Age and risk of HIV infection in men who have sex with men

Introduction HIV has been documented in populations of homosexual and bisexual men for more than 15 years. Despite significant changes in sexual risk behavior within the gay community, HIV continues to spread at a high rate in populations of men who have sex with men (MSM). According to the US Centers for Disease Control and Prevention 1, MSM account for 54% of the cumulative classified adult/adolescent AIDS cases in the United States through 1996, and 48% of incident cases in 1996. An additional 7% of cumulative and 5% of 1996 incident cases were attributed to a dual exposure category of MSM and injecting drug use. These figures indicate that sexual exposure among MSM continues to be a major source of HIV infections in the United States. The question of whether, and to what extent, young age is a risk factor for HIV infection in MSM has generated considerable interest and debate 2–6. Some studies point to a ‘young age risk effect’, whereas others do not. The investigations as a whole vary considerably in research designs, outcome measures, operationalizations of the age variable, and other methodological dimensions. Thus, different conclusions may be reached when different subsets of studies are considered. The absence of a comprehensive published review of the literature has stifled progress in this area. This article provides such a review and examines whether unprotected anal intercourse is more common and risk of HIV seroconversion higher in younger than older MSM. An examination of this question has theoretical as well as applied significance. An understanding of age-related risk for HIV infection is a step toward understanding the dynamics of the HIV/AIDS epidemic and gauging the possibility that incidence of infection in this population may escalate in the future. Unprotected anal intercourse is the primary mode of HIV transmission in MSM 7–9. If this risk behavior is more prevalent in new cohorts of young MSM than in older MSM, then waves of infection may persist and new types of tailored interventions may be needed to reach those at risk 6. The bulk of the empirical literature examining age as a risk factor for contracting HIV in MSM falls within two groups of studies: (i) the association of age and HIV seroconversion, and (ii) the association of age and unprotected anal intercourse. We review these two sets of investigations and then discuss potential behavioural, biobehavioural and psychosocial mediating mechanisms and other explanations of the findings. Methods Identification of articles Literature from 1982 through 1997 available in MEDLINE, AIDSLINE, and PsycLIT databases were searched using various keywords (HIV, AIDS, gay, homosexual, men, age, seroconversion, unsafe sex, risk behavior, anal intercourse, and numerous psychosocial terms such as norms, beliefs, attitudes, perceptions, knowledge, self-efficacy, mood, affect, and behavioral intentions). Additional papers were found by checking reference lists of articles identified in these database searches. We identified 36 studies published in scientific journals in which age was examined as a predictor of HIV seroconversion or unprotected anal intercourse in MSM. All were conducted in Western countries (Australia, Canada, US, and Northern European nations). Three of the US behavioral studies 10–12 and one of the US seroconversion studies 13 were not included in the review because they sampled very restrictive age ranges (i.e., 18–25, 17–22, 18–29 years) that precluded optimal examination of the young age risk effect. Thus, 32 studies were included in the review. Research designs and measures Prospective cohort studies provide a clear test of the association between age and risk of contracting HIV. These studies examine rates of conversion from HIV-seronegative status at baseline to HIV-seropositive status at a later point in time. All of the cohort studies reviewed here used standard procedures for Western countries to determine HIV serostatus: a double enzyme-linked immunosorbent assay with Western blot confirmation. Other studies have examined the prevalence of HIV infection in specific age-groups at particular timepoints. A cross-sectional design, however, does not provide a clear examination of the research question because that design contains an inherent sampling bias. At any given point of assessment, older men have had a disproportionately longer time period in which to become infected compared with their younger counterparts who are likely to have become sexually active more recently. Thus, a larger proportion of older than younger men may be living with HIV disease. Any possible increased risk of new infections in younger MSM could be offset by the extended time period for infection to occur and become detected in older MSM. Cross-sectional studies were not considered in this review. Cross-sectional and longitudinal assessments of unprotected anal intercourse were included. Some analyses distinguished between respondent-insertive and respondent-receptive behavior; the latter carries higher risk of contracting HIV 7,9. A few studies assessed different types of anal intercourse patterns (e.g., likelihood of always using a condom during anal intercourse, percentage who had engaged in unprotected anal intercourse outside of a monogamous relationship). All of the studies assessed behavior with self-reports; most used retrospective periods (e.g., past month, 3 months, 6 months) that have been found to yield reliable responses 14,15 and acceptable levels of predictive validity (i.e., subsequent documentation of seroconversion) 16–19. Age was operationalized as a continuous or categorical variable. Categorizations varied widely across studies (e.g., 30 years) 50 in whom HIV seroprevalence is higher 10. Future research should examine this possibility, and should sample a wide age range, have respondents provide information about sociodemographic characteristics of their sex partners, measure and statistically model respondent's age as a continuous variable predicting seroconversion and sexual risk behaviors, and test both linear and curvilinear (e.g., quadratic, cubic) age trends. By examining age as a continuous variable, one eliminates the need to make arbitrary cut-off points in the age distribution. If age categories must be used for descriptive purposes or to reduce skewness, 5-year age intervals may provide adequate sensitivity. In any event, a standardized approach to operationalizing age is greatly needed so that studies can be systematically compared. Summary A better understanding of the mechanisms underlying age-related risk of HIV infection will help address the practical situation. For MSM living today, particularly in North America, risk of contracting HIV appears to be higher among younger than older MSM. This is especially troubling given that most young men began their sexual histories in an era when HIV-protective practices are believed to be widely known. Behavioral scientists and practitioners alike must address the implications of these findings for developing targeted prevention interventions and treatment services for younger MSM. Acknowledgement The authors thank J. Moore, M. Rogers, L. Koenig, B. Bartholow, and the reviewers for their insightful comments.