In nutritional epidemiology, it is a real challenge to set up well-controlled and randomized intervention studies. Nonetheless, such studies are needed to generate causal proof when the observational studies remain overly uncertain (as explained elsewhere on this website) and the mechanisms too hypothetical. It is extremely difficult, however, to control the eating behaviour of a sufficiently large group of people over a period that is long enough to monitor the development of chronic disease.
Therefore, achieving causal conclusions through controlled intervention studies can usually only be done within short periods and at the level of surrogate outcomes for chronic diseases, such as blood pressure and blood lipid markers, markers for inflammation or oxidative stress, and glycemic parameters. Such studies have been done to some degree and failed in most cases to show adverse effects on risk factors, even with high consumption of red meat, further underlining the inconsistency of observational associations.
