Let a Hundred Flowers Bloom

The Trap

In the spring of 1956, Mao Zedong invited China to criticise its rulers. "Let a hundred flowers bloom," ran the slogan he made public on 2 May, "let a hundred schools of thought contend." Intellectuals, scientists, artists and ordinary citizens were urged to say openly what was wrong with the Party and the young state. For a year, little happened; the invitation met the silence of people who had learned to be careful. Then, in the late spring of 1957, the encouragement became insistent, and the criticism came. Letters arrived in their millions. Students at Peking University put up a "Democratic Wall." For a few weeks, the hundred flowers bloomed.

On 8 June 1957, the People's Daily told them to stop. An editorial announced that enemies of the Party had abused the invitation to attack it, and the language turned botanical and lethal: among the fragrant flowers, it transpired, there had grown "poisonous weeds." The Anti-Rightist Campaign followed. Somewhere between four hundred thousand and half a million people who had answered the call were branded "rightists" and sent to labour camps, internal exile or struggle sessions; some were executed (MacFarquhar, 1974). Many had said nothing more dangerous than what they had been asked to say.

Historians still argue about what Mao intended. Jung Chang and Jon Halliday read the episode as a premeditated snare: "Mao was setting a trap," they write, "inviting people to speak out so that he could use what they said as an excuse to victimise them" (Chang and Halliday, 2005). Mao's own words in July 1957 are hard to read any other way — only by letting the poisonous weeds emerge from the ground, he said, can we easily uproot them. Others are less sure. Jonathan Spence describes a "muddled and convoluted dispute within the Party" over how to handle dissent rather than a single cold plan (Spence, 1990); on this reading the opening was genuine and the reversal was panic, dressed afterwards as design.

For what follows, the dispute need not be settled. Whether the reversal was planned from the first or arrived as fear in the costume of principle, the experience of the person who spoke was identical: invited to be candid, then punished for the candour. That structure — solicit the criticism, then turn on the critic — is neither Maoist nor even, in the end, political. It is one of the oldest moves an institution can make against the people inside it. And it is alive, in quieter clothing, in the modern health service.

The Engine

If the move is old, the question is why it recurs — why institutions that have every reason to want the truth so reliably punish the people who bring it. The temptation is to blame bad rules or weak processes. The research points somewhere less comfortable: to a recognisable kind of leadership.

Since the early 2000s, personality psychologists have studied a cluster of three traits they call the Dark Triad: Machiavellianism, narcissism and subclinical psychopathy (Paulhus and Williams, 2002). The word "subclinical" matters — these are not psychiatric diagnoses but ordinary, measurable variations in personality, found in functioning people who hold jobs, run teams and chair boards. What binds the three is a callous and manipulative orientation toward other people. Machiavellianism, named for the strategic amorality of The Prince and brought into psychology by Christie and Geis (1970), describes a calculated, long-horizon pursuit of self-interest. Narcissism, in its grandiose form, couples entitlement and a hunger for admiration with a self-image too brittle to survive contradiction. Subclinical psychopathy combines shallow affect and low empathy with impulsivity and an absence of remorse.

These traits are not confined to the margins of working life. A meta-analysis of the Dark Triad and workplace behaviour by O'Boyle and colleagues, pooling data across dozens of studies, found that counterproductive work behaviour "was associated with increases in all 3 components of the DT" (O'Boyle et al., 2012). Other research suggests the traits do not merely survive in organisations but can be rewarded by them: individuals high in Machiavellianism are disproportionately successful at acquiring authority, and the literature on "abusive supervision" — defined by Tepper (2000) as a manager's "sustained display of hostile verbal and nonverbal behaviors" — locates most workplace mistreatment not among peers but flowing downward through the hierarchy. The phenomenon has a popular name: Babiak and Hare (2006) call it the snake in the suit.

What is most useful here is that the three traits do not retaliate in the same way. The distinction Paulhus and Williams (2002) draw between the calculating Machiavellian and the impulsive psychopath, together with the finding that each trait manipulates through a different channel — charm, self-presentation, or open intimidation (Jonason et al., 2012) — implies three quite different signatures of reprisal. The Machiavellian does not lose his temper; he reaches for process. Retaliation is routed through performance reviews, restructurings and procedure — slow, deniable, documented. The narcissist experiences criticism as a wound to the self, and answers it personally and vindictively. The subclinical psychopath neither plans elaborately nor takes the matter personally; he simply does not register the cost to the other person. Recognising which signature is at work is, as we shall see, half of recognising that retaliation is happening at all.

