Analysis of the norovirus contamination hotspots at a nationwide scale using physical and statistical mapping methods

Abstract

Norovirus is the leading cause of gastroenteritis worldwide, and the role of environmental conditions in norovirus contamination and transmission has gained increasing interest among public health managers. The goal of this study was to identify hotspots of norovirus contamination at a nationwide scale and examine the spatiotemporal characteristics of the anthropogenic environmental factors affecting these hotspots. National-scale data on the environmental occurrence of norovirus collected at 1,070 sites from 2014 to 2016, which consisted of binary classes of the detection or non-detection of noroviruses in various environmental media, such as agricultural products, human feces from public toilets, irrigation water, surface soil, and stream water, were used. The spatiotemporal environmental factors were investigated for their potential correlation with the occurrence of norovirus using physical and statistical mapping methods: a geographic information system, spatial scan statistics, and a self-organizing map (SOM). The spatial environmental factors affecting two (Hotspot A and B) of the four hotspots with relatively high log-likelihood ratios significantly differed from those affecting the nondetection areas. Hotspot A exhibited higher representative values than the non-detection sites in 12 environmental factors, whereas Hotspot B showed higher representative values than the non-detection sites in 8 environmental factors. A total of 10 norovirus contamination risk areas were identified based on the selected environmental factors. These areas matched well with the districts in which norovirus outbreaks frequently occurred. SOM analysis clearly differentiated the spatiotemporal characteristics of the environmental factors affecting the two hotspots. In addition, influential factors associated with norovirus contamination within hotspots were quantitatively evaluated using the Mutual Information (MI) method and Bayesian Graphical Network (BGN). This study provides a novel approach for predicting the areas at risk of norovirus contamination based on environmental factors and can be useful for establishing strategies that help national institutions prioritize the management of high-risk areas via mapping.