Duration of human immunodeficiency virus infection and likelihood of giving birth in a Medicaid population in Maryland.

The objective of this study was to examine the effect of duration of human immunodeficiency virus (HIV) infection on a woman's likelihood of giving birth. Using longitudinal data from the Maryland state Human Immunodeficiency Virus Information System and a retrospective cohort design, the authors compared 1,642 women with acquired immunodeficiency syndrome (AIDS) to 8,443 uninfected women enrolled in the Medicaid program between 1985 and 1995. The decade before AIDS diagnosis was divided into four 2.5-year periods. Proximity to AIDS diagnosis served as a proxy for duration of infection. An extension of the Cox model was used to estimate the relative risk for giving birth, with adjustment for covariates and repeated outcomes. The average number of births per 100 person-years was 6.0 for HIV-infected women and 11.1 for uninfected women (adjusted relative risk = 0.63; 95% confidence interval (CI): 0.57, 0.68). Accounting for duration of infection, the adjusted relative risks for birth among HIV-infected women, as compared with uninfected women, were 0.85 (95% CI: 0.71, 1.03), 0.74 (95% CI: 0.63, 0.86), 0.55 (95% CI: 0.47, 0.64), and 0.45 (95% CI: 0.38, 0.55) for successive 2.5-year periods before AIDS diagnosis. Demographic characteristics, contraception, abortion, fetal loss, or drug use could not fully explain the reductions. These results suggest that HIV-infected women experience a progressive reduction in births years before the onset of AIDS. This may compromise estimation of HIV prevalence and interpretation of time trends from serosurveillance of pregnant women.