The press release left Emory in late August with a headline built to travel: high-dose vitamin D may improve cognition among those at risk of dementia. Underneath it sat fifty-four people, most of whom had done little more than answer a questionnaire about the pills they already take. What the headline left out is the number that makes the study worth a second look: the big randomized trials that found vitamin D does nothing for memory tested 2,000 IU a day, and the signal here showed up only at 5,000.

DOSE ON THE TABLE
2,000IU/day
Big null trials (VITAL)
5,000IU/day
Emory signal
The randomized trials that found no cognitive benefit tested less than half the dose where the new signal appeared. Source: VITAL (2021); Pak et al., Sleep Medicine (2026)

Start with what Victoria Pak’s team at the Nell Hodgson Woodruff School of Nursing actually did. They took older adults carrying two problems at once, disturbed sleep and mild cognitive impairment, the pairing that often sits at the leading edge of dementia. They asked each person how much vitamin D they were taking, then scored everyone on the Montreal Cognitive Assessment, a thirty-point screen used to flag early decline. The people reporting at least 5,000 IU a day scored more than 13 percent higher than the people taking none, and the gap held after adjusting for age, sex, body mass index, and education. Lower doses did nothing.

Read that back and the caveats arrive on their own. This is a snapshot, everyone measured once, no before and no after. The dose was self-reported, which means it rests on what fifty-four people remembered about their own habits. And a snapshot cannot tell you which way the arrow points: whether the vitamin sharpened the mind, or whether the people whose minds were still sharp were simply the sort who buy 5,000 IU capsules and remember to swallow them. “May improve” is a causal claim, and the design underneath it can support an association and not one inch more. Pak, to her credit, used the softer word herself, and her team noted that 5,000 IU sits above the 4,000 IU that regulators set as the safe daily ceiling, so any real test would have to watch for toxicity too.

So on the merits of this one study, the skeptic’s reflex is correct. Fifty-four people, one time point, a questionnaire. File it under interesting, not settled.

Except the establishment reflex, the one that treats vitamin D and the brain as a closed case, is built on ground just as soft, and nobody sends out a press release about that. The “no benefit” verdict comes mostly from VITAL, the large randomized trial whose cognitive substudies followed thousands of older adults and found a pooled difference in cognitive decline of 0.01, at a p-value of 0.39. Flat. Last year the VitaMIND randomized trial added another null, and three prior randomized trials had already come up empty. That is a real body of evidence, and it is the reason a low-dose vitamin D recommendation for memory would be irresponsible.

But look at what those trials gave people. VITAL used 2,000 IU a day, and the Emory signal appeared only at 5,000, so the randomized trials that found nothing tested less than half the dose. The Emory study, for its part, assigned no dose at all: it watched what fifty-four people were already swallowing and scored them once. Neither camp has run the experiment that would actually settle it.

That gap is the exact seam a real experiment is supposed to close, and no one has closed it: a pre-registered, randomized trial of 5,000 IU a day in precisely this population, older adults with disturbed sleep and emerging cognitive slippage, the “critical window” Pak described. The tools to run it are ordinary. The molecule is off patent and costs pennies. Which is where the funding math turns ugly. The drugs getting the billion-dollar trials and accelerated approval in this disease are the amyloid antibodies, and lecanemab, sold as Leqembi, is the template: it clears plaque, carries a documented risk of brain swelling and brain bleeding, and lists at $26,500 a year before the genetic tests and repeat brain scans its safety profile demands. The cheapest thing on the shelf got a fifty-four-person pilot funded through two National Institute on Aging grants and a headline that oversold it, because a careful description would not travel.

LECANEMAB LIST PRICE
$26,500per year, before scans and monitoring
The amyloid antibody's annual price for early Alzheimer's; the off-patent vitamin costs pennies. Source: Alzheimer's Association, 2026

Pak wrote “may.” Her institution’s headline used the same word and set it in type large enough to imply something firmer. The honest sentence and the marketed one meant different things, and no randomized trial on the record has tested 5,000 IU a day in the exact population where the signal turned up. The study that would tell the two sentences apart is cheap, obvious, and still unfunded, while the antibody it would compete against bills by the tens of thousands.

Sources

  1. Sleep Medicine – Pak et al., “Vitamin D supplement intake is associated with better cognition in persons with sleep disturbance and mild cognitive impairment” (2026)
  2. Scientific Reports – “Effect of vitamin D on cognitive decline: results from two ancillary studies of the VITAL randomized trial” (2021)
  3. JAMDA – “Impact of Vitamin D Supplementation on Cognition in Adults With Mild to Moderate Vitamin D Deficiency: Outcomes From the VitaMIND Randomized Controlled Trial” (2025)
  4. Emory University – “High-dose vitamin D may improve cognition among those at risk of dementia” (2026)
  5. ScienceDaily – “This common vitamin was linked to 13% better cognitive scores” (2026)
  6. Discover Magazine – coverage with dose thresholds and adjusted covariates (2026)
  7. Alzheimer’s Association – Lecanemab (Leqembi) approved for early Alzheimer’s: risks and cost (2026)