain equations, where 20 imputed data sets were generated and analyses were combined using Rubin’s rules (Azur et al., 2011; Rubin, 1987; Wulff & Ejlskov, 2017). Sums of PHQ-8 scores at baseline and 6-month follow-up were passively imputed, which were then dichotomized and combined to give a measure of chronic depressive symptoms. Summed score of substance use at 6-month follow-up was also passively imputed. Summed score of substance use at baseline was included as an auxiliary variable to guide imputation. All of the covariates were included as regular variables. Participants with missing baseline auxiliary variables required for the multiple imputation model (n = 9; <2%) were excluded prior to imputation because their inclusion prevented estimation of the chained equations (N = 491). Separate multiple imputations were carried out prior to each generalized SEM estimations for IPV victimization and perpetration.

Since depressive symptoms were categorized into 3 groups to test for hypothesis 3, coefficients, standard errors, and 95% confidence intervals (CI) were calculated for intermittent and chronic depressive symptoms, with no depressive symptoms as the reference category. To decompose the effects, nlcom command was used to generate estimates of direct effects, indirect effects via substance use at 6-month, and total effects in the association between different chronicity levels of depressive symptoms from baseline to 6-month follow-up and IPV experiences at 12-month follow-up.

Results

Descriptive analyses

Table 1 presents the sociodemographic make-up of the 424 SMM participants in the study, along with their depressive symptoms across baseline and 6-month follow-up, substance use scores, and any experiences of IPV victimization and perpetration at baseline and 12-month follow-up. The mean age of the participants was 32 and a majority were non-White (33.0% Latino/Hispanic, 31.4% Black and African American, and 16.5% Asian American and Pacific Islander), had college degrees (41.3% college graduates, 34.7% graduate school degrees), living in a place they rent (64.6%), and employed full-time or part-time (81.1%). One in eight (12.5%) had reported depressive symptoms at both baseline and 6-month, and 15.8% reported in one of the timepoints.

Table 1.

Distribution of intimate partner violence victimization and perpetration across key variables and covariates among 424 sexual minority men

	Total	Any IPV victimization (12M)			Any IPV perpetration (12M)
Variablesa		Yes	No		Yes	No
	N (%)	N (%)	N (%)	p valueb	N (%)	N (%)	p valueb
Total:	424 (100)	108 (25.5)	316 (74.5)		76 (17.9)	348 (82.1)
Age (BL)c	32.8 ± 6.3	32.9 ± 6	32.8 ± 6.4	0.942	33.5 ± 6.4	32.7 ± 6.2	0.272
Race and ethnicity (BL)				0.698			0.518
Latino/Hispanic	140 (33.0)	37 (26.4)	103 (73.6)		22 (15.7)	118 (84.3)
Black and African American	133 (31.4)	33 (24.8)	100 (75.2)		29 (21.8)	104 (78.2)
Asian American and Pacific Islander	70 (16.5)	17 (24.3)	53 (75.7)		13 (18.6)	57 (81.4)
White	63 (14.9)	14 (22.2)	49 (77.8)		8 (12.7)	55 (87.3)
Multiracial/Other	18 (4.2)	7 (38.9)	11 (61.1)		4 (22.2)	14 (77.8)
Highest attained education level (BL)				0.344			0.585
HS Graduate (includes GED)	15 (3.5)	6 (40.0)	9 (60.0)		3 (20.0)	12 (80.0)
Some college	87 (20.5)	20 (23.0)	67 (77.0)		18 (20.7)	69 (79.3)
College graduate	175 (41.3)	40 (22.9)	135 (77.1)		26 (14.9)	149 (85.1)
Graduate school degree	147 (34.7)	42 (28.6)	105 (71.4)		29 (19.7)	118 (80.3)
Residential situation (living in a…; BL)				0.045			0.756
Home I own	109 (25.7)	18 (16.5)	91 (83.5)		17 (15.6)	92 (84.4)
Home I rent	274 (64.6)	78 (28.5)	196 (71.5)		51 (18.6)	223 (81.4)
Someone else’s home at no cost	41 (9.7)	12 (29.3)	29 (70.7)		8 (19.5)	33 (80.5)
Employment status (BL)				0.235			0.147
Employed (full-time or part-time)	344 (81.1)	85 (24.7)	259 (75.3)		58 (16.9)	286 (83.1)
Self-employed	30 (7.1)	10 (33.3)	20 (66.7)		9 (30.0)	21 (70.0)
Home maker or student	22 (5.2)	3 (13.6)	19 (86.4)		2 (9.1.0)	20 (90.9)
Out of work	28 (6.6)	10 (35.7)	18 (64.3)		7 (25.0)	21 (75.0)
Depressive symptoms (BL and 6M)				< 0.001			< 0.001
None	304 (71.7)	57 (18.8)	247 (81.3)		39 (12.8)	265 (87.2)
Intermittent	67 (15.8)	23 (34.3)	44 (65.7)		16 (23.9)	51 (76.1)
Chronic	53 (12.5)	28 (52.8)	25 (47.2)		21 (39.6)	32 (60.4)
Substance use (6M)c, d	4.9 ± 5.5	6.3 ± 5.6	4.5 ± 5.4	0.004	6.0 ± 5.8	4.7 ± 5.4	0.071
Any IPV victimization (BL)				< 0.001
Yes	155 (36.6)	70 (45.2)	85 (54.8)
No	269 (63.4)	38 (14.1)	231 (85.9)
Any IPV perpetration (BL)							< 0.001
Yes	110 (25.9)				46 (41.8)	64 (58.2)
No	314 (74.1)				30 (9.6)	284 (90.4)

