HIV and intimate partner violence among methadone-maintained women in New York City

ARTICLE IN PRESS

Social Science & Medicine 61 (2005) 171–183 www.elsevier.com/locate/socscimed

HIV and intimate partner violence among methadonemaintained women in New York City Nabila El-Bassel, Louisa Gilbert, Elwin Wu, Hyun Go, Jennifer Hill Social Intervention Group, Columbia University School of Social Work, 1255 Amsterdam Avenue, New York, NY 10027,USA Available online 28 January 2005

Abstract Intimate partner violence (IPV) has been recognized as a risk factor for HIV and sexually transmitted infections (STIs) among women, particularly among those who are drug involved. This study examines the temporal relationships between sexual and/or physical partner violence (IPV) and sexual risk of HIV/STI transmission in a longitudinal study with a random sample of 416 women enrolled in methadone maintenance treatment programs in New York City. Two hypotheses are tested: whether sexual risk-related factors or risk reduction behavior leads to subsequent IPV (H1); and whether IPV decreases likelihood of subsequent risk reduction behavior (i.e., requesting to use condoms) or increases likelihood of certain sexual risk-related factors (i.e., inconsistent condom use, having unprotected anal sex, having more than one partner, exchanging sex for drugs or money, having had an STI, being HIV positive, having a partner who engaged in HIV risk) (H2). Participants were interviewed at three waves: baseline, six months and twelve months. Hypotheses were examined using propensity score matching and multiple logistic regression analyses. The prevalence rate of any physical or sexual IPV was 46% at baseline. Findings for H1 indicate that women who reported always using condoms at wave 2 were significantly less likely than women who reported inconsistent or no condom use to experience subsequent IPV at wave 3. Similarly, increased risk of IPV at wave 3 was associated with self-reported STIs (OR ¼ 2.0, p ¼ :03), and unprotected anal sex (OR ¼ 2.0, po:01); always requesting that partners use condoms was associated with a significant decrease in subsequent IPV (OR ¼ .18, po:01). Findings for H2 suggest that IPV at wave 2 decreased the subsequent likelihood of always using condoms at wave 3 (OR ¼ .41, po:01) and always requesting that a partner use condoms (OR ¼ .42, p ¼ :02). The implications of the findings for HIV prevention interventions for women on methadone are discussed. r 2004 Elsevier Ltd. All rights reserved. Keywords: Intimate partner violence; HIV; Drug use; USA

Introduction

Corresponding author. Social Intervention Group, Columbia University School of Social Work, 1255 Amsterdam, New York, NY 10027, USA. Tel.: +1 212 851 2126; fax: +1 212 854 8549. E-mail address: [email protected] (N. El-Bassel).

Over the past decade, HIV/STIs and intimate partner violence (IPV) have emerged as significant co-occurring public health problems affecting a large number of women in drug treatment programs (Amaro, 1995; Amaro, Fried, Cabral, & Zuckerman, 1990; Cunningham, Stiffman, Dore, & Earls, 1994; El-Bassel, Gilbert, & Rajah, 2003; El-Bassel, Gilbert, Rajah, Foleno, &

0277-9536/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.socscimed.2004.11.035

ARTICLE IN PRESS 172

N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183

Frye, 2000a; Fernandez, 1995; Gilbert, El-Bassel, Schilling, Wada, & Bennet, 2000b; Worth, 1989; Wyatt, 1991). Studies have shown that past year prevalence rates of physical and sexual IPV among women in drug treatment have ranged between 25% and 57% (Brewer, Fleming, Haggerty, & Catalano, 1998; Chermack, Fuller, & Blow, 2000; El-Bassel, Gilbert, Schilling, & Wada, 2000b). These past year prevalence rates for IPV are 3–5 times higher than those found in epidemiological surveys of community-based samples of women, which range between 8% and 16% (Caetano, Nelson, & Cunradi, 2001; Straus & Gelles, 1990; Tjaden & Thoennes, 1998). Cross-sectional research suggests that IPV is associated with a number of sexual HIV/STI transmission risk related factors, including (1) engaging in unprotected sex (Amaro, 1995; Amaro et al., 1990; Cunningham et al., 1994; Fernandez, 1995; Gilbert et al., 2000a; Wingood & DiClemente, 1997; Worth, 1989), (2) higher rates of STIs (El-Bassel et al., 2000b; Gilbert, El-Bassel, Schilling, Catan, & Wada, 1998; Wu, El-Bassel, Witte, Gilbert, & Chang, 2003), (3) sex with multiple partners (Gilbert et al., 2000a), (4) engaging in unprotected anal sex (ElBassel et al., 2000a), (5) positive HIV status (North & Rothenberg, 1993; Rothenberg & Paskey, 1995), (6) trading sex for drugs or money (El-Bassel et al., 2000b), and (7) having a risky sexual partner (e.g., one who injects drugs, is HIV positive and/or has had sex with multiple partners) (Gilbert et al., 2000a; Wu et al., 2003). Sexual risk reduction behaviors such as requesting or insisting that a partner use condoms have also been found to be associated with IPV (Gilbert et al., 2000a; Wu et al., 2003). To date, however, research has yet to elucidate the possible temporal relationships between sexual HIV/STI transmission risk related factors and experiencing IPV among drug-involved women: Do such sexual HIV/STI transmission risk-related factors and/or sexual risk reduction behaviors, such as requesting that a partner use condoms, lead to IPV? Does experiencing IPV lead to an increase in a woman’s risk of sexual HIV/STI transmission or to a decrease in sexual risk reduction behaviors? The first temporal direction considered—that sexual HIV/STI risk-related factors or sexual risk reduction behavior leads to IPV—can be explained through the context of gender roles and power imbalances in relationships. The balance of power in intimate heterosexual relationships is influenced by the gender role norms and beliefs of both partners (Marin, Gomez, Tschann, & Gregorich, 1997). Men who subscribe to traditional gender roles are likely to exert control over the sexual relationship and to decide when, where and how sex occurs. Within this context, a woman’s request to use condoms represents a challenge to prevailing traditional beliefs about gender roles, raising questions regarding the male’s level of control over the sexual

