John Murphy, M.D., M.P.H., D.P.H. President COVID-19 Long-haul Foundation<\/p>\n\n\n\n
Post\u2013acute sequelae of SARS-CoV-2 infection (PASC), commonly termed Long COVID, represents a heterogeneous post-infectious condition characterized by multisystem symptoms persisting beyond acute viral illness. Despite rapid advances in immunology, vascular biology, and neuroinflammation, therapeutic development has proceeded in a landscape of incomplete mechanistic certainty and limited diagnostic standardization. This has created a clinically consequential phenomenon: therapeutic misclassification<\/strong>, in which patients are assigned treatments targeting incorrect or partially relevant biological pathways due to overlapping symptom clusters, non-specific biomarkers, and evolving disease models.<\/p>\n\n\n\n This review examines the risk of therapeutic misclassification in Long COVID, with emphasis on immune, thromboinflammatory, autonomic, and neurocognitive phenotypes. It synthesizes evidence from clinical trials, observational cohorts, mechanistic studies, and guideline statements to demonstrate how diagnostic ambiguity propagates into treatment heterogeneity and potential iatrogenic harm. The analysis further explores how competing mechanistic frameworks\u2014viral persistence, immune dysregulation, endothelial dysfunction, microvascular injury, and metabolic impairment\u2014have led to divergent therapeutic strategies without robust endotype stratification.<\/p>\n\n\n\n Long COVID sits at the intersection of infectious disease, immunology, cardiology, neurology, and vascular medicine. Yet unlike classical syndromic conditions in which pathophysiology is well-defined prior to therapeutic consensus, Long COVID has developed in reverse: treatment hypotheses have proliferated before mechanistic convergence<\/strong>.<\/p>\n\n\n\n This inversion of traditional translational medicine has produced a central clinical risk: patients with similar symptom profiles may receive fundamentally different\u2014and sometimes opposing\u2014therapeutic interventions, including:<\/p>\n\n\n\n The absence of validated biomarkers for definitive disease subtyping amplifies this variability.<\/p>\n\n\n\n Therapeutic misclassification is defined here as:<\/p>\n\n\n\n The assignment of a treatment strategy based on an incorrect or incomplete inference of the dominant biological mechanism driving disease in a given patient.<\/p>\n<\/blockquote>\n\n\n\n In Long COVID, misclassification arises from three structural features:<\/p>\n\n\n\n Unlike classic diagnostic error, therapeutic misclassification does not require a wrong diagnosis\u2014only a wrong mechanistic attribution<\/strong>.<\/p>\n\n\n\n Multiple partially validated hypotheses now coexist in the literature:<\/p>\n\n\n\n Evidence suggests persistent immune activation and T-cell dysfunction in subsets of patients, including altered cytokine profiles and inhibitory receptor upregulation.\u00b9<\/p>\n\n\n\n Some studies propose ongoing viral antigen persistence as a driver of immune activation, although direct replication-competent virus detection in Long COVID remains inconsistent.\u00b2<\/p>\n\n\n\n Markers of endothelial activation (e.g., von Willebrand factor, ICAM-1) and impaired microcirculatory regulation suggest vascular involvement.\u00b3<\/p>\n\n\n\n Studies report fibrin structural abnormalities, platelet activation, and impaired fibrinolysis in subsets of patients, though reproducibility varies across laboratories.\u2074<\/p>\n\n\n\n Reduced oxidative phosphorylation capacity and altered metabolomic signatures have been documented, linking energy failure to fatigue phenotypes.\u2075<\/p>\n\n\n\n Each model is biologically plausible, but none is universally present across all patients.