Objective: Zimbabwe is one of the poorest countries in the world, just emerging from a broken health care system. The objective is to figure out variables affecting hospital charge for Zimbabwe.
Material and methods: The variables used are sex, smoking status, number of children, region, age and body mass index. The first six of these are factors and the remaining are covariates. A mixture model was fitted to describe the dependence of hospital charge on these variables.
Results: A mixture model with five components each having a reversed Gumbel distribution was found to give an adequate fit. Both the covariates and all but one of the factors were found to be significant. Estimates of value at risk of hospital charge are given for all combinations of the factors.
Conclusions: The results suggest that the hospital charge could be higher for females, higher for smokers, higher if the patient had more children and higher if the patient is older. Further, estimates of value at risk given suggest, for example, that a 90 year old female not smoking and having no children and an average body mass index will have a hospital charge less than Z$49851 with probability 0.999.
Keywords: Estimation; Expectation maximization algorithm; Value at risk.
© 2023 The Author(s).