Skip to content

Statistics — vascular dementia and vascular cognitive impairment

Last updated: 2026-09-02 (second deepening pass; eleven sections and twelve conflict rows added, to 34 sections and 27 conflicts). Conflicting estimates are preserved side by side. Every row includes population, method, and a live-resolved PMID.

Global all-dementia context

Measure Estimate Year/population Method Source
people living with dementia 57.4m (50.4–65.1) global, 2019 GBD modeling PMID 34998485
forecast prevalence 152.8m (130.8–175.9) global, 2050 demographic/risk forecast PMID 34998485

These are not vascular-dementia-specific totals.

Pre- and post-stroke dementia

Measure Estimate (95% CI) Population Method Source
pre-stroke dementia 14.4% (12.0–16.8) hospital cohorts systematic review/meta-analysis PMID 19782001
pre-stroke dementia 9.1% (6.9–11.3) population cohorts same PMID 19782001
early post-stroke dementia 7.4% (4.8–10.0) first-ever stroke, prior dementia excluded same PMID 19782001
early post-stroke dementia 41.3% (29.6–53.1) recurrent stroke, prior dementia included same PMID 19782001
later cumulative incidence 3.0% (1.3–4.7) per year hospital cohorts after year 1 same PMID 19782001
MCI among recent lacunar stroke 47% SPS3 English-speaking baseline, n=1,636 neuropsychological z ≤ −1.5 PMID 23034910
WMH → incident stroke HR 3.3 (2.6–4.4) 22-study meta-analysis longitudinal MRI PMID 20660506
WMH → incident dementia HR 1.9 (1.3–2.8) same same PMID 20660506
WMH → death HR 2.0 (1.6–2.7) same same PMID 20660506
any post-stroke NCD 53.4% (46.9–59.8) 16 hospital studies/3,087 meta-analysis PMID 30504699
mild post-stroke NCD 36.4% (29.0–43.8) same meta-analysis PMID 30504699
major post-stroke NCD 16.5% (12.1–20.8) same meta-analysis PMID 30504699

Predictors

Predictor Effect Population/method Source
baseline cognitive impairment → later post-stroke dementia RR 3.10 (2.77–3.47) 89-study meta-analysis; 160,783 patients PMID 38101426
baseline cognitive impairment → later post-stroke cognitive impairment RR 2.00 (1.66–2.40) same PMID 38101426
diabetes → later post-stroke cognitive impairment RR 1.29 (1.14–1.45) same PMID 38101426
AF (presence/history) → later post-stroke cognitive impairment RR 1.29 (1.04–1.60) same PMID 38101426
moderate/severe WMH → later post-stroke cognitive impairment RR 1.51 (1.20–1.91) same PMID 38101426

Autopsy composition

Pathology category Share among dementia cases Population/method Source
Alzheimer pathology + infarcts 38.0% community autopsy cohort PMID 17568013
Alzheimer pathology alone 30.0% same PMID 17568013
vascular dementia alone 12% same PMID 17568013

Blood-pressure prevention

Outcome Effect (95% CI) Population/method Source
probable dementia, intensive vs standard HR 0.83 (0.67–1.04) SPRINT MIND RCT PMID 30688979
MCI, intensive vs standard HR 0.81 (0.69–0.95) SPRINT MIND RCT PMID 30688979
dementia/cognitive impairment OR 0.93 (0.88–0.98); ARR 0.39% 12 RCTs/92,135; mean 4.1y PMID 32427305
dementia OR 0.87 (0.75–0.99) 5-trial IPD/28,008 PMID 36282295
dementia RR 0.93 (0.84–1.02) 9 RCTs (1,041/29,029 vs 1,090/28,653) PMID 29927845
PROGRESS dementia RRR 12% (−8 to 28) 6,105 prior stroke/TIA; 3.9 y PMID 12742805
PROGRESS cognitive decline RRR 19% (4–32) same PMID 12742805
preDIVA dementia HR 0.92 (0.71–1.19) 3,526 adults 70–78; 6.7 y PMID 27474376
ASPREE dementia trigger HR 1.03 (0.91–1.17) 19,114; aspirin 100 mg; 4.7 y PMID 32213642
SPRINT MRI WMH growth −0.54 cm³ (−0.87 to −0.20) intensive vs standard n=449 paired MRI PMID 31408137
LACI-2 ISMN cognitive impairment aOR 0.55 (0.36–0.86) 363 lacunar stroke; 1 y PMID 37222252

Symptomatic cognitive drugs

Intervention Cognitive effect vs placebo Adverse-event effect Method Source
donepezil 5 mg ADAS-Cog MD −0.92 (−1.44 to −0.40) OR 1.22 (0.94–1.58) network meta-analysis PMID 33704781
donepezil 10 mg MD −2.21 (−3.07 to −1.35) OR 1.95 (1.20–3.15) same PMID 33704781
galantamine 16–24 mg MD −2.01 (−3.18 to −0.85) OR 1.57 (1.02–2.43) same PMID 33704781
rivastigmine 3–12 mg MD 0.03 (−3.04 to 3.10) OR 3.21 (0.36–28.88) same PMID 33704781
donepezil 5/10 mg −1.65/−2.09 ADAS-Cog withdrawal 10.1%/16.3% vs 8.8% 616-person RCT, 24 weeks PMID 12939421
galantamine ADAS-Cog −1.8 vs −0.3 discontinuation 13% vs 6% 788-person RCT, 26 weeks PMID 17664404

CAA diagnostic accuracy

Cohort Sensitivity Specificity Source
Boston v2.0 derivation 74.8% (65.4–82.7) 84.6% (71.9–93.1) PMID 35841910
temporal validation 92.5% (79.6–98.4) 89.5% (66.9–98.7) PMID 35841910
geographic validation 80.2% (70.8–87.6) 81.5% (61.9–93.7) PMID 35841910
autopsy standard 74.5% (65.4–82.4) 95.0% (83.1–99.4) PMID 35841910
Boston v2.0 probable CAA, non-haemorrhagic 28.6% (13.2–48.7) 65.3% (44.3–82.8) PMID 38710005
lecanemab ARIA-E 12.6% CLARITY-AD n=898 treated PMID 36449413
donanemab ARIA-E 24.0% (52 symptomatic) TRAILBLAZER-ALZ 2 n=860 treated PMID 37459141
CAA-ri clinical improvement with immunosuppression 94% vs 50% untreated (OR 16.0, 2.72–94.1) 48-person observational series PMID 32568365
atypical antipsychotic death in dementia RCTs OR 1.54 (1.06–2.23); 3.5% vs 2.3% 15 trials, 5,110 randomized PMID 16234500

Cognitive screening

Instrument Accuracy Population/method Source
full MoCA AUC 0.950 (0.868–0.988) 34 VaD patients, single validation PMID 22676901
short MoCA AUC 0.936 (0.849–0.981) same PMID 22676901
MMSE AUC 0.860 (0.754–0.932) same PMID 22676901
MoCA for VMCI AUC range 0.87–0.93 systematic review; heterogeneous PMID 31050033

Community incidence of vascular dementia (added 2026-08-31)

