Psychogenic vs. Biomedical Models
The most consequential debate in ME/CFS history has been whether the illness is fundamentally psychological or biological. This debate has shaped research funding, clinical treatment, and patient experience for decades.
1 Historical Context
The psychogenic model of ME/CFS emerged in the 1970s–1980s, influenced by the observation that no single pathogen could be consistently identified as a cause, that routine laboratory tests were often normal, and that fatigue was associated with depression. The “biopsychosocial” model proposed by Wessely, White, and colleagues framed ME/CFS as a condition perpetuated by deconditioning, illness beliefs, and fear avoidance—treatable through exercise and cognitive restructuring. This model achieved institutional dominance in the UK (through the PACE trial and NICE’s pre-2021 guidelines), influencing insurance decisions, disability assessments, and clinical training. Patients who resisted exercise therapy or psychological treatment were characterized as having “unhelpful illness beliefs.”
The psychogenic framing carries measurable stigma consequences beyond its direct clinical effects. Attribution of ME/CFS to controllable or psychological causes is the strongest predictor of perceived stigma (n=499; Froehlich et al. 2022), and this perceived stigma mediates the relationship between negative causal attributions and worse functional status, social role satisfaction, and health-related quality of life (Froehlich et al. 2022). From the patient perspective, stigma operating through healthcare interactions is often reported as more distressing than the physical symptoms themselves (Terman, Cotler, and Jason 2019) (Deale and Wessely 2001).
The persistence of this framing is documented even in the post-COVID era. Ranque and Cogan (2025) published a narrative review in Frontiers in Medicine arguing that long COVID should be managed primarily as a “functional somatic disorder” (FSD), treatable through CBT and graded exercise (Ranque and Cogan 2025). They reported an online survey of 240 French senior internists in which 89% endorsed FSD as a possible cause of long COVID, 63% selected it as the primary cause, and only 9% identified a biological mechanism. Notably, 46% reported unwillingness to continue managing long COVID patients after negative standard workups, and 29% preferred not to see long COVID patients at all. The survey does not demonstrate a biological or clinical mechanism—it documents the depth to which the functional paradigm has been internalized by a substantial segment of the medical profession, shaping clinical encounters before the evidence has been fully evaluated.
2 Evidence for Biological Basis
The accumulated biological evidence against a psychogenic model is now overwhelming:
- Immune dysfunction: Reduced NK cell cytotoxicity, T cell exhaustion, altered cytokine profiles, and autoantibodies—none of which are features of psychological conditions (Hardcastle et al. 2016) (Che et al. 2025)
- Metabolic dysfunction: Impaired mitochondrial respiration, hypometabolic state, and cellular energy deficits documented by metabolomics and Seahorse analyzer studies (Naviaux et al. 2016) (Mandarano et al. 2020)
- Autonomic dysfunction: Objective documentation of POTS, orthostatic hypotension, and impaired heart rate variability (Hoad et al. 2008) (Newton et al. 2007)
- Neuroinflammation: PET imaging demonstrating microglial activation (Raijmakers et al. 2021)
- Neurotransmitter deficiency: CSF catecholamine deficiency documented by the NIH study (Walitt et al. 2024)
- Objective exercise intolerance: Two-day CPET demonstrating measurable physiological deterioration after exertion—impossible to produce through psychological mechanisms—though group-average VO₂ decline is contested (positive studies (Keller et al. 2024) vs null replication (Mancini et al. 2026)); elevated RPE and chronotropic incompetence are consistent findings across studies
- Physiological (not psychological) PEM features in deep phenotyping: The NIH deep-phenotyping study, sometimes cited to support a psychological explanation of PEM, in fact documented multiple physiological abnormalities—autonomic dysfunction, differential cerebrospinal fluid catecholamine and metabolite profiles, and lower post-exercise cortisol responses—and found that peak measures did not correlate with effort preference (Walitt et al. 2024) (Appelman et al. 2025). This is inconsistent with the claim that effort avoidance dictates V̇O2max or explains PEM.
3 The Role of Psychological Factors
Acknowledging the biological basis of ME/CFS does not deny that psychological factors affect the patient experience. Depression, anxiety, and trauma are common—as consequences of chronic disabling illness, not as causes. Psychological suffering deserves treatment in its own right, but psychological treatment cannot address the underlying pathophysiology. The critical distinction is between “ME/CFS causes psychological distress” (correct) and “psychological distress causes ME/CFS” (unsupported).
