Investigators pooled age-stratified results from population-based TB prevalence surveys and observed a consistent, steep rise in bacteriologically confirmed pulmonary tuberculosis prevalence with advancing adult age. The dataset comprised 82 surveys across 30 countries with roughly 4.6 million participants, according to the preprint. Data indicated that every 10-year increase in age corresponded to a 30% increase in prevalence (95% credible interval: 22–38%). Short, direct comparison: prevalence in those aged 75+ was reported as 796 per 100,000 versus 127 per 100,000 in the 15–19 group. Longer view: the oldest adults in these surveys had a 6.4-fold higher prevalence than the youngest adults, a pattern the authors found consistently across settings.
The study team used Bayesian meta-regression to estimate these age trends. They reported positive age gradients in all countries included, with an interquartile range for the percentage increase per 10-year age step of 23–40%. The authors also noted that age gradients declined slowly over the study period covered by the included surveys. The preprint provides the pooled effect estimates and credible intervals; detailed model specifications and priors are available in the manuscript and supplementary materials linked by the authors.
Data indicated heterogeneity by region and by the TB outcome definition. The authors reported that age gradients were lower in Africa compared with other regions. They also found smaller age gradients when analyzing smear-positive TB than when analyzing bacteriologically confirmed pulmonary TB. Short statement: outcome definition changed the magnitude of the age effect. Longer interpretation: where surveys relied on smear positivity alone, the age-related rise in measured prevalence was attenuated relative to analyses that included culture or other bacteriological confirmation methods, as the authors described in their stratified analyses.
The preprint describes these contrasts directly; the full survey-level diagnostic algorithms and any between-survey measurement differences are reported within the manuscript and supplementary files cited by the authors. The source did not report every survey’s detailed protocol in the abstract, so readers should consult the supplementary material for survey-by-survey methods and outcome definitions.
The authors examined how population-level HIV prevalence modified age trends. They reported that countries with HIV prevalence ≥5% showed flatter age gradients in TB prevalence above age 40. In other words, the typical increase in TB prevalence with older age was less pronounced in higher-HIV settings, according to the study team’s stratified meta-regression. This observation points to an interaction between HIV epidemiology and age-specific TB risk at the population level; the preprint presents this as a cross-country correlation assessed within the modeling framework.
The paper lists these subgroup results as part of sensitivity and stratified analyses. The authors noted regional differences and HIV-stratified patterns without suggesting causal mechanisms in the abstract; mechanistic explanations and further subgroup details appear in the full text and supplementary materials.
To assess case detection by age, the investigators calculated notification-to-prevalence ratios using age-stratified notification data linked to survey prevalence estimates. Data indicated a negative age gradient in these ratios: notification-to-prevalence ratios declined by 10% for each 10-year increase in age (95% credible interval: 7–14%). Short takeaway: older adults had lower notification relative to their modeled prevalence. Longer interpretation: the authors framed these lower ratios in older age groups as evidence consistent with poorer case detection or longer diagnostic delays among older individuals, as reflected by lower notifications per prevalent case.
The preprint provides these ratios and the modeled age effects; the abstract reports the percent decline per decade but does not include detailed country-level notification tables. Those tables and the methods used to align survey-derived prevalence with routine notification data are present in the manuscript and supplementary files the authors supplied.
The authors reported that age gradients in TB prevalence declined slowly over the study period spanned by included surveys. They presented pooled estimates that integrate surveys from 1993 through 2025, as described in the abstract. Short note: changes over time were gradual. Longer note: the paper treats temporal change as part of the meta-regression framework and reports an overall slow decline in the magnitude of age-related increases in prevalence.
Limitations stated within the preprint include the usual caveats for pooled survey analyses: heterogeneity across surveys, variable outcome definitions, and the observational, cross-sectional nature of prevalence surveys. The abstract does not enumerate every methodological limitation; readers should consult the manuscript and supplementary material for complete descriptions. The preprint has not been peer-reviewed, and the authors explicitly caution that the work should not be used to guide clinical practice until formal review is complete.
Unanswered clinical and programmatic questions that the source leaves open include the precise causal pathways driving older-age excess prevalence and reduced notification-to-prevalence ratios, the operational barriers that lead to poorer detection among older adults, and how different diagnostic algorithms across surveys may bias age comparisons. The manuscript contains more detail but does not resolve these questions in the abstract. Funders listed for the study include the Wellcome Trust, NIHR, and the U.S. National Institutes of Health, and the authors declared no competing interests.