Physical inactivity is the last of the Lancet Commission's fourteen modifiable dementia risk factors to get its own post here, and the one where we have least to gain from writing it. That is why it needed writing. This is not a comparison of exercise against supplements — that lives in our honest comparison of exercise against anything you could buy from us. This post covers what that one does not: how the evidence is graded, what the trials found, where the dose-response curve is steepest, and why the strongest recommendation in this field rests on weaker evidence than almost anyone reporting it lets on.
The only strong recommendation of its kind
WHO's 2019 guideline on risk reduction of cognitive decline and dementia graded every intervention using GRADE. On physical activity: "Physical activity should be recommended to adults with normal cognition to reduce the risk of cognitive decline." Quality of evidence: moderate. Strength of recommendation: strong.
That pairing is close to unique in the document. Tobacco cessation is also strong, but on low-quality evidence. The strong recommendations for hypertension and diabetes are strong for treating those conditions per existing WHO guidelines — narrowed to dementia risk specifically, both fall to conditional on very low-quality evidence, worth remembering when reading about blood pressure and the brain or the glucose-lowering trials.
The grading also changes by population. For people who already have mild cognitive impairment, WHO says physical activity "may be recommended" — quality low, strength conditional. Most articles quoting the strong recommendation apply it to both groups.
What the observational evidence shows
The cohort literature is large and consistent. A 2022 meta-analysis of 58 prospective studies found physical activity associated with lower incidence of all-cause dementia (pooled relative risk 0.80, 95% CI 0.77 to 0.84, n=257,983). A 2026 update took this to 74 studies and 4,227,297 participants over a median 10 years and landed identically: RR 0.80 (95% CI 0.75 to 0.86). Heterogeneity was very high (I² 97.4%), funnel plots suggested publication bias, and certainty was rated very low to low.
A 2026 analysis of the Nurses' Health Study and Health Professionals Follow-Up Study — 107,036 participants, activity re-measured every two to four years from 1986, dementia followed to 2023 — put the top quartile at a hazard ratio of 0.72 (95% CI 0.68 to 0.76) against the bottom. Its endpoint combined self-reported diagnoses with deaths attributed to dementia, softer than clinical adjudication.
The reverse-causation test, which this factor passes
People in the years before a dementia diagnosis move less. Apathy, gait change and early executive difficulty all reduce activity, so low activity can be a consequence rather than a cause — the problem that makes late-life readings misleading for body weight and LDL cholesterol.
Here the answer goes the other way from those two, and we should say so rather than reach for the tidier parallel. The 2022 review was built to test whether follow-up length explained the association, and it did not: it held in follow-ups of 20 years or more, and neither baseline age, follow-up length nor study quality significantly moderated it. The 2026 cohort analysis used a four-year lag for the same reason, and the step study below re-ran its analysis excluding everyone diagnosed in the first two years without the results changing.
The trials, which complicate it considerably
This is the part that gets omitted. A 2024 meta-analysis restricted to randomised trials of at least 12 months in non-demented, community-dwelling older adults — 16 trials, 11,402 participants, low risk of bias — found long-term exercise alone had no significant effect on global cognition (9 trials, SMD 0.05, 95% CI −0.04 to 0.13, p=0.25), on moderate-certainty evidence. It did not alter risk of mild cognitive impairment (RR 0.98, 95% CI 0.73 to 1.31, p=0.79) or dementia onset (RR 0.61, 95% CI 0.25 to 1.52, p=0.19), both non-significant on very low to low certainty.
Once dementia is established it is worse. The DAPA trial randomised 494 people with mild to moderate dementia across 15 English regions, 2:1, to four months of supervised moderate-to-high-intensity aerobic and strength training plus usual care, or usual care alone. Fitness improved (six-minute walk distance +18.1 m, 95% CI 11.6 to 24.6). At 12 months mean ADAS-Cog was 25.2 in the exercise arm against 23.8 in usual care: an adjusted between-group difference of −1.4 points (95% CI −2.6 to −0.2, p=0.03) in the direction of greater cognitive impairment in the exercise group. Higher ADAS-Cog means worse. The authors called the difference small and its clinical relevance uncertain, and found nothing in secondary outcomes or prespecified subgroups. An earlier Cochrane review of 17 trials in 1,067 people with dementia likewise found no clear cognitive benefit (SMD 0.43, 95% CI −0.05 to 0.92, p=0.08).