<\/p>\n\n\n\n A defining feature of Long COVID is that distinct biological processes converge on a shared symptom phenotype<\/strong>:<\/p>\n\n\n\n This convergence produces a false clinical inference:<\/p>\n\n\n\n Similar symptoms imply similar biology.<\/p>\n<\/blockquote>\n\n\n\n This assumption is frequently incorrect and forms the conceptual basis of therapeutic misclassification.<\/p>\n\n\n\n Perhaps the most prominent example of therapeutic misclassification risk in Long COVID involves anticoagulation strategies.<\/p>\n\n\n\n Proponents of a thromboinflammatory model propose that:<\/p>\n\n\n\n However, major guidelines emphasize caution regarding extended anticoagulation in non-hospitalized post-COVID populations due to:<\/p>\n\n\n\n The central risk arises when:<\/p>\n\n\n\n This leads to pathway over-attribution based on non-specific symptomatic response<\/strong>.<\/p>\n\n\n\n Another major domain of therapeutic risk involves immune-targeting interventions.<\/p>\n\n\n\n In inflammatory-dominant phenotypes, immunomodulation may be rational; however, in immune-exhaustion\u2013dominant patients, further suppression may worsen:<\/p>\n\n\n\n Conversely, immune stimulation in patients with autoimmune or hyperinflammatory signatures may exacerbate:<\/p>\n\n\n\n Thus, bidirectional immune-targeting errors are structurally likely in absence of biomarker stratification<\/strong>.<\/p>\n\n\n\n A particularly underappreciated form of therapeutic misclassification occurs in patients with neurocognitive and autonomic symptoms.<\/p>\n\n\n\n Without objective biomarkers, patients may be classified as having:<\/p>\n\n\n\n rather than:<\/p>\n\n\n\n This can lead to inappropriate treatment pathways including:<\/p>\n\n\n\n A key structural driver of misclassification is the absence of:<\/p>\n\n\n\n Even promising biomarker candidates (cytokines, endothelial markers, fibrin abnormalities) show:<\/p>\n\n\n\n This creates a diagnostic vacuum filled by mechanistic inference rather than definitive classification<\/strong>.<\/p>\n\n\n\n Therapeutic misclassification extends beyond individual care into research methodology.<\/p>\n\n\n\n Trials frequently include:<\/p>\n\n\n\n If only a subset of patients responds to a therapy, mixed enrollment produces:<\/p>\n\n\n\n Conversely, uncontrolled studies may produce:<\/p>\n\n\n\n Therapeutic misclassification in Long COVID raises several ethical concerns:<\/p>\n\n\n\n Importantly, the ethical challenge is not uncertainty itself, but action taken without sufficient mechanistic resolution<\/strong>.<\/p>\n\n\n\n Therapeutic misclassification in Long COVID arises from a fundamental mismatch between:<\/p>\n\n\n\n This mismatch produces a clinical environment in which:<\/p>\n\n\n\n Long COVID represents a paradigmatic case of post-infectious heterogeneity in which overlapping clinical phenotypes conceal fundamentally distinct biological mechanisms. In this context, therapeutic misclassification is not an incidental clinical error but an expected systemic outcome of incomplete endotype resolution<\/strong>.<\/p>\n\n\n\n Mitigating this risk requires the development of:<\/p>\n\n\n\n Without these advances, therapeutic strategies risk remaining empirically driven rather than biologically targeted, perpetuating variability in outcomes and potential iatrogenic harm.<\/p>\n\n\n\n If Part I established that therapeutic misclassification in Long COVID arises from mechanistic uncertainty and phenotypic overlap, Part II addresses its more consequential downstream expression: the misalignment of pharmacologic class with underlying biological endotype<\/strong>.