We have all heard that eating fish is good for us, and the proposition is valid both empirically, and epidemiologically. But it begs a vital question just the same: good for us, compared to what?
What we know is that in general, people with diets that routinely include fish have better health outcomes than people with diets that don't. There are inevitably, however, systematic differences in such diets. People eating more fish may well benefit from fish-specific attributes, such as the omega-3 content of fatty fish, notably salmon, trout, and albacore tuna. They (we) may also benefit, however, because fish as a protein source supplants other, less salutary fare. Is the observable benefit principally derived from what the diet includes, or excludes? This is not an easy question to answer even when we wrestle with it directly, which to date we have mostly not done.
Fish is clearly good for us in the context of the prevailing, typical American diet -- propagating around the globe, alas -- in which the meat of pigs (i.e., bacon) is routinely garnish for the meat of cattle (i.e., burgers). Such meat-on-meat is further embellished as a matter of routine with everything from a bun of refined white flour (to say nothing of additions of not only rather copious salt, but also high fructose corn syrup), to ketchup (with more high fructose corn syrup), to such ominously vague concoctions as "special" sauce. Trading up in such context is not a challenge.
But what of the addition of fish to an already quite optimal vegetarian diet? Can we say if human health outcomes are demonstrably better with an optimized vegan diet, or an optimized pescetarian diet?
To date, to the best of my knowledge, we cannot.
To the extent that we mean what really matters most when we say "human health outcomes," that is not a problem we are likely to redress any time soon. What really matters most is the combination of longevity and vitality; life in years, and years in life.
To show beyond meaningful doubt that any contender for best diet laurels is truly "the" champion using such a metric is a daunting challenge. We know, for instance, that nutritional effects, and inchoate taste preferences, originate in utero. We know as well that diet quality continues to influence health quality across the full expanse of the lifespan.
To prove, therefore, that Diet A is "better" than Diet B, to say nothing of Diets C-Q waiting in the wings, likely requires the randomization of women early in their pregnancies to the competing options. Because health differences are apt to be small when variations on the theme of "optimal" are being compared, the sample size would have to be huge -- probably well into the thousands.
So, thousands of pregnant women would need to adhere to their assigned diet throughout their pregnancies, and then adhere to it some more during a required, standardized period of breast-feeding. Nutrients, after all, are transmitted through breast milk -- as are flavors.
The true study participants, the babies, only get into the act once they are weaned and start eating solid food. At that point, the babies need to adhere to the assigned diet, forever.
Since our study metric includes longevity, and since people eating (and living) optimally see their 100th birthday far more often than the rest of us, we should plan for our study to run the better part of 100 years. The challenges are by now self-evident, I trust. Leaving alone the logistical nightmare of managing perfect fidelity to assigned diets by thousands of mothers at first, and then their newborns, we are left with a study that will cost an oppressively staggering sum, and will outlive any of the investigators involved at the start.
Suffice to say, it has not been done, and I advise against holding your breath while waiting.
Of course, we might content ourselves with lesser metrics. We certainly could compare equitably high-quality versions of vegetarian and pescetarian diets in a relatively short-term randomized trial, looking at such measures as cardiac risk markers, inflammatory cytokines, body composition, and more. Not only could such a trial be done, but frankly, it should be. My lab is working on related protocols as we speak, so we might get into the game.
We have had limited forays in this direction, looking over the years at the health effects of various foods. We have reported, for instance, that daily egg ingestion produces no discernible harms in healthy adults, hyperlipidemic adults, or even adults with coronary artery disease. But of course, this egg intake is against the backdrop of a rather typical American diet. We may well be better off eating more eggs, and fewer doughnuts -- which was the thinking that motivated this line of inquiry in the first place.
My vegan colleagues, however, are quick to point out that the alleged "harmlessness" of eggs, and cholesterol, would disappear if superimposed on an optimal, plant-based diet at baseline. They may well be right, and it is a reminder to ask always about the attribution of effects to any given element of diet: compared to what?
In other cases, we have studied foods of such clearly established health benefit that our findings are of more reliably universal relevance. Nuts, for example, figure in all of the world's healthiest diets, and are consistently associated with an array of health benefits in epidemiologic research. In such context, our findings related to beneficial effects of walnuts in Type 2 diabetes and pre-diabetes may, perhaps, be spared the "compared to what?" proviso. The best of foods are good for us compared to just about anything.
By and large, though, my lab is much more about "how to" rather than "what to" when it comes to eating. We accept that the basic theme of salutary eating has been rather clear for decades, and that making far better use of what we already know is crucial, eminently possible, and much neglected. I thus hope other researchers take up the mantle for such studies as vegetarian versus pescetarian, because my team and I contend there may be a bigger fish to fry. Most of our time and energy is directed to a project all about putting what we do reliably know about diet and health to far better use for the benefit of people and planet alike.
There is, in general, something rather fishy about most claims regarding the single best way to eat. Namely, such tales tend to wag at us far more dogma than data. The matter is only decisively settled with research that for the most part has not been done, and at its logical extreme -- almost certainly never will be.
As for fish per se, I am quite convinced that eating them is good for us in general, although maybe not compared to every conceivable baseline diet. But increasingly, I am preoccupied that eating them is clearly not good for the fish, or the oceans. The environmental impacts of our dietary choices are profound, and should be of increasing concern to all of us with an affection for this planet, or our children for that matter.
Claims about any given food or diet that go beyond the evidence are, in essence, fish tales- notorious for exaggeration. They are unfortunate, as they tend to obscure the reliability, volume, and consistency of evidence underlying genuinely common ground. Let us make our stand there -- waders, optional.
Director, Yale University Prevention Research Center; Griffin Hospital
President, American College of Lifestyle Medicine
Founder, The True Health Initiative