The question of how we should go about deciding what we should eat leads the Science Examiner to ask how we should go about deciding what we should research about what we should eat.

---

I went to a very interesting science event the other night: a talk by Christopher Gardner, a nutritional scientist from Stanford, entitled: A Solution to the Omnivore’s Dilemma: The Perspective of a Nutrition Scientist. (View talk at ScienceCafeSF.com)

The "Omnivore's Dilemma" is a reference to a problem posed by the journalist Michael Pollan: As omnivores, we humans could eat anything in principle, but what should we eat in practice? In his book, Pollan points out that a lot of people and corporations have a vested interest in consumers eating specific kinds of foods - or "food-like-products" made by the agri-industrial complex - foods that are not necessarily the best for us to eat from a health or nutrition perspective.

In his book "The Omnivore's Dilemma,"  Pollan explores a lot of the issues surrounding modern industrial food production, and arrives at some fairly general conclusions about what we should eat, such as: "eat food, not too much, mostly plants" and "don't eat anything that your great-grandmother wouldn't recognize as food."

That last suggestion pretty much eliminates everything in the middle of the typical supermarket (all the packaged stuff) - leaving the products on the periphery (e.g., produce, meat, and fish) as the only viable hunting ground. Even the stuff on the periphery should be treated suspiciously, according to Pollan, because that hamburger most likely came from a cow that was being fed things that are similar to the packaged stuff in the middle of the supermarket: you also have to be careful about what you eat eats.

Christopher Gardner, in his talk, posed the question: Why don't we have a solid answer to the question of what we should eat? Haven't scientists figured this out by now?

The answer, surprisingly, is a resounding "No!" Scientists have not made a lot of progress on this question.

Gardner's lucid discussion of why it is so difficult to answer this question gives us a peek into how science addresses questions about diet, health, and disease, and, more importantly what science can and cannot (or will and will not) be able to do in addressing these questions in the future.

Ultimately, it boils down to the fact that the relationship between food, nutrients, health, and disease is maddeningly complex. Add to this the fact that humans are incredibly diverse in their physiological responses to the things they are exposed to, and you get the result that, to a first approximation, there is no one diet that will be right for all people all of the time.

Gardner describes himself as a classic "feed 'em and bleed 'em" scientist - one who studies the impact of specific nutrients on humans by dividing a large population of people randomly into two groups, one of which is fed X and the other of which is fed Y (or a placebo) for a month, or a year, or ....,  then measures some specific physical quantity (cholesterol in the blood, chemicals in the urine, ...) and looks for a statistically significant difference between the two groups.

To illustrate, he put up the following slide:



"I fill in the red parts," Gardner said, "that's what I do for a living."

And therein lies both the power and the limitation of the basic tools of the scientific method: it lends itself brilliantly to a very small set of highly specific questions. One simply can't answer scientifically a question like: Which is better for your health - an Asian diet or the Mediterranean diet? "What are you going to do," asks Gardner, "feed 1,000 people the wrong diet for 40 years and see what they die of?"

Rather, a scientist would perform an experiment more like the following:  Pick a disease of interest (say heart disease). Take a specific diet (e.g., the Mediterranean diet), and pick one kind of food from that diet (perhaps even just one nutrient from one kind of food). Since you can't wait for 40 years, you identify a specific "risk factor" for the disease of interest (say level of LDL cholesterol in the blood.) Divide your sample population randomly into two groups, and apply the "gold-standard" clinical trial method shown above.

Since there will be a good deal of variance within each group in terms of physiological reaction, what else they  eat or consume during the study, how rigorously they obey the pill/placebo schedule, etc, the answer will be a quite highly statistical one about the relative impact or lack thereof of that specific component of that specific diet on that specific risk-factor for that specific disease for a non-existent average person if that person were to ingest that substance at that dose level for the rest of their life.

Often, after millions of dollars are spent to run such a clinical trial study, it will be found that there is no statistically significant result, or worse yet, that the tested substance increased, rather than decreased, the risk factor for the disease.

Hence the problem. It is incredibly difficult and expensive to answer even the simplest questions about diet and health in a rigorous and definitive way.

We are never going to know which diet - Asian or Mediterranean - saves more people from heart disease in the long run because it is simply too difficult to pose that question in a scientifically rigorous manner.

Ultimately, someone has to pay for those studies, and so it is often only the groups with the deepest pockets, and thus with the most vested interest in making a profit off of the outcome, that get to decide what questions related to diet and health get "well-posed" scientifically.

"Are the Broccoli farmers going to band together and fund a clinical trial on the relative benefits of doubling your intake of broccoli as compared with, say, red meat?" asks Gardner? "It ain't gonna happen."

This leads us to my twist on Michael Pollan's question, which I call "The Omniscient's Dilemma:" As scientists, we humans could ask any question in principle, but which scientific questions should  we ask in practice? 

Setting aside, for the sake of discussion, the expense and logistical difficulty associated with posing a particular question, what are the most important questions about diet and health that we should be posing scientifically today?

As in the case of the question of what foods we should eat, there are a lot of vested interests that act politically and financially to bias the questions that we do actually ask in practice. Not surprisingly, the primary funders of large, expensive clinical trials are the big pharmaceutical companies. They are naturally going to be more interested in discovering drugs they can sell to treat chronic diseases than they are in identifying dietary choices consisting of readily available foods that will prevent us from getting the chronic diseases in the first place.

Could we do more of these latter sorts of studies? Of course. But unless we are able, as concerned citizens, to have more of a say in how the government spends our tax dollars on scientific research, in Gardner's words: "It ain't gonna happen."