Table 4.

Adjusted logistic regression for dichotomized GOS (favorable outcome: GOS = 4 and 5, and unfavorable outcome: GOS < 4) for core model and extended model with and without the addition of PTCI. AUC and AIC for each model are represented along with the ANOVA comparison between the model with and without the addition of PTCI. We used Nahelkerke’s Pseudo R² for consistency with the original IMPACT study

	Core model			Extended model
	OR	95% C.I.	p	OR	95 C.I.	p
Age	1.03	1.01–1.06	0.002	1.01	1.01–1.07	0.002
GCSm	0.34	0.14–0.77	0.012	0.33	1.27–7.14	0.015
Pupils	2.76	1.08–7.35	0.035	1.79	0.64–5.05	0.263
PTCI				4.52	1.88–11.39	< 0.001
AUC	0.73 (95% C.I.: 0.66–0.82)			0.79 (95% C.I.: 0.71–0.83)
AUC comparison	Model with PTCI is better than the model without PTCI p = 0.0007
Nagelkerke’s R2	0.21			0.33
Age	1.03	1.01–1.06	0.003	1.03	1.01–1.06	0.004
GCSm	0.29	0.12–0.07	0.007	0.27	0.10–0.67	0.007
Pupils	3.11	1.14–8.90	0.027	2.09	0.68–6.48	0.190
Hypotension	2.24	0.73–7.12	0.160	3.29	0.99–11.60	0.055
Hypoxia	1.58	0.41–6.01	0.495	2.00	0.43–11.63	0.37
SAH	1.23	0.49–3.19	0.652	1.30	0.46–3.80	0.62
EDH	0.93	0.28–0.41	0.882	7.96	0.26–2.28	0.68
Marshall CT score	–	–	0.991	–	–	0.99
PTCI				6.84	2.58–19.7	< 0.001
AUC	0.74 (0.65–0.81)			0.82 (0.69–0.85)
Nagelkerke’s R2	0.26			0.39
AUC comparison	Model with PTCI is better than the model without PTCI p = 0.00008

GCSm Glasgow Coma Scale motor score, OR proportional odds ratio, C.I. confidence interval, SAH traumatic subarachnoid hemorrhage, EDH epidural hematoma. Hypotension and hypoxia refer to both pre-hospital and intra-hospital periods