Iatrogenic harm in eating disorder treatment: a lived experience-led call for epistemic justice and accountability Authors: Laurence Cobbaert Published: 2026 ABSTRACT Psychological treatments for eating disorders (EDs) can be beneficial, but treatment benefit is not, by itself, evidence of safety. A treatment may reduce ED symptoms while still causing iatrogenic harm (IH), and current evidence does not allow confident claims about whether IH is more common, less common, or as common as treatment benefit, because harms remain poorly defined, inconsistently monitored, and inadequately reported. This correspondence argues that the under-recognition of IH in ED care is not only a methodological problem, but also an epistemic one. When patients report being harmed or traumatised by treatment, their accounts may be dismissed, pathologised, or reframed as resistance, non-compliance, lack of insight, or treatment-interfering behaviour. Such responses can compound harm and prevent services from learning from patient testimony. The article further argues that IH is not experienced uniformly. Marginalised communities may encounter distinct forms of harm when treatment models are neuronormative, non-affirming, culturally incongruent, or inattentive to structural inequity. Person-centred ED care must therefore ask not only whether a treatment works on average, but for whom it is safe, meaningful, accessible, and accountable. The way forward requires prospective IH monitoring, patient-reported outcomes, open-ended and qualitative methods, repair-oriented practice, and lived experience-led co-design. Lived experience involvement is essential because people who have experienced harm can help define what counts as harm, which outcomes matter, how disclosures should be interpreted, and what accountability requires. Claims about the benefits or evidence-based status of psychological treatments for EDs are incomplete when safety is inferred from benefit or effectiveness alone while IH remains underdefined, undermeasured, and underreported. Background Psychological treatments for eating disorders (EDs) can be beneficial for some people. Acknowledging iatrogenic harm (IH) does not dispute this; it challenges the assumption that benefit implies safety. Effectiveness and safety are distinct questions, and evidence-based practice requires risk-benefit assessment that weighs effectiveness alongside harm and patient values [1, 2]. This matters where ED care is delivered amid limited options and expectations to engage with prescribed or manualised recovery models, including cognitive behavioural therapy (CBT) or family-based therapy (FBT) [3–5]. Psychological treatment may reduce ED symptoms, but this does not, by itself, establish that treatment is safe [1, 2]. Because a treatment may achieve its intended therapeutic effect while still causing adverse effects, symptom reduction does not preclude IH [3, 6–8]. This distinction is especially important when IH is not prospectively defined, systematically monitored, or reported [1]. Indeed, Kinnaird et al. [9] found that only three out of 23 randomised controlled trials of psychological treatments for anorexia nervosa defined unwanted events, none reported deterioration measures beyond weight loss or ED symptom deterioration, and only one measured patient-reported harms. Honkalampi et al. [10] similarly found that only 30% of randomised controlled trials of psychological interventions reported negative effects. Lived experience-led scholarship documents treatment-related trauma, invalidation, epistemic injury, neglect, and failures to consider identity, trauma, and social context in ED care [3, 7, 11–15]. Qualitative ED research provides more specific examples: Alford et al. [6] reported that autistic people described some ED treatment modalities as misattuned and, for some, potentially traumatic, while Sandsten et al. [8] reported that inpatient care could leave people with anorexia nervosa feeling kept alive but not supported to live. Furthermore, co-designed research shows that people with lived experience identify priorities and harms that conventional outcomes may miss [16]. Together, these findings suggest that harms remain underdeveloped in clinical research on ED psychological treatments, despite evidence of IH. The evidence base used to justify ED psychological treatments may privilege clinician-defined benefits while failing to capture patient-reported harms [3, 7, 13, 14]. Current evidence therefore does not allow confident claims that IH is more common, less common, or as common as treatment benefit; rather, IH has not been assessed adequately enough to support such comparisons [1, 9, 10]. IH should also be distinguished from negative experiences in psychological treatment, including emotional distress, symptom worsening, poor therapeutic fit, loss of trust, shame, boundary violations, unmet expectations, cultural bias, identity invalidation, and premature termination [17]. Not all constitute IH; some may be transient and self-limiting. However, they warrant investigation when unwanted, enduring, harmful, attributed by patients to treatment, or contributing to deterioration, trauma, stigma, loss of trust, relational harm, disengagement, or future avoidance of care [17–19]. This article provides a lived experience-led overview of IH under-recognition in ED treatment, with attention to epistemic injustice; shows that IH is not experienced uniformly across marginalised communities when treatment frameworks fail to account for identity, culture, socio-political context, and structural inequity [20–24]; and argues that lived experience-led involvement, co-design, and participatory research are necessary to shape definitions of harm, outcomes, and service reform. It asks whether claims about ED treatment benefits are complete when safety is inferred from effectiveness while harm is absent from supporting evidence. Under-recognition