did the cochrane review pool 17 anti-amyloid trials of 18 months each with 20,342 people?
“this cockrine analysis looked at 20,342 individuals in 17 different studies 18 months each study”
a guest, a neurologist said on The Diary of a CEO, 2026-11-12, at 35:58. hear it in context → the speaker cited: a Cochrane analysis that was just published.
first, the part that holds
the review is real, it is a cochrane review of amyloid-beta-targeting antibodies, and two of its headline numbers match exactly: 17 studies and 20,342 participants. it was published on 16 april 2026, so "just published" is a fair description seven months later.
what we found when we went and looked
we read the review's abstract through its pubmed record, along with cochrane's plain-language summary and press release. the review included "17 studies with 20,342 participants," exactly as the neurologist said. the "18 months each study" part doesn't hold up. the abstract says "eleven studies lasted 18 months, four lasted 24 months, and two lasted more than 24 months." 18 months is the review's main analysis timepoint, not the length of every trial. at that timepoint each pooled result also drew on a subset of trials and people: 13 studies and 9,895 people for cognition, 9 studies and 8,053 people for dementia severity. none pooled all 20,342 people.
assertion by assertion
| assertion | verdict | what the record says |
|---|---|---|
| a cochrane review of anti-amyloid alzheimer's drugs was recently published | confirmed | the cochrane review of amyloid-beta-targeting monoclonal antibodies for people with mild cognitive impairment or mild dementia due to alzheimer's disease (cd016297) was published 16 april 2026, with evidence current to 7 august 2025. the drugs in scope were aducanumab, bapineuzumab, crenezumab, donanemab, gantenerumab, lecanemab, ponezumab, remternetug and solanezumab. [1][2][3] |
| the review included 17 studies | confirmed | abstract: "overall, we included 17 studies with 20,342 participants." the press release also says "17 clinical trials." [1][2][3] |
| the review included 20,342 participants | confirmed | 20,342 matches the abstract, the plain-language summary ("involved 20,342 people") and the press release exactly. [1][2][3] |
| each study lasted 18 months | misattributed | the abstract says "eleven studies lasted 18 months, four lasted 24 months, and two lasted more than 24 months." the review admitted trials lasting at least 12 months and analysed data at 12, 18, 24 and over 24 months. 18 months is the timepoint its main conclusions rest on, so the number is real, but it describes the analysis, not every trial. six of the 17 trials ran longer than 18 months. [1][2][5] |
| (implied framing) the 18-month results pool all 20,342 participants | misattributed | each 18-month result used a subset. cognition: 13 studies, 9,895 people. dementia severity: 9 studies, 8,053 people. functional ability: 3 studies, 3,478 people. any brain swelling (aria-e): 11 studies, 13,595 people. 20,342 is the total across all included trials, not the number of people behind any single 18-month estimate. [1] |
both directions
for the speaker: the headline numbers are right: 17 studies and 20,342 participants. the review's own conclusions are built on 18-month results: "the effect ... on cognitive function and dementia severity at 18 months is trivial, while on functional ability, it is small at best." eleven of the 17 trials did last 18 months, which is the most common length. so "18 months" is the review's central number, just attached to the wrong thing. the plain-language summary also says the studies "did not last very long," which fits the neurologist's broader point that the trial evidence is short-term.
against: "18 months each study" contradicts the abstract. six of the 17 trials ran 24 months or longer, and the review analysed 12-, 24- and over-24-month data as well. the speech also implies that 20,342 people fed the 18-month verdict. they didn't: the cognition estimate used 9,895 people and the function estimate 3,478. a further point applies to the review rather than the speaker. published commentaries argue that pooling first-generation antibodies that failed with the newer ones that won approval hides real differences between drugs. one notes high heterogeneity on dementia severity (i² 67%, and 76.7% for subgroup differences). another faults the review for leaning on fixed 18-month estimates.
what this cannot say
the full review on cochranelibrary.com blocked our fetch (http 403). we read the abstract through the pubmed record, plus cochrane's plain-language summary and press release, so we did not check trial-by-trial durations in the characteristics-of-studies tables. we are relying on the abstract's 11/4/2 breakdown. we checked one quoted sentence against a timestamp as described, not the full episode, so surrounding context could change how the line is read. this check covers the numbers only. it does not judge whether the class-level pooling is the right way to read the evidence, which is under live debate in the published literature. nothing here is advice about any person's treatment.
sources, each one fetched and read
- cochrane review authors, amyloid-beta-targeting monoclonal antibodies for people with mild cognitive impairment or mild dementia due to alzheimer's disease, cochrane database of systematic reviews, 2026 apr 16;4(4):CD016297, doi:10.1002/14651858.CD016297 (pubmed 41985900 abstract) What we read there: "overall, we included 17 studies with 20,342 participants." "eleven studies lasted 18 months, four lasted 24 months, and two lasted more than 24 months." "we analysed data at 12, 18, 24, and over 24 months of treatment." the 18-month cognition result used 13 studies and 9,895 people; dementia severity, 9 studies and 8,053; functional ability, 3 studies and 3,478; any aria-e, 11 studies and 13,595 (107 more per 1,000). Fetched 2026-11-13 (HTTP 203). https://pubmed.ncbi.nlm.nih.gov/41985900/
- cochrane, plain language summary of CD016297, 2026 What we read there: "we found 17 studies that were carried out in different countries and involved 20,342 people." the studies "did not last very long"; main results are reported at 18 months; evidence current to 7 august 2025; published 16 april 2026. Fetched 2026-11-13 (HTTP 200). https://www.cochrane.org/evidence/CD016297_are-medicines-anti-amyloid-monoclonal-antibodies-reduce-build-abnormal-proteins-brain-effective
- cochrane news release, anti-amyloid alzheimer's drugs show no clinically meaningful effect, 16 april 2026 What we read there: "the new review examined data from 17 clinical trials with a total of 20,342 participants." effects were "absent or trivial, falling well below established thresholds for the minimum clinically important difference." Fetched 2026-11-13 (HTTP 200). https://www.cochrane.org/about-us/news/anti-amyloid-alzheimers-drugs-show-no-clinically-meaningful-effect
- commentary authors, analytical framing of amyloid beta monoclonal antibody trials in alzheimer's disease can obscure clinical effects, alzheimer's & dementia, 2026;22(7):e71646 What we read there: argues the review pools "clinically and mechanistically heterogeneous interventions"; reports i² = 67% across effects and 76.7% for subgroup differences on cdr-sb. Fetched 2026-11-13 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC13408426/
- commentary authors, beyond the cochrane review: re-evaluating anti-amyloid therapy in alzheimer's disease, 2026 (pmc) What we read there: the review "synthesized data from 17 clinical trials, encompassing over 20,000 participants" and "primarily interpreted therapeutic efficacy through fixed-timepoint estimates, most notably at the 18-month mark." Fetched 2026-11-13 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC13494676/
speakers are described by role, not named. one sentence is quoted under fair use, with attribution and a link to the original at the timestamp; we do not re-host audio or video. nothing on this page is a reason to start, stop or change a medication. if you have the file behind anything marked unverifiable, the research board is where people can answer. adam