Introduction

HIV continues to disproportionately affect men who have sex with men (MSM) compared to any other group globally [1] and in the US [2]. MSM populations represent about 2 % of the US population but account for approximately six in ten (63 %) of the estimated new infections in the US and close to eight in ten (78 %) of all men who have been newly diagnosed with HIV [2]. In 2011, the majority of persons living with HIV were MSM who were also injection-drug users [2]. With the high rates of HIV among MSM populations, disclosure of HIV status to partners becomes an integral factor in the prevention and treatment of HIV [3].

Disclosure of HIV status to partners may have numerous benefits. Increasing awareness of risk may help to motivate partners to seek HIV testing, reduce risky sexual behaviors, and decrease HIV transmission [4]. Benefits to the individual who discloses may include social, physical, psychological dividends [5] such as social support, better access to HIV treatment and care, and increased opportunities to discuss the reduction of HIV risk with partners [4]. Nevertheless, the decision to disclose HIV status to partners may have implications in all arenas of life [6]. The consequences theory of HIV disclosure states that people with HIV are likely to disclose their status to partners if the benefits or rewards for disclosing their status outweigh the costs or risks [5]. Costs may include valuable things that are given up for a reward of greater or equal value, or consequences that would be otherwise avoided [5], such as blame, abandonment, discrimination, disruption of relationships, and physical and/or emotional abuse [4].

Partner violence continues to be a major public health problem in the US [7]. Violence between partners is not only a public health issue among heterosexual couples but also among same-sex couples [7, 8], and is sometimes found to be more prevalent among same-sex couples compared to heterosexual couples [9–11]. Among younger populations with same-sex partners, partner violence is also commonplace with one in four adolescents from a population-based sample reporting partner violence victimization, and approximately one in ten reporting physical partner violence [12]. Among a sample of MSM in the Chicago area, approximately 32 % reported partner violence, 21 % reported a history of verbal abuse, 19 % reported physical abuse and 19 % reported sexual abuse. The majority of men reporting any partner violence reported experiencing more than one type of partner violence [13], which highlights the interconnectivity of different forms of partner violence. Among a probability-based sample of MSM living in four US cities (Chicago, Los Angeles, New York, and San Francisco), 34 % reported psychological partner violence, 22 % reported physical partner violence and 5 % reported sexual partner violence [9]. Risky sexual behavior, such as unprotected anal sex [13], younger age [9], lower education level [9], being HIV-positive [9], experiencing depression [13] and substance abuse [13, 14] have been shown to be risk factors for partner violence among MSM populations.

The association between depressive symptoms, substance use and partner violence is complex. Several studies have shown a link between depression [15, 16], substance use and abuse [17–20], and other mental health disorders [21, 22] and partner violence among men and women. However, specifically among MSM couples, methamphetamine use, rock or crack cocaine use, and heavy drinking have been associated with partner violence victimization and aggression [14]. Alcohol intoxication, substance use problems including substance use with sex, hard drug use, and mental health diagnoses (including depression, bipolar disorder, and any psychiatric/emotional disorders) have also been associated with partner violence among MSM populations [13].

Theoretically, the psychoactive effects from the use of substances, such as impaired judgment, and having depressive symptoms may be related to partner violence victimization associated with HIV disclosure in a number of ways. Impairment from substance use may result in poorly timed or inappropriately delivered information on HIV serostatus to a partner based on poor judgement. For example, being under the influence of substances is likely to prohibit the accurate reading of a partner’s possible negative reaction to such sensitive information. It is equally likely that cognitive impairment may lead to the inability to leave a violent situation. Depressive symptoms may be associated with disclosure-associated partner violence due to the indifference a person may feel towards leaving a violent situation. For example, a history of depression may be associated with placing elevated focus on a maintenance of a relationship with a partner [23], albeit casual. As a result, an individual may be more willing to tolerate violence [23] from his partner.

