INTRODUCTION

Substantial research has found that drug-involved men are at disproportionately high risk of engaging in transmission risk behaviors for HIV and other sexually transmitted infections (STIs)1–4 as well as of perpetrating intimate partner violence (IPV) against women.5–14 Over the past decade, mounting evidence also has documented multifaceted relationships between experiencing IPV and HIV/STI transmission risks among drug-involved women.14–16 However, to our knowledge, only a handful of recent studies have examined the relationships between perpetration of IPV and HIV/STI transmission risks among heterosexual men and only one study has focused on drug-involved men.

This nascent research has suggested multiple relationships between male perpetration of IPV and HIV transmission risks. A cross-sectional study among 104 male inmates that examined their responses to condom requests by their female partners found that the severity of male perpetration of IPV was associated with more coercive responses to condom request messages that implied women’s infidelity.17 A cross-sectional study of a random sample of 726 sexually active individuals found that perpetration of IPV was associated with having unprotected intercourse.18 Another recent cross-sectional study of a nonrandom sample of 283 sexually active, young adult men recruited from an urban community health center found that participants who reported perpetration of IPV during the past year were significantly more likely to report (1) inconsistent or no condom use, (2) forced vaginal sex without a condom, and (3) sex with multiple female partners.19 A cross-sectional study with a nonrandom sample of 273 men in methadone maintenance treatment programs (MMTPs) found that men who reported perpetrating IPV were almost 4 times more likely to have had more than one intimate partner and 2.6 times more likely to have had sex with a drug injecting sexual partner than their counterparts.20 This emerging research on the relationship between perpetration of IPV and HIV/STI transmission risks among drug-involved men in heterosexual relationships remains limited in terms of cross-sectional designs, nonrandom samples, or failure to adequately control for potentially confounding variables. Longitudinal research with improved methods of control for potential confounders is needed to provide stronger evidence for direct associations between perpetration of IPV and different HIV/STI transmission risks among drug-involved men.

Research on the relationship between experiencing IPV and HIV/STI transmission risks among heterosexual women has elucidated several pathways linking IPV and HIV/STI transmission risks that may also exist among male perpetrators, including (1) engaging in unprotected sex,14,21–34 (2) higher rates of STIs,29,35–39 (3) sex with multiple sexual partners,28,29,40 (4) disclosure of an STI or positive HIV status,41,42 (5) trading sex for drugs or money,36,43 (6) having a risky sexual partner (e.g., one who injects drugs is HIV-positive and/or has had sex with multiple partners),28,29,33,36,38,43–45 (7) forced sex,43,46,47 and (8) injecting drug use.20

This study builds on previous research linking perpetration of IPV and HIV/STI transmission risks by examining the temporal relationships between perpetration of IPV and different HIV/STI transmission risks among a random sample of men in MMTP using a longitudinal design and propensity score matching. This study examines two hypotheses. The first hypothesis (H1) is: self-reported sexual HIV/STI transmission risks at baseline (i.e., wave 1) will increase the likelihood of perpetrating physical and/or injurious IPV at 6- and 12-month follow-up assessments (i.e., waves 2 and 3) over the subsequent year. The second hypothesis (H2) is: perpetration of physical and/or injurious IPV at wave 1 will increase the likelihood of sexual HIV/STI transmission risks at subsequent waves 2 and 3 (i.e., 6- and 12-month follow-up).

MATERIALS AND METHODS

Random Sampling and Recruitment Procedures

We randomly selected 1,300 men to be screened for eligibility for this study from a total population of 2,067 men who were enrolled as patients in seven MMTP clinics in Harlem, New York City. Eligible participants included men, aged 18 or over, who were enrolled at an MMTP for at least 3 months and who reported having had a sexual relationship with a woman during the past year described as a girlfriend, spouse, regular sexual partner, or the mother of his children. MMTP counselors notified potential participants of their selection for the study and invited them to contact research assistants (RAs). Once a potential participant made contact with an RA, the RA would describe the study and give him an opportunity to ask any questions. If the potential participant expressed interest in the study, the RA would complete informed consent and conduct a brief screening interview to determine eligibility.

