An observational study of urinary sodium excretion (a surrogate of dietary salt intake) and it’s association with cardiovascular mortality was published in JAMA last week. The trial involved 3681 patients in Europe without cardiovascular disease at baseline. The details are below:
Observational cohort study comprising the primary Outcome cohort, from which two further follow-up cohorts were derived (BP cohort and HTN cohort):
– Outcome cohort (n=3681) – full set of baseline measurements available; used for mortality study
– BP cohort (n=1499) – full set of baseline and follow-up measurements; used for analysis of 24hr urine Na with BP
– HTN cohort (n=2096) – those without baseline HTN; used for analysis of incidence of HTN
Participants of the FLEMENGHO and EPOGH studies
Incomplete 24hr urine collections
Cardiovascular disease at baseline
At follow-up study date, further exclusions were based on losses to follow-up, incomplete 24hr urine collections, those taking anti-HTN medications, interval development of CV disease
Primary – Adjudicated all-cause and non-fatal and fatal cardiovascular outcomes
Others – incidence of HTN, X-sectional and longitudinal BP analysis
– Median follow-up 7.9 years; 47.3% male; 4.1% Diabetic; 28.4% smoke
– Mean 24hr urine sodium was 178mmol
– Adjusted all-cause mortality did not associate with urine Na excretion
– Adjusted CV mortality increased at the lowest tertile of urine Na excretion – adjusted HR for the lowest tertile of urine Na excretion 1.56 (95% CI 1.02, 2.36)
– Adjusted nonfatal CV events did not associate with urine Na excretion
– In adjusted analyses, a 100mmol increase in urine Na excretion associated with a 1.71mmHg increase in SBP, but no change in DBP
There have been a number of articles recently raising some eyebrows in relation to dietary salt intake and outcomes, particularly in diabetics. The article above is no exception, suggesting that those in the higher tertile of urine sodium excretion actually have less fatal CV outcomes during follow-up. The response from experts in the field has been swift and many concerns with the study have been raised.
Firstly, it’s observational – a major issue;
Second, there were only 84 CV-related deaths in the cohort of 3681 participants, making it hard to draw any firm conclusions due to lack of statistical power;
Third, there are concerns over the reliability of 24hr urine collections at the best of times, although they did measure 24hr urine creatinine excretion in an attempt to include only complete collections – unfortunately this automatically creates a certain type of selection bias, as only subjects with complete collections were evaluated – we don’t know anything about those who did not have complete collections;
Fourth, the baseline urinary Na was based on one single collections – an average of two or three would have been more robust.
It seems that extreme caution is required in interpreting these results – as ever, future research and randomized controlled trials have been called for – whether they happen or not is another matter.