Home › Research › What the research found › Every row
Every row of the evidence table
One row per finding: what was tested, which way the result went, and how the study was built. The answer by area, and where the studies disagree, are on the main page.
Every row
Filter
Showing 4 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.
| Exposure | Question | Direction | Outcome measured | Design | People | Effect size, as the source reported it | Source |
|---|---|---|---|---|---|---|---|
| APOE genotype Genes (not modifiable) | Prevention |
Mixed
whyDirection is for APOE genotype after 90: ε2 carriers had lower dementia risk, while the ε4 association was not statistically significant. Women and Black participants had higher risk than men and Asian participants. Age-standardised incidence was 116.82 per 1000 person-years. |
Incident all-cause dementia after age 90 by APOE genotype, sex and race | cohort study | 805 | The age-standardised incidence rate was 116·82 cases per 1000 person-years (95% CI 93·69-139·96). In Fine-Gray models, dementia risk was higher in female than in male participants (subdistribution hazard ratio [sHR] 1·89, 95% CI 1·30-2·76) and Black than Asian participants (sHR 1·75, 1·07-2·88), lower in APOE ε2 carriers than in non-carriers (sHR 0·39, 0·17-0·88), and not significantly higher in APOE ε4 carriers than in non-carriers (sHR 1·51, 0·92-2·47) |
The lancet. Healthy longevity
2026 ✓ Peer-reviewed PMID 42551464 |
| Polygenic risk / GWAS loci Genes (not modifiable) | Prevention |
Mixed
whyThe intervention-control cognitive difference was statistically significant in the higher-genetic-risk halves and not in the lower halves, but neither genetic score significantly modified the intervention effect overall (P > 0.46); an exploratory sex difference emerged for the Alzheimer score. |
Annual change in overall cognition (modified NTB) with a 2-year multidomain lifestyle intervention, by Alzheimer and coronary genetic risk score | randomised trial | 1,177 | The intervention-control difference in annual overall cognition change (95% confidence interval) for participants with AD-GRS above/below the median (i.e. higher/lower risk) was 0.032 (0.002-0.063) versus 0.017 (-0.011 to 0.045), and for CAD-GRS above/below the median was 0.031 (0.002 to 0.059) versus 0.016 (-0.012 to 0.044). AD-GRS or CAD-GRS were not significantly related to the intervention effect overall (P > 0.46), but for AD-GRS there were differences between females and males (P = 0.024) |
Brain : a journal of neurology
2026 ✓ Peer-reviewed PMID 40747850 |
| Polygenic risk / GWAS loci Genes (not modifiable) | Prevention |
Mixed
whyThe study's own results point different ways: the polygenic score added to APOE for predicting AD/MCI and for progression from amyloid-positive MCI to AD, but amyloid deposition was predicted by APOE alone. |
Prediction of AD, MCI, amyloid deposition and MCI-to-AD progression (AUC) by polygenic risk score with and without APOE | case-control study | 770 | AD and MCI are predicted by both APOE genotype and PRS (area under the curve [AUC] = 0.82% and 68%, respectively). Amyloid deposition is predicted by APOE only (AUC = 79%). Further progression to AD of individuals with MCI and amyloid-positive status is predicted by PRS over and above APOE (AUC = 67%) |
Annals of Neurology
2019 ✓ Peer-reviewed PMID 31199530 |
| Rare copy-number variants (22q11.21 dosage, ABCA1 / ABCA7 deletions) Genes (not modifiable) | Prevention |
Mixed
whyThe two kinds of variant point opposite ways: deletions and loss-of-function alleles in ABCA1 and ABCA7 and 22q11.21 deletions raised AD risk, while 22q11.21 duplication lowered it (odds ratio 0.34 at exome-wide significance in replication). Not modifiable. |
Alzheimer's disease case status (early- and late-onset) by rare copy-number variant burden and 22q11.21 dosage | case-control study | 22,319 | ABCA1 (odds ratio [OR] = 5.77 [95% confidence interval 2.25; 17.06], p = 0.0002) and ABCA7 deletions contribute to this deletion burden (OR = 2.29 [1.44; 3.65], p = 0.0006), while CTSB LoF alleles appear as candidates (OR = 5.03 [1.50; 20.71], p = 0.0089) ... Replication in 33,977 affected individuals and 362,322 control subjects confirmed association for 22q11.21 dosage with exome-wide significance (ORSCARF2 = 0.34 [0.21; 0.53]; mega-p value = 5.52 × 10-7) |
American Journal of Human Genetics
2026 ✓ Peer-reviewed PMID 42648288 |