So: strong observational evidence for prevention, no trial evidence that exercise prevents dementia, none that it treats it. Three separate claims, routinely reported as one — the same shape as hearing loss, where a strong association meets a null trial.
Why the recommendation is still strong
Because GRADE does not set strength on certainty alone. It weighs benefits against harms, and exercise has large, certain benefits for cardiovascular health, function and mortality, with minimal harms. You do not need the dementia evidence to be conclusive to recommend it strongly — only the downside to be negligible. That is a legitimate basis for a strong recommendation, not proof that exercise prevents dementia.
Dose-response: the first few thousand steps do most of the work
A UK Biobank study of 78,430 adults aged 40 to 79 with wrist accelerometers recorded 866 dementia cases over a median 6.9 years. The optimal dose was 9,826 steps a day (HR 0.49, 95% CI 0.39 to 0.62). The minimal dose — where risk reduction reached half the observed maximum — was 3,826 steps a day (HR 0.75, 95% CI 0.67 to 0.83). Half the available benefit arrived below 4,000 steps. Intensity counted too: a peak 30-minute cadence of 112 steps a minute carried HR 0.38 (95% CI 0.24 to 0.60).
The flattening at the top of the curve probably reflects sparse data rather than a real ceiling, by the authors' own assessment, and UK Biobank's 5.5% response rate makes it unrepresentative. A separate meta-analysis of leisure-time activity found a linear relationship across 0 to 45 MET-hours a week, so the curve's shape is not settled.
Sedentary time is a separate exposure
It is not the inverse of exercise volume. A 2023 UK Biobank analysis of 49,841 adults aged 60 and over used accelerometer-derived sedentary time and found 414 dementia cases over a mean 6.7 years, with a non-linear association. Against a median of 9.27 hours a day, hazard ratios were 1.08 (95% CI 1.04 to 1.12) at 10 hours, 1.63 (95% CI 1.35 to 1.97) at 12 hours and 3.21 (95% CI 2.05 to 5.04) at 15 hours. Number of sedentary bouts per day was not associated with risk (HR 1.00, 95% CI 0.99 to 1.01, p=0.89) — on this evidence it is total hours, not how they are broken up.
What counts, and one finding that surprised us
The guideline language is about volume and intensity, not gyms. WHO asks adults for 150 to 300 minutes of moderate-intensity aerobic activity a week or 75 to 150 vigorous, plus muscle-strengthening on two or more days, and states plainly that "doing some physical activity is better than doing none." Walking, stairs, gardening and housework all count toward volume. In the 2026 cohort analysis walking beat vigorous exercise on the dementia endpoint (HR 0.76 versus 0.89).
Domain matters, too. The 2026 meta-analysis split activity by where it happens: leisure-time activity was associated with lower dementia risk (RR 0.76, 95% CI 0.65 to 0.88), and household activity too (RR 0.85, 95% CI 0.74 to 0.98) — but from a single study. Occupational activity went the other way (RR 1.20, 95% CI 1.01 to 1.42), as did commuting (RR 1.10, 95% CI 1.03 to 1.18). Certainty was very low to low, so this is a signal, not a verdict. It does mean "I'm on my feet all day at work" is unsupported as a substitute for deliberate activity. If you have a cardiac, respiratory or musculoskeletal reason to be careful, ask your GP where to start before you start.
Mechanism, and what it is not
The mechanistic work is interesting; none of it is a clinical outcome. The best-known trial randomised 120 older adults to a walking programme or a stretching control for a year; the aerobic group's anterior hippocampus grew about 2%, volume gains correlated with serum BDNF, and caudate and thalamus were unaffected. What it measured was MRI volume plus a spatial memory task — not dementia incidence, not diagnosis, not function. A hippocampal volume change, a BDNF rise or better cerebral blood flow is a plausible pathway, not an outcome, and anyone selling you something on the strength of one should be asked which they mean.
