���ɱ����ϲ�

First published in issue 3 of SOLVE magazine, 2021

The system draws on the vast accumulation of vital signs data – the temperature readings, blood pressure, pulse and breathing rates – that indicate the direction in which a patient’s health is travelling. The data also, crucially, includes details of what happened to the patient – if people recovered or if they succumbed to infections, cardiac failure or other poor outcomes.

To a new breed of data scientists, such data that links measurements and outcomes offers unprecedented opportunities to innovate clinical practices. Professor Jim Briggs of the ���ɱ����ϲ� is one of these scientists.

“We knew what patient outcomes were from anonymised hospital data. Statistical and artificial intelligence (AI) analysis allowed us to see which particular combinations of vital signs were most likely to lead to bad outcomes,” Professor Briggs says. “The idea was to then use those signals to distinguish the high-risk patients earlier and alert nurses and doctors.”

The resulting predictive model is called ViEWS (VitalPAC Early Warning Score) because it uses data collected by a handheld device (called VitalPAC) that is being rolled out across British hospitals so nurses can record vital signs digitally. 

ViEWS was developed by a larger collaboration that included clinicians Gary Smith and Paul Schmidt and data manager Paul Meredith, all from ���ɱ����ϲ� Hospitals, and Professor Briggs’ colleague Professor David Prytherch. Those trials showed that ViEWS saves lives – an estimated 200 lives in one year in ���ɱ����ϲ� alone. The impact is that dramatic.

Over the past 10 years, lessons learnt through the development of ViEWS have informed standards set by the Royal College of Physicians (RCP) for assessing the severity of acute illness. Those insights are now embedded in the National Early Warning Score (NEWS).

NEWS relies on measurements of seven vital signs to assess risk. The higher the risk, the higher the score. What the physicians struggled with is the same issue the ���ɱ����ϲ� team mastered: where to set the thresholds that trigger a warning?

For example, the RCP recommended a score of three in any single category (such as blood pressure) meets the threshold for a warning. But the ���ɱ����ϲ� analysis found that when other vital sign measures were mostly normal, a score of three means a patient isn’t any sicker than someone with a combined score of four or five.

“Our analysis found that aggregate scores – that take into account changes across all vital signs – are more important than any single vital sign,” Professor Briggs says.

The threshold analysis attracted keen international attention. 

“We calculated that an aggregate figure of five to trigger an alert saves lives without adding to the workload of medical staff. Basically NEWS allows workloads to be targeted towards patients that need attention the most.”

A second version of NEWS, called NEWS2, has now been published and is mandated for use in all acute hospitals and ambulance services. NEWS2 has also been adopted worldwide, including by hospitals in Europe, India and the USA.

“People keep inventing early warning systems, but our algorithms keep coming out ahead because of how we set the thresholds, incorporating lessons learnt using big data,” Professor Briggs says.

When COVID-19 hit, the ���ɱ����ϲ� team was able to quickly crunch additional data and showed that NEWS2 was also effective for managing patients. The World Health Organization now recommends the use of NEWS2 for the clinical management of suspected or confirmed COVID-19 patients. 

Aiding this international uptake was the decision to publish the ViEWS model rather than commercialise it, and the RCP also deciding to make NEWS and NEWS2 freely available.

“For me, the motivation was the opportunity to work with some amazing clinicians and to be able to really make a difference to clinical practices,” Professor Briggs says. “There are people around the country – around the world – who perhaps wouldn’t be alive today if it hadn’t been for the ���ɱ����ϲ� team’s work.”

Next on the research agenda? A project to assess how frequently nurses should do the rounds and take people’s vital signs. “This is an ongoing evolution where data scientists and clinicians can join forces in a process of innovation that is really about learning the deeper, subtler or complex lessons hidden in data that virtually every hospital holds,” Professor Briggs says. 

Explore related courses, research and stories