Discriminant Analysis of Factors Influencing the Occurrence of Chikungunya Disease in Households in the Community. Songkhla Province, Thailand

Abstract

From the situation of Chikungunya epidemic in Thailand and Songkhla province which occurs continuously and tends to increase. The researcher therefore conducted a study of this cross-sectional descriptive research. The objective was to analyze factors that could identify household groups that were at risk and not at risk of Chikungunya disease in communities in Songkhla province. The study was conducted in 2 groups of 566 persons, divided into 283 household members who had had Chikungunya disease and 283 households who had never had Chikungunya disease. It was found that there were 3 variables that could differentiate the household groups into households that were at risk and not at risk of Chikungunya disease: (1) prevention and control behavior of Chikungunya disease, (2) awareness of Chikungunya Disease, and (3) the Breteau Index. Such relationship could be written into a group classification equation in the raw score form of the function Y = (-1.996) + 0.443 (B) -0.222 (PS) - 0.003 (BI), and the group classification equation as a score of standard function Z = 0.962 (B) - 0.478 (PS) - 0.255 (BI). The coefficients of equation for predicting membership as at risk and not at risk households could be utilized through linear classification functions to produce an equation to forecast the classification of new households, as follows: forecasting equations for household groups that were not at risk for Chikungunya disease was D = -181.722 + 7.044 (PS) + 3.799 (B) + 0.064 (BI); and the forecasting equation for households that were at risk of Chikungunya disease was D = -179.899 + 7.246 (PS) + 3.397 (B) + 0.067 (BI). Validation of the classification function using the cross-validation method or leave-one-out revealed that the proposed functions could correctly identify household groups that were at risk and not at risk of Chikungunya disease by an average of 68.00%. For the households that were not at risk of Chikungunya disease, the functions could predict correctly up to 69.61 percent; and for the household groups at risk of Chikungunya disease, the function can predict correctly by 64.66%. From the research results, it was recommended that household members should promote behavior on preventing and controlling Chikungunya disease. Good awareness of Chikungunya disease should be promoted and the data on Breteau Index should be collected as well.