The seasonality of varicella in the tropical climates of Colombia: A statistical and mathematical modeling study

Abstract

AbstractBackgroundVaricella remains a major burden in many tropical regions, where low- to middle-income countries report the highest mortality rates. Understanding when and where varicella incidence increases could help us understand viral transmission and optimize the allocation of resources. Due to the lack of surveillance data, the epidemiology of varicella in the tropics has not been well characterized. Here, we assessed varicella seasonality and analyzed its correlation with climatic variables across Colombia.MethodsWe compiled an extensive dataset of weekly varicella reports in children up to the age of ten in 25 municipalities across Colombia. We used generalized additive models to describe the seasonality of varicella in each municipality. Using clustering methods and matrix correlation tests, we then compared the spatial variability in varicella seasonality with that in five meteorological variables across the municipalities. Finally, we developed a mathematical model to examine whether the influence of the climate on transmission rates could result in the observed seasonal patterns of varicella in Colombia and other Central American countries.FindingsVaricella seasonality was markedly bimodal, with a more pronounced peak early in the year in northern municipalities (≈week 15), while later in the year (≈week 40) in southern municipalities, including Bogotá. This spatial gradient was strongly correlated with humidity (Mantel statistic = 0·412, p-value = 0·001) but not temperature (Mantel statistic = 0·077 and p-value = 0·225). Furthermore, a mathematical model that included a small, negative association between humidity and transmission was able to reproduce the observed spatial patterns in Colombia and México. This model also predicted a latitudinal gradient in other tropical countries of Central America, where the seasonality of varicella has not yet been characterized.InterpretationThese results demonstrate a large variability in varicella seasonality across the tropical climates of Colombia. They further suggest that seasonal fluctuations of humidity explain the varicella epidemics calendar in Colombia and other Central American countries. More generally, our results highlight the need to carefully consider the subnational heterogeneity of climates when studying the seasonal epidemiology of varicella and assessing the impact of control measures.FundingCore funding from the Max Planck Institute for Infection Biology of the Max Planck Society, Berlin, Germany.