Paper 13 Ok

C 2004) AIDS and Behavior, Vol. 8, No. 4, December 2004 (
DOI: 10.1007/s10461-004-7322-5

Predictors of Unprotected Anal Intercourse Among HIV-Positive Latino Gay and Bisexual Men Paul J. Poppen,1,2 Carol A. Reisen,1 Mar´ıa Cecilia Zea,1 Fernanda T. Bianchi,1 and John J. Echeverry1 Received July 16, 2003; revised March 31, 2004; accepted April 12, 2004

This study examined sexual behaviors in a sample of 155 HIV-positive Latino gay and bisexual men. Nearly half the sample had engaged in unprotected anal intercourse in the past 12 months; unprotected anal intercourse was more likely when the partner was also HIVpositive. Separate regression models predicted the number of receptive and insertive partners for unprotected anal intercourse. Participants reported both more unprotected insertive and receptive partners if they had sex under the influence of alcohol or drugs. Older participants and those with lower levels of Latino acculturation reported having more partners with whom they took the receptive role during unprotected anal intercourse, whereas those with higher levels of depression reported having more partners with whom they took the insertive role. Hierarchical set logistic regression revealed that the dyadic variable of seroconcordance added to the prediction of unprotected anal sex with the most recent male partner, beyond the individual characteristics. Results show the importance of examining both individual and dyadic characteristics in the study of sexual behavior. KEY WORDS: Latino; MSM; HIV-positive; sex risk.

INTRODUCTION

one partner is HIV-positive, a situation in which there is actual risk. The current paper examines unprotected anal sex in a sample of HIV-positive Latino MSM. Latinos are an ethnic group affected by HIV disproportionately, with incidence rates nearly four times those of non-Latino whites (CDC, 2002). Male same-sex sexual contact is a common means of transmission for Latinos: more than half of the Mexican, Cuban, Central and South American men with AIDS in this country contracted the virus from an infected male sex partner (CDC, 2000). Despite HIV education and prevention efforts, the incidence of unprotected anal intercourse remains substantial among MSM, and levels for Latinos seem to be similar to those of other ethnic groups. Because it is rare to have a random sample of MSM, estimates of unprotected sex are derived from convenience samples of differing characteristics, with studies typically reporting percent of participants who engage in unprotected anal intercourse in a given time period. In a recent multi-city study

Men who have sex with men (MSM) continue to represent the largest number of new HIV cases annually in the United States, accounting for more than 40% of all new cases (CDC, 2002). Condoms greatly decrease the likelihood of sexual transmission from an infected partner: consistent and correct use of condoms is estimated to reduce the risk of transmission by at least 70% and perhaps as much as 95% (Pinkerton and Abramson, 1997). Condoms continue to be a critical tool for HIV prevention, but on many occasions they are not used. It is important, therefore, to understand factors and processes associated with having sex without a condom—especially when 1 George

Washington University, Washington, District of Columbia. 2 Correspondence should be directed to Paul J. Poppen, Department of Psychology, 2125 G Street, NW, George Washington University, Washington, District of Columbia 20052; e-mail: [email protected].

379 C 2004 Springer Science+Business Media, Inc. 1090-7165/04/1200-0379/0


380 (Boston, Chicago, Denver, New York, San Francisco, and Seattle) of more than 4,000 HIV-negative MSMs 16 and older, about half of the men had engaged in unprotected receptive anal sex and slightly more had engaged in unprotected insertive anal sex in the last 6 months (Koblin et al., 2003). In an ethnically diverse sample of young men, ages 23–29 (the Los Angeles Young Men’s Survey), over 30% of respondents had engaged in unprotected receptive anal sex in the last 6 months, again with a slightly higher percent reporting unprotected insertive sex (Bingham et al., 2003). Latinos were similar in incidence of unprotected sex to African Americans and non-Latino whites. In another ethnically diverse sample, from Texas, over half reported engaging in unprotected anal sex during the past month, with 45% of the Latino subsample reporting at least one episode (Lye Chng and Geliga-Vargas, 2000). Marks et al. (1999) reviewed studies that examined sexual risk behaviors of individuals who knew they were HIV-positive. In studies of MSM, estimates of unprotected anal sex ranged from 10% in the last 2 months in a sample from Los Angeles HIV outpatient clinics to nearly 60% in the last year in a San Francisco study of young men. In the Chicago MACS and AIM intervention, no differences were found in the proportion of HIV-positive and HIV-negative men reporting unprotected anal intercourse (Ostrow et al., 1999); however, Lye Chng ´ and Geliga-Vargas (2000) found that HIV-positive men were more likely to have unprotected sex than HIV-negative men. Understanding Unprotected Anal Intercourse for HIV-Positive MSM

Poppen, Reisen, Zea, Bianchi, and Echeverry believe that a fuller understanding of the dynamics influencing unprotected anal intercourse requires an examination of the characteristics of the dyadic context, not just of the individual. Factors such as the relationship and seroconcordance vary with the partner, and the same individual may behave differently depending on the circumstances. In addition, it is critical to identify factors that are associated with unprotected anal intercourse for HIV-positive MSM. Kalichman (2000) argued that interventions to change sexual risk behaviors of HIV-positive men must differ from those for HIV-negative men, because the factors associated with continuing risk practices vary in the two groups. In this paper, we examined unprotected anal intercourse in relation to characteristics of both the individual and the dyad in a sample of HIV-positive men. Demographic Factors HIV is not evenly distributed across socioeconomic strata, and many studies have examined demographic factors in relation to unprotected sex. Variables such as education and age have been shown to covary inconsistently with sex risk. Lower levels of education have been associated with greater likelihood of unprotected anal intercourse in some samples of MSM (Janssen et al., 2000; Kalichman et al., 1998). There is also some support for the contention that younger men are more likely to engage in unprotected sex (McAuliffe et al., 1999); however, other studies have failed to find this relationship (Ekstrand et al., 1999). Depression

