Simulation training for rare medications in the intensive care unit - A study with bivalirudin

Abstract

The purpose of this study was to assess whether simulation training can improve the clinician's ability to predict the effect of bivalirudin infusion. Six clinicians with experience using bivalirudin and six without experience (Groups Exp and NoExp) entered predictions for partial thromboplastin time while viewing a running display of clinical data obtained retrospectively from intensive care unit patients who had received bivalirudin infusion after cardiac surgery. All clinicians entered guesses for the same sequence of 30 patients. Average guessing-errors were analysed using analysis of variance and linear regression. All physicians evaluated 30 patients (813 partial thromboplastin time guesses overall) in less than two hours. Average errors in Groups Exp and NoExp decreased from 9.4 and 11.7 seconds in the first tercile, to 8.2 and 8.4 seconds in the last tercile of patients, respectively. The guessing-errors of Group NoExp were significantly higher than Group Exp in the first and second terciles, with no significant difference in the third tercile. Linear regression indicated a significantly steeper learning curve in Group NoExp than Exp. Brief simulation training using retrospective patient data improved the ability of inexperienced clinicians to predict the effect of bivalirudin as compared to experienced clinicians.