Compléments alimentaires, confrontés aux preuves
Magnesium
Solidité des preuves: Faible
Mis à jour 2026-08-07
Ce guide n’est pas encore traduit dans votre langue ; il est affiché en anglais.
Réponse courte
Does magnesium help insulin resistance?
Modestly, and mostly if you are short of it. Pooled trial data show magnesium lowering fasting glucose in people with diabetes and improving two-hour glucose tolerance in people at high risk. The effect on insulin resistance itself did not reach statistical significance, and trials shorter than about four months tend to show nothing.
Avant d’en prendre
Parlez-en à votre prescripteur avant d’ajouter ceci à un médicament, en particulier hypoglycémiant. N’arrêtez jamais un traitement prescrit au profit d’un complément.
Interactions
Mildly additive with glucose-lowering medication. The effect is small — smaller than berberine's — but it is real enough to have shown up in pooled trials, so it belongs on the list rather than being rounded to nothing.1
Prudence
Reduces absorption of tetracycline and quinolone antibiotics, and of bisphosphonates. Separate the doses by several hours rather than stopping either.
Prudence
Proton pump inhibitors taken long-term lower magnesium, and some diuretics increase its loss. If you take either, magnesium status is worth measuring rather than guessing at.
Prudence
Qui ne doit pas en prendre
Avoid supplementing without medical supervision if you have impaired kidney function. The kidneys clear magnesium, and when they cannot, it accumulates to dangerous levels.
À éviter
Loose stools are the usual dose-limiting effect, and they arrive well before any metabolic benefit would. Oxide is the worst offender; citrate and glycinate are better tolerated.
Prudence
Notre verdict
A reasonable thing to correct if you are deficient, and a poor bet as a glucose intervention on its own. Cheap enough that the downside is small — but "cheap and harmless" is not the same as "works".
Ce que nous pouvons dire et vendre chez vous
Les allégations de santé pour ces produits ne sont pas autorisées dans votre région. Cette page ne comporte donc aucun lien commercial : elle est purement informative.
Aucun lien commercial n’est affiché dans votre région.
Comment lire les étiquettes de preuve
Chaque affirmation de cette page porte une étiquette indiquant le poids qu’elle peut supporter. Elle décrit la solidité du résultat, non le prestige de la revue.
- Preuves solidesUn grand essai randomisé, une cohorte suivie longtemps, ou la concordance de plusieurs d’entre eux.
- Preuves émergentesRéel, mais mince — petit, court, observationnel ou non encore répliqué.
- ContestéDes chercheurs compétents sont en désaccord. Nous ne tranchons pas à votre place.
What the pooled trials show
A meta-analysis of eighteen double-blind randomised trials split the question by population. In people who already had diabetes, magnesium lowered fasting glucose. In people at high risk but not yet diabetic, it improved glucose two hours into an oral glucose tolerance test.
The measure most people actually care about — insulin resistance, as HOMA-IR — moved in the right direction but not significantly, and the trials disagreed with each other substantially.
Ce que disent les preuves
Across 18 double-blind randomised trials, magnesium supplementation reduced fasting plasma glucose in people with diabetes.1
Preuves émergentes
In people at high risk of diabetes, magnesium improved two-hour glucose after an oral glucose tolerance test.1
Preuves émergentes
The effect on insulin resistance measured by HOMA-IR was a non-significant trend, with the confidence interval crossing zero and high inconsistency between trials.1
Preuves émergentes
Why the timescale matters
Magnesium is not a supplement whose effect you will watch appear on a sensor over a fortnight. Where benefits show up in trials, they tend to require months of supplementation, which is longer than most people persist and far longer than the wear of a single CGM sensor.
If you are wearing a sensor to see whether magnesium is doing something, you will almost certainly conclude it is not — regardless of whether it eventually would.
What the label can and cannot say
Magnesium is a nutrient rather than a botanical, so unlike most of this page it holds authorised claims in the EU — for contributing to normal energy-yielding metabolism, muscle function and psychological function, among others.
None of those authorised claims is about glucose control or insulin sensitivity. A product marketed to you for blood sugar is using the general permission to imply a specific effect the regulator has not approved.
Questions fréquentes
- Which form of magnesium is best?
- For tolerability, citrate or glycinate over oxide, which is poorly absorbed and most likely to cause loose stools. No form has been shown superior for glucose outcomes specifically.
- Should I take magnesium if my levels are normal?
- The trials that show benefit are mostly in people with diabetes or at high risk, and the mechanism is plausibly repletion. If you are not short of it, expect less.
- How long before I would see anything?
- Months rather than weeks in the trial data. A two-week sensor wear is the wrong instrument for answering this question.
Références
- 1.Veronese N, Watutantrige-Fernando S, Luchini C, Solmi M, Sartore G, Sergi G, et al. (2016). Effect of magnesium supplementation on glucose metabolism in people with or at risk of diabetes: a systematic review and meta-analysis of double-blind randomized controlled trials. European Journal of Clinical Nutrition. 10.1038/ejcn.2016.154