The Amplifier

A personality, however, cannot retaliate against anyone on its own. It needs an organisation around it — and organisations turn out to be remarkably good at converting a single difficult character at the top into a settled culture below.

Part of the mechanism is silence. Morrison and Milliken (2000) describe "organisational silence" as a self-reinforcing condition in which employees collectively withhold what they know about problems, having learned that speaking is unsafe or futile. Its mirror image is what Edmondson (1999) calls psychological safety: the shared belief that a team is safe for interpersonal risk-taking — for the question, the admission, the bad-news report. Where psychological safety is absent, silence is not apathy but adaptation. Padilla, Hogan and Kaiser (2007) supply the fuller picture in their "toxic triangle," in which destructive leadership becomes possible only through the conjunction of three elements: a destructive leader, susceptible followers, and a conducive environment. The leader is necessary but never sufficient; the culture does the rest.

The subtler mechanism is mimicry. In one of the foundational papers of organisational sociology, DiMaggio and Powell (1983) observed a paradox: "rational actors make their organizations increasingly similar as they try to change them." Organisations in the same field come to resemble one another, they argued, not because similarity makes them work better but because it makes them appear legitimate — through coercive pressure from regulators, through the normative influence of shared professions, and above all through mimetic imitation of whichever peers are seen as modern and safe. A decade earlier, Meyer and Rowan (1977) had supplied the crucial corollary: organisations routinely adopt the formal structures and vocabulary of reform as "myth and ceremony," and then quietly decouple that ceremonial layer from what they actually do.

This is how the language of openness propagates through a sector without the substance of it. A vocabulary of reform — "psychological safety," "freedom to speak up," "compassionate leadership," "just culture" — spreads from organisation to organisation as a badge of modernity, adopted because every reputable peer has adopted it, and then held at arm's length from daily practice. The badge certifies the institution to outsiders; the behaviour beneath need not change at all. Awards are the purest form of this armour, because most workplace accreditations assess what an organisation says about itself. Once an institution has been certified an excellent place to work, the certificate becomes a cognitive defence against anyone who says otherwise: we are an award-winning employer, so the person complaining must be the malcontent, the underperformer, the outlier. The honour does not merely fail to detect the problem. It actively shields it.

The Script

Suppose, despite the silence, someone speaks. What happens next is patterned enough to be named.

The psychologist Jennifer Freyd built her early work on betrayal trauma theory — the finding that harm inflicted within a trusted, depended-upon relationship wounds more deeply than the same harm from a stranger, precisely because it violates the trust the victim needs in order to function (Freyd, 1996). With Carly Smith she extended the idea to organisations. "Institutional betrayal," they write, "refers to wrongdoings perpetrated by an institution upon individuals dependent on that institution, including failure to prevent or respond supportively to wrongdoings" (Smith and Freyd, 2014). The harm is doubled: there is the original wrong, and then there is the betrayal by the body that was meant to protect against it — a second injury their data link to elevated anxiety, dissociation and other trauma outcomes.

The characteristic move of such institutions, when confronted, Freyd named with an acronym in 1997: DARVO — Deny, Attack, and Reverse Victim and Offender. The perpetrator, she writes, "may Deny the behavior, Attack the individual doing the confronting, and Reverse the roles of Victim and Offender such that the perpetrator assumes the victim role and turns the true victim — or the whistle blower — into an alleged offender" (Freyd, 1997). The inclusion of the whistleblower is not incidental. The pattern Freyd first identified in interpersonal abuse transposes exactly onto the institutional case: the organisation denies the substance of the disclosure, attacks the credibility and character of the person who made it, and recasts itself as the aggrieved party — the victim of a disgruntled employee, or of an agenda-driven smear. Experimental work since has shown DARVO to be both common and effective, increasing observers' tendency to blame the victim (Harsey, Zurbriggen and Freyd, 2017).

