A role for the early warning score in early identification of critical postoperative complications

Academic Article


  • Objective: We examined whether an early warning score (EWS) could predict inpatient complications in surgical patients. Background: Abnormal vitals often precede in-hospital mortality. The EWS calculated using vital signs has been developed to identify patients at risk for mortality. Methods: Inpatient general surgery procedures with National Surgical Quality Improvement Project data from 2013 to 2014 were matched with enterprise data on vital signs and neurologic status to calculate the EWS for each postoperative vital set measured on the ward. Outcomes of major complications, unplanned intensive care unit transfer, and medical emergency team activation were classified using the Clavien-Dindo system as grade I to V. Relationship with EWS and timing of complication was assessed using Kruskal-Wallis test and linear regression accounting for clustering with generalized estimating equation. Results: Among 552 patients admitted to the ward postsurgery, 68 (12.3%) developed at least one grade I to III complication and 37 (6.7%) developed a grade IV/V complication. The mean maximum EWS was significantly higher preceding grade IV/V complications (10.1) compared with grade I to III complications (6.4) or across the hospital stay in patients without complications (5.4; P < 0.01). EWS significantly increased in the 3 days preceding grade IV/V complications (P < 0.001) and declined in patients without complications in the 3 days before discharge (P<0.001). A threshold EWS of 8 predicted occurrence of grade IV/V complications with 81% sensitivity and 84% specificity. Conclusions: Critical postoperative complications can be preceded by rising EWS. Interventional studies are needed to evaluate whether EWS can reduce the severity of postoperative complications and mortality for surgical patients through early identification and intervention.
  • Published In

  • Annals of Surgery  Journal
  • Digital Object Identifier (doi)

    Author List

  • Hollis RH; Graham LA; Lazenby JP; Brown DM; Taylor BB; Heslin MJ; Rue LW; Hawn MT
  • Start Page

  • 918
  • End Page

  • 923
  • Volume

  • 263
  • Issue

  • 5