A quarter of bariatric surgery candidates had no organ dysfunction — at the same BMI

The Lancet Diabetes and Endocrinology Commission proposed splitting obesity into two conditions: clinical obesity, meaning excess adiposity accompanied by obesity-related organ dysfunction, and preclinical obesity, meaning excess adiposity with function preserved [s1]. It is a conceptual argument about what counts as disease. Whether it changes anything in practice is an empirical question, and a study published in JAMA Network Open on August 11 is among the first to test it somewhere consequential.

Bariatric surgery is a good place to look, because patient selection there still runs largely on BMI thresholds rather than on clinical condition [s1].

What was done

Researchers audited records from four high-volume tertiary centres — in the UK, Spain, France and Brazil — covering adults who underwent a primary metabolic bariatric surgery, either laparoscopic gastric bypass or sleeve gastrectomy, between January 1, 2014 and December 31, 2025 [s1]. Revisional procedures and multiple operations within the same year were excluded [s1].

Each of the 2,316 included patients was classified as having clinical or preclinical obesity using a pragmatic operationalisation of the Commission's framework [s1].

An important design constraint: patient-level data were not shared between centres, so all comparisons were made within each cohort separately, with no direct statistical comparison across countries [s1].

The split

Of 2,316 patients, 1,709 (73.8%) met criteria for clinical obesity and 607 (26.2%) for preclinical obesity [s1]. Clinical obesity was the most common category at every centre, ranging from 62.7% in the Brazilian cohort to 79.3% in the UK cohort [s1].

Then the finding the whole exercise turns on. Mean BMI ranged from 40.2 (SD 4.9) — preclinical obesity in the Brazilian cohort — to 48.5 (SD 9.0) — preclinical obesity in the UK cohort, and BMI levels and the distribution across BMI categories (<40, 40–49, ≥50) were similar between the clinical and preclinical groups [s1].

Same BMI. Different disease state.

What differed instead

Age, consistently. Patients with clinical obesity were significantly older in every cohort: 48.9 years (SD 11.7) versus 39.5 (SD 9.7) in the UK; 45.6 (SD 11.0) versus 34.7 (SD 10.4) in France; 47.1 (SD 9.3) versus 40.0 (SD 9.8) in Spain; 48.3 (SD 11.7) versus 40.0 (SD 9.4) in Brazil — all P < .001 [s1].

Anaesthetic risk was higher, with a higher American Society of Anesthesiologists classification in every cohort [s1]. So was comorbidity burden, measured by Charlson Comorbidity Index [s1], and estimated ten-year cardiovascular risk on the Framingham score — in the French cohort, a median of 8.00 (IQR 3.00–13.00) against 2.00 (IQR 1.00–4.00) [s1].

Thirty-day major postoperative complications, defined as Clavien-Dindo grade 3 or above, were higher with clinical obesity in the French cohort — 4.0% versus 1.1% (P = .04) [s1]. That difference reached significance in one of the four cohorts, which is worth stating plainly rather than generalising from.

The preclinical group was also more female: 80.1% versus 67.2% [s1].

What it means and what it does not

The argument the data support is narrow but real. If two groups of surgical candidates have effectively the same BMI distribution but meaningfully different age, comorbidity, operative risk and cardiovascular risk, then BMI is not sorting them — and BMI is what selection currently runs on [s1]. The authors conclude the framework has potential value as a clinically meaningful basis for surgical practice [s1].

The argument the data do not support is that the framework should be used to ration surgery. Preclinical obesity means organ function is preserved now. In a cohort where those patients are on average eight to ten years younger [s1], that is at least partly a statement about timing. Nothing in a cross-sectional audit says what happens to them at 50.

Other limits are structural. This is retrospective and cross-sectional, drawn from four tertiary centres that are not representative of surgical practice generally, and the classification used a pragmatic operationalisation rather than the Commission's full criteria [s1]. Long-term prognosis was framed as an objective but the reported outcomes are baseline risk estimates and 30-day complications, not observed long-term events [s1].

The wider pattern

This is the second demonstration this year that obesity classification is doing more work than it appears to. A Swiss cohort analysis published in June applied three definitions to the same population and found obesity prevalence of 17.4%, 33.9% and 37.5% depending on which was used, with the overweight category shrinking correspondingly [s2].

Between them, the two studies make a consistent point from opposite directions. Change the definition and the population changes. Hold the definition — BMI — and the population it produces turns out to contain two rather different groups.

This article is informational and is not medical advice.

Sources

• New Obesity Definition and Clinical Obesity Prevalence for Global Metabolic Bariatric Surgery — JAMA Network Open, 2026-08-11
• Obesity prevalence varies markedly by definition: multiple cross-sectional and prospective studies — Nutrition Research, 2026-06-16