Background: Recognizing early signs of clinical deterioration of hospitalized patients is thought to improve patient outcomes by activating more attentive care in a timely fashion. Early warning system scoress are tools used by care teams to potentially predict a patient's risk of deterioration and facilitate changes in management.
Objective: To systematically review the evidence on the predictive ability of Early Warning System (EWS) scores and the impact of EWS interventions on health and resource related outcomes.
Data Sources: MEDLINE, CINAHL, and Cochrane Library through April 2013 and hand search of reference lists.
Study Selection: Independent dual review to identify English language studies of early warning systems tested with adult patients admitted to medical or surgical wards. Non-systematic reviews, opinions, and case series were excluded.
Data Extraction: Data were extracted by 2 reviewers on the population, setting, sample size, duration, model discrimination and calibration, health outcomes, and resource utilization. Quality was assessed as applicable using a modified version of the Quality in Prognosis Studies (QUIPS) assessment tool for observational studies. For predictive ability, primary outcomes were model discrimination on 48-hour mortality, cardiac arrest, or pulmonary arrest. Outcomes for impact of EWS implementation included 30-day mortality, cardiovascular events, use of vasopressors, number of ventilator days, respiratory failure, and resource utilization.
Results: Of 13,595 citations reviewed, 17 studies of 11 unique models met criteria. All were based on some combination of vital signs and clinical evaluation. Six observational studies tested in large urban hospitals in developed countries found a strong predictive value for death (AUROC 0.88-0.93) and cardiac arrest (AUROC 0.77-0.86) within 48 hours. Eleven observational cohort studies with historical controls provided evidence on the impact of EWS implementation but were insufficient to draw firm conclusions due to methodological limiitations.
Conclusions and Relevance: Current early warning system scores perform well for predicting cardiac arrest and death within 48 hours although the impact on health outcomes and utilization of resources remains uncertain. Efforts to more rigorously assess their performance and effectiveness are needed as use becomes more widespread.