IPV intimate partner violence, BL baseline, 6M 6-month follow-up, 12M 12-month follow-up, HS high school, GED general education development

^a.Time points at which the variables were collected are noted in parentheses

^b.P-values for chi-square tests (for categorical covariates) and analysis of variance (for continuous covariates) with any IPV victimization or perpetration at 12-month follow-up

^c.Continuous variables presented as mean ± standard deviation

^d.Summed total score of scale measuring substance use in the past 3 months that was adapted from question 2 from NIDA ASSIST V2.0. Scores composed of both frequency and number of types of substances used, with higher scores indicating greater use of substances.

Prevalence of IPV victimization and perpetration reported at their 12-month follow-up was 25.5% and 17.9%, respectively, which were lower than the baseline prevalence of each type of IPV (36.6% victimization, 25.9% perpetration). Of note, those who reported IPV victimization at baseline had a greater likelihood of reporting abuse at 12-month follow-up, and the perpetrators at baseline were also more likely to report perpetration at 12-month follow-up. The prevalence of both IPV victimization and perpetration increased with the chronicity of depressive symptoms. IPV victimization was also shown to have bivariate associations with residential situation and substance use, with substance use score at 6-month follow-up being statistically significantly higher among participants who reported IPV victimization at 12-month follow-up. (Table 1)

Cross-lagged panel model between depressive symptoms and IPV experiences

Figure 3 presents the results of the cross-lagged panel model analyses between depressive symptoms and IPV victimization (Panel A) and perpetration (Panel B). All autoregressive pathways among depressive symptoms were statistically significant. On the other hand, only the autoregressive pathways between adjacent timepoints were statistically significant for number of IPV victimization and perpetration. Of the cross-lagged pathways, only the pathway from depressive symptoms at baseline to number of IPV victimization at 6-month follow-up was statistically significant, providing partial support for hypothesis 1.

Figure 3.

Depressive symptoms and number of IPV experiences across three timepoints: a cross-lagged panel model analysis (N = 500)

Note: All pathways included the following covariates measured at baseline: age (continuous), race and ethnicity, highest attained education level, place of residence, and employment. Within-wave associations between IPV and depression were modeled by estimating a covariance at baseline and residual covariances at subsequent waves. Missing data were handled using full information maximum likelihood, and standard errors were obtained using 500 bootstrap samples.

