What is CARSs?

The quality and safety of care, especially in hospitals, remains a priority for the NHS. Studies report that about 1 in 10 patients admitted to hospital suffers some form of harm attributable to poor quality of care. For example, about 5% of deaths in English hospitals are preventable if care was optimised. Another example is avoidable deaths from sepsis – an immediate life-threatening condition which can be difficult to diagnose but is treatable with early antibiotic medication. It is estimated that 10,000 sepsis related deaths per year may be preventable with better quality of care.

Care typically occurs in busy NHS hospital settings where staff are working in complex systems under time, resource (human and non-human) and information constraints. In such challenging environments, human factors research suggests that a key underlying latent factor to ensuring high quality care relates to the situational awareness and mental workload of staff which includes the (in)ability to integrate disparate pieces of information and determine what is important.

A major strategy to address information chaos is the use of electronic patient records with automated tools designed to support the clinical decision-making process by reducing mental workload and raising situational awareness.

Following a two-year study, we have developed four validated automated risk scores which we called as Computer Aided Risk Scoring Systems (CARSS).

• Computer aided risk score for predicting mortality using NEWS2 data only (CARM_N)

CARM_N is a mortality risk calculator developed using age gender and physiological variable including those that make up the NEWS2 score. Under current guidelines, National Early Warning Score (NEWS2) is not used to provide an explicit estimate of the patient’s risk of death. The CARM_N score is a computer-aided NEWS model predicting the mortality risk (i.e. 0.2 = 20%, 0.8 = 80% during this hospital admission) for emergency medical admissions, so this limits the extension of our findings to non-medical and elective admissions. The CARM_N model is externally validated and are automated within the electronic record (CPD). NEWS2 is repeatedly updated for each patient according to our hospital observation protocols, and changes in NEWS over time are reflected in changes in the CARM_N mortality risk with each update.

• Computer aided risk score for predicting mortality using NEWS2 & Blood test results data (CARM_NB)

CARM_NB is a mortality risk calculator similar to CARM_N but developed using commonly requested blood results in addition to age, gender and physiological variable including those that make up the NEWS2 score. Under current guidelines, National Early Warning Score (NEWS2) is not used to provide an explicit estimate of the patient’s risk of death. The CARM_NB score is a computer-aided model predicting the mortality risk (i.e. 0.2 = 20%, 0.8 = 80% during this hospital admission) for emergency medical admissions, so this limits the extension of our findings to non-medical and elective admissions. The CARM_NB model is externally validated and are automated within the electronic record (CPD). NEWS2 is repeatedly updated for each patient according to our hospital observation protocols and blood tests may be repeated if clinically indicated and changes in NEWS and blood results over time are reflected in changes in the CARM_NB mortality risk with each update.

• Computer aided risk score for predicting sepsis using NEWS data only (CARS_N)

CARS_N is a sepsis risk calculator developed using age gender and physiological variable including those that make up the NEWS2 score. Under current guidelines, National Early Warning Score (NEWS) is not used to provide an explicit estimate of the patient’s risk of sepsis. The CARS_N score is a computer-aided NEWS model predicting the sepsis risk (i.e. 0.2 = 20%, 0.8 = 80% during this hospital admission) for emergency medical admissions, so this limits the extension of our findings to non-medical and elective admissions. The CARS_N model is externally validated and is automated within the electronic record (CPD). NEWS is repeatedly updated for each patient according to our hospital observation protocols, and changes in NEWS over time are reflected in changes in the CARS_N sepsis risk with each update.

• Computer-aided risk score for predicting sepsis using NEWS & Blood test results data (CARS_NB)

CARS_NB is a sepsis risk calculator similar to CARS_N but developed using commonly requested blood results in addition to age, gender and physiological variable including those that make up the NEWS score. Under current guidelines, National Early Warning Score (NEWS2) is not used to provide an explicit estimate of the patient’s risk of sepsis. The CARS_NB score is a computer-aided model predicting the sepsis risk (i.e. 0.2 = 20%, 0.8 = 80% during this hospital admission) for emergency medical admissions, so this limits the extension of our findings to non-medical and elective admissions. The CARS_NB model is externally validated and are automated within the electronic record (CPD). NEWS is repeatedly updated for each patient according to our hospital observation protocols and blood tests may be repeated if clinically indicated and changes in NEWS and blood results over time are reflected in changes in the CARS_NB sepsis risk with each update.

These four scores provide staff with real-time estimates of the patient’s risk of death and sepsis during their stay in hospital.

Problem 

• Decision making is complex
• Too many data items to consider
• Too many patients to consider
• Not enough time and staff

Solution 

• CARSS is designed to help Doctors to identify those patients who are most at risk without any extra burden on the Doctor
• Automated risk scoring systems
• Requires no additional data
• Uses clinical variables
• Evidence based
• Subject to ongoing research and development

Our automated risk scores rely on two routinely collected clinical data sets – the patients vital signs data as defined by the National Early Warning Score (NEWS) and routine blood test results – and in most instances will be available within 60-minutes of unplanned admission to a ward and will be automatically updated in real-time as soon as the NEWS or blood results are updated.

NEWS is based on a range of vital signs (e.g. respiration (breathing) rate, blood pressure, heart rate, temperature) and the biomedical data is obtained from routine blood tests (e.g. albumin levels (indicate how the liver is coping), haemoglobin (cells that carry oxygen around the body),  white cell count (which may indicate infection).

Our work has shown that staff view our automated scores favourably because the scores (a) are automated, (b) require no additional data collection and (c) combine 16 clinical variables whilst aiming to support the clinical decision-making process with real time estimates of the patient’s risk of death and sepsis.