Paperonium

Tuberculosis · Lancet Reg Health Am · 2026

All three tuberculosis numbers in the paper’s summary are bigger than in its own results.

More crowded homesThe summary3.25% more1.5 times as bigIts own results2.13% moreMore residents aged 60+The summary5.92% more1.3 times as bigIts own results4.41% moreMore people with primary careThe summary0.30% fewer1.2 times as bigIts own results0.25% fewer−1%0%1%2%3%4%5%6%
More crowded homesThe summary3.25% more1.5 times as bigIts own results2.13% moreMore residents aged 60+The summary5.92% more1.3 times as bigIts own results4.41% moreMore people with primary careThe summary0.30% fewer1.2 times as bigIts own results0.25% fewer0%2%4%6%

Change in tuberculosis deaths for each extra 1 in 100 homes or people

The paper says

Its opening summary says a town’s tuberculosis deaths rise 3.25% per extra 1 in 100 crowded homes (over two people per bedroom), 5.92% per extra 1 in 100 residents aged 60+, and fall 0.30% per extra 1 in 100 people served by local primary care.

Its own data show

The paper’s own results section, reporting the same analysis, says 2.13% more, 4.41% more and 0.25% fewer, and its main table agrees. The summary’s figures are about 1.5, 1.3 and 1.2 times as big. None of the paper’s other analyses, including extra ones published alongside it, gives all three.

These effects compound, like interest. If one São Paulo town has 10 more crowded homes in every 100 than another, the results’ figure means 23.5% more deaths there; the summary’s means 37.7%. For older residents: 54.0% versus 77.7%. Both versions agree on which way each factor pushes deaths, but the summary, where most readers get their numbers, overstates how much.

“The summary” means the paper’s abstract: the short summary printed at its top, which is the part most readers see.

Read the paper →Full technical write-up →

Socioeconomic dynamics and tuberculosis mortality in São Paulo: a population-level analysis using spatial modeling · Lancet Reg Health Am (2026) · doi:10.1016/j.lana.2026.101572