We have cited the Lancet Commission's list of modifiable dementia risk factors repeatedly without ever writing about the item on it that is easiest to find and easiest to treat. This fixes that.
High blood pressure is on that list. It produces no symptoms, it is measured in ninety seconds, and it is one of the few entries where there are randomised trials rather than only observational associations. Those trials are also more equivocal than the headlines were, and the honest version of this story includes a primary endpoint that missed significance. Nothing we sell treats or prevents hypertension, and nothing in this article should be read as suggesting otherwise.
The mechanism: small vessels, not big ones
When people hear "blood pressure and the brain" they think stroke. Stroke is the dramatic version, and worth knowing the warning signs for. But the more common story is quieter and cumulative.
The brain is supplied by deep, narrow arterioles that branch off larger vessels at close to full arterial pressure. They have little capacity to buffer it. Under sustained hypertension those small vessels thicken, stiffen and narrow — a process called arteriolosclerosis — and the tissue downstream ends up chronically under-perfused.
What that looks like on a scan is white matter hyperintensities: patches of damage in the deep white matter where the long connecting fibres run. They accumulate with age in almost everyone and faster in people with untreated hypertension. Because white matter is the wiring rather than the processing, the cognitive signature is characteristic — slowed processing speed, harder-won attention, reduced executive function, rather than the early dense memory loss typical of Alzheimer's disease.
Two things follow. This is why hypertension appears on dementia risk lists at all, and why it contributes to both vascular cognitive impairment and the mixed pathology most brains show at autopsy. And it is why the effect is slow: you do not feel your arterioles thickening.
SPRINT MIND: the trial that matters, reported honestly
SPRINT MIND is the best test anyone has run, and its result is more interesting than either the enthusiasts or the dismissers made it.
9,361 adults aged 50 or over with hypertension, without diabetes or prior stroke, across 102 US and Puerto Rican sites, randomised to a systolic target of under 120 mmHg (n=4,678) versus under 140 mmHg (n=4,683). The parent trial was stopped early in 2015 because intensive treatment reduced cardiovascular events and all-cause mortality. Cognitive follow-up continued to 2018, with a median intervention period of 3.34 years and median total follow-up of 5.11 years.
The primary cognitive outcome was probable dementia, and it did not reach statistical significance. 149 cases in the intensive group versus 176 in the standard group — 7.2 versus 8.6 per 1,000 person-years, hazard ratio 0.83, 95% CI 0.67 to 1.04. That interval crosses 1. Reported as "lowering blood pressure prevents dementia," which many outlets did, that is exactly the error this blog exists to avoid.
The secondary outcomes did reach significance. Mild cognitive impairment: 14.6 versus 18.3 per 1,000 person-years, HR 0.81 (0.69 to 0.95). The composite of MCI or probable dementia: 20.2 versus 24.1, HR 0.85 (0.74 to 0.97).
The authors' own explanation for the null primary result is the one worth carrying: because the trial stopped early and produced fewer dementia cases than expected, it may simply have been underpowered for that endpoint. A non-significant result in an underpowered trial is not evidence of no effect. It is also not evidence of an effect. Both halves of that sentence matter.
The imaging substudy, including the finding nobody quotes
A subset had brain MRI at baseline (670) and after a median of 3.97 years (449). The primary imaging outcome fit the mechanism above: white matter lesion volume grew by 0.92 cm³ in the intensive group versus 1.45 cm³ in the standard group, a between-group difference of −0.54 cm³ (95% CI −0.87 to −0.20). Tighter blood pressure control, less new white matter damage.
The secondary outcome went the other way. Total brain volume fell more in the intensive group — 30.6 cm³ versus 26.9 cm³, a difference of −3.7 cm³ (95% CI −6.3 to −1.1). Whether that reflects genuine harm, reduced brain fluid volume, or something else is unresolved. The authors described all the differences as small. Anyone citing the white matter result without this one is giving you half the paper.
What the wider evidence adds
A 2020 meta-analysis pooled 14 randomised trials and 96,158 participants. Blood pressure lowering was significantly associated with reduced dementia or cognitive impairment: 7.0% versus 7.5%, odds ratio 0.93 (95% CI 0.88 to 0.98), with no heterogeneity between trials. Note the absolute size — an absolute risk reduction of 0.39% (0.09% to 0.68%) over a mean 4.1 years. Real, consistent and small. The same analysis found no significant effect on cognitive test scores.
The most striking single trial is older. Syst-Eur randomised 2,418 people over 60 with isolated systolic hypertension to nitrendipine-based treatment or placebo and reported a 50% lower dementia incidence, 3.8 versus 7.7 cases per 1,000 patient-years. The caveat is the event count: 11 cases versus 21, p=0.05, over a median 2.0 years. A halving built on 32 events is a genuine finding and a fragile one, and the authors' own framing was modest — treating 1,000 patients for five years might prevent 19 cases.
