Predictors of HIV disclosure to secondary partners and sexual risk behavior among a high-risk sample of HIV-positive MSM

Abstract

This study sought to identify predictors of HIV disclosure and serodiscordant unprotected anal intercourse (SDUAI) among HIV-positive men who have sex with men (MSM). Between January 2005 and April 2006, 675 HIV-positive MSM were recruited into the Positive Connections intervention trial held in six US cities with intentional over-sampling of HIV-positive MSM of Color (74%) and men engaging in unprotected anal intercourse (UAI) in the previous year. Baseline survey data showed 30 and 31%, respectively, of participants disclosed to none or some of their secondary sex partners in the last 90 days. Greater disclosure to secondary partners was associated with having fewer sexual partners, being extremely out as MSM, longer HIV diagnosis, knowledge of CD4 count, detectable viral load and being white. Disclosure to all secondary partners was associated with lower SDUAI. Recommendations for prevention for HIV-positive MSM include the promotion of serodisclosure to all secondary partners and increasing comfort with, and outness about, one’s sexuality.

Introduction

The Centers for Disease Control and Prevention (CDC) has prioritized interventions to reduce sexual risk taking among HIV-positive persons (Janssen & Valdiserri, 2004). Disclosure of HIV-positive status to sexual partners may reduce risk of transmission in serodiscordant partnerships (Parsons et al., 2005). However, the association between HIV serodisclosure and sexual risk behavior is complex, varying by person-, partner- and situation-level factors (Simoni & Pantalone, 2004). Elucidating the relationship between disclosure and risk remains a critical health priority.

Studies show high rates of disclosure among MSM with their primary partners (79–89%) but not with casual or secondary partners (34–41%) (Hart et al., 2005; Wolitski, Parsons, & Gomez, 2004; Wolitski, Rietmeijer, Goldbaum, & Wilson, 1998). However, the association between disclosure to secondary partners and HIV risk is unclear, as some studies report a positive association (Wolitski et al., 1998) while others show no relationship (Crepaz & Marks, 2003; Marks & Crepaz, 2001). More understanding of serodisclosure strategy is necessary, as several studies have shown that inconsistently serodisclosing to sexual partners is associated with higher risk than consistently disclosing or not disclosing (Hart et al., 2005; Parsons et al., 2005).

Factors associated with higher serodisclosure include disclosure self-efficacy and intention to disclose (Hart et al., 2005) and reporting fewer sex partners (Wolitski et al., 2004). The degree of being “out” about one’s sexual orientation has been associated with improved mental health adjustment (Rosser, Bockting, Ross, Miner, & Coleman, 2008) and may serve as behavioral rehearsal for coming out as HIV-positive. Finally, changes in disease status (CD4 count, viral load, AIDS diagnosis) may impact serodisclosure by modifying beliefs of transmissibility and perceptions of responsibility (Hart et al., 2005; Kalichman & Nachimson, 1999).

As part of a larger study to test new interventions for HIV-positive MSM, the aim of this paper was to examine the complex relationship among disclosure to secondary partners, demographic factors, partner type, disease markers, outness and sexual risk among a racially and ethnically diverse sample of HIV-positive MSM.

Methods

Participants

Six-hundred-and-seventy-five HIV-positive MSM were recruited in six cities: Seattle (n = 114; 17%), Washington, DC (n = 71; 11%), Boston (n = 64; 9%), New York (n = 177; 26%), Los Angeles (n = 146; 22%) and Houston (n = 103; 15%). Inclusion in the study required participants to: be male, be 18 years or older, self-identify as HIV-positive, report at least one occasion of unprotected anal intercourse (UAI) with a man in the past year and to be English speaking.

Procedures

Study procedures were approved by the University of Minnesota Institutional Review Board and community-based Institutional Review Boards. Participants were recruited and telephone screened for eligibility by partner AIDS service organizations (ASOs). On the first day of the trial, participants completed a baseline pencil-and-paper survey. These self-reported data were not validated further. Participants received $200 compensation for completing the entire study.

Measures

Sociodemographic variables and psychosocial scales are shown in Table 1 and the Appendix, respectively. The two primary outcomes were serodisclosure and serodiscordant UAI (SDUAI). Participants reported their sexual behavior with secondary partners of HIV-positive, HIV-negative or unknown serostatus (as defined by the participant) in the last three months and to how many total partners they had disclosed their HIV-positive serostatus. In response to the bimodal distribution of these data, we created a 3-level variable indicating disclosure to none (0%), some (1–99% of partners) or all (100%) secondary partners. Serodisclosure and SDUAI was defined as UAI with a male partner of negative or unknown HIV status (Golden, Brewer, Kurth, Holmes, & Handsfield, 2004; Rawstorne et al., 2007). After summing SDUAI during the last three months, we dichotomized this count into no SDUAI (0) or at least one incident of SDUAI (1).

Table 1.

Participant demographics (n = 675).

