Table 4.

Cerebral microbleeds and the risk of dementia

	Dementia		Alzheimer’s dementia
	n/N	HR (95% CI)	n/N	HR (95% CI)
	Model I
No microbleeds	39/3911	Reference	28/3911	Reference
Any microbleeds	33/930	2.02 (1.25–3.24)	25/930	2.10 (1.21–3.64)
Lobar*	21/648	1.81 (1.05–3.11)	17/648	2.00 (1.08–3.71)
Deep or infratentorial±	12/282	2.39 (1.23–4.61)	8/282	2.15 (0.97–4.78)
	Model II
No microbleeds	26/3088	Reference	18/3088	Reference
Any microbleeds	21/728	1.59 (0.88–2.89)	16/728	1.67 (0.83–3.36)
Lobar*	15/512	1.65 (0.86–3.17)	11/512	1.66 (0.77–3.59)
Deep or infratentorial±	6/216	1.40 (0.55–3.52)	5/216	1.58 (0.56–4.45)
	Model III
No microbleeds	36/3743	Reference	26/3743	Reference
Any microbleeds	28/868	1.73 (1.03–2.90)	21/868	1.83 (1.00–3.33)
Lobar*	17/603	1.55 (0.86–2.81)	14/603	1.70 (0.87–3.32)
Deep or infratentorial±	11/265	2.42 (1.18–4.96)	7/265	2.34 (0.98–5.63)

Values represent adjusted hazard ratios (HR) (95% confidence interval [CI]) for incident dementia in participants with microbleeds compared to those without microbleeds.

Model I: adjusted for age, sex, and education level.

Model II: as Model I, additionally adjusted for APOE ε4 carriership (versus ε3/ε3) and a propensity score of cardiovascular risk factors which included hypertension, total and HDL cholesterol, smoking status, diabetes mellitus, lipid-lowering and antithrombotic medication use.

Model III: as Model I, additionally for lacunes, intracranial volume, white matter lesion volume.

n/N= number of dementia cases/total number of participants per strata. Numbers differ for model I to III because missing values for APOE genotype in model II were not imputed, and because white matter lesion volume was calculated in 4,611 participants in model III.

^*Lobar microbleeds with or without cerebellar microbleeds.

^±Deep or infratentorial microbleeds with or without lobar microbleeds.