* p < 0.05, ** p < 0.01, *** p < 0.001; IPV, intimate partner violence.

Depressive symptoms over time as latent variable and IPV experiences and mediation by substance use (SEM)

Figure 4 presents the results of the SEM analyses examining associations between depressive symptoms across baseline and 6-month follow-up and number of IPV experiences at 12-month follow-up, mediated by substance use at 6-month follow-up. Panel A shows that all pathways were statistically significant in positive directions. Both total and direct effects of depressive symptoms over time on greater number of IPV victimization in the future were statistically significant, and this association was partially mediated by substance use (15.6% of the total effect), consistent with hypothesis 2. Depressive symptoms over time were also associated greater number of IPV perpetration in the future (Panel B). However, contrary to hypothesis 2, there was no association between substance use at 6-month follow-up and IPV perpetration at 12-month follow-up, leading to no mediation effect via substance use for IPV perpetration.

Figure 4.

Association between depressive symptoms from baseline to 6-month follow-up and number of IPV experiences at 12-month follow-up, mediated by substance use at 6-month follow-up (N = 500)

Note: All pathways included the following covariates measured at baseline (age (continuous), race and ethnicity, highest attained education level, place of residence, and employment) and corresponding measure of baseline IPV experiences. Missing data were handled using full information maximum likelihood, and standard errors were obtained using 500 bootstrap samples.

* p < 0.05, ** p < 0.01, *** p < 0.001; IPV, intimate partner violence.

Chronic and intermittent depressive symptoms and IPV experiences and mediation by substance use (generalized SEM)

Figure 5 presents the results of the generalized SEM analyses examining the associations between depressive symptoms and each of IPV victimization and perpetration, with substance use as a potential mediator. Chronic depressive symptoms, compared to no depressive symptoms, showed statistically significant positive direct associations with both IPV experiences at 12-month follow-up; however, no mediation through substance use was observed. In contrast, intermittent depressive symptoms demonstrated a statistically significant positive indirect effect with IPV victimization at 12-month via substance use, but no statistically significant total or direct effect, indicating a full mediation effect. Specifically, intermittent depressive symptoms were associated with greater substance use, which in turn was associated with increased likelihood of IPV victimization, but not perpetration.

Figure 5.

Association between depressive symptoms by chronicity over 6 months and any experiences of IPV at 12-month follow-up, mediated by substance use at 6-month follow-up (N = 491)

Note: All pathways included the following covariates measured at baseline (age (continuous), race and ethnicity, highest attained education level, place of residence, and employment) and corresponding measure of baseline IPV experiences. Missing data were handled using multiple imputation, using baseline substance use as an auxiliary predictor.

* p < 0.05, ** p < 0.01, *** p < 0.001; IPV, intimate partner violence.

These results provide partial support hypothesis 3 regarding IPV victimization. Direct effect on IPV victimization was present among those with chronic depressive symptoms, but indirect effect via substance use was only present for those with intermittent depressive symptoms, thereby implying a complexity in the relationships between depression, substance use, and IPV victimization among SMM. With regard to IPV perpetration, the results did not support hypothesis 3.

Discussion

This study examines how depressive symptoms contribute to subsequent IPV experiences among SMM using longitudinal data. Through cross-lagged panel model, we found that depressive symptoms predicted future IPV victimization, with no evidence for the reverse pathway among SMM in relationship. Findings from mediation analyses demonstrated statistically significant total and direct effects between depressive symptoms and both future IPV victimization and perpetration, after controlling for baseline IPV experiences. These findings are consistent with prior research linking depression to both IPV victimization and perpetration using longitudinal data on young SMM (Stults et al., 2021) and women and men in different-sex relationships (Bacchus et al., 2018; Saunders et al., 2023; Yu et al., 2019) and extend the extant literature on IPV among SMM, which has been predominantly cross-sectional (Hong et al., 2025; Stults et al., 2021). In addition, substance use mediated the relationship between depressive symptoms and future IPV victimization, but not IPV perpetration. Further investigation found that the mediation through substance use in the pathway to increased vulnerability to IPV victimization may differ by the chronicity of depression among SMM, where the mediation through substance use may be more salient among those whose depressive symptoms are not chronic.

