Spatial Analysis of Severe Fever with Thrombocytopenia Syndrome Virus in China Using a Geographically Weighted Logistic Regression Model

Abstract

Abstract:Severe fever with thrombocytopenia syndrome (SFTS) is caused by severe fever with thrombocytopenia syndrome virus (SFTSV), which has had a serious impact on public health in parts of Asia. There is no specific antiviral drug or vaccine for SFTSV and, therefore, it is important to determine the factors that influence the occurrence of SFTSV infections. This study aimed to explore the spatial associations between SFTSV infections and several potential determinants, and to predict the high-risk areas in mainland China. The analysis was carried out at the level of provinces in mainland China. The potential explanatory variables that were investigated consisted of meteorological factors (average temperature, average monthly precipitation and average relative humidity), the average proportion of rural population and the average proportion of primary industries over three years (2010)(2011)(2012). We constructed a geographically weighted logistic regression (GWLR) model in order to explore the associations between the selected variables and confirmed cases of SFTSV. The study showed that: (1) meteorological factors have a strong influence on the SFTSV cover; (2) a GWLR model is suitable for exploring SFTSV cover in mainland China; (3) our findings can be used for predicting high-risk areas and highlighting when meteorological factors pose a risk in order to aid in the implementation of public health strategies.

Impaired Water Hazard Zones: Mapping Intersecting Environmental Health Vulnerabilities and Polluter Disproportionality

“…The present study supports previous research that demonstrates GWLR generally offers a better model fit when compared to global models like the GLR. Likewise, this study supports the notion that GWLR is a powerful tool for uncovering spatial non-stationarity in the predictors of binary outcomes [56][57][58][59][60][61][62][63][64][65][66][67][68][69][70][71][72][73].…”

“…This presentation of the GWLR results maintains comparability with previous GWR-based research on environmental inequality outcomes [70]. This new visualization strategy also contributes to a growing literature that uses FDR-correction techniques in the presentation of GWLR results [58,61,65].…”

“…Since the introduction of geographically weighted regression (GWR) techniques [54,55], studies have used GWR to model spatial non-stationarity in a wide variety of multivariate statistical relationships. These include applications of geographically weighted logistic regression (GWLR) to model local variation in the predictors of binary outcomes spanning multiple subfields in human and physical geography [56][57][58][59][60][61][62][63][64][65][66][67][68]. Within the environmental inequality outcomes literature, spatial non-stationarity has been identified in the aggregate-level population vulnerability predictors of industrial air-toxic releases in New Jersey, USA [69], estimated lifetime cancer risk from cumulative ambient air-toxic pollution in Florida, USA [70], and vegetation land cover in shrinking and growing US cities [71].…”

“…I compared overall goodness-of-fit of the GLR and GWLR models with the percent deviance explained and AIC statistics. Lower AIC values and higher percent deviance explained indicate better model fit [57,58,62,63,[65][66][67][68]80].…”

“…More recent research is attentive to these issues of abnormality, multiplicity, and spatial dependence [20,58,61,65]. Guided by this work, I used the 1.96 and FDR-corrected critical t-value thresholds in the process of classifying the GWLR results in the Bay-Delta.…”