- Poster presentation
- Open Access
Evaluation of available data on physiological track and trigger warning systems
© BioMed Central Ltd 2006
- Published: 21 March 2006
- Critical Illness
- Cardiopulmonary Resuscitation
- Timely Recognition
- Senior Clinician
- Predict Patient Outcome
Physiological track and trigger warning systems (TTs) use periodic observation of vital signs (tracking) with predetermined criteria (trigger) for requesting attendance of a senior clinician or critical care outreach team (CCOT). There has been a proliferation of such systems in recent years, but with little formal evaluation. There is no clear evidence identifying an ideal system for timely recognition of critically ill patients.
To assess the ability of different TTs to predict patient outcomes within and across hospitals, in different age groups, wards and specialties. To identify the best TT for timely recognition of critical illness.
Cohort study of data from 31 acute NHS hospitals in England and Wales. Participants varied by data source; predominantly all patients seen by CCOT or all patients on selected wards. Patient outcome was a composite of death, admission to critical care, 'do not attempt resuscitation' or cardiopulmonary resuscitation. Primary assessment was by sensitivity and positive predictive value, secondary assessment by specificity and negative predictive value.
Fifteen datasets met predefined quality criteria and were included. Sensitivity and positive predictive value were low with median (quartiles) values of 43.3 (25.4, 69.2) and 36.7 (29.3, 43.8), respectively. Specificity and negative predictive value were generally acceptable, with median (quartiles) values of 89.5 (64.2, 95.7) and 94.3 (89.5, 97.0), respectively. Within hospitals there were differences in the discrimination of TTs in relation to age, ward and specialty, but these were not consistent across hospitals.
We were unable to establish the best existing TT or develop a new high-quality TT for timely recognition of critical illness due to wide variation in the datasets. Sensitivity of existing TTs is very low, meaning that a high number of patients requiring intervention are likely to be missed if clinicians rely solely on these systems for identifying deteriorating patients. The low sensitivity may be due, in part, to sudden deterioration and infrequent measurement of vital signs. It is probable that using a TT improves identification of critical illness but it should be used as an adjunct to clinical judgment. The challenge is to increase the sensitivity of TTs while maintaining acceptable specificity.