#197 - The science of obesity & how to improve nutritional epidemiology | David Allison, Ph.D.

Episode Summary

Executive Summary: Peter Atiyah and David Allison discuss obesity science, emphasizing that obesity is a complex biological trait shaped by genetics, environment, and measurement limits—not simply willpower. Allison argues that surgery and newer drugs have strong evidence, while many public-health and nutrition-epidemiology claims are weak or misleading. The conversation broadens into reproducibility, scientific rigor, and the danger of confusing science with advocacy.

Main Topics: Allison’s scientific background and approach (Priority: 5/5): Allison explains how psychology training, statistical methods, and a lifelong habit of questioning evidence shaped his unemotional, data-first approach to obesity research. Obesity as a biological and multifactorial trait (Priority: 5/5): The discussion frames obesity as influenced by genetics, physiology, behavior, environment, and social context, rejecting simplistic explanations based on choice alone. History of obesity research and treatment (Priority: 5/5): They trace the field from early behavioral approaches to the rise of public-health framing, then to pharmacotherapy, bariatric surgery, and modern GLP-1-era treatment. BMI, confounding, and the obesity paradox (Priority: 5/5): Allison critiques BMI as a blunt tool, explains confounding in mortality studies, and discusses age-, race-, and sex-specific differences in BMI-mortality relationships. Nutritional epidemiology and methodological failure (Priority: 5/5): A major theme is that food-frequency questionnaires, weak associations, and poor handling of clustering/confounding make much of nutritional epidemiology unreliable for causal claims. Science, advocacy, and trust (Priority: 5/5): The conversation closes on the idea that public distrust often reflects confusion between science and advocacy, not rejection of the scientific method itself. What actually works for obesity (Priority: 4/5): Allison argues that surgery, pharmaceuticals, and possibly broader educational/social interventions have more promise than most school- or community-based public-health programs.

Key Arguments: Obesity should be treated as a biological variable governed by the same laws of physics, probability, and statistics as any other trait. The field became more politically and publicly salient in the early 1990s after NHANES-3 showed a sharp rise in obesity prevalence. Genetics play a major role in obesity; identical twin correlations for BMI are very high, around 0.9. Bariatric surgery is the most effective life-changing treatment currently available and has evidence for mortality benefit. Fen-phen helped legitimize obesity pharmacotherapy but also showed the need for rigorous safety monitoring. BMI is useful for population-level surveillance but is a weak individual-level clinical tool because it ignores body composition and fat distribution. The obesity paradox and U-shaped BMI-mortality curves are heavily confounded by smoking, illness, socioeconomic status, stigma, and selection bias. Throwing out early deaths or other post hoc exclusions in observational studies does not reliably fix confounding and can reduce power or worsen bias. Nutritional epidemiology often overstates certainty because of measurement error, confounding, and weak causal inference. Public-health interventions for obesity have largely failed to show meaningful effects, whereas surgery and drugs have clearer evidence. Science is not in crisis in the sense of collapse, but it does need greater rigor, honesty, and separation from advocacy. Trust in science may be less the issue than trust in specific scientific communicators and institutions, especially in nutrition and COVID-era discourse.

Data Points: PhD year: 1990 - David Allison received his PhD from Hofstra University in 1990. Postdoctoral fellowships: 2 - He completed two postdoctoral fellowships, at Johns Hopkins and the New York Obesity Research Center. Scientific publications: 500+ - Allison has authored over 500 scientific publications. National Academy of Medicine election: 2012 - He was elected to the National Academy of Medicine in 2012. Twin BMI correlation: ~0.9 - Allison says monozygotic twins separated at or near birth have BMI correlations nearly as high as twins reared together. Weight gain to BMI unit: ~6 pounds per BMI unit - He uses this approximation to explain how the mortality nadir shifts with age. Average U.S. weight gain: ~1 pound per year - Used in the discussion of how BMI mortality nadirs may move rightward over time. BMI underweight threshold: <18.5 - Allison references the standard BMI cutoff for underweight. BMI obesity threshold: 30 - He notes BMI 30 as the conventional beginning of obesity. Sumo wrestler BMI: ~43-44 - Used as an example of very high BMI. Mortality reduction after bariatric surgery: ~50% or more - Across studies, Allison says surgery often shows about a 50% reduction in mortality, sometimes more. Fen-phen withdrawal: Late 1990s, around 1997 - He estimates fenfluramine was withdrawn in the late 1990s after valvulopathy concerns. Cluster randomized trial issue: Incorrect analysis can invalidate results - He explains that clustering and nesting must be accounted for or results may be wrong. Public-health intervention effect: Little to none in many trials - Allison argues school/community/public-health obesity interventions have not shown meaningful effects.

Pivotal Quotes: "“The data, the methods used to collect the data, which give them their probative value, and the logic, which connects the data and the methods to conclusions. Everything else is not science.”" — David Allison: On what counts as science and why ad hominem or rhetorical persuasion should not substitute for evidence. "“No crisis but no cause for complacency.”" — Harvey Feinberg (as cited by David Allison): Describing the National Academies’ view on reproducibility and rigor in science. "“We have shot ourselves in the credibility foot with our obfuscation and our exaggeration and our hype.”" — David Allison: On why trust in nutrition science has eroded and why the field needs reform.

Implications: Listeners should be skeptical of simplistic obesity narratives and weak nutrition claims. The strongest current tools are surgery and medications, while science must better separate evidence from advocacy and improve rigor, transparency, and study design.