Objective: Although early warning scores were originally derived as bedside tools for alerting the medical staff, they may serve as decision rules for the admission of medical patients. We conducted this study to investigate the ability of the Modified Early Warning Score (MEWS) to identify a subset of patients at risk of deterioration, who might benefit from an increased level of attention.
Design: Prospective, single centre, cohort study.
Setting: A 64-bedded medical ward in a public, non-teaching Hospital in Italy.
Patients: All patients consecutively admitted from 15th November 2005 to 9th June 2006.
Interventions: On admission, the attending physician measured five physiological parameters (systolic blood pressure, pulse rate, respiratory rate, body temperature and level of consciousness) and calculated the MEWS. The main outcome measures were in-hospital mortality and a composite of mortality and transfer to a higher level of care. A secondary end-point was the length of stay for discharged patients.
Measurements and results: In all, 1107 patients were admitted; 621 (56.1%) were women and 486 were men. Patients of female gender were also older (mean age 80.6 years) than men (mean age 77.1; p < 0.05). Of 1107, 995 patients (89.9%) were older than 64 years. A total of 966 patients were discharged, 102 deceased and 39 were transferred. In comparison with the lowest score, the risk of death was incremental among all the MEWS categories, as well as the risk of the combined outcome of death and transfer, and highly significant (risk of death, χ2 for trend 136.307; risk of death or transfer, χ2 for trend 105.762; p < 0.00001 for both). Patients with MEWS ≤ 4 were discharged after a mean stay of 8.3 days, and alive patients with MEWS of five or more were discharged after a mean stay of 9.4 days (p = ns). A patient with a MEWS of zero at admission has a very low probability to die or to be transferred because of clinical instability (OR 0.14, 95% CI: 0.08–0.24).
Conclusions: We have confirmed that the MEWS, even when calculated once on admission, is a simple but highly useful tool to predict a worse in-hospital outcome.