External validation and comparison of formulae estimating 24-h sodium intake from a fasting morning urine sample

OBJECTIVE: The relationship between sodium intake and cardiovascular events is controversial, but most large epidemiological studies estimated sodium intake using formulae based on single urine samples, the validity of which is debated. We evaluated sodium intake estimating formulae in a large cohort of adult patients. DESIGN AND METHODS: Patients were asked to collect 24-h urine the day before admission. Validity of the 24-h urine collection was assessed by comparing creatinine clearance from this collection to the mean creatinine clearance from six fractionated urine samples. Only collections with creatinine clearance within ±15% of fractionated clearance were considered valid. The Kawasaki, INTERSALT and Tanaka formulae, using a morning fasting urine sample obtained upon admission, were compared with 24-h urine sodium excretion. The relationship between sodium intake, either measured or estimated, and blood pressure was assessed. RESULTS: Amongst 2278 patients referred to our physiology department between September 2006 and August 2016, 1018 had complete 24-h urine collections and were included in this analysis. Mean age was 51 ± 14 years and mean sodium excretion was 3624 ± 1614 mg/day. The intraclass correlation coefficient was higher for the Kawasaki (0.54; 95% confidence interval, 0.48-0.60), than for the INTERSALT (0.38; 0.33-0.42, P < 0.001), and Tanaka (0.42; 0.37-0.46, P < 0.001) formulae. The Kawasaki formula displayed the lowest mean bias (248; 157-339 mg/day). There was a significant positive association between measured sodium intake and blood pressure, and the Kawasaki formula yielded a similar association. CONCLUSION: All formulae have poor precision and accuracy and are not suitable for estimating individual sodium intake. This does not dismiss their potential value for assessment of sodium intake in population studies.