Digestery
Everyday supplements › Magnesium and Sleep › Studies

Pooled review (meta-analysis)

Pooled review of 3 trials: magnesium shortened falling asleep, but total sleep unclear

Mah and colleagues · BMC complementary medicine and therapies · 2021

Original title: Oral magnesium supplementation for insomnia in older adults: a Systematic Review & Meta-Analysis.

Under review. A person has not yet checked this summary against the paper. It is written from the PubMed abstract. Until a person checks it, search engines are asked not to list this page.

What the study found

Mixed

Time to fall asleep was 17.36 min less with magnesium than placebo (95% CI - 27.27 to - 7.44). Total sleep time rose 16.06 min, not statistically clear. All trials had moderate-to-high bias; evidence low to very low quality.

The study at a glance

Type of study
Pooled review (meta-analysis)
Who or what was studied
3 studies, 151 people · older adults with insomnia
What was tested
Magnesium by mouth (under 1 g, up to three times a day) vs placebo or no treatment
What it measured
The question itself: sleep quality and insomnia
Who has a stake
No stake declared
No competing interests declared.
Journal
BMC complementary medicine and therapies, 2021
Source
PubMed record 33865376. This page is based on the abstract only.

How much this study counts

  • ✓Pools the results of 3 studies with 151 people.
  • •A pooled review is only as strong as the studies it pools.
  • ✓Measured the question itself: sleep quality and insomnia.
  • ✓No competing interests declared.
  • •Summarized from the PubMed abstract only. The full paper may hold details the abstract leaves out.

✓ Strength– Limit• Note

These points come from fixed rules about the type of study, its size and length, what it measured, and who has a stake. Every study on Digestery gets the same rules.

Where this study sits

This paper is not on the ladder. It is a pooled review (meta-analysis), and reviews and opinion pieces discuss studies that are already counted on the steps below.

↑ Stronger evidence

  1. Trials in people: sleep quality and insomnia

    1

    1 benefit

  2. Trials in people: other measures

    6

    4 benefit, 2 no difference

  3. Trials in people: lab tests or safety

    0

    None among the papers we read.

  4. Population studies

    3

    2 mixed, 1 no difference

  5. Blood levels and genes

    3

    2 benefit, 1 no difference

↓ Weaker evidence

  • Points to benefit
  • Mixed
  • No difference
  • No direction
  • Points to harm

How it fits with the other research

The short answer to “Does magnesium help you sleep?”

Not clearly; trials disagree.

Confidence: very low. Based on 16 on-topic papers.

Read the full answer
  • Quoted in “What we found”: In three pooled trials of 151 older adults with insomnia, people on magnesium fell asleep 17.36 min sooner than on placebo; the trials had moderate-to-high risk of bias.
  • Part of the finding “Trials in poor sleepers give mixed results”.
  • Summed up in “What reviews found”: Pooled three trials in 151 older adults with insomnia. People fell asleep 17.36 min sooner on magnesium; a 16.06 min gain in total sleep was not clear. Evidence quality was low to very low.
  • Marked as a key paper in the research timeline: Pooled trials: faster sleep onset in older adults.
  • Open item for a person: Mah 2021 and Lopresti 2026 likely pool some of the same trials, possibly Nielsen 2010; check overlap before counting them as separate evidence.

Other reviews on this question

See all 16 studies on this question

How this page was made

Label

First read of the abstract
Claude Sonnet
Independent second read
Claude Haiku
Differences settled
The reviewer corrected 1 field after reading the abstract. Claude Opus 5.5.
Headline
Claude Sonnet, from the checked study record. A script checked its numbers and wording.
Numbers
Checked against the abstract by a script
“How much this study counts”
Fixed rules, the same for every study
Source check by a person
Not yet. Required before search engines may list this page.

Spot an error? Email a flag with a link to your source, or use the form on the question page.