JOINT LONGITUDINAL AND SURVIVAL MODELLING OF HIV IN THE UPPER WEST REGION

Abstract

The main aim of this research was to develop a joint longitudinal and survival model for HIV patients in the Upper West Region. The research covers a period of 8years starting from January 2006 to December 2014 with a total population of 119 HIV patients who were on two different ART regimens. Ninety-one (76.5%) of the population were females and 28 (23.5%) were males, of which 29 (24.4%) were alive, 78 (65.5%) were lost to follow-up and 12 (10.1%) experienced the event (death). Multivariate test used to study the pattern of change of CD4 count showed that the pattern of change of mean CD4 count for alcohol drinking patient and non-drinking patients were insignificantly different as well as the case for cigarette smokers and non-smokers. Pattern of change of mean CD4 count for educated and non-educated patients were also not significant. However, the pattern of change of mean CD4 count for married and unmarried patients were significantly different as well as that of males and females. The study revealed that AZT/3TC/NVP contribute better to patients survival time and CD4 count growth using weibull model and linear mixed effects model respectively. The two models further unveiled that factors such as preCD4 count, gender, and duration of treatment (months) significantly determines HIV patient’s CD4 count, whilst drug regimen, age and preCD4 determine the survival of patient. Results from the joint model indicated a significant association between the repeated CD4 count measurement and survival time of HIV patients.