Much research has addressed factors associated with having unprotected anal intercourse, but most has focused on those who are HIV-negative. Theoretical models have emphasized an individual’s perceptions, beliefs, experiences, feelings, and demographic characteristics as predictors of sexual risktaking (e.g., Becker and Joseph, 1988; Catania et al., 1990; Fisher and Fisher, 2000), and most empirical studies have tested the relationship between these individual-level constructs and unprotected anal intercourse. Thus, research has typically addressed the association between fairly stable personal characteristics or traits and unprotected sex. It is important to note, however, that sexual behavior can vary depending on circumstances (Bajos and Marquet, 2000; Corbin and Fromme, 2002). We

Another predictor of unprotected anal sex that has been examined is depression. Although most theories posit that negative mood states would be associated with more sexual risk-taking (Marks et al., 1998a; Strathdee et al., 1998), the argument has also been made that depression would be associated with greater caution and therefore less risk-taking (Frijida, 1988). Empirical studies have yielded inconsistent findings, and recently Crepaz and Marks (2001) concluded from a meta-analysis that there was little evidence that depression was related to greater sexual risk-taking. There were, however, substantial correlations—both positive and negative—among the studies included in this meta-analysis, thus creating a mean effect size near zero. This situation

Unprotected Anal Sex and HIV-Positive Latino Men could indicate that the relationship between depression and sexual risk is dependent on other factors, including characteristics associated with the partner or the dyadic context of the sexual encounter. Acculturation Another factor relevant for immigrant groups is acculturation, the process of acquiring beliefs and behaviors from a new culture, while preserving or modifying traditions of the culture of origin. Latinos in the US have attitudes, beliefs, and expectations stemming from their US and Latino cultures to varying degrees. Cultural norms and scripts about sexual behavior and romantic love from both cultures can influence sexual behavior and condom use (D´ıaz, 1998). Level of acculturation into the US has been shown to be associated with sexual attitudes and behavior in samples of Latinos (Mar´ın et al., 1998; Mason et al., 1995). Marks et al. (1998b) found that Latinos who were more acculturated into US society had higher levels of sexual risk than did less acculturated Latinos. On the other hand, San Doval et al. (1995) found that more acculturated Latinos had more positive attitudes toward condoms. Substance Use Although the evidence is not consistent (e.g., Leigh and Stall, 1993; Stall and Purcell, 2000), many studies have reported an association between alcohol or drug use and unprotected sex (e.g., Dolezal et al., 1997; Ross et al., 2001). In the research literature, substance use is often treated as an individuallevel variable. In these cases, a positive correlation can be interpreted as showing that people who drink or use drugs also tend to take sexual risks. In contrast, other studies examine substance use occurring in conjunction with the sexual encounter and provide a more compelling test of a causal relationship (Dolezal et al., 2000; Hays et al., 1997b; Kelly and Kalichman, 1998; Stall et al., 1986). The presence of alcohol and drugs can then be seen as part of the context in which the dyad is acting. Serostatus Serostatus is assessed on the individual level, but can be used to characterize the individual or

381 the dyad. As an individual characteristic, it affects a person’s motivation for protected sex. HIV-negative men may be motivated to protect themselves and avoid potential infection with HIV, as well as other STDs. HIV-positive men may also be motivated to protect themselves from additional strains of the virus or other STDs; however, this risk is usually perceived as less threatening (Kalichman, 2000). In contrast, a primary reason that many HIV-positive men use condoms is to protect their sexual partners, not themselves (Wenger et al., 1994). As a dyadic characteristic, the issue concerns seroconcordance, the match or mismatch of the serostatus of the partners. Hays et al. (1997a) reported that HIV-positive seroconcordant couples had the highest incidence of unprotected anal intercourse, followed by HIV-negative seroconcordant couples. In this situation where partners have the same serostatus, a “negotiated safety” can be achieved regardless of the use of condoms (Davidovich et al., 2000). When one partner is HIVpositive and the other HIV-negative, however, the perceived risk of transmission is higher, and as would therefore be expected, reported rates of unprotected sex are lower (Hays et al., 1997b). Moreover, disclosure of positive serostatus to a partner has been associated with greater likelihood of unprotected sex when the partner was also positive, but lower likelihood of unprotected anal intercourse when the partner was negative, in both Latino (Marks et al., 1991) and mixed ethnic samples (DeRosa and Marks, 1998).

Relationship With Partner The dyadic characteristic of relationship between partners has consistently been associated with sexual risk behavior. Among MSM, there have been many reports of higher rates of unprotected anal intercourse between more committed partners than between less committed partners (e.g., Davidovich et al., 2000; Elford et al., 2001; Hays, Kegeles et al., 1997; Lye Chng and Geliga-Vargas, 2000). Moreover, this association has also been found in studies doing within-subject comparisons (Buchanan et al., 1996; ´ Carballo-Dieguez et al., 1997; Myers et al., 1999): Thus, a man is more likely to have unprotected sex with his primary partner than with his casual partners. There are several reasons that might explain the influence of relationship status on sex risk behaviors.

