What can a broader perspective on NHS data teach us about patient safety?

The Organization for Economic Cooperation and Development estimates that one patient in 10 is harmed while receiving care in hospital, with 50 per cent of these cases being avoidable. Meanwhile, the Royal College of Emergency Medicine has found long accident and emergency waits have contributed to more than 250 excess deaths each week in the UK.

As well as the incredible human costs, unsafe care also has a financial impact: an estimated 15 per cent of total hospital activity and expenditure in OECD countries is attributable to dealing with adverse events.

Of course, almost no healthcare professional would willingly administer unsafe care. The causes behind safety incidents are often systemic: inadequate training, insufficient resources at a given moment, poorly designed processes, and poor communication.

Managing this problem, therefore, means measuring the problem. And this requires data.

On the surface, there is no shortage of data about safety within the NHS. According to the NHS Learning from Patient Safety Events service, 2.5 million patient safety events are recorded each year. Organisations are compelled to record incidents, and robust digital systems exist to manage the data sets.

These incident reports shouldn’t just sit idle on servers for eternity. LFPSE encourages its member organisations to actively learn from them and improve care and outcomes. A consistent national system for patient safety event data should make it easier for staff to record incidents and provide better quality, more accessible centralised data to support decision-making.

However, incidents represent just one of many data sets across the NHS. And we know that many factors influence patient safety. Is there an opportunity to look more broadly at what is happening in a healthcare organisation and identify areas to improve safety outside of the incidents themselves, which have all happened in the past? In short, yes.

Alison Leary, professor of workforce modelling at London South Bank University believes the NHS could be taking a lead from other “high reliability” industries such as aviation, oil and gas, and nuclear power. In these sectors, failures have catastrophic results, and each has developed practices to mitigate risks and ensure safe operations for workers, customers, and the public. Key to this is generating insight from across the business, looking at diverse data sets to predict and prevent safety incidents.

Understanding things like staff and patient experience, workforce data, incident reporting data, and integrating all of those things together can give a lot more insight. And doing it in real time would allow you to see the signals as they arise at the minute.

Professor Leary estimates that most healthcare organisations she has worked with run hundreds of applications. She says, to safety incident data, “you could easily add in things like prescribing data. You could add in the [electronic patient record]. You could add in the administrative data sets.”

The potential here is a shift from reactive patching of safety gaps to proactive prevention of harm. Instead of looking at previous incidents, establishing common faults, and correcting them, a broader view of data would allow the NHS to spot organisational behaviours that lead to safety incidents and act before they happen.

“Understanding things like staff and patient experience, workforce data, incident reporting data, and integrating all of those things together can give a lot more insight. And doing it in real time would allow you to see the signals as they arise at the minute,” says Professor Leary.

This real-time insight provides high-reliability organisations with vital sensitivity to operations – essentially oversight of how the organisation is functioning and clear red flags when something isn’t right.

Workforce data — something the NHS has in abundance and almost always in a digital form — is an obvious candidate for insight. If a ward doesn’t have enough staff rostered for a certain shift, this directly affects that ward’s ability to provide safe care. Professor Leary’s work has highlighted how important staff proficiency is to safety, too.

She also points to several other factors that indicate low safety performance – many of which are cultural. These include higher levels of absence and sickness, high turnover, and low levels of incident reporting – in particular low-harm or near-miss reports. Profanity in incident reports and examples of staff mutually reporting each other can also be red flags.

“These are quite often dismissed as just petty behaviour,” says Professor Leary. “In other safety-critical industries, they will be taken very seriously. In healthcare, culturally, they’re not.”

Predicting and flagging safety risks and future healthcare needs is already possible with some electronic systems. Queen’s Nursing Institute chief executive Crystal Oldman says rostering systems used in community nursing can identify if a planned shift will have insufficient staff, for example. But she also points to the Benson Model – an online tool that helps community healthcare groups to improve performance. It links into NHS systems and provides reports, enabling providers to track and compare performance in key areas.

“[The Benson Model is] looking at the demographics and the epidemiology, and is able to predict not just today, tomorrow, next week, but actually to predict this time next year what you’re going to need,” says Dr Oldman.

One case study offers a window into what’s possible when healthcare organisations work collaboratively, responding to a broad evidence base, and delivering measurable outcomes that extend beyond what might be considered traditional safety metrics.

