FitProven
Evidence Breakdown
Based on 9 studies
THE BIGGER QUESTION
Whether this one statement holds is settled above. What to actually do about it is a wider question, weighed across every claim that bears on it.
- Is it bad to drink electrolytes every day? — this claim is one of 2 weighed there
Evidence map
For & against, at a glance
Meta-Analysis
Neutral
Filippini T et al. · 2020Journal of the American Heart AssociationDose-response meta-analysis of 32 RCTs (≥4 weeks) of potassium supplementation, modelled with 1-stage cubic splines. The relationship was U-shaped: blood pressure fell up to about 30 mmol/day of added potassium, the reduction weakened beyond that, and blood pressure *rose* above differences of roughly 80 mmol/day. The BP-lowering effect was **stronger** in hypertensive participants and at higher levels of sodium intake — though it was not exclusive to them: the paper reports a small decrease in mean BP with added potassium up to 20–30 mmol/day "in both normotensive and hypertensive trials", and the normotensive estimate rests on only 5 trials. The rise in BP at high potassium was considerably more evident in hypertensives on antihypertensive medication (starting around 60 mmol/day) than in untreated hypertensives (around 110 mmol/day). Together this supports targeted potassium repletion, especially in people with high blood pressure and high sodium intake, rather than blanket high-dose electrolyte supplementation — and it confirms that more is not better.
0.90
Dose-response meta-analysis of 32 RCTs (≥4 weeks) of potassium supplementation, modelled with 1-stage cubic splines. The relationship was U-shaped: blood pressure fell up to about 30 mmol/day of added potassium, the reduction weakened beyond that, and blood pressure *rose* above differences of roughly 80 mmol/day. The BP-lowering effect was **stronger** in hypertensive participants and at higher levels of sodium intake — though it was not exclusive to them: the paper reports a small decrease in mean BP with added potassium up to 20–30 mmol/day "in both normotensive and hypertensive trials", and the normotensive estimate rests on only 5 trials. The rise in BP at high potassium was considerably more evident in hypertensives on antihypertensive medication (starting around 60 mmol/day) than in untreated hypertensives (around 110 mmol/day). Together this supports targeted potassium repletion, especially in people with high blood pressure and high sodium intake, rather than blanket high-dose electrolyte supplementation — and it confirms that more is not better.
Design Meta-Analysis (1.0) × quality 0.90 = impact 0.90
View sourceMeta-Analysis
Neutral
Zhang X et al. · 2016HypertensionMeta-analysis of 34 randomized double-blind placebo-controlled trials with 2,028 participants. Magnesium supplementation at a median 368 mg/day for a median 3 months lowered systolic blood pressure by 2.00 mm Hg (95% CI 0.43–3.58) and diastolic by 1.78 mm Hg (95% CI 0.73–2.82), with corresponding rises in serum magnesium. This is the best case for the claim — magnesium is an electrolyte, dietary shortfalls are common, and daily supplementation produces a real if small effect — but the effect size is clinically modest and most commercial "electrolyte" mixes are sodium-dominant with only token magnesium, so it does not validate the products actually being sold.
0.80
Meta-analysis of 34 randomized double-blind placebo-controlled trials with 2,028 participants. Magnesium supplementation at a median 368 mg/day for a median 3 months lowered systolic blood pressure by 2.00 mm Hg (95% CI 0.43–3.58) and diastolic by 1.78 mm Hg (95% CI 0.73–2.82), with corresponding rises in serum magnesium. This is the best case for the claim — magnesium is an electrolyte, dietary shortfalls are common, and daily supplementation produces a real if small effect — but the effect size is clinically modest and most commercial "electrolyte" mixes are sodium-dominant with only token magnesium, so it does not validate the products actually being sold.
