30 Years of Diet Trials: What Matters Most in Keto, Vegan & Mediterranean Diet | Christopher Gardner

Summary

1. Nutrition science has to balance rigor with real-world relevance

Christopher Gardner describes nutrition research as a constant trade-off between scientific control and generalizability. Researchers can tightly control what people eat, but that may not reflect ordinary life. Studies in which participants shop, cook, and make their own choices are more realistic, but adherence varies, and researchers cannot assume people followed the assigned diet exactly.

Gardner’s philosophical training shaped his view that nutrition rarely offers a single unquestionable answer. He favors weighing evidence carefully, being transparent about uncertainty, and avoiding claims broader than a study can support. In his shorthand, the practical principle is Michael Pollan’s: “Eat food. Not too much. Mostly plants.”

2. A null finding can be useful—and easily misrepresented

Gardner recounts a large garlic trial prompted by claims that garlic supplements lower cholesterol.

• After early work raised questions about whether garlic compounds were being delivered effectively, Gardner collaborated with garlic chemist Larry Lawson and received a $1.4 million NIH grant.
• The study compared fresh garlic, aged garlic, enteric-coated garlic, and a placebo in roughly 200 participants with moderately elevated LDL cholesterol. The team carefully prepared and tested the garlic and food seasonings used in the trial.
• After six months, none of the garlic preparations lowered LDL cholesterol compared with placebo.
• Gardner emphasizes that this answered a narrow question: whether these particular preparations lowered LDL in this particular group. It did not establish that garlic is harmful or rule out every possible effect of garlic on health.
• Media headlines turned the null result into a broader claim that garlic was harmful. Gardner argues that researchers and journalists should not overstate what a study demonstrates.

The example illustrates why an apparently disappointing “nothing happened” result may still help people make decisions and correct unsupported claims.

3. Diet studies need clear definitions and honest adherence reporting

Terms such as “keto,” “low-carb,” “low-fat,” and “Mediterranean” can describe quite different diets in different studies. Gardner says researchers should specify both:

• What diet participants were assigned, and
• What they actually ate and how closely they followed it.

In his trials, participants are told before randomization what the possible diets involve and are encouraged not to enroll if they would drop out upon receiving an unwanted assignment. Gardner says being candid about this helps retention and protects the usefulness of the results. He also stresses that imperfect adherence is not simply a research failure: it is part of what happens when dietary advice meets ordinary life.

Long-term randomized comparisons of sharply different diets are especially difficult. People may not agree to be assigned to a diet they dislike, and maintaining the assignment for years would be hard to verify. Thus, researchers often cannot run the definitive trial people might want.

4. Pre-registration helps prevent “data dredging”

Gardner explains that researchers can measure many outcomes, which creates a risk of highlighting whichever ones happen to change while overlooking the original question.

• In his fermented-food study, the team registered a primary composite measure of cytokine response before analyzing the results.
• That primary measure did not change.
• The researchers also measured many other inflammatory markers. Of about 90 markers, 20 considered pro-inflammatory decreased in the fermented-food group, 70 did not change, and none increased.
• Gardner presents these other findings as interesting but secondary and exploratory, not as proof that fermented foods improve health. A new study would be needed to test them directly.

His broader lesson is to identify the main outcome in advance, report null results, and label follow-up findings clearly.

5. A fermented-food study is not a blanket health recommendation

The conversation cautions against turning a study of particular foods, servings, or biomarkers into a universal recommendation. The fermented-food trial does not settle which products are best, how much to eat, for whom, or whether changes in microbiome diversity or inflammation translate into better long-term health.

Gardner also notes that fermented foods have a long culinary history and can be approached as enjoyable foods, not only as microbiome interventions. He favors research that takes taste, cooking, and sustained habits seriously, while resisting commercial claims that microbiome science can already prescribe precise, personalized products.

6. Several principles are shared across popular diets

In work comparing popular diets with American Heart Association nutrition criteria, Gardner says the diets differed substantially but shared several basic principles:

• Choose more whole foods.
• Eat more vegetables.
• Reduce added sugar.
• Reduce highly processed foods and refined grains.

