Observation-based early warning scores to detect impending critical illness predict in-hospital and overall survival in patients undergoing allogeneic stem cell transplantation.

Observation-based early warning scoring systems (EWSS) have been developed to improve the outcome of critically ill patients by triggering early critical care intervention. To date, none of these scoring systems have been evaluated in cancer patients or stem cell transplant (SCT) recipients. The aim of this study was to validate 3 established EWSS (modified early warning score [MEWS], patient-at-risk score [PARS], and Leed's early warning score [LEWS]) in adult recipients of Allogeneic SCT (Allo-SCT) and to determine their usefulness at predicting survival. We retrospectively analyzed the physiologic observations during the initial admission of 43 Allo-SCT recipients. Respiratory dysfunction was the most common (40 patients, 93%) event. All 3 EWSS revealed high accuracy in predicting in-hospital survival. The cutoff level associated with a high risk of in-hospital mortality was 7. Of 8 patients with a LEWS = 7, 6 died during their initial admission, whereas no patient with a lower score died (specificity 95%, sensitivity 100%). Acute clinical deterioration during the initial admission appeared to have an adverse effect on overall survival: in-hospital survivors with a LEWS >3 during their admission had a shorter median survival than patients with LEWS < or = 3, P = .018. This is the first study to validate EWSS in Allo-SCT and demonstrate that these systems are highly predictive of in-hospital and overall survival.