Response: Re: Magnesium Intake and Depression in Adults

To the Editor: We thank Slumiers and colleagues¹ for highlighting the challenges of interpreting outcomes from cross-sectional, population-based data. Preclinical and clinical studies indicate low magnesium intake may be associated with depressive symptoms. For instance, Singewald et al² reported mice consuming a diet with very low magnesium content—consisting of only 10% of the daily requirement—showed depressive behavior. Therefore we chose to divide magnesium intake into quintiles to explore the possibility that an association might be seen only in those consuming much less than the estimated average requirement. Sluimers and colleagues are correct that dietary recall is not a perfect marker of hypomagnesemia. That the association is still significant despite the variation introduced by recall methods supports its robustness. Further, intake is amenable to intervention by patients, providers, and policymakers, making it an especially valuable target of analysis.

We agree that social and medical factors are stronger correlates of depression than magnesium intake, as Table 2 (p. 253) shows. We included these variables in the multivariate analysis to explore the possibility that they confound the association of magnesium intake and depression, as suggested by the correspondents.³ That magnesium intake remains significantly associated with depression in the models argues against confounding by these variables. Of course, residual confounding by unmeasured variables is a possibility, which is why prospective randomized trials are needed. We acknowledge that the relationship between magnesium and depression may be bidirectional, and we can only report an association. However, our study adds to the overall body of literature in support of exploring this relationship further. Finally, we see no virtue in a nonrandomized prospective trial because any such study would be subject to the same biases the authors point out in their letter.