Proportionate Substantiation: Ethical Limits on Therapeutic Claims in Mental Health Care

Quick takeaways

• Two promises can both be true and still differ. “Our therapists are trained in a treatment that has been tested” and “our method works where others fail” are very different claims. The second needs far more evidence than the first.

• Evidence that a treatment works is not evidence that a clinician will help you. A study shows a therapy helps a group of people on average. It doesn’t show that a given clinic delivers it well, does better than others, or will work for you.

• Confidence is not proof. Clinicians tend to rate themselves above average, and published studies can look more positive than the full record. Patients can’t easily check any of this.

• The stronger the claim, the more evidence it needs. A clinician who can’t back up a big claim should make a smaller, honest one.

• No one can guarantee recovery. A fair promise says what is known and what isn’t, and it commits to a plan: agree on goals, review progress, and change course if the treatment isn’t helping.

• Be wary of vague numbers. A “success rate” means little without what counts as success, how many people started, and how many were followed up.

• Ranking clinicians is risky. Scorecards can push providers to avoid the hardest cases, so any reporting system has to be tested for that.

• The article doesn’t claim most professionals mislead patients. It argues that claims should be easy to check, and it proposes research to find out how common the problem really is.

Abstract

Patients choosing a mental health professional must rely on claims they cannot readily verify. Evidence that a treatment works does not show that a particular clinician delivers it well, that the clinician outperforms others, or that the treatment will help this patient. This ethical analysis argues for a principle of proportionate substantiation: the stronger and more specific a therapeutic claim, the more evidence its maker owes. The argument rests on three features of care. Patients cannot readily verify expertise, professionals invite reliance on it, and the resulting choices are consequential and costly to reverse. Research on provider self-assessment, selective reporting, therapist effects, and therapist matching shows why confidence needs external support. The article offers a four-question standard for claims, a three-part response owed when treatment is not helping, and conditions under which outcome reporting could support accountability without penalizing clinicians who treat complex cases. It answers four objections and sets out a research agenda that could confirm, narrow, or refute the diagnosis of overstatement. Uncertainty justifies modest promises. It does not justify confident promises that give way to appeals to uncertainty once outcomes disappoint.

Keywords: professional ethics; therapeutic claims; expertise; trust; mental health care; outcome reporting

Two websites

Consider two hypothetical clinic websites, each describing care for depression that has not responded to medication. The first reads:

Our clinicians are trained in cognitive behavioral therapy, which has been tested in people whose depression did not respond to antidepressants. In one primary care trial, 46% of patients who received it alongside usual care had responded at six months, compared with 22% of those who received usual care alone. We will review your progress with you after six sessions and discuss alternatives if you are not benefiting.

The second reads:

Our proprietary method succeeds where other treatments fail. Nine in ten clients recover.

Each could be literally true. A patient choosing between them, however, is not being offered the same kind of reason. The first makes a modest claim about training and about evidence for an intervention, and it commits the clinic to a response if the intervention fails for this patient. The second makes a comparative claim about the provider’s own results, using an undefined outcome among an unspecified group, and it invites the reader to predict their own recovery. Few patients can tell what kind of evidence each claim would need, and most will hear both as expertise speaking.

This article defends a principle of proportionate substantiation: the strength and specificity of a therapeutic claim determine the evidence needed to support it. The principle governs two moments, the representation of capability before care and the accountability that follows when expected benefit does not occur. It does not assert that mental health professionals routinely mislead patients, and it does not estimate how often that happens. The duty to substantiate holds whatever the prevalence of exaggeration. The case for particular reforms depends on how common and how concentrated exaggeration is, and Section 10 treats that as an empirical question the evidence could answer against the argument.

The article is an ethical analysis informed by selected research, not a systematic review. It builds on scholarship about trustworthy communication and outcome monitoring (Muir et al., 2019; O’Neill, 2004) and on the substantiation doctrine in advertising law, and it applies them to the path from public promise to referral decision to clinical follow-through. Section 2 grounds the obligation. Section 3 separates the kinds of evidence a claim can rest on, and Section 4 explains why confidence alone cannot stand in for it. Section 5 shows how uncertainty constrains promises, Section 6 sets out a standard for claims, and Section 7 describes what is owed when care is not working. Section 8 addresses outcome reporting, Section 9 answers objections, and Section 10 proposes research.

