We used latent class trajectory models to identify NT-proBNP trajectories over time. This is a specialized form of finite mixture modeling designed to identify latent classes of individuals following similar progressions of a determinant over time (14 ). Our models used second-order polynomials. After data standardization, we calculated the posterior probabilities of participants for each trajectory and then assigned participants
post hoc to the trajectory with the highest probability. We estimated the best-fitting number of trajectories based on a minimum Bayesian Information Criterion while maintaining the posterior probabilities by class (>0.70) and class size (≥2% of the population) (15 (
link)).
Continuous variables are presented as means and standard deviations (SD) and compared by Student’s
t-test. Categorical variables are shown as percentages and frequencies and compared using the χ2 test or Fisher’s exact test, as appropriate. A two-sided
P < 0.05 was considered statistically significant. We used multivariable logistic regression models to identify predictors of distinct NT-proBNP trajectories. The candidate variables were selected
a priori for inclusion in the univariable logistic regression models Variates with
P < 0.05 were then entered into a multivariate model to identify independent factors by the regression stepwise method. Time-to-first event curves are displayed using Kaplan–Meier estimates and compared by the log-rank test. Hazard ratios (HRs) and 95% confidence intervals (CIs) are estimated by Cox proportional hazards regression models. Adjustments were made for baseline variables [age, sex, body mass index (BMI), Society of Thoracic Surgeons (STS) score, diabetes, hypertension, chronic obstructive pulmonary disease (COPD), estimated glomerular filtration rate (eGFR), prior stroke, atrial fibrillation/flutter (AF), left ventricular ejection fraction (LVEF), NYHA class and NT-proBNP levels] and procedural complications (new or aggravated atrioventricular block, vascular complications, annular rupture, coronary obstruction, circulation collapse, aortic regurgitation paravalvular ≥ moderate and aortic regurgitation transvalvular ≥ moderate), and 30-day post-TAVR outcomes (NYHA ≥ Class III, myocardial infarction, stroke, disabling stroke, bleeding, life-threatening bleeding, new permanent pacemaker, new atrial fibrillation, and renal dysfunction). Covariates for this analysis were selected
a priori based on historical prognostic relevance or clinical judgment, which were further selected in the multivariate analyses based on their statistical significance. Statistical analyses were performed using R statistical software (version 4.0.3).
Zhou Y., Zhu Q., Hu P., Li H., Lin X., Liu X., Pu Z, & Wang J. (2023). NT-proBNP trajectory after transcatheter aortic valve replacement and its association with 5-year clinical outcomes. Frontiers in Cardiovascular Medicine, 10, 1098764.