Mary Dixon-Woods calls for fresh thinking on patient safety at APPG event

The modern patient safety movement is now about a quarter of a century old, if its origins are traced to landmark reports such as Liam Donaldson’s An Organisation with a Memory in 2001. Some progress has been made in that time, Mary said, but not as much as many might have hoped. Avoidable harm remains pervasive, with consequences felt at every level, from individual patients and families to the wider health system.

She told the group that some of the obstacles to progress lie in ways of thinking that came to dominate patient safety early on. Some of that thinking may have been helpful when the field was emerging, she suggested, but may now be getting in the way. In some areas there has been under-learning from other industries and sectors, in others over-learning, and in some cases the wrong learning altogether. To move forward constructively, she outlined three capabilities that need to be strengthened.

Identifying and understanding the problem

The first capability, she said, is the ability to identify and understand the problems to be solved, because without that, any efforts to improve are akin to offering treatment based on misdiagnosis. While the NHS spends significant time and resource collecting data, including around 3 million patient safety incident reports each year in England, it still lacks strong systems for assessing risk and does not do enough to turn the data it has into meaningful action.

Analyses at local level, where much of this work takes place, are often weaker than they should be, Mary argued. Too often, they focus on the sharp end of care, while giving much less attention to the blunt end of systems and structures.

“Things like supply chains, procurement decisions, legal advice, and workforce wellbeing don’t get the attention they deserve” she said, “and cumulation of learning about similar incidents across organisations isn’t anything like as effective as it should be.”

More broadly, she questioned the epidemiological logic that often drives the safety incident approach, arguing that it consumes resources that could be better spent on more sophisticated forms of safety assessment that understand healthcare as a highly complex sociotechnical system.

The no-blame mantra and transgressive behaviour

Mary also raised concerns about unrecognised forms of bias in how the system understands the problems it needs to solve. She suggested that the “no-blame” mantra, which emerged for understandable reasons 25 years ago, has been largely founded on a “medical error” paradigm: the idea that largely unintentional errors arise from good people working in bad systems.

An unintended consequence, she argued, is that transgressive behaviour by individuals or teams can be set apart from scrutiny within a patient safety framework. Contrary to patient safety legend, she said, it is not the case that there is no such thing as a bad apple, only a bad system.

“Some of the biggest scandals involving avoidable harm are due to transgressive behaviour and not classic medical error like giving a patient the wrong drug. This is something we’ve learned many times over, from Shipman onwards.”

The Care Act explicitly recognises the problems of abuse and neglect, she noted, but the system remains poorly equipped to deal practically with issues involving transgressive behaviour. Handling by professional regulators is not always suitable, while institutional HR processes are often poorly engineered for this purpose and especially weak in dealing with team issues.

That, she suggested, points to the need for proper HR systems in the NHS, an employment law framework that is fit for purpose, and a real commitment to organisational and workforce development.

Why recommendations are not enough

The second capability, Mary said, is developing good solutions. There is no shortage of recommendations, she told the audience, and in fact there are far too many. The result, she said, is “priority thickets,” with organisations overwhelmed and action stymied.

“One maternity leader told me she had 903 recommendations to take forward in one trust. This is not sensible, and it doesn’t work – not least because recommendations are not solutions.” 

When solutions are proposed, at whatever level in the system, their design is often suboptimal, she argued. Too often, they are simple, once-off bodges that do little to recognise complexity and interdependency. They may fail to build on previous learning, involve too little input from patients and staff, lack human factors expertise, and undergo too little simulation and testing.

Over-learning and under-learning

As one example of over-learning from other industries, Mary pointed to adoption of the hierarchy of risk controls, which can downplay the role of training as a solution. Yet since expertise is critical to safety in healthcare, training can in fact be the right solution, she said, provided it is high quality. She cited the PROMPT programme for obstetric emergencies, led by the late Tim Draycott, as an example of multiprofessional training with proven benefits. The Avoiding Brain Injury in Childbirth programme, now being rolled out across England, is based on the same principles.

An example of under-learning, she suggested, is healthcare’s default to local quality improvement methods for many different types of safety issue. While those methods have a valuable role, they also have serious limitations when solutions need to be designed and implemented at scale rather than each place having its own way of doing things.

“Having 147 different obstetric early warning systems is not only a terrible waste, but it also introduces risk. It’s something that should be designed once, and designed right, from the start – a lesson that has now been painfully learned.”

She also argued that the system does not do as well as it should in learning from places that are doing particularly well. Positive deviants, she said, can teach the NHS a great deal: knowing what good looks like, and what is being done to achieve it, can go a long way.

Why evaluation matters

The third capability is evaluation, an area where Mary said the system still shows striking deficiencies. Too often, evaluation is treated as discretionary or even frivolous, she warned, when in fact it is crucial. In a review of more than 50 major maternity safety improvement programmes in the NHS over a 13-year period, only 15 had been evaluated.

“Worryingly, most of the programmes showed significant weaknesses, while few involved patients or families, and none of them addressed equity.”

The result is improvement waste, with large amounts of time and energy invested in initiatives that do not work and ultimately do not improve safety.

Mary said the principles of a learning system model offer one way forward. That means building the underpinning infrastructures that give the NHS the capability to understand problems, develop good solutions and evaluate them properly.

That, she argued, will require large-scale data and other forms of intelligence, co-design methods, and a commitment to determining what works, what does not, and why.

Key message

Mary concluded that improving patient safety will require stronger capabilities for identifying problems, designing solutions, and evaluating impact. Without that, she suggested, the system risks continuing to generate activity and recommendations without enough learning or measurable change.