Why professional claims create special obligations

The duty to substantiate arises from the relationship a professional invites. A prospective patient is asked to rely on knowledge they cannot easily assess, to undertake an intervention whose benefit is uncertain, and to bear costs that include money, adverse effects, emotional effort, and delay in reaching alternatives. The professional controls much of the information that would justify that reliance. Where one party invites reliance and holds the evidence, that party owes an explanation of what it claims and on what basis.

Economists call services of this kind credence goods: the recipient may be unable to judge what was needed or how well it was delivered, even afterward (Dulleck & Kerschbamer, 2006). Experience does not close the gap. A patient who improves cannot tell whether treatment, the passage of time, or other events deserve the credit. A patient who does not improve cannot tell whether the treatment was poor or merely unsuccessful. Outcomes, patient experiences, and reputation carry real information, but they do not reliably identify a practitioner’s causal contribution. Competition and reputation therefore cannot be assumed to resolve the problem, because the feedback they depend on may be incomplete, selective, or difficult to interpret.

Respect for autonomy requires more than a statement that outcomes are uncertain. A therapeutic claim supplies a reason to choose one course of care over another. If the reason is unsupported, the choice may be fully voluntary and still rest on a misleading basis. A closing “results vary” does not substantiate “exceptional results.” Disclosure describes a limit, whereas substantiation shows whether the positive assertion deserves weight. A patient’s willingness to accept risk cannot supply evidence the clinician lacks.

Invited reliance adds a further obligation. Professionals speak from a position that reasonably signals relevant knowledge, so patients need not treat therapeutic assertions as ordinary salesmanship. A practitioner who benefits from that authority should take reasonable care to prevent a foreseeable, material misunderstanding about capability. The duty covers interpretations the speaker can reasonably anticipate, not every inference any reader might draw. Where a qualification would substantially change the apparent reason for choosing care, it belongs in the representation itself.

Fairness explains who should bear the work of justification. Patients already bear the consequences of treatment decisions and are poorly placed to investigate the claims used to win their trust. Requiring them to disprove an asserted advantage reverses the burden, because the person making a claim should supply its basis. This does not make the professional responsible for every poor outcome. It assigns responsibility for an avoidable informational disadvantage to the party best placed to prevent it.

The principle is not new in kind. United States advertising law has long held that an advertiser who makes an objective claim must have a reasonable basis for it, and that the amount and type of substantiation required depends on what the advertisement communicates to consumers, including by implication (Federal Trade Commission, 1984). A claim that a product is clinically proven implies controlled evidence, and a claim that customers are satisfied does not. Claims about health attract stronger demands because errors cost more. The clinical setting gives reason for greater care, not less, because the audience is distressed, the stakes are personal, and the claimant is trusted as a healer rather than discounted as a seller.

These considerations do not require every practitioner to conduct a trial or publish personal success rates. They require the representation to fit the available evidence. A clinician with sound training but no reliable comparative outcome data may accurately describe that training and the evidence for the treatment. What is unavailable is the further claim of demonstrated superiority. Where stronger evidence cannot reasonably be obtained, the claim should become narrower, and the obligation does not disappear.

The aim is well-founded reliance. O’Neill (2004) distinguishes communication that helps people judge a claim from disclosure that merely makes information available. Applied here, communication should let a patient understand the claim, question its basis, and recognize its material limits. An inaccessible evidence archive cannot correct a misleading headline, and a signed acknowledgment does not establish understanding. Professionals should offer a usable explanation and a real opportunity to ask questions, without requiring patients to become their own clinical researchers.

Three kinds of claim, three kinds of evidence

Three propositions are often allowed to run together: a treatment benefits a studied population, a particular practitioner delivers it effectively, and that practitioner will help this individual. Each needs different evidence, and support for one is not support for the next. These claims are not a simple hierarchy. Comparative superiority needs comparative data, an individual prognosis needs individualized assessment, and neither is merely a stronger version of the other. Table 1 sets out the distinctions.

