Researchers led by Shaoqiang Li and Zihui Wang assembled data from 201 countries and territories spanning 1990 to 2021, drawing on the Global Burden of Disease Study, the World Health Organization Global Health Observatory, the World Bank, the Organization for Economic Co-operation and Development and the NCD Risk Factor Collaboration. Their broader screen covered 182 candidate exposures, of which 26 relevant to multidrug-resistant tuberculosis were reported in detail across the domains of air pollution, nutrition, sanitation, education and physical activity. Because the exposures of interest are difficult to measure reliably in individuals but are well recorded at the population level, the team used a two-stage ecological estimator that first adjusted each outcome for sex ratio, diabetes and HIV/AIDS, then regressed the adjusted residual on each candidate exposure in turn.

The global picture they describe is starkly uneven. Prevalence of multidrug-resistant tuberculosis ranged from 0.0003 to 0.2173 per thousand across countries in 1990, and by 2021 the spread had widened to between 0.0001 and 1.256 per thousand. The five countries with the highest prevalence in 2021 were Somalia, Eswatini, Eritrea, Lesotho and Mozambique, all in sub-Saharan Africa, while Slovenia, Andorra, Barbados, Iceland and San Marino sat at the bottom with rates below 0.0006 per thousand. Of an estimated 1.11 million prevalent cases in 2021, South Asia accounted for 56.2 percent and sub-Saharan Africa for 21.6 percent, followed by Southeast Asia, East Asia and Eastern Europe. Men consistently showed higher prevalence and mortality than women, a gap that widened with advancing age, and rates rose with age in both sexes.

The strongest associations in the analysis belonged to factors that had never before been linked to drug-resistant tuberculosis in the published literature. Unsafe sanitation topped the ranking, with a one-interquartile-range increase in exposure corresponding to 13.9 additional prevalent cases per 100,000 people, followed by unsafe water sources at 11.2 and polluting cooking fuels at 10.7. The same three factors led for mortality, at 2.87, 2.74 and 2.64 deaths per 100,000 respectively. Absence of handwashing facilities, household air pollution from solid fuels, iron deficiency, vitamin A deficiency and several dietary components, including diets low in fruits, vegetables and calcium, also emerged as previously unreported correlates. Air pollution and mean years of schooling, by contrast, had already been flagged in earlier studies.

The mechanistic reasoning behind these findings draws on individual-level evidence that the ecological design itself cannot test. Poor sanitation facilitates faecal-oral transmission of enteric infections, which may drive chronic gut inflammation and systemic immune suppression. Air pollution damages the respiratory mucosal barrier, impairs ciliary function and promotes oxidative stress, and laboratory work has shown that exposing human mononuclear cells to fine particulate matter reduces their capacity to control the growth of Mycobacterium tuberculosis and suppresses interferon-gamma production. Nutrition operates through two distinct pathways: malnutrition impairs macrophage function, T-cell activation and antibody production, increasing the likelihood that latent infection progresses to active disease, and undernutrition also alters the absorption and distribution of first-line tuberculosis drugs, a pharmacokinetic variability that by itself can select for resistance even in patients whose adherence is complete.

Correlation patterns across age and sex added further texture to the results. Associations grew stronger in successively older age groups, a trend that held for 25 of the 26 exposure-outcome pairs and is consistent with cumulative exposure over a lifetime and with the slower clearance of the tuberculosis bacterium reported in aged hosts. Correlations were higher in women than in men for 23 of the 26 pairs, yet on the absolute scale the increment per interquartile range of exposure was larger in men for 22 of the pairs, reflecting their higher baseline rates of disease. The demographic distribution of the burden itself remained stable across the three decades studied, with the excess in men plausibly reflecting greater occupational exposure, smoking and alcohol consumption, and clustered occupations such as mining and construction in high-burden settings.

