The problem with nutritionists conducting a meta-analysis is that the results are irrelevant most of the time.

Diet is a complex subject, asking people about what they eat and getting relevant is a hard.

Anything except controlling what everyone eats for a span of 3–4 months, on a large and diverse sample, will yeld no significant result because people eat a lot of stuff and their bodies are very different…

Take, for example, an analysis of eating bread with one cohort of people eating ~50 grams of bread a day and the other ~500 grams a day. The whole study can be biased by the former cohort drinking more beer, or eating more cereals, or eating wraps.

The problem with “meta analyses” which are usually based on very broad dietary results is that these kind of things are not taken into consideration. They are correlation studies on small and poorly documented samples… and a correlation on a small samples based on weak data is just a coincidence.

You can create an epidemiological study that will say about anything that you want it to say.

Nutrition in the modern day and age is very much a word of mouth thing. The fact that not eating breed improve life quality is a fact that many people, coming from a variety of diets and backgrounds, have observed.

In the end, people will chose a diet that makes them feel well, not eating bread seems to work for a surprisingly large amount of people.