Every finding in the evidence table: 382 rows, filterable

Every row

Filter

Area All 382 Care, carers and services 140 Medicines 85 Genes (not modifiable) 35 All modifiable factors together 27 Diet and supplements 23 Exercise and activity 21 Sleep, mood and social life 19 Hearing and vision 8 Alcohol, tobacco and cannabis 6 Infection and vaccination 5 Heart and metabolic health 4 Injury, delirium and falls 4 Environment 3 Cognitive activity 2

Question Both Prevention 190 Treatment 192

Direction Any Benefit 144 Harm 105 No effect 38 Mixed 37 No direction reported 58

Design Any meta-analysis 57 randomised trial 44 phase 3 trial 15 systematic review 9 phase 2 trial 10 cohort study 136 phase 1 trial 2 cross-sectional study 31 case-control study 14 modelling study 20 regulatory decision 7 company announcement 1 coverage determination 1 None 35

Showing 8 of 382 rows. One row per study per exposure: a study that tested two different things appears twice, once for each.

Scroll the table sideways to see every column.

Research and outcomes, 8 rows. Every row links to its source.
Exposure	Question	Direction	Outcome measured	Design	People	Effect size, as the source reported it	Source
Anti-amyloid antibody Medicines	Treatment	Mixed why A statistically detectable but very small cognitive effect (ADAS-Cog SMD -0.11) alongside 107 more cases of ARIA-E per 1,000. The reviewers' own conclusion is that removing amyloid does not seem to produce clinically meaningful effects — the sharpest disagreement in this table with how the individual trials were reported.	ADAS-Cog and CDR-SB at 18 months; ARIA	meta-analysis	20,342	ADAS-Cog SMD -0.11 (95% CI -0.16 to -0.06); ARIA-E absolute risk difference +107 per 1,000	Cochrane Database of Systematic Reviews 2026 ✓ Peer-reviewed PMID 41985900
Anti-amyloid antibody Medicines	Treatment	Mixed why CHMP refused the drug, then on re-examination recommended it only for people with one or no APOE4 copy, where the ARIA rates are lower. Benefit on cognition and harm from ARIA in the same document, with the restriction as the regulator's way of balancing them.	CDR-SB at 18 months and ARIA rates, in the APOE4-restricted group	regulatory decision	1,521	"After 18 months of treatment, patients treated with Leqembi had a smaller increase in CDR-SB score than those who received placebo (1.22 versus 1.75)"; "8.9% of those with only one or no copy of ApoE4 experienced ARIA-E, compared with 12.6% of all patients ... 12.9% of patients in the restricted population experienced ARIA-H compared with 16.9% of the broader population"; placebo figures 1.3% and 6.8%	European Medicines Agency 2024 ⚠ Regulatory decision — not a study no DOI or PMID exists
Anti-amyloid antibody Medicines	Treatment	Mixed why Benefits were statistically present but every one fell below the minimum clinically important difference, against a number needed to harm of 9 for ARIA-oedema. Agrees with the Cochrane review and not with the trial reports.	Cognitive and functional scales against the MCID; ARIA	meta-analysis	23,202	CDR-SB mean difference -0.18 (95% CI -0.34 to -0.03), below the MCID; ARIA-oedema NNH 9	Annals of Family Medicine 2024 ✓ Peer-reviewed PMID 38253509
Anti-amyloid antibody Medicines	Treatment	Mixed why Approved on amyloid reduction (surrogate) while reporting higher ARIA rates than placebo; the notice reports no cognitive outcome.	Amyloid PET (surrogate marker); ARIA rates	regulatory decision	not stated	not established — the library recorded "aria-e in 10% of lecanemab patients vs 1% placebo; aria-h in 6% vs 5%". investors.biogen.com timed out on repeated attempts…	U.S. Food and Drug Administration (announced via Biogen press release) 2023 ⚠ Regulatory decision — not a study no DOI or PMID exists
Anti-amyloid antibody Medicines	Treatment	Mixed why Two identically designed trials disagreed: EMERGE met its primary endpoint, ENGAGE did not.	CDR-SB at 78 weeks	phase 3 trial	3,285	EMERGE: CDR-SB difference -0.39 vs. placebo (95% CI -0.69 to -0.09; P=.012); ENGAGE: CDR-SB difference +0.03 vs. placebo (95% CI -0.26 to 0.33; P=.833)	Journal of Prevention of Alzheimer's Disease 2022 ✓ Peer-reviewed PMID 35542991
Antihypertensives / BP lowering Medicines	Prevention	Mixed why Antihypertensive treatment improved cognitive scores by a small amount (SMD 0.06, moderate certainty) but the reduction in dementia incidence was not statistically significant (OR 0.89, low certainty). The drug-class rankings rest on one trial per class and are described as highly uncertain.	Cognitive function scores and incident dementia in hypertensive older adults (four RCTs)	meta-analysis	16,823	Antihypertensive treatment significantly improved cognitive function scores (SMD, 0.06; 95% CI, 0.03-0.09; P < .001, GRADE moderate certainty) and showed a nonsignificant trend toward reduced dementia incidence (OR, 0.89; 95% CI, 0.76-1.05; P = .172, GRADE low certainty)	Journal of the American Medical Directors Association 2026 ✓ Peer-reviewed PMID 41962914
Antihypertensives / BP lowering Medicines	Prevention	Mixed why Antihypertensive use tracked lower Alzheimer's risk but not lower non-Alzheimer dementia risk, where treated and untreated hypertension looked the same. A single baseline blood pressure reading predicted neither.	Alzheimer's and non-Alzheimer dementia	meta-analysis	31,250	Untreated vs treated hypertension: Alzheimer's dementia HR 1.42 (95% CI 1.08-1.87)	Neurology 2024 ✓ Peer-reviewed PMID 39141884
Metformin / antidiabetic Medicines	Prevention	Mixed why The split inside this one paper is the finding: the 41 observational studies show large benefits (SGLT2i OR 0.56, GLP-1RA OR 0.58) while the 23 randomised trials show no difference. When design and result disagree this cleanly, the trials are the ones to believe.	Incident dementia	meta-analysis	3,462,926	Observational: SGLT2i OR 0.56 (95% CI 0.45-0.69), GLP-1RA OR 0.58 (0.46-0.73) vs non-users; RCTs: no difference	Alzheimer's Research & Therapy 2024 ✓ Peer-reviewed PMID 39716328

How this table was built, and where it is weak →

Every row of the evidence table

Every row