Development of a risk stratification model for permanent pacemaker implantation after Transcatheter Aortic Valve Replacement (TAVR)
T Lv, S Qiao, Y Han, F Ren, T Chen, M Dong, T LiuAbstract
Background
Permanent pacemaker implantation (PPI) is a common complication after transcatheter aortic valve replacement (TAVR), and preoperative identification of high-risk patients is crucial for clinical decision-making. Most existing PPI prediction models after TAVR focus on clinical, echocardiographic and procedural factors, while the value of electrocardiographic (ECG) indicators remains under-explored. This study aimed to establish and validate a clinical prediction scoring model for PPI after TAVR integrating ECG indicators.
Methods
A total of 157 TAVR patients were enrolled and stratified by PPI status (No PPI, n=136; PPI, n=21). Univariate analysis (P<0.1) screened variables, and multivariate Logistic regression identified independent PPI predictors. A clinical prediction scoring system was built based on the regression coefficients of these predictors, with patients stratified into low-, medium-, and high-risk groups by total score. The model was evaluated using ROC and calibration curves.
Results
Univariate analysis identified 15 variables with P<0.05. Multivariate Logistic regression showed that valve implantation depth (OR=1.430, 95%CI:1.092–1.915, P=0.011), aortic valve peak velocity (OR=0.994, 95%CI:0.989–0.998, P=0.003), DIIISA (an ECG parameter, OR=4.055, 95%CI:1.375–12.870, P=0.011), moderate/severe aortic regurgitation (AR) (OR=6.215, 95%CI:1.463–39.193, P=0.025), and prosthetic valve size (OR=0.616, 95%CI:0.432–0.830, P=0.003) were independent predictors of PPI. Height (OR=0.938, 95%CI:0.866–1.010, P=0.099) was retained due to its clinical relevance despite marginal significance. The clinical prediction scoring model was constructed by converting regression coefficients into integer weights: valve implantation depth (+2 points/SD increase), DIIISA (an ECG parameter, +8 points/SD increase), moderate/severe AR (+10 points if present), prosthetic valve size (-3 points/SD increase), while aortic valve peak velocity and height had no contribution (scoring weight=0). Patients were stratified into three risk groups based on the total score (-3 to 20 points, calculated after continuous variable standardization): low-risk (≤5 points, PPI risk <20%), medium-risk (6–12 points, PPI risk 20%–50%), and high-risk (≥13 points, PPI risk >50%). The model showed good discrimination (AUC=0.8697, sensitivity=76.19%, specificity=86.03%) and calibration (mean absolute error=0.023).
Conclusions
The constructed clinical prediction scoring model for PPI after TAVR, which innovatively integrates ECG indicators (mainly DIIISA) with valve implantation depth, moderate/severe AR, and prosthetic valve size, has good discrimination and calibration. Stratifying patients into low-, medium-, and high-risk groups by total score, it facilitates preoperative PPI risk stratification clinically, helps clinicians formulate individual prevention strategies, and fills the gap of insufficient ECG application in existing PPI prediction models.Figure