As previously stated, one potential cost of disclosing HIV status to a partner is violence. Theoretically, among individuals with casual sex partners, the violence that may occur after disclosing their serostatus could be due to a reaction out of fear from their sex partners that they were put at risk of contracting HIV and even more so fearful that transmission of HIV may have occurred. Nevertheless, research focusing on HIV-disclosure associated partner violence tend to concentrate on women. These studies have shown that women experience abuse (sexual, physical and/or emotional) directly related to HIV disclosure or that occurred only after their HIV serostatus was disclosed [3, 24–27]. For example, Shamu and colleagues [3] explored disclosure-associated partner violence among pregnant women in Harare, Zimbabwe and showed that gender inequity, history of partner violence, risky sexual behaviors were associated with partner violence after HIV disclosure. Positive HIV status was linked to partner violence and negative reactions by male partners immediately after disclosing HIV status [3]. Among this population, the study showed the interconnectivity of partner violence, HIV status and HIV disclosure. Populations at risk for partner violence tend to be the same groups at risk for HIV infection [26]. However, studies examining risk factors for violence associated with disclosure of HIV status among sexual minority populations are lacking. Research exploring the interconnectivity between depressive symptoms, substance use and partner violence after HIV disclosure specifically among MSM populations is extremely limited. To our knowledge, no study has examined potential risk factors for partner violence associated with HIV disclosure among sexual minority populations. The objective of this study was to explore the association between depressive symptoms, substance use, and verbal and/or physical abuse by a partner associated with disclosure of HIV status among MSM. Determination of risk factors for violence associated with disclosure of HIV status to partners is crucial. As a result, public health practitioners, and health care providers of MSM populations living with HIV may be aware of these factors, and incorporate the necessary resources to address the interconnectivity of depressive symptoms, substance use and abuse associated with disclosure of status to partners, especially among MSM populations. We hypothesize that men who abuse substances and/or have depressive symptoms may be more likely to experience violence after disclosing their HIV status to their partners.

Methods

Data were obtained from 340 participants at the baseline assessment of a longitudinal randomized-controlled trial of an intervention designed to assist HIV-positive MSM in disclosing their serostatus to casual sex partners. This disclosure intervention study was conducted across two metropolitan statistical areas (MSA; Columbus, Ohio and Tampa, Florida) from December, 2009 through December, 2014 and consisted of participants who were HIV-positive, at least 18 years old, were sexually active, had two or more partners in the last 12 months (at least one of which was a man), and indicated an interest in learning more about disclosing their serostatus to sexual partners. In addition, men were English speaking and planned on living in the MSA for at least 1 year. Separate IRB approval was garnered for both recruitment locations and participants were treated in accordance with the “Ethical Principals of Psychologists and Code of Conduct” [28].

Study Participants and Recruitment

The average age of participants (n = 340) was 42.1 years (SD = 11.0 years). The average time since diagnosis was 13.6 years (SD = 8.3 years). Self-reported sexual orientation was mostly gay (n = 267, 78.5 %), and 74.4 % (n = 253) indicated that they had sex only with men. Most participants reported being single (n = 238, 70 %). Across study sites, there were statistically significant differences by age where the mean age of men from Tampa was 44.5 years (SD = 10.3 years) and from Columbus, 39.5 years (SD = 11.2 years). There were also statistically significant differences in substance use scores where the mean score among men from Tampa was 13.0 (SD = 5.7) and from Columbus, 14.9 (SD = 5.7). There were no statistically significant differences by income, race/ethnicity, educational level, employment status, mean Center for Epidemiologic Studies-Depression (CES-D) scores and partner violence status.

Participants were recruited in several ways. First, individuals were recruited through advertising efforts with local/state AIDS service organizations (ASOs). Caseworkers were informed about the study and handouts were made available to distribute to potential participants or through newsletters and direct mailings. ASOs also featured calls for participation on their websites, and sent materials to clients for recruitment efforts. Recruitment materials were also placed at HIV-related venues and forums held throughout the MSAs, and at local eating and drinking establishments. Finally, advertisements were placed in local daily newspapers.

Data Collection

Prior to randomization, participants completed a baseline questionnaire administered using audio-computer assisted self-interviewing (ACASI). Social desirability has been identified as a potential problem in the collection of reliable data on serostatus disclosure [29, 30]. However, ACASI has been demonstrated to be associated with more complete reporting of potentially stigmatized drug, sex and HIV risk behaviors [31, 32]. Additionally, ACASI has been both accepted and preferred as a method of data collection by participants in HIV risk-related studies [32]. Data were collected on person-level characteristics (i.e., items and scales asked only once for each participant) and encounter-level characteristics (i.e., repeated measures for each participant). Person-level items included measures of demographic characteristics, global sexual/disclosure activities (i.e., count measures of specific activities during the prior 30 days), and measures relevant to sexual activity/disclosure.

