How data can shape a safer NHS

The Patient Safety Strategy published today rightly says that new digital technologies have the potential to create “transformational improvements” in keeping patients from harm. The possibilities here are growing rapidly and in interesting ways, as technological change means the NHS accumulates more and more information about care activities, resources and outcomes.

Data in the consulting room

Digitising health care will bring richer clinical data into the hands of clinicians, in real time. This provides the opportunity for data to help inform the treatment of individual patients. 

Electronic health records offer the ability to develop much more sensitive measures of risks and harms, building on what is already possible from retrospective analysis of routine data. This could also provide the opportunity for real-time surveillance and feedback to improve care. One example already in action is the use of National Early Warning Score (NEWS) for spotting patients whose condition is deteriorating. This calculates a score based on vital signs like pulse rate so that an effective response can be triggered when people take a turn for the worse.

In general practice, meanwhile, the PINCER initiative uses data to identify risks and harms from prescribing medication, flagging them up so that pharmacists can work with practices to resolve issues and stop them happening again.

Identifying gaps and disconnects in care

Health and care record systems are also being developed which join up data from different care settings, both for direct care, and for analysis.

The development of linked datasets will allow us to measure safety at the interface between services – for example, to identify gaps in care and risks of harm from poor coordination when patients get handed over from a hospital to care at home. Linked datasets which cover the local population, using GP data, also provide the opportunity to develop measures of safety for particular cohorts of patients, for example people with multiple long-term conditions.  

This is going to be increasingly important to spot risks and issues as more patients are supported across a team in general practice, rather than by an individual doctor, and there is more reliance on phone or online communication.    

Increasingly health data may also come directly from patients themselves, through wearables and apps. This could provide the opportunity to understand safety and risk in a much wider range of care settings, including at home. This is where patients spend the majority of their time, yet it is also the biggest gap in our knowledge.

However, making better use of data from new digital technology won’t happen automatically. There are three major obstacles to overcome.

Staff needed

Making sense of potentially huge quantities of clinical data requires analytical knowledge, systems knowledge and clinical knowledge. Individuals with this combination of skills are in short supply. The trained staff that do exist have been focused on performance and regulation, rather than quality improvement and service transformation. There is a need to build a better joint understanding among analysts, clinicians and others about how to use data effectively at different levels – from direct care, at ward or team level, through to whole organisation or population level.   

Methods and standards for analysis

There are currently few protocols or consistent approaches published for the analysis of many of the data sources which are becoming available from digital systems. Our current project analysing vital signs data has already identified that there is no standardised way of analysing NEWS data, despite this being rolled out nationally. Developing and sharing methods and standards for analysis of similar electronic health record data needs to be a priority.

Planning for measurement from the outset

Underlying these challenges, planning for how data will be analysed and used needs to start at the outset of digital developments. Too often thinking about how data will be managed and deployed is left until after the technology that generates it has already been rolled out, by which time opportunities to capture data in a way that will support analysis are lost. This can lead to loss of faith between frontline clinicians who want to access their data, and analytics teams who don’t have the means or capacity to extract and make use of it.  

At the Nuffield Trust, we have recently published a paper using routine data to measure which patients come to harm in hospital, and we intend to build on this. I hope the Patient Safety Strategy will encourage a drive to make this easier for those who want to conduct and use similar analysis across the NHS in future. Using the information the health service increasingly generates routinely has the potential to bring about a historic shift in our ability to keep patients safe and prevent harm, and that makes it too good an opportunity to pass up.