O’Donnell et al. 2014 — PURE study: Urinary sodium and potassium excretion, mortality, and cardiovascular events
Citation
O’Donnell M, Mente A, Rangarajan S, McQueen MJ, Wang X et al. (PURE Investigators). “Urinary Sodium and Potassium Excretion, Mortality, and Cardiovascular Events.” New England Journal of Medicine 2014;371(7):612–623. doi:10.1056/NEJMoa1311889. PMID 25119607.
Study design
Prospective observational study within the PURE (Prospective Urban Rural Epidemiology) cohort. 101,945 participants from 17 countries (Argentina, Bangladesh, Brazil, Canada, Chile, China, Colombia, India, Iran, Malaysia, Pakistan, Poland, South Africa, Sweden, Turkey, United Arab Emirates, and Zimbabwe). Sodium and potassium intake estimated from spot urine samples using the Kawasaki formula (a single-morning void, adjusted for age, sex, creatinine). Primary outcome: composite of cardiovascular death, non-fatal myocardial infarction, stroke, and heart failure.
Key findings
J-shaped relationship between sodium excretion and composite cardiovascular outcome:
| Sodium excretion level | Outcome association |
|---|---|
| < 3 g/day (low) | OR 1.27 — increased risk vs reference |
| 3–6 g/day (reference range) | Lowest risk |
| ≥ 7 g/day (high) | OR 1.15 — increased risk |
- High sodium (≥7 g/day) increased composite outcome risk by 15%, driven especially by participants with hypertension (interaction p<0.001)
- Low sodium (<3 g/day) was associated with higher risk than the reference range — the J-curve’s low-end arm
- Potassium: higher estimated potassium excretion was inversely associated with composite outcomes vs. <1.50 g/day potassium reference (i.e., potassium was protective)
- The combined high-sodium/low-potassium pattern was most hazardous
Why this paper is contested
Methodological criticisms that have been raised by independent investigators:
-
Spot urine vs 24-hour urine: The Kawasaki formula applied to a single morning void systematically underestimates sodium in populations with genuinely low intake (circadian variation means early-morning spot urine overestimates daily excretion in some individuals and underestimates in others). The low-sodium group’s apparent intake (~1.6 g/day median) may substantially misclassify true sodium intake, inflating apparent risk in the low-sodium arm.
-
Sick-quitter / reverse causation: Patients with pre-existing cardiovascular disease, diabetes, heart failure, or CKD are both more likely to restrict sodium (disease-directed) AND more likely to experience adverse outcomes — creating a spurious association between low sodium and poor outcomes that is not causal. This is the “sick-quitter” effect well-documented in alcohol epidemiology.
-
Cross-country confounding: The J-curve result was driven primarily by data from China and other high-sodium-intake countries; within the low-sodium groups, there was disproportionate representation of populations with concurrent metabolic disease.
-
External validity: The 3–6 g/day “optimal range” claimed by PURE is substantially higher than current WHO (<2 g/day) and AHA (<1.5–2.3 g/day) recommendations; replication in controlled feeding studies has not confirmed a J-curve.
Current evidence status
The 2025 Kong umbrella review (21 meta-analyses, 91 outcomes) found no J-shaped curve in pooled dose-response analyses, with a linear relationship between lower sodium and better outcomes. The PURE J-curve remains an observational finding with methodological concerns. kong-2025-sodium-umbrella-review provides the current synthesis that this page should be read alongside.
gap/contradictory-evidence — the PURE J-curve finding contrasts with the predominant direction of the meta-analytic literature. See sodium-restriction § J-curve controversy.
Relationships
- Contested by: kong-2025-sodium-umbrella-review (umbrella review finding no J-curve)
- Context: sodium-restriction (full controversy framing)
- Companion PURE paper (BP, same cohort): doi:10.1056/NEJMoa1311889 (Mente et al. — BP dose-response analysis is a separate paper)