The Predictions Dengue Haemorrhagic Fever Patients Using Poisson Regression Model

Abstract

The objective of this study was to develop and validate a Poisson regression model to verify
predictors of Dengue haemorrhagic fever (DHF) patients in Kreang Sub-District, Cha-Uat District,
Nakhon Si Thammarat, Thailand. DHF cases, human behaviors, types and number of mosquitoes
breeding sites were collected from 160 households by survey in year 2016 from March to July.
Stratified sampling technique was used to sampling procedure. Mosquito larvae from positive
containers were then identified into species level by using stereomicroscope. Poisson regression is a
regression technique available for modeling variables that describe count or discrete data of the
occurrences of some event over a specified interval. SPSS for Windows version 21 was used to
data processing. The results showed that DHF cases were count data and were assigned to response
variables. Its mean was approximately equal its variance which suggested that a Poisson regression
model would be appropriate. The variable of the mosquito breeding site was assigned and estimated
using the maximum likelihood method. The Poisson regression was statistically significant model. The
log of the DHF was predicted with the predictors: log (DHF) = -2.970+1.584(CT)+2.277(TCT)
+1.386(WJ). Human behaviors were the important factors to promote the DHF widespread,
especially to facilitated the mosquito breeding. So that if we understanding we would be able to
predict the DHF incidence and prevention