In order to synthesize the available evidence, the Harvard project convened a panel of experts, chaired by Steven Teutsch, a medical epidemiologist formerly with the U.S. Centers for Disease Control and Prevention and now at Merck and Company. Other panel members included David Bellinger (Harvard University), William Connor (Oregon Health Sciences University), Penny Kris-Etherton (Pennsylvania State University), Robert Lawrence (Johns Hopkins University), David Savitz (University of North Carolina), and Bennett Shaywitz (Yale University). The panel identified important health effects to consider, assessed the dose-response relationships between fish consumption (or its constituents) and health outcomes, and developed an overall health effects model. In addition to Joshua Cohen, Harvard scientific staff included Colleen Bouzan and Ariane König, and principal investigator, George Gray, executive director of the Harvard Center for Risk Analysis.

The study found that if pregnant women were to eat the same amount of fish but replace fish high in mercury with fish low in mercury, cognitive development benefits, amounting to about 0.1 IQ points per newborn baby, could be achieved with virtually no nutritional losses. However, if pregnant women were to decrease their fish consumption by one-sixth, the loss of omega-3 fatty acids during pregnancy would cut the nutritional benefit by 80%. If other adults were to also decrease their fish intake by one-sixth, then risks from coronary heart disease and stroke would increase. For example, among 65 to 74 year old men, the annual mortality risk would increase by nearly 1 in 10,000.

The study also found that increasing fish consumption among individuals who were not going to become pregnant would substantially decrease stroke and coronary heart disease risks. Much of this benefit appears to be associated with getting people to eat at least some fish (e.g., one meal a week), rather than no fish at all.

Cohen explained that the problem with fish advisories is that we do not know what their overall impact on the population might be. “Depending on how the population reacts, that impact could very well be negative.” Because of the potential downside, Cohen urges the government to carefully evaluate the pros and cons. He concluded, “Before the government issues advisories, it needs to gather data on how people actually will react, how those changes in behavior will influence nutrient intake and exposure to contaminants, and how those changes in intake and exposure will translate into changes in health. In other words, before we put an intervention into action, we need to estimate its real world impacts – both its benefits and its countervailing risks.”