Last month, eggs were good for you. This month, according to a study making rounds on social media, they might increase your risk of cardiovascular disease. Next month, I'd bet money they'll be fine again. If you have followed nutrition science over the past two decades, you have probably developed a reasonable response to dietary headlines: ignore them entirely.
This is understandable, but unfortunate. Because buried beneath the contradictions, the clickbait, and the genuinely confusing methodology, there are real insights about food and health that could meaningfully improve your life.
The methodology problem
To understand why nutrition studies contradict each other, you first need to understand how most nutrition research actually works. The gold standard of medical research — the randomized controlled trial — is nearly impossible to conduct properly for nutrition. You cannot randomize 10,000 people to eat only broccoli for five years. So researchers rely heavily on observational studies that track what people eat and correlate it with health outcomes.
A 2018 analysis published in JAMA Internal Medicine examined 50 common dietary ingredients and found that virtually every food had been associated with both increased and decreased cancer risk across different studies (Schoenfeld & Ioannidis, 2013).
The food frequency questionnaire problem
Most observational nutrition studies rely on food frequency questionnaires — surveys that ask participants to recall what they ate. A validation study found that participants underreported caloric intake by an average of 30%, with obese individuals underreporting by up to 50% (Subar et al., 2015).
We are building much of our nutritional knowledge on data that is, to put it charitably, unreliable.
The confounding variable maze
People who eat more vegetables also tend to exercise more, smoke less, earn higher incomes, and have better access to healthcare. When a study finds that vegetable consumption is associated with lower heart disease risk, how much of that benefit comes from the vegetables themselves?
As biostatistician Sander Greenland has argued, residual confounding is an unavoidable limitation of observational research (Greenland, 2017). You can adjust for the variables you measure, but you cannot adjust for the ones you do not.
What we can actually trust
Despite these limitations, there are nutritional principles supported by evidence strong enough to have survived decades of scrutiny:
Dietary patterns matter more than individual foods. The PREDIMED trial enrolled 7,447 participants and demonstrated a 30% reduction in cardiovascular events from a Mediterranean dietary pattern (Estruch et al., 2018).
Ultra-processed foods are probably harmful. A prospective cohort study of over 44,000 French adults found that a 10% increase in ultra-processed food consumption was associated with a 14% increase in all-cause mortality (Schnabel et al., 2019).
Fiber intake is consistently beneficial. A meta-analysis in The Lancet encompassing 185 prospective studies found 15-30% reductions in all-cause mortality, cardiovascular mortality, and colorectal cancer with higher fiber intake (Reynolds et al., 2019).
The media amplification problem
A 2020 analysis found that 58% of nutrition news articles failed to mention study limitations, 72% did not distinguish between observational and interventional research, and 44% made causal claims from correlational data (Haber et al., 2018).
How to read nutrition news
Check the study type. Randomized controlled trials are stronger than observational studies. Animal studies should almost never change your diet.
Look for replication. When multiple studies from different groups converge on the same conclusion — that is evidence worth acting on.
Be suspicious of dramatic claims. Real dietary effects are typically modest — 10-30% risk reductions.
Follow the dietary pattern. The evidence consistently supports whole dietary patterns — Mediterranean, DASH, traditional Japanese — characterized by whole foods, plants, and minimal ultra-processing.
Nutrition science is not as broken as the headlines make it seem. The best dietary advice has been remarkably stable for decades: eat mostly whole foods, emphasize plants, and be deeply skeptical of anyone selling you a miracle.
References
- Estruch, R., et al. (2018). Primary prevention of cardiovascular disease with a Mediterranean diet. NEJM, 378(25), e34.
- Greenland, S. (2017). The need for cognitive science in methodology. American Journal of Epidemiology, 186(6), 639–645.
- Haber, N., et al. (2018). Causal language and strength of inference. PLOS ONE, 13(5), e0196346.
- Reynolds, A., et al. (2019). Carbohydrate quality and human health. The Lancet, 393(10170), 434–445.
- Schnabel, L., et al. (2019). Ultra-processed food consumption and risk of mortality. JAMA Internal Medicine, 179(4), 490–498.
- Schoenfeld, J. D., & Ioannidis, J. P. A. (2013). Is everything we eat associated with cancer? AJCN, 97(1), 127–134.
- Subar, A. F., et al. (2015). Addressing current criticism of self-report dietary data. Journal of Nutrition, 145(12), 2639–2645.