Survival And Prognostic Factors Of Hiv/Aids, Tb And Co-Infection In Pru District

ABSTRACT

Survival analysis is a method for analysing the occurrence of a given event. This research seeks to evaluate the survival of HIVIAIDS, TB and co-infected patients and to identify the major prognostic factors that influence their survival. In this study, survival data for HIV/AIDS, TB and HIVITB co infection patients were obtained from St. Mathias Hospital in Yeji, Pru District of Brong Ahafo Region of Ghana. The data was fitted using both the Cox model and accelerated failure time model. The AFT (Gamma) was the best model for HIVIAIDS and HIVITB co-infection survival data. However, the Cox proportional hazard model was best for the TB data based on the AIC and BIC values. The study revealed that none of the covariates significantly interact at 10% significance level. The diagnostic checks on the Cox-Snell residual plot of the gamma model shows that it has the best predictive power because it is closer to the bisector. Cox model revealed that the proportionality assumption was satisfied. The martingale residual plot of the continuous covariates indicate that for each of the covariates, the plot do not show trend and the resulting smoothed plots (LOESS) are approximately horizontal straight lines. This confirms that the martingale residual plots have a linear relationship with the survival time. Hence, the model is adequate. The HIVITB co-infection patients experienced the worse survival rate. The study deduced that Weight significantly determines the patient's survival among all the three categories. Therefore health authorities should be very cautious and pay much attention to patients who weighed below the minimum weights.