Paperonium

Tuberculosis · PLoS Med · 2019

The summary ties diabetes to spreading TB: its number is backwards, its evidence overstated tenfold.

The summarymeans less diabetesin spreadersIts own Figure 5more diabetes in spreaders0.19×0.021.15.14×0.936.450.01×0.1×1×10×100×no differenceless oftenmore often
The summarymeans less diabetesin spreadersIts own Figure 5more diabetesin spreaders0.19×0.021.15.14×0.936.450.01×1×100×no differenceless oftenmore often

Spreaders’ odds of diabetes as a multiple of others’, with the paper’s printed ranges

The paper says

The summary says diabetes is “a risk factor” for spreading tuberculosis (TB). Its number is a ratio, 0.19; above 1 means spreaders have diabetes more often than other patients, below 1 less often. Chance alone, it says, would produce this under 1 time in 333.

Its own data show

The paper’s Figure 5 compares 23 spreaders (patients judged likely to have infected others) with 61 other patients in the same groups of linked cases: 5 spreaders (22%) and 3 others (5%) had diabetes. It prints the ratio as 5.14, not 0.19, and chance alone would produce this gap about 1 time in 33.

The ratio compares odds: 5 diabetic spreaders to 18 without, versus 3 to 58 among others. Spreaders’ odds are about five times higher (Figure 5: 5.14); dividing the other way gives 0.19. The summary’s only named risk factor rests on a backwards number and a chance figure (under 1 in 333) that its own data put at 1 in 33.

“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 →

High-resolution mapping of tuberculosis transmission: Whole genome sequencing and phylogenetic modelling of a cohort from Valencia Region, Spain · PLoS Med (2019) · doi:10.1371/journal.pmed.1002961