Cause-specific hazards of antiretroviral therapy programmatic adherence, defaulting and mortality among HIV/AIDS patients who had attained CD4 count recovery after antiretroviral therapy initiation in South Africa

Background Interruptions in treatment due to non-adherence or defaulting may compromise sustained immunological recovery among HIV patients who achieved viral suppression after antiretroviral therapy (ART) initiation. This study identified determinants of time to HIV-related mortality, ART defaulting, and adherence among PLWH who achieved CD4 count recovery. Methods A retrospective cohort study analysed data from 726 HIV-positive patients, who achieved CD4 recovery between June 2004 and August 2013. Programmatic adherence, mortality, and defaulting were modelled as competing events to assess CD4 recovery trajectories, with outcomes mutually exclusive during follow-up. Programmatic adherence was defined as continuous engagement in care, measured by attending scheduled clinic visits and obtaining timely ART refills without interruptions of 90 or more consecutive days. The cumulative incidence function estimated probabilities and evaluated covariate effects, including age, sex, tuberculosis status, location, baseline CD4 count, and viral load. Bivariate cause-specific competing risk analyses were conducted for baseline characteristics, and variables with p < 0.20 were included in multivariable cause-specific hazard models. Results Gender significantly predicted treatment default, with females having a higher hazard of default than males (aCSHR = 2.161; CI: 1.182–3.958). A higher baseline CD4 count was associated with an increased hazard of defaulting (aCSHR = 1.005; CI: 1.000–1.010). Age was the only significant predictor of mortality; increasing age was associated with greater mortality risk (aCSHR = 1.091; CI: 1.036–1.149). Females had a higher hazard of achieving adherence than males (aCSHR = 1.314; CI: 1.099–1.571). Patients without tuberculosis had a lower hazard of adherence than those with tuberculosis (aCSHR = 0.570; CI: 0.468–0.694). Rural residence was associated with lower adherence compared to urban residence (aCSHR = 0.616; CI: 0.516–0.735). Conclusion Female sex and higher baseline CD4 count were associated with increased risk of defaulting, while older age predicted mortality following CD4 recovery. Adherence was associated with female sex, urban residence, tuberculosis co-infection, and viral load. Mortality estimates were limited by a few events, leading to imprecision in some covariates. Targeted retention strategies, strengthened rural health services, and monitoring of older patients are essential to sustain long-term treatment outcomes after immunological recovery.