Table 3.

Multivariable Linear Regression Models for Percent Change in LVMi from Baseline to 30 days after TAVR

Variable	β-estimate (%)	p-value
Model 1 – Clinical Model
Intercept	−9.0	0.28
Age (per year)	0.1	0.48
Male sex	−5.7	<0.001
Baseline LVMi (per 1 g/m2 increase)	0.1	<0.001
STS score (per 1 unit increase)	−0.3	0.02
Pacemaker at baseline	−3.1	0.02
Model 2 – Clinical + Baseline Echo
Intercept	−17.9	0.04
Age (per year)	0.1	0.36
Male sex	−5.7	<0.001
Baseline LVMi (per 1 g/m2 increase)	0.1	<0.001
STS score (per 1 unit increase)	−0.2	0.09
Pacemaker at baseline	−2.9	0.04
Midwall fractional shortening (per 1% increase)	0.3	<0.001
Model 3 – Clinical + Baseline Echo + 30d Echo
Intercept	−17.2	0.05
Age (per year)	0.1	0.30
Male sex	−5.7	<0.001
Baseline LVMi (per 1 g/m2)	0.1	<0.001
STS score (per 1 unit)	−0.2	0.07
Pacemaker at baseline	−3.2	0.02
Midwall fractional shortening (per 1% increase)	0.3	<0.001
30 day transvalvular mean gradient (per 1 mmHg increase)	−0.3	0.05

Linear regression models evaluating variables associated with % D in LVMi from baseline to 30 days post-TAVR. A positive β signifies regression/decrease in LVMi from baseline to 30 days post-TAVR.

Model 1: Selection model including age, sex, and baseline LVMi (forced variables) and other variables selected from among the clinical variables (Table 1) that differ (p≤0.10) between those with greater vs. lesser regression at 30 days.

Model 2: Selection model including the same variables as Model 1 in addition to baseline echo variables (Table 2) that differ (p≤0.10) between those with greater vs. lesser regression at 30 days (variables were excluded if they were a part of the calculation of LVMi).

Model 3: Selection model including the same variables as Model 2 in addition to 30 day post-TAVR echo variables (Table 2) that differ (p≤0.10) between those with greater vs. lesser regression at 30 days (variables were excluded if they were a part of the calculation of LVMi).

Abbreviations: STS, Society of Thoracic Surgery; LVMi, left ventricular mass index.