Measures

Verbal and/or Physical Abuse Associated with Disclosure

Verbal and/or physical abuse associated with disclosure was measured using a single item: “In the last 30 days, how many times were you verbally or physically abused by a partner after disclosing?” Approximately 99.4 % of participants (338/340) answered this item, with responses ranging from 0 to 10, with an average of 0.19 times.

Depressive Symptoms

Depressive symptoms were measured using the CES-D Mood [33] scale, which is a 20 item scale used to measure depressive symptoms in the general population. Items were scored using a 4-point Likert-type scale with values ranging from Rarely or None of the time (0) to Most or All of the time (3). Scores were summed to arrive at an overall score with values ranging from 0 to 60, with higher scores indicative of more depressive symptoms. The CES-D was chosen because of its good reported internal consistency (α = 0.85) and validity [33]. Depressive symptoms were also examined as a binary variable to determine participants who were at risk for clinical depression (CES-D ≥ 16) [34]. For the current study, the standardized Cronbach’s α for the CES-D was 0.94.

Substance Use

To assess substance use participants completed the substance use items of the Substance Abuse and Mental Illness Symptoms Screener (SAMISS) [35]. These seven items inquired about participants’ frequency, amount, and perceived problematic use of alcohol, illicit drugs, and prescription drugs and were summed to derive an overall substance abuse score. For most of the items, a Likert-type response scale ranging from Never (1) to 4 or more times a week (5) was used. Specific items were then summed to determine whether the participant was considered positive for substance use problems if their score exceeded a specified threshold [35]. For the current study, the standardized Cronbach’s α for the SAMISS was 0.77.

Potential Confounders

Confounders considered were shown to be associated with depression and substance use, and partner violence based on literature review a priori. Statistically significant differences in sociodemographic characteristics such as age [36–38] and income [37] are associated with depression. Substance use has also been shown to differ by age [39–44] and income [41, 43]. Previous studies have shown differences in age [45, 46] and income [45, 47] to be linked to partner violence. Therefore, the current study considered age (continuous) and monthly income ($0–$500, $501–$1000, ≥$1001) as potential confounders.

Analytic Approach

Participants were excluded if they were missing on all depressive symptoms, substance use, and partner violence questions (n = 2). The resultant sample size was 338.

Descriptive statistics were used to determine the distribution of sociodemographic characteristics among men reporting and not reporting partner violence after HIV disclosure. Two sets of multivariable logistic regression models were used to obtain adjusted odds ratios and 95 % confidence intervals depicting the association between depressive symptoms, substance use, and partner violence after disclosure of HIV status. Each confounder was placed in different models with depressive symptoms, and substance use as separate exposure variables, and partner violence victimization after disclosure as the outcome. However, no confounder changed the effect estimate between depressive symptoms, substance use and partner violence victimization >10 %. Nevertheless, adjusted models were still examined as recent research has recommended that potential confounding variables should not be discarded based on a lack of change in estimates but should still be included in the model based on theoretical evidence [48]. As a result, three nested models were used for each exposure measure: (1) an unadjusted model, (2) a model adjusted for age, and (3) a model adjusted for age and income. Exact logistic regression was used to assess the association between depressive symptoms and substance use as binary variables (where more information would be lost compared to analyzing them as continuous scores), and verbal and/or physical abuse associated with HIV disclosure. All analyses were performed with SAS software Version 9.4 (SAS Institute, Cary, NC).