Table 1. What each kind of claim asserts and what can support it

Note. Examples are illustrative.

The CoBalT trial illustrates the intervention row. Among primary care patients whose depression had not responded adequately to antidepressants, adding cognitive behavioral therapy to usual care produced a six-month response rate of 46%, against 22% for usual care alone. Response meant substantial symptom reduction, not lasting recovery (Wiles et al., 2013). That finding supports offering the treatment to a defined population. It says nothing about whether a given clinic delivers it as it was delivered in the trial, and it does not tell any patient whether they will fall among the 46% or the other 54%. The prognosis rows illustrate a consequential misuse of population evidence. A rate describes a group. It can inform a qualified statement of probability to a person, but it cannot become a promise.

Provider-level claims are not empty categories, since therapists do differ. A systematic review found therapist effects in 19 of 20 studies, with a weighted average of about 5% of outcome variance and estimates ranging from 0.2% to 29%. Therapist samples were often smaller than recommended, and few studies were designed to estimate therapist effects (Johns et al., 2019). These figures describe how much outcome variation lies between therapists. They do not give the share of any patient’s improvement caused by therapist skill, and a small share of variance can still conceal differences that matter to patients. The findings support treating provider-level differences as real and worth testing. They also show that such differences are imprecisely estimated, which is why provider claims are hard to substantiate and why a clinician’s own conviction is not evidence.

Why confidence needs outside checks

If clinicians’ confidence in their own results were reliable, outside substantiation would matter less. Three lines of evidence suggest it cannot be relied on alone.

Self-assessment

Walfish et al. (2012) found that one quarter of surveyed mental health providers rated themselves at the 90th percentile relative to peers, and none rated themselves below average. Respondents were rating themselves against professional peers generally, not against the other respondents, and a self-selected sample could contain more than its share of strong clinicians. Without independent performance measurement, the study therefore cannot identify which respondents were overconfident. It nevertheless raises a calibration concern and shows why a profession’s favorable view of its own abilities cannot substantiate claims of exceptional performance.

Selective reporting

The evidence clinicians receive can itself be distorted. Turner et al. (2008) found that published reports made 94% of antidepressant trials appear positive, compared with 51% under FDA assessments. These are trial classifications, not patient success rates. The gap shows how selective publication can make a treatment’s evidence look stronger than the fuller record. An update found improved transparency for newer antidepressants alongside continuing reporting bias (Turner et al., 2022). A credible criticism must recognize both the distortion and the improvement.

Outcome definitions

STAR*D shows why the definition of an outcome matters as much as its size. The original report used a self-report measure to summarize remission. A later reanalysis challenged the outcome measure and inclusion decisions, and the original investigators disputed the reanalysis (Pigott et al., 2023; Rush et al., 2006, 2023). The present argument does not depend on choosing a cumulative remission estimate from that debate. It depends on making analytic choices and their consequences visible whenever rates are used to shape therapeutic expectations.

These studies concern different mechanisms and settings, and none shows that most professionals mislead patients or that effective treatment is unavailable. They justify treating confidence, whether a clinician’s own or a literature’s apparent consensus, as a claim to be checked rather than as evidence. A fair assessment distinguishes particular practitioners, services, and institutions instead of turning evidence of bias into a collective accusation.

Uncertainty must constrain promises as well as defenses

Treatment uncertainty is not an exceptional fact discovered after failure. When it is known in advance, it should shape the representation that precedes care. A service cannot imply dependable recovery to attract patients and then present the unpredictability of recovery as a complete answer when they question the result. Uncertainty may explain an outcome. It does not retrospectively justify the earlier promise.

This is an obligation of consistency, not a guarantee of success. New clinical information can change an assessment. An intervention can be reasonable without working, and treatment resistance can warrant further assessment rather than a declaration of futility. The international consensus on difficult-to-treat depression emphasizes identifying remediable obstacles and pursuing outcomes that matter to the patient when sustained remission remains elusive (McAllister-Williams et al., 2020). Candor should therefore describe both the limits of an intervention and the continuing responsibilities of care.

