Spatiotemporal heterogeneity and the long-term impact of meteorological, environmental and socio-economic factors of scrub typhus in China from 2012 to 2018

Large-scale outbreaks of scrub typhus combined with the emergence of this vector-borne rickettsiosis in new areas indicate that this disease remains seriously neglected. This study aimed to explore the long-term changes and regional leading factors of scrub typhus in China, so as to provide fresh insights for the prevention and control of this disease. In this study, a Bayesian space-time hierarchical model (BSTHM) was used to identify the long-term spatiotemporal heterogeneity of scrub typhus and quantify the association between meteorological factors and scrub typhus in southern and northern China from 2012 to 2018. GeoDetector model was used to quantify the dominant forces of environmental and socioeconomic factors in the Northern and the Southern China. Scrub typhus often appeared in summer and autumn (June to November), and epidemically peaked in October, with obvious temporal seasonality. Spatially, the hot spots (high-risk regions) were concentrated in the south, on the contrary the cold spots (low-risk regions) in the north. In addition, the main meteorological factor, average temperature, gave a significant impact in both areas. The average temperature increased by 1 °C, resulting in a decrease of 1.10% in southern China and an increase of 0.96% in northern China in the risk of scrub typhus. The determinant environmental and socio-economic factors of scrub typhus in the two areas were altitude and per capita GDP, with q-values of 0.91 and 0.87, respectively. Meteorological, environmental and socio-economic factors had a significant impact on the distribution of scrub typhus, with obvious seasonality and spatial heterogeneity. This study provides helpful suggestions and basis for reasonably allocating resources and controlling the occurrence of scrub typhus. Author summary Scrub typhus is a natural-focus disease caused by the bite of chigger mite larval. In this study, we use BSTHM to capture the overall temporal trend and spatial hot spots of scrub typhus, and quantify the relationship between the disease and major meteorological factors. Meanwhile, Geodetector model was used to quantify the influence of other potential risk factors and estimate the spatio-temporal heterogeneity of scrub typhus. The results showed that scrub typhus had significant seasonality, with a q value of 0.52, and spatial heterogeneity, with a q-value of 0.64. Scrub typhus mainly occurred in summer and autumn, and high-risk areas were mainly distributed in southern China (Yunnan, Hainan and Guangdong). These heterogeneity were closely related to the vector and host. Whether in the South or the north, scrub typhus was closely related to risk factors such as temperature, per capita GDP, NDVI, altitude and the percentage of children aged 0-14. These results suggest that the relevant departments should strengthen the monitoring of the ecological environment, the host and vector of Orientia tsutsugamushi, and strengthen the risk awareness, so as to prevent and control the possible increased risk of scrub typhus under these meteorological, environmental and socio-economic conditions. Considering the differences in different regions, resources should be allocated reasonably.

occasionally. This disease is characterized of fever, eschar, rash and lymph node enlargement, and cloud(http://www.gscloud.cn/) (Fig.2). part represented common spatiotemporal variation of scrub typhus, whereas the local part revealed 115 the spatiotemporal heterogeneity of the incidence of scrub typhus throughout the whole study period.
where q represents the non-linear relation between the decisive socioeconomic factors and scrub 152 typhus. The value ranges from 0 to 1, and a higher value of q-statistic suggesting a higher 153 determinant power of a risk factor or the heterogeneity of a target variable. and ℎ are the numbers 154 of provences in the total study area and in the ℎ-th stratum(ℎ = 1,2,…, ) ,respectively. 2 and 2 ℎ 155 represent the variances in scrub typhus incidence in the entire countries in the ℎ-th stratum, 156 respectively.

180
Among the six hot spots, the upward trends of Yunnan, Guangdong, Guangxi and Hainan were 181 faster than the overall trend. Consequently, the risk of these regions might be higher than the overall 182 risk and continue to face high incidences in the future. Jiangxi and Fujian showed the same trends 183 as the overall trend, which indicated that these regions would still be hot areas in the future.

184
Therefore, relevant prevention and control departments should focus on these provinces (Fig.4).

185
Among the eleven cold spots, the increasing trend of Xinjiang was the same as the overall trend, so 186 its current risk level would remain constant in the future. The other 10 provinces showed lower 187 increasing than the overall trend, which indicated that the incidence risk of these provinces would Table 1 193 The   Table 3 201 The q values (q 1 , q 2 ) calculated for the association between scrub typhus and environmental and 202 socioeconomic factors in northern and southern China, respectively.

217
The effect of other potential meteorological factors could not be ignored, except for precipitation.

218
For example, in southern China, a 1% increase in RH was associated with a 0.3540% risk reduction 219 (RR: 0.9965). A 1-h increase in total solar hours was related to a 0.0772% increase in scrub typhus 220 risk,with a corresponding RR of 0.9992 (Table 1).

221
In the north of China, in addition to the average temperature, RH also had a nonnegligible impact 222 on this disease. A 1% increase of RH was related to a 0.7552% increase in scrub typhus risk, with 223 a corresponding RR of 1.0080. A 1 hour increase in total solar hours was associated with an 224 increased of 0.1653% in the risk of scrub typhus (RR: 1.0020). In addition, the estimated coefficients 225 for precipitation was not significant (Table 2).

247
The results showed that the spatial distribution of scrub typhus in China was inhomogeneous. The

253
The spatial heterogeneity may be attributed to, but not limited to, the following reasons. On the one 254 hand, regional differences may lead to differences in vectors and hosts between the two regions,

276
The risk of scrub typhus also presented obvious temporal heterogeneity.

304
Monthly average RH was negatively correlated with the risk of scrub typhus in the south,which is 305 different from expection, because higher RH means more mites. This negative correlation can be 306 attributed ,but not limited, to the following two reasons: One is that the water source for mites to 307 survive is water vapor(Clopton and Gold, 1993), so RH has always been a crucial factor in 308 determining the number of Chigger Mites. However, higher humidity is also not conducive to the

325
This study has some limitations. First of all, we used provincial data to explore the association at 326 the group level, which may lead to an ecological fallacy inevitablely, but this did not influence the