382 Men may feel emotionally safer with their primary partners, regardless of actual risk of transmission (Remien et al., 1995). Among primary partners, serostatus of both partners is often known, and knowledge of seroconcordance can lead to a decision to forego condom use (Kippax et al., 1993). Moreover, qualitative studies have suggested that condoms are sometimes seen as a barrier to intimacy and are therefore not used by committed couples (Flowers et al., 1997). Models to Predict Unprotected Anal Intercourse The current study examined sex without condoms in several ways. We described rates of unprotected anal intercourse in our sample of HIV-positive Latino MSM. We also tested models predicting the number of insertive and receptive partners for unprotected anal intercourse during the last 12 months. Predictors included the following individual characteristics: age, education, depression, acculturation into the US, alcohol and substance use, and whether the respondent had a main partner. We hypothesized that younger age, less education, and more substance use during sex would be associated with having more unprotected anal sex partners. In addition, we anticipated that men who had a main partner would have a lower number of partners for unprotected anal intercourse in the past year. We included measures of depression and acculturation in Latino culture, but we did not have directional hypotheses due to mixed findings in the literature. In addition, we tested a model predicting unprotected anal intercourse with the most recent sexual partner using both dyadic and individual factors. We hypothesized that the inclusion of the dyadic factors of seroconcordance and relationship between partners would add significant explanatory power beyond the individual factors of education, age, Latino acculturation, depression, and substance use during sex. Moreover, we hypothesized that there would be a greater likelihood of unprotected anal sex with HIVpositive partners, who would therefore be seroconcordant with our HIV-positive respondents, and with main partners. METHODS Procedure Data for this study were collected as part of a larger study on disclosure of HIV-positive sta-

Poppen, Reisen, Zea, Bianchi, and Echeverry tus of Latino gay men conducted in Washington, DC and New York City (Zea et al., 2004). Latino HIV-positive gay and bisexual men represent a hidden population that cannot be accurately enumerated, and therefore it is impossible to derive a truly representative sample of the whole population. We used a variety of approaches described in the literature as effective with hidden populations, including targeted sampling, key informant sampling, and snowball sampling (Carlson et al., 1994; Watters and Biernacki, 1989). Recruitment took place throughout the community at clinics, community organizations, public events (e.g., Pride Day and AIDS Walk), gay businesses, clubs, and bars. We used flyers and advertisements, as well as direct appeals, to inform potential participants about the study. Screening was based on the following inclusion criteria: being 18 years or older, HIV-positive, Latino, male, and self-identified as gay or bisexual. Some screening was done at the time of the direct appeals, and for these cases, we do not have information to estimate eligibility or refusal rates. Other screening was done when prospective participants called to inquire about the study in response to flyers and advertisements. Of those calling in, 80% met the criteria for eligibility; of these, 62% completed interviews. Participants completed a survey in either Spanish or English. Existing measures were translated from English into Spanish and translated back using the procedures suggested by Brislin (1986) and Mar´ın and Mar´ın (1991) to ensure equivalence. A panel of experts from different Spanish-speaking countries reviewed the resulting Spanish version to eliminate regional expressions that would not be universally recognized by Spanish speakers. The survey was administered using computer assisted self-interview technology (Audio-CASI). A native Spanish speaker provided instruction in the use of computers for the survey and was available throughout the administration to answer any questions. Surveys were administered either in small groups or individually, according to the preference of the participant. Responses were indicated by touching the computer screen, therefore computer skills were not required. Moreover, participants with reading difficulties could listen to the audio presentation of the questions and answers. About 87% responded to the survey in Spanish. Participants were reimbursed for their participation in the study and given a list of resources for HIV/AIDS services in their area.

Unprotected Anal Sex and HIV-Positive Latino Men Participants Participants were 155 HIV-positive Latino gay men ranging from 18 to 67 years of age, with a mean 38.5 years; 57% were recruited in New York City and 43% in Washington, DC. Fifty-one percent had education beyond high school, attending or graduating from trade school or college, while 49% had high school education or less. (This dichotomization was used in later model testing.) Seventy-six percent of the sample earned less than $800 per month. About 90% of the participants were immigrants, with 40% coming from South America, 26% from the Caribbean, 20% from Central America, and 13% from Mexico. Forty-nine percent of the 140 immigrants came to the US before age 25. Among a list of reasons for coming to the US, the most frequently endorsed were to improve financial situation (46%), to live a homosexual life more openly (34%), and to get HIV medication (26%). The samples from Washington and New York were comparable, but differed in a few ways. Although there were no significant differences in education, the New York participants were older and had lower income, with a larger proportion receiving some form of government aid (75% vs. 35%). In addition, a greater proportion of the DC sample came from Central America, while a greater proportion of the NY sample came from South America and the Caribbean.

Measures Sexual Behavior We asked questions about sexual behaviors in the last 12 months, in the last 30 days, and with the most recent partner. For the period of the last 12 months, we asked participants to indicate the number of male partners with whom they had had anal sex without a condom and the number of female partners with whom they had had anal or vaginal sex without a condom. For the previous 30 days, we asked whether the participants had sex in conjunction with alcohol or other drugs. We also asked about sexual behaviors with the most recent partner, including whether participant had ever had unprotected anal intercourse with that person. The main outcome variables for this study were the number of partners for anal intercourse without a condom in the last 12 months with whom the participant took the re-

383 ceptive role and with whom the participant took the insertive role. In addition, anal intercourse without a condom with the most recent partner was a main outcome.

Partner Variables To determine partner status, participants were asked if they currently had a main partner. A main partner was defined as a person with whom the participant had “an ongoing intimate sexual and emotional relationship.” The value 1 was given to those with a current main partner, 0 to those without. In order to characterize the dyadic characteristics of the most recent sexual encounter, participants were also asked the serostatus of the most recent partner, as well as whether that partner was their main partner. Two dichotomous variables were created: most recent partner was main partner (vs. not) and dyad was seroconcordant (both positive vs. discordant or unknown concordance).