Design Meta-Analysis (1.0) × quality 0.80 = impact 0.80
View sourceMeta-Analysis
Con
Aburto NJ et al. · 2013BMJSystematic review and meta-analysis of 37 RCTs of blood pressure in adults plus 14 cohorts and 5 RCTs of cardiovascular outcomes. Reducing sodium intake lowered systolic blood pressure by 3.39 mm Hg (95% CI 2.46–4.31) and diastolic by 1.54 mm Hg, and higher sodium intake was associated with increased stroke risk (RR 1.24, 95% CI 1.08–1.43), with no adverse effect on blood lipids, catecholamines, or renal function. For a population that already over-consumes sodium, routinely adding more via a daily supplement pushes in the direction shown here to raise blood pressure and stroke risk.
0.85
Systematic review and meta-analysis of 37 RCTs of blood pressure in adults plus 14 cohorts and 5 RCTs of cardiovascular outcomes. Reducing sodium intake lowered systolic blood pressure by 3.39 mm Hg (95% CI 2.46–4.31) and diastolic by 1.54 mm Hg, and higher sodium intake was associated with increased stroke risk (RR 1.24, 95% CI 1.08–1.43), with no adverse effect on blood lipids, catecholamines, or renal function. For a population that already over-consumes sodium, routinely adding more via a daily supplement pushes in the direction shown here to raise blood pressure and stroke risk.
Design Meta-Analysis (1.0) × quality 0.85 = impact 0.85
View sourceRCT
Con
Hew-Butler TD et al. · 2006British Journal of Sports MedicineProspective randomized trial at the 2001 Cape Town Ironman triathlon. 145 triathletes volunteered and were randomly assigned to sodium tablets (620 mg table salt each) or an identical starch placebo; of those, "114 ingested the supplemental tablets and had complete data collected before and after the race" — 53 sodium, 61 placebo. A further 299 triathletes already enrolled in other medical research were added as a non-randomised third comparison group, which is where the often-quoted total of 413 comes from; only 114 athletes are in the randomised comparison. The sodium group ingested an additional 3.6 (2.0) g of sodium over the race. There were "no significant differences between the sodium, placebo, and no supplementation groups with regard to age, finishing time, serum sodium concentration before and after the race, weight before the race, weight change during the race, and rectal temperature", and hyponatremia was not prevented by supplementation nor caused by its absence. If ~12 hours of continuous exercise does not require supplemental sodium, an ordinary day plainly does not.
0.56
Prospective randomized trial at the 2001 Cape Town Ironman triathlon. 145 triathletes volunteered and were randomly assigned to sodium tablets (620 mg table salt each) or an identical starch placebo; of those, "114 ingested the supplemental tablets and had complete data collected before and after the race" — 53 sodium, 61 placebo. A further 299 triathletes already enrolled in other medical research were added as a non-randomised third comparison group, which is where the often-quoted total of 413 comes from; only 114 athletes are in the randomised comparison. The sodium group ingested an additional 3.6 (2.0) g of sodium over the race. There were "no significant differences between the sodium, placebo, and no supplementation groups with regard to age, finishing time, serum sodium concentration before and after the race, weight before the race, weight change during the race, and rectal temperature", and hyponatremia was not prevented by supplementation nor caused by its absence. If ~12 hours of continuous exercise does not require supplemental sodium, an ordinary day plainly does not.
Design RCT (0.8) × quality 0.70 = impact 0.56
View sourceMeta-Analysis
Neutral
Filippini T et al. · 2020Journal of the American Heart AssociationDose-response meta-analysis of 32 RCTs (≥4 weeks) of potassium supplementation, modelled with 1-stage cubic splines. The relationship was U-shaped: blood pressure fell up to about 30 mmol/day of added potassium, the reduction weakened beyond that, and blood pressure *rose* above differences of roughly 80 mmol/day. The BP-lowering effect was **stronger** in hypertensive participants and at higher levels of sodium intake — though it was not exclusive to them: the paper reports a small decrease in mean BP with added potassium up to 20–30 mmol/day "in both normotensive and hypertensive trials", and the normotensive estimate rests on only 5 trials. The rise in BP at high potassium was considerably more evident in hypertensives on antihypertensive medication (starting around 60 mmol/day) than in untreated hypertensives (around 110 mmol/day). Together this supports targeted potassium repletion, especially in people with high blood pressure and high sodium intake, rather than blanket high-dose electrolyte supplementation — and it confirms that more is not better.
