Dietary Magnesium and Migraine in Adults

A Cross-sectional Analysis of the National Health and Nutrition Examination Survey 2001-2004

Margaret Slavin PhD, RDN; Huilun Li MS; Manisha Khatri BPS; Cara Frankenfeld PhD


Headache. 2021;61(2):276-286. 

In This Article


This study analyzed the dietary, supplemental, and total intake of magnesium in individuals 20- to 50-years old with migraine, as well as their association with migraine according to these different sources. Women with migraine consumed less dietary and total magnesium than women without migraine, partially confirming our hypothesis. No significance was found regarding levels of supplement intake alone for any analysis. Furthermore, the highest consumers of dietary and total magnesium intake had lower odds of migraine.

The RDA was used as a benchmark of magnesium adequacy in this analysis. The dietary and total magnesium intake of the majority of participants fell below their RDA; fewer than one in four of all participants met their RDA through diet and 1 in 3 met their RDA through diet plus supplements, and even fewer individuals with migraine reached this benchmark. These low intake levels are consistent with other observations of low magnesium intake in the United States.[35,36] As compared to the RDA, the individuals in Q1 were consuming one-third to one-half of their recommended magnesium intake, whereas the individuals in Q4 were near or above their RDA for magnesium. Individuals who achieved their RDA through a combination of diet and supplements had lower odds of migraine in the adjusted model.

The RDA is a nutrient intake reference value which is estimated to meet the needs of 98% of a specific age-gender group in the United States by the National Academy of Medicine (formerly the Institute of Medicine) and is generally considered as a goal for dietary planning of individuals.[37] Specifically, the magnesium RDA was set using results from balance studies, which involve highly controlled dietary conditions designed to achieve zero balance, where magnesium intakes (food, water, supplements, pharmaceuticals) are tuned to approximately equal magnesium losses (urine, feces, sweat, dermal losses);[37] the RDA is therefore set at a level below which magnesium deficiency would occur, but this level is not necessarily the optimal level of consumption for chronic disease prevention. It is not clear whether the RDA for magnesium is sufficient to support migraine prevention, but it is reasonable to suspect that levels below the RDA might put individuals at risk for migraine related to magnesium deficiency, which is further supported by our results.

Magnesium has long been of interest for the prevention of migraine, due to the large number of proposed mechanisms where its bioactivity overlaps with migraine pathophysiology. Such proposed mechanisms include magnesium's role in blocking the NMDA receptor and therefore blocking cortical spreading depression induced by glutamate, limiting vasoconstriction actuated through serotonin, regulating release of inflammatory mediators CGRP and substance P, influencing production of nitric oxide, and contributing as a co-factor in energy metabolism.[4,5,38,39] However, there is no easily accessible clinical measurement of magnesium status, complicating efforts to understand the impact of magnesium status on migraine outcomes.

Serum magnesium levels are tightly controlled by absorption in the intestine, resorption in the kidneys, and deposition/resorption in the bone; prolonged depletion is possible before changes in serum magnesium levels are apparent, limiting its utility as an indicator of the body's long-term magnesium status.[40] The accepted serum magnesium range of 1.82–2.32 mg/dL may be too low to detect chronic latent magnesium deficiency (CLMD), a subclinical deficiency of intake that is not apparent in serum levels due to resorption from the bone,[36,41] and it has been suggested that a serum magnesium level up to 2.06 mg/dL may indicate some level of deficiency.[41] The magnesium load test is believed to be a more reliable method for identifying a deficiency, but the required 24-hour urine collection is cumbersome for routine use,[42] while spot check urinary magnesium reflects short-term consumption only, and is suggested to be effective for use in population studies but not individuals.[40] There is some evidence to suggest that erythrocyte magnesium may be a more relevant accessible value than serum magnesium in assessing patients with migraine,[6] while the serum magnesium/calcium quotient may be a more sensitive indicator of magnesium status more generally.[42] These barriers to assessing magnesium status highlight the importance of dietary assessment as a means of understanding an individual's magnesium status.

CLMD is suspected in many U.S. adults and is also associated with development of other chronic diseases, including type 2 diabetes mellitus, metabolic syndrome, cardiovascular disease, and osteoporosis.[36] Adequate magnesium intake therefore has additional relevance, as many of these conditions are comorbid with migraine.

