Table 3.

Multivariable models on GOSE and mortality at 6-month follow-up in OzENTER-TBI (Model 1 and 2)

OzENTER-TBI	6-month GOSE N = 141 patients, 92 GOSE ≤ 4			6-month mortality N = 141 patients, 39 died
Model 1	ORa	95% CI	p value	ORa	95% CI	p value
PaO2max (for 10 mmHg increase)	1.01	0.98–1.04	0.433	1	0.97–1.04	0.898
PaO2mean (for 10 mmHg increase)	1.01	0.96–1.07	0.656	1.05	0.99–1.11	0.118
ΔPaO2mean (for 10 mmHg increase)	1.03	0.96–1.12	0.376	1.08	1–1.18	0.054

	6-month GOSE N = 141 patients, 92 GOSE ≤ 4			6-month mortality N = 141 patients, 39 died
Model 2	ORb	95% CI	p value	OR*	95% CI	p value
FiO2max (for 0.1 increase)	1.06	0.89–1.26	0.492	1	0.83–1.23	0.963
FiO2mean (for 0.1 increase)	1.02	0.77–1.34	0.911	1.32	0.98–1.8	0.069
ΔFiO2mean (for 0.1 increase)	1.15	0.79–1.69	0.483	1	0.68–1.48	0.981

Model 1. Adjusted odds ratio with 95% confidence intervals effect of exposure to high blood oxygen levels within 7 days of ICU admission on GOSE and mortality at 6-month follow-up. Validation on OzENTER-TBI. Standard logistic regression models adjusted for age, pupillary reactivity (both reactive, one reactive, both unreactive), GCS Motor (any flexion, none/extension, localizes/obey), Injury Severity Score, and, once at a time, PaO_2max, PaO_2mean and ΔPaO_2mean for OzENTER-TBI with a dummy variable for center. Model 2. Adjusted odds ratio with 95% CI of GOSE and mortality at 6-month follow-up in TBI patients exposed to high supplemental oxygen administration within 7 days of ICU admission in OzENTER-TBI. Standard logistic regression models adjusted for age, pupillary reactivity (both reactive, one reactive, both unreactive), GCS Motor (any flexion, none/extension, localizes/obey) and, once at a time, FiO_2max, FiO_2mean and ΔFiO_2mean for OzENTER-TBI with a dummy variable for center. Full models with all covariates estimates are reported in the Supplemental material

^aOR is for 10 mmHg increase in PaO₂ covariate

^bOR regards 0.1 increments in FiO₂ covariate