Skip to content

Prevention and risk-factor control

TL;DR — Preventing vascular cognitive impairment is principally the prevention of stroke and cumulative vascular brain injury. Randomized blood-pressure evidence supports a small reduction in dementia or cognitive impairment overall (OR 0.93, 95% CI 0.88–0.98; absolute risk reduction 0.39% over about 4.1 years), but individual trials and subtype analyses are not uniformly positive (Hughes 2020, PMID 32427305; van Middelaar 2018, PMID 29927845). SPRINT MIND reduced MCI (HR 0.81, 95% CI 0.69–0.95) but not the underpowered probable-dementia endpoint (HR 0.83, 95% CI 0.67–1.04) (Williamson 2019, PMID 30688979). A 2025 cluster-randomized trial in 33,995 rural Chinese adults achieved a 22.0 mm Hg systolic contrast over 48 months and reduced all-cause dementia (RR 0.85, 95% CI 0.76–0.95), which is the first positive adequately powered randomized dementia-prevention result and suggests the earlier nulls reflect insufficient blood-pressure separation rather than absence of effect (He 2025, PMID 40258956). AF anticoagulation, diabetes care, lipids, smoking cessation, exercise, and multidomain programs have strong vascular or general cognitive rationales, but evidence for preventing specifically adjudicated vascular dementia remains thinner.

Why the cognitive benefit is small

Hughes's primary analysis is an absolute risk reduction of 0.39% (0.09–0.68) over 4.1 years, or about one dementia/cognitive-impairment event prevented per 256 people treated. SPRINT MIND's dementia HR 0.83 crossed the null after early termination for cardiovascular benefit. SPS3's 11 mm Hg systolic contrast after lacunar stroke did not move CASI scores. preDIVA's 6.7-year nurse-led vascular-care program in 3,526 older adults produced HR 0.92 (0.71–1.19) for dementia. These results are consistent with a real, small, population-level effect that is easy to miss in shorter or less selected trials (Hughes 2020, PMID 32427305; Williamson 2019, PMID 30688979; Pearce 2014, PMID 25453457; Moll van Charante 2016, PMID 27474376).

Prevention levels

Level Population Target
Primordial life course prevent vascular-risk development
Primary risk factors/no stroke prevent covert and overt brain injury
Secondary after stroke/TIA prevent recurrent injury
Tertiary established VCI slow further vascular damage and disability

Blood pressure

Evidence Population/design Cognitive result
SPRINT MIND intensive vs standard target probable dementia HR 0.83 (0.67–1.04); MCI HR 0.81 (0.69–0.95) (Williamson 2019, PMID 30688979)
JAMA meta-analysis 14 RCTs eligible (96,158); 12 trials/92,135 in primary analysis dementia/cognitive impairment OR 0.93 (0.88–0.98); ARR 0.39% (0.09–0.68) over mean 4.1 years (Hughes 2020, PMID 32427305)
IPD meta-analysis five placebo-controlled trials, 28,008 dementia OR 0.87 (0.75–0.99), mean BP difference 10/4 mmHg (Peters 2022, PMID 36282295)
earlier meta-analysis nine trials dementia RR 0.93 (0.84–1.02), nonsignificant (van Middelaar 2018, PMID 29927845)

Hughes included 12 trials/92,135 participants in the primary dementia/cognitive-impairment analysis (7.0% vs 7.5% over mean 4.1 years; I²=0%). van Middelaar included nine trials (seven drug, two lifestyle/combined) with 1,041/29,029 (3.6%) vs 1,090/28,653 (3.8%) dementia cases over median 3.9 years; three trials specifying subtypes found no significant effect on Alzheimer disease or vascular dementia separately (van Middelaar 2018, PMID 29927845).

