Weight ScienceEvaluating the Evidence for a Paradigm Shift

What if the standard treatment for a major health condition not only fails most of the time, but makes things worse? That's not a rhetorical provocation. It's the conclusion that Linda Bacon and Lucy Aphramor reach after assembling decades of long-term data in their review "Weight Science: Evaluating the Evidence for a Paradigm Shift." The failure rate they document is striking. A National Institutes of Health panel found that most people who lose weight regain one third to two thirds of it within one year, and almost all of it within five years. If that's the baseline for the treatment, the question isn't whether the paradigm needs scrutiny. It's why the scrutiny took this long. The dominant clinical premise is simple: reduce body weight, reduce disease, and extend life. And in the short term, behavior-change interventions—diet, exercise, and lifestyle modification—do produce weight loss. The problem shows up when you follow people past the first year. The Women's Health Initiative, the largest randomized dietary trial ever run, enrolled more than 20,000 women who reduced caloric intake by an average of 360 calories per day and increased their physical activity. After almost eight years, the mean weight change was negative 0.1 kilograms—essentially nothing—and average waist circumference had actually increased by 0.3 centimeters. Bariatric surgery tells a similar story: weight loss peaks around one year post-operation, and gradual regain is common after that. The expected health benefits don't reliably materialize either. Epidemiologic analyses pooling over 350,000 subjects from 26 studies found that the "overweight" category was associated with equal or greater longevity than "normal" weight. Bacon and Aphramor describe an obesity paradox in which people with type 2 diabetes, hypertension, cardiovascular disease, and chronic kidney disease often show greater survival when heavier. In controlled weight-loss trials for type 2 diabetes, initial improvements in glycemic markers reverted to baseline within six to eighteen months—even when weight loss was reportedly maintained. One liposuction study the authors cite removed an average of 10.5 kilograms of fat with no improvement in blood pressure, triglycerides, cholesterol, or insulin sensitivity over ten to twelve weeks. The causal story connecting lower weight to better health is much less clean than clinical guidelines suggest. Then there are the harms—and this is where the evidence becomes harder to set aside. Repeated cycles of weight loss and regain, called weight cycling, are linked to increased inflammation, hypertension, insulin resistance, dyslipidemia, and poorer cardiovascular outcomes. Bacon and Aphramor report that weight cycling alone can account for all of the excess mortality statistically associated with obesity in both the Framingham Heart Study and analyses of the National Health and Nutrition Examination Survey. Dieting predicts future weight gain. It reduces bone mass, elevates cortisol, and promotes food and body preoccupation. There is also emerging evidence that weight loss releases persistent organic pollutants stored in fat tissue. Two cited studies found higher blood concentrations of these pollutants in people who had lost weight. The psychological toll is documented just as clearly. Body dissatisfaction predicts binge eating, lower physical activity, and increased weight gain over time. Weight-focused clinical practice is associated with reduced self-esteem, increased eating-disordered behavior, and weight stigma that deters people from seeking care at all. And body mass index—the number that anchors almost every clinical conversation about weight—turns out to be a weak proxy for health except at statistical extremes. One cited study found that using body mass index misidentified 31 percent of the population. Sixteen point three million people at normal weight with abnormal metabolic profiles were missed, while fifty-five point four million people classified as overweight or obese were flagged as ill when they weren't. This is the landscape Bacon and Aphramor survey before introducing Health at Every Size, or HAES. HAES is a weight-neutral clinical framework—not a philosophy or a social movement, though it has supporters in both—but a structured intervention with specific behavioral targets. It asks patients to practice body acceptance rather than pursue weight change, to develop what the literature calls intuitive eating rather than impose cognitive dietary restraint, and to adopt what the authors call joyful movement integrated into daily life rather than prescriptive exercise aimed at burning calories. Intuitive eating means relearning to read the body's signals—hunger, fullness, energy, mood, satiety, and digestive comfort—and making food choices based on that information rather than external rules. Clinicians working within HAES are asked to stop treating weight as the primary target of care, to drop weight-biased language, and to focus on modifiable behaviors where evidence actually shows health benefit. The framework has been endorsed by the Academy for Eating Disorders, the Binge Eating Disorder Association, the Eating Disorder Coalition, and several civil rights organizations. None of that is evidence of efficacy on its own, but it does tell you HAES has been formalized enough that it can be studied in randomized trials—and those trials are what the review's central argument rests on. Bacon and Aphramor identified six published randomized controlled trials with an explicit HAES or intuitive-eating focus. The sample sizes were modest—group sizes typically ranged from the low twenties to the mid-sixties—and follow-up ran from twenty-six to seventy-eight weeks. That's a small evidence base. But what's notable is its consistency. All six trials reported statistically detectable improvements in psychological and behavioral outcomes. Eating behaviors, binge eating, self-esteem, body image, and mood improved across studies. Several trials also showed physiological gains. The most detailed direct comparison comes from a trial by Bacon and colleagues, which randomized thirty-nine women to HAES and thirty-nine to a conventional diet program. At fifty-two weeks, the HAES group showed reductions in LDL cholesterol and systolic blood pressure, increased physical activity, reduced binge eating, and improvements in self-esteem, depression, body dissatisfaction, and interoceptive awareness—the ability to read internal body signals. The diet group did not show those psychosocial benefits. And the dropout rate tells its own story: eight percent in the HAES arm, forty-two percent in the diet arm. That's not a rounding error. Ciliska's trial similarly found reduced diastolic blood pressure, reduced binge eating, and improved self-esteem and depression at fifty-two weeks. Goodrick's trial, with follow-up extending to seventy-eight weeks, showed sustained reductions in binge eating and improved exercise in the non-diet arm. One important caveat: in some trials, physiological improvements in HAES arms were not statistically different from comparison conditions. Rapaport and colleagues saw improvements in cholesterol, blood pressure, activity, dietary quality, and emotional well-being in the HAES group—but those gains were not statistically distinguishable from the control group. The review is honest about this. And critically, no randomized HAES study produced weight gain, and none reported adverse changes in any measured variable. The conventional intervention can't claim the same. That asymmetry carries ethical weight, and Bacon and Aphramor make the argument directly. If a treatment reliably fails long-term and is associated with documented physiological and psychological harms, continuing to recommend it without disclosure raises questions of beneficence and informed consent. They call for clinicians to accurately convey the limited long-term impact of weight-loss interventions rather than implicitly promising durable results. The paper also connects the weight-focused paradigm to structural inequity. Body weight tracks with socioeconomic class, chronic stress from weight stigma affects metabolic risk, and weight discrimination in clinical settings reduces quality of care and deters help-seeking. A HAES-informed clinical ethic would reframe care around health behaviors, remove stigmatizing language, and target social determinants rather than the number on the scale. The evidence base for HAES is real but limited. Six randomized controlled trials with group sizes in the dozens and follow-up under two years cannot settle every clinical question. Bacon and Aphramor explicitly call for larger trials, more diverse populations, and research into how best to deliver HAES to different groups. Those gaps are genuine. But the review's core challenge stands independent of them: the conventional paradigm has had decades and enormous resources, and its long-term evidence is weak, its harms documented, and its ethical foundations increasingly questioned. The open question isn't really whether the evidence justifies attention to HAES. It's whether health systems are built to decouple health from weight—institutionally, clinically, culturally—and whether the will exists to find out. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.