The Welsh Risk Pool is the indemnifier for all health boards, trusts and special health authorities in the country. It has access to a nationwide data platform – Datix Cymru – giving it unique insight into the causes behind claims, redress cases, and safety incidents.

From this data, it became clear maternity cases featured disproportionately in the levels of avoidable harm and in the cost of claims. This is also seen in other UK nations. In light of this, Wales introduced a training programme – Practical Obstetric Multi-Professional Training, or PROMPT for short – which had already been developed at North Bristol Trust in the year 2000 and had been shown to improve maternity outcomes in diverse settings across the globe.

Jonathan Webb, a paramedic and head of safety and learning at NHS Wales felt PROMPT had the potential to address the recurring causes of claims and redress cases.

“We were seeing women talking about disorganisation in an emergency, and staff were talking about disorganisation in an emergency and a lack of awareness,” he says. “Not a lack of clinical skill, but a lack of human factors and situational awareness, and that’s where PROMPT Wales delivers.”

Under the PROMPT Wales model, teams are trained with colleagues on the sites where they work, so they can practice with the people, equipment, and spaces that will be used in real maternity emergencies.

“It’s more than just training,” says Sarah Hookes, assistant head of safety and learning at the Welsh Risk Pool and a registered midwife. “We know that our teams are already competent with the actual clinical management of obstetric emergencies, but it’s about what happens every day in clinical practice. By training together in the clinical area, they’re building relationships whilst testing their systems and processes.”

Babies are having better outcomes in Wales. Women are having better outcomes in Wales, and staff in Wales are feeling more positive towards safety in their units.

The programme is in place across all seven health boards in Wales, and 90 per cent of maternity staff have attended a session in the past year. PROMPT Wales has proved successful in several domains since its introduction in 2019. The number of low APGAR scores — the clinical measure of a baby’s health in the first few minutes after birth — has been significantly reduced across Wales. So has the number of cases of postpartum haemorrhage.

As well as measuring clinical improvements, the team also collects feedback from staff. The aim is to understand wider cultural markers — including the perception of management, teamwork, stress recognition, and safety climate — that may be indicators of how safe a healthcare setting is overall.

“We used a validated safety attitude questionnaire tool before we implemented PROMPT Wales and after we implemented the programme,” says Mr Webb. “There’s an improvement across all domains.

“Babies are having better outcomes in Wales. Women are having better outcomes in Wales, and staff in Wales are feeling more positive towards safety in their units.”

Work is ongoing to constantly evolve the programme further, both in the training and improvements in care and in the way that data is collected and used to inform improvements. With the benefit of hindsight, Mr Webb believes that more baseline data could have been collected before 2019. He also sees the potential to integrate other data sets.

“We’re working with our colleagues in Swansea Bay to look at what all the data in these uniform systems tells us,” he says. “So, we’ve got data on staffing levels, we’ve got data on incidents, we’ve got data on complaints, we’ve got data on claims.”

Modern technology offers the opportunity to interrogate that data using artificial intelligence to uncover new patterns. “We have a unified platform that collects unified data,” says Mr Webb. “We need to use the benefits of AI and interrogate the data more, but we’ve got a unique insight in the Welsh Risk Pool, and we must leverage that for the benefit of all patients and staff.”

Ms Hookes and Mr Webb are keen to emphasise that the programme would not have been possible without collaboration between the different stakeholders involved.

“Whilst we maintain central coordination of PROMPT in Wales, it is our colleagues in the health boards, the local facilitators, who actually make it happen,” says Ms Hookes.

Mr Webb adds: “I’ve worked in the NHS for over 35 years. I’ve worked in Wales for the last eight and I’ve never known anywhere that collaborates as well as Wales.”

One of the lessons from PROMPT Wales’s success is that joined-up data can help promote the joined-up thinking required to improve patient safety on a systemic level.

Will Browne is the co-founder and chief technology officer at healthcare data consultancy Emrys Health. He thinks the goal of using non-obvious data sets to inform patient safety decisions is “perfectly realistic”. One hurdle, however, lies in the complex web of integrations between the numerous (but finite) data systems at work within the NHS. Most of these, he notes, are built by external consultancies, work that is often unknowingly duplicated across trusts.

“For a national health service, creating 30 different interoperable integrations shouldn’t be a huge issue,” he says. “The code that links systems is a fundamental part of what the NHS should have as part of its intellectual property. Because if the NHS doesn’t own it, who does?”

