Comparative Analysis of Seasonality of Childhood Infections in Seven United States Cities in the Pre-Vaccination era

The recent resurgence of vaccine preventable childhood infections in North America highlights how we may still benefit from a greater understanding of their epidemic dynamics. Seasonality is observed for several of these infections but underlying patterns and drivers remain incompletely understood. We used weekly disease incidence data from Project TYCHO™ for New York City, Chicago, Baltimore, Atlanta, New Orleans, Denver, and Los Angeles from 1924 to 1947 and a time-series susceptible-infected-recovered model to assess seasonal patterns and drivers (temperature, school term, and absolute humidity) of measles, scarlet fever, pertussis, and diphtheria. Measles and scarlet fever exhibited peaks in transmission during early winter months, diphtheria showed a peak in fall, while pertussis peaked in early spring. We found strong positive Pearson correlation coefficients of seasonal basic reproductive numbers (R₀) between measles and scarlet fever within six cities (range: 0.78-0.90). We also observed strong positive correlations of the seasonal R₀ of measles between cities (0.77-0.97) and when comparing R₀ estimates between cities for scarlet fever (0.49-0.96). We observed weaker correlation of R₀ estimates for pertussis (0.16-0.79) and diphtheria (0.31-0.92). Correlation of estimated seasonal forcing with climate data suggested that absolute humidity accounted for much of the annual variability seen in measles. The examined drivers failed to explain variability observed in scarlet fever, diphtheria, or pertussis. Understanding the seasonality, amplitude, and driving mechanisms of epidemics of childhood diseases are important for informing preparedness and vaccine implementation.