Two obesity-diagnosis systems agree on advanced disease, split early

For most of the last two decades, the working critique of body mass index has been that a single number cannot capture how sick obesity makes a given person. Two systems built to fix that problem have now been run against each other on the same very large population — and a study in Obesity, published 2 September, finds they agree on the sickest patients and disagree on everyone else [s1].

That is a more useful result than it sounds, because the disagreement falls exactly where the clinical stakes of a definition are highest: at the boundary where a person is first labelled as having a disease rather than a risk factor.

The two systems

The Edmonton Obesity Staging System (EOSS), proposed in 2009, grades obesity from stage 0 to 4 on the basis of existing comorbidities and functional impairment rather than weight alone [s3]. The Lancet Commission's Diagnostic Model for Obesity (DMO), published in January 2025, takes a different cut: it separates preclinical obesity — excess adiposity with organs still working — from clinical obesity, defined by demonstrable organ dysfunction attributable to that adiposity [s2]. One grades severity along a ladder; the other draws a line between disease and not-yet-disease.

The new analysis applied a modified version of each framework to roughly 411,000 participants in the UK Biobank and compared how they sorted the same people [s1].

Where they agreed, and where they didn't

About one-quarter of participants were classified as having obesity under both frameworks, and most of those were assigned to advanced stages, with high concordance for established disease [s1]. In other words, when someone already has organ damage or significant comorbidity, both systems tend to catch it and to rate it as serious.

The divergence was concentrated in the early stages [s1]. The DMO captured a broader spectrum of mild or subclinical organ dysfunction, while EOSS emphasised more advanced manifestations of obesity-related disease [s1]. The authors attribute the mismatch to how each system operationalises particular domains — metabolic, cardiovascular and mental-health among them — rather than to a deep conceptual disagreement [s1]. They describe the two as complementary approaches reflecting distinct concepts of disease severity, not rivals [s1].

One further finding underlines how unstable early-stage assignment can be. When BMI was combined with different measures of fat distribution, roughly half of participants were reclassified [s1]. That is a large number, and it is a caution: which body-composition parameter a clinician reaches for can move a person across the preclinical/clinical boundary.

Why this matters

The Lancet Commission framing was designed in part to change who gets treated. Under it, a diagnosis of clinical obesity is meant to justify treatment on its own terms, while preclinical obesity is a risk state warranting monitoring and risk reduction [s2]. If two credible systems disagree about who lands on each side of that line, then the label a patient receives — and, downstream, their access to obesity-directed care or a trial — may depend on which framework a clinic or payer adopts.

This study cannot say which system is closer to the truth, and it does not try to. It is a cross-classification exercise in one predominantly middle-aged, largely European-ancestry cohort, and the UK Biobank is known to be healthier and less deprived than the general population, which limits how far the stage distributions generalise [s1]. Both frameworks were also modified to fit the available data, so the comparison is between operational approximations, not the definitions as their authors wrote them [s1]. A domain the biobank captures poorly — the mental-health component both systems include — is one of the places the authors trace the divergence to, which is a reminder that a classification is only as good as the data feeding it [s1].

The reclassification finding is the most operationally uncomfortable. That combining BMI with different fat-distribution parameters moved roughly half of participants across categories means the early-stage line is not a fixed property of a patient but partly an artefact of which measurement a clinic happens to use [s1]. For the sickest patients that instability barely matters — both systems find them regardless — but for the large group hovering at the threshold of a disease label, the tool of measurement can be decisive, and neither framework yet specifies one canonical way to measure adiposity [s1].

What to watch

The open question is prognostic: does being called clinical obesity under the DMO, or a higher EOSS stage, actually predict worse outcomes better than BMI alone, and do the two systems predict the same events? This paper maps where they diverge; it does not test which division tracks future heart attacks, diabetes or death. That validation — ideally in cohorts more diverse than the UK Biobank — is what would tell a clinician which line to trust when a patient sits in the contested early zone.

This article is informational and does not constitute medical advice.

Sources

• [s1] Divergent Obesity Classification Between the Edmonton Obesity Staging System and the Lancet Commission Model. Obesity, 2 September 2026. https://doi.org/10.1002/oby.70295
• [s2] Definition and diagnostic criteria of clinical obesity. The Lancet Diabetes & Endocrinology, 14 January 2025. https://doi.org/10.1016/S2213-8587(24)00316-4
• [s3] A proposed clinical staging system for obesity. International Journal of Obesity, 3 February 2009. https://doi.org/10.1038/ijo.2009.2