Hepatitis A incidence rate estimates from a pilot seroprevalence survey in Rio de Janeiro, Brazil

BACKGROUND: To assess the impact of water sanitation and sewage disposal, part of a major environmental control programme in Rio de Janeiro, we carried out sero-prevalence studies for Hepatitis A virus (HAV) in three micro-regions in Rio de Janeiro. Each region varied with regard to level of sanitation. We are interested in assessing the discriminating power of age-specific prevalence curves for HAV as a proxy for improvement in sanitation. These curves will serve as baseline information to future planned surveys as the sanitation programme progresses. METHODS: Incidence rate curves from prevalence data are estimated parametrically via a Weibull-like survival function, and non-parametrically via maximum likelihood and monotonic splines. Sera collected from children and adults in the three areas are used to detect antibodies against HAV through ELISA. RESULTS: We compare baseline incidence curves at the three sites estimated by the three methods. We observe a strong negative correlation between level of sanitation and incidence rates for HAV infection. Incidence estimates yielded by the parametric and non-parametric approaches tend to agree at early ages in the microregion showing the best level of sanitation and to increasingly disagree in the other two. CONCLUSION: Our results support the choice of HAV as a sentinel disease that is associated with level of sanitation. We also introduce monotonic splines as a novel non-parametric approach to estimate incidence from prevalence data. This approach outperforms current estimating procedures.