Abstract

Background Despite being key components of global tuberculosis policy, poverty reduction and social protection interventions have been inconsistently implemented. We aimed to characterise how poverty and interrelated personal risk factors increase tuberculosis risk in Peru to inform the design of locally appropriate, person-centred, equity-oriented interventions. Methods We undertook a case-control study among people aged 15 years and over in 32 communities in Peru between 2016 and 2019. Cases (n = 2337) were people diagnosed with any form of tuberculosis. Controls (n = 981) were people living in randomly selected households in the same communities. We derived measures of household poverty from three dimensions (physical, human, and financial capital) and investigated the associations between these; personal risk factors more specifically linked to health (e.g. smoking); and tuberculosis. We used logistic regression to calculate adjusted odds ratios (aOR), 95% confidence intervals (95% CI), and population attributable fractions (PAF). A directed acyclic graph was used to inform the analytical approach. Results Household poverty was strongly associated with tuberculosis (aOR = 3.1; 95% CI: 2.3–4.2 for people from the ‘poorer’ versus ‘less poor’ half of households). There was a non-linear social gradient across deciles of household poverty, with odds of tuberculosis increasing exponentially as poverty deepened (aOR = 12.6; 95% CI: 6.8–23.2 for the ‘poorest’ decile versus the ‘least poor’ decile). Overall, tuberculosis burden could be halved by reducing poverty in the ‘poorer’ half of households to the level of the ‘less poor’ half (PAF = 47%; 95% CI: 40–54). For key personal risk factors, we estimated PAF for alcohol excess (PAF = 12.3%, 95% CI: 7.2–17.2); underweight (PAF = 10.3%, 95% CI: 8.7–11.8); smoking (PAF = 8.8%, 95% CI: 3.8–13.5); HIV (PAF = 5.7%, 95% CI: 4.6–6.7); and diabetes (PAF = 4.6%, 95% CI: 3.3–6.0). We also identified other important risk factors including previous tuberculosis (PAF = 14.8%, 95% CI: 11.6–17.9); incarceration (PAF = 9.5%, 95% CI: 6.8–12.1); and lower social capital (PAF = 4.1%, 95% CI: 2.6–5.6). Most personal risk factors, particularly education and substance misuse, tuberculosis exposures (e.g. incarceration and homelessness), and undernutrition, exhibited a social gradient across quintiles of household poverty and were more prevalent in people living in poorer households (Cochran-Armitage test for linear trend P < 0.001 for variables showing these social gradients). Conclusions Interventions addressing multidimensional household poverty and interrelated personal risk factors could substantially reduce tuberculosis burden. Our results provide an evidence base for designing person-centred, equity-oriented interventions; and support more effective implementation of poverty reduction and social protection within the global tuberculosis response.