HIV‐associated morbidity and mortality in a setting of high ART coverage: prospective surveillance results from a district hospital in Botswana

INTRODUCTION: Antiretroviral therapy (ART) has significantly improved survival in Africa in recent years. In Botswana, where adult HIV prevalence is 21.9%, AIDS‐related mortality is estimated to have declined by 58% between 2005 and 2013 following the initial wide‐spread introduction of ART, and ART coverage has steadily increased reaching 84% in 2016. However, there remains little data about the burden of HIV and its impact on mortality in the hospital setting where most deaths occur. We aimed to describe the burden of HIV and related morbidity and mortality among hospitalized medical patients in a district hospital in southern Botswana in the era of widespread ART coverage. METHODS: We prospectively reviewed medical admissions to Scottish Livingstone Hospital from December 2015 to November 2017 and recorded HIV status, demographics, clinical characteristics and final diagnoses at discharge, death or transfer. We ascertained outcomes and determined factors associated with mortality. Results were compared with similar surveillance data collected at the same facility in 2011 to 2012. RESULTS: A total of 2316 admissions occurred involving 1969 patients; 42.4% were of HIV‐positive patients, 46.9% of HIV‐negative patients and 10.7% of patients with unknown HIV status. Compared to HIV‐negative patients, HIV‐positive patients had younger age (mean 42 vs. 64 years, p < 0.0001) and higher mortality (22.2% vs. 18.0%, p = 0.03). Tuberculosis was the leading diagnosis among mortality cases in both groups but accounted for a higher proportion of deaths among HIV‐positive admissions (44.5%) compared with HIV‐negative admissions (19.4%, p < 0.0001). Compared with similar surveillance in 2011 to 2012, HIV prevalence was lower (42.4% vs. 47.6%, p < 0.01), and among HIV‐positive admissions: ART coverage was higher (72.2% vs. 56.2%, p < 0.0001), viral load suppression was similar (78.6% vs. 80.3%, p = 0.77), CD4 counts were higher (55.0% vs. 44.6% with CD4 ≥200 cells/mm(3), p < 0.001), mortality was similar (22.2 vs. 22.7%, p = 0.93), tuberculosis diagnoses increased (27.5% vs. 20.1%, p < 0.01) and tuberculosis‐associated mortality was higher (35.9% vs. 24.7%, p = 0.05). CONCLUSIONS: Despite high ART‐coverage in Botswana, HIV‐positive patients continue to be disproportionately represented among hospital admissions and deaths. Deaths from tuberculosis may be contributing to lack of reduction in inpatient mortality. Our findings suggest that control of HIV and tuberculosis remain top priorities for reducing inpatient mortality in Botswana.