Diabetes Mellitus as a Risk Factor for Tuberculosis in HIV-Infected and Uninfected Populations

Tuberculosis (TB) is one of the leading infectious causes of morbidity and mortality worldwide. Diabetes and HIV are important risk factors that contribute to the persistently high TB incidence globally. However, the importance of diabetes on TB risk in key sub-populations is not well-established. Additionally, body mass index (BMI) affects both TB and diabetes risk, and may interact with diabetes to change the risk of TB. Finally, sub-clinical TB may also be an important contributor to TB-associated morbidity and mortality but is not well characterized. We used data from two observational cohorts to: (1) estimate the combined impact of diabetes low and elevated BMI on latent and active TB burdens of disease in southern India; (2) estimate the prevalence of diabetes, associated risk factors, and its impact on 12-month clinical outcomes among adults testing positive for HIV in South Africa; and (3) estimate the prevalence of sub-clinical TB in this same population of HIV positive adults in South Africa and associated demographic and clinical characteristics. We found that diabetes was not associated with LTBI in Pondicherry, India. Diabetes-BMI interaction for active TB was statistically significant on both the additive and multiplicative scales. Compared to participants without diabetes, the greatest risk of active TB disease associated with diabetes was among overweight/obese. The burden of TB attributable to diabetes was highest in the low BMI group, suggesting routine screening of low BMI diabetic patients for active TB could be worthwhile in this setting. Among people living with HIV in Durban, South Africa the prevalence of diabetes was moderate and associated with well-established, traditional risk factors. Limiting diabetes screening to BMI >25 kg/m2 would capture 85% of participants who screened positive for diabetes, suggesting this could be a useful tool in resource limited settings. Over 12 months of follow-up, screening positive for diabetes was associated with a higher hazard of death but not other HIV-related outcomes. The prevalence of sub-clinical TB was 1.1% and not associated with any of the demographic or clinical characteristics we investigated. A significant limitation to these findings was incomplete diabetes testing, which is logistically complex and expensive, highlighting an opportunity for diabetes testing strategies that can be more readily implemented worldwide. Additionally, our work provides support for prospective cohort studies of diabetes patients in more diverse settings. Successful clinical management of incident active TB or HIV in the context of increasingly common non-communicable diseases and limited resources remains an important area of research.