Causes of death and associated factors over a decade of follow-up in a cohort of people living with HIV in rural Tanzania

BACKGROUND: Nearly half of HIV-related deaths occur in East and Southern Africa, yet data on causes of death (COD) are scarce. We determined COD and associated factors among people living with HIV (PLHIV) in rural Tanzania. METHODS: PLHIV attending the Chronic Diseases Clinic of Ifakara, Morogoro are invited to enrol in the Kilombero and Ulanga Antiretroviral Cohort (KIULARCO). Among adults (>/= 15 years) enrolled in 2005-2018, with follow-up through April 2019, we classified COD in comprehensive classes and as HIV- or non-HIV-related. In the subset of participants enrolled in 2013-2018 (when data were more complete), we assessed cause-specific mortality using cumulative incidences, and associated factors using proportional hazards models. RESULTS: Among 9871 adults (65% female, 26% CD4 count < 100 cells/mm(3)), 926 (9%) died, among whom COD were available for 474 (51%), with missing COD mainly in earlier years. The most common COD were tuberculosis (N = 127, 27%), non-AIDS-related infections (N = 72, 15%), and other AIDS-related infections (N = 59, 12%). Cardiovascular and renal deaths emerged as important COD in later calendar years, with 27% of deaths in 2018 attributable to cardiovascular causes. Most deaths (51%) occurred within the first six months following enrolment. Among 3956 participants enrolled in 2013-2018 (N = 203 deaths, 200 with COD ascertained), tuberculosis persisted as the most common COD (25%), but substantial proportions of deaths from six months after enrolment onwards were attributable to renal (14%), non-AIDS-related infections (13%), other AIDS-related infections (10%) and cardiovascular (10%) causes. Factors associated with higher HIV-related mortality were sex, younger age, living in Ifakara town, HIV status disclosure, hospitalisation, not being underweight, lower CD4 count, advanced WHO stage, and gaps in care. Factors associated with higher non-HIV-related mortality included not having an HIV-positive partner, lower CD4 count, advanced WHO stage, and gaps in care. CONCLUSION: Incidence of HIV-related mortality was higher than that of non-HIV-related mortality, even in more recent years, likely due to late presentation. Tuberculosis was the leading specific COD identified, particularly soon after enrolment, while in later calendar years cardiovascular and renal causes emerged as important, emphasising the need for improved screening and management.