An Application of Transformed Distribution: Length of Stay in Hospitals

Abstract

Length of stay in hospitals are mostly characterized as asymmetric, right skewed and leptokurtic in nature. Earlier studies have considered parametric distributions like gamma, Pareto, lognormal for studying length of stay of patients in hospitals. However, in this study we have proposed transformed distributions to be the best choice for characterizing the length of stay. For this study, we have considered paediatric asthma dataset and identified that transformed Weibull-Pareto as the best fit. For a comparative purpose we have also provided the results of gamma, lognormal, and Pareto distribution. Maximum likelihood approach is considered to estimate the unknown parameters of the Transformed distribution followed by goodness of fit tests to examine the suitability of the fitted distributions. The results provide a direction for modelling the length of stay in hospitals due to different medical problems which require hospitalization.