Analyzing Risk Factors of Toxocara cati Infection on Pet Cats Using A Logistic Regression Model

Abstract

Regression analysis is one of statistical methods used to model the relationship of explanatory variables and a single response variable. Logistic regression is applied on a categorical response variable typically a binary variable. In this research, the response variable was infection of Toxocara cati with 0 as negative infection and 1 as positive infection. The data was obtained from a survey conducted by a postgraduate student majoring in Veterinary Public Health of Bogor Agricultural University in 2014. The best model was developed by means of best subsets method which resulted in four risk factors namely were sex, availability of defecation sand, frequency of going outside, and repetition of giving anthelmintic. The resulted model showed 70.0% correct in prediction but the sensitivity was only 40.2%. It showed that the resulted model not so good in prediction. It was also described by the area under ROC curve that was only 68.1%. Based on this model, it was shown that risk factors of Toxocara cati infection were the availability of defecation sand and repetition of giving anthelmintic. The significance level of these factors was 5%...