<\/p>\n\n\n\n In practice, this is where abstract diagnostic ambiguity becomes clinical harm or inefficacy\u2014because treatment decisions are not neutral hypotheses but active biological interventions.<\/p>\n\n\n\n Long COVID presents a particularly high-risk environment because multiple high-impact drug classes are simultaneously plausible yet mechanistically incompatible across patient subgroups.<\/p>\n\n\n\n The rationale for anticoagulant and antiplatelet therapy in Long COVID stems from:<\/p>\n\n\n\n These findings support a thromboinflammatory model<\/strong>, which is biologically coherent in a subset of patients.\u00b9<\/p>\n\n\n\n Therapeutic misclassification occurs when:<\/p>\n\n\n\n This leads to a critical inference error:<\/p>\n\n\n\n vascular signaling abnormality \u2260 clinically significant thrombosis<\/p>\n<\/blockquote>\n\n\n\n In non-thrombotic phenotypes, anticoagulation may:<\/p>\n\n\n\n Thus, anticoagulation becomes a high-risk intervention when applied outside validated thrombotic endotypes<\/strong>.<\/p>\n\n\n\n Immunomodulation in Long COVID presents a symmetric classification problem:<\/p>\n\n\n\n This creates a bidirectional therapeutic risk landscape<\/strong>.<\/p>\n\n\n\n Glucocorticoids may be appropriate in acute hyperinflammatory states, but in post-acute disease:<\/p>\n\n\n\n Conversely, withholding immunosuppression in true inflammatory phenotypes may allow:<\/p>\n\n\n\n Because inflammatory biomarkers (e.g., IL-6, TNF-\u03b1) overlap across multiple phenotypes, clinicians risk:<\/p>\n\n\n\n treating cytokine elevation as a uniform therapeutic target rather than a context-dependent signal<\/p>\n<\/blockquote>\n\n\n\n Antiviral therapy in Long COVID assumes:<\/p>\n\n\n\n While biologically plausible, evidence remains heterogeneous across tissues and cohorts.\u00b2<\/p>\n\n\n\n Antiviral misclassification occurs when:<\/p>\n\n\n\n This creates a causal inversion error<\/strong>:<\/p>\n\n\n\n symptom response \u2260 proof of viral persistence<\/p>\n<\/blockquote>\n\n\n\n One of the most consequential misclassification domains in Long COVID is neurocognitive and autonomic symptom presentation.<\/p>\n\n\n\n Patients presenting with:<\/p>\n\n\n\n may be misclassified as having:<\/p>\n\n\n\n This arises because:<\/p>\n\n\n\n However, emerging evidence suggests contributions from:<\/p>\n\n\n\n Misclassification here can lead to:<\/p>\n\n\n\n A frequent clinical assumption is that persistent fatigue reflects:<\/p>\n\n\n\n This leads to graded exercise prescriptions.<\/p>\n\n\n\n However, in a subset of Long COVID patients:<\/p>\n\n\n\n Thus, treating metabolic impairment as deconditioning produces:<\/p>\n\n\n\n physiologic stress amplification rather than recovery<\/p>\n<\/blockquote>\n\n\n\n Most Long COVID trials enroll patients based on:<\/p>\n\n\n\n This results in biologically mixed cohorts.<\/p>\n\n\n\n When a therapy targets a specific pathway (e.g., endothelial dysfunction), mixed enrollment leads to:<\/p>\n\n\n\n This is one of the most important but underrecognized consequences of misclassification.<\/p>\n\n\n\n Conversely, uncontrolled studies may show improvement in:<\/p>\n\n\n\n without controlling for:<\/p>\n\n\n\n Leading to:<\/p>\n\n\n\n false attribution of efficacy to incorrect biological mechanisms<\/p>\n<\/blockquote>\n\n\n\n Therapeutic misclassification is not solely a clinical error\u2014it is structurally reinforced by system-level factors:<\/p>\n\n\n\n These factors collectively create a high-entropy therapeutic environment<\/strong>.<\/p>\n\n\n\n The most important conceptual advance in understanding therapeutic misclassification is recognizing it as:<\/p>\n\n\n\n an emergent property of incomplete disease modeling under high phenotypic similarity<\/p>\n<\/blockquote>\n\n\n\n Rather than isolated clinician error, it arises from:<\/p>\n\n\n\n Thus, misclassification is structural, predictable, and system-wide<\/strong>.