relationship as well as regarding the level of intimacy, trust, and commitment in the relationship (Amaro, 1995; El-Bassel et al., 1998). If a woman suspects infidelity, injection drug use, or other risky behaviors, requesting that her partner use condoms or get tested for HIV may be interpreted by him as a lack of trust or care (El-Bassel et al., 2000a; Gilbert et al., 2000b; Kelly & Kalichman, 1995; Kelly et al., 1993). It may imply to some men that she has engaged in risky behaviors and be perceived as a sign of infidelity and a breach of gender role expectations on her part (El-Bassel et al., 2000a). Such perceptions threaten the stability of the couple, increasing the likelihood of abuse (O’Leary & Wingood, 2000), as some men resort to using physical and/or sexual IPV as a mechanism to repair their masculine self-esteem and maintain power. Some women may try to safeguard themselves from HIV/STI transmission by refusing sex or, at least, refusing unprotected sex or asking their partners to use condoms. In retaliation to the refusal, the partner may react violently towards the woman (ElBassel et al., 1998, 2000a; Gilbert et al., 2000b). Similarly, a woman’s disclosure of an STI or positive HIV status may raise questions of infidelity and trigger episodes of IPV (Gielen, O’Campo, Anderson, Keller, & Faden, 2000). Existing research has found positive STI or HIV status to be associated with IPV (Gielen et al., 2000; Gielen, O’Campo, Faden, & Eke, 1997; North & Rothenberg, 1993; Rothenberg & Paskey, 1995). It is not clear whether this association is related to disclosure of positive STI or HIV status or a result of other mediators factors, such as relationship power dynamics, that make condom use more unlikely, and consequently, HIV/STIs more likely. Furthermore, because drug-dependent women are often considered ‘‘sexually promiscuous’’ or ‘‘damaged goods,’’ they are perceived as violating traditional gender role norms, and thus, viewed as more deserving of abuse (Miller, 1990). Several studies have documented the perilous and degrading circumstances under which women who exchange sex for money or drugs operate, where coercive sex is common and condom use infrequent (Edlin et al., 1994; El-Bassel et al., 1996; Fullilove, Lown, & Fullilove, 1992; Irwin et al., 1995; Lown, Winkler, Fullilove, & Fullilove, 1993). The second direction posits that the experience of IPV leads to unprotected sex and a greater likelihood of HIV and other STIs as abused women are less likely to request or insist that their partners use condoms to avoid risking future IPV. Experiencing IPV and engaging in unsafe sex occurs as an extension of the unequal distribution of sexual, social and economic power between men and women and the low social status of drug-involved women. Gender theory provides a useful conceptual framework for understanding how gender roles reinforce differentials in sexual relationship power that create a pretext for IPV and leave women

ARTICLE IN PRESS N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183

vulnerable to HIV/STIs (Amaro, 1995; Zierler & Krieger, 1997). After experiencing IPV, some women may be hesitant to attempt to negotiate condom use as well as be afraid to refuse unprotected sex (El-Bassel et al., 1998; Gilbert et al., 2000b). This context of fear of violence renders women unable to shield themselves from HIV/STI transmission. Fear of IPV has been implicated as a risk factor for having unprotected sex in several qualitative and quantitative studies (El-Bassel et al., 1998; Gilbert et al., 2000b; Morrill & Ickovics, 1996; Wingood & DiClemente, 1997). Alternatively, the bi-directional relationships between IPV and HIV/STI transmission risk may be mediated by several psychosocial factors that have been found to be independently associated with both IPV and HIV/STI risk. Childhood sexual abuse has been independently associated with both IPV (Boyd, 1993; Downs, Miller, Testa, & Panek, 1992; Gilbert, El-Bassel, Schilling, & Friedman, 1997; Marshall & Rose, 1990) and with sexual HIV risk behaviors (Wingood & DiClemente, 1997; Wyatt et al., 2002; Zierler et al., 1991). Posttraumatic Stress Disorder (PTSD) has also been found to be independently associated with experiencing IPV (Dansky, Byrne, & Brady, 1999; Schiff, El-Bassel, Engstrom, & Gilbert, 2002) and engaging in HIV risk behaviors (Wyatt et al., 2002). Similarly, research has demonstrated that psychological distress is associated with experiencing IPV (Gilbert et al., 1997; Schiff et al., 2002) and HIV risk behavior (Camacho, Brown, & Simpson, 1996; El-Bassel, Ivanoff, Schilling, Borne, & Gilbert, 1997a). Drug and/or alcohol use by women and/or their partners may also influence the relationship between IPV and sexual HIV risk behaviors. A growing body of research has documented associations between drug and alcohol use and IPV (Berenson, Stiglich, Wilkinson, & Anderson, 1991; El-Bassel et al., 1997a; El-Bassel & Witte, 2001; El-Bassel, Witte, Wada, Gilbert, & Wallace, 2001; Miller, 1990; Miller et al., 1990). Numerous studies have linked drug and alcohol use with having unprotected sex, sex with a risky partner, sex with multiple partners, exchanging sex for money or drugs as well as with HIV and other STIs (Chiasson et al., 1990; Edlin et al., 1992, 1994; El-Bassel et al., 1997b). Studies have also linked having a low income, financial dependency and unemployment to both IPV (Mason & Blankenship, 1987; Straus, 1984) and HIV-risk behaviors (Monti-Catania, 1997; Zierler, 1997). Most studies examining the relationship between IPV and sexual HIV/STI risk-related factors have used nonrandom samples of women and cross-sectional designs. Neither the study designs nor the methodologies used in these past studies are conducive to estimating causal effects. Cross-sectional designs generally do not allow for a time lapse between when a potential causing agent is measured and outcomes are observed. Moreover,

173

causal estimation in any kind of observational study must rigorously address potential bias that may arise due to lack of randomization to treatment and control groups; rigorous approaches must account for shortcomings noted in the theoretical and empirical statistical literature regarding traditional or common approaches such as covariance adjustment which may be insufficient to remove bias due to potentially strong assumptions regarding model specification (Allison, 2002; Dehejia & Wahba, 2000; Rosenbaum, 1984; Rubin, 1973, 1979; Rubin & Thomas, 2000). To address these gaps, we conducted a longitudinal, panel study with a random sample of 416 women in MMTPs to examine the temporal relationship between sexual HIV/STI transmission risk-related factors and IPV. Using propensity score matching to minimize bias arising from the non-experimental design, we tested the following two hypotheses:

Hypothesis 1. (H1): Sexual HIV/STI transmission risk factors and/or risk reduction behavior increase the likelihood of subsequent IPV—We tested whether women who reported sexual HIV/STI transmission risk factors (i.e., being HIV positive, having an STI, inconsistent condom use, engaging in unprotected anal sex, or having more than one concurrent partner) or sexual risk reduction behaviors (i.e., requested or insisted that partner use condoms) were more likely to report experiencing physical or sexual IPV at a subsequent time point compared to women who did not report one of these risk factors or risk reduction behaviors. In this hypothesis the HIV/STI risk indicators are our predictor variables of interest and the IPV variables are our outcomes.

Hypothesis 2. (H2): IPV decreases the likelihood of sexual risk reduction behavior and increases the likelihood of subsequent risk of HIV/STI transmission—We tested whether women who reported IPV were at greater odds than women who did not report IPV to report elevated risk for HIV/STI transmission at the subsequent assessment time point. In this hypothesis, the IPV variables are our predictor variables of interest and the HIV/STI-risk indicators are our outcomes.

Methods Recruitment and eligibility We randomly selected 753 women from the total population of women who were enrolled in 14 MMTPs in New York City. Eligibility criteria were: (1) being female between the ages of 18–55 years; and (2) during

ARTICLE IN PRESS 174

N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183

the past year, having had a sexual or dating relationship with someone whom she described as her boyfriend, girlfriend, spouse, regular sexual partner, or the father of her children. A total of 559 women agreed to participate in a 15-min screening interview and 427 met eligibility criteria. Of those eligible, 416 women agreed to participate. Study participants were interviewed at baseline (wave 1) and received follow-up interviews at 6 months (wave 2) and 12 months (wave 3). Data were collected between 1997 and 2000. We excluded 11 women who reported having had a sexual or dating relationship with same sex partners at all three waves because of their substantially different patterns of sexual risk behaviors for STI/HIV transmission. Thus, data are presented on 405 women. The Institutional Review Boards of the participating MMTPs and Columbia University (the research institution) approved the protocol for this study.

