Paperonium

Neurology · PLoS Biol · 2020

Smoking and MS: the summary says 1.61, the main text 1.16. An admitted typo, still unfixed.

The summary0.92 to 1.61, includes 1.5The main text, twice0.92 to 1.16, rules out 1.5Earlier studies of smokers1.5×, measured differently1.61×1.16×0.8×1×1.2×1.4×1.6×smoking makes no difference
The summary0.92 to 1.61,includes 1.5The maintext, twice0.92 to 1.16, rulesout 1.5Earlier studiesof smokers1.5×, measureddifferently1.61×1.16×0.8×1×1.2×1.5×2×smoking makesno difference

Times as likely to get MS with smoking: best guess (dot) and range (bar)

The paper says

The study uses genes that make people likelier to start smoking, instead of comparing smokers with non-smokers. Its summary says smoking makes multiple sclerosis (MS) 1.03 times as likely, with a range of 0.92 to 1.61, and suggests smoking does not raise MS risk.

Its own data show

The main text gives the range twice as 0.92 to 1.16; 1.61 is 1.16 with two digits swapped. In the reviewers’ comments, published with the paper, a reviewer asked whether 0.92 to 1.61 was right; the authors called it a typo for 1.16 and said they had fixed it. The published summary still says 1.61.

A range is the span of values the data can’t rule out: 0.92 to 1.16 means anything from 8% less likely to 16% more likely. The paper says its result contradicts earlier studies finding smokers 1.5 times as likely to get MS (a rough comparison, but the paper’s own). The correct range excludes 1.5; the typo lets it back in.

“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 →

Little evidence for an effect of smoking on multiple sclerosis risk: A Mendelian Randomization study · PLoS Biol (2020) · doi:10.1371/journal.pbio.3000973