<\/p>\n\n\n\n Although full guideline development lies beyond this review, emerging mitigation strategies include:<\/p>\n\n\n\n Therapeutic misclassification in Long COVID is most evident at the level of pharmacologic intervention, where anticoagulants, immunomodulators, antivirals, neuropsychiatric agents, and metabolic therapies may be applied to biologically mismatched patient subgroups.<\/p>\n\n\n\n This mismatch is not incidental but structurally embedded in current diagnostic frameworks. Without biomarker-defined endotypes, treatment selection remains vulnerable to inference errors, leading to variable efficacy, potential harm, and inconsistent clinical trial outcomes.<\/p>\n\n\n\n The resolution of this problem requires a shift from phenotype-driven medicine to mechanism-anchored therapeutic stratification<\/strong>.<\/p>\n\n\n\n The central conclusion of this review is that therapeutic misclassification in Long COVID is not an incidental phenomenon arising from clinical uncertainty, but rather a systematic consequence of biological heterogeneity combined with incomplete mechanistic resolution<\/strong>. Across immune, vascular, neurocognitive, autonomic, and metabolic domains, overlapping symptom phenotypes obscure fundamentally distinct underlying disease processes. As a result, treatment strategies are frequently inferred from symptom clusters rather than validated biological endotypes.<\/p>\n\n\n\n This creates a recurring error pattern in clinical practice: phenotype convergence without pathway convergence<\/strong>, leading to mechanistically mismatched therapy.<\/p>\n\n\n\n A useful conceptual model is to define Long COVID as a high-entropy syndrome<\/strong>, characterized by:<\/p>\n\n\n\n In such systems, classical diagnostic reasoning\u2014which assumes relatively stable mapping between phenotype and mechanism\u2014becomes unreliable.<\/p>\n\n\n\n This is particularly relevant when interventions are mechanism-specific (e.g., anticoagulants, immunomodulators, antivirals), because incorrect pathway assignment leads directly to therapeutic misclassification.<\/p>\n\n\n\n Current clinical frameworks rely heavily on symptom clusters such as:<\/p>\n\n\n\n However, symptom clustering alone fails to resolve biological heterogeneity because:<\/p>\n\n\n\n Thus, symptom-based classification is necessary for clinical communication but insufficient for therapeutic precision.<\/p>\n\n\n\n A defining feature of Long COVID is mechanistic overlap across disease domains<\/strong>:<\/p>\n\n\n\n This interdependence means that:<\/p>\n\n\n\n a single symptom cluster may arise from multiple independent or interacting biological systems<\/p>\n<\/blockquote>\n\n\n\n Therefore, assigning a single dominant mechanism without biomarker confirmation risks systematic misclassification.<\/p>\n\n\n\n Therapeutic misclassification has several clinically significant consequences:<\/p>\n\n\n\n Patients may receive therapies targeting the wrong mechanism while the dominant pathology remains unaddressed.<\/p>\n\n\n\n Interventions such as anticoagulation or immunosuppression may introduce harm when applied outside their biologically appropriate subgroup.<\/p>\n\n\n\n Heterogeneous enrollment obscures treatment effects in responsive subgroups, leading to false-negative results.<\/p>\n\n\n\n Observational improvement in unstratified populations may be incorrectly generalized as proof of mechanism.<\/p>\n\n\n\n The review supports a transition from phenotype-based classification to endotype-based classification<\/strong>, defined as:<\/p>\n\n\n\n biologically distinct disease subgroups characterized by reproducible molecular or functional signatures<\/p>\n<\/blockquote>\n\n\n\n In Long COVID, candidate endotypes include:<\/p>\n\n\n\n Importantly, these are not mutually exclusive and may coexist within individuals.