Measurement Sexual HIV risk was assessed with several measures focusing on the 6 months prior to the assessment time point, such as frequency of condom use among sexually active women (always, sometimes, never), frequency of requesting condom use among women who do not always use condoms (always, sometimes, never), having had unprotected anal sex (yes/no), having had more than one partner (yes/no), exchanging sex for drugs or money (yes/no), having had an STI (yes/no), and being HIV positive. IPV was measured using the Revised Conflict Tactics Scales (CTS2) (Straus, Hamby, Boney-McCoy, & Sugarman, 1996). The CTS2 measures the extent of intimate partners, sexual, physical and injurious attacks on one another during the last 6 months (Straus, Hamby, Boney-McCoy, & Sugarman, 1996a). The CTS2 provides a scoring system for prevalence of different types and severity of IPV (i.e., mild and severe physical violence, mild and severe sexual coercion, and mild and severe injuries) in the last 6 months and lifetime. For each of the 39 items in the CTS2, subjects were asked how frequently their intimate partner enacted a certain behavior on them. Responses were scored on a 7 point Likert scale, ranging from never to 20 or more times in the past 6 months. If respondents indicated that they experienced one or more incidents for any item in a subscale in the past 6 months, they were categorized as positive for that subscale. These three subscales (i.e., physical, sexual, injurious) provide an overall prevalence of IPV across all types, including minor and severe components, which we defined for the purpose of this paper as ‘‘physical and/or sexual IPV.’’ Internal consistency of the CTS2 scales is high (.85) (Straus et al., 1996a).

Observed potential confounders Potential confounders are variables that we hypothesize may be related to specific attributes of the predictor of focal interest, the outcome variable, or both. In a non-randomized controlled study, if one simply compares the mean outcomes across two groups that differ with respect to a particular predictor variable, differences in outcome variables may be the result of differences in the predictor variable, the confounders, or both. Unbiased estimation of the effects of a specific predictor on an outcome variable must separate out the effects of these confounders. Thus, in the absence of randomization, estimation of causal effects of a particular predictor variable requires ‘‘controlling for’’ effects of confounders (more fully described in ‘‘Data Analysis’’ below). The selection of the following confounders included in the data analyses was empirically guided based on findings noted above in the literature review on factors associated with both IPV and sexual HIV/STI transmission risk-related factors among women. Socio-demographics and relationship characteristics included age, ethnicity, level of education, incarceration in the past 6 months, homelessness in the past 6 months, employment status in the past 6 months, and average monthly income over the past 12 months. Relationship characteristics included length and type of relationship, and number of intimate partners in the past year. Childhood sexual abuse was measured using the Childhood Sexual Abuse Interview (CSAI), which assessed experiences ranging from an invitation to do something sexual, to sexually oriented touching, to intercourse. It was scored by self-report of the presence or absence of each item. Two subscales were found to have good construct validity and reliability (El-Bassel et al., 2000b). Touching/exposure was measured by summing five items (e.g., did anyone ever touch your body, breasts or private sexual parts; did anyone ever show you their private sexual parts?); and penetration was quantified with three items (e.g., did anyone ever put his penis in your mouth, or put their mouth on your private sexual parts; did anyone have intercourse with you?) (Finkelhor, 1979; Sgroi, 1982). A positive indicator for abuse also required that force was used, or that the perpetrator was at least 5 years older than the victim at the time of occurrence, or a relative. Mental health status: The Brief Symptom Inventory (BSI) (Derogatis, 1993; Derogatis & Savitz, 1999) was used to assess psychological distress. The BSI includes a global severity index that provides a flexible overall assessment of psychological status. It has good internal consistency .71–.74 and has been tested with a wide range of populations (Derogatis & Melisaratos, 1983). Post-traumatic stress disorder was assessed using the Post-traumatic Stress Diagnostic Scale (PDS) (Foa, 1995), which is a 49-item self-report instrument designed to aid in the diagnosis of PTSD. The structure and

ARTICLE IN PRESS N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183

content of the PDS mirror the DSM-IV diagnostic criteria for PTSD. Relationship dependencies: Financial dependencies included housing dependency measured by one item (whether the woman or her partner held the lease on their housing, contribution to the household expenses measured by one item (whether the women and her partner contribute the same, or whether one partner contributed more). Drug dependency was measured by one item (whether the partner paid for the woman’s drugs). Joint drug activity was assessed with one item (whether the woman and her partner shared and/or split drugs with each other). Drug use was assessed by asking the participant whether she had used any illicit drugs (i.e., crack cocaine, marijuana, heroin, tranquilizers, speed) in the past 6 months. Alcohol use was assessed by an alcohol quantity frequency index from the Drug and Alcohol Use and Risk Behavior Questionnaire (El-Bassel, Ivanoff, Schilling, Gilbert, & Safyer, 1995), which measures the amount and frequency of consumption of alcohol in the past 6 months. For this paper, participants who reported any alcohol use in the past 6 months were categorized as positive for alcohol use. Perceived social support was assessed using the Multidimensional Scale of Perceived Social Support (MSPSS), a 12-item instrument designed to assess the extent to which respondents perceive the are given social support from: family, friends, and a significant other (Zimet, Dahlem, Zimet, & Farley, 1988). The MSPSS has been studied with a variety of populations and has excellent internal consistency, with alphas of .91 for the total scale and .90 to .95 for the subscales (Zimet et al., 1988).

Data analysis Step I: Addressing missing data Dropout rates were 17%, and 24% at waves 2 and 3, respectively. No statistically significant differences were observed on background variables, IPV or HIV risk variables between dropouts and non-dropouts. We used multiple imputation (Rubin, 1987; Schafer, 1997) via the MISSING library in S-Plus 6.0 to address missing data. Multiple imputation is advantageous with respect to minimizing bias versus alternative approaches to handling missing data such as complete case analyses when data may not be missing completely at random (see Allison, 2002 for a review of issues related to bias and missing data). This not only increases the sample size but facilitates generalization of the findings to the population from which it was randomly sampled. MI also relies on weaker (i.e., more plausible) assumptions than competing missing data techniques such as

