Objectives: To determine the incidence and risk factors of mortality for all HIV-infected patients receiving antiretroviral treatment at public and private healthcare facilities in the Botswana National HIV/AIDS Treatment Programme.

Design: We studied routinely collected data from 226 030 patients enrolled in the Botswana National HIV/AIDS Treatment Programme from 2002 to 2013.

Methods: A person-years (P-Y) approach was used to analyse all-cause mortality and follow-up rates for all HIV-infected individuals with documented antiretroviral therapy initiation dates. Marginal structural modelling was utilized to determine the effect of treatment on survival for those with documented drug regimens. Sensitivity analyses were performed to assess the robustness of our results.

Results: Median follow-up time was 37 months (interquartile range 11-75). Mortality was highest during the first 3 months after treatment initiation at 11.79 (95% confidence interval 11.49-12.11) deaths per 100 P-Y, but dropped to 1.01 (95% confidence interval 0.98-1.04) deaths per 100 P-Y after the first year of treatment. Twelve-month mortality declined from 7 to 2% of initiates during 2002-2012. Tenofovir was associated with lower mortality than stavudine and zidovudine.

Conclusion: The observed mortality rates have been declining over time; however, mortality in the first year, particularly first 3 months of antiretroviral treatment, remains a distinct problem. This analysis showed lower mortality with regimens containing tenofovir compared with zidovudine and stavudine. CD4 cell count less than 100 cells/μl, older age and being male were associated with higher odds of mortality.