of IH and epistemic injustice IH remains under-recognised in psychological treatments for EDs because harms are poorly measured and patient testimony about IH may be dismissed when it challenges clinical or institutional authority [3, 7, 8, 12, 13]. Clinical literature reinforces this: when trials document benefits more systematically than harms, silence may be mistaken for safety. Poor harms monitoring does not show that harms did not occur [1]. Yet adverse effects in ED psychotherapy trials are inconsistently defined, monitored, and reported [9], mirroring wider psychotherapy research in which negative effects are rarely monitored [10]. IH may then be treated as unlikely or exceptional, and patient reports may be dismissed as anecdotal, exaggerated, or attributable to psychopathology [3, 7, 12, 13, 22]. Qualitative and lived experience-led research documents harmed or traumatised patients having their accounts reframed as defiance, resistance, lack of insight, manipulation, non-compliance, treatment-interfering behaviour, or psychopathology [7, 8, 12, 13, 25, 26]. Sandsten et al. [8] illustrate this clinically: participants described not feeling seen or heard, and reported that questioning care or expressing frustration could lead to being labelled “manipulative” or “difficult.” Downs [7] similarly describes “a profound resistance” to the possibility that treatment caused harm, with professional defensiveness replacing listening and compassion. Downs [13] conceptualises this as an “ignorance culture,” in which alarms raised by people with lived experience are ignored or minimised. These dynamics constitute epistemic injustice. The harm is not only that patients experience IH, but that their capacity to recognise and report it is undermined. In psychiatry, epistemic injustice occurs when testimony is assigned less credibility because of assumptions attached to psychiatric diagnosis or distress [27]. Fisher [28] similarly describes patient voices being silenced when systems privilege professional interpretation over lived experience testimony. In ED care, naming harm can itself be reinterpreted as illness, resistance, or lack of insight. Downs [12] therefore argues that trauma frameworks for EDs must include epistemic injury and IH within services. This aligns with broader qualitative psychotherapy literature: Curran et al. [29] identified silencing, devaluation, blame, and misuse of power as adverse therapy processes through which psychological treatment can become harmful. This does not require assuming clinician intent. Harm can occur unintentionally in healthcare, including psychological treatment. The ethical question is how clinicians and services respond once harm is named. Denial, dismissiveness, or defensiveness can compound the original harm [7, 12, 13], whereas reflection, openness, and accountability preserve the possibility of healing, repair, and learning. Causing harm and responding to reports of harm are therefore not the same ethical act: the former may be unintended; the latter involves a conscious choice to listen, self-reflect, and repair, or to preserve existing clinical interpretations by shifting the problem onto the patient. IH is not experienced uniformly: identity and context IH is not experienced uniformly. Harms produced by psychological treatment are shaped by the fit between treatment assumptions and a person’s embodied, cultural, neurodevelopmental, and socio-political realities. This matters for person-centred ED care because a treatment can be described as the ‘gold standard’ while still being experienced as invalidating, inaccessible, exclusionary, misattuned, or harmful by people whose needs were not centred in its development or evaluation [6, 20–24]. To avoid implying an exhaustive list, this section refers broadly to marginalised communities and uses selected examples from the literature to illustrate how IH may arise. Neurodivergent support needs are seldom accounted for in psychological treatments for EDs [16, 22]. Autistic people and their parents have described some ED treatment modalities, including CBT and FBT, as misattuned, with adverse effects such as feeling misunderstood, mental health decline, and treatment-related trauma [6]. Autistic women have also reported poorer ED service experiences and lower perceived benefit than non-autistic women, including in relation to CBT [30]. These findings reflect broader concerns that neurodivergent embodiment, sensory experience, communication, and testimony may be insufficiently recognised in ED care [20]. Neurodiversity-affirming ED care therefore requires adaptations to sensory, executive functioning, and communicative needs [31]. Similar concerns extend beyond ED settings, including evidence of harm in autistic adults’ experiences of mental health services and adults with ADHD describing CBT as unhelpful, overwhelming, or harmful [32, 33]. Gender is another context in which IH can occur when treatment is not affirming. For trans and gender diverse (TGD) people, harm may arise when ED treatment fails to recognise gender dysphoria, embodiment, minority stress, transition-related goals, and the relationship between gender and eating distress. Studies have identified negative ED treatment experiences, misgendering, stereotypes, lack of provider training, discordance between traditional ED treatment and gender-affirming care, and programmes designed around thin, white, cisgender women [34–36]. IH must also be understood in relation to culture, racism, colonisation, and structural inequity. Harm can arise when Western, Eurocentric, clinical models are treated as culturally neutral and universally transferable. Cobbaert [22] argues that CBT in EDs should not be described as universally evidence-based, because evidence claims are limited by the populations, outcomes, and service contexts represented in the underlying trials. This is important where dominant ED evidence bases