The trial that changes the blood-pressure question

Every estimate above comes from trials that were not designed to prevent dementia, achieved modest blood-pressure separation, or ran too briefly for dementia to accumulate. One trial fixes all three problems at once. In rural China, 326 villages containing 33,995 adults aged ≥40 with uncontrolled hypertension were cluster-randomized to usual care or to a stepped-care protocol delivered by trained non-physician community health providers under primary-care supervision, targeting systolic <130 and diastolic <80 mm Hg. Over 48 months the net contrast was 22.0 mm Hg systolic (95% CI 20.6–23.4) and 9.3 mm Hg diastolic (8.7–10.0) — roughly double the separation achieved in the placebo-controlled trials pooled by Cochrane — and the prespecified primary outcome of all-cause dementia was reduced: risk ratio 0.85 (95% CI 0.76–0.95, P=0.0035). Serious adverse events were less frequent in the intervention group (RR 0.94, 0.91–0.98, P=0.0006) (He 2025, PMID 40258956, NCT03527719).

This large cluster-randomized trial had incident dementia as its primary outcome and was positive. A 15% relative reduction on a 22 mm Hg contrast is consistent with the possibility that limited blood-pressure separation and follow-up contributed to earlier nulls, but cross-trial comparison cannot establish that explanation. Three limitations bound the inference: it was open-label with blinded endpoint adjudication, dementia was all-cause rather than subtype-adjudicated (so the vascular-dementia-specific effect is unknown), and the population — untreated hypertension in a rural low-resource setting, mean age well below most Western trials — resembles Syst-Eur's population more than SPRINT's.

The Syst-Eur outlier

The single largest reported antihypertensive dementia effect in the literature is also the oldest. In the double-blind phase of Syst-Eur (2,418 people aged ≥60 with isolated systolic hypertension, median 2.0 years), nitrendipine-based treatment cut dementia incidence by 50%, from 7.7 to 3.8 cases per 1,000 patient-years (21 vs 11 events, P=0.05), on a systolic/diastolic difference of only 8.3/3.8 mm Hg (Forette 1998, PMID 9802273). Open-label extension to a median 3.9 years, by which time roughly half the former control group had also started nitrendipine, put the reduction at 55% (7.4 to 3.3 per 1,000 patient-years; 43 vs 21 cases, P<0.001), with an adjusted hazard ratio of 0.38 (95% CI 0.23–0.64) for nitrendipine use and 20 dementias prevented per 1,000 patients treated for 5 years (95% CI 7–33) (Forette 2002, PMID 12374512).

That effect is roughly ten times the pooled estimate. Three readings compete and none has been settled: (i) the event counts are small (21 vs 11 in the blinded phase) and the P value is exactly 0.05, so the estimate may be an early-stopping-style overshoot; (ii) dihydropyridine calcium-channel blockade may protect the brain beyond blood pressure — an exploratory network meta-analysis of the four placebo-controlled trials that adjudicated dementia (Syst-Eur, PROGRESS, SCOPE, HYVET-COG; 16,823 participants) ranked calcium-channel blockers highest for dementia prevention (SUCRA 94.7%) while diuretic ± ACE-inhibitor combinations ranked highest for cognitive-score improvement (SUCRA 95.3%), with the authors themselves warning that one study per node makes the ranking unreliable (Ding 2026, PMID 41962914); or (iii) the population — untreated isolated systolic hypertension at a mean MMSE of 29 — carried far more preventable vascular injury than modern trial populations already on treatment. The Cochrane review that pools four placebo-controlled trials reporting incident dementia in people without prior cerebrovascular disease finds no effect (236/7,767 vs 259/7,660; OR 0.89, 95% CI 0.72–1.09, very low certainty), alongside a possible small MMSE benefit (MD 0.20, 95% CI 0.10–0.29, very low certainty) achieved on a −9.25/−2.47 mm Hg contrast; it attributes the null largely to study durations far too short for dementia to accumulate, and to placebo-group drop-in (Cunningham 2021, PMID 34028812). Ding's pairwise dementia estimate is OR 0.89 (95% CI 0.76–1.05), numerically close to Cunningham's OR 0.89 (0.72–1.09), but the two four-trial sets are not the same: Ding includes PROGRESS (prior stroke/TIA), whereas Cunningham is restricted to people without prior cerebrovascular disease (Cunningham 2021, PMID 34028812; Ding 2026, PMID 41962914). The shared point estimate is therefore not independent replication.