	Median	IQR
Age	42	37–48
Annual income	$10,080	$6,000–$22,000
Year of diagnosis	1995	1990–2000
CD4 count	428	259–618
Viral load	2800	170–22,000
	n	%
Race/ethnicity
African American	300	45
Caucasian	168	25
Hispanic	157	23
Other	44	7
Education
HS or less	268	40
Some college	253	38
Bachelor’s degree or more	152	23
Sexual orientation
Gay	528	80
Bisexual	110	17
Hetero or other	22	3
Taking antiretroviral medication	492	75

Analysis

Serodisclosure to none, some or all secondary partners was modeled using ordinal logistic regression, assuming proportional odds, onto single predictors. Significance was determined by model likelihood ratio (LR) tests with an alpha threshold of .05. A full proportional odds logistic regression model of serodisclosure comprised variables that were significant in single-predictor models. Akaike’s Information Criterion (AIC) (Akaike, 1974) for backward variable selection on the full model with a threshold of 0 failed to remove any variables. Ordinary (binary) logistic regression was used to model SDUAI. Analyses were conducted using SAS 9.1 for Windows.

Results

Demographics are presented in Table 1. Participants were primarily men of color (n = 502; 75%), lower income and well-educated.

Serodisclosure and risk behavior

Most participants reported secondary sexual partners (n = 525); 320 reported primary relationships. As disclosure to primary partners was very common (88%), no further analyses were conducted on men without secondary partners. Disclosure to secondary partners varied; of 445 men reporting on disclosure to secondary partners, 132 (30%) reported disclosing to none, 138 (31%) to some and 175 (39%) to all. Unprotected anal intercourse in the last three months with secondary partners was common (n = 389; 74%). The proportion of participants engaging in UAI was similar with seroconcordant (n = 320; 72%) and serodiscordant (n = 271; 66%) secondary partners.

Modeling serodisclosure

Single predictor logistic regression was used to model disclosure to secondary partners. Greater disclosure was associated with white racial identity, outness as MSM, knowledge of CD4 count, detectable viral load, years since diagnosis (3rd order polynomial, LR = 7.95; df = 3; p <0.05) and number of partners (log-transformed, LR = 6.77; df = 1; p<0.01) (Table 2). Age, income, education, employment status, female partners, sexual orientation and recent diagnosis with a non-HIV STI were not significantly associated with serodisclosure.

Table 2.

Logistic regression model showing the adjusted relationship of demographic and health characteristics of HIV-positive MSM to the likelihood of HIV serodisclosure to secondary partners (n = 396).^a

Factor	Level	Single predictor models		Multivariate model
		Odds ratio	0.95 Confidence bounds	Odds ratio	0.95 Confidence bounds	AIC
Race/ethnicity	White (reference)					3.92
	Black	0.47	0.31, 0.72	0.5	0.31, 0.81
	Latino/Hispanic	0.53	0.33, 0.87	0.52	0.30, 0.91
	Other	0.44	0.21, 0.95	0.64	0.27, 1.51
CD4	‘Do not know’ (reference)					5.53
	Value reported	1.99	1.18, 3.37	2.09	1.05, 4.16
Viral load	Undetectable (reference)					11.85
	‘Do not know’	0.83	0.51, 1.38	1.37	0.73, 2.58
	Detectable	1.82	1.22, 2.70	1.85	1.21, 2.85
Out as MSM	Less out (reference)					32.63
	Extremely out	2.16	1.51, 3.09	2.11	1.40, 3.14

^aFull model also included number of secondary partners and years since diagnosis shown in Figure 1.

We created a full model of all significant variables from the simple models. Based on AIC, all variables in Table 2 and Figure 1 remained significant predictors of serodisclosure.

Figure 1.

Adjusted log odds likelihood of serodisclosure to secondary partners as a function of numbers of secondary partners and years since diagnosis.^aaFull model also included race/ethnicity, CD4 count, viral load and out as MSM shown in Table 2.

Serodisclosure and SDUAI

Serodisclosure to secondary partners was related to reduced risk behavior (adjusted for partner number). Compared to men who consistently disclosed, those disclosing to none (OR = 4.37; 95%CI: 2.64, 7.22) or some (OR = 4.79; 95%CI: 2.89, 7.95) of their secondary partners had significantly greater odds of SDUAI.

To adjust for other possible confounding variables, we constructed a full model. No factors related to serodisclosure (Table 2) were associated with SDUAI and thus were not included. Of eight psychosocial factors examined (altruism, condom self efficacy, compulsive sexual behavior, HIV-related behavioral intention, internalized homonegativity, depression and anxiety, social support, sexual comfort), only condom self-efficacy substantially modified the relationship between serodisclosure and SDUAI. Adjusting for condom self-efficacy revealed an even greater impact of serodisclosure on risk of SDUAI (Table 3).

Table 3.

Multivariable logistic regression model showing the adjusted relationship of serodisclosure to unprotected anal sex in serodiscordant partners (n = 388).