Latino Acculturation Latino acculturation reflects the degree to which participants have retained or internalized Latino culture, regardless of whether they were born in Latin America or the US. Latino acculturation was assessed with 11 items adapted from a scale developed by Zea et al. (2003), which measures three aspects of acculturation: language competence (e.g., “How well do you speak Spanish on the phone?”), cultural identification (e.g., “I feel that I am part of culture”), and cultural knowledge (e.g., “How well history?”). Audio-CASI prodo you know gramming enabled us to substitute the name of the appropriate country of origin for each participant in the blanks indicated above. For example, a participant from Colombia would be asked, “How well do you know Colombian history?” The response format was a 4-point Likert scale, with higher scores indicating greater Latino acculturation. The Cronbach’s alpha for this sample was .77.

Depression The short form of the Beck Depression Inventory (Beck and Beck, 1972) was administered as the measure of depression. This 13-item scale has been

384

Poppen, Reisen, Zea, Bianchi, and Echeverry

used with Spanish-speaking populations (Zea et al., 1996, 1999). The measure had an internal consistency of .89 in the current study. Higher scores indicated greater depression, and a score of 16 represents the criterion for clinical depression. Substance Use Two separate questions asked how often the respondent had sexual relations under the influence of alcohol or of drugs in the past 30 days. These questions were combined into a single dichotomous variable indicating whether the respondent had sex under the influence of either alcohol or drugs at least once during this period (coded as 1), versus not at all (coded as 0). This variable is a measure of alcohol and drug use in conjunction with sex, but it is not associated with a particular episode between partners. Therefore, it provides information about the individual behavior but not about any specific dyadic encounter.

RESULTS Descriptive Statistics

tire sample of 155 men, which includes those who did not have any sex in the last 12 months. It is noteworthy that the distribution of number of partners is highly skewed, and therefore, the means are considerably higher than the medians. With the most recent partner, about a third have had unprotected anal intercourse. Unprotected sex was much more likely, however, when there was seroconcordance. When the partner was HIV-negative or of unknown serostatus, over three-quarters of the participants reported always having protected sex with that partner. In contrast, when both partners were HIV-positive, slightly less than half reported always having protected sex. As can be seen in Table I, a substantial proportion (about two-fifths) of the total sample had a current main sexual partner. In general, the participants were not depressed; despite the challenge of having a chronic illness, less than 10% scored above the criterion score for clinical depression. About a fifth of all participants reported using alcohol or using drugs in conjunction with sex in the last 30 days. A substantial number (14%) reported having engaged in sex in order to obtain drugs or money since testing HIV-positive. Models Predicting Number of Unprotected Anal Sex Partners in the Last 12 Months

Over 85% of the sample reported having sex with a man at least once in the last 12 months and nearly half reported unprotected anal intercourse with a man (see Table I). In contrast, only 10% reported having had sex with a woman in the last 12 months, with about half of those reporting unprotected sex episodes. Table I is based on the en-

Analyses predicting number of male partners with whom participants reported having had unprotected anal intercourse were performed on the subsample of those who had sex during the preceding 12 months (N = 133). Table II shows separate regressions for the number of partners with whom

Table I. Reported Sexual Behaviors and Other Characteristics of HIV-Positive Latino MSM (N = 155)

Last 12 months with a man Any sex at all Unprotected anal insertive Unprotected anal receptive Last 12 months with a woman Any sex at all Unprotected vaginal Unprotected anal Last 30 days with a man Sex with alcohol Sex with drugs Has a main partner now Clinical depression

Engaged in behavior

Average number of partners

Median

Range of partners

85.8% 48.4% 42.6%

18.4 5.2 4.0

4.0 0.0 0.0

0–200 0–120 0–80

10.3% 5.2% 3.2%

0.6 0.1 0.2

0.0 0.0 0.0

0–40 0–6 0–9

18.1% 12.9% 42.6% 7.7%

Unprotected Anal Sex and HIV-Positive Latino Men Table II. Variables Predicting Number of Unprotected Anal Sex Partners, Last 12 Months (N = 132) Number of unprotected partners with whom participant took the insertive role Overall model F (6, 126) = 2.53, p < .05, R2 = 10.7 Intercept Education level Age Latino acculturation Depression Has current main partner Used substance during sex Number of unprotected partners with whom the participant took the receptive role Overall model F (6, 126) = 3.33, p < .005, R2 = 13.7 Intercept Education level Age Latino acculturation Depression Has current main partner Used substance during sex

Parameter

t-value

−.680 .038 .003 −.113 174 −.088 .160

−1.90 1.20 0.92 −1.34 2.34∗ −1.28 2.13∗

−.622 −.006 .008 −.164 .138 .014 .161

−1.79 −0.21 2.26∗ −2.01∗ 1.91 −0.20 2.21∗

Note. Signs of coefficients reflect the direction of the relationships in the raw, untransformed data. ∗ p < .05.

the participant reported taking the insertive role and taking the receptive role for unprotected sex. Predictors included education level, age, Latino acculturation, depression, whether participant currently had a main partner, and whether participant reported using alcohol or drugs during sex in the last 30 days. For the most part, predictor variables were uncorrelated with one another; the only exceptions were that older participants were less likely to have main partners and were lower in Latino acculturation, and that those with greater education were less depressed. The strength of these relationships was small, with all Pearson correlation coefficients less than .20. Because the distributions of the outcome variables of number of insertive and receptive partners were heavily, positively skewed (see Table I), an inverse transformation was used. The skewness approached zero for the resulting transformed variables. The kurtosis of the transformed variables was greatly improved, although the concentration of participants with 0 or 1 partner made it impossible to correct the kurtotis completely. The overall models predicting number of unprotected male partners were significant, both for partners with whom the participant took the in-

385 sertive role (F (6, 126) = 2.53, p < .05) and for partners with whom the participant took the receptive role (F (6, 126) = 3.33, p < .01). Participants reported both more insertive and more receptive partners for unprotected sex if they had sex under the influence of alcohol or drugs. Participants with higher levels of depression reported having more unprotected partners with whom they took the insertive role, whereas those with lower levels of Latino acculturation reported having more unprotected partners with whom they took the receptive role. In addition, older men had more partners with whom they were receptive than younger men. In the analyses mentioned above, we combined alcohol and drug use in a single variable. We made this decision because both types of substances can affect level of inhibition and judgment. Moreover, the relatively low proportion of participants who reported using substances in conjunction with sex provided another reason to combine variables. In order to examine the contribution of each variable, however, we ran follow-up analyses of the models separately for drugs and for alcohol. Results indicated that the significant effect of the combined variable was principally, but not solely, due to the influence of drug use.