HYVET-COG is the placebo-controlled trial in the oldest population and it is individually null. Among 3,336 participants aged ≥80 with systolic 160–200 mm Hg followed a mean 2.2 years on indapamide ± perindopril, 263 incident dementias occurred at 38 per 1,000 patient-years on placebo and 33 on treatment (HR 0.86, 95% CI 0.67–1.09) on a 15/5.9 mm Hg contrast; pooling it with the other placebo-controlled trials moved the combined estimate to HR 0.87 (0.76–1.00, P=0.045) (Peters 2008, PMID 18614402). The trial stopped early for stroke and mortality benefit, which truncated dementia accrual — the same structural problem that later affected SPRINT MIND. Early stopping reduced follow-up and power for the slower cognitive endpoint.

PROGRESS assigned 6,105 people with prior stroke/TIA to perindopril ± indapamide versus placebo: dementia 193/3,051 (6.3%) vs 217/3,054 (7.1%), RRR 12% (95% CI −8% to 28%); cognitive decline 9.1% vs 11.0%, RRR 19% (4–32%), driven by events associated with recurrent stroke (Tzourio 2003, PMID 12742805). SPS3 found no CASI difference between systolic targets <130 vs 130–149 mm Hg after recent lacunar stroke (Pearce 2014, PMID 25453457).

SPRINT MRI substudy (n=670 baseline, 449 follow-up MRI) showed smaller WMH volume increase with intensive treatment (0.92 vs 1.45 cm³; between-group difference −0.54 cm³, 95% CI −0.87 to −0.20) but a slightly greater decrease in total brain volume (−3.7 cm³, −6.3 to −1.1) (SPRINT MIND Investigators 2019, PMID 31408137). A later post-hoc analysis of the same substudy replaced single-marker WMH with a confirmatory-factor-analysis global SVD score integrating periventricular WMH, white-matter free water, and basal-ganglia perivascular spaces: over a median 3.9 years, intensive treatment produced markedly less SVD progression (Cohen's d = −0.40, 95% CI −0.62 to −0.17) with a clear dose-response by attained systolic reduction — 21.2% (7.4–35.0%) less progression for 0–10 mm Hg, 26.3% (13.1–39.5%) for 10–20 mm Hg, and 39.4% (24.2–54.5%) for ≥20 mm Hg (Charisis 2026, PMID 42614618). The imaging effect is therefore large and dose-dependent while the dementia effect is small and uncertain — the clearest instance in this knowledge base of an imaging surrogate outrunning its clinical endpoint (see OQ-7). PRESERVE network analysis (82 lacunar-stroke participants) found intensive BP improved DTI weighted global efficiency (P=0.002) without changing conventional WMH or volume metrics (Pflanz 2022, PMID 35977831).

Differences reflect trial selection, endpoint definitions, intervention contrast, follow-up, and power. A pooled small benefit should not be converted into a universal intensive target. Orthostatic hypotension, falls, renal effects, and frailty require monitoring; beta-blockers (OR 7.76, 95% CI 2.51–24.03) and tricyclic antidepressants (OR 6.30, 2.86–13.91) have the largest drug-induced orthostatic-hypotension odds versus placebo (Bhanu 2021, PMID 34752479). Orthostatic hypotension itself is associated with slightly lower MMSE (MD −0.51, 95% CI −0.85 to −0.17) and higher cognitive-impairment odds (OR 1.19, 1.00–1.42) in older adults, with uncertain causal direction (Iseli 2019, PMID 30825549).

