Background

While factors that drive early mortality among people living with HIV (PLWH) initiating antiretroviral therapy (ART) in sub-Saharan Africa (SSA) have been described, less is known about the predictors of long-term mortality for those with ART experience.

Methods

PLWH and on ART attending two HIV treatment clinics in Moshi, Tanzania were enrolled from 2008 through 2009 and followed for 3.5 years. Demographic, psychosocial, and clinical information were collected at enrollment. Plasma HIV RNA measurements were collected annually. Cause of death was adjudicated by two independent reviewers based on verbal autopsy information and medical records. Bivariable and multivariable analyses were conducted using Cox proportional hazard models to identify predictors of mortality.

Results

The analysis included 403 participants. The median (IQR) age in years was 42 (36–48) and 277 (68.7%) participants were female. The proportion of participants virologically suppressed during the 4 collection time points was 88.5%, 94.7%, 91.5%, and 94.5%. During follow-up, 24 participants died; the overall mortality rate was 1.8 deaths per 100 person-years. Of the deaths, 14 (58.3%) were suspected to be HIV/AIDS related. Predictors of mortality (adjusted hazard ratio, 95% confidence interval) were male sex (2.63, 1.01–6.83), secondary or higher education (7.70, 3.02–19.60), receiving care at the regional referral hospital in comparison to the larger zonal referral hospital (6.33, 1.93–20.76), and moderate to severe depression symptoms (6.35, 1.69–23.87).

Conclusions

As ART coverage continues to expand in SSA, HIV programs should recognize the need for interventions to promote HIV care engagement for men and the integration of mental health screening and treatment with HIV care. Facility-level barriers may contribute to challenges faced by PLWH as they progress through the HIV care continuum, and further understanding of these barriers is needed. The association of higher educational attainment with mortality merits further investigation.