Measure Estimate (95% CI) Population/year Method Source
vascular-dementia incidence 1.5 per 1,000 person-years Rotterdam Study, 7,046 dementia-free adults ≥55, 15,135 person-years, 1990–1993 baseline three-stage screening, prospective cohort PMID 9521184
all-dementia incidence 10.7 per 1,000 person-years same same PMID 9521184
all-dementia incidence by age band 0.6 → 97.2 per 1,000 py across 5-year bands same same PMID 9521184
lifetime dementia risk from age 55 0.33 women / 0.16 men same Kaplan-Meier from incidence PMID 9521184
vascular-dementia incidence 1.4 (0.6–2.3) per 1,000 person-years NEDICES, central Spain, 3,891 followed a median 3.2 y European-standardized, population survey PMID 17727890
all-dementia incidence 10.6 (8.9–12.3) per 1,000 person-years same same PMID 17727890
VaD share of incident dementia 11.2% (18/161 cases; AD 71.4%) same same PMID 17727890
cumulative vascular-dementia risk to age 95 0.04 in both sexes EURODEM, 4 European cohorts, 528 incident cases, 28,768 person-years pooled Poisson/log-linear PMID 10599770
cumulative Alzheimer risk to age 95 (from 65) 0.22 women / 0.09 men same same PMID 10599770
AD incidence at age 90 81.7 (63.8–104.7) women vs 24.0 (10.3–55.6) men per 1,000 py same same PMID 10599770
vascular-dementia prevalence (primary or contributing) 4.2% (AD 5.4%; AD/VaD ratio 1.5) Honolulu-Asia Aging Study, 3,734 Japanese-American men aged 71–93 age-standardized prevalence 7.6% overall PMID 8805729
>1 possible cause of dementia 26% of cases same same PMID 8805729

Mortality and cost (added 2026-08-31)

Measure Estimate (95% CI) Population/method Source
any dementia vs no dementia, all-cause mortality HR 5.90 (3.53–9.86) 78 studies; 63,125 with dementia, 152,353 controls PMID 36097997
non-Alzheimer vs Alzheimer dementia, all-cause mortality HR 1.33 (1.21–1.46) same PMID 36097997
survival from diagnosis, Alzheimer disease mean 5.8 years (SD 2.0) same PMID 36097997
survival from diagnosis, non-Alzheimer vs Alzheimer −1.12 years (−1.52 to −0.72) same PMID 36097997
survival from onset, vascular dementia vs Alzheimer −1.27 years (−1.90 to −0.65) same; only VaD and DLB reached significance PMID 36097997
Lewy body dementia mortality (highest of all subtypes) HR 17.88 (5.87–54.46) same PMID 36097997
informal care hours, 24 brain health disorders 9.4 billion (7.8–11.3) hours, 2021 204-country modelling study PMID 41748237
forgone earnings from informal care US$1.7 trillion (1.5–2.1), 2021 same; stroke = most hours, dementia = most lost earnings PMID 41748237
annual growth in informal care hours since 2000 3.2% (2.9–3.5) same PMID 41748237
vascular dementia in AF with low conventional stroke risk HR 1.68 (1.33–2.12) vs matched non-AF controls 36,340 AF vs 117,298 controls, UK primary care, median 5 y PMID 38839900
all-cause dementia in the same cohort HR 1.17 (1.04–1.32); Alzheimer HR 0.85 (0.70–1.03) same PMID 38839900

Post-stroke cognitive trajectory and prediction (added 2026-08-31)

Measure Estimate (95% CI) Population/method Source
trajectory-group membership at ~3.6 months low −3.27 SD (17%), medium −1.23 SD (48%), high +0.71 SD (35%) STROKOG, 1,149 patients, 9 hospital cohorts, latent-class growth analysis PMID 37072222
change over year 1, high-performance group +0.22 SD/year (0.07–0.36) same; low and medium groups not significant PMID 37072222
large-artery vs small-vessel stroke → low-performance group RRR 2.77 (1.32–5.83) same PMID 37072222
diabetes → low-performance group RRR 3.78 (2.08–6.88) same PMID 37072222
global cognitive decline beyond 1 year post-stroke −0.053 SD/year (−0.073 to −0.033) STROKOG IPD, 1,488 patients, median 2.68 y PMID 34775838
excess decline vs stroke-free controls −0.078 SD/year (−0.11 to −0.045) same, subgroup analysis PMID 34775838
post-stroke dementia incidence 8.7 per 100 person-years (655/2,663 over median 2.0 y) STROKOG, 12 studies, 10 countries PMID 41525568
5-year post-stroke dementia model discrimination C 0.81 (0.75–0.87) development; 0.70 (0.67–0.73) internal-external validation Fine-Gray with death as competing risk PMID 41525568
CAIDE score applied to stroke cohorts AUC 0.53 4 harmonized cohorts, 12–18 months PMID 32568644
ANU-ADRI applied to stroke cohorts C 0.66 same PMID 32568644
Brief Dementia Screening Indicator in stroke cohorts C 0.61 same PMID 32568644
OCS-based 6-month PSCI model C 0.76 (0.71–0.80) internal; 0.74 (0.68–0.80) external 430 development + external OCS-Care validation PMID 41794047

Criteria performance against autopsy (added 2026-08-31)

Criteria Sensitivity Specificity Population Source
ADDTC possible VaD 0.63 0.64 113 autopsied demented patients PMID 9305324
NINDS–AIREN possible VaD 0.58 0.80 same PMID 9305324
Hachinski Ischemic Score 0.43 0.88 same PMID 9305324
ADDTC possible VaD 0.70 0.78 89 autopsied (20 VaD, 23 mixed, 46 AD) PMID 11772694
NINDS–AIREN possible VaD 0.55 0.84 same PMID 11772694
DSM-IV vascular dementia 0.50 0.84 same PMID 11772694
ADDTC probable VaD 0.25 0.91 same PMID 11772694
NINDS–AIREN probable VaD 0.20 0.93 same PMID 11772694
ICD-10 vascular dementia 0.20 0.94 same PMID 11772694
ADDTC / NINDS–AIREN / HIS possible VaD 0.56–0.58 (all three) 0.74 / 0.73 / 0.66 110 autopsies in the oldest-old, 36 confirmed VaD PMID 16580095
mixed dementia misclassified as VaD ADDTC 54%, NINDS–AIREN 29%, HIS 18% 113 autopsies PMID 9305324
pure VaD cases in the oldest-old with no stroke history 42% 110 autopsies PMID 16580095
interrater reliability, 4 criteria sets κ 0.30 (ADDTC) to 0.61 (original HIS) 25 case vignettes, 7 centres PMID 10681076

Neuropathologic composition, large series (added 2026-08-31)

