Oral magnesium supplements decrease high blood pressure (SBP>155 mmHg) in hypertensive subjects on anti-hypertensive...

Previously, we examined 44 human studies involving oral magnesium (Mg) supplementation for hypertension (HT), sorting them according to HT status, Mg dose and anti-hypertensive medication usage. We found that while some studies reported a significant lowering of blood pressure with Mg supplementation, others did not. We present here our first meta-analysis of a uniform subset from this series of studies. Seven studies, involving 135 hypertensive subjects on anti-hypertensive medication continuously for at least six months, with no more than a two-week washout and with a mean starting systolic blood pressure (SBP)>155 mmHg, demonstrated a mean change of -18.7 mmHg [95% CI=-14.95 to -22.45] p<0.0001 and an effect size test (Cohen's d)=1.19, i.e. a large and highly significant effect. Meta-analysis of diastolic blood pressure (DBP) for these same seven studies showed a mean change in DBP of -10.9 mmHg [95% CI=-8.73 to -13.1], p<0.0001, with an effect size test (Cohen's d)=1.19. Other studies from our original collection, approaching, but not meeting the >155 mmHg starting SBP values or not complying as regards anti-hypertensive medication usage, showed mean changes in both SBP and DBP with oral Mg that, while not approaching the high-responder values of the present study, appeared to include some high-responder subjects combined with low- or non-responder subjects. This uniform subset of seven studies showed a strong effect of Mg treatment in hypertension, which is in stark contrast to results of three other meta-analyses. Using non-uniform sets of studies, the small effects reported in previous meta-analyses may reflect a blending of dissimilar studies, which acted to seriously underestimate the potential of Mg in hypertension in some (but not all) subjects. Within studies, blending of non-, moderate and highresponder subjects in any one study might mask strong effects of Mg treatment in some subjects.