Previous research has noted that compromised cognitive and affective abilities, feelings of worthlessness, and low levels of energy that come with depression interfere with decision-making and reduce motivation to seek help or take action, consequently increasing one’s vulnerability to IPV victimization (Iverson et al., 2011). Also, social isolation and loneliness that are characteristic of depression may make depressive IPV survivors more vulnerable to be exploited by their partners (Elmer & Stadtfeld, 2020; Ge et al., 2017). Especially for those who are survivors of ongoing violence and control in their relationships, instances of IPV and symptoms of depression are likely to feed into one another, preventing them from breaking the cycle of victimization (Iverson et al., 2011; Iverson et al., 2022). For SMM survivors of IPV, struggles that stem from their poor mental health exist in conjunction with experiences unique to them. For example, enacted and internalized homophobia and their adverse mental health sequelae (Hoy-Ellis, 2023; Meyer & Frost, 2013), barriers to appropriate resources, and concealment of IPV experiences out of shame and denial of victimhood due to these experiences’ relation to masculinity and sexual minority identities (Huntley et al., 2019) contribute to their continued susceptibility to IPV victimization.

At the same time, research has shown that internalization and dysregulation of negative affect that are inherent in chronic depression may manifest as aggression (Berg et al., 2019; Birkley & Eckhardt, 2015; Dutton & Karakanta, 2013; Maloney et al., 2023). This helps explain the link between depression to IPV perpetration among SMM. Individuals who chronically experience depressive symptoms may become more irritable, impulsive, and prone to hostile communication, which can escalate into emotional or physical abuse against their partners (Birkley & Eckhardt, 2015; Dutton & Karakanta, 2013).

It is also important to interpret the association between depressive symptoms and IPV perpetration within the context of high prevalence of SMM who are both survivors and perpetrators of IPV (i.e., bidirectional IPV) (Kirschbaum et al., 2023; Wagner et al., 2024). Bidirectional IPV was also prevalent among SMM in our study: 95 reported both experiences of IPV victimization and perpetration at baseline, which was 22.4% of the analytic sample and 61.3% and 86.3% among those who experienced IPV victimization and perpetration, respectively (results not shown). Increased likelihood of perpetration of IPV among chronically depressed SMM may be understood both as the instigation of abuse that prompts retaliation from partners and as a dysregulated defensive response to partner violence. For example, survivors of IPV experiencing depressive symptoms may respond to IPV with violence instead of with nonviolent strategies such as de-escalation, partly due to limited cognitive capacities and increased impulsivity. Also, having chronic depressive symptoms may increase the likelihood of being involved in aggravated verbal conflicts, where both parties contribute to escalation of conflict, which may result in physical violence (Berg et al., 2019). Given the potential for such hostile relationship dynamics that may give rise to patterns of reciprocal abuse and mutual escalation, our findings highlight depressive symptoms as a critical point of intervention for reducing IPV in the SMM community.

Findings were mixed for SMM with less chronic depressive symptoms, illustrating the complexities of syndemic relationships among IPV, depression, and substance use. A full mediation effect through substance use was observed between intermittent depressive symptoms and subsequent IPV victimization, with no evidence of a direct association. On the other hand, there was no association observed between intermittent depressive symptoms and subsequent IPV perpetration. The difference in the findings is attributed to substance use being associated with subsequent victimization but not perpetration of IPV, after controlling for corresponding baseline IPV experiences and depressive symptoms. Overall, these findings suggest that the impact of depression and substance use on future IPV among SMM varies by both the chronicity of depressive symptoms and the type of IPV (victimization versus perpetration).