Measure Estimate Population/method Source
participants with ≥1 neuropathology 94% (≥2: 78%; ≥3: 58%; ≥4: 35%) 1,079 autopsied with 2+ cognitive evaluations PMID 29244218
Alzheimer disease occurring in isolation 9% (present in 65%) same PMID 29244218
distinct neuropathologic combinations observed >230, each in <6% of cohort same PMID 29244218
person-specific share of cognitive loss due to AD mean ~50%; individual range 22–100% same PMID 29244218
decedents with any cerebrovascular pathology 80% (of these, 37% single, 63% mixed) 1,474 Rush decedents, 32 CVD combinations PMID 34601898
faster decline vs no CVD mixed CVD profiles only; single CVD profiles did not differ same PMID 34601898
participants with mixed pathology >80%; 280 unique copathology combinations 1,633 ROS/MAP decedents, mean age at death 90.4 PMID 41543852
empirical neuropathologic profiles infarcts+vessel disease 15.9%; LATE-NC+hippocampal sclerosis 12.3%; Lewy bodies 21.7%; ADNC+CAA 9.7%; low pathology 40.4% same, hierarchical clustering PMID 41543852

Small-vessel disease markers and progression (added 2026-08-31)

Measure Effect (95% CI) Population/method Source
baseline WMH volume → dementia HR 1.31 per SD (1.02–1.67) RUN DMC, 503 sporadic SVD, median 13.2 y PMID 37073486
DWI-positive lesions → dementia HR 2.03 (1.01–4.04) same PMID 37073486
PSMD → dementia HR 1.24 per SD (1.02–1.51) same PMID 37073486
WMH progression → dementia HR 1.76 per SD (1.18–2.63) same PMID 37073486
dementia incidence in sporadic SVD 108/498 (21.5%) over median 13.2 y same (AD 38, VaD 34, mixed 26) PMID 37073486
genetically determined WMH → Alzheimer disease OR 1.43 (1.10–1.86) two-sample MR, up to 75,024 AD cases PMID 38776079
genetically determined WMH → incident all-cause dementia HR 1.02 (1.00–1.04), null individual-level longitudinal, 10,699 cases PMID 38776079
systolic BP → lacunar stroke OR 1.41 (1.30–1.54) MR, GWAS n=757,601 exposure PMID 38818733
genetically proxied CCB → WMH volume β −0.287 (−0.408 to −0.165) same PMID 38818733
genetically proxied CCB → basal-ganglia PVS OR 0.85 (0.81–0.89) same PMID 38818733
enlarged perivascular spaces → MMSE 0.001 (−0.007 to 0.008), null UNIVRSE, 5 population cohorts, 3,575 people PMID 30068634
enlarged perivascular spaces → general cognitive factor 0.002 (−0.001 to 0.006), null same PMID 30068634
chronic cortical microinfarct prevalence 35% (24–49%) RUN DMC-Intense, 54 SVD patients, monthly 3T MRI PMID 32065609
acute cortical microinfarct cumulative incidence 13% (6–24%); all disappeared on follow-up MRI same, 472 DWI scans over median 39.5 weeks PMID 32065609
intensive vs standard BP → global SVD factor change Cohen's d −0.40 (−0.62 to −0.17) SPRINT MRI substudy, 442 with paired scans, median 3.9 y PMID 42614618
dose-response by attained SBP reduction 21.2% (7.4–35.0) / 26.3% (13.1–39.5) / 39.4% (24.2–54.5) less progression for 0–10 / 10–20 / ≥20 mm Hg same PMID 42614618

Biomarker validation (added 2026-08-31)

Marker Result Population/method Source
PSMD → general cognition β −0.8 (−1.2 to −0.4); explains cognition beyond WMH MarkVCID-1 n=396, replicated in CHARGE (6,172), Rush (287), UC Davis (567) PMID 39569745
MRI free water → executive composite associated cross-sectionally; predicted accelerated decline (P=0.0026) over ~1.29 y MarkVCID sub-cohorts + 5 legacy cohorts PMID 36523847
cerebrovascular reactivity → MoCA and executive IRT positive at all 3 sites, prespecified hypothesis MarkVCID, n=263, 5% CO₂ PMID 38951718
WMH growth/regression protocol reliability test-retest ICC 0.969 (growth) / 0.937 (regression); cross-site ICC 0.995 / 0.990 MarkVCID multi-site PMID 37840499
WMH growth → memory associated (P<0.028); WMH regression not associated with any outcome same PMID 37840499
plasma NfL strongest association with SVD burden (WMH, MSMD, fiber density) 76 memory-clinic patients; 41 ADNI validation PMID 42252508
plasma p-tau217 strongest association with AD-related neurodegeneration same PMID 42252508
plasma GFAP associated with both SVD and AD markers same; WMH association replicated in MESA PMID 42252508, 41630612
plasma Aβ42/40 lower ratio associated with amyloid PET and with microbleeds MESA, n=251 PMID 41630612

Haemorrhage risk and anticoagulation (added 2026-08-31)

Measure Estimate (95% CI) Population/method Source
symptomatic ICH on anticoagulation, microbleeds present 9.8 per 1,000 patient-years (4.0–20.3) CROMIS-2, 1,447 AF + recent stroke/TIA, 3,366 patient-years PMID 29778365
symptomatic ICH, no microbleeds 2.6 per 1,000 patient-years (1.1–5.4); adjusted HR 3.67 (1.27–10.60) same PMID 29778365
HAS-BLED discrimination for symptomatic ICH C 0.41 (0.29–0.53) same PMID 29778365
HAS-BLED + microbleeds C 0.66 (0.53–0.80) same PMID 29778365
HAS-BLED + microbleeds + diabetes + anticoagulant type C 0.74 (0.60–0.88) same PMID 29778365
>10 basal-ganglia perivascular spaces → symptomatic ICH HR 8.96 (2.41–33.4) on 14 events CROMIS-2 post hoc, 1,386 participants, mean 2.34 y PMID 32934168
starting vs avoiding anticoagulation after intracranial haemorrhage: recurrence 8/101 (8%) vs 4/102 (4%); adjusted HR 2.42 (0.72–8.09) SoSTART RCT, 67 UK hospitals PMID 34487722
pooled: any stroke or cardiovascular death HR 0.68 (0.42–1.10), I²=0% COCROACH IPD meta-analysis, 412 participants, 4 trials PMID 37839434
pooled: ischaemic major adverse cardiovascular events 9/212 (4%) vs 38/200 (19%); HR 0.27 (0.13–0.56) same PMID 37839434
pooled: haemorrhagic major adverse cardiovascular events 15/212 (7%) vs 9/200 (5%); HR 1.80 (0.77–4.21) same PMID 37839434
CAA-related ICH recurrence 23% (18–28%), I²=96.7% 30-study meta-analysis PMID 38020625
disseminated cortical superficial siderosis → recurrence OR 3.21 (2.25–4.58) same PMID 38020625
cortical superficial siderosis → recurrent ICH (within RCT) HR 7.7 (1.4–42.2) on 13 events PRESTIGE-AF imaging subanalysis, n=313, median 1.4 y PMID 41197104
probable CAA category → recurrent ICH not associated (P>0.2) same PMID 41197104
Edinburgh CT-only criteria, 5-year recurrence 12% low / 16% intermediate (aSHR 1.68, 1.21–2.32) / 26% high risk (aSHR 2.97, 1.50–5.89) IPD meta-analysis, 1,620 patients, 8 cohorts PMID 40975099
LAAO in CAA-related ICH with AF: ischaemic stroke 5.16 per 100 patient-years (1.36–17.48) 102 pooled participants, median 9.4 months PMID 39694822
LAAO in CAA-related ICH with AF: ICH 2.73 per 100 patient-years (0.41–13.94) same PMID 39694822