0.90
Dose-response meta-analysis of 32 RCTs (≥4 weeks) of potassium supplementation, modelled with 1-stage cubic splines. The relationship was U-shaped: blood pressure fell up to about 30 mmol/day of added potassium, the reduction weakened beyond that, and blood pressure *rose* above differences of roughly 80 mmol/day. The BP-lowering effect was **stronger** in hypertensive participants and at higher levels of sodium intake — though it was not exclusive to them: the paper reports a small decrease in mean BP with added potassium up to 20–30 mmol/day "in both normotensive and hypertensive trials", and the normotensive estimate rests on only 5 trials. The rise in BP at high potassium was considerably more evident in hypertensives on antihypertensive medication (starting around 60 mmol/day) than in untreated hypertensives (around 110 mmol/day). Together this supports targeted potassium repletion, especially in people with high blood pressure and high sodium intake, rather than blanket high-dose electrolyte supplementation — and it confirms that more is not better.
Design Meta-Analysis (1.0) × quality 0.90 = impact 0.90
View sourceMeta-Analysis
Neutral
Zhang X et al. · 2016HypertensionMeta-analysis of 34 randomized double-blind placebo-controlled trials with 2,028 participants. Magnesium supplementation at a median 368 mg/day for a median 3 months lowered systolic blood pressure by 2.00 mm Hg (95% CI 0.43–3.58) and diastolic by 1.78 mm Hg (95% CI 0.73–2.82), with corresponding rises in serum magnesium. This is the best case for the claim — magnesium is an electrolyte, dietary shortfalls are common, and daily supplementation produces a real if small effect — but the effect size is clinically modest and most commercial "electrolyte" mixes are sodium-dominant with only token magnesium, so it does not validate the products actually being sold.
0.80
Meta-analysis of 34 randomized double-blind placebo-controlled trials with 2,028 participants. Magnesium supplementation at a median 368 mg/day for a median 3 months lowered systolic blood pressure by 2.00 mm Hg (95% CI 0.43–3.58) and diastolic by 1.78 mm Hg (95% CI 0.73–2.82), with corresponding rises in serum magnesium. This is the best case for the claim — magnesium is an electrolyte, dietary shortfalls are common, and daily supplementation produces a real if small effect — but the effect size is clinically modest and most commercial "electrolyte" mixes are sodium-dominant with only token magnesium, so it does not validate the products actually being sold.
Design Meta-Analysis (1.0) × quality 0.80 = impact 0.80
View sourceMeta-Analysis
Con
Aburto NJ et al. · 2013BMJSystematic review and meta-analysis of 37 RCTs of blood pressure in adults plus 14 cohorts and 5 RCTs of cardiovascular outcomes. Reducing sodium intake lowered systolic blood pressure by 3.39 mm Hg (95% CI 2.46–4.31) and diastolic by 1.54 mm Hg, and higher sodium intake was associated with increased stroke risk (RR 1.24, 95% CI 1.08–1.43), with no adverse effect on blood lipids, catecholamines, or renal function. For a population that already over-consumes sodium, routinely adding more via a daily supplement pushes in the direction shown here to raise blood pressure and stroke risk.
0.85
Systematic review and meta-analysis of 37 RCTs of blood pressure in adults plus 14 cohorts and 5 RCTs of cardiovascular outcomes. Reducing sodium intake lowered systolic blood pressure by 3.39 mm Hg (95% CI 2.46–4.31) and diastolic by 1.54 mm Hg, and higher sodium intake was associated with increased stroke risk (RR 1.24, 95% CI 1.08–1.43), with no adverse effect on blood lipids, catecholamines, or renal function. For a population that already over-consumes sodium, routinely adding more via a daily supplement pushes in the direction shown here to raise blood pressure and stroke risk.