175

complete case analyses, last observation carried forward, etc. (Rubin, 1987). Step II: Temporal sequencing of measures for causal hypothesis testing To maintain integrity with respect to inferring a causal chain of events, analyses were designed such that attributes of potential ‘‘causes’’ were obtained from assessments that preceded the measurement of the ‘‘effect.’’ For H1, the predictor variables of interest are HIV risk indicators measured at wave 2, and the outcome variable of experiencing IPV was assessed from data collected during wave 3. For H2, the predictor variable is IPV at wave 2, and the outcome variable is HIV risk behavior assessed at wave 3. For both sets of hypotheses, attributes of confounders were obtained from assessments at wave 1. Step III: Estimating effects of predictor variables without bias In the absence of randomization or experimental design, such as in observational studies, causal inference must address the lack of randomization that may lead to biased estimates of the effect of predictor variables on outcome variables. Attempting to control for these differences by relying solely on covariance adjustment in regression equations has been shown to be flawed because such approaches rely on strong parametric assumptions (for example, linearity), and, when groups are very dissimilar, may force extrapolations of the model over portions of the covariate space where there are no data (Dehejia & Wahba, 1999; Rubin, 1979). To address these issues, we used propensity score matching before computing estimates of the causal effects between each sexual HIV/STI transmission risk indicator and IPV described above. Propensity score matching is a technique used in observational studies to select groups that have similar values for observed confounders and differ only with respect to a predictor variable of interest (Connors et al., 1996; Dehejia & Wahba, 1999; Rosenbaum & Rubin, 1983, 1985; Stenestrand & Wallentin, 2001). In effect, propensity score matching reconstructs a sample that mimics the results of a randomized clinical trial where ‘‘treatment’’ and ‘‘control’’ groups (that is, groups with different attributes of the predictor variable such as HIV risk for H1 or IPV for H2) that are similar, on average, with respect to observed potential confounding covariates (Rosenbaum, 1984; Rosenbaum & Rubin, 1983). Briefly, propensity scores—formally, the probability that a person belongs to a group sharing a specific attribute on the predictor variable of interest conditional on attributes for the confounders—for each individual are calculated using logistic regression of the predictor variable of interest on the confounding covariates.

ARTICLE IN PRESS 176

N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183

Then, for an individual with a particular attribute for a predictor variable, a matched comparison group member is chosen as the individual who has the different predictor variable attribute and has the closest propensity score (i.e., nearest-neighbor matching). We used matching with replacement, which tends to yield less biased estimates than matching without replacement (Dehejia & Wahba, 2000). In these procedures, unmatched individuals are therefore not included, resulting in a different sample size for each analysis; inclusion of these non-comparable individuals would result in invalid estimates of causal effects. To judge whether the matching has been successful, the confounder distributions should be compared across the matched groups. For this study, we used two tests, the Kolmogorov–Smirnov and Shapiro–Wilk tests, to confirm that the matching procedures resulted in groups that were substantially similar on average with respect to confounders and differed only with respect to the predictor variable of interest. The dangers of using regression are kept to a minimum once the groups being compared have been forced to be similar in terms of the confounders, e.g., after propensity score matching. It has been demonstrated that combining matching with subsequent covariance adjustments (e.g., first matching and then performing regression on the matched sample) for all confounding covariates or a subset of key variables, is superior to either matching or covariance adjustment used alone (Rubin, 1973, 1979; Rubin & Thomas, 2000). After propensity score matching procedures were used to select a final well-matched sample of participants, we used multiple logistic regression to test each hypothesis. Adjusted odds ratios (ORs) and their associated 95% confidence intervals (CIs) were examined to test the hypotheses. The ORs were adjusted by including standard background variables of age, ethnicity, education, drug and alcohol use as well as the baseline report of the outcome variable for each set of hypotheses (i.e., baseline IPV reports for H1 and baseline HIV risk measures for H2). Assuming that the background variables we have controlled for represent all potential confounders, these estimates can be considered to be valid estimates of causal effects.

Table 1 Socio-demographic characteristics of the study sample Variable Age (years) Ethnicity African American White Latina Education level o High school High school or GED 4 High school Homelessness in past 6 months Marital status Single, never married Separated, divorced, or widowed Married Monthly income ($)

n x¯ ¼ 39:9

% SD ¼ 6.7 (%)

126 88 191

31 21 47

238 80 87 40

59 20 21 10

187 136 82 x¯ ¼ 624

46 34 20 SD ¼ 797

quarter (26%) said that they relied on their partners to pay for or supply drugs. Almost half (45%) of the participants reported that they shared or split drugs with their partners. History of childhood sexual abuse The prevalence of childhood sexual abuse among this sample was high. Fifty six percent had experienced touching or exposure and 28% had experienced penetration before age 16. Mental health status The average score of the global severity index of the BSI was .89. This mean is comparable to those found among other drug-involved female populations (ElBassel et al., 1997b), but it is substantially higher than the estimate of the mean global severity index of .37 for the general female population (Derogatis, 1993). Prevalence of IPV

Findings Socio-demographic characteristics of the sample that were collected at baseline are presented in Table 1. The average length of intimate relationships was 8.8 years (SD ¼ 7.8). In about two-thirds (68%) of the relationships, the women did not hold the housing lease. Half of the women reported that they contributed more than their partners to the household expenses. About one-

Prevalence of different types of IPV perpetrated by male intimate partners in the past 6 months reported at each wave of the study is presented in Fig. 1A. The prevalence rate of physical and/or sexual IPV was 46%, 41% and 31% at baseline, 6-, and 12-months follow-up respectively. These prevalence rates are high, but consistent with what has been reported in other studies (Brewer et al., 1998; Chermack et al., 2000; El-Bassel et al., 2000b). Fig. 1 illustrates that the prevalence of IPV for each wave decreased slightly over time.

ARTICLE IN PRESS N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183

Sexual & Physical IPV

50

Prevalence (%)

40 Combined

30

20 Severe Minor

10

0 0

(A)

6 Time (months)

12

Prevalence (%)

80

Did not always request condoms

70 Did not always use condoms

STI 20

HIV-positive HIV-positive partner

10

Unprotected anal sex Multiple concurrent partners

0 0

(B)

6 Time (months)

at wave 2 and subsequent physical and/or sexual IPV at wave 3. Support for this hypothesis was found for women who reported having an STI and women who reported engaging in unprotected anal sex. Women who reported always using condoms at wave 2 were significantly less likely than women who reported inconsistent or no condom use to experience IPV at wave 3. Similar support was found for women who always request that their partner use condoms compared to women who never or sometimes request condom use. Our findings did not demonstrate support for IPV as a consequence of having multiple concurrent partners or as a function of the woman or her partner’s HIV status. H2: Whether IPV leads to increased risk of sexual HIV/ STI transmission

HIV Risks

90

177

12

Fig. 1. Prevalence (with standard errors) versus time for (A) sexual and/or physical IPV, and (B) HIV risk reduction indicators ðN ¼ 406Þ:

Table 2B presents findings for each hypothesis testing whether the experience of IPV is associated with subsequent sexual HIV/STI transmission risk factors and/or a decrease in subsequent risk reduction behavior. Reporting experiencing IPV at wave 2 was associated with significantly lower likelihood of always using condoms and always requesting that a partner use condoms at wave 3. At the 95% level of confidence, we did not detect a significant association with IPV and subsequent reports of the following HIV/STI risk or risk reduction indicators: women’s HIV status, self-reported STIs, unprotected anal sex, having multiple concurrent partners, and partner’s HIV status.

