F1000Research (Mar 2018)
A comparison of spatial heterogeneity with local cluster detection methods for chronic respiratory diseases in Thailand [version 2; referees: 2 approved]
Abstract
Background: The Centers for Disease Control and Prevention reported that deaths from chronic respiratory diseases (CRDs) in Thailand increased by almost 13% in 2010, along with an increased burden related to the disease. Evaluating the geographical heterogeneity of CRDs is important for surveillance. Previous studies have indicated that socioeconomic status has an effect on disease, and that this can be measured with variables such as night-time lights (NTLs) and industrial density (ID). However, there is no understanding of how NTLs and ID correlate with CRDs. We compared spatial heterogeneity obtained by using local cluster detection methods for CRDs and by correlating NTLs and ID with CRDs. Methods: We applied the spatial scan statistic in SaTScan, as well as local indices of spatial association (LISA), Getis and Ord’s local Gi*(d) statistic, and Pearson correlation. In our analysis, data were collected on gender, age, household income, education, family size, occupation, region, residential area, housing construction materials, cooking fuels, smoking status and previously diagnosed CRDs by a physician from the National Socioeconomic Survey, which is a cross-sectional study conducted by the National Statistical Office of Thailand in 2010. Results: According to our findings, the spatial scan statistic, LISA, and the local Gi*(d) statistic revealed similar results for areas with the highest clustering of CRDs. However, the hotspots for the spatial scan statistic covered a wider area than LISA and the local Gi*(d) statistic. In addition, there were persistent hotspots in Bangkok and the perimeter provinces. NTLs and ID have a positive correlation with CRDs. Conclusions: This study demonstrates that all the statistical methods used could detect spatial heterogeneity of CRDs. NTLs and ID can serve as new parameters for determining disease hotspots by representing the population and industrial boom that typically contributes to epidemics.
Keywords