Of the 1,300 randomly selected men, 25 men were excluded from screening because they did not demonstrate sufficient English-speaking proficiency to provide informed consent and three men were excluded from screening because of a severe cognitive or psychiatric impairment that interfered with their ability to provide informed consent. Of the remaining 1,272 randomly selected men, 774 men agreed to participate and completed a 15-min screening interview, 194 refused to participate in the study, and 304 missed two or more screening appointments and did not participate. Of the 774 men who were screened, 499 met eligibility criteria of whom 356 (71%) agreed to participate and completed a baseline assessment.

Data for the longitudinal study were collected between 1999 and 2003. Eligible participants were assessed with repeated measures at baseline (wave 1), at 6-month follow-up (wave 2), and at 12-month follow-up (wave 3). For all three assessments, male RAs administered face-to-face structured interviews, which averaged 1.5 h in length. Compensation entailed $5 for participating in the screening, $30 for participating in the face-to-face baseline interview, $35 for the 6-month interview, and $40 for the 12-month interview. The RAs received 24 h of training in recruitment and interviewing skills. The Institutional Review Boards of the participating MMTPs and Columbia University approved the protocol for this study.

Measurement

The repeated assessments covered self-reported sociodemographics, relationship characteristics, drug use and drug risk behaviors, sexual HIV risk behaviors, self-reported STIs, and participant reports on partner’s drug-related and sexual HIV risk behaviors. Information on a maximum of two current female intimate partners was elicited from participants.

Sociodemographic and relationship characteristics included: age, race/ethnicity, education, employment status, length of stay in most recent methadone maintenance treatment, number of intimate partners in past 6 months, length of relationship with intimate partner, type of relationship, and contribution of participant and partner’s to household expenses.

The Drug Use and Risk Behavior Questionnaire was developed by the investigators to provide frequency counts of binge drinking, injecting drug use, and using crack/cocaine, heroin, marijuana, and other illicit drugs in the past 6 months. Internal consistency was assessed with 800 subjects and yielded α reliability of 0.80.48,49 For the purpose of this study, we dichotomized use of each drug so that any use was coded as 1 and nonuse was coded as 0. Binge drinking was defined as drinking five or more alcoholic drinks within a 6-h period, which is considered a standard definition of a binge-drinking episode for men.50

Self-reported data on sexual HIV/STI risks were measured by the Sexual Risk Behavior Questionnaire (SRBQ). The internal consistency of the SRBQ has yielded an α reliability of 0.80.44 The SRBQ ascertains sexual behaviors within past 6 months, including (1) frequency of condom use (i.e., always, sometimes, or never) during vaginal and anal sex with intimate, casual, and paying partner(s); (2) number of sexual partners; (3) risk factors associated with partners reported by participants (i.e., partners who inject drugs, are HIV-infected, have had STIs, or who have had sex with other concurrent partners in the past 6 months); (4) self-reported STIs in past 6 months; and (5) self-reported HIV status.44 A positive response to an HIV/STI transmission risk at wave 2 (6-month follow-up) and/or wave 3 (12-month follow-up) was coded as 1 for that outcome; a negative response to the HIV risk outcome for both waves 2 and 3 was coded as 0. With respect to condom use and condom request variables, participants indicating at waves 2 and 3 that they always used or requested condoms were coded as “always = 2” whereas participants reporting they sometimes used or requested condoms were coded as “sometimes = 1.” Participants who reported never using or requesting condoms at wave 2 and wave 3 were coded as “never = 0.”

Perpetration of physical and injury-related IPV was assessed using the revised Conflict Tactics scales (CTS2).49 The CTS2 contains two subscales measuring physical and injury-related IPV in the past 6 months, which provide an overall prevalence of IPV that we define as “physical and/or injury-related IPV.” We examined IPV across intimate partners at each wave. The CTS2 also has a subscale that assesses sexual coercion, which was considered as an HIV/STI risk variable because it contains items on coerced sex without condoms. Internal consistency of the CTS2 subscales ranges between 0.79 and 0.95.49 At all waves respondents who reported perpetrating any physical and/or injury-related IPV in the prior 6 months were coded as 1 for IPV and those who did not report perpetrating any physical and/or injury-related IPV in prior 6 months were coded as 0.