Conclusion HIV risk The prevalence of HIV infection remained about 20% throughout the course of the study; the prevalence of self-reported STIs diagnosed during the 6 months prior to assessment was a few percentage points higher. The most prevalent HIV risk behavior reported was not always requesting condoms (ranging from 78 to 85%); followed by not always using condoms with regular partners (66–72%); engaging in unprotected anal sex (9–17%); and having multiple, concurrent partners (4–11%). The prevalence over time of selected HIV/ STI transmission risk indicators is presented in Fig. 1B. H1: Whether HIV/STI risk factors or sexual risk reduction behavior leads to IPV Table 2A presents the findings for each hypothesis on the temporal relationship between sexual HIV/STI transmission risk factors and risk reduction behavior

To our knowledge, this is the first longitudinal investigation of the temporal relationships between an array of sexual HIV/STI transmission risk factors and IPV among a random sample of women in drug treatment. The study examined the temporal relationships between sexual HIV/STI transmission risks and IPV at two points in time over a 1-year period (waves 2 and 3), controlling for potentially confounding variables collected at baseline (wave 1). The methodology of this study significantly improves on earlier cross-sectional studies with non-random samples. The use of a more advanced analytical procedure designed specifically for causal inference in combination with multiple logistic regression analysis also represents an improvement in statistical procedures, making it possible to account for the influence of potentially confounding factors on the relationships between HIV/STI transmission risks and IPV, in a more robust way. The study findings indicate multiple temporal relationships between sexual HIV/STI transmission risk factors and IPV. Significant associations between

178 Table 2 (A) Estimates of the temporal relationships between HIV risk indicators and subsequent IPV and (B) Estimates of the temporal relationships between IPV and subsequent HIV risk indicators Any IPV at wave 3 n

.51 (.18, 1.4)

.23

332

2.0 (1.1, 3.6)

.03

369

2.0 (1.3, 3.1)

o.01

391

.53 (.29, .97)

.05

358

.18 (.08, .41)

o.01

381

1.5 (.68, 3.3)

.21

381

.54 (.12, 2.4)

.43

354

Participant’s HIV status at wave 3 OR (95% CI)

p

n

STI at wave 3 OR (95% CI)

Unprotected anal sex at wave 3 Condom use consistency at wave 3 p

n

OR (95% CI)

p

n

IPV at wave 2 .21 (.03, 1.6) .61 246 .88 (.53, 1.5) .34 322 1.8 (.58, 5.5) .21 323 No (0) versus Yes (1) Condom request consistency at wave 3 Multiple concurrent partners at wave 3 Partner’s HIV status at wave 3 OR (95% CI) IPV at wave 2 .42 (.22, .82) No (0) versus Yes (1)

p

n

OR (95% CI)

p

n

OR (95% CI)

p

n

.02

318

3.1 (.89, 11)

.07

320

.87 (.46, 1.7)

.67

310

OR (95% CI)

p

n

.41 (.24, .71)

o.01

320

ARTICLE IN PRESS

(B)

p

N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183

(A) Participant’s HIV status at wave 2 Negative (0) versus Positive (1) STI at wave 2 No (0) versus Yes (1) Unprotected anal sex at wave 2 No (0) versus Yes (1) Condom use consistency at wave 2 Never or sometimes (0) versus always (1) Condom request consistency at wave 2 Never or sometimes (0) versus always (1) Multiple concurrent partners at wave 2 No (0) versus Yes (1) Partner’s HIV status at wave 2 Negative (0) versus Positive (1)

OR (95% CI)

ARTICLE IN PRESS N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183

inconsistent or non-condom use and IPV and between inconsistent or no requests for partners to use condoms and IPV were found for both hypotheses, consistent with previous cross-sectional studies linking inconsistent condom use and IPV (Amaro, 1995; Amaro et al., 1990; Cunningham et al., 1994; Fernandez, 1995; Gilbert et al., 2000a; Wingood & DiClemente, 1997; Worth, 1989; Wyatt, 1991). The bi-directional relationship between condom use and IPV may be indirectly or directly related to a woman’s attempts to negotiate condom use with her partner or to her experience of IPV. Women who always use condoms in their intimate relationships may not need to negotiate with their partners because norms of consistent condom use have already been established. In contrast, women who sometimes use condoms and attempt to initiate or re-negotiate their use may increase their risk of IPV. Their partners may perceive their inconsistent or new requests to use condoms as a sign of infidelity, a relapse to injection drug use or as an accusatory suggestion of such risky behaviors by their partners (Amaro, 1995; El-Bassel et al., 2000a; Wingood & DiClemente, 1997). The act of a woman requesting to use condoms may also be perceived by her male partner as a breach of gender role expectations and as a threat to the male’s sexual decision-making power. Such perceptions may lead to relationship conflict and IPV as some men resort to using physical and/or sexual IPV in retaliation or as a mechanism to repair their self-esteem and maintain power (El-Bassel et al., 1998, 2000b; Gilbert et al., 2000b). The mechanisms linking bi-directional relationships between IPV and condom use may arguably differ for women who never use or request condoms versus those who sometime use or request condoms. Women who never use condoms or never request that their partners use condoms may weigh the risk of unsafe sex with the risk of IPV and may legitimately decide they are safer not broaching the subject of condoms. This may explain the significant findings on inconsistent or non-condom use leading to IPV for H2. Alternatively, the bi-directional relationships between non-condom use and IPV found for H1 and H2 may occur as an extension of similar relationship power dynamics (i.e., relationship dependencies, couple communication skills, male dominance and gender imbalances) that contribute both to IPV and a woman’s inability to negotiate consistent condom use (El-Bassel et al., 1998, 2000a; Gilbert et al., 2000b). Women who experience these types of relationship power dynamics may submit to unprotected sex and be more subject to physical and/or sexual IPV when they are dependent on their partners for money, drugs, or social support. The relationship between engaging in unprotected anal sex at wave 2 and experiencing IPV at wave 3 found for H1 may also reflect relationship dynamics similar to

179

those described above, or alternatively, may suggest an overlap between these two variables as women often experience anal sex as a form of sexual IPV. In our qualitative research with women on methadone, several participants mentioned anal sex as a type of sexual abuse and also reported that sexual violence often occurred after physical violence (El-Bassel et al., 2000a; Gilbert et al., 2000b). Self-reported STIs in the previous 6 months doubled a woman’s subsequent risk of experiencing IPV, consistent with previous cross-sectional research (El-Bassel, Gilbert, Wada, Witte, & Schilling, 2000b; Gilbert et al., 1998). However, IPV at wave 2 was not significantly associated with self-reported STIs at wave 3. The significant findings supporting the hypothesis that recent STIs increase the risk of subsequent IPV may be related to relationship conflict over the disclosure of an STI to a partner and its potential association with infidelity. Alternatively, it is possible that this significant association is linked to the similar set of relationship power dynamics that lead to both IPV and unprotected sex, which may result in an STI. Contrary to previous research (North & Rothenberg, 1993; Rothenberg & Paskey, 1995), neither a woman’s self-reported HIV status nor a partner’s HIV status were significantly associated with IPV for H1 or H2. The lack of a relationship between a woman’s HIV status and IPV for H1 in contrast to the significant relationship for recent STIs may be due to the longer duration of time of HIV infection and disclosure. Since the vast majority of HIVpositive women had learned of their HIV status prior to participating in the baseline of the study, it is unlikely that this would have the same impact on relationship dynamics as a disclosure of a recent STI.

