COS 61-1
Spatiotemporal variation of childhood disease epidemics spanning the 20th century
Spatial and temporal variation in infectious disease incidence is pertinent to understanding the ecology governing disease transmission. Knowledge of this variation allows for the optimization of control strategies to mitigate disease burden. To date, the testing of hypotheses regarding the mechanisms that generate patterns of disease across a landscape have been limited by the lack of longitudinal disease data. Here for the first time, we take advantage of a recently digitized historical dataset of childhood infections spanning the 20th century in the United States. Using these data, we ask (1) are all notifiable childhood diseases seasonal, (2) does heterogeneity exist in the seasonal incidence of childhood disease, and if so, is (3) heterogeneity driven by human demography and/or environmental factors? To answer these questions, we performed wavelet spectral analyses on weekly state-level incidence time-series for nine childhood infections. Heterogeneity in the seasonal timing of epidemics was investigated by quantifying the variation in seasonal phase angles explained by the latitude and longitude of population centers. Additionally, a comparative analysis was performed to characterize the pairwise phase relationships between diseases. Lastly, the mechanisms underlying disease seasonality were explored by coupling these disease data with complimentary state-specific monthly natality and monthly climatic data.
Results/Conclusions
Seasonal periodicity in disease incidence was observed in all of the infections in our study. In addition to the annual periodicity, complex multi-annual cycles were discovered in pertussis, mumps, and measles. We revealed a clear annual geographic gradient in epidemic timing for polio, diphtheria, scarlet fever, rubella, typhoid fever, and chickenpox. Crucially, multiple diseases shared similar geographic gradients, suggesting a common mechanism driving those patterns. A north-south gradient was observed in polio, typhoid fever, scarlet fever, and diphtheria and was associated with variation in seasonal birth timing and amplitude, as well as multiple climatic variables. Within each disease, epidemics peaked 6-19 weeks apart across the United States. This could facilitate persistence via source-sink dynamics, which have been previously observed for measles in England and Wales. The sequential appearance of epidemics within each state may provide insight into the ecological interference between pathogens, and the mechanisms governing disease seasonality.