<\/p>\n\n\n\n To reduce therapeutic misclassification, we propose a structured, three-tier framework integrating biomarker assessment, mechanistic stratification, and treatment alignment.<\/p>\n\n\n\n Identify probable Long COVID and establish symptom domains without assigning mechanism.<\/p>\n\n\n\n A clinical phenotype map<\/strong>, not a treatment decision.<\/p>\n\n\n\n Assign patients to dominant biological domains using measurable circulating markers.<\/p>\n\n\n\n A weighted endotype probability profile<\/strong>, not a definitive diagnosis.<\/p>\n\n\n\n Match therapy to dominant biological signal rather than symptom cluster.<\/p>\n\n\n\n Treatment selection should follow biological dominance, not symptomatic resemblance.<\/p>\n<\/blockquote>\n\n\n\n To reduce harm, any therapeutic decision should be constrained by three safeguards:<\/p>\n\n\n\n No mechanism-targeted therapy should be initiated solely on symptom presentation.<\/p>\n\n\n\n Exclude incompatible therapies across domains (e.g., anticoagulation in absence of coagulation signal).<\/p>\n\n\n\n Repeat biomarker evaluation due to dynamic disease expression.<\/p>\n\n\n\n To reduce misclassification bias in research:<\/p>\n\n\n\n Trials should predefine biological subgroups using biomarker panels.<\/p>\n\n\n\n Outcomes should correspond to targeted pathway (e.g., endothelial function, cytokine normalization).<\/p>\n\n\n\n Allow real-time subgroup refinement based on interim biomarker response.<\/p>\n\n\n\n Symptom-based enrollment alone should be considered insufficient.<\/p>\n\n\n\n Therapeutic misclassification raises distinct ethical issues:<\/p>\n\n\n\n Regulatory frameworks may need to evolve toward mechanism-informed authorization pathways<\/strong>.<\/p>\n\n\n\n Key limitations include:<\/p>\n\n\n\n These limitations mean the proposed framework is conceptual and translational rather than definitive clinical guidance<\/strong>.<\/p>\n\n\n\n Therapeutic misclassification in Long COVID represents a predictable consequence of high biological heterogeneity combined with incomplete mechanistic resolution. Symptom overlap across immune, vascular, neurocognitive, and metabolic domains leads to frequent assignment of therapies that may not correspond to the dominant disease mechanism in individual patients.<\/p>\n\n\n\n To address this, we propose a structured framework integrating:<\/p>\n\n\n\n This model shifts clinical reasoning from symptom-driven empiricism to mechanism-constrained precision medicine<\/strong>, with the goal of reducing iatrogenic risk, improving trial validity, and enabling biologically coherent treatment strategies.<\/p>\n\n\n\n Conceptual Origins, Diagnostic Uncertainty, and the Structural Basis of Treatment Error John Murphy, M.D., M.P.H., D.P.H. President COVID-19 Long-haul Foundation Abstract Post\u2013acute sequelae of SARS-CoV-2 infection (PASC), commonly termed Long […]<\/p>\n","protected":false},"author":2,"featured_media":15075,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[119,289,290,421],"tags":[],"class_list":["post-15000","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-diagnostics-covid-19","category-long-haul-disease","category-long-term-effects","category-pasc"],"_links":{"self":[{"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=\/wp\/v2\/posts\/15000","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=15000"}],"version-history":[{"count":4,"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=\/wp\/v2\/posts\/15000\/revisions"}],"predecessor-version":[{"id":15012,"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=\/wp\/v2\/posts\/15000\/revisions\/15012"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=\/wp\/v2\/media\/15075"}],"wp:attachment":[{"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=15000"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=15000"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/cov19longhaulfoundation.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=15000"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