Prevention trials, expanded (added 2026-08-31)

Trial / synthesis Effect (95% CI) Population/method Source
Syst-Eur, double-blind phase dementia 7.7 → 3.8 per 1,000 py (21 vs 11 cases, P=0.05); 50% reduction 2,418 adults ≥60, isolated systolic hypertension, median 2.0 y, nitrendipine-based PMID 9802273
Syst-Eur, open extension 7.4 → 3.3 per 1,000 py (43 vs 21 cases, P<0.001); adjusted HR 0.38 (0.23–0.64) median 3.9 y; 20 dementias prevented per 1,000 treated 5 y (7–33) PMID 12374512
Cochrane antihypertensives, incident dementia OR 0.89 (0.72–1.09), very low certainty 4 placebo-controlled trials without prior cerebrovascular disease, 236/7,767 vs 259/7,660 PMID 34028812
Cochrane antihypertensives, MMSE MD 0.20 (0.10–0.29), very low certainty 5 placebo-controlled trials; contrast −9.25/−2.47 mm Hg PMID 34028812
antihypertensives, cognitive-function score SMD 0.06 (0.03–0.09), GRADE moderate 4 RCTs (Syst-Eur, PROGRESS, SCOPE, HYVET-COG), 16,823 participants — not the same four-trial set as Cunningham (PROGRESS included) PMID 41962914
antihypertensives, dementia incidence OR 0.89 (0.76–1.05), GRADE low same four-trial set as the Ding SMD row, not Cunningham PMID 41962914
drug-class ranking (exploratory, 1 study per node) CCB highest for dementia prevention (SUCRA 94.7%); diuretic ± ACEi highest for cognitive improvement (SUCRA 95.3%) same PMID 41962914
lipid-lowering therapy, incident cognitive impairment or dementia OR 0.96 (0.74–1.26); 1.33% vs 1.36% over mean 34.5 months 15 RCTs, 139,169 participants PMID 40794911
statins specifically OR 0.90 (0.67–1.21) same PMID 40794911
rhythm vs rate control, dementia HR 0.74 (0.62–0.89), I²=62% 14 studies, 193,830 AF patients, mostly observational PMID 38369630
AF ablation → vascular dementia HR 0.58 (0.42–0.80), I²=31% same PMID 38369630
US POINTER, structured vs self-guided 0.243 vs 0.213 SD/y; difference 0.029 SD/y (0.008–0.050), P=0.008 2,111 randomized, 2 y, 89% completed PMID 40720610
LatAm-FINGERS, systematic vs flexible 0.31 vs 0.20 SD/y; difference 0.11 SD/y (0.06–0.15), P<0.0001 1,065 across 11 Latin American countries, 2 y PMID 42442374
exercise and/or intensive vascular risk reduction, PACC exercise 0.1 (−0.1 to 0.2), P=0.37; IRVR −0.1 (−0.3 to 0.03), P=0.12 (abstract prints unsigned 0.1; arm means +0.2 vs +0.3 imply a negative contrast); interaction P=0.13 513 randomized 2×2 factorial, 24 months PMID 41870419
ACHIEVE hearing intervention, 3-year cognition difference 0.002 SD (−0.077 to 0.081), P=0.96 977 adults 70–84; prespecified cohort interaction P=0.010 PMID 37478886

Drug trials, expanded (added 2026-08-31)

Intervention Effect (95% CI) Population/method Source
memantine 20 mg, MMM 300 ADAS-Cog difference 2.0 points (0.49–3.60); CIBIC-plus 60% vs 52%, P=0.227 321 randomized / 288 ITT, France, 28 weeks PMID 12105362
memantine 20 mg, MMM500 ADAS-Cog −1.75 (−3.02 to −0.49); CGI-C no difference 579 randomized / 548 ITT, 54 UK centres, 28 weeks PMID 12409683
memantine, pooled ADAS-Cog/11 MD −2.20 (−3.24 to −1.15), moderate certainty; GBS −1.93 (−4.69 to 0.84) 2 trials, n=752 / n=595 PMID 42635820
galantamine, pooled ADAS-Cog/11 −2.01 (−3.18 to −0.85); ADAS-Cog/13 −2.51 (−3.87 to −1.15); NPI −0.13 (−4.05 to 3.79) 2 trials, n≈1,330 PMID 42635820
EGb 761, pooled SKT −2.65 (−5.17 to −0.12), I²=92.8%, very low certainty 3 trials, n=283 PMID 42635820
VaD placebo-controlled trials available 16 trials, 5,668 participants; all but one published ≤2012 systematic search 1990–2025 PMID 42635820
donepezil 5 mg (SMD framing) SMD −1.11 (−1.88 to −0.34) 11-RCT network meta-analysis PMID 35048806
donepezil 10 mg, CDR-SB / EXIT25 −0.25 (−0.44 to −0.06) / −1.47 (−2.79 to −0.15) same PMID 35048806
Cerebrolysin, cognition SMD 0.36 (0.13–0.58), very low quality Cochrane, 3 trials, 420 people PMID 31710397
Cerebrolysin, global response RR 2.69 (1.82–3.98), very low quality Cochrane, 2 trials, 379 people PMID 31710397
Cerebrolysin RCT ADAS-Cog+ 10.6 vs 4.4 points (LS mean difference −6.17, P<0.0001); combined responders 67.5% vs 27.0% 242 VaD participants, 24 weeks PMID 20656516
Actovegin (ARTEMIDA) ADAS-Cog+ change −6.8 vs −4.6; difference −2.3 (−3.9 to −0.7), P=0.005 at 6 months 503 randomized within 1 week of ischaemic stroke, MoCA ≤25 PMID 28432265
butylphthalide (NBP) ADAS-Cog −2.46 vs −1.39, P=0.03; CIBIC-plus 57.1% vs 42.1%, P=0.01 281 subcortical VCIND, 15 Chinese centres, 24 weeks PMID 26086183
butylphthalide, Cochrane re-analysis adjusted MD −1.07 (−2.02 to −0.12) ADAS-Cog-12, very low certainty same trial re-assessed PMID 35833913
Ginkgo in dementia, SKT MD −1.86 (−3.48 to −0.24), I²=96%, low certainty Cochrane, 9 studies, 2,801 participants PMID 41641880
Ginkgo in dementia, ADL International Scale MD −0.19 (−0.35 to −0.03), I²=91% Cochrane, 8 studies, 2,571 PMID 41641880
Ginkgo in MCI, ADAS-Cog MD −0.07 (−0.67 to 0.51), moderate certainty (no effect) Cochrane, 2 studies, 508 PMID 41641880
Ginkgo serious adverse events in MCI RR 0.95 (0.82–1.09), high certainty Cochrane, 3 studies, 714 PMID 41641880
CONIVaD combination all endpoints null; nimodipine adherence >75% dose in only 15% 62 randomized, 12 months PMID 33855653
COMCID cilostazol MMSE change −1.8 vs −1.3 at 96 weeks, null 159 with MCI, MMSE 22–28, CDR 0.5 PMID 38048134
cilostazol, recurrent ischaemic stroke OR 0.68 (0.57–0.81) 17 trials, 10,225 participants, mostly Asia-Pacific PMID 32646330
cilostazol, haemorrhagic stroke OR 0.43 (0.29–0.64) 16 trials, 9,736 PMID 32646330
LACI-2 at 6 months, ISMN composite adjusted OR 0.74 (0.55–0.99) 363 lacunar stroke PMID 42535538
LACI-2 at 6 months, combination cognition adjusted common OR 0.40 (0.21–0.78) same PMID 42535538
OxHARP sildenafil, pulsatility (primary) 0.02 (−0.01 to 0.05), P=0.18, null 3-way crossover, 65 with valid data PMID 38832504
OxHARP sildenafil, cerebrovascular reactivity 0.83 cm/s per mm Hg (0.23–1.42), P=0.007 same PMID 38832504
MINERVA minocycline, microglial signal / BBB permeability RR 1.01 (0.98–1.04) / RR 0.97 (0.91–1.03), both null 44 SVD patients, 3 months PMID 38629936

