Three of four surgery candidates were already ill, and body mass index showed none of it

The same index, two different patients

The Lancet Commission's framework asks something other than what the scale asks. Instead of grading obesity by body mass index alone, it separates excess adiposity that has already produced organ dysfunction — clinical obesity — from preclinical obesity, where function is still preserved. Applied retrospectively to 2,316 people who underwent primary gastric bypass or sleeve gastrectomy at four centres in the United Kingdom, Spain, France and Brazil between 2014 and 2025, it placed 1,709 of them, or 73.8 percent, in the clinical group and 607, or 26.2 percent, in the preclinical one.^[1]

What makes the split worth attention is that body mass index did not predict it. Mean index ran between 41.2 and 47.5 in the clinical group and between 40.2 and 48.5 in the preclinical one — overlapping ranges, effectively the same patients by weight. What separated them was everything else: the clinical group was roughly ten years older, sicker by ASA score, and carried consistently higher Framingham cardiovascular risk in every cohort where it was recorded. Reading an endometriosis cohort in which the hazard ratio was highest in women with a body mass index under 30, I argued earlier that the index marks risk poorly at the level of a single patient. This audit says the same thing from the other end of the scale.^{[1], [2]}

Where does the label come from?

The framework's usefulness depends entirely on who gets called clinically obese, and here the audit is honest about its own machinery. The databases it drew on were assembled before the commission published its report, so organ dysfunction was never assessed prospectively against the new criteria. It was reconstructed afterwards from International Classification of Diseases coding and whatever the clinical documentation held. The authors write that this introduces potential intercentre variability: hospitals code differently, and the framework inherits that variation intact.^[1]

The authors go further and name the direction the error most likely runs. An audit of routine data, they write, precluded definitive causal attribution of organ dysfunction to obesity, raising the possibility of overattribution and therefore potential overestimation of clinical obesity. Put plainly: some of that 73.8 percent may be people whose hypertension or sleep apnoea would have arrived regardless, now folded into an obesity diagnosis because they were heavy and the code was there. The opposite reading is equally available: coding holds only what a clinician thought worth documenting, so genuine organ dysfunction that never reached a note would pull the true figure higher. Both errors live in the same dataset, and this design cannot separate them.^[1]

Who was counted?

Then the denominator. Every one of the 2,316 was already a candidate for metabolic bariatric surgery at a high-volume tertiary centre: referred, worked up, accepted for an operation. Mean index ran between 41.2 and 48.5 across the centres. A general practitioner deciding whether a patient's weight has begun to cost them an organ faces a very different population, and that is exactly the decision the clinical-obesity label is meant to inform. The framework was validated here on people for whom the major treatment decision had already been made. Two of the four datasets, the UK and Brazilian ones, came from clinical audit databases rather than complete institutional registries, so even within surgery the sample falls short of a clean census.^[1]

What the audit does establish is worth keeping. Patients labelled clinically obese carried more disease, scored higher on operative risk and lost slightly less weight at twelve months: 34.0 percent against 36.1 percent in Spain, 28.5 against 32.2 in France, 31.0 against 33.5 in Brazil. Small gaps, consistently in the same direction, on an endpoint surgeons already track. That justifies recording the classification prospectively; triaging anyone with it needs more. The test I would watch for is a cohort that applies the criteria at the clinic door rather than reconstructing them from codes, and reports outcomes beyond weight — cardiovascular events, medication burden, thirty-day complications. If such a cohort reports before the end of 2027 and the clinical group still separates on a hard outcome, the label has earned its place in the pathway. Until then it describes patients well and decides nothing for them.^[1]