often reflect white and Western samples, while diagnostic frameworks and outcome measures may privilege Eurocentric assumptions over cultural, colonial, and socio-political drivers of distress [22]. Mainstream psycho-behavioural therapies may also fail to address systemic oppression and normative differences between cultural traditions [37]. For example, Māori experiences of EDs remain under-researched, with stereotypes, discrimination, and culturally incongruent treatment creating risks of IH and barriers to recovery [24]. Similarly, there are no clear policy recommendations for ED prevention or management specific to First Nations Peoples in Australia, despite evidence of unmet need [23]. Toward person-centred accountability Person-centred ED care requires systems that actively look for IH, treat patient-reported IH as valid and worthy of investigation, and respond when it is reported. Monitoring must move beyond conventional, decontextualised indicators such as symptom and weight change. These outcomes matter, but they cannot capture the psychological, relational, epistemic, identity-based, and trauma-related harms described in qualitative and lived experience-led research [3, 6–8, 12, 13]. IH monitoring should ask directly about treatment-related trauma, coercion, loss of trust, relational harm, identity invalidation, disengagement, and future avoidance of care. IH should therefore be defined prospectively, assessed through open-ended questions, supplemented by qualitative methods, and reported transparently [38]. Patient-reported outcomes are essential because clinician-rated measures can miss domains that matter most to the person receiving care [39, 40]. However, measurement alone is not accountability. Asking about IH without acknowledgement, validation, listening, humility, learning, and improvement risks extracting testimony without meaningful response. When IH is reported, ED services must ask what happened, what the treatment or service contributed, what repair is needed, and what must change to prevent repetition. Repair-oriented practice requires not merely recording IH, but acknowledging it, validating the person’s experience, examining the service’s role, offering meaningful repair, and using what has been learned to change practice [7, 13]. Lived experience involvement and co-design are central because IH is both a problem of missing data and a problem of epistemic justice [20, 27, 28]. Patient-reported outcomes, qualitative interviews, and feedback forms can document experience, but they do not necessarily change who defines the problem, selects outcomes, decides what counts as evidence, interprets findings, or designs the response. Co-design addresses this by involving people with lived experience as active partners in shaping the framework itself. In mental health research, this can improve relevance, acceptability, trust, and relational safety by valuing multiple forms of knowledge [41–44]. This is especially important in ED care, where IH is often misrecognised. Clinical frameworks may fail to anticipate, measure, or believe patient-reported IH, particularly when IH-related disengagement is interpreted through narratives of resistance, non-compliance, lack of insight, treatment-interfering behaviour, or psychopathology [3, 6–8, 12, 13]. Experiential knowledge can help identify forms of IH that clinician-defined outcomes may miss, particularly where harms are relational, epistemic, identity-based, or trauma-related [9, 41, 44]. Lived experience-led involvement must shape what counts as IH, which outcomes are selected, how feedback is collected, how disclosures are interpreted, and how services respond when harm is reported [16, 45–47]. This involvement must also be attentive to structural inequity. Marginalised communities may experience IH through different mechanisms, including neuronormativity, non-affirming care, cultural incongruence, racism, coloniality, and exclusion. Indigenous and First Nations scholarship further shows that culturally responsive ED research and policy must be led, informed by, and accountable to the communities most affected [23, 24]. Conclusion Psychological treatments for EDs can be beneficial, and recognising IH does not undermine this. Rather, it strengthens the ethical and evidentiary foundations of care by insisting that treatment benefit and treatment safety are distinct questions [1, 2]. A treatment may reduce ED symptoms while still causing harm, and symptom reduction should not be treated as evidence that IH did not occur [1, 3, 6–8]. Because IH remains under-recognised in clinical practice and poorly monitored and reported in clinical research, current evidence cannot determine whether IH is more common, less common, or as common as treatment benefit [9, 10, 17–19]. Meaningfully addressing this issue requires recognising that different communities may experience different forms of IH. Marginalised communities may encounter distinct harms when treatment frameworks are neuronormative, not gender-affirming, culturally incongruent, or inattentive to structural inequity, including food insecurity [6, 20–24, 30–37]. Person-centred ED care must therefore ask not only whether a treatment works on average, but for whom it is experienced as safe, meaningful, accessible, and accountable [20–24]. The way forward requires prospective IH monitoring, patient-reported outcomes, open-ended and qualitative methods, accountability and repair-oriented practice, and co-designed clinical research [1, 16, 38–47]. Claims about the benefits or evidence-based status of psychological treatments for EDs remain incomplete when safety is inferred from effectiveness metrics alone, while IH remains underdefined, undermeasured, and unaddressed [1, 9, 10, 17–19]. Addressing IH in ED care means ensuring that those harmed by treatment are heard, believed, and offered opportunities for repair and service improvement [7, 12, 13, 16, 45–47].