Factor	Level	Odds ratio	0.95 Confidence bounds
Serodisclosure to secondary partners	All (reference)
	Some	5.33	2.96, 9.60
	None	6.1	3.39, 10.97
Condom self-efficacy	Highest quartile (reference)
	3rd quartile	3.35	1.59, 7.05
	2nd quartile	4.27	2.10, 8.70
	Lowest quartile	7.29	3.50, 15.17
Number of secondary partners	2 partners (25th percentile, reference)
	6 partners (75th percentile)	2.27	1.62, 3.17

Discussion

The generalizability of these results is limited as a result of its cross-sectional design, recruitment of high-risk MSM from HIV epicenters and self-reported HIV status, serodisclosure and risk behaviors. In addition, the survey design precluded examination of variation in serodisclosure to secondary partners by partner serostatus, which limits a more comprehensive understanding of disclosure and risk dynamics.

Although roughly equal proportions of participants serodisclosed to all, some, or none of their secondary sex partners, only a strategy of consistent disclosure to all secondary partners was associated with lower sexual risk-taking. Our findings differ from two previous studies (Hart et al., 2005; Parsons et al., 2005) reporting inconsistent disclosure associated with highest risk. Although demographic (white race), psychosexual (“out” as MSM), behavioral (fewer numbers of sexual partners) and medical (knowing one’s CD4 count, detectable viral load, years since diagnosis) factors were associated with HIV serodisclosure to secondary partners, only serodisclosure and condom self-efficacy was associated with SDUAI. These findings suggest that HIV-prevention programs for HIV-positive men should focus on disclosure to all secondary partners.

A novel finding in the study was that men who reported being extremely “out” as MSM (43% of the sample) were 2.1 times more likely to serodisclose to all of their secondary partners. Being out about one’s sexuality may serve an important function of gaining greater social support overall for men in the study. In a prior study of Latino men, social isolation was associated with less communication about serostatus (Zea, Reisen, Poppen, & Diaz, 2003). Similarly, among young Black MSM, those who were not out about their sexual orientation were less likely than those out to have more male sexual partners in their lifetime, more UAI with male sexual partners and to have been diagnosed as HIV-positive (Crawford, Allison, Zamboni, & Soto, 2002). However, racial and ethnic differences in openness about sexual orientation may not extend beyond sexual relationships, as other studies found that Black HIV-positive MSM were either less likely (Simoni, Mason, & Marks, 1997) or equally likely (Crawford et al., 2002) as other HIV-positive MSM to disclose their sexual orientation to members of their support network.

In sum, the results of this study suggest that future intervention programs should encourage consistent serodisclosure to secondary sexual partners and promote outness and comfort with sexual orientation within a community-appropriate approach.

Appendix

Appendix. Psychosocial measures.

Scale	Source/adapted from	Description	Alpha
Sexual comfort	Marin, Gomez, Tschann, & Gregorich, 1997	Six-item, Likert-type scale measuring comfort with sexuality and one’s body from very uncomfortable (1) to very comfortable (4)	.84
Internalized  homonegativity	Bell & Weinberg, 1981	Four-item likert-types scale measuring participants’ acceptance of negative views about their own homosexuality. High scores indicate more internalized homonegativity.	.88
Mental health	Derogatis, Yevzeroff, & Wittelsberger, 1975	This scale consists of 13 items drawn from the Depression and Anxiety subscales of the Brief Symptom Checklist. Items measure depression symptoms and anxiety symptoms using a Likert-type format where high scores indicate more depression or anxiety.	.93
Social support	Zimet, Dahlem, Zimet, & Farley, 1988	Six-item Likert-type scale assessing the amount of support that participants receive from friends and a “special person”.	.91
Compulsive sexual  behavior inventory	Coleman, Miner, Ohlerking, & Raymond, 2001	Thirteen-item Control subscale from the CSBI (Coleman et al., 2001) and consists of items that measure a sense of lack of control over one’s sexual behavior. High scores indicate more lack of control over sexual behavior.	.92
Altruism	Nimmons, Acree, & Folkman, 2000	Seven-item Likert-type scale measuring the degree to which participants are willing to behave in a manner that is self-sacrificing in order to avoid spreading the HIV virus. High scores indicate more altruistic beliefs.	.91
Condom self-efficacy	Marin, Gomez, Tschann, & Gregorich, 1997	Fourteen-item Likert-type scale measuring Bandura’s (1981) concept of self-efficacy with respect to using condoms in multiple situations and settings. High scores indicate more self-efficacy.	.95
Social norms	Marin, Gomez, Tschann, & Gregorich, 1997	Three-item Likert-type scale measuring the attitudes of the participants’ friends to using condoms. High scores indicate attitudes that support using condoms in multiple contexts.	.77
HIV behavioral  intention	Items created for this study	Three-item Likert type scale measuring intent to disclose serostatus, discuss safer sex and refuse unsafe sex.
HIV disease	Rosser et al., 2002	Participants asked: “What was your most recent CD-4 count/viral load?” with response options being a quantifiable number, “I have never had a CD-4/viral load taken”, “don’t know”, and on the question for viral load, “undetectable”. Participants were asked CD-4 count and viral load separately.
Outness as MSM	Kinsey, Pomeroy, & Martin, 1948	Participants were asked “How open (out) are you as gay, bisexual or a man attracted to men?” with answer options ranging from 1 (not at all) to 7 (extremely).