TABLE 2.

ORs (and 95% CIs) of cognitive impairment by quartiles of serum tHcy levels¹.

	Serum tHcy levelsa
	Quartile 1	Quartile 2	Quartile 3	Quartile 4	p-value for trend2
Model 1	1.00 (reference)	1.538 (1.005–2.354)	2.335 (1.542–3.534)	5.004 (3.267–7.665)	<0.001
Model 2	1.00 (reference)	1.539 (1.002–2.363)	2.174 (1.430–3.305)	4.851 (3.152–7.466)	<0.001
Model 3	1.00 (reference)	1.490 (0.953–2.329)	1.715 (1.105–2.662)	3.562 (2.260–5.616)	<0.001
Model 4	1.00 (reference)	1.486 (0.961–2.297)	1.962 (1.280–3.006)	4.254 (2.740–6.603)	<0.001
Model 5	1.00 (reference)	1.554 (1.003–2.408)	2.019 (1.319–3.089)	4.533 (2.925–7.024)	<0.001
Model 6	1.00 (reference)	1.438 (0.928–2.229)	1.893 (1.234–2.904)	4.140 (2.666–6.430)	<0.001
Model 7	1.00 (reference)	1.532 (0.997–2.356)	1.941 (1.268–2.971)	4.056 (2.604–6.319)	<0.001
Model 8	1.00 (reference)	1.539 (1.002–2.363)	2.175 (1.427–3.315)	4.855 (3.141–7.504)	<0.001
Model 9	1.00 (reference)	1.494 (0.967–2.310)	1.957 (1.279–2.994)	4.398 (2.837–6.817)	<0.001

¹ORs and 95% CIs were calculated with the use of the binary logistic regression model. ²Tests for the trend were conducted by treating the quartiles as a continuous variable and assigning the median for each quartile. ^aThe variable was transformed to the log scale. Model 1: Adjusted for age, sex, and education. Model 2: Model 1 plus additional adjustments for hypertension, history of stroke, HDL-C, and current smoking. Model 3: Model 2 plus additional adjustment for the total CSVD burden score. Model 4: Model 2 plus additional adjustment for pWMH (score = 3). Model 5: Model 2 plus additional adjustment for dWMH (score ≥ 2). Model 6: Model 2 plus additional adjustment for presence of lacune. Model 7: Model 2 plus additional adjustment for presence of deep CMB. Model 8: Model 2 plus additional adjustment for presence of lobar CMB. Model 9: Model 2 plus additional adjustment for BG-EPVS (score ≥ 2).