Data Analysis

Of the 356 participants, who completed the baseline interview (wave 1), 287 (81%) completed the 6-month follow-up interview (wave 2) and 278 (78%), the 12-month interview (wave 3). Similarities on the baseline measures were found among those who were retained at follow-ups and dropouts, except on age, ethnicity, and injection drug use. Multiple imputation via the Multivariate Imputation by Chained Equations module in Stata8 was used to reduce the potential for bias resulting from missing data and differential attrition.51–54

Propensity scores were calculated using attributes for observed confounders and treatment variables observed at wave 1 (baseline). Propensity score matching is a technique used in observational studies to select groups, which are similar on average with respect to potential confounders.55–59 The confounders included (1) sociodemographic variables (i.e., age, race/ethnicity, education, length of time in MMTP, and employment status), (2) relationship characteristics (i.e., type of intimate relationship(s), length of relationship, and financial dependency), and (3) substance use (i.e., participant’s use of illicit drugs and binge drinking in past 6 months, participant’s reports on intimate partner’s substance use in past 6 months). The selection of these potential confounders was based on previous research on factors associated with both IPV and sexual transmission risks among drug-involved, heterosexual men and women.14,20 The matching procedure was performed by “PSMATCH2” (a shareware module in Stata8).60 The diagnostics of balance on all covariates were also conducted for adequacy of matched groups.61

After selecting a final sample of participants using propensity score matching, multiple logistic regression analyses were conducted to test each hypothesis. Causal effect sizes were estimated by odds ratios (ORs) and their associated 95% confidence intervals (CI), adjusting for the same set of confounders used in the propensity score matching. For all evaluations testing H1, we compared participants who reported perpetrating physical or injurious-related IPV in the prior 6 months at wave 1 (baseline) with men who did not perpetrate such violence on HIV risk outcomes reported in wave 2 (6-month follow-up) and/or wave 3 (12-month follow-up). For all evaluations testing H2, the treatment variables are HIV risk factors measured at wave 1 and the outcome variable is perpetration of physical and injury-related IPV at wave 2 and/or wave 3.

RESULTS

Sociodemographic and Relationship Characteristics

Sociodemographic and relationship characteristics collected at wave 1 are presented in Table 1. The majority of participants self-identified as Latino or African American. The men’s mean age was almost 44 years old (SD = 8.5) and their average level of education was 11.6 years (SD = 2.3). Almost half of the sample (47%) was unemployed and about one tenth had been incarcerated in the past 6 months. Participants had been in methadone maintenance treatment for an average of 8.3 years (SD = 7.4). In terms of relationship characteristics, the average length of relationship with the main intimate partner was 10.5 years (SD = 9.2). More than half indicated that they were married or had a common-law marriage with their main intimate partners. Likewise, more than half were living with their partner. The mean age of their main partners was almost 40 years old (SD = 8.9). The majority of partners were Latina or African American. About half of the partners were unemployed.

TABLE 1 Sociodemographic characteristics and relationship characteristics (N = 356)
Full size table

Substance Abuse-related Characteristics

In the 6 months before wave 1 (baseline), about half of the participants (50%) reported heroin use, 37% reported crack/cocaine use, 38% marijuana use, 72% reported any illicit drug use, and 26% reported binge drinking. At wave 1 (baseline), 42% of the participants reported that their intimate partners used an illicit drug in the prior 6 months, 22% reported partner use of crack/cocaine, 18% had a partner who used heroin, and 29% reported their partners engaged in binge drinking.

Prevalence of IPV

Prevalence rates of perpetrating physical and/or injurious IPV during the prior 6 months as reported at each wave of the study are presented in Table 2. The prevalence of perpetrating physical and/or injurious IPV was 28% for wave 1 (baseline), 34% for wave 2 (6-month follow-up), and 31% for wave 3 (12-month follow-up). About half (47%) of the sample reported perpetrating physical and/or injurious IPV during the follow-up period covering waves 2 and 3.

