Imperial College London

Q&A: How digital technology can help spot sepsis


The Sepsis Big Room team at St Mary's Hospital

The Sepsis Big Room team at St Mary's Hospital

Digital alert systems and Toyota-style ‘big rooms’ are helping healthcare staff catch and treat sepsis early to improve patient outcomes.

Early diagnosis and intervention are key for treating sepsis – a rare but life-threatening response to an infection. Digital tools that alert clinical staff to review potential sepsis patients immediately have been introduced in hospitals worldwide, including at our partner Imperial College Healthcare NHS Trust (ICHT) hospitals.

Dr Kate Honeyford, a research associate at our School of Public Health, has led an analysis to determine the impact of a digital sepsis alert on patient outcomes at ICHT. For World Sepsis Day, we spoke to Kate to find out more. 

Why is it important to act fast in suspected cases of sepsis?

It can be really difficult to tell whether someone has sepsis when they arrive at A&E. The progression from an apparently mild infection to organ failure can be rapid, with patients finding themselves in intensive care within hours of seeking medical attention. It can take 24 hours for a blood test to confirm sepsis but treatment needs to start much earlier.

The good news is that rapid treatment with intravenous antibiotics is extremely effective at treating sepsis. Evidence suggests that hospitals should be aiming to administer antibiotics within one hour of diagnosis and that this will save lives.

What are digital alerts and how can they help with spotting sepsis early?

Screening scores based on routine clinical measurements, sometimes supplemented by blood tests are commonly used. The introduction of electronic health records means that these scores can be linked to digital alerts which determine the risk automatically and notify nurses and doctors that a patient may be at risk of sepsis.

A digital sepsis alerting system was introduced across ICHT hospitals (St Mary’s, Charing Cross and Hammersmith Hospitals) in March 2016.

How have you been involved in sepsis alert?

As a research associate specialising in the statistical analysis of health data, I’ve been involved in evaluating the impact of the alert on patient outcomes at ICHT. The study was a collaboration between our Faculty and ICHT with funding from the Imperial NIHR Biomedical Research Centre. I am also involved in the Sepsis Big Room.

What is a Sepsis Big Room?

Sepsis Big Room – which coincided with the introduction of the alert – is a weekly coached meeting for everyone interested in sepsis, led by Dr Anne Kinderlerer. The flat structure means everyone has a voice. There are specific problems the group is trying to solve. Every meeting starts with a patient story which maintains focus and allows a progression from ‘micro’ to ‘macro’.

The Big Rooms (Oobeya) approach was adapted from Toyota’s methodology and is led by the central flow coaching academy hosted by Sheffield Teaching Hospitals NHS Foundation Trust.

How did you find the experience of taking part in a Sepsis Big Room?

For me, sitting in on the weekly Sepsis Big room has been an amazing experience and a fast learning curve, particularly when it comes to the many acronyms used in hospitals! Listening to discussions about what the data actually means to people who care for patients and how sepsis care initiatives, including the alert, work in wards has made the work much more meaningful.

How was the sepsis alert introduced?

The alert was introduced in a phased approach, with ‘silent running’ occurring across the hospitals. This meant I was able to identify patients who alerted but clinicians didn’t know, these acted as the controls in my natural experiment. There were also alerts which were visible to clinicians, these were the intervention group.

Compared to a randomised clinical trial, in a natural experiment, the two groups are not as likely to be balanced and this means comparing the two groups may not be valid. We used propensity scores to balance the two groups, and adjusted the statistical models for all measured confounders. Our chosen patient outcomes to investigate were informed by discussions in the Big Room and national priorities.

Did the introduction of the sepsis digital alert have an impact on patient outcomes?

Yes, we found that there was a measurable increase in the number of patients receiving IV-antibiotics within one hour of the alert.

Our work is particularly interesting because we didn’t just look at the outcomes of patients with a confirmed diagnosis of sepsis, we looked at the group of patients who triggered the alert. This means our results are resistant to changes in diagnosis and recording practice which can be affected by policy change and the alert itself.

What’s next?

At the moment the alert is having a positive impact but we want to refine it to make it more specific (ie removing unnecessary alerts). For example, we are looking at individual clinical triggers in the alert and which are important in predicting patient’s need for treatment and we are also examining whether there are appropriate modifications for the alert so it is useful in the maternity and paediatric wards. We also need to know if the introduction of the alert led to unintended consequences.

This work is funded by NIHR Imperial BRC and NIHR Health Informatics Collaborative.

Image credit: Imperial College Healthcare NHS Trust.



Ellyw Evans

Ellyw Evans
Faculty of Medicine Centre

Click to expand or contract

Contact details

Show all stories by this author

Leave a comment

Your comment may be published, displaying your name as you provide it, unless you request otherwise. Your contact details will never be published.