Clinic-based diabetes screening at the time of HIV testing and associations with poor clinical outcomes in South Africa: a cohort study

BACKGROUND: HIV clinical care programs in high burden settings are uniquely positioned to facilitate diabetes diagnosis, which is a major challenge. However, in sub-Saharan Africa, data on the burden of diabetes among people living with HIV (PLHIV) and its impact on HIV outcomes is sparse. METHODS: We enrolled adults presenting for HIV testing at an outpatient clinic in Durban. Those who tested positive for HIV-infection were screened for diabetes using a point-of-care hemoglobin A(1c) (HbA(1c)) test. We used log-binomial, Poisson, and Cox proportional hazard models adjusting for confounders to estimate the relationship of diabetes (HbA(1c) ≥ 6.5%) with the outcomes of HIV viral suppression (< 50 copies/mL) 4–8 months after antiretroviral therapy initiation, retention in care, hospitalization, tuberculosis, and death over 12 months. RESULTS: Among 1369 PLHIV, 0.5% (n = 7) reported a prior diabetes diagnosis, 20.6% (95% CI 18.5–22.8%, n = 282) screened positive for pre-diabetes (HbA(1c) 5.7–6.4%) and 3.5% (95% CI 2.7–4.6%, n = 48) for diabetes. The number needed to screen to identify one new PLHIV with diabetes was 46.5 persons overall and 36.5 restricting to those with BMI ≥ 25 kg/m(2). Compared to PLHIV without diabetes, the risk of study outcomes among those with diabetes was not statistically significant, although the adjusted hazard of death was 1.79 (95% CI 0.41–7.87). CONCLUSIONS: Diabetes and pre-diabetes were common among adults testing positive for HIV and associated with death. Clinic-based diabetes screening could be targeted to higher risk groups and may improve HIV treatment outcomes.