\n\n\n\n1. Introduction: The Problem of Treating a Mechanistically Unresolved Disease<\/h2>\n\n\n\n
\n
\n\n\n\n2. Defining Therapeutic Misclassification<\/h2>\n\n\n\n
\n
\n
\n\n\n\n3. Mechanistic Pluralism in Long COVID: A Double-Edged Problem<\/h2>\n\n\n\n
3.1 Immune dysregulation and exhaustion<\/h3>\n\n\n\n
3.2 Persistent viral antigen or reservoirs<\/h3>\n\n\n\n
3.3 Endothelial and microvascular dysfunction<\/h3>\n\n\n\n
3.4 Coagulation abnormalities and immunothrombosis<\/h3>\n\n\n\n
3.5 Metabolic and mitochondrial dysfunction<\/h3>\n\n\n\n
\n\n\n\n4. The Core Driver of Misclassification: Phenotypic Convergence Without Pathway Convergence<\/h2>\n\n\n\n
Biological pathway<\/th> Shared clinical output<\/th><\/tr><\/thead> Immune dysregulation<\/td> fatigue, malaise<\/td><\/tr> Microvascular dysfunction<\/td> exercise intolerance<\/td><\/tr> Neuroinflammation<\/td> cognitive impairment<\/td><\/tr> Metabolic failure<\/td> post-exertional symptoms<\/td><\/tr> Autonomic dysfunction<\/td> tachycardia, orthostasis<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n \n
\n\n\n\n5. Anticoagulation Misclassification: A Central Example<\/h2>\n\n\n\n
5.1 Rationale for anticoagulation hypothesis<\/h3>\n\n\n\n
\n
5.2 Clinical counter-evidence and guideline caution<\/h3>\n\n\n\n
\n
5.3 Misclassification risk<\/h3>\n\n\n\n
\n
\n\n\n\n6. Immunomodulatory Misclassification<\/h2>\n\n\n\n
6.1 Immunosuppression risk<\/h3>\n\n\n\n
\n
6.2 Immunostimulation risk<\/h3>\n\n\n\n
\n
\n\n\n\n7. Autonomic and Neuropsychiatric Misclassification<\/h2>\n\n\n\n
\n
\n
\n
\n\n\n\n8. The Role of Biomarker Ambiguity<\/h2>\n\n\n\n
\n
\n
\n\n\n\n9. Clinical Trial Contamination by Misclassification<\/h2>\n\n\n\n
9.1 Heterogeneous enrollment<\/h3>\n\n\n\n
\n
9.2 Dilution of treatment effect<\/h3>\n\n\n\n
\n
9.3 False-positive mechanistic attribution<\/h3>\n\n\n\n
\n
\n\n\n\n10. Ethical Dimensions of Therapeutic Misclassification<\/h2>\n\n\n\n
\n
\n\n\n\n11. Conceptual Summary<\/h2>\n\n\n\n
\n
and<\/li>\n\n\n\n\n
\n\n\n\nConclusion <\/h2>\n\n\n\n
\n
Part II \u2014 Drug-Class\u2013Specific Misclassification, Clinical Trial Failure Modes, and System-Level Drivers of Iatrogenic Error<\/h3>\n\n\n\n
\n\n\n\n12. Introduction: From Conceptual Misclassification to Pharmacologic Consequence<\/h2>\n\n\n\n
\n\n\n\n13. Anticoagulants and Antiplatelet Agents: Overextension of the Thromboinflammatory Model<\/h2>\n\n\n\n
13.1 Biological rationale driving use<\/h3>\n\n\n\n
\n
\n\n\n\n13.2 Misclassification pathway<\/h3>\n\n\n\n
\n
\n
\n\n\n\n13.3 Clinical consequence<\/h3>\n\n\n\n
\n
\n\n\n\n14. Immunosuppressive Therapy: The Bidirectional Error Problem<\/h2>\n\n\n\n
14.1 Two opposing risks<\/h3>\n\n\n\n
Error type<\/th> Mechanism<\/th><\/tr><\/thead> Over-suppression<\/td> immune exhaustion or viral persistence worsened<\/td><\/tr> Under-suppression<\/td> autoimmune or hyperinflammatory state untreated<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
\n\n\n\n14.2 Steroid misclassification example<\/h3>\n\n\n\n
\n
\n
\n\n\n\n14.3 Mechanistic ambiguity as a driver of error<\/h3>\n\n\n\n
\n
\n\n\n\n15. Antiviral Therapy Misclassification: The Persistence Hypothesis Problem<\/h2>\n\n\n\n
15.1 Competing assumptions<\/h3>\n\n\n\n
\n
