Another inquiry, the same warning?
What the Thirlwall Inquiry asks of PSIRF, corporate governance and a just culture
The difficult question is not whether accountability matters. It is whether accountability, learning and system redesign are being joined together well enough to make patients safer.
The publication of the Thirlwall Inquiry report on 15 September 2026 demands a careful response. First and foremost, its subject is the profound harm experienced by babies and families, and the repeated failures of management, governance, safeguarding and escalation at the Countess of Chester Hospital. For those of us who have not lived through such harm, its scale and consequences are almost unimaginable. That truth must remain at the centre of the response. It makes getting the system right overall, not merely reacting after catastrophe, even more pressing. The report makes strong, practical recommendations, including Board-level monitoring of deaths, predetermined escalation routes for concerning patterns, strengthened safeguarding arrangements, more effective inspection, and a barring system for NHS managers moving towards statutory regulation. These recommendations deserve serious attention and timely action.[1][2]
There is, however, a wider question that the NHS must be willing to hold alongside that necessary accountability: are repeated inquiries uncovering an endless supply of uniquely “bad” managers, or are they also revealing persistent conditions in which warning signs are discounted, uncomfortable information is softened, reputation becomes defensive, and action is delayed?
That is not an argument against the report, regulation or consequences for misconduct. It is an argument for making sure that the response does not become narrower than the problem.
Accountability is necessary, but punishment is not a safety system
The Thirlwall Inquiry is explicit that deliberate harm, negligence, recklessness and serious misconduct cannot be explained away as “the system”. Its discussion of NHS culture also challenges the limitations of a simplistic “no blame” approach. That distinction matters because a just culture is not, and has never been, a no-blame culture, despite how it is sometimes reported. Whether we use James Reason’s framing, with its fair boundary between acceptable and unacceptable behaviour, or Sidney Dekker’s later restorative framing, accountability remains present and important. The difference is in what accountability is for and how it is exercised: not as a reflex search for someone to punish, but as a fair process that learns, responds to conduct proportionately, repairs harm and makes future care safer.[3][18][19]
Dekker’s work is particularly useful here. He distinguishes backward-looking, retributive accountability (who crossed a line and what sanction should follow) from forward-looking, restorative accountability: who was harmed, what they now need, whose obligation it is to respond, and what must change. Restorative accountability does not let people “off the hook”. It asks them, and the organisation around them, to give an account, meet obligations, rebuild trust and contribute to safer systems. Reason and Dekker approach the boundary differently, but neither removes accountability from just culture.[18][19]
Promethean HD’s resources on Just Culture in Healthcare and Beyond Blame: Human Factors, Just Culture and Governance explore this distinction in more detail, including the practical relationship between systems learning, restorative accountability and fair governance.
Some people will push boundaries, knowingly break rules or place patients at risk. These cases require proportionate investigation and action. But safety research and operational experience repeatedly suggest that they are the exception, not the default explanation for failure. Most people come to work intending to do a good job. If every adverse outcome is treated primarily as evidence of a defective individual, the organisation may remove a person while leaving intact the pressures, incentives, information gaps and governance weaknesses that shaped the outcome.
Even when the decisions under scrutiny were made by senior leaders, we still need to “get in the tunnel”. Dekker’s Field Guide to Understanding ‘Human Error’ asks investigators to reconstruct the unfolding situation from the inside: what people could see and know at the time, the goals and pressures they were balancing, the cues they noticed or discounted, and why their actions made sense locally before the outcome was known. This is the opposite of reading the story backwards with hindsight. Understanding is not exoneration. It is the disciplined work needed to explain performance and prevent recurrence.[23]
The recurring lesson: concerns are raised, but the system does not respond
Francis, Berwick, Kirkup and Ockenden differ in scope and context, but familiar themes recur: patients and families are not heard; staff struggle to speak up; data and concerns do not travel reliably to decision-makers; organisations become defensive; and declared learning is not consistently translated into sustained improvement.[4][5][6][7] The 2026 Nottingham maternity review again set out local actions, system-wide learning and immediate and essential actions after reviewing more than 2,500 family cases.[8]
The Thirlwall report goes further by reviewing three decades of inquiry recommendations. Its conclusion is stark: while some important changes have been delivered, many recommendations were not implemented, were only partly implemented, or could not be shown to have been effective. It highlights the volume, duplication and variable quality of recommendations, weak tracking, unclear ownership and the risk that improvement is lost through structural change.[9]
