Estimating the efficacy of interventions to prevent mother-to-child transmission of HIV in breast-feeding populations: development of a consensus methodology.

ABSTRACT

Postnatal transmission of HIV through breast milk complicates both the design of effective interventions to prevent mother-to-child transmission of HIV (PMTCT) and their evaluation. Estimated long-term efficacy in five African trials (four with peri-partum antiretrovirals and one with artificial feeding) varied from 25 to 50 per cent. This variation may be due, at least in part, to differences in analytical methodology. To facilitate direct comparison between trials, a methodological consensus approach to the analysis and presentation of the results of PMTCT trials was developed. The initial methodology used and results presented from African trials with available long-term efficacy data were reviewed during a workshop in Bordeaux, France, in September 2000. A consensus approach for evaluating efficacy applicable across PMTCT studies was developed. There are four typical situations defined by duration of follow-up (short versus long), and the available demographic (vital status) and biological data (single versus repeat HIV testing). Efficacy can be assessed from the risk of infection directly or from HIV-free survival by combining infection and death as a single endpoint. Studies should report results in a standardized format including infection, weaning, mortality and loss to follow-up. New statistical methods that account for the unknown date when a child would first test positive for HIV, for weaning as a competing risk for HIV infection, and for increased risk of death among HIV-infected children should be used in analysing data from PMTCT studies with repeat HIV testing. All estimates should be reported with confidence intervals. This standardized methodology that allows direct comparison between studies is now being applied to four randomized clinical trials.