\n\n\n\n15.2 Misclassification pathway<\/h3>\n\n\n\n
\n
\n
\n\n\n\n15.3 Clinical consequence<\/h3>\n\n\n\n
\n
\n\n\n\n16. Neuropsychiatric Misclassification: Functional Labeling of Biological Disease<\/h2>\n\n\n\n
16.1 Diagnostic drift risk<\/h3>\n\n\n\n
\n
\n
\n\n\n\n16.2 Structural driver of misclassification<\/h3>\n\n\n\n
\n
\n
\n\n\n\n16.3 Clinical consequence<\/h3>\n\n\n\n
\n
\n\n\n\n17. Metabolic and \u201cDeconditioning\u201d Misclassification<\/h2>\n\n\n\n
17.1 The deconditioning assumption<\/h3>\n\n\n\n
\n
\n\n\n\n17.2 Misclassification risk<\/h3>\n\n\n\n
\n
\n
\n\n\n\n18. Clinical Trial Misclassification: The Hidden Driver of Evidence Failure<\/h2>\n\n\n\n
18.1 Heterogeneous enrollment problem<\/h3>\n\n\n\n
\n
\n\n\n\n18.2 Dilution of therapeutic signal<\/h3>\n\n\n\n
\n
\n\n\n\n18.3 False mechanistic validation<\/h3>\n\n\n\n
\n
\n
\n
\n\n\n\n19. System-Level Drivers of Misclassification<\/h2>\n\n\n\n
19.1 Fragmented specialty care<\/h3>\n\n\n\n
\n
19.2 Absence of validated biomarkers<\/h3>\n\n\n\n
\n
19.3 Rapid therapeutic diffusion<\/h3>\n\n\n\n
\n
19.4 Patient heterogeneity<\/h3>\n\n\n\n
\n
\n\n\n\n20. Conceptual Synthesis: Misclassification as a Systems Failure, Not an Individual Error<\/h2>\n\n\n\n
\n
\n
\n\n\n\n21. Toward a Prevention Framework (Pre-NEJM Translation Concept)<\/h2>\n\n\n\n
21.1 Biomarker-stratified prescribing<\/h3>\n\n\n\n
\n
21.2 Endotype-first diagnostic logic<\/h3>\n\n\n\n
\n
21.3 Mechanism-constrained trial design<\/h3>\n\n\n\n
\n
21.4 Adaptive therapeutic algorithms<\/h3>\n\n\n\n
\n
\n\n\n\nConclusion (Part II)<\/h2>\n\n\n\n
DISCUSSION<\/h2>\n\n\n\n
1. Principal Finding of This Review<\/h3>\n\n\n\n
\n\n\n\n2. Long COVID as a \u201cHigh-Entropy Clinical Syndrome\u201d<\/h3>\n\n\n\n
\n
\n\n\n\n3. Limitations of Symptom-Driven Taxonomy<\/h3>\n\n\n\n
\n
\n
\n\n\n\n4. Mechanistic Overlap as the Core Source of Misclassification<\/h3>\n\n\n\n
\n
\n
\n\n\n\n5. Consequences of Therapeutic Misclassification<\/h3>\n\n\n\n
5.1 Under-treatment of correct pathways<\/h4>\n\n\n\n
5.2 Exposure to unnecessary pharmacologic risk<\/h4>\n\n\n\n
5.3 Apparent treatment failures in clinical trials<\/h4>\n\n\n\n
5.4 Reinforcement of incorrect mechanistic models<\/h4>\n\n\n\n
\n\n\n\n6. Need for Endotype-Based Medicine<\/h3>\n\n\n\n
\n
\n
\n\n\n\nFRAMEWORK FOR PREVENTION OF THERAPEUTIC MISCLASSIFICATION<\/h2>\n\n\n\n
7. Overview of Proposed Framework<\/h3>\n\n\n\n
\n\n\n\nTIER I \u2014 PHENOTYPIC SCREENING (Clinical Entry Layer)<\/h2>\n\n\n\n
Objective:<\/h3>\n\n\n\n
Components:<\/h3>\n\n\n\n
\n
Output:<\/h3>\n\n\n\n
\n\n\n\nTIER II \u2014 BIOMARKER STRATIFICATION (Mechanistic Layer)<\/h2>\n\n\n\n
Objective:<\/h3>\n\n\n\n
Candidate domains and markers:<\/h3>\n\n\n\n
Immune domain<\/h4>\n\n\n\n
\n
Endothelial domain<\/h4>\n\n\n\n
\n
Coagulation domain<\/h4>\n\n\n\n
\n
Neuroimmune domain<\/h4>\n\n\n\n
\n
Metabolic domain<\/h4>\n\n\n\n
\n
Output:<\/h3>\n\n\n\n
\n\n\n\nTIER III \u2014 THERAPEUTIC ALIGNMENT LAYER (Intervention Mapping)<\/h2>\n\n\n\n
Objective:<\/h3>\n\n\n\n
Example alignment rules:<\/h3>\n\n\n\n
\n
Principle:<\/h3>\n\n\n\n
\n
\n\n\n\n8. SAFETY LAYER: MISCLASSIFICATION GUARDRAILS<\/h2>\n\n\n\n
8.1 Biological confirmation requirement<\/h3>\n\n\n\n
8.2 Cross-domain exclusion check<\/h3>\n\n\n\n
8.3 Temporal reassessment requirement<\/h3>\n\n\n\n
\n\n\n\n9. IMPLICATIONS FOR CLINICAL TRIAL DESIGN<\/h2>\n\n\n\n
9.1 Stratified enrollment<\/h3>\n\n\n\n
9.2 Mechanism-specific endpoints<\/h3>\n\n\n\n
9.3 Adaptive enrichment design<\/h3>\n\n\n\n
9.4 Avoid phenotype-only inclusion criteria<\/h3>\n\n\n\n
\n\n\n\n10. ETHICAL AND REGULATORY IMPLICATIONS<\/h2>\n\n\n\n
\n
\n\n\n\n11. LIMITATIONS OF CURRENT EVIDENCE BASE<\/h2>\n\n\n\n
\n
\n\n\n\n12. CONCLUSION<\/h2>\n\n\n\n
\n
\n\n\n\nReferences <\/h2>\n\n\n\n
\n