Healthcare does not lack recommendations. It lacks consistent evidence that recommendations have been prioritised, resourced, implemented, tested and sustained. The Government’s proposed single tracker for major maternity and neonatal recommendations is therefore welcome, but tracking activity cannot be the end point. Boards need evidence of changed conditions and better outcomes, including evidence that improvement is holding under operational pressure.[2]
PSIRF should help us resist the pull towards simple explanations
This is where the Patient Safety Incident Response Framework matters. PSIRF is not simply a different investigation template. At its best, it changes the question from “who failed?” to “how did the system make this outcome possible, and what needs to change?” Rasmussen’s work on risk management in dynamic socio-technical systems provides an important foundation. It shows how decisions and adaptations across government, regulators, organisations, management, staff and the conditions of work can interact over time. AcciMap makes those relationships visible across system levels, rather than compressing a complex event into a single root cause or a linear chain. Used alongside methods such as SEIPS, it helps investigators examine how external, organisational, technological, task, environmental and individual factors combined to shape performance.[10][24]
Rasmussen also helps explain why safety can erode without a single reckless decision. In a changing system, local actors adapt to pressures for efficiency, workload and performance. Each adjustment may appear reasonable, yet the system can migrate towards the boundary of safe operation. An AcciMap therefore does more than catalogue contributory factors: it helps boards and investigators see how policy, regulation, governance, resources, supervision and frontline work are connected, and where stronger constraints or feedback are needed.[24]
That systems orientation is needed urgently. A 2026 study of 300 patient safety investigations found that contributing factors were disproportionately focused on people and that most recommendations were weak or medium strength. The authors concluded that simple linear thinking persists despite attempts to be systems-focused.[11]
A punitive climate can make that problem worse. Fear encourages defensiveness, silence and retrospective tidying of the story. Sir Jeremy Hunt has argued that blame and litigation can obstruct candour and learning, while also stressing that a just culture is not an accountability-free culture. His distinction is useful: people should be accountable for conduct, competence, honesty and integrity, but human error itself should not automatically trigger punishment.[12]
The practical test for PSIRF is therefore not whether an organisation can name SEIPS in an investigation report. It is whether the learning response creates a sufficiently safe space to understand work as done, involves patients, families and staff meaningfully, identifies system-level contributors, and produces actions strong enough to change the conditions of work.
For readers looking for a practical introduction, Promethean HD’s What Is PSIRF? guide explains the framework’s purpose, its move away from the Serious Incident Framework and the role of proportionate, compassionate, systems-based learning. The PSIRF Systems Approach to Learning resource develops the practical use of human factors methods in learning responses.
Corporate governance: from counting harm to sensing risk
Many quality and safety reports remain dominated by lagging indicators: incidents, complaints, mortality, claims, vacancies and overdue actions. These measures matter, but they describe harm or pressure that has already materialised. A board cannot govern safely through the rear-view mirror alone.
NHS England’s 2024 Insightful Provider Board guidance encourages boards to be open, curious and problem-sensing; to triangulate signals from different sources; to test whether board reports match frontline reality; and to distinguish reassurance from assurance. It also describes a fair and just culture as one that considers wider systemic issues, understands what led to behaviours and outcomes, and balances learning with accountability.[13]
The Thirlwall recommendations on real-time data, Board-level monitoring and predetermined escalation routes reinforce that challenge. Statistical signals cannot explain why something is happening, but they can prompt timely review. The report’s formulation is especially important: healthcare needs both “the statistics and the stories”.[14]
For boards, this means developing a richer set of leading indicators. Examples might include deterioration in psychological safety, repeated workarounds, unresolved safety actions, recurring temporary staffing gaps, delays in escalation, weak attendance at safety huddles, abnormal variation between services, concerns moving between committees without closure, or a widening gap between executive assurance and patient or staff experience. None is a perfect predictor. Together, they can show where the system is becoming brittle.
The challenge is scale and complexity. A healthcare provider can resemble many different industries operating under one roof: diagnostics, pharmacy, facilities, logistics, catering, digital infrastructure, emergency response, complex clinical services and more. That complexity makes a single dashboard seductive, but insufficient. Good governance requires disciplined triangulation, operational curiosity and direct exposure to how work is actually being delivered.