Other industries, especially commercial ones like retail, are much better at using these broad data sets to understand their business and their customers.

Indeed, when commercial levers are being pulled, any organisation will tend to get better at dealing with data. Mr Browne says the NHS is capable of creating and using high-quality data sets when the incentive is there.

In acute care, for example, there is good data about the patient interactions that take place, and this is used to inform national decisions. But the reason this data exists and is used is because of the financial incentive, ie, it supports commissioning decisions. Other similar data sets, in community care, for example, are of lower quality and consequently less use.

“Until we shift that entire data landscape to outcomes being as important as activity, we’re always going to be really limited in the data and the intelligence we get from data,” says Professor Leary.

Mr Browne suggests providers could standardise on collecting patient satisfaction data after every single healthcare interaction. This is feasible in technical terms but would require buy-in from the centre to make it a reality.

After every single visit to your healthcare provider, why do you not get to comment on that? You could very simply have that in the same way that you do when you visit any establishment or jump in any Uber.

“After every single visit to your healthcare provider, why do you not get to comment on that?” asks Mr Browne. “You could very simply have that in the same way that you do when you visit any establishment or jump in any Uber, there’s a five-star review rating. Why not? There’s nothing stopping us doing that, except to have the political will to put in place.”

In addition to collecting more of certain data types, the NHS still has work to do on its own data science capabilities – at least according to those who spend time close to the service.

Professor Leary notes that there are “huge amounts of potential business intelligence,” but the NHS doesn’t employ sufficient data scientists or experts in human factors, safety science, or workforce modelling to translate that into new insight that can be used to inform change programmes. Mr Browne is of a similar opinion.

“The old school view of a health analyst is someone who moves data between spreadsheets and makes tables,” says Mr Browne. “That job doesn’t exist anymore and shouldn’t exist anymore.

“You should have people writing data pipelines to connect things and automate things, and you should have people who can use those data pipelines to build models, to create insight, to understand what’s happening in a multifactorial world, and that latter group has statisticians and data scientists.”

This would help, among other things, to predict the demands that today’s and tomorrow’s population will place on the NHS. Dr Oldman believes such a forward-looking perspective is important to the near-term strategy of the NHS as a whole.

“There’s plenty of data around the workforce itself,” says Dr Oldman. “We haven’t got the data on needs and what the population needs are. So, I would like to see more demand modelling.”

The potential to learn from data on a greater scale than within one organisation — as has happened with maternity care in Wales — provides some optimism. Ms Hughes believes that safety incident data is in danger of being “trapped within individual organisations”. She would like central leadership to take a greater role in coordinating safety improvements.

“You have lots of organisations undertaking investigations and reviews into patient incidents, highlighting issues that may need not a response just within their organisation, but a response across their [integrated care system], or nationally,” she says. “Where does that go?”

Ms Hughes believes there needs to be greater coordination at a system level, both regionally and nationally, when it comes to learning from and acting on patient safety investigations. This includes the design and development of improvement programmes, including front-line practitioners, human factors and patient safety experts, with learning widely shared across the healthcare system.

“To ensure safety, any changes must be designed into the care in line with new standards, guidelines, and standard operating procedures, and this needs to be across the system,” she says.

For more than a decade, there has been a necessary and important focus on both back office and patient administration technology in healthcare.

While there are still more gains to be realised in both these areas, there is also now a realisation that the many technologies and processes that sit in the “space between” these systems have a profound impact on the daily operations of hospitals and community providers, and therefore on patients, service users, and staff alike.

It is in this space between where much of the “operational glue” of care delivery is really managed. It is here where productivity is lost, where culture and staff experience are created, and where harm occurs.

The operational glue – whether that is theatre management systems, staff deployment systems, patient transport systems or even safety systems – shapes how waiting lists and other performance targets are delivered.

At RLDatix, we believe that when these systems are truly connected, when they are owned and influenced by clinical teams across pathways of care, and when the data is liberated and presented as intelligence to the right people at the right time, these technologies have the potential to remove administrative waste, create better working lives, and improve care delivery, radically altering efficiency, productivity, and care.

RLDatix is connecting hospital and community operations through its health and care operations platform, which includes the Allocate and Datix systems, bringing workforce availability and deployment together with clinical activity, and safety and incident management. We are focusing on ensuring that in the future back office, hospital and community operations and patient administration systems connect where needed for more sustainable and safer care delivery.

Liz Jones, chief customer officer, RLDatix