Imaging automation (added 2026-08-31)

Measure Result Population/method Source
automated WMH segmentation, SE-UNet Dice 0.675 (0.666–0.685) internal; 0.722 (0.719–0.726) external trained on 2,408 scans, validated on 6,013; 8,421 acute ischaemic stroke patients PMID 39013565
human–human Dice reliability (benchmark) 0.744 (0.738–0.751) same PMID 39013565
automated vs manual WMH volume correlation r = 0.933 (SE-UNet); CCC 0.841–0.956 externally same PMID 39013565
cases below uncertainty index 0.35 86% of external cases, mean Dice 0.744 same PMID 39013565
ML models on SVD markers, controls vs Alzheimer dementia AUC 0.88 (0.85–0.92) meta-analysis of 16 of 75 eligible studies PMID 40775365
ML models on SVD markers, controls vs cognitive impairment AUC 0.84 (0.74–0.95) same PMID 40775365
studies testing external generalisability 5 of 75 same PMID 40775365

Monogenic disease frequency (added 2026-08-31)

Measure Estimate (95% CI) Population/method Source
genetically confirmed clinical CADASIL prevalence 1.98 per 100,000 adults (1.24–3.00) west of Scotland register, 22 cases in 7 pedigrees PMID 15834040
predicted NOTCH3 mutation carrier prevalence 4.14 per 100,000 adults (3.04–5.53) same, adding 37 predicted carriers PMID 15834040
CADASIL prevalence, later Glasgow series 4.6 per 100,000 adults; mutation prevalence 10.7 per 100,000 49 pedigrees, 21 NOTCH3 mutations (61% exon 4) PMID 24840674
median age at first stroke, women 57 years same PMID 24840674
median age at first stroke, men by diagnosis era 46 y (pre-2006) → 56 y (post-2006), P=0.034 same PMID 24840674
genetic CADASIL prevalence 1 in 277 individuals 129,933 healthy Korean adults, NGS of 3 NOTCH3 variants PMID 40982447
NOTCH3 allele frequencies (Korean) p.Arg544Cys 0.07%, p.Arg640Cys 0.07%, p.Arg75Pro 0.04% same PMID 40982447
high-risk EGFr domains → stroke in community carriers OR 10.81 (5.46–21.37) 2,574 CADASIL patients + 1,647 population; validated in 434 + 1,003 PMID 36535904
medium-risk EGFr domains → stroke OR 1.81 (0.84–3.88) same PMID 36535904
risk metric vs NOTCH3 aggregation load / signalling activity P=0.006 / P=0.88 same PMID 36535904

Cognitive screening across populations (added 2026-08-31)

Measure Result Population/method Source
optimal MoCA cutoff in less-WEIRD populations lower than in WEIRD populations systematic review: 28 MMSE, 39 MoCA, 5 OCS accuracy studies, 17 countries PMID 37480233
optimal MMSE and OCS subtest cutoffs similar across WEIRD and less-WEIRD populations same PMID 37480233
accuracy studies in South American, African, non-Chinese Asian stroke populations none identified same PMID 37480233
OCS normative sample / acute stroke characterization 140 healthy participants / 208 acute stroke patients within 3 weeks development and validation study PMID 25730165
OCS memory and receptive-communication subtests impairment rates lower than expected → relatively insensitive independent evaluation, 316 acute-stroke-unit patients PMID 37186035

Patient-reported burden (added 2026-08-31)

Measure Estimate (95% CI) Population/method Source
stroke survivors with any unmet need ≥12 months post-stroke 84% (median 4 of 20 domains) 765 community-dwelling Australian survivors PMID 25042019
greater disability → unmet need, work domain aOR 7.0 (3.0–17.0) same PMID 25042019
memory problems → unmet need, leisure domain aOR 2.1 (1.0–4.2) same PMID 25042019
being 1–2 years post-stroke → unmet need, work domain aOR 3.4 (1.5–7.8) same PMID 25042019
CADA-PRO internal consistency / test-retest Cronbach α 0.95 / ICC 0.88 developed in 44, validated in 89 (43 CADASIL, 46 other SVD) PMID 39234671
CADA-PRO correlations in non-CADASIL SVD only with HADS and Starkstein Apathy Scale same PMID 39234671
psychoeducation studies covering non-language cognitive domains very few of 30 included studies scoping review, 1996–2023, 9 high-income countries PMID 39819901

Vascular-dementia-specific prevalence and incidence, by country (added 2026-09-02)

Measure Estimate (95% CI) Population / year Method Source
vascular dementia prevalence 1.6% (1.5–1.7); 3.92m cases (3.64–4.22) China, ≥60 y, 2015–2018, n=46,011 national multistage stratified cluster survey with in-person testing PMID 33271079
vascular dementia prevalence 0.96% (0.63–2.1) China, ≥18 y, 26 studies, 100,923 subjects, 1999–2019 meta-analysis including Chinese-language databases PMID 34178760
vascular cognitive impairment prevalence 1.54% (1.14–1.93); 2.91% at ≥80 y China, 81 studies, 784,846 participants, 1980–2023 random-effects meta-analysis, I²>90% PMID 40889799
vascular cognitive impairment incidence 0.29 per 100 person-years (0.21–0.41) China, 10 incidence studies same PMID 40889799
vascular dementia prevalence 116 per 10,000 (86–157) community, ≥50 y, 47 studies meta-analysis by region, age, sex PMID 31884487
vascular dementia prevalence, sex ratio at 60–69 y 56 per 10,000 men vs 32 per 10,000 women (1.8×) community same PMID 31884487
vascular dementia incidence 9.5 / 1,000 py (AD 14.6; combined 3.8; DLB 1.4) Hisayama, Japan, 828 adults ≥65, 17 y, 91.2% morphologically evaluated prospective community cohort with autopsy PMID 18977814
10-year survival, dementia vs matched controls 13.6% vs 29.3%; HR 1.67 (1.31–2.13) Hisayama, ages 65–89 same; no significant difference between subtypes PMID 18977814
"pure" vascular dementia share of demented autopsies 12.3% overall; 15.0% at age 60 falling to 8.7% at 90+ 1,700 consecutive Vienna autopsies, mean age 84.3 hospital autopsy series, consensus criteria PMID 21504129

Do not average the three Chinese estimates. A pooled VCI prevalence (1.54%) that is no higher than a national VaD prevalence (1.6%) is internally inconsistent; the denominators (≥18 vs ≥60), the criteria, and the ascertainment differ. The Hisayama incidence is roughly six times the Rotterdam and NEDICES figures in the community-incidence section above, despite stronger (91% morphological) case verification.