A Norwegian qualitative study of 16 healthcare professionals who use psychologically and behaviourally informed approaches to CFS found that explanation alone is consistently insufficient: patients need embodied experiences of improvement — felt change in their own bodies — before the proposed explanatory “map” becomes credible and before their confidence that change is possible takes hold (Bakken, Mengshoel, and Synnes 2026). The HCPs interpreted this pattern through the construct of epistemic trust: the extent to which therapeutic input is experienced as credible, personally relevant, and applicable (Fonagy and Allison 2014).
This relational reading has empirical backing independent of the psychobehavioural model’s content. In a prospective CFS cohort (n=217), early therapeutic alliance, positive outcome expectations, and early change in fatigue-perpetuating beliefs jointly predicted improvement during cognitive behavioural therapy, explaining roughly a quarter of the variance in post-treatment fatigue (Heins, Knoop, and Bleijenberg 2013). A meta-analysis of more than 30,000 psychotherapy patients found a robust alliance–outcome association (r≈0.28) that is not specific to any one treatment model (Flückiger et al. 2018).
The implication is that what HCPs describe as the active ingredient — a validating relationship plus first-hand experience of change — is separable from the specific BPS causal claims (stress regulation, “reprogramming”) that the map also carries. This is consistent with the paper’s own healing-norms stance: validation precedes any reappraisal, and change is earned through the patient’s experience, not imposed by explanation. It does not validate the BPS model as a causal account of ME/CFS; it validates a relational mechanism that is equally available to biomedical and pacing-based care. This reading also accommodates the finding that belief-content change — catastrophizing — does not by itself explain CBT outcomes (Catastrophizing: A Weaponised Construct): the operative factor is the relationship and expectation, not the specific cognitive content.
Certainty: 0.53 (Heins 2013, ME/CFS-specific prospective, 0.55; Flückiger 2018, general-population meta-analysis discounted ×0.75 to 0.53; Bakken 2026 qualitative, 0.35).
Evidence type: prospective observational (Heins), meta-analysis of RCTs (Flückiger), qualitative HCP perspectives (Bakken).
Severity: unknown — none of the alliance studies stratify ME/CFS severity, and Bakken 2026 does not report patient severity.
Limitations: The alliance–outcome association is correlational, not a demonstrated causal mechanism; no RCT has isolated the relational/trust component from the BPS content in CFS. Reverse causality and baseline-severity confounding are unaddressed — more severely affected patients see more clinicians and may accumulate more alliance-relevant encounters — and retrospective “embodied improvement” cannot be distinguished from natural history or spontaneous remission narrated after the fact. The epistemic-trust construct is theoretically grounded but not yet validated as a measurable driver of CFS recovery (Fisher, Fonagy, and Zilcha-Mano 2025).
Replication: The alliance–outcome association is extensively replicated in general psychotherapy (Flückiger meta-analysis); in CFS specifically it rests on the single Heins cohort.
Consequence: Reframes the BPS debate away from “does the model explain the illness” toward “does the relationship and the patient’s felt improvement do the work” — a point that matters for both clinicians (build trust first, whatever model they hold) and patients (agency and validation are legitimate, not concessions).
The same Norwegian qualitative study documents the clinician side of a field the paper has elsewhere described from the patient side. The 16 healthcare professionals — all purposively recruited because they already use psychologically and behaviourally informed approaches — operationalised the biopsychosocial “map” in two overlapping orientations: one framing fatigue as cumulative stress and overload with impaired regulation, the other as maladaptive learning, symptom sensitisation, and self-sustaining predictive processes (“retraining the brain”) (Bakken, Mengshoel, and Synnes 2026).
Critically, these clinicians did not report confident certainty. They described supporting recovery within a “delicate balancing act”: being sufficiently challenging to support change while avoiding loss of trust, and navigating between validation and reappraisal, hope and realism, and agency and blame. They reported residual diagnostic uncertainty and explicit concern about how their approach would be viewed “if future research were to establish a treatable biological dysfunction with little or no role for the processes they currently targeted.” Some reported being criticised for “not believing patients” or minimising suffering, and several advised caution regarding certain patient organisations (Bakken, Mengshoel, and Synnes 2026).
This is the clinician-side mirror of the epistemic-injustice tension the paper documents from the patient side (Psychogenic vs. Biomedical Models): the contestation over CFS does not only harm patients — it also leaves clinicians insecure, defensive, and sometimes at odds with patient communities. It does not excuse psychosomatic attribution, but it shows that the BPS camp is internally heterogeneous and uncertain rather than a monolith confidently asserting the illness is psychological.
