Obesity: study compares BMI with more comprehensive criteria

The same body mass index can cover very different situations: a high body weight with no detected functional impairment, or excess adipose tissue already associated with health problems. A study published in “Nature” compares this broader approach with BMI alone, using data from 844 108 adults followed across four cohorts. It sheds light on possible reclassifications, without in itself establishing a new rule for managing care.

BMI measures body size, not overall health

Body mass index is calculated by dividing weight by height squared. It provides a simple measure for classifying body size, but does not directly measure either the amount or distribution of fat. At the same BMI, muscle mass, body shape or abdominal fat can differ. The index also does not, on its own, describe how organs function or the effects of excess adipose tissue on everyday activities.

A Lancet commission proposed in 2025 that BMI be supplemented with other measurements, notably waist circumference and the waist-to-height ratio. It also distinguishes clinical obesity, where excess adiposity is accompanied by organ dysfunction or limitations in everyday activities, from preclinical obesity, where excess adiposity is present without functional impairment detected under the criteria used. “Preclinical” therefore does not mean there is no risk or no need for prevention.

Four cohorts, varying results

The study, published in “Nature” on 11 November 2026, reanalyses data already collected in China, the United Kingdom, the United States and Catalonia. The researchers compared BMI-based classification with a definition combining BMI, waist circumference and the waist-to-height ratio. Here is the proportion classified as obese in each sample analysed:

These percentages describe the participants included in each cohort, not all the residents of these countries or regions. The samples differ in composition and follow-up methods: the table compares the study’s results, not the health of national populations.

The change works in both directions. Depending on the cohort, 1.2 to 3.1 % of participants were classified as obese by BMI without meeting the waist circumference criteria used. Conversely, 5.2 to 8.6 % of participants classified as overweight according to BMI met the combined adiposity criteria corresponding to obesity. BMI alone can therefore place some people in a category that abdominal measurements do not confirm, while leaving others outside that category.

Distinguishing a weight category from functional impairment

The researchers then separated clinical obesity from the preclinical form. For the former, they looked for health problems or functional limitations among the criteria defined by the Lancet commission. The available data came from existing health records, with codes and measurements that varied from one cohort to another. The authors applied a statistical method to estimate, at group level, what proportion of the observed health problems could be attributed to obesity.

This estimate cannot determine the cause of a health problem in a given individual. A disease present in someone classified as having clinical obesity is not automatically caused by their weight. The researchers also note that the records did not include the same assessments everywhere, which may affect classifications.

In three cohorts with sufficient follow-up, the clinical form was associated with higher all-cause mortality than the reference group selected by the study. The association was more modest for the preclinical form. This is an adjusted statistical comparison, not an individual prediction: it does not mean that a person concerned has a given probability of dying, or that the classification causes a difference in mortality. The US cohort was not included in this analysis because its follow-up period was too short.

In the three cohorts with ten years of follow-up, estimates adjusted using the attribution method indicate that between 77.4 % and 91.3 % of people classified as having preclinical obesity at the outset did not move into the clinical category during that period. For the US cohort, the available follow-up was shorter and the estimate covers four years. These results show that trajectories are not identical; they do not guarantee that a situation will remain stable, or that it would be desirable to wait before taking action.

A tool for interpretation, not yet proof of improved care

Researcher Marie Spreckley considers the results to support the value of going beyond BMI alone. She also notes the limitations of existing records, differences in measurement between cohorts and the risk of residual confounding factors. She points out that some of the contrast observed between clinical and preclinical obesity is to be expected, since the presence of functional impairment is part of the definition of the clinical form.

The study reanalyses cohort data; it is not a trial in which patients were assigned to two diagnostic methods to compare their care. It therefore does not show that adopting this classification improves treatment, quality of life or survival. The authors argue for its adoption worldwide; this conclusion is their recommendation, not a universal medical instruction that has come into force.

The practical point is more specific: BMI can remain a screening measure, but it does not sum up either the distribution of fat or a person’s state of health. Additional measurements and functional assessment can provide other information, to be interpreted in their clinical context. The thresholds studied do not constitute a new universal standard to apply to oneself. The study received no industry funding; however, several authors declare consulting, research or remuneration links with companies in the pharmaceutical or medical device sectors.

References

• L’essentiel: overview of the study
• Nature: study on cohorts and obesity classification
• Science Media Centre: reaction from Marie Spreckley

A tape measure, illustrating anthropometric measurements that complement BMI. Photograph from 2007.

Pastorius — Wikimedia Commons — CC BY 3.0.