The Impact of Using Different Methods to Assess Completeness of 24‐Hour Urine Collection on Estimating Dietary Sodium

Abstract

The standard for population‐based surveillance of dietary sodium intake is 24‐hour urine testing; however, this may be affected by incomplete urine collection. The impact of different indirect methods of assessing completeness of collection on estimated sodium ingestion has not been established. The authors enlisted 507 participants from an existing community study in 2009 to collect 24‐hour urine samples. Several methods of assessing completeness of urine collection were tested. Mean sodium intake varied between 3648 mg/24 h and 7210 mg/24 h depending on the method used. Excluding urine samples collected for longer or shorter than 24 hours increased the estimated urine sodium excretion, even when corrections for the variation in timed collections were applied. Until an accurate method of indirectly assessing completeness of urine collection is identified, the gold standard of administering para‐aminobenzoic acid is recommended. Efforts to ensure participants collect complete urine samples are also warranted.

Increased dietary sodium increases blood pressure and causes hypertension with associations in healthy populations for increasing cardiovascular events.1, 2 To support public health policy and to plan effective health promotion programs, it is important to establish reliable benchmark levels of dietary sodium, which also permit evaluation of potential changes over time among various target groups.3

Dietary sodium can be estimated by using either food frequency questionnaires or a 24‐hour recall approach. Both methods tend to underestimate sodium intake and rely on accurate information on the sodium content of foods or questionnaires that have been validated for dietary sodium.4 Since 90% or more of ingested sodium is excreted by the kidneys, the gold standard for assessing sodium intake in populations is 24‐hour urinary sodium excretion.

Collecting 24‐hour urine samples is problematic, however, with a high participant burden and the potential for undercollection or overcollection of urine. Several methods have been developed to assess whether 24‐hour urine samples are complete. The gold standard involves administering para‐aminobenzoic acid (PABA) three times during the urine collection. The administered PABA is almost completely excreted in the urine during a 24‐hour urine collection. Many indirect methods for assessing urine collection are commonly used to avoid the expense and complexity of utilizing PABA. The potential impact of using these different indirect methods to assess completeness of urine collection, to our knowledge, has not been adequately assessed.

We recruited volunteers from the Champlain Community Heart Health Study (CCHHS)5 in Eastern Ontario to conduct a 24‐hour urine survey for dietary sodium and to explore related methodological issues. The findings were expected to guide a possible nationwide population salt intake surveillance program.

Methods

Study Population

In 2009, participants of the CCHHS were invited, in a random sequence, to join a 24‐hour dietary sodium urine survey called the Champlain Heart Urinary Sodium Study (CHUSS). Recruitment was stopped after 507 volunteers consented. Volunteer participants were requested to fill out a self‐administrated questionnaire and undergo physical measurements, after which participants were scheduled for 24‐hour urine collection, according to a previously published protocol.6 The questionnaire items included demographic information, the presence of any chronic diseases, medications, tobacco and alcohol use, family health history, and physical activity. The physical measures included height, weight, waist and hip circumference, and blood pressure. PABA was not administered during collection as part of the study. Ethical approval was obtained from the University of Ottawa Heart Institute review board.

Urine Collection

The enlisted participants were instructed to collect urine for 24 hours in special containers (4 L WM HDPE, product no: CA73560‐018 from VWR Canada). No preservative was added to the containers as use of boric acid was not approved by the ethics review board. Participants were requested to report incomplete collections, including the total number of hours and minutes of urine collection if less or more than 24 hours, which we rounded to the nearest hour. Containers were then forwarded to the medical laboratory of the Ottawa Hospital for analysis.

Laboratory Methods

The urine volume was measured, the sample mixed, and an aliquot of 30 mL was taken. The sodium concentration was determined by direct ion‐selective electrode methods. Measurements were performed with an Optima Clinical Chemistry Analyzer (BioSys Laboratories, Pasadena, CA). For total urine volumes collected for less or more than 24 hours, a proportional adjustment was made to standardize to 24 hours. Total 24‐hour excretion was calculated by multiplying the measured sodium concentration with the total urine volume. The time‐adjusted urine volume was then multiplied by the reported sodium concentration (mmol/L) and the atomic weight of sodium (22.989 mg/mmol) to calculate the sodium intake (mg/24 h). A factor of 113 was used for creatinine.7

Data Restrictions

Because of relatively few respondents in the oldest and youngest age groups, data analysis was restricted to participants aged between 40 and 69 years. Records with inadequate information to calculate body mass index were also excluded. Different methods were used to assess the impact on estimated dietary sodium. However, for the analyses presented to ensure overall relevance to other studies, samples were included only if there was (1) a minimum urine volume of 0.5 L, and (2) a self‐reported complete collection (missing no more than a “few drops”).