When the exposure matters: midlife versus late life

Trial populations are old; the exposure may not be. In 15,744 ARIC participants aged 44–66 at baseline followed 25 years to 1,516 adjudicated dementia cases, midlife diabetes carried HR 1.77 (95% CI 1.53–2.04) — nearly as large as APOE ε4 (HR 1.98, 1.78–2.21) — with midlife smoking HR 1.41 (1.23–1.61), hypertension HR 1.39 (1.22–1.59) and even prehypertension HR 1.31 (1.14–1.51) (Gottesman 2017, PMID 28783817). A midlife metabolic exposure conferring approximately the same dementia hazard as the strongest common genetic risk factor is the single most quotable argument for treating vascular risk as a dementia intervention.

The attributable fractions have since been computed across the life course in the same cohort, with 33 years of follow-up. Having at least one of hypertension, diabetes or current smoking accounted for 21.8% (95% CI 14.3–29.3) of dementia by age 80 when measured at ages 45–54, 26.4% (19.1–33.6) at 55–64, and 44.0% (30.9–57.2) at 65–74 — but only 2–8% of dementia occurring after age 80. Fractions were higher in APOE ε4 non-carriers (33.3–61.4% from age 55), in Black participants (25.5–52.9% from age 45) and in women (29.2–51.3% from age 55) (Smith 2025, PMID 40455489). Two consequences follow directly. Vascular prevention has a deadline: the attributable fraction collapses for dementia after 80, so a trial recruiting 80-year-olds is targeting a window in which the exposure explains almost nothing — which is a better explanation of HYVET-COG's null than trial duration alone. And the benefit is concentrated in the groups least represented in prevention trials.

Population-attributable estimates outside high-income Western settings are larger and differently composed. Using 75,214 person-waves from the China Health and Retirement Longitudinal Study, the combined PAF for the Lancet Commission's 12 modifiable risk factors was 45.4% in 2011 and 52.5% in 2018 (change not statistically significant), with low education the largest single contributor (mean individual weighted PAF 11.3%), followed by depression, hypertension, smoking and physical inactivity; PAFs were higher in lower-income and rural groups, and the socioeconomic and rural–urban gaps stayed constant or widened over the period (Chen 2024, PMID 38872868). Roughly half of dementia being attributable to modifiable factors — against the roughly 40% usually quoted — is what one expects where risk-factor control is less complete, and it is the quantitative basis for expecting a trial like the Chinese cluster-randomized study above to succeed where Western trials did not.

Atrial fibrillation

AF causes embolic stroke and is associated with cognitive decline. Anticoagulation prevents AF-related stroke when indicated, but comparative dementia-risk evidence between direct oral anticoagulants and warfarin is largely observational (Branco 2023, PMID 37405677; Fong 2023, PMID 38095004).

Question Evidence boundary
anticoagulate AF? use stroke/bleeding net benefit, not dementia alone
DOAC vs warfarin cognition? observational confounding remains
microbleeds? increase ICH risk but are not a binary rule (Charidimou 2017, PMID 29117953)
rhythm control? cognitive endpoint evidence incomplete (see below)

Rhythm control is the AF question with the largest reported cognitive effect and the weakest design behind it. Pooling 14 studies covering 193,830 AF patients — predominantly observational, with ROBINS-I rather than randomized-trial risk assessment — rhythm control was associated with lower dementia risk than rate control (HR 0.74, 95% CI 0.62–0.89; I²=62%), and catheter ablation specifically with lower overall dementia (HR 0.62, 95% CI 0.56–0.68), Alzheimer disease (HR 0.78, 95% CI 0.66–0.92), and vascular dementia (HR 0.58, 95% CI 0.42–0.80; I²=31%) (Guo 2024, PMID 38369630). Effects of that magnitude in observational ablation cohorts are the classic signature of confounding by indication — people selected for ablation are younger, fitter, and less comorbid than those left on rate control — and no randomized trial has yet tested a cognitive endpoint. The vascular-dementia-specific hazard ratio is the most interesting number here precisely because it is the one a stroke-mediated mechanism would predict, and therefore the one most easily explained without invoking a direct cognitive effect of rhythm.

