By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
Peter Attia MD269 - Good vs. bad science: how to read and understand scientific studies
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
At a glance
WHAT IT’S REALLY ABOUT
Peter Attia and Bob Kaplan Explain How to Read Scientific Studies
- Peter Attia and Bob Kaplan walk through how a study moves from hypothesis, null hypothesis, design, power analysis, ethics approval and pre-registration to execution, and how observational, experimental and meta-analytic studies differ.
- They explain how bias and confounding such as healthy user bias, recall bias with food frequency questionnaires, and performance bias can undermine observational studies and even randomized trials, using examples like PREDIMED and a hypothetical potato diet trial.
- The statistical core covers power, p-values, relative versus absolute risk, hazard ratios, number needed to treat and confidence intervals, with worked numeric examples showing why effect size and uncertainty matter.
- The discussion turns to why trials stop early (safety, benefit, futility), how peer review and publication bias work, how pre-registration and registered reports help, and how journal impact factor is calculated.
- Attia closes with his paper-reading routine: abstract, methods, figures and legends, then discussion, and stresses that skill comes with practice and that media reporting often lacks rigor.
IDEAS WORTH REMEMBERING
5 ideasA meta-analysis is only as good as its constituent studies
Attia says pooling garbage studies yields a garbage meta-analysis, so you must examine each study within it, and the evidence pyramid wrongly implies meta-analyses always beat randomized trials.
Observational nutrition research is limited by healthy user bias and recall problems
People who avoid meat tend to differ in many health behaviors, and food frequency questionnaires depend on unreliable memory, so Attia argues causality cannot be inferred and that such questionnaires should be abandoned.
Randomized trials can still be biased
Attia describes performance bias, where one arm gets more attention or coaching, and notes PREDIMED had randomization problems because entire clinics were assigned to one group, requiring reanalysis.
Power determines whether a null result means anything
An underpowered study may fail to detect a real difference, while an overpowered one can flag clinically irrelevant effects as statistically significant, so sample size relative to effect size matters.
Relative risk without absolute risk can mislead
A 25% relative increase in breast cancer in the Women's Health Initiative was only four to five cases per thousand, and NNT, calculated from absolute risk reduction, shows how many people must be treated to prevent one event.
WORDS WORTH SAVING
5 quotesSo if you do a meta-analysis of a bunch of garbage studies, you get a garbage meta-analysis.
— Peter Attia
you can't really talk about relative risk without knowing absolute risk, and sometimes they don't give you enough data in the paper to do that, and it's infuriating actually.
— Peter Attia
when something doesn't work it is just as important as when it does work.
— Peter Attia
you could make a case that one of the most efficacious dietary interventions known to man is having somebody watch what you eat every meal.
— Peter Attia
you just have to roll up your sleeves and accept the fact that you're, you're not going to be able to read these things in the bathtub, uh, you know, on a lazy Sunday morning.
— Peter Attia
High quality AI-generated summary created from speaker-labeled transcript.
