Thromb Haemost 2010; 104(06): 1258-1262
DOI: 10.1160/TH10-07-0426
New Technologies, Diagnostic Tools and Drugs
Schattauer GmbH

The inter-rater reliability of the Pulmonary Embolism Severity Index

Montserrat Fraga
1   Division of General Internal Medicine University of Lausanne, Lausanne, Switzerland
,
Patrick Taffé
2   Institute of Social and Preventive Medicine, University of Lausanne, Lausanne, Switzerland
,
Marie Méan
1   Division of General Internal Medicine University of Lausanne, Lausanne, Switzerland
,
Olivier Hugli
3   Emergency Department, University of Lausanne, Lausanne, Switzerland
,
Stéphanie Witzig
1   Division of General Internal Medicine University of Lausanne, Lausanne, Switzerland
,
Gérard Waeber
1   Division of General Internal Medicine University of Lausanne, Lausanne, Switzerland
,
Drahomir Aujesky
4   Division of General Internal Medicine, Bern University Hospital, Bern, Switzerland
› Author Affiliations
Further Information

Publication History

Received: 01 July 2010

Accepted after minor revision: 30 July 2010

Publication Date:
24 November 2017 (online)

Zoom Image

Summary

The Pulmonary Embolism Severity Index (PESI) is a validated clinical prognostic model for patients with acute pulmonary embolism (PE). Our goal was to assess the PESI’s inter-rater reliability in patients diagnosed with PE. We prospectively identified consecutive patients diagnosed with PE in the emergency department of a Swiss teaching hospital. For all patients, resident and attending physician raters independently collected the 11 PESI variables. The raters then calculated the PESI total point score and classified patients into one of five PESI risk classes (I-V) and as low (risk classes I/II) versus higher-risk (risk classes III-V). We examined the inter-rater reliability for each of the 11 PESI variables, the PESI total point score, assignment to each of the five PESI risk classes, and classification of patients as low versus higher-risk using kappa (κ) and intra-class correlation coefficients (ICC). Among 48 consecutive patients with an objective diagnosis of PE, reliability coefficients between resident and attending physician raters were > 0.60 for 10 of the 11 variables comprising the PESI. The inter-rater reliability for the PESI total point score (ICC: 0.89, 95% CI: 0.81–0.94), PESI risk class assignment (κ: 0.81, 95% CI: 0.66–0.94), and the classification of patients as low versus higher-risk (κ: 0.92, 95% CI: 0.72–0.98) was near perfect. Our results demonstrate the high reproducibility of the PESI, supporting the use of the PESI for risk stratification of patients with PE.