When to Break the Fast?

This is the graphical take  home message in a recent look at the National Health and Nutrition (NHANES) cohort recently published in JACC. Does this move us forward in answering the question – should we eat breakfast?

Short answer: Not really.

Why not? Because association does not imply causation. Even if the association is supported by common sense or confirms your suspicions, biases, or dogma.

Observational evidence has been often cited to promote breakfast eating in schoolchildren – kids who eat breakfast tend to do better in school (in some studies). It is easy to cook up good reasons why this might be true – and to then advocate for breakfast eating.

However, it is also easy to imagine lots of other reasons why kids who eat breakfast do better in school that have nothing to do with eating (socio-economic status, education level of parents, proximity to school, work status of parents, structure of the household, etc,)

The study published this week in JACC adds weight to the association between breakfast skipping and cardiovascular disease and stroke. In the big picture, cohort based sub-studies like this NHANES piece are exploratory. They help us see signals in very broad strokes.  When coupled with knowledge of physiology, other observational studies, and experimental evidence, they can make compelling arguments for more directed research. There are tons of specific questions that beg to be asked about people who do or don’t eat breakfast:

• Are they shift workers?
• Do they eat a late supper or a midnight snack?
• What is their socio-economic status, and their education level?
• Do they eat alone or with others?, and most bluntly:
• Why, in their subjective opinion, don’t they eat breakfast?

 Why don’t the researchers ask all of these questions and more? No doubt they would like to, but they have another 100 issues that they care just as much about, and there is a limit to how many hours patients will take to fill out all of the surveys. Moreover, as interesting as such questions could be, it only makes it a more detailed observation. The strength of the association may be more precise, but causation is never fully affirmed based on observational studies – to cite axiom #1 of literature review: “More research is needed.”

And what we want is research that will move us closer to causation- i.e. evidence from categories 2-5 below:

1. Epidemiological signals such as the one that this study highlights.
2. Biological plausibility (I would include animal research in this category)
3. Natural Experiments
4. Trials looking at surrogate outcomes (often risk factor profiles)
5. Trials looking out clinical outcomes that matter (these are ultra-rare because the outcomes you care about can take decades to occur)

Not surprisingly, the authors of this NHANES based study use their findings to promote breakfast eating: … these findings, writes Wei Bao, MD, PHD, of the University of Iowa – and a lead author of the study … “underscore the importance of eating breakfast as a simple way to promote cardiovascular health and prevent cardiovascular morbidity and mortality,” In contrast, the Editorialist, urged caution in the over interpretation of these observational findings. Why is caution the right approach, even if the observations mesh with common sense?

There are two reasons:

1. Breakfast skippers are different than breakfast eaters in many ways (this is also a well documented observation). Although valiant analytic efforts are made to account for these differences, it is highly unlikely that you could even name them all, much less perfectly adjust for them, and
2. Because the cost of being wrong could be onerous, and you may not realize it for decades, if ever. Before you scoff at that statement consider what happened with hypertension. When observational studies informed us that hypertension was associated with cardiovascular disease we immediately hopped on the bandwagon, and pharmacological treatment of hypertension became the standard of care. However, it took us a couple of decades to really confirm that this was a good idea – ie reduced outcomes – and then only if you used the “correct” treatments. It turns out that some of the treatments had side effects that increased cardiovascular and other risks. Sometimes the pill really is too bitter to be swallowed.