1442. Spatiotemporal Clusters of Varicella and the Regional Risks through Bayesian Approach: A National Five-year Cohort Analysis
Kwan Hong,Jeehyun Kim,Sujin Yum,Raquel Elizabeth Gómez Gómez,Byung Chul Chun
DOI: https://doi.org/10.1093/ofid/ofab466.1634
2021-11-01
Open Forum Infectious Diseases
Abstract:Abstract Background Since varicella epidemics repeatedly occurred in Korea, it is essential to control varicella outbreaks preemptively in the targeted region. Therefore, we aimed to reveal spatiotemporal clusters of varicella and the regional risk factor of varicella incidence at the national level. Methods All varicella cases (defined as ICD-10 codes, B01-B09) from 2013 to 2017 in Korea were extracted from National Health Insurance Service. Of the total, 566,978 cases were realigned spatially by 250 administrative districts of Korea and temporally by a week. Spatial autocorrelation was tested using the global Moran’s I statistics using Monte Carlo simulation. Kulldorff’s prospective space-time scan statistics were used to reveal space-time clusters of varicella. Possible risk factors were extracted from the Korean Statistical Information Service and Community Health Survey of Korea, including hand hygiene perceptions, alcohol and smoking status, the proportion of children under 15 years old, the number of households, and household income by regions. After selecting significant risk factors through non-spatial generalized linear models, a conditional autoregressive spatiotempoal model with Bayesian extension was applied to estimate the regional factors affecting varicella incidence. Results There was spatial autocorrelation using Global Moran’s I statistics (P< 0.01). When the maximum cluster size was limited to 10% of the population, 17 spatiotemporal clusters were detected in specific regions in Korea (figure 1). Low perception of hand hygiene, the high proportion of alcohol drinking and cigarette smoking, high children proportion, low number of familial member, and low household income were associated with varicella spatiotemporal incidence (odds ratio: 0.97, 1.01, 2.31, 1.10, 0.99, 0.99, respectively; 95% credible intervals of all risk factors did not include 1). Figure 1. Space-time prospective clusters of varicella in Korea using varicella incidence from 2013 to 2017. Relative risks ratio of each cluster is described at the point. Conclusion Varicella incidence shows spatiotemporal clustering patterns in specific regions. Since regional factors such as the perception rate of hand hygiene, child proportion, alcohol drinking, cigarette smoking, and low household income affect varicella’s spatiotemporal incidence, strategies for targeted control of high-risk regions are strongly recommended. Disclosures All Authors: No reported disclosures
immunology,infectious diseases,microbiology