While numerous studies have reported an association between eating fish and better health, most such research is observational. Clinical trials involving food are usually short and small and look at markers such as lipids, endothelial function, and blood pressure. Long-term randomized clinical trials with hard endpoints like heart attacks and dementia usually use supplements compared to placebos. (In the case of fish oil supplements, results of clinical trials have been frustratingly conflicting or disappointing.) It’s nearly impossible to do a large long-term randomized controlled trial involving food (say, randomly tell thousands of people to eat fatty fish twice a week for several years, and thousands of others to eat no fish).
Unlike clinical trials, observational population studies comparing fish consumption and health outcomes can’t demonstrate cause and effect—only correlation. Such research also depends on people recalling what they’ve eaten and accurately filling out food questionnaires. Another challenge is that people who eat fish on their own may do other healthful things, such as exercise more, or avoid unhealthy habits like smoking; researchers try to control for these factors but “residual confounding” may persist.