Derivation and Internal Validation of a Mortality Prediction Tool for Initial Survivors of Pediatric In-Hospital Cardiac Arrest*. Issue 3 (March 2018)

• Record Type: Journal Article
• Authors: Holmberg, Mathias J.; Moskowitz, Ari; Raymond, Tia T.; Berg, Robert A.; Nadkarni, Vinay M.; Topjian, Alexis A.; Grossestreuer, Anne V.; Donnino, Michael W.; Andersen, Lars W.
• Is Part Of: Pediatric critical care medicine. Volume 19:Issue 3(2018)
• Publication Date: 2018-03
• Subjects: heart arrest -- mortality -- pediatrics -- prediction model -- prognosis; Pediatric intensive care -- Periodicals; Pediatric emergencies -- Periodicals; 618.05
• DOI: 10.1097/PCC.0000000000001416
• Languages: English
• ISSNs: 1529-7535

Abstract

Abstract : Objectives: To develop a clinical prediction score for predicting mortality in children following return of spontaneous circulation after in-hospital cardiac arrest. Design: Observational study using prospectively collected data. Setting: This was an analysis using data from the Get With The Guidelines-Resuscitation registry between January 2000 and December 2015. Patients: Pediatric patients (< 18 yr old) who achieved return of spontaneous circulation. Interventions: None. Measurements and Main Results: The primary outcome was in-hospital mortality. Patients were divided into a derivation (3/4) and validation (1/4) cohort. A prediction score was developed using a multivariable logistic regression model with backward selection. Patient and event characteristics for the derivation cohort ( n = 3, 893) and validation cohort ( n = 1, 297) were similar. Seventeen variables associated with the outcome remained in the final reduced model after backward elimination. Predictors of in-hospital mortality included age, illness category, pre-event characteristics, arrest location, day of the week, nonshockable pulseless rhythm, duration of chest compressions, and interventions in place at time of arrest. The C -statistic for the final score was 0.77 (95% CI, 0.75–0.78) in the derivation cohort and 0.77 (95% CI, 0.74–0.79) in the validation cohort. The expected versus observed mortality plot indicated good calibration in both the derivation and validation cohorts. The score showed a stepwise increase in mortality with an observed mortality of less than 15% for scores 0–9 and greater than 80% for scores greater than or equal to 25. The model also performed well for neurologic outcome and in sensitivity analyses for events within the past 5 years and for patients with or without a pulse at the onset of chest compressions. Conclusions: We developed and internally validated a prediction score for initial survivors of pediatric in-hospital cardiac arrest. This prediction score may be useful for prognostication following cardiac arrest, stratifying patients for research, and guiding quality improvement initiatives. Abstract : Supplemental Digital Content is available in the text.