A prospective cohort study across 13 sites in Benin and Togo (n=769) exploited a natural policy experiment: Benin provides in-kind food to all drug-susceptible TB patients; neighboring Togo does not. Recipients of nutritional support showed 57% higher odds of ≥5% weight gain by month 2 (aOR 1.57, 95% CI 1.13–2.19), 92% higher odds of ≥10% weight gain by month 6, a 40.6-meter greater improvement in 6-minute walk test distance, and dramatically higher medication adherence (aOR 3.43, 1.81–6.51). Critically, mortality was 68% lower among food-support recipients (aOR 0.32, 0.11–0.93).
Tuberculosis and undernutrition form one of medicine's most vicious cycles — TB worsens malnutrition, and malnutrition impairs immune function, enabling TB progression. Despite WHO acknowledging this bidirectional relationship, nutritional support has remained inconsistently integrated into national TB programs globally. This study's natural-experiment design, using Beninese non-recipients as an internal control, substantially strengthens causal inference compared to typical observational work — a meaningful methodological advantage. The adherence finding is particularly striking, suggesting food support may address a fundamental barrier to treatment completion beyond mere caloric supplementation, potentially reducing selection pressure for drug resistance. Limitations include the relatively short follow-up, lack of sputum conversion differences suggesting nutritional support aids recovery without accelerating bacterial clearance, and the observational architecture despite the clever design. As a preprint not yet peer-reviewed, these results require independent validation before informing policy, but the mortality signal is compelling enough to warrant urgent attention from TB program administrators.