Representations and clinical decisions should be evaluated separately. A clinician may overstate likely benefit and then provide competent care. Another may describe uncertainty accurately and then fail to act on deterioration. Neither assessment can stand in for the other. Review should ask whether the original claim was supported and whether the later response was reasonable in light of the information available at the time.

The distinction protects patients and professionals alike. Patients need not prove that treatment caused a poor outcome before questioning a misleading claim, and clinicians need not promise recovery to demonstrate competence. Accountability concerns what was represented, what supported it, what was reasonably done, and how the plan changed when benefit failed to appear. A favorable outcome does not validate an unsupported promise, and an unfavorable outcome does not invalidate reasonable care.

A standard for substantiating therapeutic claims

The proposed standard asks four questions: what is being claimed, what supports it, what the audience is likely to understand, and what responsibility follows. It develops existing obligations rather than replacing them. The American Psychological Association’s ethics code already prohibits false or deceptive statements about competence, services, and their scientific basis or degree of success (American Psychological Association, 2017).

First, identify the object of the claim. Training, use of an evidence-supported intervention, observed service outcomes, comparative superiority, and individual prognosis are distinct assertions, and the evidence offered must address the one actually made. A citation to treatment research cannot establish a provider’s exceptional results, and personal experience cannot establish that an approach succeeds where alternatives fail. Table 2 lists common claims and the support each needs. The list is not a strict hierarchy, because comparative claims and individual predictions need different kinds of evidence, not simply more of it.

Table 2. Common claims and the support each requires

Note. Wording and figures are illustrative.

Second, define material terms. A success rate needs an outcome, a denominator, an assessment method, and a time horizon. Satisfaction, symptom improvement, remission, and sustained functioning are not interchangeable. A completion-only rate should not be presented as the experience of everyone who entered care. Numerical precision without these details can mislead more effectively than a vague assurance.

Third, make qualifications prominent enough to shape the message. Whether a statement misleads depends on its reasonable interpretation in context, not on whether each isolated sentence is literally true. An evidence summary should sit alongside a strong claim, with fuller methods available where relevant. Institutions should test understanding with intended audiences, including people with limited health literacy, rather than assume that a disclaimer neutralizes the main impression.

Fourth, connect the claim to an accountable clinical process. Before treatment, patient and clinician should agree on goals, a reasonable review point, and what insufficient benefit would mean for the plan. That makes the promise assessable during care rather than leaving its meaning to be contested after the relationship has broken down.

Applied to the two websites in Section 1, The first website distinguishes training from intervention evidence, identifies a relevant trial result, and provides a review plan. Its training claim would still require verification. The second asserts comparative superiority and implies a near-guarantee, and it supplies none of what a performance or comparative claim needs: no outcome definition, no denominator, no follow-up, no comparison. The standard does not prohibit confident recommendations. It requires confidence to remain answerable to its grounds.

Responsibility when treatment is not helping

The standard for public claims must carry through into the care those claims invite. When a patient is not improving, accountability requires three things: recognition, a reasoned decision, and follow-through.

Recognition. The clinician should first discuss the gap between expected and observed benefit, including the patient’s account of functioning and burdens. A symptom score should inform that conversation and not overrule it. The review point should reflect the intervention and the clinical circumstances, and significant deterioration warrants earlier assessment.

A reasoned decision. Review should consider the treatment’s rationale, the adequacy of its delivery, tolerability, diagnostic alternatives, practical barriers, and the person’s preferences. NICE’s depression guidance offers an existing clinical example of reviewing contributing factors and the diagnosis after an inadequate response (National Institute for Health and Care Excellence, 2022). Continuing the same approach can be reasonable, but it should come with a stated rationale, a further review point, and signs that would prompt a change. Reassurance alone is not a treatment review.

Follow-through. Modification, consultation, or referral should identify who will act, how the patient will take part, and what support remains during a transition. A referral does not itself establish that alternative care can be reached. Clinicians cannot remove every service constraint, but institutions should make such constraints visible rather than relabel them as patient resistance or biological futility. The record should preserve the reasons for the decision and the agreed next step, which supports continuity and later review without equating documentation with good care.