Certainty: 0.35 (qualitative, reflexive thematic analysis; purposive sample of HCPs who endorse the approach, so it cannot represent the range of clinician views).
Evidence type: qualitative HCP perspectives (Bakken 2026); contextualised against the paper’s existing patient-side critique (Bayliss et al. 2014) (Schomerus et al. 2026).
Severity: N/A — the study reports clinician experience, not patient outcomes.
Limitations: Single Norwegian specialist-healthcare context; the HCP sample is selected for a specific therapeutic orientation, and the patients they describe were likely receptive to it. The study does not address patients who do not improve or who deteriorate under these approaches, which the authors themselves flag as a needed direction.
Replication: Not yet replicated; this is one qualitative study. It converges with patient-side recovery-narrative work (Ingman, Chalder, and Lawrence 2025) in its emphasis on shifts in illness understanding, but the HCP perspective is novel.
Consequence: Humanises the contested field without endorsing it — showing that even clinicians inside the psychobehavioural camp are uncertain and vulnerable to the same polarisation the paper describes, which bears on how patients and clinicians might rebuild trust rather than entrench opposition.
4 Patient Advocacy and the Psychogenic Harm
The psychogenic model has caused measurable harm to patients:
- Treatment harm: GET prescribed under the psychogenic model has worsened many patients (Kindlon 2011)
- Diagnostic delay: Physicians trained to view ME/CFS as psychological often fail to diagnose or investigate the condition, resulting in years of delayed treatment. The diagnostic labyrinth is well-documented—across 12 qualitative studies, patients report navigating systems where their symptoms are repeatedly dismissed as psychological, somatoform, or not severe enough to warrant investigation (Bayliss et al. 2014)
- Insurance and disability denial: Patients report that psychogenic framing has contributed to denial of disability benefits and insurance claims
- Research underfunding: Classification as a psychological condition reduced biomedical research funding for decades
- Stigma: Patients report that the psychogenic label is the single most damaging aspect of their interaction with the medical system, more distressing than the illness itself. Perceived stigma independently predicts suicidal ideation after controlling for depression (McManimen et al. 2018) (Johnson et al. 2022). The healthcare dismissal experience—being disbelieved, trivialised, or labelled—is the most consistent finding across qualitative research spanning 1999–2024 (Asbring and Närvänen 2002) (Guise, McVittie, and McKinlay 2010) (Melby and Nair 2024)
- Media reinforcement: Content analysis of 280 UK newspaper articles found 70% emphasised psychological characterisations of ME/CFS, amplifying public stigma (Knudsen et al. 2011)
- Name-based stigma: The diagnostic label “chronic fatigue syndrome” generates more negative attributions than alternative names, indicating the illness name itself propagates stigma (Jason et al. 2002) Patient advocacy organizations have been instrumental in challenging the psychogenic model, demanding data transparency (PACE trial reanalysis), and advocating for guideline changes. The 2021 NICE guideline revision (National Institute for Health and Care Excellence 2021), which explicitly rejected the biopsychosocial model for ME/CFS, represents a landmark victory for evidence-based medicine driven in significant part by patient advocacy. The academic counterargument continues to build. Spanoghe, Molmans et al. (2026) published a peer-reviewed commentary in Frontiers in Medicine directly rebutting the FSD framing proposed by Ranque and Cogan (Spanoghe et al. 2026). Their critique identifies five structural weaknesses in the FSD position: (1) the original article is a narrative review without systematic evidence evaluation, risk-of-bias assessment, or GRADE-quality hierarchisation; (2) the FSD concept itself has poorly defined boundaries, heterogeneous criteria, and functions as a residual category rather than a well-delimited diagnostic entity; (3) the literature selection omits substantial biomedical evidence including viral persistence, neuroinflammation, endothelial dysfunction, skeletal muscle abnormalities, mitochondrial dysfunction, and autonomic impairment; (4) attributing primary etiological weight to psychological factors without longitudinal or causal analyses risks confusing consequences with causes; and (5) framing long COVID as predominantly psychosomatic in a context of scientific uncertainty causes measurable harm—distress, self-doubt, loss of agency, and isolation. The authors advocate instead for treating long COVID as a heterogeneous post-infectious multisystemic condition requiring biomedical sub-classification, mechanistic investigation, and targeted interventions.
