TUBERCULOSIS RISK FROM LOW BODY MASS INDEX, DIABETES, AND THEIR CO-OCCURRENCE IN LOW AND MIDDLE-INCOME COUNTRIES: AN INDIVIDUAL AND POPULATION-LEVEL EPIDEMIOLOGICAL ASSESSMENT

Friday, January 8, 2016
Foyer, G/F (Jockey Club School of Public Health and Primary Care Building at Prince of Wales Hospital)

Lea Prince, MA, PhD, Centers for Health Policy and Primary Care and Outcomes Research, Department of Medicine, Stanford University, Stanford, CA, Jason Andrews, Stanford University School of Medicine, Stanford, CA, Sanjay Basu, MD, PhD, Stanford University, Stanford, CA and Jeremy D. Goldhaber-Fiebert, PhD, Stanford Health Policy, Centers for Health Policy and Primary Care and Outcomes Research, Department of Medicine, Stanford University, Stanford, CA
Purpose: Globally, tuberculosis (TB) prevalence has declined over the past decade, but its risk factors have varied over time and across populations. We sought to understand the relationship between TB and two key risk factors—the persistent risk factor of low body mass index (BMI) and the increasingly prevalent risk factor of diabetes. We examined the relationship at both the individual and population levels to support projection of future TB trends and efforts to target TB control to higher risk groups.

Method(s): We analyzed two datasets describing self-reported TB, diabetes, and BMI: India’s National Family Health Survey (NFHS) wave 3, (n=184,733) and the World Health Survey (39 low- and middle-income countries; n=129,193). Multivariate logistic regressions assessed the individual-level relationship between TB, diabetes, and low BMI while accounting for other traditional TB risk factors. We estimated regression coefficients with and without diabetes/low BMI interaction terms to assess whether risk factor co-occurrence further elevated TB risk. We performed a population-level analysis examining how TB incidence and prevalence varied with the prevalence of diabetes/low-BMI co-occurrence.

Result(s): In NFHS, the multivariate model that assumed independence of diabetes and BMI as TB risk factors predicted a TB risk for individuals with diabetes that was always higher than those without diabetes at similar BMI levels (diabetic: 2.50% at low BMI; 0.81% and normal BMI; 0.37% at high BMI; non-diabetic: 0.63% at low BMI; 0.20% and normal BMI; 0.09% at high BMI). There was no statistically significant difference in the predicted probabilities of TB when diabetes and BMI were interacted in a second multivariate model. Findings were similar in the WHS, though the BMI gradient was steeper in both diabetic and non-diabetic individuals, likely reflecting HIV and other unmeasured TB risk factors at lower BMI levels. The population-level analysis found that diabetes/low-BMI co-occurrence was associated with elevated TB risk, though given that the prevalence of co-occurrence is generally ≤0.5% its predicted effect on TB incidence and prevalence is <0.2 percentage points and not consistently statistically significant.

Conclusion(s): Concerns about the need to coordinate control efforts around the nexus of diabetes and low BMI co-occurrence may be premature as we find that while both are substantial risk factors for TB in low and middle-income countries, their interaction has not produced substantial excess burden.

See more of: SHORT-FORM ORAL PRESENTATIONS: SESSION 1

See more of: Making Difficult Clinical and Policy Decisions: The Example of Ageing and End of Life Care in Asia-Pacific