Mofidi R, McBride OMB, Green BR, Gatenby T, Walker P, Milburn S. Validation of a Decision Tree to Streamline Infrainguinal Vein Graft Surveillance.
Ann Vasc Surg 2016;
40:216-222. [PMID:
27890844 DOI:
10.1016/j.avsg.2016.07.082]
[Citation(s) in RCA: 4] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/26/2016] [Revised: 06/30/2016] [Accepted: 07/14/2016] [Indexed: 11/15/2022]
Abstract
BACKGROUND
Duplex ultrasound (DU)-based graft surveillance remains controversial. The aim of this study was to assess the ability of a recently proposed decision tree in identifying high-risk grafts which would benefit from DU-based surveillance.
MATERIALS AND METHODS
Consecutive patients undergoing infrainguinal vein graft bypass from January 2008 to December 2015 were identified from the National Vascular registry and enrolled in a duplex surveillance program. An early postoperative DU was performed at a median of 6 weeks (range: 4-9 weeks). Grafts were classified into high risk or low risk based on the findings of the earliest postoperative scan and 4 established risk factors for graft failure (diabetes, smoking, infragenicular distal anastomosis, and revision bypass surgery) using a classification and regression tree (CRT). The accuracy of the CRT model was evaluated using area under receiver operator characteristic (AROC) curve.
RESULTS
About 278 vein graft bypasses were performed; 29 grafts had occluded by the first surveillance visit; 249 vein grafts were entered into surveillance. Sixty-four (23%) developed critical stenosis. Overall 30-month primary patency, primary-assisted patency, and secondary patency rates were 71.2%, 77.2%, and 80.1%, respectively. AROC for prediction of graft stenosis or occlusion was 83% (95% confidence interval [CI]: 78-87%). The sensitivity and specificity of the CRT model for prediction of graft stenosis or occlusion were 95% (95% CI: 88-98%) and 52.2% (95% CI: 45-60%).
CONCLUSIONS
A prediction model based on commonly recorded clinical variables and early postoperative DU scan is accurate at identifying grafts which are at high risk of failure. These high-risk grafts may benefit from DU-based surveillance.
Collapse