Schomerus et al. (2026) articulate five specific, interconnected harms that follow from psychologizing LC and ME/CFS, whether primary or secondary:
Harm 1: Iatrogenic deterioration from activation therapy. A psychological attribution implies that deconditioning, avoidance, and illness beliefs must be treated with graded activity. Patient surveys document that 72% of LC respondents reported “damaging or incorrect treatments, predominantly activating therapies during rehabilitation or medical recommendations to exercise” (Hammer et al. 2025). The US patient survey by Eckey et al. (2025, n=3,867) found that 50% of those who experienced graded exercise therapy reported substantial worsening, with a further 30% reporting moderate or mild deterioration; only 10% experienced any improvement (Eckey et al. 2025). These harm reports are biologically anchored: PEM is driven by impaired energy metabolism (Chapter Energy Metabolism and Mitochondrial Function) and exertion above the lowered threshold produces measurable tissue damage (Appelman et al. 2024).
Harm 2: Invalidation of illness experience. When symptoms are framed as products of dysfunctional illness beliefs rather than biological disease, the patient’s body perception—their only available instrument for managing PEM through pacing—is pathologized. Psychologization undermines the patient’s capacity for self-management by reclassifying pacing as “avoidance” and genuine PEM as “catastrophizing” (Section Inflammation Changes How the Brain Senses the Body). Büchner et al. (2025, n=1,156) found that 46.8% of LC patients were told their symptoms were psychological; this experience was strongly associated with perceived stigma (\(p < 0.001\)) and mediated worse mental health outcomes (Büchner et al. 2025).
Harm 3: Neglect of biomedical diagnosis and treatment. A psychological framework reduces the perceived need for biomedical investigation. Individual symptoms and comorbidities—POTS (postural orthostatic tachycardia syndrome), MCAS (mast cell activation syndrome), orthostatic hypotension—are somatically treatable in many cases, yet patients report that psychosomatic attribution prevents appropriate workup. The D-A-CH consensus explicitly recommends systematic exclusion of treatable somatic conditions before considering psychiatric differentials (Hoffmann et al. 2024). A psychological “F-diagnosis” can produce negative cascading effects across treatment, employment law, and insurance as physical complaints receive less attention in the presence of psychiatric labels—a well-documented pattern called diagnostic overshadowing, where a mental health label reduces the perceived need to investigate physical symptoms.
Harm 4: Medico-legal disadvantage. In disability assessment, insurance evaluation, and workers’ compensation, the assumption of psychological causation is practically unfalsifiable: rejection of the psychosomatic disease model is itself interpreted as evidence of its relevance (lack of insight, somatization). Unlike in therapeutic contexts where the patient can decline an unhelpful model, in medico-legal settings the attribution carries binding consequences.
Harm 5: Family and social harm. The same logic that pathologizes patients also pathologizes their caregivers. Parents of children with ME/CFS who protect their child from activity that triggers PEM can be accused of facilitating avoidance, reinforcing illness behavior, or even perpetrating Münchhausen-by-proxy syndrome—with documented cases of child protection proceedings and loss of custody (Schomerus et al. 2026). Similar dynamics affect adult children caring for parents with ME/CFS and parents with ME/CFS who are assessed for childcare capacity.
(Certainty: 0.70 for the existence of each documented harm (individually supported by quantitative surveys, clinical consensus, and qualitative evidence per above). The framework’s claim that these harms are causally mediated by psychologization — rather than generic to contested illness — has not been formally validated. A simpler alternative hypothesis is that any contested chronic illness generates similar harms; psychologization is one form of contested-illness dismissal, and the harms it produces may not be psychologization-specific. The framework is strongest when read as a taxonomy of documented harms requiring prospective testing of causal links. Replication status: partially — each sub-component has independent supporting evidence, but the framework as a whole and its causal claims have not been empirically validated against alternative models. Falsifiability: The framework predicts that psychologization exposure → functional decline is mediated by pacing disruption (Harm 2), testable via the prospective cohort design proposed in Prospective Cohort of Psychologization → Functional Decline. The framework would be weakened if psychologization exposure does not independently predict functional decline after controlling for objective baseline severity, or if a generic “contested illness” measure predicts decline equally well.)
Drug trials routinely monitor and report adverse events as a regulatory requirement, but psychosocial intervention trials for ME/CFS — despite evidence that activation-based approaches cause harm in a substantial minority of patients — operate under far laxer standards. The general pattern of inadequate harm monitoring in GET/CBT trials is independently documented: Kindlon (2011, (Kindlon 2011)) found systematic underreporting of harms in ME/CFS psychosocial trials, and Cooper et al. (2025, (Cooper and Papadopoulos 2025)) demonstrated via systematic review that GET carries a 3–5× higher harm rate than pacing. Schomerus et al. (2026) note a parallel pattern in LC psychotherapy trials: a systematic review found that most included studies provided no adverse effects information — though this specific statistic (reported as a secondary citation within the Schomerus essay) should be treated with caution pending dedicated independent verification (Schomerus et al. 2026). PEM, the decisive clinical parameter for safety, is frequently not assessed or not assessed differentially as an adverse outcome in psychosocial trials. Many studies lack control groups, making it impossible to distinguish genuine treatment effects from spontaneous recovery or non-specific effects such as placebo and natural history.