Cost of vascular dementia (added 2026-09-02)

Measure Estimate Population / year Method Source
annual direct healthcare cost, vascular dementia US$14,387 100,000-member US Medicare HMO, 1999–2002 administrative claims PMID 16155348
annual direct cost, other dementias / AD / CVD-no-dementia / controls US$10,716 / $7,839 / $8,254 / $5,494 same same (all P<0.0001 vs VaD) PMID 16155348
annual direct cost, vascular dementia US$5,112 (AD $4,625; FTD $4,924; between-group P>0.05) Argentina, 104 patients, 2002–2008 matched cost study PMID 21044400
hospitalization cost, VaD vs AD significantly higher for VaD (P<0.001) Argentina, same post-hoc comparison PMID 21044400
societal cost, VaD vs AD 23% higher for VaD (P=0.02) rural Nordanstig cohort, Kungsholmen project, Sweden Resource Utilization in Dementia instrument, societal perspective PMID 16191249
informal care, AD vs PD 55.73 h/week and $17,492/y vs 15.8 h/week and $3,284/y; no data for other dementias 21-study review systematic review PMID 23509789

The hospital-heavy, medication-light cost profile replicates from the US to Argentina, which argues it is a property of the disease's care pathway rather than of one payment system. No contemporary VaD-specific societal cost estimate exists.

Population attributable fractions (added 2026-09-02)

Measure Estimate (95% CI) Population Method Source
dementia by age 80 attributable to ≥1 of hypertension, diabetes, smoking measured at 45–54 y 21.8% (14.3–29.3) ARIC, n=7,731, 33 y follow-up population attributable fraction PMID 40455489
same, measured at 55–64 y 26.4% (19.1–33.6) ARIC, n=12,274 same PMID 40455489
same, measured at 65–74 y 44.0% (30.9–57.2) ARIC, n=6,787 same PMID 40455489
dementia occurring after age 80 attributable to those factors 2–8% ARIC same PMID 40455489
combined PAF, 12 Lancet Commission modifiable factors 45.4% (2011) → 52.5% (2018), change not significant CHARLS, China, 75,214 person-waves time-series PAF PMID 38872868
largest single contributor in China low education, mean individual weighted PAF 11.3% same same PMID 38872868
midlife diabetes → 25-y incident dementia HR 1.77 (1.53–2.04) ARIC, 15,744 adults aged 44–66, 1,516 cases Cox regression, fully adjusted PMID 28783817
midlife hypertension / prehypertension / smoking HR 1.39 (1.22–1.59) / 1.31 (1.14–1.51) / 1.41 (1.23–1.61) same same PMID 28783817
APOE ε4 (reference comparison) HR 1.98 (1.78–2.21) same same PMID 28783817

Blood-pressure prevention, randomized (added 2026-09-02)

Trial Contrast achieved Dementia result Source
CRHCP cluster-randomized (33,995 adults, 326 villages, rural China, 48 months) 22.0 mm Hg systolic (20.6–23.4); 9.3 mm Hg diastolic (8.7–10.0) all-cause dementia RR 0.85 (0.76–0.95), P=0.0035; serious adverse events RR 0.94 (0.91–0.98) PMID 40258956
HYVET-COG (3,336 adults ≥80, mean 2.2 y, placebo-controlled) 15 / 5.9 mm Hg 38 vs 33 per 1,000 patient-years; HR 0.86 (0.67–1.09); pooled with other placebo trials HR 0.87 (0.76–1.00, P=0.045) PMID 18614402
Syst-Eur double-blind phase (2,418 adults ≥60, median 2.0 y) 8.3 / 3.8 mm Hg 7.7 → 3.8 per 1,000 patient-years (21 vs 11 events, P=0.05), 50% reduction PMID 9802273
Syst-Eur open extension (median 3.9 y) 7.4 → 3.3 per 1,000 patient-years, 55%; HR 0.38 (0.23–0.64); 20 dementias prevented per 1,000 treated 5 years (7–33) PMID 12374512

The dose–response across these four rows is the strongest argument that the pooled nulls reflect insufficient achieved contrast rather than absence of effect. None of the trials adjudicated vascular dementia separately.

Exercise and non-drug intervention in vascular cognitive impairment (added 2026-09-02)

Intervention Population Result Source
progressive aerobic training, 6 months thrice weekly vs usual care + education 70 adults, mild subcortical ischaemic VCI, mean age 74 ADAS-Cog −1.71 (−3.15 to −0.26, P=0.02); not sustained at 12 months (−0.63, −2.34 to 1.07); 6-min walk +30.35 m (5.82–54.86); diastolic BP −6.89 mm Hg (−12.52 to −1.26) PMID 27760869
progressive resistance training, 12 months vs balance-and-tone active control 91 adults, SVD with MCI; 76 completed ADAS-Cog-Plus −0.18 (−0.35 to −0.01, P=0.04); females −0.27 (−0.49 to −0.05, P=0.02), males null; CRP −2.93 (−5.36 to −0.49) PMID 41795685
citicoline 1 g/day for 12 months vs no citicoline (open label) 347 first-ever ischaemic stroke, mean age 67.2 attention–executive OR 1.721 (1.065–2.781) at 6 months, 2.379 (1.269–4.462) at 12 months; mRS ≤2 57.3% vs 48.7% (P=0.186) PMID 23406981
donepezil 24-week trial (study 307) 603 NINDS-AIREN probable/possible VaD ADAS-Cog effect size −1.90 (5 mg, P=0.001), −2.33 (10 mg, P<0.001); ADFACS −1.31 both doses (P=0.02); AE withdrawal 11.1/11.1/21.8% PMID 12970516

The aerobic-exercise ADAS-Cog effect exceeds donepezil 5 mg (−0.92) and matches galantamine (−2.01) from the Cochrane pooling; no trial has compared them directly.

Behavioural and psychological symptoms by VCI subtype (added 2026-09-02)

Subtype Apathy Depression Irritability Other
unspecified VCI 54.29% 43.48% 38.76%
subcortical VCI 62.01% 52.11% 44.73%
mixed dementia 61.65% 45.68% sleep disturbance 44.63%; hallucinations 26.64%
VCI–no dementia 44.97% 32.75% anxiety 30.07%

Pooled from 35 NPI-based studies, n=5,805 (PMID 41043167). These figures conflict in direction with the registry-versus-clinic comparison of agitation recorded in the conflicts table below.

