Premium
The validation and reproducibility of the pulmonary embolism severity index
Author(s) -
CHAN C. M.,
WOODS C.,
SHORR A. F.
Publication year - 2010
Publication title -
journal of thrombosis and haemostasis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.947
H-Index - 178
eISSN - 1538-7836
pISSN - 1538-7933
DOI - 10.1111/j.1538-7836.2010.03888.x
Subject(s) - medicine , pulmonary embolism , intraclass correlation , severity of illness , cohen's kappa , mortality rate , retrospective cohort study , risk assessment , cohort study , clinical psychology , computer security , machine learning , computer science , psychometrics
Summary. Background : Rapid, accurate risk stratification is paramount in managing patients with acute pulmonary embolism (PE). The PE Severity Index (PESI) is a simple tool that risk stratifies patients with acute PE. Objectives : We sought to validate the PESI as a predictor of short‐ and intermediate‐term mortality and to determine the inter‐rater variability. Patients/Methods : We retrospectively identified all patients with acute PE between October 2007 and February 2009. Two clinicians reviewed charts and independently scored PESI blinded to each other and to patient outcomes. Thirty‐ and 90‐day mortality served as study endpoints and vital status was assessed via the Social Security Death Index. To facilitate analyses, raw PESI score was converted into risk class groups (I–V) and further dichotomized into low risk (I–II) vs. high risk (III–V) groups. Intraclass correlation and the kappa statistic were used to determine inter‐rater variability. Results : The cohort included 302 subjects (mean age, 59.7 ± 17.2 years; 44% male). All‐cause 30‐ and 90‐day mortalities were 3.0% and 4.0%, respectively. The mortality rate increased as raw PESI score increased. Risk of death correlated with risk class ( P < 0.001). There were no deaths in risk classes I–III, but 30‐ and 90‐day mortality for class V were 9.2% and 10.5%, respectively. Overall, mean PESI scores were similar between observers: 103.3 ± 39.3 and 96.5 ± 37.6 ( P = NS). The inter‐rater variability was good (kappa = 0.69; P < 0.0001). Conclusions : The PESI correlates with 30‐ and 90‐day mortality. It represents a reproducible scoring tool to risk stratify patients with acute PE.