Sensitivity of a Simple Noninvasive Screening Algorithm for Chronic Thromboembolic Pulmonary Hypertension after Acute Pulmonary Embolism

• Main Authors: Yvonne M. Ende-Verhaar, Dieuwertje Ruigrok, Harm Jan Bogaard, Menno V. Huisman, Lilian J. Meijboom, Anton Vonk Noordegraaf, Frederikus A. Klok
• Format: Article
• Language: English
• Published: Georg Thieme Verlag KG 2018-01-01
• Series: TH Open
• Subjects: pulmonary embolism; cteph; screening algorithm; sensitivity; reproducibility
• Online Access: http://www.thieme-connect.de/DOI/DOI?10.1055/s-0038-1636537
• ISSN: 2512-9465; 2512-9465

Abstract Background Recently, we constructed a noninvasive screening algorithm aiming at earlier chronic thromboembolic pulmonary hypertension (CTEPH) detection after acute pulmonary embolism (PE), consisting of a prediction score and combined electrocardiography (ECG)/N-terminal pro-brain natriuretic peptide (NT-proBNP) assessment. The aim of this study was to confirm the algorithm's sensitivity for CTEPH detection and to evaluate the reproducibility of its individual items. Methods Two independent researchers calculated the prediction score in 54 consecutive patients with a history of acute PE and proven CTEPH based on clinical characteristics at PE diagnosis, and evaluated the ECG and NT-proBNP level assessed at the moment of CTEPH diagnosis. Interobserver agreement for the assessment of the prediction score, right-to-left ventricle (RV/LV) ratio measurement on computed tomography pulmonary angiography, as well as ECG reading was evaluated by calculating Cohen's kappa statistics. Results Median time between PE diagnosis and presentation with CTEPH was 9 months (interquartile range: 5–15). The sensitivity of the algorithm was found to be 91% (95% confidence interval [CI]: 79–97%), indicating that 27 of 30 cases of CTEPH would have been detected when applying the screening algorithm to 1,000 random PE survivors with a 3% CTEPH incidence (projected negative predictive value: 99.7%; 95% CI: 99.1–99.9%). The interobserver agreement for calculating the prediction score, RV/LV ratio measurement, and ECG reading was excellent with a kappa of 0.96, 0.95, and 0.89, respectively. Conclusion The algorithm had a high sensitivity of 91% and was highly reproducible. Prospective validation of the algorithm in consecutive PE patients is required before it can be used in clinical practice.