The headline number is startling: an estimated 18.1 million people were living with anorexia nervosa or bulimia nervosa in 2023. But the most useful question is not whether eating disorders are suddenly everywhere. It is what a global estimate can—and cannot—tell us about a problem that health systems often fail to see.
A new Global Burden of Disease analysis published in *JAMA Psychiatry* on October 28, 2026, assembled data from systematic reviews, population surveys and vital-registration systems across 204 countries and territories. The authors estimate that 3.92 million people had anorexia and 14.2 million had bulimia in 2023. About 41% of those affected were aged 10 to 24, and roughly 5.5 million were male.
Those estimates broaden a conversation too often reduced to a narrow stereotype: young, female and affluent. Yet the study's real contribution is more complicated. It shows both a substantial rise in measured burden and the limits of the instruments used to measure it.
A 71% rise is not a 71% rise in risk
Between 1990 and 2023, the estimated number of people living with anorexia or bulimia rose by 71%. That does not mean an individual's risk rose by the same amount. The world's population grew and aged over those three decades, increasing the number of cases even if underlying rates had remained unchanged.
The researchers therefore also calculated an age-standardized prevalence rate, which adjusts populations to a common age structure. That measure rose by 20.6%. In other words, demographic change explains much of the increase, but not all of it. The analysis also estimated 3.83 million disability-adjusted life years in 2023—a combined measure of years lost to ill health and premature death.
The steepest age-standardized increases appeared in East Asia, South Asia and Southeast Asia. That pattern may reflect real changes in risk. It may also reflect better recognition, expanding survey coverage, shifting diagnostic practice or changes in the statistical model. The study cannot cleanly separate those explanations.
That distinction matters. A global model is not a census, and the figures come with uncertainty intervals rather than laboratory precision. The central estimate of 18.1 million, for example, sits within an uncertainty range of 13.9 million to 24.0 million. Reporting only the midpoint creates a sharper picture than the evidence supports.
The map also shows where evidence is easiest to collect
Global comparisons can look authoritative even when the underlying data are uneven. Some countries have repeated psychiatric surveys and detailed health records; others have sparse, older or non-comparable evidence. Where observations were missing, the Global Burden of Disease model borrowed information from related places and predictors. That is a defensible way to create a complete map, but it also means some country estimates are more model-dependent than others.
The blind spots go further. The analysis covered anorexia and bulimia but not binge-eating disorder or other specified feeding or eating disorders, because the burden framework did not yet have the disability weights needed to model them. The National Institute of Mental Health lists binge-eating disorder among the common eating disorders. Excluding it means the study is not an estimate of all eating disorders—and likely understates their total burden.
Mortality is another weak point. The model estimated deaths attributed directly to anorexia, but deaths involving malnutrition, cardiac complications or suicide may be coded under another cause. It did not estimate mortality for bulimia. The apparently small death count should therefore not be read as evidence that these illnesses are medically minor.
Nor can the dataset explain why prevalence changed. Economic development, urbanization, social media, food environments and diagnostic awareness are plausible contributors discussed in the literature, but this analysis did not test them as causal mechanisms. A regional rise and a social change occurring at the same time do not establish that one produced the other.
The forecast is a planning signal, not a prediction
The authors project 21.8 million people living with anorexia or bulimia by 2050, about 20% more than in 2023. They deliberately based the forecast on 1990–2019 patterns to reduce distortion from the COVID-19 years. Even so, a quarter-century projection depends on assumptions about population, diagnosis and past trends. New prevention strategies, conflicts, migration, health-system changes or revised classifications could all bend the line.
The forecast is therefore most useful as a capacity question: if the broad direction is right, are services prepared to recognize eating disorders across genders, ages and regions? The World Health Organization's September 2026 mental-health fact sheet notes that effective psychological and family-based treatments exist, while access remains uneven. NICE guidance likewise emphasizes coordinated, multidisciplinary care rather than a single intervention.
The new study does not prove a universal cause or deliver a precise national league table. It does something more useful: it establishes that the measured burden has grown faster than population change alone would predict, while exposing how much remains unmeasured. The honest conclusion is not that the map has settled the scale of the problem. It is that even an incomplete map points toward a larger, more geographically diverse need than health systems have been built to meet.
