The cited study is real and correctly attributed. Within the prospective population-based Rotterdam Study, serum magnesium levels were measured in 9,569 participants free from dementia at baseline (1997 to 2008), who were followed for incident dementia determined by DSM-III-R criteria until January 1, 2015, using Cox proportional hazard regression to associate quintiles of serum magnesium with incident all-cause dementia, with the third quintile as reference, adjusted for age, sex, cohort, educational level, cardiovascular risk factors, kidney function, comorbidities, other electrolytes, and diuretic use.

The central finding is not what the post describes. Both low serum magnesium levels (≤0.79 mmol/L) and high serum magnesium levels (≥0.90 mmol/L) were associated with an increased risk of dementia (hazard ratio 1.32, 95% CI 1.02-1.69, and HR 1.30, 95% CI 1.02-1.67, respectively).

Both low and high levels of serum magnesium are associated with dementia risk, suggesting a U-shaped relationship.

On Alzheimer's disease specifically, which is the post's headline claim, the study did not find a statistically significant result. Sensitivity analysis showed a trend for increased risk for Alzheimer's disease among those in the first quintile of magnesium levels (HR 1.28; 95% CI 0.97-1.69) and those in the fifth quintile (HR 1.21; 95% CI 0.92-1.58). However, these rates were not statistically significant.

The lead author framed the findings as preliminary. "These results need to be confirmed with additional studies, but the results are intriguing," said first author Dr. Brenda Kieboom. The author was also surprised by the high-magnesium finding: investigators expected a linear association and were surprised that a high serum magnesium level was linked to dementia risk, with Kieboom noting "we do not have a biological explanation yet on how this would work."

The broader evidence base does not support supplementation as a dementia prevention strategy. A systematic review of three RCTs and 12 cohort studies found that evidence from the limited number of RCTs was insufficient to draw conclusions on the effects of magnesium supplements, and cohort studies showed inconsistent dose-response relationships between dietary magnesium and cognitive disorders with high heterogeneity.

Consistent U-shaped associations of serum magnesium with all-cause dementia and cognitive impairment were found in cohorts, suggesting an optimal serum magnesium concentration of approximately 0.85 mmol/L.

Magnesium is absent from the leading consensus list of dementia risk factors. The 2024 Lancet Commission report reaffirms dementia risk reduction by addressing twelve existing risk factors: low education, hearing loss, hypertension, smoking, obesity, depression, physical inactivity, diabetes, excessive alcohol consumption, traumatic brain injury, air pollution and social isolation , with high cholesterol and vision loss added as the thirteenth and fourteenth factors . Magnesium status appears nowhere on that list.

METHODOLOGY AND CONTEXT

• Design: Prospective population-based observational cohort. Not a trial. Cannot establish causation.

• Sample: 9,569 participants, mean age 64.9 years, 56.6% women.

• Follow-up: Median 7.8 years, with 823 participants diagnosed with all-cause dementia.

Of those, 662 were diagnosed with Alzheimer's disease.

• Exposure measurement: Single baseline serum magnesium measurement.

• Reference group: The third quintile, 0.84 to 0.85 mmol/L.

• Stated limitations by the authors: Magnesium levels were only measured at the start of the study and could have changed during follow-up, and "serum magnesium levels do not necessarily represent total body magnesium. There can still be a magnesium deficiency if serum magnesium levels are normal; therefore, misclassification could have occurred."

• Critical threshold context: The study's "low" category was ≤0.79 mmol/L. Clinically, hypomagnesemia is defined as a serum magnesium concentration below the normal range, for example less than 0.7 mmol/L in adults, which may vary among clinical laboratories. The study's low quintile therefore sits largely inside the normal reference range and does not correspond to clinical magnesium deficiency.

• Prior evidence base at the time: Data in humans had been limited to a few small case-control studies showing conflicting results, plus a small randomized trial in participants with mild cognitive impairment , and previous experimental data had shown magnesium "might have a protective effect on learning in rats with dementia."

• Related cohort: The ARIC study of 12,040 participants over a median 24.2 years reported that low midlife serum magnesium was associated with a 24% higher risk of incident dementia, while the Copenhagen General Population Study reported an association between extreme plasma magnesium concentrations and vascular-related non-Alzheimer's dementia.