The Attack phase has a precise epistemic form. The philosopher Miranda Fricker (2007) gave it a name: testimonial injustice — the wrong done to someone "in their capacity as a knower" when prejudice causes a hearer to assign their word a "credibility deficit." Carel and Kidd (2014) showed how readily this happens in healthcare, where complainants are "vulnerable to testimonial injustice through the presumptive attribution of characteristics like cognitive unreliability and emotional instability that downgrade the credibility of their testimonies." This is the mechanism by which a complaint is answered without ever being addressed. The institution does not engage the substance of what was said; it raises a question about the person who said it. Are they quite well? Are they not rather difficult, rather angry — perhaps not entirely reliable? The disclosure is left untouched on the table while the discloser is quietly redefined as the problem. It is the Hundred Flowers in microcosm: the candour solicited, and then the candid recast as the threat.

The Pattern Is Not New

Before turning to the case at hand, it is worth establishing that none of this is hypothetical, and none of it is foreign to the National Health Service.

When Sir Robert Francis QC reported in 2013 on the catastrophe at Mid Staffordshire — where poor care contributed to the deaths of patients who should have lived — he did not describe a failure of clinical knowledge. He described a failure of culture. The trust had been governed, he found, by "an institutional culture which ascribed more weight to positive information about the service than to information capable of implying cause for concern" (Francis, 2013). Beneath it lay "a culture of fear in which staff did not feel able to report concerns" and "a culture of bullying, which prevented people from doing their jobs properly." The board had allowed "a focus on reaching national access targets, achieving financial balance and seeking foundation trust status" to displace the basic obligation of safe care. Good news was preferred to true news, and the people who carried true news were made to regret it.

The same shape appears at the Countess of Chester Hospital. There, a group of consultant paediatricians raised repeated concerns about a rise in unexplained deaths on the neonatal unit. The criminal questions arising from those deaths remain, in 2026, the subject of appeal and review, and nothing here turns on them; what is not in dispute, established by contemporaneous documents and sworn evidence to the Thirlwall Inquiry, is how the hospital treated the doctors. Its chief executive instructed the senior clinicians to write a letter of apology to the nurse they had reported (Halliday, 2023). A director told them that calling the police would be "terrible" for the hospital's reputation and would turn the unit into a "crime scene." They were warned of referral to the General Medical Council. One of them later told the inquiry that the meeting in which they were made to apologise "felt choreographed," and asked how an organisation could "start a meeting saying you followed Speak Out Safely practices and then tell seven Consultants who all have significant concerns like this that they are to apologise to the person" (Thirlwall Inquiry, 2024). The invitation to speak safely, and the demand to apologise for having done so, were issued in the same room. It is hard to imagine a cleaner reprise of the hundred flowers.

Wales has its own instance. When the maternity services at Cwm Taf Morgannwg were found, in 2019, to have failed women and babies on a scale the reviewers called one of the worst care scandals in the history of NHS Wales, it emerged that a damning internal report completed the previous year had been held back: executives, an independent review found, had treated it as a "draft" they did not want to share more widely, and there was "no evidence that the existence of the report was known to the full board" (Royal College of Obstetricians and Gynaecologists and Royal College of Midwives, 2019). Concerns about a "blame culture" and bullying had been raised internally as early as 2017. The warning existed; it was simply not allowed upward.

Three institutions, one mechanism: positive information weighted over warning, the inconvenient document held back, the people who raised the alarm recast as the difficulty. This is the script of the previous sections, performed in public. The question is whether a national body can be watched performing it not once, in retrospect, after a tragedy forces an inquiry — but continuously, in its own routine record. In Wales, it can.

The Fresh Start That Wasn't

In November 2018, the Public Accounts Committee of the National Assembly for Wales published its findings on the body then responsible for the nation's health informatics, the NHS Wales Informatics Service. Its verdict was unusually blunt. "Digital transformation requires an open culture," the Committee wrote; it "found that the culture at NWIS was the antithesis of this," and was "particularly concerned at the apparent lack of openness and transparency across the whole system" (Public Accounts Committee, 2018). The members had noticed something they could not get past: "witnesses were reluctant to be critical of progress or arrangements on the record," and written evidence from different parts of the NHS was "remarkably similar," leaving them "with the impression that we were getting a pre-prepared line." The Committee's chair described "a culture of self-censorship and denial." And the report added a warning that now reads as prophecy: this "cultural problem may be masking wider and deeper problems which we did not uncover." The same period had seen the service's data centres fail twenty-one times in six months — an outage every nine days — while the Auditor General had separately found its progress reporting "overly positive" (Wales Audit Office, 2018). (The full findings, and the evidence that the culture outlived the rebrand, are documented in Toxic Culture.)