A crucial control comparison qualified the interpretation. When the team repeated the analysis with drug-susceptible tuberculosis as the outcome, the associations were as strong or stronger for every factor except diet low in fruits. Standardized to each outcome’s own mean, the ratio of the multidrug-resistant to the drug-susceptible estimate was 0.78 for unsafe sanitation, 0.60 for polluting cooking fuels and 0.51 for vitamin A deficiency, while diet low in fruits stood at 1.04, the only factor associated more strongly with resistant disease. This pattern suggests that the identified conditions relate to tuberculosis burden as a whole rather than to drug resistance specifically, which carries significant implications for how control programmes are designed.

Sensitivity analyses tested how robust the findings were to socioeconomic confounding. After adjustment for gross domestic product per capita and health expenditure, six of the thirteen principal exposures remained significant for prevalence: unsafe sanitation, unsafe water source, absence of handwashing facilities, underweight, iron deficiency and diet low in fruits. The same six held when the Socio-demographic Index replaced the two economic covariates, and polluting cooking fuels additionally remained significant for mortality. Estimates were attenuated under this adjustment, with unsafe sanitation falling from 13.9 to 2.8 cases per 100,000 per interquartile range, but the authors note that these exposures are themselves strongly correlated with national income, at a median absolute Spearman coefficient of 0.86, so the adjustment removed much of the exposure’s own variation. Results varied by less than 13 percent across exposure windows of 5, 10, 15 and 20 years.

The regional decomposition reveals that the two areas carrying most of the burden face very different exposure profiles. South Asia, with 56.2 percent of prevalent cases, showed its largest excesses in diet low in fruits, underweight and iron deficiency, while sub-Saharan Africa, with 21.6 percent of cases but 32.2 percent of deaths, was furthest above the global median for absence of handwashing facilities, underweight and household air pollution from solid fuels. The authors emphasize that the same interventions would therefore not address both regions equally, and that because correlations were strongest in the oldest age groups, older adults in these two regions are where burden and exposure coincide most closely. Eastern Europe presents a telling exception: it carries 5.6 percent of prevalent cases with a population-weighted exposure excess of only 0.03 interquartile ranges, yet has the highest multidrug-resistant proportion of tuberculosis of any region, pointing to treatment and health-system factors that this design could not measure.

The authors are candid about the limitations inherent in an ecological study of this scale. Previous treatment history, treatment adherence and drug-susceptibility testing coverage, the factors with the largest established effects on resistance, have no complete country-year series across 201 countries over three decades. Associations identified through contrasts between countries cannot be read as effects in individuals, and both outcomes and many exposures are modelled estimates that share covariates and smoothing assumptions. National income also determines both the exposures and the capacity to detect resistant disease, warranting caution about directionality. Nevertheless, the six associations that survived socioeconomic adjustment are the least likely to be explained away by such confounding, and the findings align directionally with individual-level evidence, including a cluster-randomized trial in India in which nutritional supplementation among household contacts reduced tuberculosis incidence by 39 to 48 percent. The practical implication is that water, sanitation and clean-cooking programmes are already working on the very conditions most strongly associated with the resistant tuberculosis burden, and there is a case for planning tuberculosis control and environmental health interventions together rather than treating them as separate enterprises.

Subject of Research: Population-level environmental and nutritional factors associated with multidrug-resistant tuberculosis across 201 countries from 1990 to 2021

Article Title: Identifying overlooked factors associated with multidrug-resistant tuberculosis: an ecological study across 201 countries and regions, 1990–2021

Article References: Identifying overlooked factors associated with multidrug-resistant tuberculosis: an ecological study across 201 countries and regions, 1990–2021. (n.d.). https://doi.org/10.1016/j.jare.2026.10.028

Image Credits: AI Generated

DOI: Not provided

Keywords: multidrug-resistant tuberculosis, sanitation, air pollution, nutrition, Global Burden of Disease, cooking fuels, iron deficiency, vitamin A deficiency, handwashing, South Asia, sub-Saharan Africa, ecological study