Midlife versus late life, which is where most confusion lives
Timing changes the picture, and this is the part that makes single-number advice misleading.
The ARIC cohort followed 4,761 people with blood pressure measured across 24 years from midlife, then assessed cognition in later life. Dementia incidence per 100 person-years ran: 1.31 for sustained normotension; 1.99 for midlife normotension with late-life hypertension; 2.83 for hypertension at both stages — and 4.26 for midlife hypertension followed by late-life hypotension, the highest of all five groups.
Two conclusions. Sustained midlife hypertension carries the clearest risk, which is why the Lancet Commission's recommendation is to maintain systolic blood pressure at 130 mmHg or below from age 40. And falling blood pressure in later life after years of hypertension marks the worst outcomes — which is not a reason to tolerate high readings, but is a reason that aggressive targets in the frail elderly are a clinician's judgement rather than a rule. It also shows why low late-life readings can be a consequence of illness rather than a sign of health.
If your worry is your memory rather than your arteries, the more useful starting article is this one, and a short blood panel covers the reversible causes more cheaply.
What actually lowers blood pressure
Medication works and is a prescriber's conversation. On everything else, the trial numbers are worth having because they are smaller and more specific than the usual advice.
Diet. The DASH-Sodium trial, 412 participants, tested the DASH diet against a typical US control diet at three sodium levels. The combination — DASH plus low sodium versus control plus high sodium — produced systolic pressure 7.1 mmHg lower in participants without hypertension and 11.5 mmHg lower in those with it. That is the largest single non-drug effect here, and it required changing both diet pattern and salt, not one or the other.
Exercise. A meta-analysis of 93 trials and 5,223 participants found endurance training lowered systolic pressure by 3.5 mmHg (95% CI −4.6 to −2.3) on average. But the average conceals the useful detail: in hypertensive participants the reduction was 8.3 mmHg (−10.7 to −6.0), in prehypertensive 2.1 mmHg, and in people with normal blood pressure 0.75 mmHg with a confidence interval crossing zero — no effect. Exercise lowers blood pressure most in the people whose blood pressure is high.
Alcohol. A meta-analysis of 36 trials found that in people drinking two or fewer drinks daily, reducing intake produced no significant blood pressure change. In heavier drinkers it did: those consuming six or more per day who halved their intake saw systolic pressure fall 5.50 mmHg (−6.70 to −4.30). The wider case that moderate drinking is protective has not held up, but on blood pressure specifically the benefit of cutting down is concentrated in heavy drinkers.
Weight loss, treating sleep apnoea — the STOP-BANG questions take two minutes and untreated apnoea raises blood pressure — and stopping smoking all belong on the list.
Where supplements sit, which is nowhere
No Proco product treats, prevents or manages hypertension, and we do not sell anything that does. Magnesium has a small, inconsistent literature on blood pressure that we have gone through in full elsewhere, and it is not a reason to take magnesium instead of having your blood pressure measured. Omega-3's cardiovascular evidence is genuinely contested between major reviews. Our own comparison of exercise against supplements does not flatter us, and this is another instance of the same pattern.
The same vascular cluster drives the other Lancet factors, and insulin resistance and type 2 diabetes work through overlapping mechanisms. For the full list and what the 45% figure does and does not mean, start here, with the factor-by-factor breakdown here.
The one thing to do
Get your blood pressure measured. If you are over 40 and cannot remember the last reading, that is the action item, and it is free in most pharmacies. Everything above is downstream of a number you do not know.
This article is not medical advice and has not been evaluated by the FDA. Nothing here is intended to diagnose, treat, cure or prevent any disease, or to start, stop or change a prescribed medication. Blood pressure targets are individual and belong with a clinician who knows your history.
Sources: SPRINT MIND Investigators. JAMA. 2019;321:553-561 · Nasrallah IM, et al. JAMA. 2019;322:524-534 · Hughes D, et al. JAMA. 2020;323:1934-1944 · Forette F, et al. Lancet. 1998;352:1347-51 (Syst-Eur) · Walker KA, et al. JAMA. 2019;322:535-545 (ARIC) · Sacks FM, et al. N Engl J Med. 2001;344:3-10 (DASH-Sodium) · Cornelissen VA, Smart NA. J Am Heart Assoc. 2013;2:e004473 · Roerecke M, et al. Lancet Public Health. 2017;2:e108-e120 · Livingston G, et al. Lancet. 2024;404:572-628