We sell nothing for this, and it is the one to do
Nothing in the Proco range substitutes for physical activity, and no supplement has comparable evidence on this endpoint. The same WHO guideline that strongly recommends physical activity also contains this, on moderate-quality evidence, as a strong recommendation: "Vitamins B and E, polyunsaturated fatty acids and multi-complex supplementation should not be recommended to reduce the risk of cognitive decline and/or dementia." Polyunsaturated fatty acids means fish oil. We sell fish oil. We think the omega-3 and cognition evidence holds more than that sentence allows, and we have said why most of what is marketed for cognition fails our bar — but a guideline body graded that recommendation strong, and we are not burying it here.
If you do exactly one thing off the list of fourteen, do this one. It carries the strongest recommendation on the list, the case for it does not depend on the dementia evidence being conclusive, half the observed risk reduction sits below 4,000 steps a day, and it costs nothing. Everything we sell is a layer on top of that, or nothing at all. For other outcomes the calculus differs — see exercise and depression and exercise for anxiety.
This article is for informational purposes only and has not been evaluated by the FDA. It is not intended to diagnose, treat, cure or prevent any disease. Speak with a healthcare provider before starting a new exercise programme or supplement, particularly if you have a cardiac, respiratory or musculoskeletal condition or take regular medication.
Sources: World Health Organization. Risk Reduction of Cognitive Decline and Dementia: WHO Guidelines. Geneva: WHO, 2019 (NCBI Bookshelf NBK542783). World Health Organization. WHO Guidelines on Physical Activity and Sedentary Behaviour. Geneva: WHO, 2020 (NCBI Bookshelf NBK566048). Iso-Markku P, Kujala UM, Knittle K, Polet J, Vuoksimaa E, Waller K. "Physical activity as a protective factor for dementia and Alzheimer's disease: systematic review, meta-analysis and quality assessment of cohort and case-control studies." British Journal of Sports Medicine, 2022;56:701–709. doi:10.1136/bjsports-2021-104981. Feter N, Iso-Markku P, Markarian T, et al. "Domain-specific physical activity levels and risk of dementia: a systematic review and meta-analysis." The Lancet Public Health, 2026. doi:10.1016/S2468-2667(26)00142-8. Li Y, Li Y, Liu Y, et al. "Association of long-term physical activity levels with dementia risk and cognitive function in US adults: a prospective cohort study." The Lancet Public Health, 2026. doi:10.1016/S2468-2667(26)00123-4. Reparaz-Escudero I, Izquierdo M, Bischoff-Ferrari HA, Martínez-Lage P, Sáez de Asteasu ML. "Effect of long-term physical exercise and multidomain interventions on cognitive function and the risk of mild cognitive impairment and dementia in older adults: a systematic review with meta-analysis." Ageing Research Reviews, 2024;100:102463. doi:10.1016/j.arr.2024.102463. Lamb SE, Sheehan B, Atherton N, et al. "Dementia And Physical Activity (DAPA) trial of moderate to high intensity exercise training for people with dementia: randomised controlled trial." BMJ, 2018;361:k1675. doi:10.1136/bmj.k1675. Forbes D, Forbes SC, Blake CM, Thiessen EJ, Forbes S. "Exercise programs for people with dementia." Cochrane Database of Systematic Reviews, 2015;(4):CD006489. doi:10.1002/14651858.CD006489.pub4. Del Pozo Cruz B, Ahmadi M, Naismith SL, Stamatakis E. "Association of daily step count and intensity with incident dementia in 78,430 adults living in the UK." JAMA Neurology, 2022;79(10):1059–1063. doi:10.1001/jamaneurol.2022.2672. Xu W, Wang HF, Wan Y, Tan CC, Yu JT, Tan L. "Leisure time physical activity and dementia risk: a dose-response meta-analysis of prospective studies." BMJ Open, 2017;7:e014706. doi:10.1136/bmjopen-2016-014706. Raichlen DA, Aslan DH, Sayre MK, et al. "Sedentary behavior and incident dementia among older adults." JAMA, 2023;330(10):934–940. doi:10.1001/jama.2023.15231. Erickson KI, Voss MW, Prakash RS, et al. "Exercise training increases size of hippocampus and improves memory." PNAS, 2011;108(7):3017–3022. doi:10.1073/pnas.1015950108.