BACKGROUND

Early warning scores (EWS) are widely used to allow early recognition of the deteriorating patient. We aimed to test their ability to predict major deterioration in medical patients.

METHODS

Two cohorts were prospectively identified who were admitted to an acute medical admissions unit and to the respiratory unit but not admitted to the intensive care unit (ICU): medical-non ICU and respiratory-non ICU groups. Two further cohorts were retrospectively identified that required ICU admission from these units (medical-ICU and respiratory-ICU groups). Discriminant analysis and receiver operating characteristic curves were used to discriminate between groups, and time relationships were analysed.

RESULTS

Heart rate (HR) and respiratory rate (RR) were significantly higher--and oxygen saturation (SaO₂) significantly lower--in the medical-ICU group as compared with the medical non-ICU group, and in the respiratory-ICU group as compared with [corrected] the respiratory-non ICU group. Discriminant functions incorporating HR, RR and SaO₂ performed at least as well as existing EWS systems in predicting ICU admission.

CONCLUSIONS

Commonly used physiological parameters and existing EWS systems are useful at identifying sick patients. The discriminant functions described here appear to have a role in this setting but require validation in future studies.