ARIMA Models of Dengue Cases in Kartamantul, Based on Area Risk Classification

Abstract

Dengue is still one of public health problems in Indonesia. In this study, three temporal indices (frequency, duration and intensity indices) based on serologically confirmed cases between 2010 and 2014 in Yogyakarta Municipality, Sleman Regency and Bantul Regency (acronym: Kartamantul), which are spatially analyzed, used to determine the risk level of Dengue transmission for each village in that area in 2015. Subsequently, ARIMA models with Box-Jenkins approach for those risk classification are developed to predict the number of cases in 2015. The results show that the risk categorization yielded from those Dengue data series has relatively high concordance with risk classification resulting from Dengue data in 2015 (the Kappa coefficient: 0.593; p-value < 0.001). The best ARIMA models for both the “high” and “medium” risk villages are (0, 1, 0)(1, 1, 0)12; and for “low” risk areas it is (0, 1, 0)(0, 1, 0)12; which means that both models demonstrate a seasonal pattern. The analysis shows that the ARIMA models have relatively good predictability for the upcoming number of cases. Therefore, these analyses approach is suggested to be adopted for complementing the techniques of area stratification and transmission period which are commonly used in Dengue surveillance.