Application of log-linear models to pneumonia patients: A case study of Jigme Dorji Wangchuck National Referral Hospital (JDWNRH) in Bhutan

Abstract

© Published under licence by IOP Publishing Ltd. Pneumonia is a lung infection or swelling (inflammation) of the tissue in one or both lungs caused by bacteria Streptococcus pneumonia. Pneumonia claims the lives of the world's most vulnerable children. The aim of this study is to find the association between pneumonia variables of pneumonia patients. The recorded data of 4190 JDWNRH pneumonia patients from 2005 to 2016 were used. A test of independence was conducted for all pneumonia patients and individual gender to identify the association between any two variables using chi-square test and Cramer's V statistic. The variables with maximum Cramer's V were formulated using two-dimensional log-linear models to obtain estimate parameters, expected frequencies and standardized residuals. Three-dimensional log-linear models were applied to test the association among any three variables. The analysis of deviance is used to select the best models. The result showed that most paired variables were significantly related at p-value less than 0.01. For all data, ward has a strong association with pneumonia types, age, status of last contact and region. In addition, length of stay has association with status of last contact.