Sequencing TB in a Brazilian prison shows most transmission happens outside the cell

One of the unresolved questions in tuberculosis control is deceptively basic: does the bacterium spread mainly through prolonged close contact with a few people, or through many brief ones? The answer determines whether contact tracing — the backbone of TB programmes everywhere — can plausibly interrupt transmission.

A genomic study published in The Lancet Regional Health – Americas on 18 November tried to answer it in the setting where the stakes are highest: a prison system [s1].

The design

Researchers conducted genomic surveillance in a prison system in Central West Brazil, whole-genome sequencing Mycobacterium tuberculosis isolates and collecting detailed incarceration histories for each person [s1]. They then modelled transmission linkages as a function of different types of prison exposure, using genomic clustering as a proxy for transmission and controlling for multiple pairwise comparisons [s1].

They gathered incarceration histories for 595 individuals — mean age 31 (SD 8.5), 99% men — and sequenced 550 high-quality genomes from them [s1].

Using genomic clustering as a proxy for transmission is the standard approach and also the study's main structural limitation: two closely related genomes are consistent with transmission between those two people, or with both having been infected by an unsampled third.

What shared cell time does

The dose-response is clear. In a binomial model, a month-long increase in exposure to a person with tuberculosis inside a prison cell increased the odds of pairwise genomic clustering by 14% (odds ratio 1.14, 95% CI 1.09 to 1.19) [s1]. A six-month increase in exposure doubled the odds (OR 2.24, 95% CI 1.73 to 2.91) [s1].

So cell-sharing matters, and it matters proportionally to time.

Why that does not solve the problem

The second finding is the one with operational consequences. Most people with tuberculosis in this system had not one but many documented cell exposures: 83% (494 of 595) had at least one day-long exposure in a prison cell to another person with tuberculosis, with a median of eight unique such exposures (IQR 4 to 12) [s1].

And the system moves people constantly. Individuals with tuberculosis had been transferred a median of eight times (IQR 4 to 13) in the two years before diagnosis [s1]. That churn produces a highly connected contact network in which almost everyone is a plausible contact of almost everyone else.

The authors' interpretation is that while documented cell-level exposures explain a significant proportion of transmission, most transmission links occur outside prison cells — either from other contacts within the same prison, or from unreported or unsampled exposures [s1].

That is a difficult result for contact tracing. If the median person has eight candidate cell contacts and has been moved eight times in two years, and if most linked transmission happened somewhere other than the cell anyway, then tracing individual contacts is not a plausible route to interrupting the epidemic. The authors conclude in favour of prison-wide mass screening, tuberculosis preventive therapy, and structural interventions to reduce transmission risk in prisons and other congregate settings [s1].

The scale of what is at stake

Prisons are not a marginal contributor to TB in Latin America. A modelling study published in The Lancet Public Health in October 2024 calibrated dynamic transmission models to historical and contemporary data from Argentina, Brazil, Colombia, El Salvador, Mexico and Peru — countries holding roughly 80% of the region's incarcerated population and TB burden [s2].

It estimated that population TB incidence in 2019 was 29.4% higher (95% uncertainty interval 23.9 to 36.8) than would have been expected without the rise in incarceration since 1990, corresponding to 34,393 excess incident cases (28,295 to 42,579) across those countries [s2]. The transmission population attributable fraction for incarceration in 2019 was 27.2% (20.9 to 35.8) — higher than published estimates for HIV, alcohol use disorder or undernutrition [s2].

The same model projected that a gradual 50% reduction in prison admissions and duration of incarceration by 2034 would cut population TB incidence by over 10% in every one of the six countries except Mexico [s2]. The authors were explicit that their model excluded HIV, drug resistance, gender and age structure [s2], which is a real constraint on how literally those projections should be read.

Reading the two together

The modelling study says incarceration policy is a major driver of TB at population scale. The genomic study says that within a prison, transmission is diffuse enough that individual-level contact tracing will not contain it. Both point toward the same class of response — population-level screening, preventive therapy, and changes to the conditions and volume of incarceration itself — rather than toward better case-by-case follow-up.

The genomic study was conducted in one prison system in one Brazilian region, with 550 sequenced genomes, and its conclusions about unsampled transmission depend on how complete that sampling was [s1]. It was funded by the US National Institutes of Health [s1].