This video is a debate about the hierarchy of scientific evidence, especially in nutrition, and how to weigh mechanistic, epidemiological, and randomized controlled trial (RCT) data.
Here are the key points from the discussion between Avi and TMC:
1. Evidence Hierarchy in Science and Nutrition
- The conversation centers on how to rank different types of scientific evidence: mechanistic studies (studies of biological mechanisms, often in cells or animals), epidemiological studies (observational studies in populations), and randomized controlled trials (RCTs, considered the gold standard for causality).
- Avi argues that mechanistic data is often overvalued in nutrition, and that RCTs should be considered the highest tier of evidence for establishing causality, followed by epidemiology, with mechanistic studies as the lowest.
2. Mechanistic Data vs. RCTs and Epidemiology
- Mechanistic studies can show how something works in cells or animals, but these results often do not translate to real-world human outcomes.
- RCTs, when properly conducted, provide stronger evidence for cause and effect in humans than mechanistic or epidemiological studies.
- Epidemiological studies can sometimes track well with RCTs, especially when designed and analyzed carefully, but they are more prone to confounding factors.
3. Limitations and Fallibility of Evidence
- No type of study can prove causality with 100% certainty; all scientific inference is subject to human fallibility.
- The participants agree that while epidemiology and RCTs have limitations, mechanistic studies are even less reliable for predicting human health outcomes, especially in nutrition.
4. Anecdotal Evidence
- Personal anecdotes are considered the weakest form of evidence, especially for long-term health outcomes like cardiovascular disease.
- Anecdotes may be useful for acute, obvious effects (like allergic reactions), but not for complex, chronic diseases.
5. Application to Diets (Vegan vs. Carnivore)
- The discussion touches on the lack of robust data for carnivore diets compared to vegan diets, making it difficult to draw strong conclusions about long-term health effects.
- Both agree that more research is needed, and that personal experience should not outweigh higher-quality evidence.
6. Supplementation and Practical Considerations
- The need for supplementation (e.g., B12 for vegans, possibly others for carnivores) is discussed, with the point that most diets have “cracks” that can be addressed with supplements.
- Historical context: Many health issues in the past were due to lack of supplementation or food fortification (e.g., folate to prevent neural tube defects).
7. Openness to Revising Views
- The participants agree on the importance of being open to new evidence and willing to revise their views if presented with strong data.
3 Likes
A_User
#3
Why do people post mechanistic studies for a disease when epidemiological studies are available?
Is it because they lack knowledge of evidence-based research? It is contrary to their goals to use inferior evidence. People dream of being Napoleon but he wouldn’t do stupid mistakes like this.
L_H
#4
The hierarchy above is inevitably a simplification, helpful to laypersons but not really the scientific community. The limitations
of various studies go far beyond this single dimension.
A RCT may, for example be very weak evidence if the outcomes measured are indirect biomarkers only, or if the duration is too short or the statistical power weak.
Most scientists would like to have some mechanistic understanding in addition to solid intervention based evidence. Not only to fully explain what’s going on, but also because it potentially open doors to further hypotheses and alternative, perhaps improved interventions.
2 Likes