Results

The distribution of sociodemographic characteristics, verbal and/or physical abuse by an intimate partner, depressive symptoms, and substance use among the study population overall, and by partner violence status are presented in Table 1. In the past 30 days, 6.2 % of the population reported partner violence victimization after disclosing their HIV serostatus to a partner. Approximately six in ten respondents were at risk for clinical depression (58.6 %) and seven in ten reported substance use (72.2 %). The mean score for depressive symptoms was 20.4 (SD = 13.3) and the mean substance use score was 13.9 (SD = 5.8). Seven in ten (71.4 %) respondents reporting verbal and/or physical abuse by a partner and close to six in ten (57.7 %) not reporting disclosure-associated partner violence in the past 30 days were at risk for clinical depression. Approximately 90.5 % of the sample reporting disclosure-associated partner violence and 71.0 % of the sample not reporting disclosure-associated partner violence met the Whetten criteria [34] for substance use. Mean CES-D scores were 27.6 (SD = 13.7) and 20.0 (SD = 13.1) for participants reporting disclosure-associated partner violence and not reporting disclosure-associated partner violence, respectively. Mean substance use scores were 17.5 (SD = 6.7) among victims of disclosure-associated partner violence in the past 30 days and 13.7 (SD = 5.7) among participants not reporting disclosure-associated partner violence.

Table 1 Sociodemographic characteristics, depressive symptoms, and substance use overall and by partner violence associated with disclosure in study sample
Full size table

Logistic regression analyses depicting the association between depressive symptoms and substance use sum scores, and verbal and/or physical abuse by a partner associated with disclosure in the past 30 days are shown in Table 2. Separate crude models show that for every one-unit increase in depressive symptoms and substance use scores, there was a 4 % increase (OR 1.04; 95 % CI 1.01–1.08) and 10 % increase (OR 1.10; 95 % CI 1.03–1.18) in the odds of reporting verbal and/or physical abuse by a partner after disclosing HIV serostatus, respectively. In other words, the more someone was depressed or increased their substance use, the likelihood of experiencing verbal and/or physical abuse by a partner after disclosing their HIV status increased 4 and 10 %, respectively. After adjusting for age and income, the relationship between depressive symptoms and disclosure-associated partner violence was attenuated so that the confidence intervals included unity (OR 1.03; 95 % CI 0.99–1.07). However, the association between substance use and disclosure-associated partner violence remained statistically significant (OR 1.09; 95 % CI 1.01–1.16).

Table 2 Logistic regression analyses depicting the association between depressive symptoms, and substance use sum scores, and partner violence associated with disclosure
Full size table

Table 3 shows the results of exact logistic regression analyses depicting the association between being at risk for clinical depression, and substance use, and verbal and/or physical abuse by a partner after disclosing serostatus in the past 30 days. There were no statistically significant associations.

Table 3 Exact logistic regression analysis depicting the association between being at risk for clinical depression and substance use, and partner violence associated with disclosure
Full size table

Discussion

This study is the first to demonstrate an association between depressive symptoms, substance use, and experiencing verbal and/or physical abuse associated with disclosure of HIV serostatus. In the full models, substance use scores were statistically significantly associated with disclosure-associated partner violence, but no statistically significant relationship was seen between depressive symptoms and disclosure-associated partner violence. Associations between binary representations of being at risk for clinical depression, substance use and disclosure-associated partner violence were not statistically significant.

The results from the current study depicting a relationship between substance use and disclosure-associated partner violence support findings from previous researchers examining the link between substance use and partner violence, and confirms part of our hypothesis. Wu et al. showed that substance use, specifically heavy drinking and methamphetamine use, was linked to partner violence among Black MSM throughout the relationship period and in the past month [14]. However, the same study found that cocaine use was statistically significantly associated with lifetime partner violence but not partner violence in the past month. Tran et al. found that substance use, including ecstasy, cocaine, and methamphetamine use, was also linked to partner violence among Asian/Pacific Islander MSM [49]. The lack of statistically significant associations between substance use as a binary variable and disclosure-associated partner violence could have been due to the small cell sizes as a result of the small number of participants reporting verbal and/or physical abuse associated with disclosure in the past 30 days. Categorizing continuous variables into binary variables may result in loss of information, which in this case, due to small numbers, may also contribute to failure to reject the null hypothesis.

Substance use may be associated with verbal and/or physical abuse by a partner after disclosure for a number of reasons. As previously theorized, substance use may be associated with disclosure-associated partner violence due to psychoactive effects of substance use and/or abuse leading to impaired judgment; and/or the inability to leave a violent situation. Therefore, men with substance use may be more likely to be vulnerable and more susceptible to verbal and/or physical abuse by a partner after disclosing their HIV serostatus compared to men who do not use substances.

