SESSION TITLE: Venous Thromboembolism Posters
SESSION TYPE: Original Investigation Poster
PRESENTED ON: Wednesday, October 28, 2015 at 01:30 PM - 02:30 PM
PURPOSE: Several validated clinical prediction rules (CPRs) exist to prognosticate the risk of early mortality following pulmonary embolism (PE). We sought to compare the performance of the Pulmonary Embolism Severity Index (PESI), simplified PESI (sPESI) and Aujesky 2006 tools for predicting in-hospital mortality.
METHODS: We retrospectively collected data on 236 patients hospitalized for objectively-confirmed acute PE from a single institution between 2013 and 2014. We estimated proportions of patients classified as low- and higher-risk of early mortality according to the PESI, sPESI and Aujesky 2006 CPRs and assessed in-hospital mortality for each risk group. To measure each models’ accuracy for predicting in-hospital mortality, we calculated sensitivity, specificity and predictive values for low- and higher-risk patients. To evaluate prognostic performance, we calculated the area under the receiver operating characteristic (ROC) curve for each tool.
RESULTS: Overall in-hospital mortality was 2.1% (5/236). PESI classified 47.5% of patients as low-risk, compared to 29.8% and 25.0% for sPESI and Aujesky 2006, respectively. In low-risk patients in-hospital mortality was estimated to be 0.9% (1/113) for PESI, 0% (0/70) using sPESI and 0% (0/59) based upon Aujesky 2006. Sensitivity was 80% (95% confidence interval (CI=30%-99%) for PESI and 100% (95%CI=46-100%) for both sPESI and Aujesky 2006; while specificity was 48% (95%CI=42%-55%) for PESI, 30% (95%CI=25%-37%) for sPESI and 26% (95%CI=20%-32%) for Aujesky 2006. Positive predictive values for PESI, sPESI and Aujesky 2006 were 3.3%, 3.0% and 2.8%; while negative predictive values were 99.1% for PESI and 100% for the other two CPRs. Area under the ROC curves suggested good prognostic performance for each CPR: PESI=0.79 (95%CI=0.62-0.96); sPESI=0.86 (95%CI=0.78-0.95); Aujesky 2006=0.84 (95%CI=0.70-0.97).
CONCLUSIONS: The results of our external validation study of these three CPRs for predicting in-hospital mortality in PE patients were consistent with previous studies, including that of a recently published meta-analysis of available PE CPRs.
CLINICAL IMPLICATIONS: These three CPRs may be valuable tools to predict early mortality following a PE.
DISCLOSURE: The following authors have nothing to disclose: Whitney Saulsberry, Christine Kohn, Craig Coleman, Diana Sobieraj
No Product/Research Disclosure Information