Faced with this diagnosis, the Welsh Government did not set out to treat the culture. It replaced the name. On 1 April 2021 the NHS Wales Informatics Service was abolished and Digital Health and Care Wales (DHCW) — a new Special Health Authority with its own board, established as the direct successor to the service the Committee had just condemned — was launched as a "fresh start," backed by hundreds of millions of pounds and the vocabulary of transformation. The continuity, however, ran deeper than the discontinuity. Some eight hundred staff transferred on the first day. The leadership carried over. The founding board meeting left no published minutes at all, and the inherited estate of failing systems was presented to it not as a problem to be confronted but as a position that was, in the record's own framing, entirely positive — with no inherited difficulties acknowledged.

A culture is not a logo, and it does not change when the logo does. The Committee's 2018 finding was never shown to have been remedied; it was succeeded by a new masthead under which the same people continued the same work. A change of name is not a change of culture. What follows is what that culture, watched across five years of its own public record, has actually done.

The Mask and the Face

Read DHCW's account of itself and you encounter a model employer. Read its own minutes, surveys, audits and disclosures, and you encounter something else. The distance between the two is not noise. It is the subject. (For the same culture analysed through organisational theory, see Anatomy of Institutional Failure.)

Consider the awards. In 2022 the organisation won the British Computer Society's title of UK Best Place to Work in IT; it was shortlisted again for 2024–25, and held the Gold Corporate Health Standard and the BS76000 standard for valuing people. These are real honours, and they are also, almost without exception, self-nominated and self-described — they assess what an organisation says about itself. What the organisation did not place on the trophy shelf was the figure recorded in its own boardroom in July 2024: that sixty-five per cent of its staff reported being burnt out, or close to it. By the following summer that figure had risen by a further 3.9 percentage points, to roughly sixty-nine per cent — an increase that did not survive into the published minutes. Nor did the award sit easily beside the independent NHS all-Wales staff survey, in which the body's engagement score had fallen to seventy-six per cent. The certificate said one thing; the people inside said another; only the certificate was framed. (The awards, the burnout figures and the sickness record are documented in full in Best Place to Work?.)

Consider, next, the claim to safety. DHCW reports that 91.3 per cent of its staff would feel secure raising a concern about unethical behaviour. Set against the national picture — the whistleblowing charity Protect finds that 73 per cent of those who do raise concerns go on to report victimisation — the figure is striking, and it is contradicted from within. At least two former senior staff have brought Employment Tribunal claims connected to the raising of concerns; the merits of those claims are for the tribunal, but the existence of more than one such claim is a matter of public record, and a pattern that involves more than one person has stopped being a personnel matter. In May 2025 the organisation's own internal auditors gave only "limited assurance" — the second-lowest rating available — on the fairness of its recruitment processes; the result was recorded in a single sentence of the public minutes and discussed only in private session. A separate internal review of staff culture and wellbeing followed, and is being tracked against some fifteen corrective recommendations. (The recruitment audit and the tribunal claims are set out in Two Whistleblowers.)

Consider the doctrine the organisation chose for itself. In the same July 2024 board meeting at which the sixty-five-per-cent burnout figure was recorded, and at which an eighty-per-cent engagement score was celebrated, the board adopted a Compassionate Leadership Pledge. It did so, on the record, in about fifteen seconds and without a single question. Compassion was affirmed as policy in the same hour that the evidence of its absence was noted and passed over. Meanwhile the measurable index of distress climbed. The number of working days lost to sickness rose from 8,684 in 2021–22 to 15,846 in 2024–25 — an increase of 82 per cent in three years, against a workforce that grew by nothing like as much — and the organisation's own annual report named stress and anxiety as the single largest cause. When the board was told, in January 2024, that sickness had reached the highest level of the year, with dozens of new long-term cases in a single month, the minuted response was "Received & Discussed," and the minuted action was "None to note."