Outcome evidence without a market for easy cases

Provider outcome information is one way to make expertise easier to assess. Constantino et al. (2021) tested matching adult outpatients to therapists on the therapists’ historical performance with problems like the patient’s. Of 288 randomized patients in six community mental health clinics in Cleveland, 218, treated by 48 therapists, entered the primary modified intention-to-treat analyses. Matched patients improved more over 16 weeks than those assigned as usual, with effect sizes between 0.50 and 0.75 across outcome measures. The result comes from one trial in one setting. It used a particular outcome instrument, and one author is affiliated with a company with a commercial interest in it. It supports further testing of informed matching, not universal therapist rankings.

Public reporting can also change which patients receive care. Dranove et al. (2003) found that cardiac surgery report cards led to both better matching of patients to providers and selection behavior, with worse outcomes for sicker patients in their analysis. That finding cannot be transferred directly to psychotherapy. It does identify a mechanism any reform must address: when a score affects reputation, improving the score can become more attractive than treating the people who most need care.

The objection changes the conditions under which reporting should be introduced. An improved score is weak evidence of improvement if the service has become less accessible to complex patients. Evaluations should therefore compare trends in outcomes with referral acceptance and case complexity, with particular attention to groups vulnerable to exclusion. Apparent gains accompanied by selective entry should trigger redesign of the reporting or incentive scheme before it expands. Risk adjustment corrects for differences among the patients observed, but it cannot recover information about people screened out before treatment. A credible evaluation should follow the whole pathway from referral to acceptance, treatment entry, follow-up, and outcome. These safeguards are proposed conditions for testing a reporting system, not an established solution to selection behavior.

Missing data pose a related problem. People who drop out of follow-up may have improved, deteriorated, or met barriers of cost and access. Reports should show how many people entered care, how many supplied follow-up information, and how conclusions change under plausible assumptions about those missing. Otherwise a service can look successful because its least successful experiences became unobservable.

Comparative claims also need adequate samples and uncertainty estimates. Because therapist-level estimates are imprecise (Johns et al., 2019), a league table built on small caseloads risks ranking noise even where true differences exist. A small difference between unstable averages should not become a ranking. Initial implementation may reasonably emphasize confidential clinical feedback and independently reviewed service-level reporting. Muir et al. (2019) situate outcome monitoring within psychotherapy ethics, and Delgadillo et al. (2018) found feedback benefits that were modest and concentrated among patients not progressing as expected. Both approaches tie measurement to a response in care and avoid producing one more promotional credential.

Objections and replies

Hope and confidence are part of treatment. Clinicians who project doubt may weaken the hope and collaboration that plausibly contribute to benefit. But the standard constrains claims about capability and outcome, not confidence about the process. A clinician can honestly say that a plan is sound, that they are committed to making it work, and that they will change course if it does not. That statement offers hope without asserting a record or a prognosis the clinician cannot support.

The standard burdens small practices and invites defensive vagueness. It does not ask for data the clinician lacks. It asks that the claim shrink to fit what is known, and the cheapest compliant claim, covering training, the evidence for the intervention, and a review plan, is available to every clinician. The costly claims, such as superior results and specific success rates, are the ones that ought to cost something to make.

Patients discount marketing, so no special rule is needed. Some do, but vulnerability plausibly rises with distress, prior treatment failure, and time pressure, and professional authority plausibly changes how a statement is received. Both are hypotheses that Section 10 can test. The proper test is the reasonable interpretation of the claim by its intended audience, which is an empirical question that Section 10 proposes to study.

Ethics codes already prohibit deception. They do, and this article builds on them. Existing codes and guidelines address deceptive claims, and some address monitoring, review, and referral. This article’s contribution is to connect those obligations across promotion, treatment review, and outcome reporting, and to scale the evidence required to the strength of the claim. A systematic comparison across codes is beyond its scope. Codes also bind particular professions, while patients meet therapeutic claims from many kinds of providers, organizations, and services.

