Abstract 13198: Assessment of the Clot Real-Time Risk Prediction Model of Hospital-Associated Venous Thrombosis in Children After Congenital Heart Surgery

• Published in: Circulation (New York, N.Y.) Vol. 148; no. Suppl_1
• Main Authors: Sunthankar, Sudeep D, Moore, Ryan P, Byrne, Daniel W, Domenico, Henry J, Wheeler, Allison P, Walker, Shannon C, Kannankeril, Prince J
• Format: Journal Article
• Language: English
• Published: 07.11.2023
• ISSN: 0009-7322; 1524-4539
• DOI: 10.1161/circ.148.suppl_1.13198

Abstract only Introduction: Children with congenital heart disease (CHD) are at high risk for hospital-associated venous thromboembolism (HA-VTE) after cardiac surgery. The Children's Likelihood Of Thrombosis (CLOT) trial validated a real-time predictive model for HA-VTE using data extracted from the EHR for pediatric inpatients with an available online application (https://cqs.app.vumc.org/shiny/PediatricVTEPrediction/). Hypothesis: We tested the hypothesis that addition of CHD surgical specific data would improve the CLOT model for HA-VTE prediction in the CHD surgical population. Methods: The performance of the model in CHD patients from 2010-2022, was assessed using three iterations of the CLOT model: 1) the original CLOT model, 2) the original model refit using only data from the CHD cohort, and 3) the model updated with the addition of cardiopulmonary bypass time, surgical severity (STAT category), and BMI as covariates. The discrimination of the three models was quantified and compared using AUROC. Results: Our cohort included 1457 patients (median 2.0 IQR[0.5-5.2] y/o) following cardiac surgery. HA-VTE was present in 5% of our CHD cohort compared to 1% in the general pediatric population. The AUROC for the original (0.80 [95% CI: 0.74-0.85]), refit (0.85 [0.80-0.88]), and updated (0.86 [0.81-0.90]) models demonstrated excellent discriminatory ability within the CHD cohort (Figure 1). Conclusions: The addition of cardiac specific features improved model performance for discrimination of HA-VTE; however, this benefit was marginal compared to refitting the original model to the CHD cohort. This suggests strong predictive models, such as CLOT, can be optimized for cohort subsets without additional data extraction, thus reducing cost of model development and deployment. Identification of high-risk individuals using real-time predictive models deployed in the EHR may allow for prophylactic therapy and reduction of thrombosis in CHD surgical patients.