Consider what the organisation commissioned and then declined to publish. In 2024 it paid the consultancy Atos some £207,100 to survey how its stakeholders saw it. The resulting Stakeholder Review found that just 13.3 per cent of respondents spoke highly of the organisation; it was never published, and the several hundred people who had given their time to the survey were never shown its conclusions. It surfaced only through a freedom-of-information request. A separate external review — this one of the organisation's own culture and its handling of raised concerns — was commissioned from an outside adviser and completed shortly before the body was escalated into formal intervention. Its findings were never published either; the adviser's identity, the cost and the terms of reference were all withheld; and the resulting actions, the board was told, would be reviewed on a "biannual" basis. An organisation confident of its culture does not commission a review of it and then bury the answer. (The £207,100 stakeholder review is examined in the buried Atos review.)

Consider, finally, the record itself. A systematic comparison of what is said aloud in DHCW's board meetings with what is later published as the minutes of those meetings identifies 107 separate instances, across 26 of 37 meetings, in which material spoken in the room was removed before publication; on average, between eighty-five and ninety per cent of what was said does not survive into the official account. Since May 2025 the first draft of that account has been generated by an artificial-intelligence tool and then approved, in its softened form, by a human hand. The effect can be seen at its purest in a single field on a single form. In April 2026, thirteen months into formal intervention, the board approved a Microsoft enterprise agreement worth £226.9 million; in the box on the approval paper reserved for risks to be escalated, the entry read, in full, "No risks to escalate," and the corporate-risk and quality-impact assessments were both marked "N/A." The same posture survives into the organisation's formal accounts, whose governance statement begins to explain the decision to escalate its own intervention status — "the rationale for the decision to increase the escalation status was the ongoing challenges with pace and delivery on key national priorities, including:" — and then simply stops, the sentence abandoned after the colon, before declaring a few lines later that there are "no control issues or significant governance issues." By that point the body had been placed under Level 4 — Targeted Intervention, the second-highest tier of the Welsh Government's NHS Wales Oversight and Escalation Framework, one step below special measures — and the first time its leadership had been named, in the language of the escalation itself, as a reason.

Each of these is the reversal in miniature. The data are invited and then deleted; the concern is invited and then buried; the openness is certified and then withheld. The hundred flowers are encouraged to bloom precisely so that the record of them can be edited.

The Cure Withheld

Freyd did not stop at naming the disease. The antidote she proposes is what, in 2014, she named institutional courage (Freyd, 2014), and has since defined as "a commitment to seek the truth and engage in moral action, despite unpleasantness, risk, and short-term cost" and "a pledge to protect and care for those who depend on the institution" (Center for Institutional Courage, n.d.). It is not a slogan but a practice, and at its centre is something unglamorous and verifiable — the methodical collection of data on harm, and its transparent publication. An institution with courage counts the people it has failed, and lets the count be seen. The point is not merely moral: subsequent research finds that where institutional courage is present, the damage of institutional betrayal is measurably reduced (Smidt, Adams-Clark and Freyd, 2023).

By this test, the contrast is stark. In England, the National Guardian's Office — an independent body to which every trust's Freedom to Speak Up guardian reports — recorded 38,158 cases of staff speaking up in a single year, the highest annual total on record (National Guardian's Office, 2025). Wales has no equivalent figure to set beside it, because it has no equivalent body: there is no national guardian for NHS Wales, and no national speaking-up data of the English kind is published. DHCW, for its part, discloses none of its own — no whistleblowing data, no disciplinary data, no analysis of why people leave. The absence is not an oversight. As Freyd's framework makes plain, the surest way to ensure that institutional betrayal can never be measured is to decline, methodically, to measure it.

Nor is the machinery for measuring it robust. NHS Wales has no independent speak-up guardian of the English kind. Its 2023 framework, Speaking Up Safely, routes concerns instead through an internal "Executive Lead" and a non-executive "Board Champion" — both inside the organisation whose senior managers a person might need to challenge (Welsh Government, 2023). Even England's more independent model has been criticised on exactly this point: a guardian "paid by, and reporting to," the trust cannot reliably speak truth to it, and only "an outsider to whom the trust is obliged to listen" can do the job (Campbell, 2015) — a structural weakness later confirmed by a national study of how the role was actually resourced and deployed (Jones et al., 2022). A mechanism built to be heard inside the institution is, by design, one the institution can manage.