Limitations The current study is limited by several factors. First, although the generalizability of study findings to other low income, urban populations of women in MMTPs is greater than previous studies that used non-random samples, the findings may not be generalizable to women in the general population or to other populations of women in drug treatment. Second, while the use of propensity score matching and selection of confounding variables in this study represents a significant improvement over previous studies in controlling for potential confounders, the set of variables included in propensity score matching and as covariates in regression analysis does not exhaust the list of potential confounders which may influence the relationship between IPV and HIV risk. Future research on the relationship between HIV and IPV may benefit from the inclusion of positive psychosocial and relationship variables, which may influence HIV risk and IPV, such as relationship

ARTICLE IN PRESS 180

N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183

problem solving, couple communication skills, balance of relationship power, access and utilization of services as well as social support. Third, the associations between recent IPV and sexual risk indicators may reflect a social desirability bias: women who felt comfortable disclosing partner violence and risky sexual behaviors may have been more likely to report both than women who didn’t feel comfortable with disclosure of this personal information. Finally, Fig. 1 indicates that IPV decreased more from wave 1 to 3 compared to condom related HIV/STI risk variables. The marked decrease of IPV over time may have resulted from an MMTP treatment or may reflect a natural deterioration and cycle of a violent relationship. In either case, the differential drop between IPV and HIV risk may suggest a potential MMTP treatment effect that may have confounded study findings. Future research on the relationship between IPV and HIV/STI transmission risks should include biological testing of STIs to corroborate selfreport of STIs. Future research with larger sample sizes should also consider examining the actual numbers of women who transition from not experiencing IPV to experiencing IPV in conjunction with engaging or not engaging sexual risk behaviors and STIs over time. Such analysis would permit the ability to draw conclusions regarding causal relationships between IPV and sexual HIV/STI transmission risk factors as well as STI incidence and disclosure.

have been found to be associated with sexual HIV risk behavior among women, particularly among those who are drug involved (Wingood & DiClemente, 1997; Wyatt et al., 2002; Zierler et al., 1991). Strategies to address IPV in the context of HIV/STI prevention must take into account the balance of power in intimate relationships, level of sexual communication and negotiation skills, relationship dependencies, and an accurate appraisal of risk of IPV. These HIV/STI risk reduction intervention strategies may include interrelated cognitive awareness components and cognitive behavioral skills, such as: (1) raising awareness of the interpersonal contexts and triggers of HIV risk behavior and IPV, the meaning of the relationship, conflicts around drug use, gender roles and expectations around sex and safer sex between long-term heterosexual partners, (2) developing an accurate appraisal of risk of HIV and IPV as well as increasing safety planning, problem-solving, and sexual communication skills to negotiate successful HIV risk reduction with their partners, (3) increasing access to and utilization of services and (4) bolstering informal social support to help women establish and maintain relationship safety. Failure to address the co-occurring problems of HIV risk and IPV in an integrated approach for women in drug treatment may not only increase their risk of HIV, but may also jeopardize their recovery and safety.

Implications for intervention

Acknowledgements

Despite these limitations, this study has important implications for intervention. The high prevalence of recent or ongoing severe IPV among this random sample of women in MMTPs underscores the need to implement routine screening for IPV in these programs. Standard protocols for assessment, safety planning, treatment, and referrals to address the problem of IPV among individuals in drug treatment programs need to be designed, tested, and implemented. To date, few HIV prevention interventions have been developed and empirically tested to target the cooccurring problems of sexual HIV risks and IPV among drug-involved women. The disproportionately high prevalence rates of IPV among this population and the bi-directional causal relationships between sexual HIV risks and IPV found in this study underscore the need to develop HIV/STI prevention interventions that take into account these overlapping problems. In this study, as in others, women with a history of drug abuse tended to have experienced childhood sexual abuse, PTSD symptoms and meet PTSD criteria (Beck, Wright, Newman, & Liese, 1993; Chu, 1991; El-Bassel, Simoni, Cooper, Gilbert, & Schilling, 2001; Gilbert et al., 1997; Herman, 1992). Childhood sexual abuse and PTSD symptoms

This study was supported by NIDA Grant ] R01DA11027 awarded to Dr. El-Bassel.

References Allison, P. D. (2002). Missing data. In M. S. Lewis-Beck (Ed.), Quantitative applications in the social sciences. Beverly Hills, CA: Sage Publications, Inc. Amaro, H., Fried, L. E., Cabral, H., & Zuckerman, B. (1990). Violence during pregnancy and substance abuse. American Journal of Public Health, 80(5), 575–579. Amaro, H. (1995). Love, sex and power: considering women’s realities in HIV prevention. American Psychologist, 50, 437–447. Beck, A. T., Wright, F. D., Newman, C. F., & Liese, B. S. (1993). Cognitive therapy of substance abuse. New York, NY: Guilford Press. Berenson, A. B., Stiglich, N. J., Wilkinson, G. S., & Anderson, G. D. (1991). Drug abuse and other risk factors for physical abuse in pregnancy among white non-Hispanic, black, and Hispanic women. American Journal of Obstetrics & Gynecology, 164, 1491–1499. Boyd, C. J. (1993). The antecedents of women’s crack cocaine abuse: substance abuse, and sexual abuse, depression and

ARTICLE IN PRESS N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183 illicit drug use. Journal of Substance Abuse Treatment, 10, 433–438. Brewer, D. D., Fleming, C. B., Haggerty, K. P., & Catalano, R. F. (1998). Drug use predictors of partner violence in opiate dependent women. Violence and Victims, 13(2), 107–115. Caetano, R., Nelson, S., & Cunradi, C. (2001). Intimate partner violence, dependence symptoms and social consequences from drinking among white, black and Hispanic couples in the United States. American Journal of Addiction, 10(Suppl.), 60–69. Camacho, L. M., Brown, B. S., & Simpson, D. D. (1996). Psychological dysfunction and HIV/AIDS risk behavior. Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology, 11, 198–202. Chermack, S., Fuller, B., & Blow, F. (2000). Predictors of expressed partner and non-partner violence among patients in substance abuse treatment. Drug and Alcohol Dependence, 58(1–2), 43–54. Chiasson, M. A., Stoneburner, R. L., Lifson, A. R., Hildebrandt, D. S., Ewing, W. E., Schultz, S., & Jaffe, H. W. (1990). Risk factors for human immunodeficiency virus type 1 (HIV-1) infection in patients at a sexually transmitted disease clinic in New York City. American Journal of Epidemiology, 131(2), 208–220. Chu, J. A. (1991). The repetition compulsion revisited: reliving dissociated trauma. Psychotherapy, 28(2), 327–332. Connors, A., Speroff, T., Dawson, N., Thomas, C., Harrell, F., Wagner, D., Desbiens, N., Goldman, L., Wu, A., Califf, R., Fulkerson, W., Vidaillet, H., Broste, S., Bellamy, P., Lynn, J., & Knaus, W. (1996). The effectiveness of right heart catheterization in the initial care of critically ill patients. Support investigators. JAMA, 276(11), 889–897. Cunningham, R. M., Stiffman, A. R., Dore, P., & Earls, F. (1994). The association of physical and sexual abuse with HIV risk behaviors in adolescence and young adulthood: implications for public health. Child Abuse and Neglect, 18(3), 233–245. Dansky, B. S., Byrne, C. A., & Brady, K. T. (1999). Intimate violence and post-traumatic stress disorder among individuals with cocaine dependence. American Journal of Drug and Alcohol Abuse, 25(2), 257–268. Dehejia, R., & Wahba, S. (1999). Causal effects in nonexperimental studies: re-evaluating the evaluation of training programs. Journal of the American Statistical Association, 94, 1053–1062. Dehejia, R., Wahba, S. (2000). Propensity score matching methods for non-experimental causal studies, National Bureau Economic Research Technical Working Paper 6829. Derogatis, L., & Melisaratos, N. (1983). The Brief Symptom Inventory: an introductory report. Psychological Medicine, 13, 595–605. Derogatis, L. (1993). Brief Symptom Inventory: administration, scoring, and procedures manual. Minneapolis, MN: National Computer Systems, Inc. Derogatis, L., & Savitz, K. (1999). The SCL-90-R, Brief Symptom Inventory, and matching clinical rating scales. In M. E. Maruish (Ed.), The use of psychological testing for treatment planning outcomes assessment. Mahwah, NJ: Lawrence Erlbaum Associates.

