Effect of age at seroconversion on the natural AIDS incubation distribution. Multicenter Hemophilia Cohort Study and the International Registry of Seroconverters.

Abstract

To estimate the age-specific relative risk of progression from HIV seroconversion to onset of AIDS in hemophiliacs and homosexual men. Prospective follow-up data from HIV seroconversion to AIDS was analyzed for hemophiliacs in the Multicenter Hemophilia Cohort Study and for homosexual men in the International Registry of Seroconverters. Follow-up was censored at 1 July 1987 to obtain natural history estimates unaffected by therapies that were widely used after this date. The age-specific relative hazard of progression was estimated using nonparametric proportional hazards models and the baseline hazard function was described using spline models. Among the 373 children and adults with hemophilia and the 1020 adult homosexual men, each 10-year increment in age at seroconversion was associated with a 1.6- and 1.4-fold increase in the hazard of progression, respectively. The effect of age was highly significant among hemophiliacs. The magnitude of the effect was consistent in different cohorts of homosexual men, although it was not nominally significant. Furthermore, there was a significant increase (1.9-fold higher) in progression rates among homosexual men above rather than below 35 years of age at seroconversion. After adjusting for age, progression rates among hemophiliacs were significantly slower, possibly because Kaposi's sarcoma was rare. In both groups, the annual hazard rate increased during the first 5 years after seroconversion, but broad confidence limits prevented us from establishing hazard trends beyond this time. Estimates of the aggregate incubation time distribution were substantially shorter in older compared with younger individuals. The increasing risk with age was similar in hemophiliacs and homosexual men, suggesting that age is an endogenous host factor. These findings can be incorporated into newly developed backcalculation models to estimate HIV infection incidence by age group from age-specific AIDS surveillance data.