Research that could support or challenge the argument

The ethical duty to substantiate claims does not depend on exaggeration being widespread, but the case for particular reforms does. Research should therefore separate a defensible norm from an empirical diagnosis that could prove wrong.

A preregistered audit of claims. The audit would sample public claims across independent practices, hospitals, professional organizations, and digital services, analyzing discipline and setting separately. Independent coders would classify each claim’s strength, identify the evidence offered, and judge whether it supports the asserted intervention benefit, provider performance, or comparison. Numerical claims would be assessed for outcome definitions, denominators, follow-up, and missing data. The primary result would be the proportion of substantive claims adequately supported, with uncertainty estimates, not a collection of conspicuous offenders.

The audit supports competing predictions. If stronger promotional claims frequently lack relevant support, the findings would favor targeted correction and monitoring. If most claims are well substantiated, the broad allegation of professional exaggeration would weaken. If poor substantiation is concentrated in particular business models, a narrower response would be justified. The design must permit all three conclusions.

Audience experiments. A different explanation is that accurate wording is interpreted too broadly. Experiments could vary credentials, testimonials, claim specificity, and the prominence of uncertainty statements while holding the treatment information constant. Participants should be asked to distinguish evidence for the intervention from evidence for the provider and from a prediction about themselves. If better wording corrects misunderstanding, communication design becomes the immediate priority. If understanding is already accurate, further disclaimers may add burden without benefit.

Implementation studies. These should test whether clearer claims and contextualized outcome information improve referrals, decision quality, and responses to nonresponse. They should also track waiting time, refusal of complex cases, and access disparities. More measured trust is not necessarily a failure, since reduced confidence in an unsupported promise may be an appropriate result. The outcome that matters is whether patients can place trust more accurately while keeping access to appropriate care.

Conclusion

Patients should not have to mistake professional visibility for demonstrated capability. The challenge for the mental health professions is to make expertise easier to assess, especially where strong therapeutic claims invite heavy reliance. Evidence of bias warrants scrutiny, while the prevalence and distribution of misleading claims remain open to investigation.

The practical obligation is clear. A claim should say what is known, stay proportionate to the evidence behind it, and be tied to a credible plan for the case in which improvement does not come. Uncertainty about recovery is compatible with responsible expertise. Unsupported certainty followed by retrospective qualification is not. Patients deserve to meet the limits of a professional promise before they bear its consequences.

References

American Psychological Association. (2017). Ethical principles of psychologists and code of conduct (2002, amended effective June 1, 2010, and January 1, 2017). https://www.apa.org/ethics/code

Constantino, M. J., Boswell, J. F., Coyne, A. E., Swales, T. P., & Kraus, D. R. (2021). Effect of matching therapists to patients vs assignment as usual on adult psychotherapy outcomes: A randomized clinical trial. JAMA Psychiatry, 78(9), 960–969. https://doi.org/10.1001/jamapsychiatry.2021.1221

Delgadillo, J., de Jong, K., Lucock, M., Lutz, W., Rubel, J., Gilbody, S., Ali, S., Aguirre, E., Appleton, M., Nevin, J., O’Hayon, H., Patel, U., Sainty, A., Spencer, P., & McMillan, D. (2018). Feedback-informed treatment versus usual psychological treatment for depression and anxiety: A multisite, open-label, cluster randomised controlled trial. The Lancet Psychiatry, 5(7), 564–572. https://doi.org/10.1016/S2215-0366(18)30162-7

Dranove, D., Kessler, D., McClellan, M., & Satterthwaite, M. (2003). Is more information better? The effects of “report cards” on health care providers. Journal of Political Economy, 111(3), 555–588. https://doi.org/10.1086/374180

Dulleck, U., & Kerschbamer, R. (2006). On doctors, mechanics, and computer specialists: The economics of credence goods. Journal of Economic Literature, 44(1), 5–42. https://doi.org/10.1257/002205106776162717

Federal Trade Commission. (1984). FTC policy statement regarding advertising substantiation. Appended to Thompson Medical Co., 104 F.T.C. 648, 839.