TABLE 2 Prevalence of IPV, condom use behaviors, and HIV/STI risks at three time points
Full size table

HIV/STI Risk

At wave 1 (baseline), more than a third of the participants reported consistent condom use during vaginal sex with their intimate partners in the past 6 months. Five percent self-reported having had an STI in the past 6 months and 15% reported testing positive for HIV. About one quarter indicated that they had more than one intimate partner and 8% reported buying sex for money or drugs. Less than one quarter of the participants injected drugs in the past 6 months. In addition to these risks, 7% of the men reported that their intimate partners were HIV-positive, 6% reported that their partners had an outside partner, and 8% indicated that their partners injected drugs in the past 6 months. Prevalence rates of HIV risks were consistent over time.

Hypotheses Testing

H1: Perpetration of IPV Increases the Likelihood of Subsequent HIV/STI Risk

The findings presented in Table 3 are the adjusted ORs for reporting HIV/STI risks at wave 2 (6-month follow-up) and/or wave 3 (12-month follow-up) associated with perpetrating physical and/or injurious IPV in the prior 6 months at wave 1 (baseline). Compared to men who do not report physical and/or injurious IPV, men who report perpetrating physical and/or injurious IPV in the prior 6 months at wave 1 (baseline) were significantly more likely to indicate that they have had more than one intimate partner (OR = 2.9, CI = 1.2, 7.1) and more likely to buy sex for money or drugs (OR = 3.3, CI = 1.1, 10.4) at subsequent waves (6 and 12-month follow-up). Perpetrators of IPV at wave 1 (baseline) were also significantly more likely than nonperpetrators to report sexual coercion at subsequent waves (OR = 2.9, CI = 1.2, 6.6). In addition, perpetrators of IPV at wave 1 were marginally more likely to report not using condoms consistently with other nonmain partners at subsequent waves (OR = 0.4, CI = 0.1, 1.1). No significant associations were found between perpetration of IPV at wave 1 (baseline) and subsequent self-reported STIs, injection drug use, frequency of condom use, requests for condom use, and subsequent partner-related risk factors (i.e., injecting drugs, having more than one partner).

TABLE 3 Estimates of the temporal relationship between perpetration of IPV at baseline and subsequent HIV risk indicators at follow-up (H1)
Full size table

H2: HIV/STI Risks Increase the Likelihood of Perpetration of IPV

Table 4 contains adjusted ORs for perpetrating IPV at subsequent waves associated with reporting HIV/STI risks in the prior 6 months at wave 1 (baseline). Compared to men who reported never using condoms in the past 6 months at wave 1, men who sometimes used condoms in the past 6 months were significantly more likely to report perpetrating physical and/or injurious IPV at subsequent waves (6- and 12-month follow-up) (OR = 0.3, CI = 0.1, 0.8). Inconsistent condom use at wave 1 (baseline) was marginally associated with subsequent perpetration of IPV (OR = 0.5, CI = 0.3, 1.1). Men who reported injecting drugs in the past 6 months at baseline were significantly more likely than noninjectors to indicate perpetrating IPV at subsequent waves (OR = 3.7, CI = 1.1, 11.9). Men who indicated any sexual coercion against their partners at baseline were also significantly more likely than their counterparts to report perpetrating physical or injurious IPV at subsequent waves (OR = 2.6, CI = 1.2, 5.4). In addition, having a self-reported STI in the past 6 months at wave 1 (baseline) was marginally associated with perpetration of IPV at subsequent waves at the 90% confidence level (OR = 5.6, CI = 0.98, 31.6). No support was found for subsequent perpetration of IPV as a consequence of having a risky partner, having had sex with multiple or commercial partners, or partner-related risk factors assessed at wave 1 (baseline).

