Every row of the evidence table

Direction is for the blood test's ability to detect treatment-related amyloid clearance on PET: AUC 0.61, which the authors call suboptimal and not currently usable for that purpose. It says nothing about the drug's effect.

No significant difference in 5-year cumulative spending between spouses of people with and without dementia ($3,974, 95% CI -$3,199 to $11,477); year-5 spending was higher (+$2,748), driven by inpatient costs.

No difference from waitlist control on the primary burden outcome or on mood, WHOQOL-Bref or productivity; only the CarerQoL-VAS quality-of-life score was higher in the intervention group. The trial recruited 93 of a 194 target.

Burden fell in both arms with no significant difference between the two platforms. There was no untreated control, so the within-arm improvement cannot be separated from time or attention.

No difference between health-system care, community-based care or usual care on either primary outcome. Caregiver self-efficacy did improve in both active arms.

No benefit and no harm: screening did not change family-member quality of life, preparedness or self-efficacy, and did not raise depression or anxiety. Only 5.1% screened positive.

Difference-in-differences around ACO entry found no change in end-of-life care processes, outcomes or spending.

The primary outcome did not improve. Emergency department visits and hospitalisations were lower (RR 0.45, 95% CI 0.31-0.65), which is a secondary outcome.

Three trials comparing respite with no respite found no significant benefit on any carer outcome; the four included trials were too different to pool.

Co-primary endpoints were not met at 52 weeks. The comparator was a low dose of the same drug rather than placebo, which the authors say masked any separation; they report separation in the MCI subgroup only at 78 and 104 weeks.

The primary endpoint was not met: the low dose showed no clinically meaningful benefit and the 3 mg dose showed 32% greater progression than placebo, with more serious adverse events. Less whole-brain volume loss on MRI is a surrogate finding.

The primary clinical endpoint (CDR-SB) was not met in the overall population. The tau figures are a biomarker outcome, not a clinical one.

ADAS-Cog11 showed slower decline, but the co-primary functional endpoint (ADCS-ADL) did not differ and the authors classify the study as negative.

No difference from placebo on the primary agitation outcome or other main behavioural outcomes. As-needed psychotropic use was lower during the active period, and no serious drug-related adverse events occurred.

Primary outcome null. One secondary cognitive measure (ADAS-Exec) favoured liraglutide on an unadjusted test; ADCS-ADL and CDR-SB did not.

The trial did not meet its primary endpoint at any of three doses despite sustained target engagement (lower CSF soluble TREM2, higher osteopontin); the most frequent adverse events were MRI changes resembling ARIA.

Not effective, and drug-related QT prolongation was observed. The randomised sample was smaller than planned.

Neither choir singing nor group music therapy beat standard care. Country was the strongest predictor of the differences between sites.