Paperonium

COVID-19 · eLife · 2022

“COVID-19 hospital-stay odds 0.62 times as high, range 0.25–0.51.” A range must contain its result.

Summary, results andmain tablerange leaves out 0.62Its own chartdrawn range contains 0.620.62×0.25 to 0.51should be inside0.420.920.2×0.3×0.5×0.7×1×no difference
Summary,results andmain tablerange leavesout 0.62Its own chartdrawn rangecontains 0.620.62×0.25 to 0.51should be inside0.420.920.2×0.5×1×no difference

Times the odds of a COVID-19 hospital stay, per step up in gene-predicted IGF-1

The paper says

In its summary, results and main table, the paper says people whose genes give them more IGF-1, a growth-related hormone, have lower odds of a COVID-19 hospital stay: 0.62 times the odds, with a range (values the data can’t rule out) of 0.25 to 0.51.

Its own data show

The paper says chance alone would give a result this strong 1.8 times in 100; with a result of 0.62, that fixes the range at about 0.42 to 0.92. The paper’s own chart draws the range there too, though it labels it 0.25 to 0.51.

A range is built around its result, so it must contain it. As printed, this one says higher IGF-1 goes with at least 49% lower odds of a hospital stay; the paper’s own numbers allow as little as 8%. The 0.62 and the paper’s conclusions stand, but the summary makes the effect look bigger and surer than its data show.

“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 →

Genetically predicted high IGF-1 levels showed protective effects on COVID-19 susceptibility and hospitalization: a Mendelian randomisation study with data from 60 studies across 25 countries · eLife (2022) · doi:10.7554/eLife.79720