Every row of the evidence table

Direction is for chronic kidney disease as a source of false signal: at eGFR 45 versus 60, p-tau217 was 31-55% higher in amyloid-negative people on two immunoassays, a rise largely removed by using the %p-tau217 ratio. A test-validity result, not a health outcome.

Direction is for microglial activation: amyloid was associated with astrocyte reactivity across cortical regions only when microglia were activated, and that microglia-dependent reactivity related to cognitive impairment through tau phosphorylation and aggregation.

Direction is for higher baseline plasma GFAP: it detected abnormal CSF amyloid (AUC 0.79, 0.86 with APOE4) and predicted later conversion to Alzheimer's dementia (AUC 0.84); GFAP slopes were steeper in those who progressed.

Direction is for BEING a carer: carers were worse off than non-carers on every domain, with the largest gaps in depression and stress.

Only STRAINED carers were at higher risk (RR 1.63, 95% CI 1.00-2.65, with the interval touching 1). Carers who reported no strain showed nothing (RR 1.08), so it is the strain rather than the caregiving that carries the signal.

Direction is for having AD dementia: lifetime formal-care costs of 2,440,000 SEK against 510,000 SEK for matched controls, with prescriptions and social care rising and specialist outpatient visits falling as the disease advanced. A cost result, not a health outcome.

Direction is for greater disease severity: mean annual costs rose from SEK 99,906 in MCI to SEK 795,617 in severe AD (SEK 57,625 to 668,073 in the amyloid-confirmed subset), driven by institutionalisation, home care and daytime support while outpatient costs fell. A cost result, not a health outcome.

Direction is for being on the Alzheimer's continuum: societal costs were 52% higher in subjective cognitive decline, 170% in MCI and 307% in AD dementia than in controls, and payer costs 26%, 85% and 36% higher. A cost result, not a health outcome.

Direction is for being non-Hispanic Black or Hispanic rather than non-Hispanic White: disease-course Medicare spending was higher ($165,730 and $160,442 versus $136,326), highest for Black beneficiaries in every phase and driven by high-cost services in the last year of life. A disparities result, not a biological effect of race.

Direction is for dying with dementia rather than heart disease, cancer or other causes: total cost was higher ($287,038) and out-of-pocket spending 81% higher, consuming 32% of household wealth versus 11%.

Direction is for living in the community rather than a nursing home: unadjusted costs were lower in the community, but after adjusting for functional impairment and comorbidity community care cost €11,344 more per year, mainly through informal care. A cost result, not a health outcome.

Direction is for having dementia: the attributable yearly cost per person was $41,689-$56,290 depending on how informal care is valued, a burden the authors call similar to heart disease and cancer.

Direction is for value for money at current Australian prices: donanemab improved modelled QALYs (4.38 vs 4.01) but the ICER of $342,424 per QALY exceeded commonly cited willingness-to-pay thresholds. An unfavourable value result, not a clinical harm.

Direction is for value for money: three years of lecanemab gained 0.17 QALYs and the cost-effective price at 1 million SEK per QALY was 33,886 SEK a year, which the authors say makes it unlikely to be cost effective in Sweden at current US list prices. An unfavourable value result, not a clinical harm.

Direction is for value for money at the current price: no lecanemab strategy was cost-effective against standard care alone, whichever diagnostic test was used; CSF-guided targeted treatment would become cost-effective below $5,100 per year. An unfavourable value result, not a clinical harm.

Direction is for chronic bi-weekly intravenous infusion relative to oral treatment: administration and monitoring cost $45,208 plus $6,095 caregiver burden, eroding 44% of treatment value, against 2% for an oral drug. A cost result for a hypothetical treatment, not a clinical harm.

Direction is for value for money at the list price: the benchmark price ($8,900-$21,500 per year) sits below the $26,500 list price, the committee voted 15-0 that the drug represents low long-term value at that price and 12-3 that the evidence did not adequately demonstrate a net health benefit. An unfavourable value assessment, not a clinical harm.

Direction is for value for money at the assumed $28,000 annual price: ICERs of $981,000/QALY (aducanumab) and $193,000/QALY (donanemab) exceed the $150,000/QALY threshold, so neither was cost-effective. An unfavourable value result, not a clinical harm.

Direction is for a lower (worse) glymphatic index: it was lowest in AD dementia and predicted faster amyloid accumulation, amyloid-positive transition, atrophy, clinical progression and cognitive decline, with the cognitive effect fully mediated by amyloid PET and atrophy.

Direction is for co-existing Lewy body pathology in people with Alzheimer pathology: worse baseline cognition, faster global cognitive decline and more posterior hypometabolism than Alzheimer pathology alone. The seed amplification assay itself agreed closely with autopsy (96.3% sensitivity, 96.4% specificity in 61 cases).

Direction is for amyloid: it induced hyperconnectivity from tau epicentres to posterior tau-vulnerable regions, present already in preclinical AD, and this hyperconnectivity mediated amyloid's effect on faster tau accumulation.

Direction is for greater spatial extent of tau: it tracked memory in unimpaired people and executive function in MCI and dementia better than a temporal meta-ROI, and tau spread fastest in unimpaired and MCI participants. Most amyloid-positive participants followed Braak staging, with substantial regional heterogeneity within each stage.

Direction is for regional tau burden: more tau went with less glucose metabolism (r = -0.49) and with worse domain-specific cognition, while amyloid was diffuse and only weakly related. An imaging association, not an exposure effect.

Direction is for going undiagnosed. Whether the diagnosis itself would change survival is not tested here.

The primary cognitive outcome was significantly WORSE in the exercise arm (adjusted difference -1.4, 95% CI -2.6 to -0.2, favouring usual care). The authors conclude the programme does not slow cognitive impairment and treat the difference as of uncertain clinical importance. Fitness did improve. This is the sharpest counterweight to the observational exercise evidence.

The review's subject is the harm itself: brain swelling and microbleeds caused by the drugs.

Against placebo: ARIA-E 24.4% vs 1.9%. Risk tracks APOE4 status and baseline imaging.

Wider range of harms than regulatory alerts had flagged, with the highest risks soon after starting. No increase in ventricular arrhythmia.

Direction is for valacyclovir: cognition worsened more on the drug than on placebo on the primary outcome (between-group difference 3.93 ADAS-Cog points, P = .01); function, amyloid PET and tau PET did not differ. The authors conclude it is not recommended.

Direction is for tau seeding activity: substantial hippocampal and neocortical seeding was present even without overt neocortical tangles and predicted region-specific cognitive performance and longitudinal decline; seeding and modification profiles tracked neuritic plaque density and separated AD from PART.