181

Downs, W. R., Miller, B. A., Testa, M., & Panek, D. (1992). The long-term effects of parent-to-child violence for women. Journal of Interpersonal Violence, 7, 365–382. Edlin, B. R., Irwin, K. L., Ludwig, D. D., McCoy, H. V., Serrano, Y., Word, C., Bowser, B. P., Faruque, S., McCoy, C. B., & Schilling, R. F. (1992). High-risk sex behavior among young street-recruited crack cocaine smokers in three American cities: an interim report. Journal of Psychoactive Drugs, 24, 363–371. Edlin, B. R., Irwin, K. L., Faruque, S., McCoy, C., Word, C., Serrano, Y., Inciardi, J., Bowser, B., Schilling, R. F., & Holmberg, S. (1994). Intersecting epidemics: crack cocaine use and HIV infection in inner-city young adults. New England Journal of Medicine, 21, 1422–1427. El-Bassel, N., Ivanoff, A., Schilling, R. F., Gilbert, L., & Safyer, S. (1995). Correlates of problem drinking among drug using incarcerated women. Addictive Behaviors, 20, 359–369. El-Bassel, N., Gilbert, L., Ivanoff, A., Schilling, R. F., Borne, D., & Safyer, S. F. (1996). Correlates of crack abuse among incarcerated women: psychological trauma, social support and coping behavior. American Journal of Drug and Alcohol Abuse, 22, 41–56. El-Bassel, N., Ivanoff, A., Schilling, R. F., Borne, D., & Gilbert, L. (1997a). Skills-building and social support enhancement to reduce HIV risk among women in jail. Criminal Justice and Behavior, 24(2), 205–223. El-Bassel, N., Schilling, R. F., Irwin, K. L., Faruque, S., Gilbert, L., Von Bargen, J., Serrano, Y., & Edlin, B. (1997b). Sex trading and psychological distress among women recruited from the streets of Harlem. American Journal of Public Health, 87(1), 66–70. El-Bassel, N., Gilbert, L., Krishnan, S., Schilling, R. F., Gaeta, T., Purpura, S., & Witte, S. S. (1998). Partner violence and sexual HIV-risk behaviors among women in an inner-city emergency department. Violence and Victims, 13(4), 377–393. El-Bassel, N., Gilbert, L., Rajah, V., Foleno, A., & Frye, V. (2000a). Fear and violence: raising the HIV stakes. AIDS Education and Prevention, 12(2), 154–170. El-Bassel, N., Gilbert, L., Schilling, R. F., & Wada, T. (2000b). Drug abuse and partner violence among women in methadone treatment. Journal of Family Violence, 15(3), 209–225. El-Bassel, N., Simoni, J., Cooper, D., Gilbert, L., & Schilling, R. (2001). Sex trading and psychological distress among women on methadone. Psychology of Addictive Behaviors, 15(3), 177–184. El-Bassel, N., & Witte, S. (2001). Drug use and physical and sexual abuse of street sex workers in New York City. Research for Sex Work, 4, 31–32. El-Bassel, N., Witte, S., Wada, T., Gilbert, L., & Wallace, J. (2001). Correlates of partner violence among female streetbased sex workers: substance abuse, history of child abuse and HIV risks. AIDS Patient Care and STDs, 15(1), 41–51. El-Bassel, N., Gilbert, L., & Rajah, V. (2003). The relationship between drug abuse and sexual performance among women on methadone: heightening the risk of sexual intimate violence and HIV. Addictive Behaviors, 28(8), 1385–1403. Fernandez, M. I. (1995). Latinas and AIDS: challenges to HIV prevention efforts. In A. O’Leary, & L. S. Jemmott (Eds.),

ARTICLE IN PRESS 182

N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183

AIDS prevention and mental health (pp. 159–174). New York: Plenum. Finkelhor, D. (1979). Sexually victimized children. New York: The Free Press. Foa, E. B. (1995). Posttraumatic Stress Diagnostic Scale manual. Minneapolis: National Computer Systems, Inc. Fullilove, M. T., Lown, A., & Fullilove, R. E. (1992). Crack ‘‘hos and skeezers’’: traumatic experiences of women crack users. Journal of Sex Research, 29, 257–287. Gielen, A. C., O’Campo, P., Faden, R. R., & Eke, A. (1997). Women’s disclosure of HIV status: experiences of mistreatment and violence in an urban setting. Women and Health, 25(3), 19–31. Gielen, A. C., O’Campo, P., Anderson, J., Keller, J., & Faden, R. (2000). Women living with HIV: disclosure, violence, and social support. Journal of Urban Health, 77, 480–491. Gilbert, L., El-Bassel, N., Schilling, R. F., & Friedman, E. (1997). Childhood abuse as a risk for partner abuse among women in methadone maintenance. American Journal of Drugs and Alcohol Abuse, 23(4), 581–595. Gilbert, L., El-Bassel, N., Schilling, R. F., Catan, V., & Wada, T. (1998). Partner abuse and sexual risk behavior among women on methadone. Violence & Victims, 13(4), 1–17. Gilbert, L., El-Bassel, N., Rajah, V., Foleno, A., Fontdevila, J., Frye, V., & Richman, B. L. (2000a). The converging epidemics of mood-altering-drug use, HIV, HCV, and partner violence: a conundrum for methadone maintenance treatment. Mount Sinai Journal of Medicine, 67(5–6), 452–464. Gilbert, L., El-Bassel, N., Schilling, R., Wada, T., & Bennet, B. (2000b). Partner violence and sexual HIV risk behaviors among women in methadone treatment. AIDS and Behavior, 4(3), 261–269. Herman, J. (1992). Trauma and recovery. New York, NY: Basic Books. Irwin, K. L., Edlin, B. R., Wong, L., Faruque, S., McCoy, H. V., Word, C., Schilling, R. F., McCoy, C. B., Evans, P., Holmberg, S. D., the Multicenter Crack, C., & Team, H. I. V. I. S. (1995). Urban rape survivors: characteristics and prevalence of human immunodeficiency virus and other sexually transmitted infections. Obstetrics and Gynecology, 85(3), 330–336. Kelly, J. A., Murphy, D. A., Bahr, G. R., Koob, J. J., Morgan, M. G., Kalichman, S. C., & Stevenson, L. Y. (1993). Factors associated with severity of depression and high-risk sexual behavior among persons diagnosed with human immunodeficiency virus (HIV) Infection. Health Psychology, 12(3), 215–219. Kelly, J. A., & Kalichman, S. C. (1995). Increased attention to human sexuality can improve HIV-AIDS prevention efforts: key research issues and directions. Journal of Consulting and Clinical Psychology, 63, 907–918. Lown, E. A., Winkler, K., Fullilove, R. E., & Fullilove, M. T. (1993). Tossin’ and tweakin’: Women’s consciousness in the crack culture. In C. Squire (Ed.), Women and AIDS: Psychological Perspectives (pp. 90–105). London: Sage Publications. Marin, B. V., Gomez, C. A., Tschann, J. M., & Gregorich, S. E. (1997). Condom use in unmarried Latino men: a test of cultural constructs. Health Psychology, 15(5), 458–467.

