Joint modeling of longitudinal CD4 count and time-to-death of HIV/TB co-infected patients: a case of Jimma University Specialized Hospital

Abstract

Tuberculosis (TB) and HIV have been closely linked since the emergence of AIDS; TB enhances HIV replication by accelerating the natural evolution of HIV infection which is the leading cause of sickness and death of peoples living with HIV/AIDS. Death is the serious problem that needs to be addressed so that maximum survival time can be obtained for the HIV/TB co-infection patients. Since the longitudinally measured CD4 count measurement is correlated with survival time joint modeling are used to handle the associations between these two processes to obtain valid and efficient survival time. Objective: To indentify factors affecting change in CD4 count over time; risk factors for the survival time and associate change in CD4 count over time and time-to-death processes of HIV/TB co-infected patients. Methods: The study consists of 254 HIV/TB co-infected patients who were 18 years old or older and who were on ART follow up from first February 2009 to fist July 2014 in Jimma University Specialized Hospital, West Ethiopia. First, data were analyzed using longitudinal and survival models separately. Then, based on the separate models several joint models with different random effects and different shared parameters have been explored and compared using deviance information criteria score. Results: The median survival time was estimated 62.5 months. The linear mixed model showed functional status; weight and time effects have significant effect on the CD4 count measurement process; Cox and Weibull survival model showed base line weight; baseline smoking; separated marital status group and base line functional status have significant effect on hazard function of the survival time whereas the joint model showed subject specific base line value; subject specific linear and quadratic slopes of CD4 count process significantly affects the survival time of co-infected patient at 5% significance levels. Conclusion: The longitudinally measured CD4 count measurement marker process is significantly associated with time to death and subject specific quadratic slope growth of CD4 count measurement; base line clinical stage IV and smoking is the high risk factors that lower the survival time of HIV/TB co-infected patients. Key words: survival analysis; longitudinal analysis; Cox PH; linear mixed model; joint modeling; HIV-TB