Estimating the population distribution of usual 24-hour sodium excretion from timed urine void specimens using a statistical approach accounting for correlated measurement errors

Abstract

Background: High US sodium intake and national reduction efforts necessitate developing a feasible and valid monitoring method across the distribution of low-to-high sodium intake.

Objective: We examined a statistical approach using timed urine voids to estimate the population distribution of usual 24-h sodium excretion.

Methods: A sample of 407 adults, aged 18-39 y (54% female, 48% black), collected each void in a separate container for 24 h; 133 repeated the procedure 4-11 d later. Four timed voids (morning, afternoon, evening, overnight) were selected from each 24-h collection. We developed gender-specific equations to calibrate total sodium excreted in each of the one-void (e.g., morning) and combined two-void (e.g., morning + afternoon) urines to 24-h sodium excretion. The calibrated sodium excretions were used to estimate the population distribution of usual 24-h sodium excretion. Participants were then randomly assigned to modeling (n = 160) or validation (n = 247) groups to examine the bias in estimated population percentiles.

Results: Median bias in predicting selected percentiles (5th, 25th, 50th, 75th, 95th) of usual 24-h sodium excretion with one-void urines ranged from -367 to 284 mg (-7.7 to 12.2% of the observed usual excretions) for men and -604 to 486 mg (-14.6 to 23.7%) for women, and with two-void urines from -338 to 263 mg (-6.9 to 10.4%) and -166 to 153 mg (-4.1 to 8.1%), respectively. Four of the 6 two-void urine combinations produced no significant bias in predicting selected percentiles.

Conclusions: Our approach to estimate the population usual 24-h sodium excretion, which uses calibrated timed-void sodium to account for day-to-day variation and covariance between measurement errors, produced percentile estimates with relatively low biases across low-to-high sodium excretions. This may provide a low-burden, low-cost alternative to 24-h collections in monitoring population sodium intake among healthy young adults and merits further investigation in other population subgroups.

Trial registration: ClinicalTrials.gov NCT01631240.

Keywords: 24-hour urine collection; calibration; nutrition survey; population distribution; sodium; timed urine void; usual sodium intake.

Conflict of interest statement

Author disclosures: C-Y Wang, AL Carriquiry, T-C Chen, CM Loria, CM Pfeiffer, K Liu, CT Sempos, CG Perrine, and ME Cogswell, no conflicts of interest.

© 2015 American Society for Nutrition.

Figures

FIGURE 1

Population distributions of usual 24-h sodium excretion: observed vs. estimated from evening, overnight, or evening-overnight combination specimens for males (A) and females (B) aged 18–39 y. The observed population usual distribution of 24-h sodium excretion was derived with use of 24-h urine collections from the overall study sample (i.e., all 407 participants; 274 with one and 133 with two 24-h urine collections). The estimated usual 24-h sodium excretions were derived with use of calibrated sodium excretions from urine voids collected in the evening, overnight, or the combination of evening and overnight specimens. All observed and estimated population usual sodium excretion distributions were adjusted for day-to-day variation in excretion and day of the week.

FIGURE 2

Median bias and median relative bias of selected percentiles of usual 24-h sodium excretion estimated from one-void specimens by gender and the timing of urine void collection among 100 validation groups.

FIGURE 3

Median bias and median relative bias of selected percentiles of usual 24-h sodium excretion estimated from two-void urine combinations by gender and the timing of urine void collection among 100 validation groups.