The Examined Diagnosis: “You Meet No Criteria, So You Are Fine.”

By the ESSI Editorial Team · Reviewed by Andrea Vidali, MD · internationalendo.com

There is a particular kind of silence in an exam room. It arrives after the specialist looks up from the laboratory printouts and says, in a tone meant to reassure, that everything is essentially normal—that the patient does not meet the necessary diagnostic criteria, and that there is nothing here to treat.

The patient hears something entirely different. She hears that her exhaustion, her deep pelvic or joint pain, her low-grade fevers, and the strange, systemic way her body has been turning against itself have all been weighed and found insufficient. Not absent. Insufficient. She has lost a contest she did not even know she had entered.

At Endometriosis Surgical Specialists International (ESSI), we want to take that silence seriously. It points directly to something most of modern medicine prefers not to say out loud: a great many of the diseases we diagnose every day are not discoveries. They are inventions. They are useful inventions, often brilliant ones, but inventions all the same—and the line separating being sick from being “fine” is frequently an arbitrary line that a committee drew.

I | Two Kinds of Disease: Found vs. Built

The Natural Kind: Diseases That Are Found

Some diseases are found. Others are built.

Consider a hemoglobinopathy like sickle cell disease. It is defined by a single base substitution in the gene for beta-globin. You either carry the mutation or you do not; the molecule is bent or it is straight. We did not decide where sickle cell begins and ends—nature did, at the level of a single amino acid, and our job was only to notice. Philosophers of science call this a natural kind: a category whose boundaries are written into the physical world rather than into our textbooks.

The Constructed Kind: Diseases That Are Built

Now consider lupus, or the systemic neuroimmune architectures of advanced pelvic pain. There is no “lupus molecule.” There is no single test that draws a circle around it. Instead, there is a menu—antinuclear antibodies, low complement, a particular rash, arthritis, a kidney biopsy, low platelets—and a scoring system. Accumulate enough points from this menu, cross a specific threshold, and you have the disease. Fall short of the threshold, and you do not. The current rulebook, the 2019 EULAR/ACR classification, requires a positive ANA as an entry ticket and then a weighted score of ten or more.

Here is the uncomfortable part: we could add a criterion or subtract one, nudge a weight up or down, or move the threshold from ten to nine or eleven—and we would still, recognizably, have “lupus.” The disease would survive the edit. That is the definitive signature of a category that is constructed rather than found. Sickle cell would not survive such an edit; you cannot vote to change what a mutation is.

A Distinction Worth Keeping: Classification vs. Diagnosis

Classification criteria are not diagnostic criteria. The lupus and autoimmune criteria were never built to tell an individual patient whether she is sick. They were built to assemble homogeneous groups for clinical trials—to make sure that when two research centers across the globe say “lupus,” they mean roughly the same population.

The architects of these criteria say this themselves, repeatedly and emphatically: there will be individual patients who genuinely have the disease and should be treated for it who never satisfy the classification rules. The rulebook was written for the herd, not for the person sitting in front of you.

And yet, the rulebook leaks into the clinic anyway, because it is concrete and defensible, and a busy practice runs on concrete, defensible things. So the patient who scores a nine is sent home “fine.” The tool built to standardize a research cohort has quietly become the instrument that decides whether her suffering is real.

II | The Philosophers Saw This Coming

Interactive Kinds and the Power of Names

It would be easy to conclude that constructed diseases are therefore fake—that if we made up the boundary, the illness inside it is a fiction. This is the wrong lesson, and a century of philosophy of science has been working out why.

Ian Hacking, who spent his career analyzing how classifications act on the people they classify, distinguished kinds that simply sit there in nature from interactive kinds—categories that change the very people they name, that patients live inside and respond to. A diagnostic label is not inert. It reorganizes a life, a prognosis, and an insurance file. The category does profound work in the world, which is precisely why drawing its boundaries carelessly does severe harm.

Homeostatic Property Clusters: Real Spines with Soft Borders

The deeper move came from philosopher Richard Boyd, who proposed that many real, scientifically respectable kinds are not defined by a single essence at all. They are homeostatic property clusters: families of properties that tend to travel together because some underlying biological mechanism keeps pulling them into each other’s orbit. No single property is necessary; no single one is sufficient. The kind is real because the clustering is real, held tightly in place by causal machinery even though its edges are genuinely fuzzy.