Lipids and diabetes

Statins prevent vascular events in indicated populations; post-stroke cognitive meta-analysis found heterogeneous relations among lipids, atherosclerosis, statin use, and cognition that do not justify prescribing solely as a dementia drug (Yang 2020, PMID 31505259). The randomized evidence is now large enough to be conclusive on the null: 15 trials with 139,169 participants reporting incident cognitive impairment or dementia gave OR 0.96 (95% CI 0.74–1.26) for lipid-lowering therapy versus control (1.33% vs 1.36% over a mean 34.5 months), with no signal for statins (OR 0.90, 95% CI 0.67–1.21), PCSK9 inhibitors (OR 1.77, 95% CI 0.46–6.83), or other agents (OR 0.85, 95% CI 0.61–1.17) (Reddin 2025, PMID 40794911). This result does double duty: it removes lipid lowering from the list of plausible dementia-prevention levers, and it is the strongest available refutation of the recurring concern that statins impair cognition. Diabetes is a risk marker in post-stroke meta-analysis, but direct evidence that a specific glucose-lowering regimen prevents VaD is insufficient (Filler 2024, PMID 38101426).

Lifestyle and multidomain intervention

FINGER randomized 1,260 at-risk adults (CAIDE score ≥6, ages 60–77) to diet, exercise, cognitive training, and vascular monitoring versus general advice. The between-group annual change in neuropsychological-test-battery score was 0.022 (95% CI 0.002–0.042); adverse events were 7% versus 1%, mostly musculoskeletal pain (Ngandu 2015, PMID 25771249). preDIVA, a 6-year nurse-led vascular-care cluster trial in 3,526 community adults aged 70–78, found no dementia reduction (121/1,853 vs 112/1,601; HR 0.92, 95% CI 0.71–1.19) (Moll van Charante 2016, PMID 27474376). ASPREE found no dementia, MCI, or cognitive-decline benefit of 100 mg aspirin versus placebo in 19,114 older adults over a median 4.7 years (dementia-trigger HR 1.03, 0.91–1.17; probable AD HR 0.96, 0.74–1.24; MCI HR 1.12, 0.92–1.37) (Ryan 2020, PMID 32213642). VITATOPS B-vitamins lowered homocysteine without changing MMSE or incident cognitive impairment after stroke/TIA (Hankey 2013, PMID 23765945).

Component Plausible route Evidence caution
physical activity BP, metabolism, fitness, neuroplasticity adherence and injury
diet vascular/metabolic exposure pattern and access differ
cognitive training task-specific performance/reserve transfer to daily function
smoking cessation reduces vascular events cognitive endpoint rarely primary
sleep treatment oxygenation/BP/daytime function VaD-specific trials sparse
social engagement reserve and mood confounding in cohorts

FINGER supports a small cognitive-score effect in selected at-risk older adults; it was not a treatment trial for established vascular dementia.

