diet studies are pure bullshit, says new scientist

i also have serious concerns about RCTs, because the parameters you control for are too few, and may in fact be completely irrelevant.

---------- Forwarded message ---------
From: New Scientist Health Check <no-reply@newscientist.com>
Date: Sat, Feb 22, 2020 at 11:29 AM
Subject: Why you shouldn't get your diet advice from news headlines
To: 

Plus: a new way to save the fertility of women having cancer treatment and the violinist who played during her own brain surgery

Hello readers, This week, there were two studies in the news looking at the effects of food on our health. One claimed that a Mediterranean diet reduces frailty in old age, and the second that eating big breakfasts burns fat. Many of you have told me that you are interested in nutrition research, but I have a surprising confession for a biomedical journalist: I believe that the vast majority of published studies looking at diet can't be trusted. I'm not the only one saying this. A growing number of statisticians argue that most studies done in this area are of poor quality. To understand why, let's first look at the best kind of research, the randomised controlled trial. This is where two groups of people are randomly allocated either to take a medicine or a dummy pill that looks just like it, and the health outcomes are compared. When it comes to diet, this is time-consuming and expensive to do properly, and most trials have methodological flaws, such as being too small, short-lasting or measuring effects that are theorised to affect health, but we don't know for sure. The Mediterranean diet study, for instance, made its claims based on changes in people's gut bacteria, our knowledge of which is at a very early stage. The big breakfast study didn't show this strategy makes you lose weight, because it only lasted three days. Many nutrition studies aren't even in the form of a trial – researchers simply observe what people eat to see if certain foods correlate with health outcomes. The problem is that wealthier people tend to eat more of what are seen as health foods, and they also live longer for a whole host of other reasons. So these "observational" studies can also generate false conclusions. How do we know the field is unreliable? For a start, dietary studies are renowned for contradicting each other. That's why one day we see headlines saying red meat is bad for you, then the next day it's OK. Even where there is consensus that a certain food has an effect, the impact is often tiny next to other known lifestyle influences, like smoking or heavy drinking. This is the case for the idea that processed meat like bacon causes bowel cancer, for instance; the claimed effect is so small that if 100 people stopped eating bacon, there would be only one less heart attack. But the clincher for me is that when we put ideas about healthy eating to a proper test, in the form of a large good-quality randomised trial, they almost never stand up, as I found when I investigated this issue last year. Where does that leave those of us wanting to eat healthily? Some statisticians are calling for a pause on further studies on diet until these methodological problems can be fixed. Some sceptics even say that, in Western countries where few people have malnutrition, there may be no meaningful effects to find. Within the limits of moderation and common sense, the finer details of our diet really don't matter that much. It's a startling claim – let me know what you think of it!