. Author manuscript; available in PMC: 2021 Jun 1.

Published in final edited form as: J Am Soc Echocardiogr. 2020 Mar 26;33(6):756–762.e1. doi: 10.1016/j.echo.2020.01.014

Table 2:

Adjusted Outcomes

Mortality at 30 days
	OR	95% CI	P
^†Multivariable logistic regression overall (N=219,238)
TEE	0.77	0.73 – 0.82	<0.001
MuInvariable logistic regression: excluding mitral valve repair (N = 191,999)
TEE	0.79	0.74 – 0.83	<0.001
^*MuInvariable logistic regression: Propensity score matched (with replacement) analysis: N=186,738 exposures (N = 186,738) matched to N=186,738 (N = 14,382 weighted) controls
TEE	0.75	0.67 – 0.85	<0.001

Length of Hospitalization
	Absolute % Change	95% CI	P
^†Multiple linear regression: overall (N=219,238)
TEE	<0.01%	−0.61 – 0.62%	0.99
Multiple linear regression: excluding mitral valve repair (N=191,999)
TEE	−0.01%	−0.66 – 0.63%	0.97
Multiple linear regression: Propensity score matched (with replacement) analysis: N=186,738 exposures (N = 186,738) matched to N=186,738 (N = 14,382 weighted) controls
TEE	0.47%	−1.01 – 1.96%	0.53

Abbreviations: TEE, transesophageal echocardiography; OR, odds ratio; CI, confidence interval

^†

Full regression models for primary analysis are found in Appendix Table 2 (multivariable logistic regression for 30-day mortality) and Table 3 (multiple linear regression for length of hospitalization).

Because the propensity match was performed with replacement, we performed an unconditional multivariable logistic regression analysis on the 30-day mortality outcome in order to account for the multiplicities as weighted in the matched controls.