Table 3.

Multinomial logistic regression analysis of PFO diameter in relation to severe PVH and DWMH (Grade 3 vs. Grade 0) in young adults.

Variables	OR	95% CI		p-value
		Lower limit	Upper limit
PVH	1.522	1.228	1.885	<0.001
DWMH	1.857	1.555	2.219	<0.001

OR, odds ratio; CI, confidence interval; PVH, periventricular hyperintensity; DWMH, deep white matter hyperintensity; PFO, patent foramen ovale.

Grade 0 (no WMH) was used as the reference group. Data represent the risk of severe WMH (Grade 3) compared with no WMH (Grade 0). The multinomial logistic regression model was used because the proportional odds assumption was violated (parallel lines test, both p < 0.05). The model was adjusted for age and sex. Due to significant multicollinearity among PFO diameter, RLS grade, and RLS type (correlation coefficients r = 0.799–0.831, all p < 0.001), a forward stepwise selection approach (entry criterion p < 0.05, removal criterion p > 0.10) was applied; PFO diameter was the only variable retained in the final model.