Development of an electrocardiographic prediction model for outflow tract idiopathic premature ventricular contraction with decreased left ventricular ejection fraction

Abstract

Abstract Background The incidence of premature ventricular contraction (PVC) can be as high as 70% to 75% in general population. Even in the absence of any underlying heart disease, isolated ventricular premature contractions (IPVCs) can still lead to cardiac structural remodeling and impaired heart function during long-term follow-up. Currently, there is a lack of predictive models to assess the risk of idiopathic PVCs associated with left ventricular dysfunction, leading to over or under treatment for IPVC. Objective To develop an electrocardiographic prediction model of IPVC with decreased left ventricular ejection fraction (LVEF). Methods A retrospective analysis of 419 patients first diagnosed with idiopathic outflow tract PVC from January 2020 to December 2023 was performed. The cohort was randomly divided into training and validation sets in a 7:3 ratio. A nomogram predictive model was constructed based on multifactorial logistic regression analysis. Model discrimination was assessed using ROC curve and AUC, while calibration was evaluated with the Hosmer–Lemeshow test. Decision curve analysis (DCA) assessed clinical applicability, and net reclassification improvement (NRI) along with integrated discrimination improvement (IDI) indices were calculated to assess additional predictive value compared with traditional PVC burden. Results No significant differences in baseline data were observed between the training and validation sets (P > 0.05). The Nomogram prediction model involving PVC burden, PVC bigeminy, and SDNN ≤ 80 ms. The model showed good discrimination both in the training and validation set, with AUC of 0.912 (95% CI 0.872–0.951) and 0.894 (95% CI: 0.803–0.985), respectively. Calibration curve indicated good agreement between predicted probability and actual incidence in both datasets with Hosmer and Lemeshow goodness-of-fit (GOF) test of χ 2 = 9.832, P = 0.277 and χ 2 = 5.935, P = 0.654. DCA also showed good application value. Compared with PVC burden alone, the model showed superior predictive performance with NRI of 0.851 (95%CI 0.624–1.078; P < 0.001) and IDI of 0.152 (95%CI 0.094–0.209; P < 0.001). Conclusion The ECG prediction model for IPVC with decreased LVEF developed in this study is highly effective and easy to implement. Due to the accessible nature of Holter data, this model provides valuable insights for assessing IPVC prior to clinical decision-making regarding treatment.

Keywords

• IPVC
• Decreased LVEF
• PVC burden
• SDNN
• PVC bigeminy