Quality as the organising principle
The 2026 Quality Strategy for NHS-funded care offers an important counterweight to the risk that finance or performance eclipses safety. It proposes quality as the organising principle for NHS activity, with safety, effectiveness and experience treated as interdependent. It also calls for clearer accountability, meaningful data, transparency, supported staff and robust quality management systems.[15]
NASA’s history offers a useful caution, but the chronology matters. The phrase “faster, better, cheaper” belongs to Daniel Goldin’s tenure from 1992, not to the 1980s. The earlier Challenger inquiry nevertheless documented the same underlying danger: an accelerating flight schedule, an ineffective safety, reliability and quality assurance programme, gaps in communication, and management structures that allowed flight-safety concerns to bypass key decision-makers. Its conclusion was not that NASA had stopped caring about safety, but that the checks and balances needed to protect it had been allowed to weaken. The later “faster, better, cheaper” approach delivered genuine successes, but NASA’s own assessments and the Columbia investigation again showed how resource constraints, schedule pressure, normalised anomalies and weakened independent challenge can combine. Faster, better and cheaper may be legitimate ambitions. They cannot be allowed to become priorities that safety must work around. When safety is not designed into governance, assurance and decision-making, it can be traded away almost without anyone consciously deciding to do so.[20][21][22]
The lesson for NHS boards is not that finance is unimportant. Financial stewardship supports sustainable care. The danger arises when financial and operational metrics become the dominant language of governance, while quality is treated as a specialist report or a separate committee’s concern. If the Quality Strategy is to change practice, quality must shape resource allocation, executive attention, internal audit, risk appetite, board development and the evidence used to judge success.
So what should change?
Implement and evaluate, not simply accept. Every major recommendation should have a named owner, resources, milestones, measures of effectiveness, independent challenge and a plan for sustainability.
Balance accountability with systems learning. Make clear that a just culture is not a no-blame culture. Use fair criteria to distinguish error, adaptation and system-shaped behaviour from misconduct, recklessness and deliberate harm. Whether applying Reason’s boundary-based framing or Dekker’s restorative approach, accountability must be proportionate, transparent and directed towards learning, repair and prevention.
Strengthen PSIRF capability. Train investigators, operational leaders and boards to use systems methods well, involve those affected, and develop stronger safety actions rather than defaulting to reminders, retraining and policy revision.
Build leading indicators into governance. Combine quantitative trends with staff voice, patient experience, Freedom to Speak Up intelligence, observations of work as done, audit findings and evidence of whether actions changed risk.
Test the learning climate. Ask whether staff can describe uncertainty, mistakes and trade-offs without fear, and whether leaders respond with curiosity before judgement.
Keep quality central to corporate decisions. Require major workforce, finance, digital and service-change proposals to show their anticipated impact on safety, effectiveness and experience.
The question the Thirlwall Inquiry leaves us with
The Thirlwall Inquiry contains powerful recommendations and a clear demand for accountability. The NHS should act on them. But if the response becomes primarily a search for tighter grip, stronger sanctions and more removable “bad eggs”, we may reproduce the very conditions that make people hide uncertainty, soften concerns and protect themselves or their organisations.
The more useful question is not whether regulation or accountability should exist. It is whether they are being designed as part of a wider safety system that also creates candour, psychological safety, high-quality investigation, effective governance, strong implementation and organisational learning.
Another inquiry has identified failures to listen, escalate, learn and act. Across these inquiries, the harm to patients and families is not an abstract governance failure. For those who have not experienced it, it is almost impossible to comprehend. That should sharpen, not narrow, our response. The real test now is not how strongly the NHS condemns the past. It is whether boards, regulators and national bodies can demonstrate that the whole system has changed: that warning signals travel, concerns are heard, accountability is fair, recommendations are implemented, and improvement remains effective under pressure.
We need to get this right by design, not luck.
References
18. Reason, J. (1997), Managing the Risks of Organisational Accidents. Aldershot: Ashgate.
19. Dekker, S. (2018), Just Culture: Restoring Trust and Accountability in Your Organization, 3rd ed. Boca Raton, FL: CRC Press.
20. Johnson, S. B. (2012), “Success, Failure, and NASA Culture”, ASK Magazine, NASA.
21. Hamaker, J. W. (1999), “The Faster, Better, Cheaper Approach to Space Missions: An Engineering Management Assessment”, NASA Technical Reports Server, Document ID 19990100660.
22. Columbia Accident Investigation Board (2003), Columbia Accident Investigation Board Report, Volume 1. Washington, DC: NASA and Government Printing Office.
23. Dekker, S. (2014), The Field Guide to Understanding ‘Human Error’, 3rd ed. Farnham: Ashgate.
24. Rasmussen, J. (1997), “Risk management in a dynamic society: a modelling problem”, Safety Science, 27(2–3), 183–213. doi:10.1016/S0925-7535(97)00052-0.
Comments