Table 2.

Detailed examination of the relationship between CD34+CD133+ (log-transformed) and the risk of AD/dementia risk

CD34+CD133+ cells plus covariates	AD		All-cause dementia
	HR (95% CI)	P value	HR (95% CI)	P value
Model 1: no covariates	0.57 (0.40, 0.81)	0.002*	0.57 (0.42, 0.77)	0.003*
	n = 1,617		n = 1,640
Model 2: age, sex, years of education	0.64 (0.45, 0.93)	0.02*	0.64 (0.47, 0.87)	0.005*
	n = 1,617		n = 1,640
Model 3: Model 2 + APOE ε4 + vascular diseases	0.64 (0.44, 0.94)	0.02*	0.63 (0.45, 0.87)	0.006*
	n = 1,340		n = 1,362
Model 3 after stratification
No vascular diseasesa	1.23 (0.15, 10.18)	0.85	1.12 (0.19, 6.56)	0.90
	n = 256		n = 257
Peripheral vascular diseasesb only	0.62 (0.40, 0.94)	0.02*	0.61 (0.43, 0.88)	0.008*
	n = 832		n = 850
Cerebrovascular diseasesc only	0.58 (0.35, 0.96)	0.03*	0.53 (0.35, 0.81)	0.003*
	n = 628		n = 641

Cox proportional hazards regression models were used to study the relationship between log-transformed CD34+CD133+ cell frequency (%) and the risk of AD or all-cause dementia after adjusting for the covariates. HR with 95% CI with P values is shown. Model 1: simple association without confounders; Model 2: adjusting for age, sex, and education; Model 3: Model 2 + APOE ε4 + vascular diseases. ^a Model 3 after the stratification as no vascular diseases for those with no CHD, no HTN, no stroke, no silent infarct, no CMB, and low level of WMHI; ^b Model 3 after the stratification as peripheral vascular diseases for those with CHD or HTN; ^c Model 3 after the stratification as cerebrovascular diseases for those with stroke, silent infarct, CMB, or high level of WMHI; * P value significant < 0.05