Prognostic Accuracy of Clinical Prediction Rules for Early Post-Pulmonary Embolism All-Cause Mortality: A Bivariate Meta-Analysis.

Chest

PubMedID: 25317677

Kohn CG, Mearns EL, Parker MW, Hernandez AV, Coleman CI. Prognostic Accuracy of Clinical Prediction Rules for Early Post-Pulmonary Embolism All-Cause Mortality: A Bivariate Meta-Analysis. Chest. 2015;147(4):1043-62.
Abstract
Background:Studies suggest outpatient treatment or early discharge of acute pulmonary embolism (aPE) is reasonable for those deemed to be at low-risk of early mortality. We sought to determine clinical prediction rule (CPR) accuracy for identifying aPE patients at low-risk for mortality. Methods:We performed a literature search of Medline and Embase from January 2000-March 2014, along with a manual search of references. We included studies deriving/validating a CPR for early post-aPE all-cause mortality and providing mortality data over at least the index aPE hospitalization but =90-days. A bivariate model was used to pool sensitivity and specificity estimates using a random-effects approach. Traditional random-effects meta-analysis was performed to estimate the weighted proportion of patients deemed at low-risk for early mortality and their odds ratios for death compared to high-risk patients. Results:Forty studies (52 cohort-CPR analyses) reporting on 11 CPRs were included. The highest sensitivities were observed with the Global Registry of Acute Coronary Events (GRACE)(0.99, 95%CI=0.89-1.00), Aujesky 2006 (0.97, 95%CI=0.95-0.99), simplified Pulmonary Embolism Severity Index (sPESI)(0.92, 95%CI=0.89-0.94), PESI (0.89, 95%CI=0.87-0.90) and European Society of Cardiology (ESC)(0.88, 95%CI=0.77-0.94) tools; with remaining CPR sensitivities ranging from 0.41-0.82. Of these 5 CPRs with the highest sensitivities, none had a specificity >0.48. They suggested anywhere from 22%-45% of aPE patients were at low-risk; and that low-risk patients had a 77%-97% lower odds of death compared to those at high-risk. Conclusions:Numerous CPRs for prognosticating early mortality in aPE patients are available, but not all demonstrate the high sensitivity needed to reassure clinicians.

Background
Studies suggest outpatient treatment or early discharge of acute pulmonary embolism (aPE) is reasonable for those deemed to be at low-risk of early mortality. We sought to determine clinical prediction rule (CPR) accuracy for identifying aPE patients at low-risk for mortality.

Methods
We performed a literature search of Medline and Embase from January 2000-March 2014, along with a manual search of references. We included studies deriving/validating a CPR for early post-aPE all-cause mortality and providing mortality data over at least the index aPE hospitalization but =90-days. A bivariate model was used to pool sensitivity and specificity estimates using a random-effects approach. Traditional random-effects meta-analysis was performed to estimate the weighted proportion of patients deemed at low-risk for early mortality and their odds ratios for death compared to high-risk patients.

Results
Forty studies (52 cohort-CPR analyses) reporting on 11 CPRs were included. The highest sensitivities were observed with the Global Registry of Acute Coronary Events (GRACE)(0.99, 95%CI=0.89-1.00), Aujesky 2006 (0.97, 95%CI=0.95-0.99), simplified Pulmonary Embolism Severity Index (sPESI)(0.92, 95%CI=0.89-0.94), PESI (0.89, 95%CI=0.87-0.90) and European Society of Cardiology (ESC)(0.88, 95%CI=0.77-0.94) tools; with remaining CPR sensitivities ranging from 0.41-0.82. Of these 5 CPRs with the highest sensitivities, none had a specificity >0.48. They suggested anywhere from 22%-45% of aPE patients were at low-risk; and that low-risk patients had a 77%-97% lower odds of death compared to those at high-risk.

Conclusions
Numerous CPRs for prognosticating early mortality in aPE patients are available, but not all demonstrate the high sensitivity needed to reassure clinicians.