Criteria performance, additional (added 2026-09-02)

Measure Estimate (95% CI) Population Source
Mayo Clinic criteria (temporal link or bilateral infarction), pure VaD sensitivity 0.75, specificity 0.81; LR+ 3.9 (2.2–6.7) Rochester Epidemiology Project, 89 autopsies of 419 incident dementias PMID 12707071
pathological composition, same series AD 51%, pure VaD 13%, AD+VaD 12%, other 24% same PMID 12707071
interrater agreement, vascular dementia by ICD-10 κ 0.79 (0.70–0.87) meta-analysis, 22 studies of 7,577 screened PMID 33942363
interrater agreement, all-cause dementia by DSM-III-R κ 0.66 (0.53–0.78) same PMID 33942363
interrater agreement, Alzheimer disease by NINCDS-ADRDA κ 0.71 (0.65–0.77) same PMID 33942363
2-year progression of clinic CIND to dementia 34% overall; VCI-ND 40.0%, pre-AD 41.0% Canadian Cohort Study of Cognitive Impairment, n=146 PMID 16966831
2-year reversion of clinic CIND to normal 14% overall; psychiatric CIND 20%, CIND-NOS 30% same PMID 16966831
MCI prevalence by age (AAN guideline) 6.7% (60–64), 8.4% (65–69), 10.1% (70–74), 14.8% (75–79), 25.2% (80–84) systematic review PMID 29282327
2-year cumulative dementia incidence in MCI >65 14.9% same PMID 29282327

Small-vessel disease markers, additional (added 2026-09-02)

Measure Estimate (95% CI) Population Source
multimarker SVD score ≥2 → incident all-cause dementia HR 1.67 (1.05–2.66) adjusted for FSRP; 1.76 (1.10–2.81) adjusted for individual risk factors; Harrell c 0.82–0.83 Framingham, 1,152 MRIs, median 7.4 y PMID 40953349
severe perivascular spaces in both regions → incident dementia over 8 y OR 2.91 (1.43–5.95, P=0.003) 414 community adults aged 72–92, 3T MRI PMID 33504642
DTI-ALPS → incident dementia in established SVD HR 0.328 (0.183–0.588, P<0.001); PVS volume predicted nothing 120 lacunar stroke with confluent WMH, 3 y imaging / 5 y cognition PMID 38465597
microinfarcts at autopsy VaD 62%, AD 43%, mixed 33%, non-demented 24% 32 neuropathological studies, 10,515 people PMID 22234334
WMH GWAS loci 31 loci (21 novel) across WMH (n=42,310), FA (n=17,663), MD (n=17,467); 66 TWAS genes UK Biobank and others PMID 32358547
cognitive reserve moderation education and occupation predicted slower 3-y decline independently of WMH volume; education attenuated the WMH–7-y cognition relationship LADIS, n=615 PMID 27951523

Imaging measurement reliability (added 2026-09-02)

Measure Estimate Population Source
interrater κ, baseline visual WMH rating (Manolio, Fazekas–Schmidt, Scheltens) 0.59–0.78; Fazekas–Schmidt better than Manolio (P=0.003) 74 scans, 5 European centres, 3 raters PMID 12574557
interrater κ, WMH progression on follow-up scans 0.19–0.39 same PMID 12574557
agreement of visual scales with volumetry Kendall W 0.37–0.57 (all P<0.001); between-scale Spearman 0.712–0.806 255 Austrian Stroke Prevention Study subjects PMID 12574557
MMSE + visual medial-temporal-atrophy rating sensitivity 95% (AD), 85% (other dementias); specificity 96%; volumetry added nothing over MMSE 143 memory-clinic patients PMID 11032615
lenticulostriate pulsatility index, hypertensive vs normotensive higher PI and lower damping factor (both P=0.015); no difference in middle cerebral artery; unchanged by SVD-score adjustment 28 hypertensive vs 25 matched controls, 7T phase-contrast PMID 36722349
grey-matter ASL spatial coefficient of variation predicted 3-y memory decline, WMH progression, incident microbleeds and incident vascular events (29/368, 7.8%) 368 memory-clinic patients PMID 35871110

Post-stroke covert injury and delirium (added 2026-09-02)

Measure Estimate (95% CI) Population Source
new ischaemic lesions at 6 months after stroke 15.5% of 503 patients; 72% DWI-positive, 91% clinically covert prospective multicentre, 36-month follow-up PMID 39417418
those lesions → global cognition / mRS / recurrent stroke β −0.31 (−0.48 to −0.14) / β 0.36 (0.14–0.58) / HR 3.81 (1.35–10.69) same PMID 39417418
new brain infarct on 2-year MRI in anticoagulated AF 5.5% (68/1,227); 85.3% clinically silent; 88.2% anticoagulated at baseline Swiss-AF, mean age 71, 89.9% anticoagulated PMID 35171989
clinical stroke/TIA over the same 2 years 2.3% (28/1,227) same PMID 35171989
hospitalisation with delirium → 5-y dementia HR 2.64 (1.47–4.74) with WMD; 3.41 (1.91–6.09) without WMD OXVASC, 1,369 TIA/minor stroke, 209 dementias PMID 38310893
hospitalisation with infection → 5-y dementia HR 1.75 (1.04–2.94) only in moderate/severe white-matter disease same PMID 38310893
hospitalisation without delirium or infection HR 1.01 (0.86–1.20) same PMID 38310893
cumulative dementia after spontaneous ICH 32.0–37.4% at 5 years review of post-ICH cohorts PMID 38640161
dementia hazard by stroke subtype (5 y / 10 y) ICH 2.14 / 1.61; TIA 1.92 / 1.61; ischaemic 1.81 / 1.49 8,236 Taiwanese stroke patients, 1:1 propensity-matched PMID 31267646

Mechanism and biomarker quantities (added 2026-09-02)

Measure Estimate Population Source
plasma NfL, moderate-severe vs no SVD burden 45.2 ± 16.0 vs 34.3 ± 15.1 pg/mL; OR 1.71 (1.24–2.35) 496 non-demented ADNI participants PMID 33517704
rate of NfL change → SVD progression / WMH / lacunes OR 1.38 (1.08–1.76) / 1.41 (1.10–1.79) / 1.99 (1.42–2.77) same, 387 with longitudinal measures PMID 33517704
baseline plasma GFAP → SVD markers 9 years later WMH β 0.06 (0.01–0.10); FA β 0.08 (0.03–0.13); MD β 0.14 (0.09–0.18); NfL null for all three 5,270 UK Biobank adults aged 40–60 PMID 41461058
CAA prevalence in a population autopsy series 44.1% with neocortical CAA 211 Japanese-American men, HAAS PMID 12058090
CASI deficit, AD alone vs AD+CAA (vs non-demented no CAA) 16.6% lower vs 45.9% lower same, adjusted for plaques, tangles, infarcts, haemorrhage, APOE PMID 12058090
MCI prevalence in imaging-diagnosed CAA 79% (27/34); executive z −1.14 ± 1.07, processing speed −1.06 ± 1.12, memory −0.44 ± 1.03 prospective CAA cohort PMID 27338926
4-year ICH risk by cortical superficial siderosis 25% none / 28.9% focal / 74% (44.1–95.7) disseminated (log-rank P=0.0031) 118 Boston-criteria CAA patients, median 24 months PMID 24107862
6-month recurrent lobar ICH 7% (21/292); disseminated siderosis HR 3.92 (1.38–11.17); acute cSAH on CT HR 3.48 (1.13–10.73) 292 consecutive CAA-related lobar ICH survivors ≥55 PMID 27694268

