AEGiS-NHIVPC: Behavioral Risk Clusters and HIV Seroincidence among Vaccine Efficacy Trial Participants: Implications for HIV Vaccine Efficacy Trials and Prevention Programs

National HIV Prevention Conference

Atlanta, Georgia, USA — July 27 - 30, 2003

Behavioral Risk Clusters and HIV Seroincidence among Vaccine Efficacy Trial Participants: Implications for HIV Vaccine Efficacy Trials and Prevention Programs

Natl HIV Prev Conf 2003 July 27-30:abstract no. M1-C0204
Bartholow BG, GOLI V, Ackers M, McLellan E, Durham M, Popovic V
Centers for Disease Control and Prevention, Atlanta, GA

BACKGROUND: HIV vaccine efficacy trials require the participation of many individuals at risk of HIV to determine HIV vaccine efficacy in a timely and cost-effective manner. Understanding risk behavior and infection dynamics within the context of efficacy trials will help inform the design, enrollment, and analyses of future trials. Furthermore, the risk behavior observed in large-scale HIV vaccine efficacy trials may have implications for HIV prevention programs.

METHODS: 5,103 HIV seronegative men who have sex with men (MSM) participating in the phase III efficacy trial of AIDSVAX B/B were interviewed about their HIV risk behavior at baseline and semiannually over 36 months. HIV testing was conducted at semiannual visits using EIA with positive results confirmed by immunoblot. Two step cluster analyses were conducted using demographic variables, numbers of HIV+, HIV-, and HIV unknown status partners, proportion of partners by antibody status, and drug use including poppers, amphetamines, hallucinogens, and Viagra. Sexual risk behaviors among emergent risk clusters were compared over time using generalized estimating equations, and HIV incidence rates and confidence intervals were calculated for each risk cluster.

RESULTS: Eight risk clusters emerged that varied with regard to HIV risk behavior and annualized HIV incidence (range 1.8%-6.6%). Cluster characteristics were as follows:

Cluster Number	% Total	Cluster Description	Race	Age†	Drug Use	Number of Partners‡	Majority HIV Status of Partner	HIV Incidence (95% CI)

1	19	white non-drug users	White	younger	none	high	HIV-	2.2 (1.7-2.7)

8	18	older low-risk	White	older	none	low	HIV uk	1.9 (1.4-2.4)

7	15	older popper users	white	older	100% poppers	moderate	HIV uk	3.6 (2.9-4.3)

3	14	young party crowd	mixed	younger	>80% party drugs	high	HIV uk	4.9 (3.9-5.5)

6	12	primary discordant relationships	mixed, no Hispanic	older	low	low	HIV+	1.8 (1.3-2.4)

5	11	men of color	non-white	younger	low	low	HIV uk	3.0 (2.3-3.8)

4	8	older Viagra users	mixed	older	100% Viagra	moderate	HIV uk	2.6 (1.7-3.4)

2	3	older high risk	mixed	older	>50% poppers, 25% party drugs	high	HIV uk	6.6 (4.6-8.5)

†Younger/older age defined by median split as: younger =36 years, older >36 ‡ low =25th percentile, moderate 26-74th percentile, high=75th percentile

CONCLUSIONS: Significant variation in HIV risk behavior and incidence was observed among risk clusters. Trial enrollment criteria based on these cluster analysis results could reduce enrollment of lower risk participants and maximize HIV incidence, increse statistical power, and reduce cost of conducting future HIV vaccine trials. Furthermore, the observed risk clusters suggest that different intervention approaches should be considered for these heterogeneous clusters of MSM at-risk of HIV infection.

030727
M1-C0204