Models Predicting Unprotected Anal Intercourse With Most Recent Partner Hierarchical set logistic regressions were performed to test whether dyadic variables added significantly to the prediction of whether the participant had ever had unprotected anal intercourse with his most recent male partner, beyond the explanatory power of the individual-level variables (see Table III). In the first step, the individual characteristic variables of education, age, Latino acculturation, depression, and alcohol or drug use during sex were entered: the model was not significant. The second step involved the entry of the dyadic characteristic variables, reflecting the relationship between partners (i.e., Was this partner the participant’s main partner?) and seroconcordance (i.e., Was this partner, like the participant, HIV-positive?). The overall model was significant (χ2 (7) = 17.27, p < .05). In addition, the change in the log likelihood of the model due to the addition of the dyadic variables was also significant (χ2 (2) = 20.12, p < .001), indicating the importance of dyadic variables in predicting behavior with the most recent partner. Of the

386

Poppen, Reisen, Zea, Bianchi, and Echeverry

Table III. Logistic Set Regression Predicting Unprotected Anal Sex With Most Recent Partner (N = 132) Model

−2 log L

Overall model

Set 1: Individual variables Set 2: Inclusion of dyadic variables Final Model Intercept Set 1: Individual characteristic variables Education level Age Latino acculturation Depression Used substance during sex Set 2: Dyadic characteristic variables This is main partner Partner is HIV+

−137.57 −127.51

χ2 (5) = 7.21. χ2 (7) = 17.27∗

∗p

Parameter Wald Chi-square −.651 .08

.066 −.047 .253 −.036 −.949

.11 3.01 .24 .00 3.97∗

.281 −1.228

.33 6.79∗∗

< .05. ∗∗ p < .01.

predictor variables, seroconcordance between partners was significant. When the partner was also HIVpositive, the probability of unprotected anal sex was greater. In addition, although substance use during sex was not significant in the first step, it achieved significance after the dyadic variables were included.

DISCUSSION Most of this sample of HIV-positive Latino gay and bisexual men reported having sex in the past year, with nearly half indicating they had unprotected anal intercourse. Although these results are not derived from a representative sample, they are similar to other reports from Latino samples (e.g., ´ D´ıaz, 1998; Lye Chng and Geliga-Vargas, 2000). Because our sample included only HIV-positive men, who could potentially infect their partners, these relatively high rates of unprotected anal intercourse are noteworthy. While the median number of partners for unprotected sex during the past year was small, some individuals reported having many partners, with the maximum estimate of 120. Despite the high proportion of men who had unprotected anal intercourse, there were differences in behavior with seropositive and seronegative partners. Unprotected anal sex was much more common when there was seroconcordance, which suggests that many HIV-positive men are selective about partners with whom they have unprotected sex. By choosing to have sex only with those who are also positive,

these men are demonstrating a sense of responsibility about not endangering others who are uninfected. Our findings indicated that those participants who reported using drugs or alcohol in the context of their sexual encounters had higher numbers of both insertive and receptive partners for unprotected sex. This effect was principally due to the influence of drug use. Although there are conflicting findings in the literature on the role of substance use in sexual risk behavior, our results are consistent with others that have indicated an association between alcohol or drug use and sex risk (Hays et al., 1997b; Stall et al., 2000). Alcohol or drug use can lead to disinhibition, which can erode intentions to use condoms. Moreover, substance use is often associated with sensation seeking, and those individuals who are drawn to maximizing sensation tend to object to condoms (Dolezal et al., 1997). On the other hand, there could be a confound between this substance use variable and frequency of sex: Those who have a lot of sex would have increased opportunity both to use substances on at least one occasion and to have more unprotected partners. Future research could clarify this issue. More depressed mood was also associated with a greater number of partners for unprotected anal sex. Again, the research literature reports mixed findings (Crepaz and Marks, 2001), but our results support a link between depression and sex risk behavior. Among HIV-negative men, those who are depressed may feel devalued or unworthy and, therefore, be less inclined to use condoms to protect themselves (D´ıaz, 1999). In this sample of HIV-positive men, however, the self-protective motivation is less relevant; rather, men may seek sex as a way of easing emotional pain and coping with depression. Another explanation could be that depression leads to feelings of disconnection from other people, which could then reduce the motivation to protect potential partners from infection. Participants who had retained or internalized Latino culture to a greater degree reported having more partners with whom they took the receptive role for unprotected anal sex, but this effect did not reach significance in the prediction of number of partners with whom they took the insertive role. In traditional Latino culture there is a stigma about being in the receptive position, which is seen as a female role signifying a loss of masculinity (Parker, 1996), although in more contemporary culture, this attitude is ´ less pervasive (Carballo-Dieguez et al., 2004). Therefore, men whose attitudes are more influenced by traditional Latino culture would be less willing to take,