Design Meta-Analysis (1.0) × quality 0.85 = impact 0.85
View sourceRCT
Con
Hew-Butler TD et al. · 2006British Journal of Sports MedicineProspective randomized trial at the 2001 Cape Town Ironman triathlon. 145 triathletes volunteered and were randomly assigned to sodium tablets (620 mg table salt each) or an identical starch placebo; of those, "114 ingested the supplemental tablets and had complete data collected before and after the race" — 53 sodium, 61 placebo. A further 299 triathletes already enrolled in other medical research were added as a non-randomised third comparison group, which is where the often-quoted total of 413 comes from; only 114 athletes are in the randomised comparison. The sodium group ingested an additional 3.6 (2.0) g of sodium over the race. There were "no significant differences between the sodium, placebo, and no supplementation groups with regard to age, finishing time, serum sodium concentration before and after the race, weight before the race, weight change during the race, and rectal temperature", and hyponatremia was not prevented by supplementation nor caused by its absence. If ~12 hours of continuous exercise does not require supplemental sodium, an ordinary day plainly does not.
0.56
Prospective randomized trial at the 2001 Cape Town Ironman triathlon. 145 triathletes volunteered and were randomly assigned to sodium tablets (620 mg table salt each) or an identical starch placebo; of those, "114 ingested the supplemental tablets and had complete data collected before and after the race" — 53 sodium, 61 placebo. A further 299 triathletes already enrolled in other medical research were added as a non-randomised third comparison group, which is where the often-quoted total of 413 comes from; only 114 athletes are in the randomised comparison. The sodium group ingested an additional 3.6 (2.0) g of sodium over the race. There were "no significant differences between the sodium, placebo, and no supplementation groups with regard to age, finishing time, serum sodium concentration before and after the race, weight before the race, weight change during the race, and rectal temperature", and hyponatremia was not prevented by supplementation nor caused by its absence. If ~12 hours of continuous exercise does not require supplemental sodium, an ordinary day plainly does not.
Design RCT (0.8) × quality 0.70 = impact 0.56
View sourceShowing the 4 strongest of 9 studies. Tap any node to expand its detail.
Evidence
AGAINST (5)
AGAINST Meta-Analysis0.85 Aburto NJ, Ziolkovska A et al. (2013)
Lower sodium cut systolic BP by 3.39 mm Hg (95% CI 2.46-4.31) and diastolic by 1.54 mm Hg; higher sodium intake associated with stroke risk RR 1.24 (95% CI 1.08-1.43); no adverse effect on lipids, catecholamines or renal function
Systematic review and meta-analysis of 37 RCTs of blood pressure in adults plus 14 cohorts and 5 RCTs of cardiovascular outcomes. Reducing sodium intake lowered systolic blood pressure by 3.39 mm Hg (95% CI 2.46–4.31) and diastolic by 1.54 mm Hg, and higher sodium intake was associated with increased stroke risk (RR 1.24, 95% CI 1.08–1.43), with no adverse effect on blood lipids, catecholamines, or renal function. For a population that already over-consumes sodium, routinely adding more via a daily supplement pushes in the direction shown here to raise blood pressure and stroke risk.
Weighted 0.85 — Pooled participant count not reported per outcome, so sample_size is omitted: k=37 RCTs for blood pressure plus 14 cohorts and 5 RCTs for cardiovascular outcomes. WHO-commissioned, Cochrane-style methodology, published in the BMJ — low risk of bias and well powered. Held below 0.90 only because it is applied here indirectly (population sodium reduction, not daily electrolyte supplements), and because the funders (WHO, national governments) have a public-health policy position on sodium.