This is why a speak-up policy that is not honoured is worse than none at all. A framework like Speaking Up Safely, and the "raising concerns" procedures that sit beneath it — which promise a culture in which "no individual will suffer victimisation or detrimental treatment as a result of speaking up" (Welsh Government, 2023) — do real work the moment they are published: they tell staff that the institution wants to hear from them, and that those who come forward will be protected. Where the protection is real, the result is courage. Where it is not — where the procedure exists to be pointed to rather than followed — the policy has not failed neutrally; it has become an instrument. It issues the invitation while withholding the safety, draws the conscientious into the open on the strength of a promise the institution does not keep, and leaves them more exposed than silence would have. A speak-up policy that does not protect those who speak is not a safeguard with a flaw in it. It is the hundred flowers, printed on headed paper.

Return, at the end, to the field. The hundred flowers were invited to bloom for a reason: so that the ones that dared to could be seen, and counted, and dealt with. The darkest reading of that history is not that the criticism was punished, but that the invitation was itself an instrument of identification. An institution that solicits candour while keeping no honest record of what becomes of the candid has built the same instrument in a milder climate. It need not crush anyone. It need only ensure that the numbers are never gathered, the reviews never published, the minutes never complete.

What the Welsh Government Should Do

It need not be this way, and the choice is not the organisation's alone. The Welsh Government is no bystander here: it is the sponsoring department, the author of the escalation framework, and — since April 2026 — the authority that has placed Digital Health and Care Wales under Level 4 intervention. It holds the levers that would replace the present arrangement with something nearer to institutional courage. Five would make the difference; CareNHS sets out the fuller programme of cultural reform in Culture Reform.

First, make the count mandatory and public. Require every NHS Wales body to publish, each year, the data that openness actually consists of: concerns raised and how they were resolved, grievances and disciplinary cases, dismissals and settlement agreements, and the reasons people give for leaving. What an institution is not obliged to count, it can always claim not to know.

Second, establish genuinely independent speak-up oversight. Wales needs a national guardian function — independent of the bodies it scrutinises, and able to report in public — rather than the internal Executive Lead and non-executive Board Champion on which the 2023 framework relies. A channel that answers to the people it may need to challenge is not a safeguard; it is a filter.

Third, publish the reviews already paid for. As the body's sponsor and its escalating authority, the Government should require publication of the commissioned-but-withheld reviews — the stakeholder review, and the external review of culture and the handling of concerns — together with their authorship, cost and terms of reference. A review whose findings are buried is not assurance; it is the opposite.

Fourth, put culture inside the intervention. Level 4 already names delivery, accountability and leadership. The intervention should carry an independent cultural review with published findings and a named senior owner answerable for them — not another pledge adopted in fifteen seconds.

Fifth, insist on an honest record. A body under targeted intervention should keep minutes that record what was said — not a softened draft generated by software and approved after the fact. The record is where accountability either survives or disappears.

None of this is radical, and none of it is costly. It is simply the decision to look. The hundred flowers were invited to bloom so that the ones that dared could be identified; the opposite of that is not louder values or another pledge, but counting — and letting the count be seen. Institutional courage begins there. Institutional betrayal begins with the decision not to.

CareNHS invites a response from any organisation discussed in this article. If a response is received, we will publish it in full.

References

Babiak, P., & Hare, R. D. (2006). Snakes in Suits: When Psychopaths Go to Work. New York: HarperCollins.

Campbell, D. (2015). NHS whistleblowers report offers no prescription for ending cover-up culture. The Guardian, 11 February.

Carel, H., & Kidd, I. J. (2014). Epistemic injustice in healthcare: a philosophical analysis. Medicine, Health Care and Philosophy, 17(4), 529–540.

Center for Institutional Courage (n.d.). What is institutional courage? Available at: https://www.institutionalcourage.org/what-is-institutional-courage (accessed June 2026).