TABLE 4 Estimates of the temporal relationship between HIV risk indicators at baseline and subsequent perpetration of IPV at follow up (H2) after propensity score matching: adjusted ORs and 95% CI
Full size table

DISCUSSION

To our knowledge, this is the first longitudinal investigation that examines temporal relationships between perpetrating IPV and HIV/STI transmission risks among a random sample of drug-involved men in heterosexual, intimate relationships. The methodology of propensity score matching in combination with multiple logistic regression in this longitudinal study allows for a more rigorous method of accounting for potentially confounding variables and inferring causality. These methodological advances improve on previous cross-sectional research, which has examined relationships between male perpetration of IPV and HIV transmission risks. Collectively, the study findings suggest multiple, temporal relationships between perpetrating IPV and several HIV/STI risk factors, which mirror relationships between these cooccurring problems found in studies among drug-involved women.

The first hypothesis that perpetration of IPV increases the likelihood of presenting subsequent HIV/STI transmission risks was supported for some HIV risks. The temporal relationship between IPV and engaging in outside relationships with other intimate or sex exchanging partners is consistent with previous cross-sectional studies, which have found a significant relationship between these two variables among heterosexual men as perpetrators19,20,62 and among heterosexual women as victims of IPV.28–30,40,45,63 These findings may suggest that the relationship instability associated with IPV may increase the likelihood that men will engage in outside relationships as an exit strategy or as another form of retaliation, which may also be considered as a form of psychological IPV. Further research is needed to examine whether perpetration of IPV among drug-involved men predicts subsequent changes in partner status (e.g., shifting primary intimate partners, engaging in concurrent relationships with different types of sexual partners, and initiating sex with casual or commercial partners) that may, in turn, result in changes in other HIV risk-related behaviors, like inconsistent condom use.

Moreover, the study findings supported temporal relationships between perpetration of physical or injurious IPV and sexual coercion, suggesting that the pathways linking perpetration of physical or injurious IPV and sexual coercion are bidirectional. Although this finding is not surprising as sexual coercion is considered a domain of IPV, it does have HIV/STI risk implications. Sexual coercion has been identified as a risk factor for HIV/STIs as it has been found to be associated with unprotected sex and rough sex resulting in vaginal lacerations that may increase the likelihood of HIV/STI transmission.64–66

In addition, the data supported the second hypothesis: an increased likelihood of subsequent perpetration of IPV was significantly associated with wave 1 reports of recent injection drug use and noncondom use. The temporal relationship found between injection drug use and perpetration of IPV is consistent with previous literature.20 Several qualitative studies suggest that the practice of injecting drugs and sharing needles between intimate partners is often pervaded with gender-based inequalities and IPV.67,68 The relationship between perpetration of IPV and inconsistent or no condom use has been supported by previous research.19,20 Failure to use condoms in intimate relationships, especially when other risk factors are present, such as having an STI or outside partners, may increase the likelihood of relationship conflict and IPV. The marginal association found between self-reported STIs and subsequent perpetration of IPV may be related to the impact of the disclosure of an STI, or alternatively, contracting STIs may be related to sexual affairs outside the intimate relationship, which may trigger IPV. Here again, future research to identify the precise sequence of multiple HIV/STI risks that lead to the perpetration of IPV may further elucidate the causal mechanisms linking these two cooccurring problems.

Limitations

The nonresponse rate limits the generalizability of findings as we do not have data on whether randomly selected men who did not participate in the screening interview may have differed from respondents in terms of the IPV, HIV risk, and background variables. This study also relies on self-reported data of sensitive behaviors that are subject to a social desirability bias, which may have influenced study findings. For example, social desirability bias may have inhibited participants from disclosing recent perpetration of IPV and other socially undesirable HIV risk behaviors.

Implications for Intervention

Despite these limitations this study has several implications for designing effective HIV prevention interventions for this population. The HIV/STI transmission risks linked with perpetration of IPV among this sample of men suggest the need for HIV prevention strategies to take into account broader relationship safety issues. To date, no HIV prevention interventions have been developed and tested for drug-involved men, who are at risk of perpetrating IPV. Over the past two decades MMTPs have played an instrumental role in reducing HIV among patients.69 MMTPs may serve as an optimal setting to launch an HIV prevention intervention, which addresses broader relationship safety issues associated with IPV and HIV/STI transmission risks among men and their intimate partners.