Table 3.

The 28-day mortality rate from initiation of invasive mechanical ventilation with dexmedetomidine use, propensity score matched cohort (1 January 2020 to 3 November 2022).

	Model 1 a		Model 2 b
	n = 3806		n = 576
Cox Regression Model	aHR (95% CI)	p	aHR (95% CI)	p
Multivariable (DEX use) c	0.81 (0.73, 0.90)	<0.001	0.95 (0.72, 1.25)	0.73
Univariable (DEX use) d	0.90 (0.81, 0.99)	0.04	1.05 (0.80, 1.36)	0.74

Abbreviations: aHR = adjusted hazard ratio; DEX = dexmedetomidine; 95% CI = 95% confidence interval. ^a Model 1: Health system sites included that lacked PaO₂/FiO₂ data. ^b Model 2: Only health system sites that had PaO₂/FiO₂ data were included. ^c Model 1 adjusted for (i) dexmedetomidine as a time-varying covariate, (ii) age, (iii) BMI, (iv) mCCI, (v) mSOFA (central nervous system and respiratory component removed), (vi) dexamethasone use, and (vii) remdesivir use; Model 2 adjusted for (i) dexmedetomidine as a time varying covariate, (ii) age, (iii) BMI, (iv) mCCI, (v) PaO₂/FiO₂, (vi) mSOFA (central nervous system component removed), (vii) dexamethasone use, and (viii) remdesivir use. In both models, adjusting for site differences was performed by including a categorical variable with levels for each data partner representing >5% of our population and then all others grouped together. ^d Model 1 and Model 2 adjusted for dexmedetomidine use as a time-varying covariate only.