Are eggs good or bad for you?

Nearly everyone I ask is sick and tired of how often we nutrition experts seem to change our minds about whether eggs are good or bad for you. In the past several weeks, alone, two powerful studies have been published in major medical journals. Of course they contradict each other. One tells us that eggs will prevent heart attacks and the other that they will kill you with a heart attack.

But a third study, also published in the past few weeks with enormous media attention, may once and for all answer the question of whether eggs are healthy or harmful. In it, they looked at people who skipped breakfast in the late 1980’s and early 1990’s and found that those who did had a higher risk of heart attacks.

{mosads}And since, according to the American Egg Board, more eggs are eaten at breakfast than any other meal, one might surmise that this proves that eating eggs reduces heart attacks.

The study, in fact, proves no such thing. But this is how far awry our conclusions can be when we try to attribute cause based on observation. Despite what the news says, none of these studies provide any information about whether eggs cause anything, or that skipping breakfast shortens your life. Each of these studies is designed for one thing only: to look for a correlation. Not to determine cause and effect.

I have no relationship with the egg industry. But I do have a conflict of interest in telling this story. The reputation of science, to which I highly identify, is under attack because of our inability to make durable recommendations.

Here is how the science works, or at least how it is supposed to work. There are two broad approaches to how we go about figuring out how things work. We can observe, or we can perturb. In observational research, we look for correlations between two phenomena. In clinical trials, we do different things to different groups and monitor the consequences. In these, we can randomly assign similar groups to receive a treatment and compare that group to those receiving a different treatment or a placebo.

Cause and effect can only be derived from a randomized trial. But, since nutritional effects are slow to occur, these studies take years to do, large numbers of willing participants and are very expensive. Further, since randomized trials require similar groups, the subjects are selected for certain characteristics and the results only really apply to those people who have these characteristics.

Since there is no profitable medication at the end of the study, we depend on the USDA and the NIH to fund these kinds of studies, but giving these agencies money for this kind of research has become quite unpopular on Capitol Hill. Moreover, getting people to make long-term changes in diet is very difficult, if not impossible, to achieve. We have a lot of excuses for not doing this kind of research and thus rely heavily on observation for our preventive recommendations.

It should be noted that what little randomized research there is for eggs demonstrates improved blood lipid (cholesterol, etc.) profile.

Sometimes, the relationship in an observational study is so strong as to be difficult to controvert. For example, there has never been a randomized human trial to test whether smoking causes cancer and emphysema. It would seem unethical because there doesn’t seem to be any lingering doubts that cigarettes cause these diseases. The observations that led to that conclusion were supported by animal research and validated by a reduction in smoking-related diseases concomitant with a reduction in smoking.

As ridiculous as it sounds, there are those that would still argue that, without a randomized human trial, the causal relationship between cigarettes and cancer and emphysema are technically unproven. But we also know that, using observation, rather than randomized trials, one could come to the conclusion that yellow fingers cause lung cancer.

Heavy smokers tend to have yellow fingers. Heavy smokers get lung cancer. Therefore, the concordance between yellow fingers and lung cancer being high, we can conclude that yellow fingers cause lung cancer.

But we know that this is not the case. Dr. Charles S. Cameron an important early voice in our understanding of the causes of cancer, summed it up this way, quoting a colleague: “If it has not been proved that tobacco is guilty of causing cancer of the lung, it has certainly been shown to have been on the scene of the crime.” The article, written in the Atlantic in 1956, is worth reading. It was published 20 years before warnings appeared on cigarette packages.

There are numerous examples in nutrition that are similar to smoking and seem incontrovertible. No one will argue that starvation will kill you, vitamin D supplementation in milk all but eliminated rickets and adding iodine to salt wiped out goiter.

But the memory of these few examples where there were huge impacts on public health through nutrition have left us looking for the next big thing. While we treat each of these conflicting egg stories as if we’ve just saved millions from heart attacks, they are not the kind of earth shattering result we would like. Moreover, the road is strewn with discarded nutritional ideas that didn’t work, or were even harmful. You simply need to look at the National Cancer Institute’s website to find data from a large randomized trial that says that antioxidants cause cancer.

Conclusions based on observation are highly subject to bias. For example, the lack of media attention to the “eggs reduce heart attacks” study caused me a great deal of concern. I was unable to find a single mention of it in the media. This, compared to the doom and gloom reports about the other, suggest a significant bias towards stories that eggs are bad. In fact, the news reporting all but overtly suggests that the ”raft of studies reassuring consumers that eggs are OK to eat” had been struck down by this one re-analysis, which needed the combined results of six studies to find a correlation strong enough to be statistically significant. Similarly, researchers will, usually unconsciously and without malice, apply their own biases, not only to their findings, but to the way the study is designed.

Reliance on observation has resulted in some important big effect benefits in the past and is still very important to monitor the health of the populace, but is clearly not adequate for fine tuning our recommendations.

Meanwhile, we scientists help the media confuse the medical information consumer and we destroy our credibility in the process, by refusing to educate the public on the weaknesses of our methods. We need to make some hard decisions about how we do our research if we really want to know what to eat, or whether it really matters and insure we’ll be listened to when we do.

David S. Seres M.D. is the director of medical nutrition, associate professor of medicine and associate clinical ethicist at Columbia University Irving Medical Center, New York and chair of the Medical Nutrition Council and member of the board of directors of the American Society for Nutrition. The views above are his own.