A predictive model for survival after in-hospital cardiopulmonary arrest

Abstract

Background: In-hospital cardiopulmonary resuscitation (CPR) has seen a steady increase in the application of technology and techniques since the introduction of closed cardiac massage in 1960. Despite this progress, there has not been a demonstrated improvement in survival rates after in-hospital cardiac arrest over the last 40 years. Identification of prognostic factors associated with survival after a resuscitation attempt can help physician decisions and patients' end-of-life choices in a pre-arrest situation.

Methods: Using an Utstein-based template we analyzed 219 consecutive adult attempted resuscitations in a large urban teaching hospital over a 3-year period. The main outcome measures were survival to discharge, 1 and 3 months. Backwards stepwise logistic regression was used to select baseline variables that predict survival at discharge, 1 and 3 months.

Results: Survival rates at discharge, 1 and 3 months were 15.1, 13.3, and 11.5%. Meaningful neurological status (cerebral performance score of 1) at discharge was achieved in 61% of survivors. Independent predictors of survival were: higher body-mass index (BMI), presence of chronic renal insufficiency (CRI), respiratory arrest, ventricular tachycardia/fibrillation (VT/VF) as initial rhythm and arrest early during the hospital stay. A risk model based on these variables demonstrated a significant fit between predicted and observed survival at discharge with goodness of fit test P-value of 0.87.

Conclusions: Survival after in-hospital cardiopulmonary arrest is poor and can be estimated by using clinical variables. If validated in a large prospective trial, this score could help physicians in attempting resuscitation, patients and families in making end-of-life decisions and hospitals in resource allocation.