Chang, J., & Halliday, J. (2005). Mao: The Unknown Story. London: Jonathan Cape.

Christie, R., & Geis, F. L. (1970). Studies in Machiavellianism. New York: Academic Press.

DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48(2), 147–160.

Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383.

Francis, R. (2013). Report of the Mid Staffordshire NHS Foundation Trust Public Inquiry. London: The Stationery Office (HC 947).

Freyd, J. J. (1996). Betrayal Trauma: The Logic of Forgetting Childhood Abuse. Cambridge, MA: Harvard University Press.

Freyd, J. J. (1997). Violations of power, adaptive blindness, and betrayal trauma theory. Feminism & Psychology, 7(1), 22–32.

Freyd, J. J. (2014). Official campus statistics for sexual violence mislead [op-ed]. Al Jazeera America, 14 July. [Coins the term institutional courage.]

Fricker, M. (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press.

Halliday, J. (2023). Doctors were forced to apologise for raising alarm over Lucy Letby and baby deaths. The Guardian, 19 August.

Harsey, S. J., Zurbriggen, E. L., & Freyd, J. J. (2017). Perpetrator responses to victim confrontation: DARVO and victim self-blame. Journal of Aggression, Maltreatment & Trauma, 26(6), 644–663.

Jonason, P. K., Slomski, S., & Partyka, J. (2012). The Dark Triad at work: how toxic employees get their way. Personality and Individual Differences, 52(3), 449–453.

Jones, A., Maben, J., Adams, M., Mannion, R., et al. (2022). Implementation of 'Freedom to Speak Up Guardians' in NHS acute and mental health trusts in England: the FTSUG mixed-methods study. Health and Social Care Delivery Research, 10(23).

MacFarquhar, R. (1974). The Origins of the Cultural Revolution, Vol. 1: Contradictions Among the People 1956–1957. New York: Columbia University Press.

Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: formal structure as myth and ceremony. American Journal of Sociology, 83(2), 340–363.

Morrison, E. W., & Milliken, F. J. (2000). Organizational silence: a barrier to change and development in a pluralistic world. Academy of Management Review, 25(4), 706–725.

National Guardian's Office (2025). Freedom to Speak Up: National Guardian's Office Annual Report 2024–25. Laid before Parliament, 24 November 2025, p. 12.

O'Boyle, E. H., Forsyth, D. R., Banks, G. C., & McDaniel, M. A. (2012). A meta-analysis of the Dark Triad and work behavior: a social exchange perspective. Journal of Applied Psychology, 97(3), 557–579.

Padilla, A., Hogan, R., & Kaiser, R. B. (2007). The toxic triangle: destructive leaders, susceptible followers, and conducive environments. The Leadership Quarterly, 18(3), 176–194.

Paulhus, D. L., & Williams, K. M. (2002). The Dark Triad of personality: narcissism, Machiavellianism, and psychopathy. Journal of Research in Personality, 36(6), 556–563.

Public Accounts Committee (2018). Informatics Systems in NHS Wales. Cardiff: National Assembly for Wales (cr-ld11822).

Royal College of Obstetricians and Gynaecologists & Royal College of Midwives (2019). Review of Maternity Services at Cwm Taf Morgannwg University Health Board. London: RCOG.

Smidt, A. M., Adams-Clark, A. A., & Freyd, J. J. (2023). Institutional courage buffers against institutional betrayal, protects employee health, and fosters organizational commitment following workplace sexual harassment. PLOS ONE, 18(1), e0278830.

Smith, C. P., & Freyd, J. J. (2014). Institutional betrayal. American Psychologist, 69(6), 575–587.

Spence, J. D. (1990). The Search for Modern China. New York: W. W. Norton.

Tepper, B. J. (2000). Consequences of abusive supervision. Academy of Management Journal, 43(2), 178–190.

Thirlwall Inquiry (2024). Transcript of Week 10, Day 2 (19 November 2024): oral evidence of Dr Stephen Brearey, p. 170. Independent Inquiry into events at the Countess of Chester Hospital.

Wales Audit Office (2018). NHS Wales Informatics Services. Cardiff: Auditor General for Wales.

Welsh Government (2023). Speaking Up Safely: A Framework for the NHS in Wales. Cardiff: Welsh Government.