THE RELATIONSHIP OF HIV PREVALENCE IN PREGNANT-WOMEN TO THAT IN WOMENOF REPRODUCTIVE AGE - A VALIDATED METHOD FOR ADJUSTMENT

Objective: To devise and validate a method for adjusting HIV seropreva lences in pregnant women to estimate population prevalences among all women in their child-bearing years. Design: Birth and termination rate s from women with known HIV infection in the United Kingdom were calcu lated according to the likely route of HIV infection and whether HIV i nfection was diagnosed. Methods: Birth and termination rates were weig hted and combined to produce summary statistics. Comparisons were then made with population birth and termination rates to derive summary re lative inclusion ratios (RIRs), the relative probabilities of includin g HIV-infected and uninfected women in seroprevalence surveys of pregn ant women. Results: The derived RIRs for women having live births were close to unity: 1.03 [95% confidence intervals (CI) 0.90-1.17] for Lo ndon and 0.80 (CI, 0.71-0.89) for elsewhere in England and Wales. This indicates that currently observed overall seroprevalences among pregn ant women having live births in London would be similar to those among all women of the same age, while elsewhere it would be slightly under estimated. Sensitivity analysis indicated that RIRs could, however, va ry three-fold (0.47-1.56) according to the proportion of diagnosed mat ernal infections and the mix of maternal HIV-exposure categories. The method was validated by using it to predict the ratio of unlinked sero prevalences between women having terminations and live births in Londo n. It predicted a ratio of 1.74 : 1, which is close to the observed ra tio of 2.07 : 1. Conclusions: Application of HIV seroprevalences from pregnant women to whole populations may need adjustment for fertility rates among HIV-infected women. A general method for this has been der ived and validated. Gathering fertility data for HIV-infected women is a useful adjunct to serosurveillance. (C) 1998 Lippincott Williams & Wilkins.