What the FINGER replications actually showed

Trial N / design Result Comparator problem
FINGER (Finland) 1,260, multidomain vs general advice, 2 y between-group annual NTB change 0.022 (95% CI 0.002–0.042) (Ngandu 2015, PMID 25771249) control arm received advice only
US POINTER 2,111, structured vs self-guided multidomain, 2 y both arms improved: 0.243 SD/y structured vs 0.213 SD/y self-guided; difference 0.029 SD/y (95% CI 0.008–0.050; P=0.008); benefit greater at lower baseline cognition (P=0.02 for interaction) and unrelated to APOE ε4 (P=0.95) (Baker 2025, PMID 40720610) both arms were active
LatAm-FINGERS 1,065 across 11 Latin American countries, systematic vs flexible lifestyle intervention, 2 y 0.31 SD/y vs 0.20 SD/y; difference 0.11 SD/y (95% CI 0.06–0.15; P<0.0001); adherence 71.6%; 82.3% completed (Crivelli 2026, PMID 42442374) both arms active; effect ~4× US POINTER
rIVR + exercise (2×2 factorial) 513, aerobic exercise and/or systolic <130 mm Hg plus atorvastatin, 2 y no effect of exercise (0.1 PACC units, 95% CI −0.1 to 0.2; P=0.37), of intensive risk reduction (−0.1 units, 95% CI −0.3 to 0.03; P=0.12; the published abstract prints this difference as unsigned 0.1, but the arm means of +0.2 vs +0.3 imply a negative contrast, which is the quantity that sits inside the interval), or of the combination (P=0.13 for interaction) (Zhang 2026, PMID 41870419) usual-care arm included

Three features of this table matter more than the individual P values. First, global cognitive scores rose over two years in every arm of every recent trial, including self-guided and flexible controls — practice effects and trial participation itself dominate the raw trajectory, so a multidomain trial without an active control cannot distinguish intervention from enrolment. Second, the between-group differences shrink as the control arm gets better: 0.11 SD/y against flexible advice in Latin America, 0.029 SD/y against a genuinely self-guided program in the US. Third, the one trial that randomized pharmacological vascular risk reduction with a usual-care arm — the design closest to the vascular-dementia causal hypothesis — was flatly null on cognition at 24 months despite hitting its blood-pressure and LDL targets (Zhang 2026, PMID 41870419). Taken together these results are compatible either with a small real multidomain benefit or with an engagement effect, and the field has not designed the trial that separates them.

Outside the vascular domain, the largest single-risk-factor prevention trial is instructive about population selection: ACHIEVE randomized 977 adults aged 70–84 with untreated hearing loss to hearing aids plus audiological counselling or health education, and found no difference in 3-year global cognitive change (−0.200 vs −0.202 SD; difference 0.002, 95% CI −0.077 to 0.081; P=0.96) — but a prespecified interaction (P=0.010) showed benefit confined to the ARIC subcohort, who were older and carried more vascular risk (Lin 2023, PMID 37478886). A prevention effect that appears only in the higher-vascular-risk stratum of a trial is exactly the design lesson VCI prevention trials have repeatedly failed to apply.

Secondary stroke prevention

Preventing recurrent stroke is strongly justified because recurrent stroke cohorts have much higher dementia burden (Pendlebury 2009, PMID 19782001). Cause-specific strategies—anticoagulation for AF, antiplatelet therapy for non-cardioembolic disease, lipid lowering, BP control, diabetes and smoking management—should follow stroke guidance rather than be improvised from cognitive outcomes.

Implementation and equity

Risk-factor efficacy requires detection, affordable medication, adherence, monitoring, and long-term follow-up. Cognitive impairment itself reduces medication-management capacity, creating a feedback loop. Intervention designs should measure caregiver workload and treatment burden, not only targets.

