Spatial association between environmental factors and melioidosis in Thailand

Abstract

Background Melioidosis is an infectious disease incurring health burden in Southeast Asia and northern Australia. It is an infectious disease that can be transmitted to both humans and animals whose contact with contaminated soil or water may expose them to the bacteria in an endemic area. This study examined the spatial association between environmental factors and melioidosis incidence in Thailand using the statistics of melioidosis registered in the National Notifiable Disease Surveillance System (Report 506), administered by the Ministry of Public Health. Data on satellite-based environmental factors, including normalized difference drought index (NDDI), normalized difference vegetation index (NDVI), normalized difference water index (NDWI), daytime and night-time land surface temperature (LST), urban and built-up area, cropland and precipitation, were obtained from Google Earth Engine. Methods The study used secondary data on melioidosis registered in the National Notifiable Disease Surveillance System (Report 506) by the Department of Disease Control, Ministry of Public Health of Thailand, and the satellite-based environmental factors, from 2006 to 2020. All variables were aggregated at the provincial level to match the spatial resolution of melioidosis incidence. An analysis of bivariate Local Indicators of Spatial Association (LISA) showed that all of the satellite-based environmental factors were statistically associated with melioidosis incidence at p-value < 0.05. In addition, the statistically significant clusters were mostly located in the Northeast region. Fixed- and random-effect regression models were also used for examining multivariable relationships. Results Both models showed similar results, indicating that all of the satellite-based environmental factors were statistically associated with melioidosis incidence at p-value < 0.001, except for precipitation in the fixed-effect model (p-value = 0.096). Particularly, the regression coefficients of NDVI, night-time LST, cropland and precipitation were positive, and those of NDDI, NDWI, daytime LST and urban and built-up area were negative. Conclusion All these results indicated that favourable climate condition and farming activity are associated with melioidosis incidence. Furthermore, the satellite data can alternatively provide timely information for future prevention policies.