Johns, R. G., Barkham, M., Kellett, S., & Saxon, D. (2019). A systematic review of therapist effects: A critical narrative update and refinement to review. Clinical Psychology Review, 67, 78–93. https://doi.org/10.1016/j.cpr.2018.08.004

McAllister-Williams, R. H., Arango, C., Blier, P., Demyttenaere, K., Falkai, P., Gorwood, P., Hopwood, M., Javed, A., Kasper, S., Malhi, G. S., Soares, J. C., Vieta, E., Young, A. H., Papadopoulos, A., & Rush, A. J. (2020). The identification, assessment and management of difficult-to-treat depression: An international consensus statement. Journal of Affective Disorders, 267, 264–282. https://doi.org/10.1016/j.jad.2020.02.023

Muir, H. J., Coyne, A. E., Morrison, N. R., Boswell, J. F., & Constantino, M. J. (2019). Ethical implications of routine outcomes monitoring for patients, psychotherapists, and mental health care systems. Psychotherapy, 56(4), 459–469. https://doi.org/10.1037/pst0000246

National Institute for Health and Care Excellence. (2022). Depression in adults: Treatment and management (NICE guideline NG222), section 1.9. https://www.nice.org.uk/guidance/ng222/chapter/Recommendations#further-line-treatment

O’Neill, O. (2004). Accountability, trust and informed consent in medical practice and research. Clinical Medicine, 4(3), 269–276. https://doi.org/10.7861/clinmedicine.4-3-269

Pigott, H. E., Kim, T., Xu, C., Kirsch, I., & Amsterdam, J. (2023). What are the treatment remission, response and extent of improvement rates after up to four trials of antidepressant therapies in real-world depressed patients? A reanalysis of the STAR*D study’s patient-level data with fidelity to the original research protocol. BMJ Open, 13(7), Article e063095. https://doi.org/10.1136/bmjopen-2022-063095

Rush, A. J., Trivedi, M., Fava, M., Thase, M., & Wisniewski, S. (2023). The STAR*D data remain strong: Reply to Pigott et al. American Journal of Psychiatry, 180(12), 919–920. https://doi.org/10.1176/appi.ajp.20230869

Rush, A. J., Trivedi, M. H., Wisniewski, S. R., Nierenberg, A. A., Stewart, J. W., Warden, D., Niederehe, G., Thase, M. E., Lavori, P. W., Lebowitz, B. D., McGrath, P. J., Rosenbaum, J. F., Sackeim, H. A., Kupfer, D. J., Luther, J., & Fava, M. (2006). Acute and longer-term outcomes in depressed outpatients requiring one or several treatment steps: A STAR*D report. American Journal of Psychiatry, 163(11), 1905–1917. https://doi.org/10.1176/ajp.2006.163.11.1905

Turner, E. H., Cipriani, A., Furukawa, T. A., Salanti, G., & de Vries, Y. A. (2022). Selective publication of antidepressant trials and its influence on apparent efficacy: Updated comparisons and meta-analyses of newer versus older trials. PLOS Medicine, 19(1), Article e1003886. https://doi.org/10.1371/journal.pmed.1003886

Turner, E. H., Matthews, A. M., Linardatos, E., Tell, R. A., & Rosenthal, R. (2008). Selective publication of antidepressant trials and its influence on apparent efficacy. The New England Journal of Medicine, 358(3), 252–260. https://doi.org/10.1056/NEJMsa065779

Walfish, S., McAlister, B., O’Donnell, P., & Lambert, M. J. (2012). An investigation of self-assessment bias in mental health providers. Psychological Reports, 110(2), 639–644. https://doi.org/10.2466/02.07.17.PR0.110.2.639-644

Wiles, N., Thomas, L., Abel, A., Ridgway, N., Turner, N., Campbell, J., Garland, A., Hollinghurst, S., Jerrom, B., Kessler, D., Kuyken, W., Morrison, J., Turner, K., Williams, C., Peters, T., & Lewis, G. (2013). Cognitive behavioural therapy as an adjunct to pharmacotherapy for primary care based patients with treatment resistant depression: Results of the CoBalT randomised controlled trial. The Lancet, 381(9864), 375–384. https://doi.org/10.1016/S0140-6736(12)61552-9