Marshall, L. L., & Rose, P. (1990). Premarital violence: the impact of family or origin violence, stress and reciprocity. Violence and Victimization, 5, 51–64. Mason, A., & Blankenship, V. (1987). Power and affiliation motivation, stress, and abuse in intimate relationships. Journal of Personality and Social Psychology, 52(1), 203–210. Miller, B. A. (1990). The interrelationships between alcohol and drugs and family violence. National Institute on Drug Abuse Research Monographs, 103, 177–207. Miller, B. A., Nochajski, T. H., Leonard, K. E., Blane, H. T., Gondoli, D. M., & Bowers, P. M. (1990). Spousal violence and alcohol/drug problems among parolees and their spouses. Women and Criminal Justice, 1(2), 55–72. Monti-Catania, D. (1997). Women, violence, and HIV/AIDS. In N. Goldstein, & J. Manlowe (Eds.), The gender politics of HIV/AIDS in women (pp. 242–251). New York: New York University. Morrill, A., Ickovics, J. (1996). Surviving abuse and HIV risk: how women’s experience of abuse shapes their heterosexual risk for HIV. Poster symposium at the XI international conference on AIDS. Vancouver, CA. North, R. L., & Rothenberg, K. H. (1993). Partner notification and the threat of domestic violence against women with HIV infection. New England Journal of Medicine, 329(16), 1194–1196. O’Leary, A., & Wingood, G. M. (2000). Interventions for sexually active heterosexual women. In J. L. Peterson, & R. J. DiClemente (Eds.), Handbook of HIV prevention (pp. 179–197). New York: Plenum Publishing Corp. Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70, 41–55. Rosenbaum, P. R. (1984). The consequences of adjustment for a concomitant variable that has been affected by the treatment. The Journal of the Royal Statistical Society, A, 147(5), 656–666. Rosenbaum, P. R., & Rubin, D. B. (1985). Constructing a control group using multivariate matched sampling methods that incorporate the propensity score. The American Statistician, 39, 33–38. Rothenberg, K. H., & Paskey, S. J. (1995). The risk of domestic violence and women with HIV infection: implications for partner notificiation, public policy and the law. American Journal of Public Health, 85(11), 1569–1575. Rubin, D. B. (1973). The use of matched sampling and regression adjustment to remove bias in observational studies. Biometrics, 29, 185–203. Rubin, D. B. (1979). Using multivariate matched sampling and regression adjustment to control bias in observational studies. Journal of the American Statistical Association, 74, 318–328. Rubin, D. B. (1987). Multiple imputation for nonresponse in surveys. New York: Wiley Publications. Rubin, D. B., & Thomas, N. (2000). Combining propensity score matching with additional adjustments for prognostic covariates. Journal of the American Statistical Association, 95, 573–585. Schafer, J. L. (1997). Analysis of incomplete multivariate data. London: Chapman & Hall.

ARTICLE IN PRESS N. El-Bassel et al. / Social Science & Medicine 61 (2005) 171–183 Schiff, M., El-Bassel, N., Engstrom, M., & Gilbert, L. (2002). Psychological distress and intimate physical and sexual abuse among women in methadone maintenance treatment programs. Social Service Review, 76(2), 302–320. Sgroi, S. M. (1982). Handbook of clinical intervention in child sexual abuse. Lexington, MA: D.C. Heath and Company. Stenestrand, U., & Wallentin, L. (2001). Early statin treatment following acute myocardial infarction and one-year survival. JAMA, 285(4), 430–436. Straus, M., Hamby, S., Boney-McCoy, S., & Sugarman, D. (1996). The Revised Conflict Tactics Scales (CTS2): development & preliminary psychometric data. Journal of Family Issues, 17, 283–316. Straus, M. A. (1984). Family violence and non-family crime and violence. In A. J. Lincoln, & M. A. Straus (Eds.), Crime and the family. Springfield, IL: Thomas. Straus, M. A., & Gelles, R. (1990). How violent are American families? Estimates from the national family violence survey and other studies. In M. Straus, & R. Gelles (Eds.), Physical Violence in American Families (pp. 95–112). New Brunswick, NJ: Transaction Publishing. Tjaden, P., & Thoennes, V. (1998). Prevalence, incidence, and consequences of violence against women: findings from the national violence against women survey. Research in brief. Washington, DC: United States Department of Justice, National Institute of Health. Wingood, G. M., & DiClemente, R. J. (1997). The effects of an abusive primary partner on the condom use and sexual negotiation practices of African-American women. American Journal of Public Health, 87(6), 1016–1018.

183

Worth, D. (1989). Sexual decision-making and AIDS: why condom promotion among vulnerable women is likely to fail. Studies on Family Planning, 20(6), 297–307. Wu, E., El-Bassel, N., Witte, S. S., Gilbert, L., & Chang, M. (2003). Intimate partner violence and HIV risk among urban minority women in primary health care settings. AIDS and Behavior, 7(3), 291–301. Wyatt, G. E. (1991). Child sexual abuse and its effects on sexual functioning. Annual Review of Sex Research, 2, 249–266. Wyatt, G. E., Myers, H. F., Williams, J. K., Kitchen, C. R., Loeb, T., Carmona, J. V., Wyatt, L. E., Chin, D., & Presley, N. (2002). Does a history of trauma contribute to HIV risk for women of color? American Journal of Public Health, 92(4), 660–665. Zierler, S. (1997). Hitting hard: HIV and violence. In N. Goldstein, & J. Manlowe (Eds.), The gender politics of HIV/ AIDS in women (pp. 207–221). New York: New York University. Zierler, S., & Krieger, N. (1997). Reframing women’s risk: social inequalities and HIV infection. Annual Review of Public Health, 18, 401–436. Zierler, S. L., Fiengold, D. L., Laufer, D., Velentgas, P., Kantrowitz-Gordon, I., & Mayer, K. (1991). Adult survivors of childhood sexual abuse and subsequent risk of HIV infection. American Journal of Public Health, 81, 572–575. Zimet, G. D., Dahlem, N. W., Zimet, S. G., & Farley, G. K. (1988). The Multidimensional Scale of Perceived Social Support. Journal of Personality Assessment, 52, 30–41.