Cox PH Regression and Homogeneous Semi-Markov Models for Identification of Risk Factors of HIV/AIDS Patients Taking HAART and Assessing the Disease Progression

Objective: The objectives of the study were to identify the determinant factors for survival time of Human immunodeficiency virus (HIV) infected patients treated with highly active antiretroviral therapy (HAART) and to observe the HIV progression of HIV infected patients in Hawassa City Adare hospital, Ethiopia. Material and Methods: This was a retrospective cohort study of 330 patients who started ART between 2008 and 2014 at Hawassa City Adare Hospital. Data were extracted from paper based medical records database and the survival of patients was estimated by the Kaplan-Meier method. Predictors of mortality were identified by Cox proportional hazards model and the progression of the AIDS patients by Homogeneous Semi-Markov Stochastic model. Results: Survival of patients was significantly related with age, sex, Tuberculosis (TB) status, HIV disclosure, functional status, substance use, baseline World Health Organization (WHO) clinical stage, baseline weight and baseline cluster of differentiation 4 (CD4) count. Results of the Cox PH model revealed that; older, TB co-infected, substance user patients, and patients with lower baseline CD4 count and weight had significantly higher risk of death or shorter survival time than their counterparts. The results of Homogeneous Semi-Markov Stochastic model indicated that AIDS patients in the first state of the disease had the highest survival probability as compared to the patients in the second, third and fourth stage of the disease. Conclusion: To minimize deaths, more attention should be given during the early phases of treatment of HIV/AIDS patients on HAART.