Comparisons of spot vs 24-h urine samples for estimating population salt intake: validation study in two independent samples of adults in Britain and Italy

Objectives: To assess the reliability and reproducibility of estimations of group mean 24-h urinary sodium (Na) excretion through timed spot urines compared to 24 h urinary Na output in two independent cross-sectional population samples including men and women and different ethnic groups.

Methods and results: Study 1 was carried out in Britain and included 915 untreated 40-59 yrs male and female participants (297 white, 326 of black African origin and 292 South Asian). Study 2 was carried out in Italy and included 148 white men (mean age 58.3 yrs). All participants provided both a 24-h urine collection and a timed urine sample as part of population surveys. Na, creatinine (Cr) and volume (V) were measured in all samples. Age, body mass index (BMI) and blood pressure (BP) were also measured. We compared the daily Na excretion through 24-h urine (gold standard) with its estimate from timed urine samples with two methods: Tanaka's predictions and Arithmetic extrapolations, and assessed them with correlation coefficients, Bland-Altman plot, prediction of quintile position and Receiver Operating Characteristic (ROC) Areas Under the Curve (AUC) for a cut-off of <100 mmol of Na/day. In Study 1 (discovery study) with the Tanaka method there were poor correlations between predicted and measured 24-h Na excretions in different ethnic groups and genders (r Spearman from 0.055 [R(2) = 0.003] in black women to 0.330 [R(2) = 0.11] in white women). The Bland-Altman plots indicated consistent bias with overestimate for low and underestimate for high intakes. ROC AUCs varied from 0.521 to 0.652 with good sensitivity (95-100%) but very poor specificity (0-9%). With the Arithmetic extrapolations correlations varied from 0.116 [R(2) = 0.01] to 0.367 [R(2) = 0.13]. Bias was detected with both Bland-Altman plots and through quintile analyses (underestimate at low levels and overestimate at high levels). Finally, ROC AUCs varied from 0.514 to 0.640 with moderate sensitivity (64-70%) but low specificity (20-53%). In Study 2 (validation study) results were consistent with the discovery phase in white men.

Conclusion: Based on these results, 24-h urinary collection for the measurement of Na excretion remains the preferred tool for assessing salt intake when compared with reported methods based on timed spot urine samples.