Fishy Dietary Advice

Researchers announced this week that the health benefits of eating fish outweigh the risks. But as far as the data indicate, the scales are empty – there appear to be no special health benefits or risks to weigh.

Harvard University researchers reported in the Journal of the American Medical Association (Oct. 18) that eating fish once or twice per week reduces the risk of heart attack by 36 percent and the risk of premature death by 17 percent, and appears beneficial for early neurodevelopment.

The researchers also reported that, while methylmercury in fish may “modestly decrease the cardiovascular benefits of fish intake [for adults]” and may “adversely affect early neurodevelopment [for children],” and dioxins and PCBs in fish may slightly increase cancer risk, the overall benefits of fish outweigh these risks.

These fishy claims ought to be thrown back.

First, this study contains no new evidence concerning the still-controversial health benefits of eating fish. The researchers merely collected all previous studies they could find concerning fish intake and health, and statistically combined the various results. While such statistical pooling may be a useful analytical technique for certain types of scientific study – such as carefully controlled clinical trials involving pharmaceutical products – its use is dubious with human population studies (epidemiology) where the researchers have little control over study subjects’ lifestyles and knowledge of their relevant medical histories.

Each of the individual fish studies is unique in terms of data collection and statistical methods – think of them essentially as apples and oranges – so that they can’t readily be mixed together to produce a meaningful result. The researchers seem to be hoping that the differences between the studies are randomly distributed so that when their results are combined the differences cancel each other out. Without empirical evidence of such canceling out, this amounts to wishful thinking rather than science.

Even giving their questionable methodology the benefit of the doubt, the researchers’ results are hardly persuasive statistically speaking. The rule of thumb with epidemiologic results claiming to show a reduction in health risk is that the reduction must be greater in size than 50 percent. Alleged reductions of 50 percent and less – particularly in studies that are not carefully controlled clinical trials – are very likely to simply represent statistical noise as opposed to valid results.

There are a number of other methodological criticisms that could be leveled at the Harvard study, but I will let the former editor of the Journal of the American Medical Association – the journal that published the study – comment on the slippery nature of dietary fish studies.

When asked by a New York Times reporter in 1998 about the dubious science behind a widely touted dietary fish study published in JAMA, then-editor Dr. George Lundberg responded by saying, “People are told that eating fish once a week is not a bad thing. What harm could it do?”

That attitude, apparently, still carries the day with JAMA’s editors.

Greenland Eskimos and other fish-eating populations may have lower rates of cardiovascular disease, as pointed out by the Harvard researchers, but other explanations – including genetics and other lifestyle differences – are more plausible explanations for such phenomena.

So what about the alleged “risks” of eating fish?

Certainly mercury in fish has been associated with neurological harm – based on a single extreme case of pollution in Minimata Bay, Japan during the 1950s. But that harm was associated with mercury in fish that was 10 to 30 times higher than the mercury level for fish currently permitted by the Food and Drug Administration. According to data presented in the Harvard study, none of the 29 types of fish and shellfish tested by the government for mercury approach any sort of danger level.

The Harvard researchers casually toss about allegations that exposure to PCBs and dioxins may increase cancer risk, and that PCB exposure in children is associated with impaired development.

These allegations are maddening, however, as the researchers made no effort to critically evaluate their sources. Alarm over PCBs and dioxins was started in the 1970s by environmentalists who exploited the fact that little if any scientific data concerning the health impacts of exposure to PCBs and dioxins existed – no doubt this data vacuum was because no one had any reason to suspect that such exposures were at all dangerous given the lack of real-world adverse impacts.

Thirty years and billions of dollars of research later, there still is no body of research (incidents of poisoning aside) credibly demonstrating that exposure to PCBs or dioxins increases the risk of any adverse health effect whatsoever.

As a result of the researchers’ failure to conclude that there are no actual risks from eating fish, widespread media stories about the study all give the incorrect impression that there are real risks to eating fish – although they are “outweighed by the benefits.”

At this point, the most that can be said is that fish and shellfish are tasty sources of protein, vitamins and minerals that have proven nutritional value. But the scientific data collected so far do not show that fish consumption wards off heart attacks, causes cancer or affects IQ.

Media reports say the study has pleased fish industry groups and outraged anti-industry environmentalists. Perhaps, but the study’s intended audience – consumers of fish – unfortunately have only been further misinformed and confused.