Modelling of smear positive tuberculosis (PTB+) treatment outcome data in Kenya for the period 2002-2007

Abstract

In this research project, we review and use two statistical methods namely poisson regression and multinomial logit regression are used to model rates of outcomes of count data. Thestudy reviews the research work on modelling of count data particularly infectious diseases. There has been wide use of the two methods in literature but little application to Tuberculosis modelling. The outcome of smear positive tuberculosis cases namely cure, treatment completed, out of control, failures, transfer outs and deaths were modelled using poisson regression and multinomial regression. The results showed that the two methods can be used to model epidemiological behaviour of the infectious Tuberculosis and the two methods can be adopted at the national and provincial levels totrack the occurrence of the outcomes over the years and across the provinces. The results can be used to track performance of country in terms of the occurrence of treatment outcomes and the models can be used in provinces and districts. The models can act as strong epidemiological analysis tools.