Tuberculosis Incidence and Body Mass Index – An Updated Systematic Literature Review and Global Analysis

Abstract

Background.  Undernutrition is the leading population-level determinant of tuberculosis, but the relationship between tuberculosis and nutritional status remains poorly understood. We refined and extended a 2009 review to establish the association between tuberculosis incidence and the full body mass index (BMI) range in key populations across high-income (HIC) and low and middle-income countries (LMIC).  Methods. We systematically searched Medline and Embase for new studies published between 2009 and 2024 investigating tuberculosis incidence and nutritional status in adults and extracted estimates of incidence and BMI distributions. To capture uncertainty, we randomly drew 1,000 datasets from these distributions. For each dataset, we fitted a log-linear model across included studies and calculated the mean reduction in tuberculosis incidence per one unit increase in BMI, with 95% uncertainty intervals (95%UI). Using random effects meta-analyses, we estimated the incidence rate ratio (IRR) and 95% confidence interval (95%CI) for underweight (BMI<18.5kg/m2) versus not underweight (BMI³18.5kg/m2). We stratified results by population group and country income status where possible. Findings. We screened 8,881 abstracts, reviewed 127 full-text articles, and included 42 studies describing 45 cohorts – 39 from our search and six from the 2009 review. We observed a consistent inverse log-linear association between tuberculosis incidence and BMI in the underweight to obese BMI range (15.0-35.0kg/m2). Using unadjusted rates, the mean reduction in tuberculosis incidence per unit increase in BMI was 12.3% (95%UI: 10.8-13.8) in 19 general population cohorts (n=24,917,920); 9.7% (95%UI: 7.3-12.3) in eight cohorts of people with HIV (n=134,031); and 16.1% (95%UI: 12.3-19.2) in two cohorts of people with diabetes (n=1,117,991). In the meta-analyses, the IRR for tuberculosis comparing underweight versus not underweight was 3.05 (95%CI: 2.46-3.77) in 17 general population cohorts (n=22,282,841); 2.80 (95%CI: 2.45-3.21) in 18 cohorts of people with HIV (n=160,222); and 4.11 (95%CI: 3.74-4.51) in three cohorts of people with diabetes (n=1,118,424). Results were consistent when using estimates adjusted for confounding. Interpretation. We demonstrated an inverse log-linear association between tuberculosis incidence and BMI across a wide range of nutritional status and populations, including LMIC, people with HIV and people with diabetes. Our findings demonstrate the importance of ensuring interventions to improve nutrition become an integral component of the global tuberculosis response.  Funding. National Institute for Health and Care Research (NIHR UK)