@gladstonebrookes.bsky.social
luv soy, luv seed oils, hate animal abuse. simple as
Typo in one of the summary tables, I think.
360 out of 1000 is 36%, not 3.6%, so the absolute risk difference should be 3% (not 0.3%), which is clinically significant under their criteria (so not "little to no difference" as is currently written).
360 out of 1000 is 36%, not 3.6%, so the absolute risk difference should be 3% (not 0.3%), which is clinically significant under their criteria (so not "little to no difference" as is currently written).
November 4, 2025 at 11:22 AM
Typo in one of the summary tables, I think.
360 out of 1000 is 36%, not 3.6%, so the absolute risk difference should be 3% (not 0.3%), which is clinically significant under their criteria (so not "little to no difference" as is currently written).
360 out of 1000 is 36%, not 3.6%, so the absolute risk difference should be 3% (not 0.3%), which is clinically significant under their criteria (so not "little to no difference" as is currently written).
They seem to have decided that the minimal clinically important difference is 5% absolute risk difference for MACE and 1% for mortality, irrespective of follow-up time.
Though I don't see an explanation for why these were chosen.
Though I don't see an explanation for why these were chosen.
November 4, 2025 at 11:14 AM
They seem to have decided that the minimal clinically important difference is 5% absolute risk difference for MACE and 1% for mortality, irrespective of follow-up time.
Though I don't see an explanation for why these were chosen.
Though I don't see an explanation for why these were chosen.
October 18, 2025 at 9:37 AM
You're right, I agree the p-values part is actually fine, and I fully retract that part of my criticism.
Sorry for the confusion.
Sorry for the confusion.
September 22, 2025 at 7:10 AM
You're right, I agree the p-values part is actually fine, and I fully retract that part of my criticism.
Sorry for the confusion.
Sorry for the confusion.
This guy's takes on nutrition include the belief that seed oils cause school shootings by weakening your connection to God, and that heroin overdose is because of seed oils.
September 21, 2025 at 1:30 PM
This guy's takes on nutrition include the belief that seed oils cause school shootings by weakening your connection to God, and that heroin overdose is because of seed oils.
Hey @gidmk.bsky.social, following this post, I dug into the study further, and it turns out there are a *lot* more issues with it.
Only one of the estimates used is actually what they claim it is (odds ratio for the association between highest vs lowest intake of cruciferous veg and colon cancer)..
Only one of the estimates used is actually what they claim it is (odds ratio for the association between highest vs lowest intake of cruciferous veg and colon cancer)..
September 21, 2025 at 9:08 AM
...there is apparent cherry-picking that biases the result towards finding a beneficial effect of cruciferous vegetable consumption, and the sensitivity analysis is full of incorrect p-values.
pubpeer.com/publications...
pubpeer.com/publications...
PubPeer - Cruciferous vegetables intake and risk of colon cancer: a do...
There are comments on PubPeer for publication: Cruciferous vegetables intake and risk of colon cancer: a dose–response meta-analysis (2025)
pubpeer.com
September 21, 2025 at 8:59 AM
...there is apparent cherry-picking that biases the result towards finding a beneficial effect of cruciferous vegetable consumption, and the sensitivity analysis is full of incorrect p-values.
pubpeer.com/publications...
pubpeer.com/publications...
My new favourite graph.
September 20, 2025 at 7:04 PM
My new favourite graph.
So it could be a (butter)fly on the wall?
September 1, 2025 at 7:02 PM
So it could be a (butter)fly on the wall?
Great paper!
August 28, 2025 at 9:45 PM
Great paper!
And always fun when one of those outcomes is just "Changes in the gut microbiota."
August 21, 2025 at 8:03 AM
And always fun when one of those outcomes is just "Changes in the gut microbiota."
Speaking of which, how exactly are they applying the Kruskal-Wallis test to something that has been transformed into count data?
August 2, 2025 at 5:56 PM
Speaking of which, how exactly are they applying the Kruskal-Wallis test to something that has been transformed into count data?
...And the fact that the variables are not normally distributed doesn't make this categorisation necessary, as there are non-parametric methods developed for this exact situation (which they clearly know about as they refer to the Kruskal-Wallis test).
August 2, 2025 at 5:56 PM
...And the fact that the variables are not normally distributed doesn't make this categorisation necessary, as there are non-parametric methods developed for this exact situation (which they clearly know about as they refer to the Kruskal-Wallis test).