Depressive symptoms were not statistically significantly associated with verbal and/or physical abuse by a partner after disclosure, which contradicts our hypothesis. This finding suggests that depressive symptoms might not play a role in disclosure-associated partner violence from casual partners among MSM. The current findings contradicts other studies, which have shown that depressive symptoms are associated with partner violence [13, 15, 16, 50]. Nevertheless, these studies examined depression and partner violence among women [50, 51], and heterosexual couples [16]. This difference in findings could be due to differences in study populations. Houston and McKirnan examined risk factors of intimate partner abuse among MSM and found that substance abuse and depression were associated with sexual and/or physical and/or verbal abuse [13]. However, their study examined sexual, physical and verbal abuse, not associated with disclosure of HIV status.

The findings of this study should be considered with some limitations. First, the total sample size was small with only 21 respondents responding to experiencing verbal and/or physical abuse by a partner after HIV disclosure in the past 30 days. As a result, large confidence intervals were obtained for analyses examining depressive symptoms and substance abuse as binary variables, and results implied that the null hypothesis could not be rejected. Also, due to the small sample size, the most parsimonious model (adjusting for age and income) was used and did not include other covariates in the model. We did not adjust for other sociodemographic characteristics, and for substance use when using depressive symptoms as an outcome, or for depressive symptoms when using substance use as the outcome. Indeed, adjusting for one factor, which is the outcome in another model with the same confounders may lead to circular analysis [52]. Second, verbal and/or physical abuse by a partner associated with HIV disclosure was only reported in the past 30 days. Disclosure-associated partner violence in the past month may not be representative of partner violence after disclosure at other time points. Varying relationships have been found between substance use and partner violence during the relationship period and partner violence in the past-month [14]. Third, it is possible that there may be shifts in the impact of variables or intervening variables over the study period (2009–2014). As a result, relationships among variables may change overtime. Therefore, longitudinal studies collecting data on depressive symptoms and substance use with follow-up data collection on partner violence are warranted to help to mitigate the shifts in variable impact that may change overtime in a cross-sectional study design, and to compare to findings seen in the current study. Fourth, the relationship between depressive symptoms, substance use, and partner violence may have differed across study sites. Indeed, substance use was statistically different across study sites. However, we were unable to determine differences and/or similarities in relationships across study sites due to the small number of participants reporting verbal and/or physical abuse from a partner. Fifth, the question asking study participants about disclosure-associated partner violence did not ask for information on sexual violence, which has been linked to alcohol and drug problems through posttraumatic stress symptom severity [53], and to depression through avoidance coping [54].

Nevertheless, the current study had several strengths. This study was the first to examine the association between depressive symptoms, substance use, and disclosure-associated violence among MSM. Depressive symptoms and substance use were operationalized as binary and continuous variables. These definitions allowed for analysis of binary variables based on previously defined thresholds, and of continuous variables so as not to lose any information in analysis. Next, nested models accounting for multiple potential confounders were used. The use of nested models helps to determine the change in effect estimates depending on what confounders were adjusted for at each stage. A final strength of this study was the use of exact logistic regression models. This technique accounted for the small cell sizes in models using the binary variables for depressive symptoms and substance use for analysis.

Conclusions

HIV disclosure interventions geared towards MSM populations should include resources to address substance use and partner violence victimization. Nevertheless, the findings also suggest that interventions for substance use should also address verbal and/or physical abuse by a partner associated with disclosure of HIV status among MSM populations. Health care providers, especially for populations living with HIV, should also be aware of the association between substance use and disclosure-associated violence so they may provide resources addressing not only substance use, but also violence prevention to persons who want to disclose their status to an intimate partner. Given the prevalence of substance use and depressive symptoms experienced by MSM (as demonstrated in this sample), the inclusion of these variables in future studies with this population is warranted. In light of the current study findings, future researchers should address the link between other mental health measures such as anxiety, PTSD, and disclosure-associated violence victimization among MSM populations. Future researchers may also wish to assess partner violence experienced over time, specifically include sexual violence as part of their measures, and use larger sample sizes when possible.