Impact of incomplete coverage of neonatal dried blood spot screening on estimating HIV-1 seroprevalence.

The aim of this study was to determine the extent to which selective under-coverage of births to mothers more likely to be at risk of HIV-1 infection will result in a significant under-estimation of the true neonatal seroprevalence. Census data, local birth statistics, maternity data and data from the prevalence monitoring programme were used to produce a model to predict the effects of under-coverage in the uptake of neonatal metabolic screening which has been observed in babies with a mother of ethnic group black African. The adjustment factor which allows for under-coverage is the relative inclusion ratio (RIR); the probability that samples from a group at different risk of HIV infection were included in the survey divided by the probability of inclusion for samples from all other babies. The RIR was found to be close to unity (0.97), indicating a minimal bias. Under usual conditions only if the relative inclusion ratio (RIR) declined to values of 0.87 or below would there be a substantial bias. Despite some selective under representation, the results obtained from the Unlinked Anonymous HIV Monitoring Programme Dried Blood Spot Survey would seem to identify levels of prevalence in the population of child-bearing women with a good degree of accuracy and remains a useful tool for resource allocation, planning of services, provision of care and counselling.