Table 3.

Result of univariable logistic regression for potential risk factors of cerebral infarcts (CI) and multivariable logistic regression model for CI chosen by a stepwise selection algorithm in patients with childhood-onset craniopharyngioma recruited in KRANIOPHARYNGEOM 2007 between 2007 and 2019.

Effect	Comparison	p-value	Odds ratio
		(Wald)	(95% confidence interval)
	Univariable logistic regression model for potential risk factors of CI
Age at diagnosis (n = 242)	per year	0.9637	0.998 (0.908; 1.096)
Sex (n = 242)	female vs. male	0.3372	0.668 (0.294; 1.522)
Initial tumor volume (n = 220)	per cm³	0.2753	1.005 (0.996; 1.013)
Grade of HI (n = 241)	2 vs. 1 or 0	0.0762	2.711 (0.900; 8.167)
Hydrocephalus (n = 222)	yes vs. no	0.0003	6.632 (2.399; 18.331)
Gross-total resection (n = 242)	yes vs. no	0.0819	2.130 (0.909; 4.993)
Grade of HL (n = 241)	2 vs. 1 or 0	0.0237	2.598 (1.136; 5.943)
	Multivariable logistic regression model for CI chosen by a stepwise selection algorithm
Hydrocephalus	yes vs. no	0.0001	7.652 (2.695; 21.722)
Gross-total resection	yes vs. no	0.0319	2.757 (1.092; 6.958)

HI, preoperative hypothalamic involvement; HL surgical hypothalamic lesion.