Abstract Body (Do not enter title and authors here): Background: Tachyarrhythmias are a complication of congenital heart disease (CHD) surgery associated with significant morbidity and mortality. Risk scores to predict supraventricular tachycardia (SVT) and accelerated junctional rhythm/junctional ectopic tachycardia (AJR/JET) were previously published, but their performance not evaluated.
Objective: To validate risk prediction scores for post-operative SVT and AJR/JET in an external cohort.
Methods: Prospective cohort of children 0-18 years undergoing surgery for CHD from 2020-2023. In-hospital telemetry and 12-lead ECGs were reviewed daily for the presence of arrhythmia. C-statistic and Hosmer-Lemeshow tests were used to quantify discrimination and calibration of the model.
Results: From 1190 surgeries in 1109 patients (54% male, age 1.71 [IQR 0.3, 6.5] years), SVT occurred in 124 (10%) and AJR/JET in 68 (6%). The observed risk of arrhythmia was significantly higher in the prospective validation cohort than the retrospective cohort. Patients in the prospective cohort were older, more likely to have heterotaxy syndrome and with longer and more complex surgeries than the retrospective cohorts. The model accurately predicted the risk of SVT with a C-statistics of 0.76 [95% CI 0.72, 0.80] and AJR/JET with a C-statistics of 0.63 [95% CI 0.55, 0.70] for AJR/JET. However, the model was not well calibrated with Hosmer-Lemeshow p-value <0.001.
Conclusion: Risk prediction scores for post-operative arrhythmia created on retrospective cohort did not perform well in a prospective validation study. Arrhythmia outcomes were significantly higher with prospective review of telemetry than reported with diagnostic and complication codes in a retrospective cohort.