Syndromic surveillance in public health

This thesis is an investigation of temporal and spatial patterns of over the counter (OTC) sales volumes of products related to gastrointestinal illness (GI). The utilisation of health related data, such as sales of non-prescription medications, is described as 'syndromic' surveillance, because indirect indications of disease activity may precede diagnoses or laboratory-confirmed information. Various analytic approaches were used to describe and evaluate whether sales patterns of OTC anti-diarrheal and anti-nauseant medications effectively reflected community GI activity, under large outbreak and for non-outbreak conditions. A retrospective analysis showed that dramatic increases in sales volumes of relevant OTC products mirrored case numbers of GI in the early stages of two large Canadian waterborne outbreaks. Evaluation of OTC sales patterns under non-outbreak conditions, for routine surveillance of GI, involved local and provincial level comparisons of data representing three years of daily aggregate sales, laboratory-confirmed cases of reportable GI, and another syndromic data source, chief complaints from emergency room (ER) visits for GI. Overall, seasonal patterns in the two syndromic data sources were very similar: sales and ER visits were high in the winter and spring, and low in the fall and summer. Generalised Linear Models with smoothing using natural splines revealed no discernable lag time (in days) between increases in GI related ER visits and sales of OTC products. When comparing weekly frequencies of reportable laboratory-confirmed cases to OTC sales volumes, seasonal patterns for Norovirus infections corresponded well, but bacterial and parasitic infections did not, being highest in summer and fall. Spatio-temporal analysis, using isopleth maps to visualise seasonal GI risk and OTC sales, revealed areas in Ontario that were consistently high in both risk of GI and sales, and also areas that consistently had only high sales. Despite the inherent limitations of both syndromic and reportable GI data, this thesis provided a number of valuable insights as to how they are related. Results of this research highlighted the potential public health applications of monitoring GI related OTC sales patterns for early detection of major outbreaks, routine surveillance of Norovirus, and for an alternative perspective of spatio-temporal trends in GI activity.