Inter-rater Reliability of a 13-Category Arterial Doppler Waveform Classification and Practice of French Vascular Physicians

Background: Arterial Doppler Ultrasound waveform (DW) analysis allows the detection and evaluation of lower extremity peripheral artery disease. The high heterogeneity of the reported description of DW is reduced by the use of classification. However, the reliability of these classifications is either unknown or low to moderate and practices of vascular caregivers regarding the use of these classifications remain unknown. Aims: This study aims to assess the inter-observer reliability of the Saint-Bonnet classification, a 13-category DW classification. The secondary objective was to determine the utilization rate of the most common classifications and the ability of these classifications to describe DW. Methods: A national survey was conducted among all vascular physicians of French society of vascular medicine. They were invited by email to describe on a website 20 DW without and with the display of the Saint-Bonnet classification. The reliability of this classification was estimated by Fleiss' Kappa expressed with [95% confidence interval]. A semantic analysis allowed us to classify the physicians' responses according to the terms used. Finally we have evaluated for each classification the rate of misuse, i.e., the addition of a complementary term to the defined categories. Results: One hundred and ten physicians participated and only 5% of these were familiar with Saint-Bonnet classification. Fifty-four percent of vascular physicians used no classification at all. Vascular physicians used the Spronk (four-category), Descotes (five-category) and Saint-Bonnet (13-category) classifications for respectively, 31, 10, and 5%. Kappa coefficient of Fleiss (κ) was 0.546 [0.544–0.547] (p < 0.001). Reliability by category ranges from κ of 0.075 to 0.864. In multivariate analysis, the use of a classification was associated with fewer years of experience and was dependent on geographic location. Misuse rate by classification was 88, 82, and 5% using Spronk, Descotes and Saint-Bonnet classifications respectively. Conclusion: The reliability of Saint-Bonnet classification is weak to moderate by vascular physicians who are not familiar with its use. However, unlike the other classifications, it seems to be sufficiently precise so that the user does not need to complete its description. There is a significant heterogeneity in the use of arterial Doppler classifications in France.