Modeling In-Patient Morbidity and Mortality Cases of Some Infectious Diseases

Abstract

The study is an attempt to assess the appropriateness of Binomial and Poisson distribution models, in measuring and modelling the monthly distribution of in-patient morbidity and mortality cases of the top six infectious diseases according to the World Health Organisation (2011) report: malaria, HIV/AIDS, pneumonia, hepatitis B, tuberculosis and typhoid, and use the appropriate models to determine the chances at which certain number of deaths occur in a month. Data for the study were obtained from the Biostatistics Unit of the Regional Hospital in the Central Region of Ghana and covered monthly data from the period of January, 2008 to December, 2012. To determine how well a statistical model can fit a particular distribution, model errors were calculated to determine the difference in the model and actual distribution of the data set. Also, two formal goodness-of-fit tests were considered; the Kolmogorov-Simirnov test and Chi-Square Goodness of Fit test were conducted to compare the discrepancies between the reality and what the distributions are predicting. At the end of the analyses, it was found at 5% significance level that, the monthly number of in-patient morbidity and mortality cases on Malaria, HIV/AIDS and Pneumonia significantly fit Poisson distribution better than Binomial. Finally, the research found that, it is highly unlikely that more than 10 admitted patients in the Central Hospital will die of malaria, HIV/AIDS and pneumonia in a particular month, and likely that less than 4 admitted patients will die of malaria, HIV/AIDS and pneumonia in a particular month.

Keywords: In-patient, Mortality, Morbidity, Binomial Distributions, Poisson Distributions