Time Series Evaluation of Childhood Diarrhea in Abu Homos, Egypt

Background: Infectious diarrhea is a significant cause of morbidity and mortality in children living in developing regions of the world. However, the temporal distribution and mechanisms underlying fluctuations in diarrhea incidence are largely unknown. We sought to model all cause and ETEC- and Campylobacter-attributable diarrhea incidence using time series methods to assess temporal fluctuations as well as identify specific climate-related factors associated with fluctuations in disease.;Methods: Data on diarrhea incidence were obtained from a series of prospective cohort studies of pediatric diarrhea conducted between 1995 and 2006 in Abu Homos, Egypt. This agricultural community of approximately 350,000 persons is located in the Nile Delta region of northern Egypt approximately 22 miles southeast of Alexandria. Active surveillance, including twice weekly home visits, assessed for incident diarrheal episodes. Specimens were collected and tested for known bacterial enteropathogens. These data were merged with climatological data from Alexandria, Egypt obtained from the Climate Services Branch of the National Climate Data Center (Ashville, NC).;Autoregressive Integrated Moving Average (ARIMA) models were developed for all-cause diarrhea, and for Campylobacter- and ETEC-attributable diarrhea incidence. Models were also developed for common ETEC phenotypes and for diarrhea in which no pathogens were identified. Climate variables were analyzed as input series to assess their temporal association with each outcome. Multivariate models were developed and model fit was assessed by residual analysis and validated on a subset of the data series.;Findings: We found seasonal and non-seasonal fluctuations in all-cause and pathogen-specific diarrhea incidence. Additionally, there was variability in the temporal distribution of incident diarrhea across cohorts. For all-cause diarrhea, increasing rainfall was associated with increasing diarrhea incidence at both proximal and delayed lags. Univariable and multivariable analyses of ETEC-attributable diarrhea highlighted some important temporal variability by pathogen phenotype. Campylobacter-attributable diarrhea incidence showed no consistent seasonality, though there was a delayed-type, borderline significant association with precipitation. The multivariate models explained a significant proportion of the series variability and adequately modeled disease incidence in the validation subset analyses.;Interpretation: This study confirmed that precipitation plays a significant role in the incidence of all cause and pathogen-specific diarrheal diseases in a region and population for which there is significant burden. Novel findings include pathogen- and phenotype-specific associations and a common delayed association with precipitation. While climate cannot be controlled, studying the mechanism of how it impacts disease incidence may identify more proximal causes of true risk factors and result in better targeted interventions.