Funding: WHO, the Kidney Evaluation Association Japan, and the governments of Japan and the Republic of Korea
BMJ
AGAINST Prospective Cohortn=4880.80 Almond CS, Shin AY et al. (2005)
13% of finishers were hyponatremic (serum sodium 135 mmol/L or less) and 0.6% critically so; independent risk factors were weight gain during the race (OR 4.2) and finishing time over 4 h (OR 7.4), while the composition of fluids ingested was NOT an independent risk factor
Prospective cohort of 766 runners in the 2002 Boston Marathon, 488 of whom gave a finish-line blood sample: 13% were hyponatremic (serum sodium ≤135 mmol/L) and 0.6% critically so. On multivariate analysis the risk factors were substantial weight gain during the race (OR 4.2, i.e. drinking too much fluid) and finishing time over 4 hours (OR 7.4); the composition of the fluids ingested was not an independent risk factor — drinking electrolyte-containing sports drinks instead of water did not protect runners. The problem is fluid overload, not electrolyte deficiency, which undercuts the marketing rationale for daily electrolyte dosing.
Weighted 0.80 — 488 runners analysed out of 766 enrolled (only those who gave a finish-line blood sample), published in NEJM with prespecified multivariate analysis — large, prospective and independently funded. Discounted from the top band for the 64% sampling fraction (possible selection bias) and because it is a single race in a single climate, and observational rather than randomised.
Funding: US National Institutes of Health (NCRR grant RR-02172)
The New England Journal of Medicine
AGAINST Cross-Sectional0.75 Powles J, Fahimi S et al. (2013)
Global mean sodium intake in 2010 was 3.95 g/day (95% UI 3.89-4.01) — roughly 10 g of salt, nearly double the WHO limit; no world region averaged below the recommendation
Bayesian pooled analysis of 142 surveys of 24-hour urinary sodium excretion plus 103 dietary surveys across 66 countries, modelling intake for 187 countries. Global mean sodium intake in 2010 was 3.95 g/day (95% UI 3.89–4.01), roughly 10 g of salt and nearly double the WHO limit of 2 g/day; North America and Western Europe sat at 3.4–3.8 g/day and no world region averaged below the recommendation. Population-level sodium deficiency, the premise of daily sodium supplementation, does not exist anywhere measured.
Weighted 0.75 — Not a participant-level study, so no sample_size: it is a Bayesian pooling of 142 surveys of 24-hour urinary sodium excretion plus 103 dietary surveys across 66 countries, modelled out to 187 countries. Estimates for unsurveyed countries are modelled and correspondingly uncertain, and the senior author declares food- and nutrition-industry consulting. It scores high for its design class anyway because the question it answers — is any population eating too little sodium — turns on breadth of measurement rather than causal inference, and no dataset is broader.
Funding: Bill & Melinda Gates Foundation, via the Global Burden of Diseases 2010 study; one author supported by MRC grant MR/K005901/1
BMJ Open
AGAINST RCTn=1140.70 Hew-Butler TD, Sharwood K et al. (2006)
No difference in post-race serum sodium, weight change, body temperature or finishing time with ~3.6 g supplemental sodium
Prospective randomized trial at the 2001 Cape Town Ironman triathlon. 145 triathletes volunteered and were randomly assigned to sodium tablets (620 mg table salt each) or an identical starch placebo; of those, "114 ingested the supplemental tablets and had complete data collected before and after the race" — 53 sodium, 61 placebo. A further 299 triathletes already enrolled in other medical research were added as a non-randomised third comparison group, which is where the often-quoted total of 413 comes from; only 114 athletes are in the randomised comparison. The sodium group ingested an additional 3.6 (2.0) g of sodium over the race. There were "no significant differences between the sodium, placebo, and no supplementation groups with regard to age, finishing time, serum sodium concentration before and after the race, weight before the race, weight change during the race, and rectal temperature", and hyponatremia was not prevented by supplementation nor caused by its absence. If ~12 hours of continuous exercise does not require supplemental sodium, an ordinary day plainly does not.
Weighted 0.70 — The randomised trial is smaller than it looks: 145 athletes were randomised and 114 had complete pre- and post-race data (53 sodium, 61 placebo); the often-quoted 413 adds 299 non-randomised observational athletes. A placebo-controlled field trial across ~12 h of racing is strong and direct evidence, but 114 analysed is modest and the paper carries no funding statement.