Unprotected Anal Sex and HIV-Positive Latino Men or admit to taking, the receptive role with a larger number of partners. This finding highlights the importance of cultural factors in sexual behavior. Older men also reported having more partners with whom they took the receptive role than did younger men. Because transmission of HIV is less likely when the infected partner is in the receptive role, some HIV-positive men choose to take this position as a means to limit risk to their partners (Parsons et al., 2003). It is possible that older men employ this strategy more than younger men, and that this behavioral approach to risk reduction is responsible for the observed relationship. We examined the question of the role of the dyadic context in sexual risk behavior. Although many models of sexual risk are based solely on individual characteristics, we argued that factors describing the context in which the sexual encounter occurs should also be addressed in order to account for the variety of behaviors that a person may display in different situations. The set of dyadic variables added explanatory power to the prediction of whether unprotected anal sex had ever occurred with the most recent partner. The only significant predictors were seroconcordance and substance use during sex. Unprotected anal intercourse was much more likely when the partners were seroconcordant, which in this sample meant that both were HIVpositive, and if substances were used in conjunction with sex. The importance of seroconcordance in determining sexual behavior is consistent with previous research showing that unprotected anal intercourse is most likely to occur between serocondordant partners, with higher rates for seropositive men than seronegative men (Hays et al., 1997b). Many MSM use information about the serostatus of both partners to determine whether they can engage in unprotected sex without posing a risk to themselves or their partners (Kippax et al., 1993). Although previous research has shown fairly consistently that unprotected anal intercourse is more likely to occur between main or primary partners than between casual partners, in our HIVpositive sample, relationship type was not significant. We do not believe that relationship type is unimportant; however, for HIV-positive men the serostatus of the partner is more relevant in determining sexual behavior, because of implications of risk to the partner. In addition, there was some overlap between the two variables in our analyses, as more than half of the main partners were also HIV-positive.

387 After controlling for the dyadic variables, substance use during sex also was predictive of unprotected anal intercourse with the most recent partner. Before the dyadic variables were included in the model, substance use approached significance (p < .06), but did not achieve it. The change from nonsignificance to significance was presumably due to the fact that the inclusion of the dyadic factors explained additional variance, and thus resulted in a smaller error term. The substance use variable summarized behavior over the previous 30 days and did not specifically refer to sex with the most recent partner; therefore, we classified it as an individual difference variable. One might argue, however, that this variable also captured information about the dyadic context. The responses of those participants who indicated that they had not used alcohol or drugs at any time with sex would also apply to behavior with the most recent partner, and therefore could be seen as dyadic. However, for those who reported they had used substances during sex, we cannot tell whether they were referring to behavior with this partner or with others. An improvement to the current study would be a more precise measure of substance use associated with specific sexual partners or encounters. It is important to note that depression was associated with the number of partners for unprotected anal sex, but not with unprotected anal intercourse with the most recent partner. To the extent our measure of depression reflects a dispositional characteristic, we can interpret this discrepancy as illustrative of the need to distinguish between individual-level and dyadic-level influences and outcomes. In this case, the outcome variable measuring number of partners summarizes a person’s behavior for a year. It is not surprising that it can be predicted from characteristics of the individual, such as depression. In contrast, we expected that the prediction of behavior with a specific partner would be influenced to a greater extent by the dyadic context, including aspects of the relationship between partners or seroconcordance. Our results with the outcome variable measuring unprotected anal intercourse with the most recent partner are consistent with this expectation. The failure of depression to predict this outcome is thus explicable. These findings suggest that dyadic variables such as seroconcordance are of critical importance in the prediction of sexual behavior with a specific partner for HIV-positive men. The characterization of the dyad in this study was limited and did not include

388 features such as duration of relationship and level of intimacy. Future research and interventions should address the impact of other dyadic dimensions on unprotected anal sex, and theoretical models should be developed to account for the role of such factors. Moreover, the possible moderation of individual characteristics by dyadic factors should also be considered. Research examining sexual behavior, as well as interventions promoting safer sex practices, should address differences in motivations of HIVpositive and HIV-negative men. ACKNOWLEDGMENTS This study was supported by a grant from the National Institutes of Health (NIMH 1 R01 MH 60545). REFERENCES Bajos, N., and Marquet, J. (2000). Research on HIV sexual risk: Social relations-based approach in a cross-cultural perspective. Social Science and Medicine, 50, 1533–1546. Beck, A. T., and Beck, R. W. (1972). Screening depressed patients in family practice: A rapid technique. Postgraduate Medicine, 52, 81–85. Becker, M. H., and Joseph, J. G. (1988). AIDS and behavioral change to reduce risk: A review. American Journal of Public Health, 78, 394–410. Bingham, T. A., Harawa, N. T., Johnson, D. F., Secura, G. M., MacKellar, D. A., and Valleroy, L. A. (2003). The effect of partner characteristics on HIV infection among African American men who have sex with men in the Young Men’s Survey, Los Angeles, 1999–2000. AIDS Education and Prevention, 15(Suppl. A), 39–52. Brislin, R. W. (1986). The wording and translation of research instruments. In W. J. Lonner and J. W. Berry (Eds.), Field methods in cross-cultural psychology (pp. 137–164). Beverly Hills, CA: Sage. Buchanan, D. R., Poppen, P. J., and Reisen, C. A. (1996). The nature of partner relationship and AIDS sexual risk-taking in gay men. Psychology and Health, 11, 541–555. ´ Carballo-Dieguez, A., Remien, R. H., Dolezal, C., and Wagner, G. (1997). Unsafe sex in the primary relationships of Puerto Rican men who have sex with men. AIDS and Behavior, 1, 9–17. ´ Carballo-Dieguez, A., Dolezal, C., Nieves, L., D´ıaz, F., Decena, C., and Balan, I. (2004). Looking for a tall, dark, macho man. . . sexual-role behaviour variations in Latino gay and bisexual men. Culture, Health, and Sexuality, 6, 159–171. Carlson R. G., Wang, J., Siegal, H. A., Falck, R. S., and Guo, J. (1994). An ethnographic approach to targeted sampling: Problems and solutions in AIDS prevention research among drug and crack-cocaine users. Human Organization, 53, 279– 286. Catania, J. A., Kegeles, S. M., and Coates, T. J. (1990). Towards an understanding of risk behavior: An AIDS risk reduction model (ARRM). Health Education Quarterly, 17, 53–72. CDC. (2000). US HIV and AIDS cases reported through December, 1999. Year end edition. HIV/AIDS surveillance report, 11. Atlanta, GA: Centers for Disease Control and Prevention.

