Population and Metapopulation Ecology of Childhood Diseases

Researchers have long used mathematical models and empirical data to explore the population ecology of childhood diseases such as measles and whooping cough. These diseases have proven ideal model systems for studying population dynamics over space and time. Here we present a novel dataset of weekly measles and whooping cough case reports in pre-vaccine era U.S. cities and states, along with a previously- studied dataset of measles in England & Wales. We first estimate per-population disease reporting probabilities. We find that disease reporting is highly variable over space and between diseases, and correlated with socioeconomic covariates including ethnic composition and school attendance. Using these reporting estimates, we infer the long-term, marginal distribution of disease incidence for each population. This describes a probabilistic measure of disease persistence that compares favorably with a classic threshold persistence measure, critical community size (CCS). The U.S. and England & Wales exhibit similar patterns of measles incidence distributions: larger populations show higher mean viincidence and lower variance. The per-time probability of local extinction (conditioned on population size) is higher in the U.S. than in England & Wales, likely due to larger distances between U.S. cities. Finally, we use observed persistence and inferred incidence distributions to estimate the per-time probability of true persistence. Estimated persistence of whooping cough is much higher than persistence of measles (conditioned on population size). We find that cryptic persistence (the difference between observed and estimated persistence) of whooping cough is most common in small populations, while for measles cryptic persistence is most common in medium-sized populations that hover at the edge of extinction. Our results show that variation in disease reporting can significantly affect meta- population estimates of disease persistence, such as CCS. The distributional estimates of incidence presented here explicitly account for incomplete reporting, providing summaries of long-term ecological patterns that are comparable between metapopulations. These measures can provide disease control programs with valuable information on where disease incidence is expected to be higher or lower than expected based on population size alone.BiologyDoctoralUniversity of New Mexico. Biology Dept.Wearing, Helen J.Erhardt, Erik B.Moses, Melanie E.Brown, James H