Meanwhile, the appropriateness of the RDA levels has been questioned on the basis of new balance studies, which suggest that current RDA levels may be falsely high.[41] They instead suggest that dietary intake above 250 mg/day for a 70 kg adult may be sufficient to prevent CLMD, although needs vary with weight.[41] This value roughly corresponds to the lowest two Qs of dietary magnesium consumption in the current study, which had mean intakes of 140 and 229 mg/day, suggesting that many U.S. adults would still be considered to be consuming inadequate magnesium even with this lower target magnesium intake level. As such, the interpretation of the overall trend detected here, where lower intake of magnesium is associated with higher odds of migraine, remains unchanged in light of this questioning of the RDA.

The study is strengthened by NHANES' use of well-validated dietary assessment methodology, large sample size, extensive dataset allowing for intentional selection of inclusion/exclusion and confounding variables, and multi-year, stratified, clustered four-stage sampling design. We recognize that the exclusions and sample size prevent the sample from being fully representative of the U.S. population; however, the sample does provide strong external validity to the U.S. population between the ages of 20 and 50.

Several limitations of the study warrant acknowledgment and consideration in interpreting the results. First, the cross-sectional nature of this work means that a temporal relationship between magnesium intake and migraine cannot be established. It is possible that the presence of migraine disease drove choices that influenced magnesium intake, by increasing or decreasing intake. For example, the data from this analysis were collected between 2000 and 2004, while the American Academy of Neurology evidence-based guidelines issued in September 2000 provided a mixed review of magnesium, having classified magnesium as Grade B ("probably effective") for migraine prevention, but the clinical impression of effect was interpreted that "few people get clinically significant improvement."[43] Given the weak recommendation, the nascent stage of migraine dietary supplement research at the time of data collection, and the similar rates of magnesium supplementation between migraine and control groups, we estimate the likelihood of this influence to be low on this data. Other statistical considerations include the smaller sample size present in the sub-analyses (by gender or supplement-only analysis), which may have resulted in a loss statistical power. Also, given the differences observed between the migraine and control group across gender, socioeconomic status, BMI, and alcohol consumption, it remains possible that there is residual confounding after adjustment in multivariate models. Individuals designing new data collection may consider these factors in future studies in a more systematic way.

Only adults within the ages of 20–50 years were included in the study and the data are not generalizable to adults beyond this range. Also, the assessment of migraine was limited to a single question about migraine and severe headache in the past 3 months, and the prevalence of individuals with migraine in the current analysis was 24.8%, which is higher than the migraine prevalence rates in the United States, 20.7% of females and 9.7% of males.[44] Furthermore, data are not available on other features of the participants' migraines, such as severity or the presence of aura and other symptoms. Nonetheless, previous analyses have supported the agreement of this migraine assessment with criteria for migraine and probable migraine,[17] and the demographics of the migraine population are similar to previous works.[24–32] Others have published on migraine with this dataset on the basis of supportive data from the American Migraine Prevalence and Prevention study, as discussed in the methods section.[7,16,45] These data therefore provide meaningful insight where there is otherwise a dearth of epidemiological data connecting diet and migraine. This study counts elemental magnesium consumption levels; however, multiple properties of the diet and supplements influence the bioavailability of magnesium which cannot be accounted for in this analysis.[46] Finally, the age of the data (collected 2001–2004) is also of note; however, it is used because migraine data were not collected in NHANES after 2004. While the data may not fully reflect current food consumption patterns, magnesium was named a "shortfall nutrient" in the most recent 2015–2020 Dietary Guidelines for Americans, highlighting that there remains a high prevalence of inadequate intake of magnesium in the U.S. population.[47]

In the current analysis, the presence of low dietary magnesium levels in a majority of individuals with migraine, combined with prior evidence that magnesium supplementation at levels below the RDA has shown prophylactic efficacy, is suggestive that a significant proportion of individuals with migraine may have an underlying magnesium deficiency. It remains highly unlikely that any level of dietary magnesium would be sufficient to prevent all migraine headaches for 3 months (as measured in the migraine variable here) in those individuals with well-established disease patterns. Thus, the detection of different odds of migraine here is suggestive of a protective effect of dietary magnesium. These observations signal important research questions for the future, including whether sufficient magnesium intake can prospectively influence development and/or progression of migraine disease—and if so—what level of intake is sufficient, as well as whether background magnesium status influences response to magnesium supplementation for migraine prophylaxis.