https://www.thelancet.com/article/S2468-2667(18)30135-X/fulltext
Dietary carbohydrate intake and mortality: a prospective cohort study and meta-analysis
My take on the study, it is a meta-analysis (a subjective interpretation) of observational studies (a subjective selection) based on dodgy methods such as food frequency questionnaires. At best it provides a hypothesis but it tells us nothing.
How good is an estimate of actual foods eaten based on a dietary questionnaire at enrollment?
This suffers from the fundamental FFQ problems - people donât remember, and people lie (not even maliciously, they just guess wildly and make stuff up).
How can we make valid inferences from self-reported recall data gathered at very long intervals?
I remember one study that came out last year, in which the food frequency questionnaire had been administered once during the study, and that was something like 25 years before the results were tabulated.
I hope the studies in this metanalysis were better-conducted than that one.
This tells you all you need to know that this is worthless:
Procedures
Participants completed an interview that included a 66-item semi-quantitative food frequency questionnaire (FFQ), modified from a 61-item FFQ designed and validated by Willett and colleagues16, at Visit 1 (1987â89) and Visit 3 (1993â95). Participants reported the frequency with which they consumed particular foods and beverages in nine standard frequency categories (extending from never or less than one time per month, to six or more times per day). Standard portion sizes were provided as a reference for intake estimation, and pictures and food models were shown to the participants by the interviewer at each examination. We used the Harvard Nutrient Database to derive nutrient intakes from the FFQ responses.16
I guess two FFQs in 4-8 years is better than two in 25 years? But âbetterâ is relative. Itâs better to get stabbed than stabbed, shot, and thrown off a building - but that doesnât mean that itâs good.
I find it amazing that âscientistsâ feel like they can learn anything from this. My academic background is in the physical sciences, where you measure things, change just one input, and measure things again. And even then itâs very hard to discern cause and effect.
Iâm amazed that thereâs any signal in all the noise from this recalled, self-reported data. (And maybe there isnâtâŚ)
Nutritional science is very hard to do, as Gary Taubes points out. Some of the difficulties are insuperable. But that doesnât excuse the reliance on very high p-values and the âconclusionsâ drawn from very small observed effects. And even when the results are statistically significant, it doesnât guarantee that theyâll be clinically useful. A little more humility on the part of the researchers and less willingness on the part of journalists to overhype results would go a long way.
As for food-frequency questionnaires, they are a joke, but no one has been able to come up with anything better.
Food log. But since itâs a lot more work no one does it unless itâs in a controlled clinic. Even then, itâs a matter of days to weeks, not years. Relevant example: how many folks on this forum actually log their food intake daily? I doubt very many.
PS: When I say âlogâ I mean weigh and measure. Everything.
Good point ⌠and people here are probably much more diet, health, and nutrition-aware than the average. Itâs been made easier with apps like MyFitnessPal and so forth, but not necessarily more accurate. (âWas that steak 6oz or 8oz?â, etc)
Precisely. Regularity rules.
Thatâs why people can more accurately complete a PFQ than a FFQ.
And we are all familiar with the adage, carbage in, garbage out!