Background: The ISTH’s SSC recently proposed a definition of pulmonary embolism (PE)-related death, aiming to increase the reproducibility of adjudication of death events in venous thromboembolism studies.
Aims: To evaluate the accuracy and interrater reliability of the ISTH definition of PE-related death.
Methods: We used data from an autopsy study that was performed at the NewYork-Presbyterian-Hospital between 01/2010 and 07/2019. Included in this study were all patients with autopsy-confirmed PE-related death (cases) during that time frame, combined with patients who died in 2018 from a cause other than PE (controls). Based on clinical summaries which were retrospectively collected from the electronical health records and autopsy reports, two adjudicators independently, blinded to case-to-control ratio and autopsy results, determined the cause of death in each patient using the ISTH definition and classification (Figure). Patients with conflicting adjudications for cause of death were independently assessed by a third adjudicator. The primary outcome was autopsy-confirmed PE-related death. We determined the sensitivity and specificity of the ISTH definition for autopsy-confirmed PE-related death, and its interrater reliability using the percentage agreement and Cohen’s kappa.
ISTH definition for PE-related death and classification of the cause of death in venous thromboembolism studies
Results: A total of 126 deaths (median age, 68 years [range, 21-94], 60 [48%] women) were adjudicated, of which 29 were autopsy-confirmed PE-related deaths. The ISTH definition’s sensitivity and specificity for autopsy-confirmed PE-related death were 45% (95%CI, 26-64%) and 99% (95%CI, 94-100%), respectively. Interrater reliability for PE-related death was substantial (percentage agreement, 94%; kappa, 0.73; 95%CI, 0.50-0.91; Table). When deaths classified in category B were also considered to be PE-related, sensitivity and specificity for autopsy-confirmed PE-related death were 83% (95%CI, 64-94%) and 74% (95%CI, 64-83%), respectively, and the interrater agreement was moderate (percentage agreement, 71%; kappa, 0.41; 95%CI, 0.24-0.57).
| Category | A2 | A3 | B1 | B2 | C |
|---|---|---|---|---|---|
| A2. Objectively confirmed PE | 5 | – | – | – | – |
| A3. PE most likely the main cause of death | 3 | 3 | 5 | – | – |
| B1. Undetermined despite information | – | 1 | 19 | – | 13 |
| B2. Insufficient information | 1 | – | 9 | – | 4 |
| C. Cause of death other than PE | – | – | 20 | – | 43 |
| Abbreviation: PE, pulmonary embolism. Subcategory A1 is not displayed, because the study design did not allow us to classify death events as autopsy-confirmed PE. |
|||||
Classification of the cause of death by adjudicator 1 (columns) and adjudicator 2 (rows)
Conclusions: Adjudication of the cause of death using the ISTH definition results in very high specificity, moderate sensitivity and good interrater reliability for PE-related death.
To cite this abstract in AMA style:
Tritschler T, Salvatore SP, Kahn SR, Garcia D, Delluc A, Kraaijpoel N, Langlois N, Girard P, Le Gal G. ISTH Definition of Pulmonary Embolism-related Death and Classification of the Cause of Death in Venous Thromboembolism Studies: An Autopsy Study [abstract]. Res Pract Thromb Haemost. 2021; 5 (Suppl 2). https://abstracts.isth.org/abstract/isth-definition-of-pulmonary-embolism-related-death-and-classification-of-the-cause-of-death-in-venous-thromboembolism-studies-an-autopsy-study/. Accessed November 29, 2023.« Back to ISTH 2021 Congress
ISTH Congress Abstracts - https://abstracts.isth.org/abstract/isth-definition-of-pulmonary-embolism-related-death-and-classification-of-the-cause-of-death-in-venous-thromboembolism-studies-an-autopsy-study/