Dataset Information

Validity of predictive equations for 24-h urinary potassium excretion based on timing of spot urine collection among adults: the MESA and CARDIA Urinary Sodium Study and NHANES Urinary Sodium Calibration Study.

ABSTRACT:

Background

24-h urine collections are the suggested method to measure daily urinary potassium excretion (uK) but are costly and burdensome to implement.

Objective

This study tested how well existing equations with the use of spot urine samples can estimate 24-h uK and if accuracy varies by timing of spot urine collection, age, race, or sex.

Design

This cross-sectional study used data from 407 participants aged 18-39 y from the Washington, DC area in 2011 and 554 participants aged 45-79 y from Chicago in 2013. Spot urine samples were collected in individual containers for 24 h, and 1 for each timed period (morning, afternoon, evening, and overnight) was selected. For each selected timed spot urine, 24-h uK was predicted through the use of published equations. Difference (bias) between predicted and measured 24-h uK was calculated for each timed period and within age, race, and sex subgroups. Individual-level differences were assessed through the use of Bland-Altman plots and correlation tests.

Results

For all equations, regardless of the timing of spot urine, mean bias was usually significantly different than 0. No one prediction equation was unbiased across all sex, race, and age subgroups. With the use of the Kawasaki and Tanaka equations, 24-h uK was overestimated at low levels and underestimated at high levels, whereas observed differential bias with the Mage equation was in the opposite direction. Depending on prediction equation and timing of urine sample, 61-75% of individual 24-h uKs were misclassified among 500-mg incremental categories from <1500 to ≥3000 mg. Correlations between predicted and measured 24-h uK were poor to moderate (0.19-0.71).

Conclusion

Because predicted 24-h uK accuracy varies by timing of spot urine collection, published prediction equations, and within age-race-sex subgroups, study results making use of predicted 24-h uK in association with health outcomes should be interpreted with caution. It is possible that a more accurate prediction equation can be developed leading to different results.

SUBMITTER: Mercado CI

PROVIDER: S-EPMC6454816 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:Given a global focus on salt reduction efforts to reduce cardiovascular risk, it is important to obtain accurate measures of salt intake on a population level. This study determined firstly whether adjustment for intra-individual variation in urinary sodium (Na) excretion using three repeated 24 h collections affects daily estimates and whether the use of repeated spot urine samples results in better prediction of 24 h Na compared to a single collection. Twenty three community-dwelling men and women from South Africa (mean age 59.7 years (SD = 15.6)) participating in the World Health Organization Study on global AGEing and adult health (WHO-SAGE) Wave 3 study collected 24 h and spot early morning urine samples over three consecutive days to assess urinary Na excretion. INTERSALT, Tanaka, and Kawasaki prediction equations, with either average or adjusted spot Na values, were used to estimate 24 h Na and compared these against measured 24 h urinary Na. Adjustment was performed by using the ratio of between-person (sb) and total (sobs) variability obtained from repeated measures analysis of variance. Sensitivity of the equations to predict daily urinary Na values below 5 g salt equivalent was calculated. The sb/sobs for urinary Na using three repeated samples for spot and 24 h samples were 0.706 and 0.798, respectively. Correction using analysis of variance for 3 × 24 h collections resulted in contraction of the upper end of the distribution curve (90th centile: 157 to 136 mmoL/day; 95th centile: 220 to 178 mmoL/day). All three prediction equations grossly over-estimated 24 h urinary Na excretion, regardless of whether a single spot urine or repeated collections corrected for intra-individual variation were used. Sensitivity of equations to detect salt intake equivalent values of ≤5 g/day was 13% for INTERSALT, while the other two equations had zero sensitivity. Correcting for intra-individual variability in Na excretion using three 24 h urine collections contracted the distribution curve for high intakes. Repeated collection of spot samples for urinary Na analysis does not improve the accuracy of predicting 24 h Na excretion. Spot urine samples are not appropriate to detect participants with salt intakes below the recommended 5 g/day.