Open questions

  • Does the 15% all-cause dementia reduction on a 22 mm Hg systolic contrast extend to adjudicated vascular dementia, and does it replicate in populations already on treatment? (He 2025, PMID 40258956)
  • If only 2–8% of dementia occurring after age 80 is attributable to hypertension, diabetes or smoking, should prevention trials stop recruiting above 75? (Smith 2025, PMID 40455489; Peters 2008, PMID 18614402)
  • Why does midlife diabetes carry a dementia hazard (1.77) approaching that of APOE ε4 (1.98), and is that hazard reversible? (Gottesman 2017, PMID 28783817)
  • Is the higher attributable fraction in China (45–52%) evidence of greater prevention headroom, or of different risk-factor measurement? (Chen 2024, PMID 38872868)
  • Which BP target optimizes cognitive benefit in frail or SVD-enriched patients? (Peters 2022, PMID 36282295)
  • Does anticoagulation reduce dementia independent of preventing clinical stroke? (Branco 2023, PMID 37405677)
  • Can recurrent-stroke prevention demonstrably lower delayed dementia incidence? (Filler 2024, PMID 38101426)
  • Which multidomain components drive the FINGER effect, and why did preDIVA not reduce dementia? (Ngandu 2015, PMID 25771249; Moll van Charante 2016, PMID 27474376)
  • Does intensive BP slowing of WMH growth translate into less dementia? (SPRINT MIND Investigators 2019, PMID 31408137)
  • Why is the Syst-Eur dementia effect an order of magnitude larger than every pooled estimate — small-numbers chance, a dihydropyridine-specific effect, or an untreated-hypertension population no longer recruitable? (Forette 1998, PMID 9802273; Forette 2002, PMID 12374512; Cunningham 2021, PMID 34028812)
  • If intensive BP control reduces global SVD burden with a clear dose-response (Cohen's d −0.40) but not dementia, is the imaging endpoint invalid or the clinical endpoint underpowered? (Charisis 2026, PMID 42614618)
  • Do multidomain trials measure intervention or enrolment, given that cognition improved in every arm including self-guided controls? (Baker 2025, PMID 40720610; Crivelli 2026, PMID 42442374)
  • Why did randomized pharmacological vascular risk reduction plus exercise fail where observational vascular-risk associations are strong? (Zhang 2026, PMID 41870419)
  • Is the ablation–vascular-dementia association (HR 0.58) causal or confounding by indication? (Guo 2024, PMID 38369630)