British Journal of Sports Medicine
AGAINST Crossover Trialn=90.40 Cosgrove SD, Black KE (2013)
No effect on 72 km time-trial performance (~171 min) and no effect on plasma sodium (pre-to-post change 0.47% placebo vs 0.56% sodium); sodium increased fluid intake by 160 mL/h, adding carried weight for no benefit
Randomized double-blind crossover in 9 well-trained cyclists completing a 72 km time trial (~171 min) in cool conditions (13.8 ± 2.0 °C) with sodium supplements or placebo. Sodium had no effect on time-trial performance and no effect on plasma sodium (change from pre- to post-race 0.47% placebo vs 0.56% sodium); it did increase fluid intake by 160 mL/h, adding carried weight for no benefit. Small sample, but the direction is clear: even in trained athletes exercising for ~3 hours, supplemental sodium changed neither performance nor electrolyte status.
Weighted 0.40 — Only 9 cyclists, though the randomised double-blind crossover design means each acts as his or her own control, which recovers some power. Independently funded with no supplement-company involvement, and a clean null in cool conditions. Capped at the low end purely on precision: n=9 cannot exclude a small effect, and a single ~3-hour trial in 14 degrees C does not generalise to heat or ultra-endurance.
Funding: University of Otago; authors declared no competing interests
Journal of the International Society of Sports Nutrition
NEUTRAL (4)
NEUTRAL Cross-Sectionaln=125810.90 Cogswell ME, Zhang Z et al. (2012)
Median usual sodium intake 3,371 mg/d — 99.4% of adults exceeded the 1,500 mg/d AHA limit and ~91% exceeded the 2,300 mg/d IOM upper limit; median potassium 2,631 mg/d, with fewer than 2% meeting the 4,700 mg/d adequate intake
Cross-sectional analysis of 24-hour dietary recalls from 12,581 US adults in NHANES 2003–2008. It cuts both ways for this claim: 99.4% of adults already consumed more sodium than the AHA limit of 1500 mg/d and ~91% exceeded the IOM upper limit of 2300 mg/d — so sodium deficiency is essentially nonexistent in the population — yet fewer than 2% of adults met the 4700 mg/d potassium adequate intake. If there is any electrolyte gap in the typical Western diet, it is potassium (and food-source potassium), not the sodium that dominates electrolyte supplements.
Weighted 0.90 — 12,581 adults from NHANES 2003-2008 — a large, nationally representative, federally funded (CDC/USDA/NIH) sample with no declared conflicts, using proper usual-intake modelling rather than raw single-day recalls. Top of the cross-sectional class. The inherent limits (self-reported 24-hour recalls, no outcome data) are the design's, which the study-type base weight already prices in.
Funding: Supported by the CDC, US Department of Health and Human Services; the Agricultural Research Service, US Department of Agriculture; and the Office of Dietary Supplements, NIH (grant no. 3R01HL091024-02S1 to ALC).
The American Journal of Clinical Nutrition
NEUTRAL Meta-Analysisn=17640.90 Filippini T, Naska A et al. (2020)
U-shaped dose-response: blood pressure fell up to ~30 mmol/day of added potassium, the reduction weakened beyond that, and BP rose above differences of ~80 mmol/day. BP-lowering was stronger in hypertensive participants and at higher sodium intakes — but a small BP decrease up to 20-30 mmol/day was seen in both normotensive and hypertensive trials, so the benefit is not confined to hypertensives. The BP rise at high potassium was much more evident in drug-treated hypertensives (from ~60 mmol/day) than in untreated ones (~110 mmol/day)
Dose-response meta-analysis of 32 RCTs (≥4 weeks) of potassium supplementation, modelled with 1-stage cubic splines. The relationship was U-shaped: blood pressure fell up to about 30 mmol/day of added potassium, the reduction weakened beyond that, and blood pressure rose above differences of roughly 80 mmol/day. The BP-lowering effect was stronger in hypertensive participants and at higher levels of sodium intake — though it was not exclusive to them: the paper reports a small decrease in mean BP with added potassium up to 20–30 mmol/day "in both normotensive and hypertensive trials", and the normotensive estimate rests on only 5 trials. The rise in BP at high potassium was considerably more evident in hypertensives on antihypertensive medication (starting around 60 mmol/day) than in untreated hypertensives (around 110 mmol/day). Together this supports targeted potassium repletion, especially in people with high blood pressure and high sodium intake, rather than blanket high-dose electrolyte supplementation — and it confirms that more is not better.