Neuropathology variance explained (added 2026-09-02)

Measure Estimate Population Source
variance in late-life cognitive decline explained by 5 pathologies together 41% (tangles 34%, global AD 22%, Lewy bodies 8%, amyloid 6%, macroinfarcts 2%) 856 ROS/MAP decedents, mean 7.5 annual evaluations PMID 23798485
variance explained by 11 pathologies together 43%; AD indices 30–36%, non-AD neurodegenerative 4–10%, cerebrovascular 3–8% 1,164 ROS/MAP decedents, up to 24 evaluations PMID 33742668
variance in onset of terminal decline / preterminal / terminal rates 28% / 32% / 19% same PMID 33742668
copathology frequency, Chinese community brain bank LATE-NC 341 > ADNC 331 > PART 231 > CAA 183 > ARTAG 144 > α-synucleinopathy 124 > AGD 107 > hippocampal sclerosis 46, of 610 National Human Brain Bank, China PMID 39901730
APOE ε4 allele frequency, same cohort 13.63% same PMID 39901730
sex difference in mixed pathology women more likely to have AD + cerebrovascular pathology; men more likely pure Lewy body disease (adjusted for age, education, race, APOE ε4) >1,500 community-dwelling decedents PMID 31128096

Known conflicts and caveats

Conflict Interpretation
7.4% vs 41.3% post-stroke dementia first vs recurrent stroke and prior-dementia exclusions differ
16.5% major NCD vs “about 10%” new dementia criteria, timing, and hospital sampling differ
BP meta-analyses significant vs nonsignificant endpoints, trial set, power, and methods differ
cognitive drug scales improve but ADL does not statistical cognition change may lack clinical importance
CAA sensitivity varies 74.5–92.5% cohort selection and sample size differ
autopsy percentages cohort-specific, not global etiologic fractions
Syst-Eur 50–55% dementia reduction vs pooled OR 0.89–0.93 small event counts (21 vs 11 blinded), a possible dihydropyridine-specific effect, and an untreated-hypertension population no longer recruitable; shown side by side, never averaged (PMID 9802273, 12374512, 34028812)
donepezil 5 mg MD −0.92 points vs SMD −1.11 raw-scale versus standardized-mean-difference framing of overlapping trials; the SMD is a within-network ordering, not an effect magnitude (PMID 33704781, 35048806)
Western reviews find no recommendable drug vs 194-RCT Asian network ranking butylphthalide and huperzine A first risk-of-bias inclusion thresholds and regional literature coverage differ; neither position has an independent cross-region replication (PMID 42635820, 39239652)
genetically determined WMH causal in two-sample MR (OR 1.43) vs null in individual-level analysis (HR 1.02) different estimands from the same construct within one paper; unresolved (PMID 38776079)
clinical CADASIL prevalence ~2–11 per 100,000 vs genetic prevalence 1 in 277 ascertainment plus genuinely low penetrance; the two numbers measure different things and must not be interchanged (PMID 15834040, 24840674, 40982447)
single vs mixed cerebrovascular pathology single CVD pathologies showed no faster decline than none; only mixtures did. Present/absent coding of "cerebrovascular disease" averages across these groups (PMID 34601898)
probable-CAA category not associated with ICH recurrence while cortical superficial siderosis was within one randomized trial's imaging subanalysis, on 13 events; category-level and marker-level risk are not interchangeable (PMID 41197104)
VaD incidence 1.4–1.5 per 1,000 py vs post-stroke dementia 7.4–41.3% community incidence and post-event prevalence answer different questions and share no denominator
SVD imaging responds strongly to BP lowering (d −0.40) while dementia endpoints do not either the imaging endpoint is not on the causal path to cognition, or the clinical endpoints are underpowered; unresolved (PMID 42614618, 34028812)
pooled Chinese VCI prevalence 1.54% no higher than national VaD prevalence 1.6% a superset cannot be smaller than its subset; denominators (≥18 vs ≥60 y), criteria, and ascertainment differ. Shown side by side, never combined (PMID 40889799, 34178760, 33271079)
Hisayama VaD incidence 9.5 vs Rotterdam 1.5 and NEDICES 1.4 per 1,000 py not explained by weaker ascertainment — Hisayama verified 91.2% morphologically. Candidate causes (stroke incidence, AD under-detection, environment) have never been decomposed (PMID 18977814, 9521184, 17727890)
VaD prevalence 1.8× higher in men at 60–69 vs sex-neutral cumulative incidence (0.04 both sexes) prevalence and incidence diverge if male case fatality is higher; no study has reconciled them (PMID 31884487, 10599770)
probable-VaD autopsy sensitivity 0.20–0.25 vs 0.75 conjunctive (NINDS-AIREN/ADDTC) versus disjunctive (Mayo) criteria architecture, in hospital versus population samples (PMID 11772694, 12707071)
interrater κ 0.30–0.61 across criteria vs κ 0.79 for ICD-10 vascular dementia vignette/multi-criteria comparison versus pooled routine practice; ICD-10's high κ may reflect a definition almost nobody meets (PMID 10681076, 33942363)
perivascular spaces: OR 2.91 for 8-year dementia vs no cross-sectional association vs no effect of PVS volume counted extremes over long horizons, population-average burden cross-sectionally, and segmented volume in severe disease are three different measurements (PMID 33504642, 30068634, 38465597)
CAA has no independent dementia effect (OR 0.96 after Braak adjustment) vs CASI 45.9% lower in AD+CAA than 16.6% in AD alone mediation framing versus interaction framing of a conditional effect; both analyses are correct and answer different questions (PMID 39481068, 12058090)
vascular dementia is the agitated dementia vs the apathetic dementia long-term-care registry (n=10,405) versus memory clinic (n=180); setting and rating instrument differ, and no study has tested that explanation (PMID 35491774, 28245482)
Alzheimer caregiving is more burdensome vs vascular caregiving is more burdensome reverses with severity: vascular greater at mild-to-moderate, Alzheimer greater at severe (PMID 25475248, 10097779)
cerebrovascular pathology explains 3–8% of cognitive-decline variance in the cited cohorts and models prevalence times effect, not effect alone, determines population impact; this descriptive variance estimate is not a causal treatment-effect ceiling (PMID 33742668)
aerobic exercise ADAS-Cog −1.71 vs donepezil 5 mg −0.92 different trial sets, no head-to-head comparison exists; the exercise effect did not persist 6 months after stopping (PMID 27760869, 33704781)
2026-08-31 registry sweep found no phase-3 successor to LACI-2; 2026-09-02 sweep found one registry absence claims decay within days; IMPACT (NCT07252544, n=3,156) was registered but not returned by the earlier query