Skip to main content

7 Flaws in Nutrition Epidemiology

Video Title: 7 Flaws in Nutrition Epidemiology
Author: Professor Bart Kay - Nutrition Science Channel
Date: April 25, 2021
Source: Watch on YouTube

Overview​

In this lecture, Professor Bart Kay explains why he views nutritional epidemiology as a flawed, "pseudo-scientific" discipline. His primary goal is to demonstrate why observational studies cannot be used to establish causal links between diet and health outcomes.

The Core Problem: Association vs. Causation​

Bart emphasizes that epidemiology is strictly observational. He uses the "Ice Cream and Shark Attacks" analogy to show that correlation does not equal causation. Just because two variables move together doesn't mean one drives the other.

The 7 Invalidating Flaws​

1. Arbitrary Selection Criteria​

Authors of meta-analyses can set arbitrary rules (e.g., English language only, specific genders) to exclude studies that might contradict their desired conclusion.

2. Publication Bias​

Scientific journals tend to publish studies that find an "effect" and ignore those that find nothing. This "file drawer effect" creates an artificial consensus in the literature.

3. P-Hacking​

Researchers may over-sample or under-sample data points specifically to push their results below the p < 0.05 threshold required for publication.

4. Adjusted Outcome Statistics (Multivariate Regression)​

Bart argues that adjusting data for "confounders" is essentially data fabrication. He cites the Adventist Health Study 2 as a prime example where he claims the adjusted results were 180 degrees opposite to the raw observations regarding red meat and mortality.

5. Relative vs. Absolute Risk​

Studies often report "Relative Risk" to make findings look more significant.

  • Example: A change from 1 in a million to 2 in a million is a 100% relative increase, but an absolute increase of only 0.0001%, which is negligible for individuals.

6. Causal Language for Non-Causal Data​

The use of terms like "risk," "hazard," or "protection" is inappropriate for observational data, as these terms imply a causal relationship that hasn't been proven.

7. Extrapolation​

Studies often look at elderly or diseased populations (where events like death are more likely to occur) and then extrapolate those findings to young, healthy individuals where the data does not apply.

Conclusion​

Bart Kay concludes that nutritional epidemiology is of "no value to any given living human being" and should be entirely disregarded when making personal dietary choices.

Bart's Hard Science Checklist

For a study to be relevant to human health, Bart Kay typically requires it to be:

  1. Done on humans (not animal models).
  2. Done In Vivo (living bodies, not petri dishes).
  3. Using Experimental controls (RCTs).
  4. Measuring Hard Outcomes (death/disease, not surrogate markers).