Abstract
Prediction tools for actionable arrhythmias after implantable cardiac monitor (ICM) implantation in patients with unexplained syncope remain limited. This dual-center retrospective study included 302 patients divided chronologically into development (n = 91) and temporal validation (n = 211) cohorts. The endpoint was the first ICM-detected arrhythmia prompting permanent pacemaker or defibrillator implantation, catheter ablation, initiation of Vaughan Williams Class I or III antiarrhythmic drugs, or oral anticoagulation. A multivariable Cox model was developed and internally validated using 1000 bootstrap resamples, repeating variable selection and model fitting. During a median follow-up of 23.4 months, 78 patients reached the endpoint, including 32 in the development cohort and 46 in the temporal validation cohort. The nomogram included age, a history of recurrent syncope, and first-degree atrioventricular block. The optimism-corrected C-index was 0.720 in development. In temporal validation, the C-index was 0.707 (95% confidence interval, 0.623–0.790), and the 12- and 24-month time-dependent areas under the receiver operating characteristic curve were 0.733 and 0.741, respectively. The nomogram showed moderate discrimination in temporal validation at the same two centers. The model underestimated risk at 12 months, whereas mean predicted risk was closer to observed risk at 24 months.