Table 6

Predictors of neurologic improvement among all patients and operative patients. Multivariable logistic regression controlling for age, sex, admission neurologic exam, and level of injury.

Variable	Univariable Logistic Regression		Multivariable Logistic Regression
Variable	OR (95% CI)	P	OR (95% CI)	P
All patients, N = 193^{^a}
Age		0.432		-
Young, <30 y (n = 75)	Reference	-	-	-
Old, ≥30 y (n = 118)	0.71 (0.29, 163)	-	-	-
Sex		0.820		-
Female (n = 25)	Reference	-	-	-
Male (n = 168)	0.88 (0.30, 3.19)	-	-	-
Injury level
Upper thoracic (ref) vs lower thoracic/lumbar (n = 29)^{^b}	5.52 (1.10, 100)	0.100	-	-
Lower thoracic (ref) vs upper thoracic/lumbar (n = 101)	3.97 (1.67, 10.56)	0.002	0.61 (0.03, 3.96)	0.656
Lumbar (ref) vs thoracic (n = 63)	0.14 (0.05, 0.33)	<0.001	0.14 (0.01, 0.83)	0.071
Insurance		0.947		-
Private (n = 20)	Reference	-	-	-
Public (n = 173)	1.04 (0.23, 3.40)	-	-	-
Time from injury to admission
Continuous, d	0.93 (0.85, 0.99)	0.093	0.93 (0.84, 1.00)	0.113
Operative		0.034		0.067
Nonoperative (n = 55)	Reference	-	Reference	-
Operative (n = 138)	3.83 (1.27, 16.61)	-	3.58 (1.02, 17.07)	-
Admission neurologic examination		<0.001		0.002
Complete: ASIA A (n = 113)	Reference	-	Reference	-
Incomplete: ASIA B–D (n = 80)	8.71 (3.38, 27.05)	-	5.74 (1.94, 18.00)	-
Operative patients, N = 138
Age		0.299		-
Young, <30 y (n = 57)	Reference	-	-	-
Old, ≥30 y (n = 81)	1.63 (0.66, 4.27)	-	-	-
Sex		0.921
Female (n = 23)	Reference	-	-	-
Male (n = 115)	1.06 (0.35, 3.94)	-	-	-
Injury level
Upper thoracic (ref) vs lower thoracic/lumbar (n = 20)	4.85 (0.93, 89.30)	0.133	-	-
Lower thoracic (ref) vs upper thoracic/lumbar (n = 71)	3.36 (1.34, 9.22)	0.012	0.47 (0.02, 2.89)	0.495
Lumbar (ref) vs thoracic (n = 47)	0.17 (0.06, 0.42)	< 0.001	0.09 (0.00, 0.51)	0.027
Time from admission to operating room
Continuous, d	1.01 (0.99, 1.03)	0.237	-	-

Abbreviation: ASIA, American Spinal Injury Association.
↵a ASIA E patients were excluded, and none got worse in this cohort; only patients with a discharge ASIA score were included.
↵b Excluded from the logistic regression model due to collinearity.