He argues that these common points matter because many people eating a standard American diet do not follow them. The diets were not equally aligned with the AHA criteria: DASH, Mediterranean, pescetarian, and ovo-lacto-vegetarian patterns ranked most favorably; keto and paleo ranked lower; several other approaches fell in between. Gardner presents this as a comparison against those criteria, not as a definitive ranking of every diet for every person.

He also discusses foods that can be controversial in diet debates—legumes, fruit, nuts and seeds, eggs, fish, and yogurt—emphasizing that their value depends partly on what they replace and on the person’s overall diet.

7. Two diet trials show why results need context

Gardner’s keto-versus-Mediterranean trial: Participants with prediabetes or type 2 diabetes followed each diet for 12 weeks in a crossover study. Both diets were designed to limit added sugar and refined grains; the keto plan excluded foods such as legumes, whole grains, and fruit, while the “Mediterranean-plus” plan included more fat than a typical Mediterranean diet. Both groups lost weight and improved some measures, including HbA1c. The keto diet reduced triglycerides more, while LDL moved in opposite directions: it rose on keto and fell on the Mediterranean-style diet. Gardner notes the study’s duration, adherence, and design limits.

Sam Klein’s study: Gardner discusses a separate, longer study in which participants were supplied with food and lost about 10% of their body weight across diet groups. He notes that the ketogenic group ate a high proportion of calories from saturated fat yet did not show the expected rise in LDL/apoB, and that liver fat decreased, with particularly strong reductions on keto. Gardner treats this as a result worth examining, not as a final verdict. He raises weight loss and the temporary metabolic state associated with active weight loss as possible factors, and says the food composition and longer-term effects need closer study.

Together, the studies show why one result should not be generalized without considering the participants, weight change, diet composition, duration, and whether the intervention reflects everyday eating.

8. Adherence can determine whether an intervention works in practice

Gardner uses a women’s health calcium example to distinguish efficacy under ideal conditions from effectiveness as a public-health strategy. In the study he describes, many participants already took calcium, making it difficult to compare supplementation with no supplementation. Later analyses focusing on people who adhered to pills suggested possible benefit, but such subgroup findings do not necessarily show that broadly offering supplements will reduce fractures across the population.

His point is that a treatment may appear to work among people who take it consistently, while an intervention that people do not follow may fail as a practical public-health policy. The same issue applies to diets: it is not enough to ask which plan can produce a result under controlled conditions; researchers also need to ask whether people can and want to sustain it.

9. Health advice should account for enjoyment, culture, and values

Gardner recalls an earlier study in which two low-fat diets both reduced LDL, but the diet richer in plant foods reduced it about twice as much. Yet participants were reluctant to keep eating the meals once the study ended, partly because they wanted variety, convenience, and the ability to eat in restaurants. The example highlights that a diet’s measurable effects do not guarantee that people will continue it.

He argues that sustainable food choices can also connect with personal priorities such as environmental impact, animal welfare, social conditions in food production, family traditions, and cooking skills. A successful eating pattern needs to work as food in people’s lives—not only as a set of nutrients.

10. The overall message

Gardner’s approach is to make claims proportional to the evidence, distinguish primary from exploratory findings, and consider how diets work outside controlled trials. He is particularly interested in food patterns that are nutritious, enjoyable, practical, and sustainable, rather than reducing food to a single nutrient or fashionable label.

Speakers and sources featured

Main speakers

• Simon Hill — host and interviewer.
• Christopher Gardner — nutrition researcher interviewed.

Researchers and other sources discussed

These people are referenced in the conversation; they are not necessarily speakers in the video:

• Michael Pollan
• Larry Lawson
• Justin Sonnenburg
• Mark Davis
• Holden Maecker
• Vayu Hillman
• Shawn Spencer
• Rob Knight
• Peter Turnbaugh
• Maria Marco
• Kevin Hall
• David Jenkins
• Sam Klein
• David Ludwig
• Lauren Cordain
• Roy Taylor

Organizations and studies mentioned

• Stanford and the Stanford nutrition and microbiome research groups
• The U.S. National Institutes of Health (NIH)
• ClinicalTrials.gov
• The American Heart Association
• Women’s Health Initiative
• PREDIMED
• The garlic and fermented-food trials discussed by Gardner

Promotional segments

• Function Health
• IM8
• Shopify