Things like wine is healthy, chocolate is healthy, berries are healthy, meat is unhealthy, eggs are unhealthy* all come from the weakest studies, 5 yearly food frequency questionnaires - where they all people to fill out how many serves of each food category the study cares about (the berry one listed about 12 direct berries out of the thirty or so that are popular; one meat one included “hamburger” in their meat category despite a burger meal usually having most its energy from the coke and fries).
The biggest problems include that the study can be warped several ways, the food list can change how people respond, the categorisation of foods changes everything, people can’t remember how many serves of whatever they had 5 years ago, so they pick the “most virtuous”, and after all that the most they can say is there might be an association between <food> and <health effect>
They report hazard ratios (how much the “wrong” food choice increases your risk of <health effect>) and usually get ~10%, everything outside nutrition needs 200% to say there’s an effect, cigarette smoking has a hazard ratio for lung cancer of 300% for a pack a day smoker**
*But “eggs are healthy” came from a study, feeding people eggs and testing their cholesterol, so if much more reliable
**The hazard ratios are usually reported as a fraction of 1 — 0.10 for many diet studies, or 3 for the smoker
Things like wine is healthy, chocolate is healthy, berries are healthy, meat is unhealthy, eggs are unhealthy* all come from the weakest studies, 5 yearly food frequency questionnaires - where they all people to fill out how many serves of each food category the study cares about (the berry one listed about 12 direct berries out of the thirty or so that are popular; one meat one included “hamburger” in their meat category despite a burger meal usually having most its energy from the coke and fries).
The biggest problems include that the study can be warped several ways, the food list can change how people respond, the categorisation of foods changes everything, people can’t remember how many serves of whatever they had 5 years ago, so they pick the “most virtuous”, and after all that the most they can say is there might be an association between <food> and <health effect>
They report hazard ratios (how much the “wrong” food choice increases your risk of <health effect>) and usually get ~10%, everything outside nutrition needs 200% to say there’s an effect, cigarette smoking has a hazard ratio for lung cancer of 300% for a pack a day smoker**
*But “eggs are healthy” came from a study, feeding people eggs and testing their cholesterol, so if much more reliable
**The hazard ratios are usually reported as a fraction of 1 — 0.10 for many diet studies, or 3 for the smoker