Abstract

Background The relationship between nutritional status and tuberculosis is critically important but poorly understood. We extended a 2009 review characterizing the relationship between body mass index (BMI) and tuberculosis risk. Methods We systematically searched for new studies published between 2009 and 2024 investigating BMI and tuberculosis risk in adults. We extracted estimates of risk in BMI categories, used resampling to assign a median BMI ‘dose’ within each category, and included these in one-stage dose–response meta-analyses, stratifying results by population group and country tuberculosis burden. We fitted linear models for comparability with the 2009 review and restricted cubic spline models to investigate nonlinear relationships and piecewise linear models. Results Our analyses showed an inverse dose–response relationship between BMI and tuberculosis risk across all populations in the full underweight to obese range (15.0–35.0 kg/m2). The spline and piecewise linear models showed a nonlinear relationship—in 22 general-population cohorts (n = 24 921 531), there was a steep per-unit reduction in risk for BMI of <25.0 kg/m2 [18.0%, 95% confidence interval (CI): 16.4–19.6], which decreased more gradually for BMI of ≥25.0 kg/m2 (6.9%, 95% CI: 4.6–9.2). In 18 cohorts of people with HIV (n = 162 609), the reduction was 15.3% for BMI of <23.0 kg/m2 (95% CI: 13.1–17.5) and 2.6% (95% CI: –3.1–7.9) for BMI of ≥23.0 kg/m2. In three cohorts of people with diabetes (n = 1 118 424), the reduction was 20.5% for BMI of <24.0 kg/m2 (95% CI: 18.4–22.6) and 13.4% (95% CI: 3.9–22.0) for BMI of ≥24.0 kg/m2. Based on the global BMI distribution, we estimated a relative risk of tuberculosis associated with undernutrition (BMI < 18.5 kg/m2) of 5.0 (95% CI: 4.2–5.9). Conclusion Our results highlight the independent importance of nutritional status as a driver of the tuberculosis epidemic.