Poppen, Reisen, Zea, Bianchi, and Echeverry CDC. (2002). US HIV and AIDS cases reported through December, 2001. Year end edition. HIV/AIDS surveillance report, 13. Atlanta, GA: Centers for Disease Control and Prevention. Corbin, W. R., and Fromme, K. (2002). Alcohol use and serial monogamy as risks for sexually transmitted diseases in young adults. Health Psychology, 21, 229–236. Crepaz, N., and Marks, G. (2001). Are negative affective states associated with HIV sexual risk behaviors? A meta-analytic review. Health Psychology, 20, 291–299. Davidovich, U., deWit, J., and Stroebe, W. (2000). Assessing sexual risk behaviour of young gay men in primary relationships: The incorporation of negotiated safety and negotiated safety compliance. AIDS, 14, 701–706. DeRosa, C. R., and Marks, G. (1998). Preventive counseling of HIV-positive men and self-disclosure of serostatus to sex partners: New opportunities for prevention. Health Psychology, 17, 224–231. D´ıaz, R. M. (1998). Latino gay men and HIV: Culture, sexuality, and risk behavior. New York: Routledge. D´ıaz, R. M. (1999). Trip to Fantasy Island. Sexualities, 2, 89–112. Dolezal, C., Meyer-Bahlburg, H. F. L., Remien, R. H., and Petkova, E. (1997). Substance use during sex and sensation as predictors of sexual risk behavior among HIV+ and HIV− gay men. AIDS and Behavior, 1, 19–28. ´ Dolezal, C., Carballo-Dieguez, A., Nieves-Rosa, L., and D´ıaz, F. (2000). Substance use and sexual risk behavior: Understanding their association among four ethnic groups of Latino men who have sex with men. Journal of Substance Abuse, 11, 323– 336. Ekstrand, M. L., Stall, R. D., Paul, J. P., Osmond, D. H., and Coates, T. J. (1999). Gay men report high rates of unprotected sex with partners of unknown or discordant HIV status. AIDS, 13, 1525–1533. Elford, J., Bolding, G., Maguire, M., and Sherr, L. (2001). Gay men, risk and relationships. AIDS, 15, 1053–1055. Fisher, J. D., and Fisher, W. A. (2000). Theoretical approaches to individual-level change in HIV risk behavior. In J. L. Peterson and R. J. DiClemente (Eds.), Handbook of HIV prevention (pp. 3–48). New York: Kluwer Academic/Plenum Publishers. Flowers, P., Sheeran, P., Beail, N., and Smith, J. A. (1997). The role of psychosocial factors in HIV risk-reduction among gay and bisexual men: A quantitative review. Psychology and Health, 12, 197–230. Frijida, N. H. (1988). The laws of emotion. American Psychologist, 43, 349–358. Hays, R. B., Kegeles, S. M., and Coates, T. J. (1997a). Unprotected sex and HIV risk-taking among young gay men within boyfriend relationships. AIDS Education and Prevention, 9, 314–329. Hays, R. B., Paul, J., Ekstrand, M., Kegeles, S. M., Stall, R., and Coates, T. J. (1997b). Actual versus perceived HIV status, sexual behaviors and predictors of unprotected sex among young gay and bisexual men who identify as HIV-negative, HIV-positive and untested. AIDS, 11, 1495–1502. Janssen, M., DeWit, J., Stroebe, W., and VanGriensven, F. (2000). Educational status and risk of HIV in young gay men. Journal of Health Psychology, 5, 487–499. Kalichman, S. (2000). HIV transmission risk behaviors of men and women living with HIV-AIDS: Prevalence, predictors, and emerging clinical interventions. Clinical Psychology: Science and Practice, 7, 32–47. Kalichman, S. C., Nachimson, D., Cherry, C., and Williams, E. (1998). AIDS treatment advances and behavioral prevention setbacks: Preliminary assessment of reduced perceived threat of HIV-AIDS. Health Psychology, 17, 546–550. Kelly, J. A., and Kalichman, S. C. (1998). Reinforcement value of unsafe sex as a predictor of condom use and continued HIV/AIDS risk behavior among gay and bisexual men. Health Psychology, 17, 328–335.