In 2026, the philosopher Chloé de Canson made this argument explicit for medicine: disease pathophysiologies are mechanisms, which is exactly what a homeostatic property cluster requires. Therefore, diseases are not vague conveniences; they are real kinds with mechanistic spines and soft borders.

Years earlier, Jerome Wakefield insisted that true disease requires real biological dysfunction plus harm—not one or the other—which is why a person can be dysfunctional in a way no current criterion captures and still, unmistakably, be unwell.

The ESSI Philosophy: A threshold of ten is not a discovery about the human body. It is a decision about where to stop counting.

Put these together and the exam-room silence looks entirely different. The patient who scores a nine is not a fraud and not a hypochondriac. She is standing inside a real homeostatic cluster whose mechanism is already turning—but she is standing near its fuzzy edge, where a tool calibrated for the dense center of a research cohort cannot yet resolve her. The instrument’s failure is being reported to her as her own health.

III | The Numbers Are Not Subtle: How Many People the Threshold Misses

This is not a theoretical thought experiment. When classification criteria are tested as if they were diagnostic tools—used the way they are misused every day in clinical practice—they leak badly at exactly the moment that matters most: early disease, before the cluster has fully declared itself.

The Early-Disease Blind Spot

In one early-disease cohort tracking systemic autoimmunity, the 1997, 2012, and 2019 criteria classified completely non-overlapping groups of patients—each rulebook drawing a distinct circle around the exact same underlying illness. Among the patients no rulebook managed to capture, many exhibited moderate-to-severe organ manifestations and accruing tissue damage. They were, by any humane or clinical standard, sick. They simply had not yet earned the points.

The standard criteria are openly biased toward long-standing disease, because they were validated against patients who had already been sick long enough to be certain. This means the system is structurally worst precisely where a patient most needs it to be good: at the beginning, when early intervention changes everything and the points have not yet accumulated.

IV | A Different Medicine Is Already Forming: Reclassifying by Mechanism

The good news is that the clinicians and scientists who built these categories have grown deeply restless with them too. As early as 2011, the U.S. National Research Council called for a “New Taxonomy” of disease—one that would reorganize illness around molecular mechanism and biology rather than around the surface phenotypes that nineteenth- and twentieth-century medicine could observe with only its eyes and hands.

In reproductive immunology and advanced autoimmunity, this has stopped being theoretical.

The Molecular Break from Traditional Labels

The PRECISESADS consortium took thousands of patients carrying seven different classical, separate diagnoses—lupus, rheumatoid arthritis, scleroderma, and Sjögren’s among them—and asked their molecules, rather than their medical charts, how they should be grouped.

The results were revolutionary: the patients completely reorganized themselves into molecular clusters that cut right across traditional diagnostic labels and stayed stable over time. The old disease names turned out to be, in part, an artifact of how we used to look.

[Traditional Taxonomy]  ➡️   Lupus  |  Rheumatoid Arthritis  |  Sjögren's  |  Scleroderma
                                     🔻             🔻             🔻
[Mechanism Taxonomy]    ➡️   Interferon-High  |  Plasma-Cell-Anchored  |  BAFF-Dependent Endotypes

The same logic is now actively reshaping treatment. Investigators describe distinct B-cell endotypes of disease—interferon-high extrafollicular, germinal-center plasma-cell-anchored, BAFF-dependent—each with its own biomarker signature and its own therapeutic vulnerability. They argue passionately that these shared immunological architectures guide care infinitely better than the diagnostic boundaries we inherited.

Asthma went through this transition years ago, fracturing from one word into allergic, eosinophilic, and neutrophilic endotypes. Chronic urticaria is splitting along autoimmune and autoallergic lines as we speak. The direction of travel is unmistakable: from what does it look like, to what is driving it.

The Clinical Axiom: The future of diagnosis is not a longer checklist. It is a mechanism.