Of note, substance use did not mediate the association between chronic depressive symptoms and IPV victimization among SMM. Putting together the results from both linear and generalized SEM analyses, the direct effect between depressive symptoms and future IPV victimization may have been driven by those who were chronically depressed, whereas the indirect pathway via substance use was driven by those who were intermittently depressed. The indirect effect via substance use was marginally significant (p = 0.050) among the chronically depressed, which may imply that a statistically significant association might have been observed with a larger sample of SMM. This may also be interpreted that the strong predictive power of chronic depressive symptoms on future IPV victimization might have overshadowed the relatively modest indirect effect operating through substance use, compared to those who experienced depressive symptoms intermittently.

Differential indirect effects via substance use in predicting future IPV victimization may be explained by the degree to which substance use adds cognitive and affective impairments depending on depressive symptom chronicity. Substance use may further exacerbate the cognitive deficits and mood instability experienced by SMM with intermittent depressive symptoms, resulting in a full mediation effect via substance use in predicting future IPV victimization, whereas among SMM who are chronically depressed, substance use and its associated detriments may not produce sufficient additional impairments in cognitive or affective functioning to further increase vulnerability to IPV victimization. These further highlight chronic depression as an important antecedent to IPV experiences and a critical point of intervention to reduce rates of IPV among SMM.

Evidence on substance use serving as a pathway connecting depressive symptoms, especially intermittent depressive symptoms, to IPV victimization points to a potential mechanism for understanding the syndemic of these health outcomes among SMM. SMM who are newly or infrequently experiencing depressive symptoms or recovering from them may turn to substance use as a means for self-medication to modulate their emotional state (Wang et al., 2018). This coping strategy may be particularly salient among SMM, given the high prevalence of co-occurrence between depressive symptoms and substance use disorders among sexual minorities who were assigned male at birth in comparison to heterosexual cisgender men (Felner et al., 2021). Taken together with our findings, intermittent depressive symptoms alone may not be sufficient to predict IPV victimization; however, substance use, in synergy with depressive symptoms, can increase vulnerability to abuse in intimate relationships between men. Impairments to cognitive function and emotional regulation influenced by depression as well as use of substance likely debilitate one’s ability to detect and respond to instances of violence (Iverson et al., 2011; Kirwan et al., 2024), which can be exploited by IPV perpetrators.

Strength and Limitations

The present study demonstrates several strengths. Use of data with three timepoints over 12 months allowed for assessment of temporal associations between depressive symptoms, substance use, and IPV experiences, improving on prior cross-sectional studies (Callan et al., 2021; Hong et al., 2025). This has also allowed for us to test for mediation via substance use, which provided important insights on potential mechanisms linking depression and IPV within the syndemic theory. By distinguishing between intermittent and chronic depressive symptoms, we provided more nuanced insights into how mental health trajectories influence IPV risk. Finally, we assessed for both IPV victimization and perpetration, providing empirical evidence on the mechanisms underlying both sides of IPV in same-sex male relationships where bidirectional IPV is prevalent.

In addition to these strengths, there are limitations to the present findings. The study used a non-random, convenience sample of SMM who were of higher education levels and computer literate, which limits the generalizability of the data. Due to the aims of the parent study focused on HIV prevention, the sample did not include SMM living with HIV, an important subpopulation to the understanding of the syndemic between mental health, substance use, and violence among SMM (Singer et al., 2017). Moreover, the number of participants reporting IPV experiences may have been too low to establish adequate power for statistical inferences, due to the overall sample size. The study also relied on self-report to measure depressive symptoms, rather than using independent or clinician diagnostic rating. It is recommended that future studies garner larger samples of SMM reporting IPV that are inclusive of people living with HIV and have greater number of participants with IPV experiences to provide more robust evidence on the syndemic factors to IPV experiences among SMM. Notwithstanding these limitations, our findings offer several implications for research and interventions addressing occurrences of violence within intimate relationships between men.

