Fitting length of stay in hospitals using transformed distributions

Abstract

Healthcare is imperative for well-being of people, wherein length of stay in hospital for treatment poses serious challenge in hospital management. Studies have shown that the distribution of length of stay in hospitals is heavy tailed in nature and lognormal, gamma, and exponential distribution are found to be suitable for fitting length of stay in hospitals. In this study, we consider transformed distributions for fitting length of stay. We compare three transformed distributions Beta-Cauchy, Gamma-Pareto, and Gamma-Exponential-Cauchy and show that the transformed Gamma-Pareto distribution is more appropriate for fitting length of stay of diabetes patients in hospitals.