Invalid methodology

Reference

O’Donnell MJ, Mente A, Rangarajan S, McQueen MJ, Wang X, Liu L, et al. for the PURE Investigators

Urinary sodium and potassium excretion, mortality, and cardiovascular events.

New Engl J Med 371, 612–23 (2014).

Claim

Obtained morning fasting urine samples from 101,945 persons taking part in the Prospective Urban Rural Epidemiology (PURE) study, a cohort study that enrolled and followed up 156,424 persons, aged 37–70 years from 628 rural and urban communities in 17 low-middle-and high-income countries. Salt intake was inferred

from estimated 24h urinary sodium excretion calculated with the Kawasaki formula applied to a morning fasting urine sample. They recorded all-cause and major CV deaths for a mean follow-up of 3.7 years. A composite outcome occurred in 3,317 participants. There was a J-shaped association between estimated sodium excretion and CV events: a higher estimated sodium excretion of >7 g (>17.5 g of salt) per day was associated with increased risks of death and major CV events, with a stronger effect among people with ‘hypertension’. On the other hand, an estimated sodium excretion that was below 3 g (7.5 g of salt) per day was also associated with an increased risk of the composite outcome.

Critique

The estimated sodium excretion using morning fasting spot urines with Kawasaki or other equations is unreliable and biased, as demonstrated by several studies, [1-2-3-4] including validations within the PURE study [5-6].

The sodium study within the PURE study may be affected by selection bias: the sodium study, in fact, included only 65% of the participants; there were fewer from India (5 v 18%) and many more from China (42% v 30%). Moreover, a high proportion of participants had preexisting ill-health (hypertension, blood pressure medication, pre-existing CHD and CVD).

The lower sodium group was grossly unbalanced compared to the higher sodium group: it was older and had fewer men, fewer Asians and smokers with more Africans and non-Asians and urban persons; they also had a lower BP, a higher LDL-cholesterol, a history of CVD and diabetes, lower level of physical activity and higher medication use (suggesting reverse causality).

The lower sodium excretion group (<3 g/day or <7.5 g of salt/day) was unable to discriminate within the range of recommended targets for populations and, therefore, was not informative for policy making.