Research Implications

The present study contributes to expanding the knowledge on the dynamic interactions among depression, substance use, and IPV over time among SMM, a relatively understudied population in IPV research (Callan et al., 2021; Laskey et al., 2019). The evidence for depressive symptoms and substance use as precedents to IPV among SMM complements past research that focused on IPV as risk factor to poor mental health (Hong et al., 2025) and substance use (Metheny et al., 2025; Metheny et al., 2024) in highlighting the synergistic relationships between these psychosocial contributors to overall health of SMM (Singer et al., 2017). Our findings support the need for additional longitudinal studies that explore the pathways in which the syndemic factors to health, including IPV, are influencing one another in SMM. Specifically, future research should explore symptomology of depression as it interacts with substance use, power dynamics, and violence occurrence in intimate relationships between men.

Clinical and Policy Implications

Our findings suggest a need for the development of IPV prevention programs tailored to specific needs of SMM that incorporate strategies to address mental health and related coping mechanisms, including substance use. Mental health providers and other service providers working with SMM population should be aware of the high burden of mental health, substance use, and IPV experiences and their synergistic relationships (Callan et al., 2021; Hong et al., 2025; Singer et al., 2017). Building onto this awareness, clinical and community settings may incorporate screening for depression and substance use to help identify SMM at higher risk of IPV. Development and adaptation of integrated care and treatment models that combine mental health, substance use treatment, and IPV prevention services for SMM are also necessary (Iverson et al., 2011; Scheer & Poteat, 2021).

For example, community-based organizations serving persons affected by IPV could integrate trauma-informed and syndemic-aware approaches to deliver culturally appropriate services for their SMM care seekers, which would be important to lowering the barriers to IPV-related service for SMM help seekers (Huntley et al., 2019). These approaches should be guided by evidence-based interventions that have been developed to target IPV, depression, and substance use, such as cognitive-behavioral therapy (CBT) designed to address depressive symptoms among survivors of interpersonal violence (Iverson et al., 2011), and interventions for substance abuse (harm reduction model) (Carrico et al., 2014), and excessive alcohol use (CBT and Motivational Interviewing) (Dimova et al., 2022) among SMM. These interventions should be adapted to be culturally responsive to the distinct needs and social contexts of SMM survivors and perpetrators of IPV.

These efforts should be complemented by policies addressing minority stress, stigma, and discrimination, as these structural factors exacerbate depression, substance use, and IPV risk. Our findings suggest that less chronic levels of depressive symptomology are also associated with increased vulnerabilities for IPV experiences. This points to a need for public health interventions that focus on early intervention on depressive symptoms among SMM as a pathway to preventing IPV, especially since chronic depressive symptoms have shown to be associated with both victimization and perpetration of IPV. There is also a need for greater investment towards raising awareness and development of syndemic-focused interventions to address multiple, co-occurring health challenges faced among SMM.

Conclusion

The current study expands on existing literature through examination of directionality of effect within the syndemic of IPV, depression, and substance use among SMM. Our findings suggest that depressive symptoms temporally preceded IPV victimization. Building onto this, we report that depressive symptoms had both total and direct effect on future instances of both IPV victimization and perpetration and that substance use mediated the effect on IPV victimization but not perpetration. We further found that mediation via substance use in predicting future IPV victimization differed by depressive symptom chronicity. These identified pathways to IPV among SMM highlight a need for multilevel trauma-informed interventions that address depressive symptoms and substance use as interrelated mechanisms underlying IPV experiences among SMM.

Public Significance Statement.

The study finds that depressive symptoms among sexual minority men are associated with increased vulnerability to victimization and a higher likelihood of perpetrating violence in intimate relationships, particularly among individuals with chronic depressive symptoms. Substance use mediates the relationship between depressive symptoms and experiences of intimate partner violence victimization. These findings highlight a need for multilevel trauma-informed interventions that address depressive symptoms and substance use as interrelated mechanisms underlying intimate partner violence experiences in this population.