References

  1. Williamson JD, et al. Intensive versus standard BP control on probable dementia. JAMA. 2019. PMID 30688979
  2. Hughes D, et al. BP lowering with incident dementia/cognitive impairment. JAMA. 2020. PMID 32427305
  3. Peters R, et al. BP lowering and prevention of dementia: IPD meta-analysis. Eur Heart J. 2022. PMID 36282295
  4. van Middelaar T, et al. BP-lowering interventions to prevent dementia. J Hypertens. 2018. PMID 29927845
  5. Bhanu C, et al. Drug-induced orthostatic hypotension. PLoS Med. 2021. PMID 34752479
  6. Branco DR, et al. Direct oral anticoagulants vs vitamin K antagonist on dementia risk in atrial fibrillation. J Thromb Thrombolysis. 2023. PMID 37405677
  7. Fong KY, et al. Dementia risk of DOACs versus warfarin. JACC Asia. 2023. PMID 38095004
  8. Charidimou A, et al. Microbleeds, anticoagulation, hemorrhage risk. Neurology. 2017. PMID 29117953
  9. Yang Z, et al. Blood lipids, atherosclerosis, statins and post-stroke cognition. Ageing Res Rev. 2020. PMID 31505259
  10. Ngandu T, et al. FINGER multidomain intervention. Lancet. 2015. PMID 25771249
  11. Pendlebury ST, Rothwell PM. Post-stroke dementia. Lancet Neurol. 2009. PMID 19782001
  12. Filler J, et al. Risk factors after stroke. Lancet Healthy Longev. 2024. PMID 38101426
  13. Tzourio C, et al. PROGRESS dementia and cognitive decline. Arch Intern Med. 2003;163:1069-75. PMID 12742805
  14. Pearce LA, et al. SPS3 cognitive function. Lancet Neurol. 2014;13:1177-85. PMID 25453457
  15. SPRINT MIND Investigators for the SPRINT Research Group. Intensive vs standard BP control and cerebral white matter lesions. JAMA. 2019;322:524-534. PMID 31408137
  16. Pflanz CP, et al. BP-lowering intensity and white-matter network integrity. Neurology. 2022;99:e1945-e1953. PMID 35977831
  17. Iseli R, et al. Orthostatic hypotension and cognition in older adults. Exp Gerontol. 2019;120:40-49. PMID 30825549
  18. Moll van Charante EP, et al. preDIVA multidomain vascular care. Lancet. 2016;388:797-805. PMID 27474376
  19. Ryan J, et al. ASPREE aspirin and dementia. Neurology. 2020;95:e320-e331. PMID 32213642
  20. Hankey GJ, et al. VITATOPS B-vitamins and cognition. Stroke. 2013;44:2232-9. PMID 23765945
  21. Forette F, et al. Prevention of dementia in randomised double-blind placebo-controlled Systolic Hypertension in Europe (Syst-Eur) trial. Lancet. 1998;352:1347-51. PMID 9802273
  22. Forette F, et al. The prevention of dementia with antihypertensive treatment: new evidence from the Systolic Hypertension in Europe (Syst-Eur) study. Arch Intern Med. 2002;162:2046-52. PMID 12374512
  23. Cunningham EL, et al. Pharmacological treatment of hypertension in people without prior cerebrovascular disease for the prevention of cognitive impairment and dementia. Cochrane Database Syst Rev. 2021;5:CD004034. PMID 34028812
  24. Ding L, et al. Impact of different antihypertensive drug classes on incident dementia in older adults: a systematic review and meta-analysis. J Am Med Dir Assoc. 2026;27:106185. PMID 41962914
  25. Charisis S, et al. Intensive versus standard blood pressure control and overall brain small vessel disease burden: a post-hoc analysis of the SPRINT randomized clinical trial. EClinicalMedicine. 2026;99:104143. PMID 42614618
  26. Guo J, et al. Effects of rhythm-control and rate-control strategies on cognitive function and dementia in atrial fibrillation: a systematic review and meta-analysis. Age Ageing. 2024;53:afae009. PMID 38369630
  27. Reddin C, et al. Association of lipid-lowering therapy with dementia and cognitive outcomes: a systematic review and meta-analysis. Age Ageing. 2025;54:afaf219. PMID 40794911
  28. Baker LD, et al. Structured vs self-guided multidomain lifestyle interventions for global cognitive function: the US POINTER randomized clinical trial. JAMA. 2025;334:681-691. PMID 40720610
  29. Crivelli L, et al. Multidomain lifestyle intervention for the prevention of cognitive decline in at-risk older adults in Latin America (LatAm-FINGERS). Lancet. 2026;408:417-429. PMID 42442374
  30. Zhang R, et al. Effects of exercise and intensive vascular risk reduction on cognitive function in older adults: a randomized clinical trial. JAMA Neurol. 2026;83:424-434. PMID 41870419
  31. Lin FR, et al. Hearing intervention versus health education control to reduce cognitive decline in older adults with hearing loss in the USA (ACHIEVE). Lancet. 2023;402:786-797. PMID 37478886
  32. He J, et al. Blood pressure reduction and all-cause dementia in people with uncontrolled hypertension: an open-label, blinded-endpoint, cluster-randomized trial. Nat Med. 2025;31:2054-2061. PMID 40258956
  33. Peters R, et al. Incident dementia and blood pressure lowering in the Hypertension in the Very Elderly Trial cognitive function assessment (HYVET-COG). Lancet Neurol. 2008;7:683-9. PMID 18614402
  34. Gottesman RF, et al. Associations between midlife vascular risk factors and 25-year incident dementia in the ARIC cohort. JAMA Neurol. 2017;74:1246-1254. PMID 28783817
  35. Smith JR, et al. Contribution of modifiable midlife and late-life vascular risk factors to incident dementia. JAMA Neurol. 2025;82:644-654. PMID 40455489
  36. Chen S, et al. Temporal trends and disparities of population attributable fractions of modifiable risk factors for dementia in China. Lancet Reg Health West Pac. 2024;47:101106. PMID 38872868