Whooh, lots to unpack here.

prospective population-based cohort study used data from 3 large cohorts: the Nurses’ Health Study (1990-2023), the Nurses’ Health Study II (1991-2023), and the Health Professionals Follow-up Study (1990-2023).

So the normal horsemen of weak science, tiny hazard ratios, low probability associations, relative risk only, observational, cohorts, food frequency questionaries, not clinically significant.

evidence pyramid

This is observational, which is hypothesis generating, not showing causation. We have healthy patient confounders, sugar and fructose confounders, etc.

Dietary intake was measured using a validated semiquantitative FFQ comprising more than 130 food items, administered at baseline and every 4 years

So every FOUR years people were asked to categorize what they had eaten in the last 4 years....

. Participants were initially categorized into quartiles of intake levels; however, due to the right-skewed distribution of these exposures, the sample sizes within each category were uneven. Therefore, we labeled the categories as levels 1 to 4 instead of quartiles 1 to 4.

They didn't like the data... so they reclassified people posthoc.

Major Issue: These FFQs don't take into account metabolic health. People on a Ketogenic LCHF, or a Zero Carb (Carnivore) diet absolutely do not have increased mortality. These FFQs can be sliced to show any association that you like, the fact they are mixing healthy patients (following guidelines) with people who don't follow any guidelines (including alcohol, sugar, fructose, etc) and seeing a overall health benefit from the guideline followers doesn't meat every single association is causative.

The Standard American Diet (SAD) is so absolutely BAD that any intervention looks good compared to it. Sadly a major component of the SAD diet is seed oils, which this paper is trying to promote. If they wanted to do real science they would hold seed oils constant in a study and add real butter on top, or have a 100% ASF fat diet vs 100% Seed oil diet (oh wait, we did that study already, they didn't like the data and hid it for 30 years - spoiler seed oils kill people)