Weighted 0.90 — 1,764 participants across 32 RCTs of at least 4 weeks, modelled as a dose-response spline rather than a single pooled estimate - the most informative design for a 'more is not better' question. Publicly funded with no author disclosures. Its weakest part is the normotensive subgroup, which rests on only 5 trials.
Funding: European Food Safety Authority (grant GP-EFSA-AFSCO-2017-01 GA09)
Journal of the American Heart Association
NEUTRAL Meta-Analysisn=20280.80 Zhang X, Li Y et al. (2016)
Magnesium (median 368 mg/day for a median 3 months) lowered systolic BP by 2.00 mm Hg (95% CI 0.43-3.58) and diastolic by 1.78 mm Hg (0.73-2.82), with corresponding rises in serum magnesium
Meta-analysis of 34 randomized double-blind placebo-controlled trials with 2,028 participants. Magnesium supplementation at a median 368 mg/day for a median 3 months lowered systolic blood pressure by 2.00 mm Hg (95% CI 0.43–3.58) and diastolic by 1.78 mm Hg (95% CI 0.73–2.82), with corresponding rises in serum magnesium. This is the best case for the claim — magnesium is an electrolyte, dietary shortfalls are common, and daily supplementation produces a real if small effect — but the effect size is clinically modest and most commercial "electrolyte" mixes are sodium-dominant with only token magnesium, so it does not validate the products actually being sold.
Weighted 0.80 — 34 randomized double-blind placebo-controlled trials, 2,028 participants, with a confirmed serum-magnesium rise showing the intervention actually landed — methodologically the strongest evidence available for the claim. Two reservations: the effect is real but clinically modest (about 2 mm Hg), and although the authors declare no conflict, one co-author's affiliation is the Center for Magnesium Education and Research, an advocacy body, and she is a cited primary-source author within the review itself.
Funding: Indiana University Health / Indiana University School of Medicine Strategic Research Initiative
Hypertension
NEUTRAL Prospective Cohortn=101945INDUSTRY0.60 O'Donnell M, Mente A et al. (2014)
J-shaped curve: risk lowest at 4-6 g/day sodium; ≥7 g/day OR 1.15 (95% CI 1.02-1.30) and <3 g/day OR 1.27 (95% CI 1.12-1.44); higher potassium excretion associated with lower risk
PURE cohort: estimated 24-hour urinary sodium and potassium in 101,945 people across 17 countries, mean 3.7 years of follow-up, 3,317 deaths or major cardiovascular events. Risk was lowest at 4–6 g/day of sodium, higher at ≥7 g/day (OR 1.15, 95% CI 1.02–1.30) and also higher below 3 g/day (OR 1.27, 95% CI 1.12–1.44); higher potassium excretion was associated with lower risk. The low-sodium arm of this J-curve is the strongest published hint that sodium can be too low — but the mean intake was 4.93 g/day, well above the point of concern, so it does not establish that a typical person needs supplemental sodium. Estimating intake from a single spot urine sample is a known weakness of this analysis.
Weighted 0.60 — 101,945 people in 17 countries with 3,317 events — the largest dataset on the question by a wide margin, which is why it is not scored lower. But sodium intake was estimated from a single fasting spot urine sample via the Kawasaki formula, a method known to distort the tails of the distribution, and the observational design cannot rule out reverse causation in the low-sodium arm, which is exactly the arm the claim depends on.
Funding: Population Health Research Institute, Hamilton Health Sciences, Canadian Institutes of Health Research and the Heart and Stroke Foundation of Ontario, plus unrestricted grants from several pharmaceutical companies (AstraZeneca, Sanofi, Boehringer Ingelheim, Servier, GlaxoSmithKline, Novartis, King)
The New England Journal of Medicine