This regulatory gap may reflect divergent institutional histories (drug and behavioural intervention trials evolved under different regulatory systems) and genuine measurement challenges (defining adverse events in psychosocial trials is inherently harder than in drug trials) rather than deliberate neglect. Regardless of the cause, the consequence is the same: inadequate safety data for interventions actively prescribed to a population where harm rates in patient surveys range from 51–74% (Kindlon 2011).
(Certainty: 0.65. The general pattern of inadequate monitoring is independently documented by Kindlon (2011) and Cooper (2025). The specific LC psychotherapy statistic is a secondary citation within Schomerus 2026 — the general pattern is well-supported without relying on it. The regulatory-asymmetry claim (different safety standards for drug vs psychosocial trials) is a structural/regulatory fact, not an empirical hypothesis. Falsifiability: a dedicated systematic review of adverse event reporting in all published ME/CFS psychosocial intervention trials would establish the rate of adequate safety monitoring and either confirm or revise the specificity of the claim. Origin: literature — derived from Schomerus 2026’s argument, with supporting evidence from Cooper 2025.)
The entire evidence base supporting the therapeutic-alliance and “recovery-conceivable” reading is anchored in ambulatory, largely mild-to-moderate patients. None of the load-bearing studies stratify by ME/CFS severity: the alliance cohort (n=217) and the recovery systematic review do not report a severe or very-severe subgroup (Heins, Knoop, and Bleijenberg 2013) (Ingman et al. 2022), and Ingman explicitly cautions that its findings may not generalise to severe ME/CFS. The HCP-perspective study reports no patient severity at all, and its purposively recruited patients were likely engaged and receptive (Bakken, Mengshoel, and Synnes 2026).
Whether embodied improvement and epistemic trust operate at all for the bedbound patient — for whom “relearning to interpret bodily signals” collides with genuine physiological fragility and post-exertional symptom exacerbation — is unevidenced. The alliance finding is therefore best read as a claim about the ambulatory majority, not as a recovery mechanism that extends to the severe and very-severe patients where psychobehavioural framing has historically done the most harm (Psychogenic vs. Biomedical Models).
Consequence: The trust-building insight should not be silently generalised to severe ME/CFS, where a “recovery is conceivable” framing has materially different risks than it does for an ambulatory patient.
The single most important evidence gap: no longitudinal study has prospectively measured psychologization exposure and tracked functional decline while accounting for other factors that might explain the link. Büchner 2025 is cross-sectional (n=1,156); all other cited studies are qualitative or cross-sectional. Proposed: n ≥ 500 newly diagnosed, 36-month follow-up, quarterly assessments including psychologization exposure, SF-36, actigraphy, PEM frequency, CPET, and biomarker panel. Primary analysis: statistical model testing baseline psychologization as predictor of functional trajectory after controlling for objective severity measures.
The recovery debate in ME/CFS partly dissolves when its central term is disambiguated. “Recovery” is used to mean at least three incommensurate outcomes: full remission to premorbid functioning — the bar physicians apply (Devendorf et al. 2019); full recovery, reached by roughly one in four patients after CBT, versus exit from diagnostic criteria, reached by over two in three in the same cohort (Knoop et al. 2007); and self-rated “better”, reported by about 44% after CBT (Ingman et al. 2022). The “recovery conceivable” framing adds a fourth referent that is neither a biological endpoint nor a functional threshold: a shift in illness understanding coupled with felt improvement and renewed hope (Bakken, Mengshoel, and Synnes 2026).
The consequence is that much of the efficacy dispute — does a psychobehavioural approach “recover” patients — is contested at the level of the word itself, not the underlying data: the same study can be read as showing that recovery is a minority outcome (full remission), a common outcome (self-rated improvement), or a matter of orientation rather than outcome (conceivability). Whether any of these is the “right” referent, and whether a single measure can serve patients, clinicians, and researchers at once, remains an open conceptual question.
Consequence: A patient and a clinician can both be honest while disagreeing entirely about whether “recovery” happened — because they are measuring different things. Naming the three referents is a prerequisite for any claim about recovery rates to mean anything.