Unprotected Anal Sex and HIV-Positive Latino Men Kippax, S., Crawford, J., Davis, M., Rodden, P., and Dowsett, G. (1993). Sustaining safe sex: A longitudinal sample of homosexual men. AIDS, 7, 257–263. Koblin, B. A., Chesney, M. A., Husnik, M. J., Bozeman, S., Celum, C. L., Buchbinder, S., Mayer, K., McKirnan, D., Judson, F. N., Hunag, Y., Coates, T. J., and the EXPLORE Study Team. (2003). High-risk behaviors among men who have sex with men in 6 US cities: Baseline data from the EXPLORE study. American Journal of Public Health, 93, 926–932. Leigh, B., and Stall, R. (1993). Substance use and risky sexual behavior for exposure to HIV: Issues in methodology, interpretation, and prevention. American Psychologist, 48, 1035– 1045. Lye Chng, C., and Geliga-Vargas, J. (2000). Ethnic identity, gay identity, sexual sensation seeking and HIV risk among multiethnic men who have sex with men. AIDS Education and Prevention, 12, 326–339. Mar´ın, G., and Mar´ın, B. V. (1991). Research with Hispanic populations. Applied social research methods series (Vol. 23). Newbury Park, CA: Sage. ´ ´ Mar´ın, B. V., Gonzalez, F. J., and Gomez, C. A. (1998). Unmarried Latino men who report sex with men: Psychocultural and demographic characteristics. AIDS and Behavior, 2, 203– 211. Marks, G., Richardson, J. L., and Maldonado, N. (1991). Selfdisclosure of HIV infection to sexual partners. American Journal of Public Health, 81, 1321–1323. Marks, G., Bingham, C. R., and Duval, T. S. (1998a). Negative affect and unsafe sex in HIV-positive men. AIDS and Behavior, 2, 89–99. Marks, G., Cantero, P. J., and Simoni, J. M. (1998b). Is acculturation associated with sexual risk behaviours? An investigation of HIV-positive Latino men and women. AIDS Care, 10, 283– 295. Marks, G., Burris, S., and Peterson, T. A. (1999). Reducing sexual transmission of HIV from those who know they are infected: The need for personal and collective responsibility. AIDS, 13, 297–306. Mason, H. R. C., Marks, G., Simoni, J. M., and Ruiz, M. S. (1995). Culturally sanctioned secrets? Latino men’s nondisclosure of HIV infection to family, friends, and lovers. Health Psychology, 14, 6–12. McAuliffe, T. L., Kelly, J. A., Sikkema, K. J., Murphy, D. A., Winett, R. A., Solomon, L. J., and Roffman, R. A. (1999). Sexual HIV risk behavior levels among young and older gay men outside of AIDS epicenters: Findings of a 16-city sample. AIDS and Behavior, 3, 111–119. Myers, T., Allman, D., Calzavara, L., Morrison, K., Marchand, R., and Major, C. (1999). Gay and bisexual men’s sexual partnerships and variations in risk behaviour. The Canadian Journal of Human Sexuality, 8, 115–126. Ostrow, D. G., McKirnan, D., Klein, C., and DiFranceisco, W. (1999). Patterns and correlates of risky behavior among HIV+ gay men: Are they really different from HIV− men? AIDS and Behavior, 3, 99–110. Parker, R. G. (1996). Behaviour in Latin American men: Implications for HIV/AIDS interventions. International Journal of STD and AIDS, 7, 62–65.

389 ´ Parsons, J. T., Halkitis, P. N., Wolitski, R. J., Gomez, C. A., and the Seropositive Urban Men’s Study Team. (2003). Correlates of sexual risk behaviors among HIV-positive men who have sex with men. AIDS Education and Prevention, 15, 383–400. Pinkerton, S. D., and Abramson, P. R. (1997). Effectiveness of condoms in preventing HIV transmission. Social Science and Medicine, 44, 1303–1312. ´ Remien, R. H., Carballo-Dieguez, A., and Wagner, G. (1995). Intimacy and sexual behavior in serodiscordant male couples. AIDS Care, 7, 429–438. Ross, M. W., Rosser, B. R. S., Bauer, G. R., Bockting, W. O., Robinson, B. E., Rugg, D, L., and Coleman, E. (2001). Drug use, unsafe sexual behavior, and internalized homonegativity in men who have sex with men. AIDS and Behavior, 5, 97– 103. San Doval, A., Duran, R., O’Donnell, L., and O’Donnell, C. R. (1995). Barriers to condom use in primary and nonprimary relationships among Hispanic STD clinic patients. Hispanic Journal of Behavioral Sciences, 17, 385–397. Stall, R., and Purcell, D. W. (2000). Intertwining epidemics: A review of research on substance use among men who have sex with men and its connection to the AIDS epidemic. AIDS and Behavior, 4, 181–192. Stall, R. D., McKusick, L., Wiley, J., Coates, T. J., et al. (1986). Alcohol and drug use during sexual activity and compliance with safe sex guidelines for AIDS: The AIDS Behavioral Research Project. Health Education Quarterly, 13, 359–371. Stall, R. D., Hays, R. B., Waldo, C. R., Ekstrand, M., and McFarland, W. (2000). The gay ’90s: A review of research in the 1990s on sexual behavior and HIV risk among men who have sex with men. AIDS, 14(Suppl. 3), S1–S14. Strathdee, S. A., Hogg, R. S., Martindale, S. L., Cornelisses, P. G. A., Craib, K. J. P., Montaner, J. S. G., O’Shaughnessy, M. V., and Schechter, M. T. (1998). Determinants of sexual risk-taking among young HIV-negative gay and bisexual men. Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology, 19, 61–66. Watters, J. K., and Biernacki, P. (1989). Targeted sampling: Options for the study of hidden populations. Social Problems, 36, 416–430. Wenger, N., Kusseling, F., Beck, K., and Shapiro, M. (1994). Sexual behavior of individuals infected with human immunodeficiency virus. Archives of Internal Medicine, 154, 1849–1854. Zea, M. C., Belgrave, F. Z., Townsend, T. G., Jarama, S. L., and Banks, S. R. (1996). The influence of social support and active coping on depression among African Americans and Latinos with disabilities. Rehabilitation Psychology, 41, 223–240. Zea, M. C., Reisen, C. A., and Poppen, P. J. (1999). Psychological well-being among Latino lesbians and gay men. Cultural Diversity and Ethnic Minority Psychology, 5, 371–379. Zea, M. C., Asner-Self, K., Birman, D., and Buki , L. (2003). The Abbreviated Multidimensional Acculturation Scale: Empirical validation with two Latino/Latina samples. Cultural Diversity and Ethnic Minority Psychology, 9, 107–126. Zea, M. C., Reisen, C. A., Poppen, P. J., Echeverry, J. E., and Bianchi, F. T. (2004). Disclosure of HIV-positive status to Latino gay men’s social networks. American Journal of Community Psychology, 33, 107–116.