Participants included in the analyses were required to have no active menstruation and a measured creatinine excretion rate (mCER). Creatinine excretion rate is often used to determine the accuracy of urine collection. Men typically excrete between 15 mg/kg/24 h and 25 mg/kg/24 h, while women excrete from 10 mg/kg/24 h to 20 mg/kg/24 h.8 However, creatinine excretion rates are affected by various factors, including muscle mass, kidney function, substances competing with creatinine for renal transporters, ethnicity, and age. Furthermore, there is significant variability within patients reflecting the most recent diet. Therefore, four creatinine‐based urine inclusion/exclusion criteria were applied and the results were compared. One analysis excluded those samples for which the mCER was biologically implausible (mCER <350 or >3500 mg/24 h). The second was to include only those individuals whose creatinine levels fell within the standard creatinine excretion rates (15–25 mg/kg/24 h for men and 10–20 mg/kg/24 h for women). The third approach was to exclude only those with low values of mCER creatinine index (the index is the mCER divided by 21, and a value <0.7 is considered low), which is equivalent to mCER <15 mg/kg/24 h.9 A linear regression model was developed in which the expected creatinine excretion rate (pCER) was estimated from a regression equation, built only on records with standard creatinine rates.10 Samples were then excluded if their 24‐hour creatinine excretion was <15% or >50% of their predicted pCER.

Results

Following a randomized sequential invitation process, 1010 of the 1443 CCHHS participants were contacted to obtain 507 who consented to collect urine samples. Exclusion criteria resulted in a sample size of 380 for the main analysis. The exclusions included 13 participants younger than 40 years, 28 older than 69 years, 28 missing height or weight or both measures, seven with 24‐hour urine collections <0.5 L, and 76 urine collections reported to be missing more than a few drops of urine. Several urine collections met more than one exclusion criteria.

A total of 66% of the participants reported collecting urine for 24 hours; the distribution was skewed to undercollection but ranged from 8 hours to 34 hours. A total of 32% of participants indicated that some urine was spilled. Of all the participants, 14% (n=76) indicated they were missing many drops of urine while 29% indicated it was only a few drops.

The Tables provide estimates of 24‐hour urine sodium when a variety of additional exclusion criteria were applied to correct for 24‐hour urine collection and or exclude incomplete urine collections. Table 1 presents the main analysis where urine collections that varied from 24 hours were adjusted to represent a 24‐hour urine collection. When the creatinine‐based exclusion criteria were removed, the analysis file contained 380 records and resulted in a mean urinary sodium excretion of 6128 mg/24 h. In contrast, the lowest value for 24‐hour urine sodium was obtained by excluding urine samples based on the mCER criteria. Applying the pCER exclusion criteria from ±15% up to ±50% changed estimated 24‐hour urine sodium minimally from around 4500 mg/24 h but very substantively impacted the sample size from 83 to 270 individuals.

Table 1.

Estimates of 24‐Hour Dietary Sodium Using Different Methods of Excluding Invalid Urine Collections^a

Completeness of Collection Criteria	Number	Mean Sodium, mg	SE Mean Sodium, mg	Low 95% CI Sodium, mg	High 95% CI Sodium, mg	Median Sodium, mg
Unadjusted for time and excluding those with mCER <350 or >3500 mg/24 h	287	4255	122	4015	4495	3866
24‐h time‐adjusted and excluding those with mCER <350 or >3500 mg/24 h	287	4447	120	4211	4682	4218
24‐h time‐adjusted and removing creatinine criteria	380	6128	259	5621	6635	4910
24‐h time‐adjusted and excluding samples with mCER <15 mg/kg/24 h	355	6274	261	5763	6786	5046
24‐h time‐adjusted and including samples with 15–25 mg/kg/24 h for men and 10–20 mg/kg/24 h for women	104	3648	93	3467	3830	3317
24‐h time‐adjusted and excluding pCER ≥15%b	83	4552	106	4345	4759	4397
24‐h time‐adjusted and excluding pCER ≥25%b	150	4530	112	4310	4750	4368
24‐h time‐adjusted and excluding pCER ≥35%b	215	4488	112	4268	4707	4303
24‐h time‐adjusted and excluding pCER ≥50%b	270	4551	125	4307	4795	4269

Abbreviations: CI, confidence interval; mCER, measured creatinine excretion rate; SE, standard error. ^aUrine samples from some participants were excluded, including participants younger than 40 or older than 69 years, those with urine volumes <0.5 L, those reporting many drops of urine missing, and those who were menstruating during the urine collection. ^bBuilt on a linear regression model to estimate creatinine excretion rate (pCER).

The data in Table 2, which exclude individuals who collected urine for more than or less than 24 hours, indicate substantively higher estimates for 24‐hour urine sodium. Similar to Table 1 when the creatinine‐based exclusion criteria were removed, the highest value for average estimated 24‐hour urine sodium was obtained (7264 mg/24 h) and the lowest value for 24‐hour urine sodium was obtained by excluding urine samples based on the mCER criteria (3910 mg/24 h).

Table 2.

