News for hackers. Stuff that matters.

A comment on Meat consumption and risk of 25 conditions: outcome-wide analyses of 475K people. Parent comment

Re:Meat consumption and risk of 25 conditions: outcome-wide ... by tankenmate on Wednesday March 03, 2021 @08:04AM (#26327775)
No, that is not what it said.

The summary says "higher BMI accounted for a substantial proportion of these increased risks suggesting that residual confounding or mediation by adiposity might account for some of these remaining associations"

That means that a substantial proportion of the increased risk is explained by a higher BMI. Unfortunately the summary doesn't say how much "substantial" is.

[ Parent | on HN ]
Re:Meat consumption and risk of 25 conditions: outcome-wide ... by bigfudge on Wednesday March 03, 2021 @05:11PM (#26335425)
To be specific, the regression models they use assume no measurement error in predictors. In this case BMI isn't even a direct measure of adiposity, so it's kind of obvious, but even if they had a direct measure of body fat the issue could remain because error in measurement of predictors would attenuate the ability of this covariate to eliminate confounding.
[ Parent | on HN ]