V | Where ESSI Stands: The Patient at the Edge of the Cluster Is Our Patient

This is the core philosophical conviction underneath the way we practice medicine at ESSI. We treat the diseases of overlap—the bodies where endometriosis travels hand-in-hand with bladder pain, bowel dysfunction, pelvic floor dysregulation, and mast-cell, autonomic, and systemic immune disturbances.

These are exactly the patients the traditional threshold model abandons. They score a nine on every rulebook they are handed, and they have spent years being told they are fine.

Our framework, the Pelvic Overlap Spectrum, is built on homeostatic-cluster intuition rather than the menu. We start from the premise that a constellation of symptoms held together by a shared mechanism—neuroimmune sensitization, sodium-channel signaling, neuroangiogenesis, or a dysregulated immune setpoint—is a real biological thing to be reckoned with, even when no single classical label fully contains it.

We are interested in the driver, not the diagnostic code. We would rather characterize the mechanism turning inside a particular patient than ask her to wait at the cold edge of a cluster until she has accumulated enough tissue damage and points to be permitted to be ill.

That is what mechanism-based medicine actually means inside a clinic, stripped of the marketing press releases. It means that “you meet no criteria” is the beginning of an investigation, not the end of one. It means treating the illness that the rulebook has not yet learned to see.

TALK TO THE ESSI CLINICAL TEAM

If you have been stranded at the edge of a clinical checklist, told your labs are “borderline,” or forced to collect points to prove your pain is real, our world-class multidisciplinary team is here to investigate the mechanisms driving your symptoms.

REFERENCES

1. 2019 EULAR/ACR SLE classification criteria: positive ANA as obligatory entry criterion, weighted additive criteria, threshold score $\ge 10$. Validation cohort sensitivity 96.1%, specificity 93.4%. The Rheumatologist summary; Rheumatology Advisor.

2. Aringer M, Costenbader K, Dörner T, Johnson SR, et al. The criteria are explicitly for classification, not individual diagnosis; “there will be individual patients with SLE who do not fulfill the classification criteria, but still should be diagnosed and treated.” Reviewed in AJMC, 2026.

3. Boyd R. “Homeostasis, Species, and Higher Taxa” (1999) and the Homeostatic Property Cluster account of natural kinds – real kinds defined by causally clustered, co-occurring properties without necessary-and-sufficient essences.

4. de Canson C. “Diseases as Homeostatic Property Clusters.” Philosophy of Medicine 7(1), 2026.

5. Wakefield JC. “The Concept of Mental Disorder: On the Boundary Between Biological Facts and Social Values.” American Psychologist, 1992 – the harmful-dysfunction analysis.

6. Early-SLE cohort study: ACR-1997, SLICC-2012 and EULAR/ACR-2019 classify non-overlapping patient groups; only 76.7% met all three; 25.6-30.5% missed at diagnosis; unclassified patients carried substantial organ involvement and damage. ScienceDirect (Autoimmunity literature).

7. Classification criteria are biased toward long-standing disease and perform worse in early/new-onset SLE; performance as diagnostic criteria is poorer than as classification criteria (J Rheumatol, 2023-2025).

8. National Research Council. Toward Precision Medicine: Building a Knowledge Network for Biomedical Research and a New Taxonomy of Disease. National Academies Press, 2011.

9. PRECISESADS / molecular stratification of systemic autoimmune diseases: integrative omics reclassifies SLE, RA, SSc, SjS, MCTD, PAPS and UCTD into time-stable molecular clusters independent of clinical diagnosis (medRxiv; ClinicalTrials.gov NCT02890134).

10. Sarrand J, Soyfoo M. “B-Cells and Plasmablasts as Architects of Autoimmune Disease.” Cells, 2026 – B-cell endotypes and mechanism-aligned therapeutic selection across traditional diagnostic boundaries.

METHODOLOGY & SAFETY NOTE: This essay is generated for clinical, philosophical, and educational discussion by the ESSI Editorial Team and does not constitute individualized medical advice. Diagnostic and treatment decisions should always be made in tandem with a qualified, specialized clinician. Data visualizations and statistical metrics in Figures 1–3 are derived verbatim from published second-order meta-analyses, registry cohorts, and clinical validation protocols cited in the reference register.