Estimates of 24‐Hour Dietary Sodium Using Different Methods of Excluding Invalid Urine Collections^a

Completeness of Collection Criteria	Number	Mean Sodium, mg	SE Mean Sodium, mg	Low 95% CI Sodium, mg	High 95% CI Sodium, mg	Median Sodium, mg
Excluding those with mCER <350 or >3500 mg/24 h	156	5040	133	4779	5301	4703
Removing creatinine criteria	247	7370	288	6805	7935	6259
Excluding samples with mCER <15 mg/kg/24 h	234	7514	289	6948	8080	6422
Including samples with 15–25 mg/kg/24 h for men and 10–20 mg/kg/24 h for women	33	3497	100	3301	3693	3195
Excluding pCER ≥15%b	49	4770	114	4546	4995	4538
Excluding pCER ≥25%b	84	4906	122	4667	5145	4645
Excluding pCER ≥35%b	112	5026	123	4786	5267	4759
Excluding pCER ≥50%b	146	5190	139	4918	5463	4850

Abbreviations: CI, confidence interval; mCER, measured creatinine excretion rate; SE, standard error. ^aUrine samples from some participants were excluded, including participants younger than 40 or older than 69 years, those with urine volumes <0.5 L, those reporting many drops of urine missing, and those who were menstruating during the urine collection. ^bBuilt on a linear regression model to estimate creatinine excretion rate (pCER).

Discussion

There was marked variation in estimates of sodium excretion in our study depending on the indirect method used to exclude incomplete 24‐hour urine collections. The range fell between 3648 mg sodium/24 h and 7264 mg sodium/24 h. This twofold difference has substantive implications on assessments of sodium intake at the population level, and is very likely to make assessments of changes or trends in intake tenuous. The degree of variation indicates a strong need to establish a valid indirect method or to apply the gold standard direct method utilizing PABA.11 The high frequency of participants indicating incomplete collections also calls for better instructions and attention to urine collection. If nearly all participants had collected complete urine samples, indirect methods of assessing completeness of urine may have produced better results.

The specific choice of creatinine‐based exclusion criteria had a significant impact on the mean sodium excretion values in the study population. When we rejected urine samples based on the biologically implausible (mCER) or the linear regression model expected creatinine excretion rate (pCER) as the inclusion criteria, the mean sodium values were all significantly lower than in the analysis with no creatinine‐based exclusion criteria. The fact that high proportions of samples fall outside the expected range of mCER for almost all methods further emphasizes the importance of rigorous application of the urine collection procedure with validity checks.

Study Limitations

The assessment of urinary sodium in this study did not make use of PABA; hence, we cannot evaluate the differing indirect methods of assessing completeness of urine collection against a gold standard. Assessing the validity of differing indirect methods of correcting completeness of urine collections compared with the gold standard of PABA excretion is an important next step. Currently, we can only comment that in the absence of a gold standard, there is little agreement among the indirect methods. Creatinine, when used as a surrogate of complete collections, can result in a wide range of values for sodium excretion. There are several potential explanations why differing formula for incomplete urine collections do not provide a consistent estimate of dietary sodium. Urinary creatinine excretion commonly varies by 25% from day to day reflecting dietary sources of creatinine, muscle breakdown, and changes in glomerular filtration rate.12 Sodium excretion is much more variable and under tight physiologic control, being impacted by state of hydration, activation of the renin‐angiotensin‐aldosterone axis, sympathetic nervous system and naturetic peptides, awake‐sleep cycle, and many commonly used drugs (eg, nonsteroidal anti‐inflammatory drugs and diuretics) as well as body sodium stores in addition to recent dietary sodium intake.13 In particular, when vascular volume is low (eg, dehydration) creatinine is secreted and is present in urine in excess of that predicted by glomerular filtration rate while sodium will be very actively reabsorbed and little will be excreted relative to the glomerular filtration rate. The differing regulation of creatinine and sodium urinary excretion makes creatinine at best a crude method of correcting a 24‐hour urine estimate of sodium excretion for lost or excess urine collection.

We adjusted incomplete collections to 24 hours assuming that the collection that was obtained reflected an average rate of sodium excretion. However, simply accounting for changes in length of collection is also likely to cause inaccuracies. Nearly all sodium ingested is excreted within 6 hours; hence, missing similar lengths of urine collections that are soon after ingestion vs many hours after ingestion will have markedly different impacts on estimates of sodium ingested.14 It was surprising that when the methods used to exclude urine samples based on inappropriate creatinine excretion were removed, estimates of sodium ingestion increased. We do not have a clear explanation for this finding, but suggest that many of the people excluded by more rigorous methods may have overcollected their urine and/or had higher dietary sodium. High dietary sodium can be associated with higher physical activity and hence muscle mass, which could have exceeded the creatinine exclusion standards of some methods.

Our findings are limited in generalizability to men and women within the age range of 40 to 69 years. Insufficient numbers of participants were recruited to permit generalization of those young and elderly outside the age range.

Conclusions

This study demonstrates that differing indirect methods of assessing completeness of urine collections can have substantive impacts on estimates of dietary sodium. PABA should be utilized in assessing completeness of 24‐hour urine collections and careful instruction and supervision utilized for the urine collections. Research to establish more pragmatic yet valid indirect methods is also warranted.15

J Clin Hypertens (Greenwich). 2016;18 581–584. DOI: 10.1111/jch.12716. © 2015 Wiley Periodicals, Inc.