A systematic review and meta-analysis of 24-h urinary output of children and adolescents: impact on the assessment of iodine status using urinary biomarkers

Purpose Urinary iodine concentration (UIC (μg/ml) from spot urine samples collected from school-aged children is used to determine the iodine status of populations. Some studies further extrapolate UIC to represent daily iodine intake, based on the assumption that children pass approximately 1 L urine over 24-h, but this has never been assessed in population studies. Therefore, the present review aimed to collate and produce an estimate of the average 24-h urine volume of children and adolescents (> 1 year and < 19 years) from published studies. Methods EBSCOHOST and EMBASE databases were searched to identify studies which reported the mean 24-h urinary volume of healthy children (> 1 year and < 19 years). The overall mean (95% CI) estimate of 24-h urine volume was determined using a random effects model, broken down by age group. Results Of the 44 studies identified, a meta-analysis of 27 studies, with at least one criterion for assessing the completeness of urine collections, indicated that the mean urine volume of 2–19 year olds was 773 (654, 893) (95% CI) mL/24-h. When broken down by age group, mean (95% CI) 24-h urine volume was 531 mL/day (454, 607) for 2–5 year olds, 771 mL/day (734, 808) for 6–12 year olds, and 1067 mL/day (855, 1279) for 13–19 year olds. Conclusions These results demonstrate that the average urine volume of children aged 2–12 years is less than 1 L, therefore, misclassification of iodine intakes may occur when urine volumes fall below or above 1 L. Future studies utilizing spot urine samples to assess iodine status should consider this when extrapolating UIC to represent iodine intakes of a population. Electronic supplementary material The online version of this article (10.1007/s00394-019-02151-w) contains supplementary material, which is available to authorized users.


Introduction
Monitoring the iodine nutrition of populations and individuals is important to identify those at risk of deficiency, as deficiencies during childhood have been linked with impaired cognitive and motor functions in schoolchildren [1,2]. Current recommendations for assessing the iodine nutrition of populations using spot urine samples from school-aged children were developed by the World Health Organization (WHO) over a number of years   [3][4][5][6][7]. Whilst the original version of these recommendations was based on the association between a daily estimated iodine intake, extrapolated from creatinine excretion, equivalent to 100 µg of iodine and reduced goiter prevalence among children and adults [8], later iterations extrapolated this value to represent a concentration, urinary iodine concentration (UIC), expressed as µg/L [6]. Whilst these recommendations were originally developed as a marker of iodine status within populations, some studies have utilized UIC from spot urine samples as an indicator of iodine intake [9][10][11][12][13][14]. In such instances, issues may arise as UIC would only be reflective of daily intake if urine volume was equivalent to 1 L. For example, in a population with a daily urinary volume of 1 L, a UIC of 100 µg/L could be extrapolated to be indicative of a daily iodine intake of 100 µg/24-h. As this value is equal to the daily intake originally associated with reduced goiter prevalence [8], this population can be classified as having a sufficient iodine intake. However, in a population of children with a daily urine volume closer to 0.5 L, a UIC of 100 µg/L may be indicative of a daily iodine intake closer to 50 µg/24-h. This could result in the misclassification of this population as iodine sufficient when their daily iodine intake may, in fact, place them at risk of developing iodine deficiency disorders.
Therefore, iodine monitoring programs which have extrapolated daily iodine intake from UIC determined from spot urine samples in populations of children and adolescents [9][10][11][12][13][14] and have not taken the lower daily urinary output into account, may be inaccurately estimating iodine intakes. As such, this may have resulted in the misclassification of populations as having sufficient iodine intakes, when their true intakes may be lower than the 100 µg/day originally associated with reduced goiter prevalence. Such misclassifications may have prevented the implementation of necessary iodine fortification programs. To date, there has been no global systematic collation of the average 24-h urine volume of children and adolescents. This information could help researchers estimate population dietary iodine intake of children and adolescents from spot urine samples. Therefore, the aim of the current study was to estimate the average 24-h urine volume measured in healthy children and adolescents, by conducting a systematic review and meta-analysis of studies which have reported the 24-h urine volume of children and adolescents aged 2-19 years.

Methods
This protocol adheres to the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015 statement [15] and was registered with the International Prospective Register of Systematic Reviews (PROS-PERO) (registration number CRD42016033682).

Information sources and search
A search strategy was developed to identify papers published up to October 2018, which have reported the 24-h urine volume of children and adolescents (> 1 year and ≤ 19 years). An electronic literature search of EBSCOHOST (MEDLINE complete, CINAHL, Academic Search Complete and Global Health) and EMBASE databases was conducted. The search strategy was developed in consultation with a research librarian. Free text keywords were used to conduct the search. Search criteria specific to each database are outlined in Table 1. The search strategy was piloted across each database to improve the effectiveness of the final search. Only peer-reviewed original research articles published in English and conducted in humans were included. It was beyond the scope of this review to include and examine sources from 'grey' literature. The reference lists of included studies identified through the search were also reviewed.

Inclusion/exclusion criteria
Only peer-reviewed original research studies were included and any reviews, meta-analyses, editorials, case reports, conference proceedings or other grey literature identified through the search were excluded at the screening stage. As the primary focus of this review was to determine the average daily urinary output of children and adolescents, only studies which reported the 24-h urinary output of healthy children and adolescents > 1 and ≤ 19 years of age were included. Where multiple published reports were available from the same study, the most recently published and/or the study with the largest sample size was included.
All papers identified from the initial electronic search process were imported into an Endnote library, and duplicates removed. Titles and abstracts were screened and studies included based on the eligibility criteria as outlined above. Two investigators (KB and MW) independently screened the titles and abstracts of the articles independently to assess eligibility for inclusion. If agreement was reached, articles were either excluded or moved to the next stage (full-text). If agreement was not reached, the article was moved to the full-text stage. Following this screening process, the full text of eligible studies was retrieved and studies which collected 24-h urine samples but did not report the final 24-h urine volume were excluded. At this stage, the reference lists of included studies were scanned, and the full text of any relevant studies retrieved and reviewed for inclusion. The PRISMA flow chart [16] was used to document the number of studies identified during the search process and those excluded and included according to the outlined eligibility criteria (Fig. 1).
As multiple studies had utilised data from the Dortmund Nutritional and Anthropometric Longitudinally Designed (DONALD) Study [17], cross-checking of study dates and participant characteristics was carried out to minimise participant overlap between the studies. Some studies (n = 7) [18][19][20][21][22][23][24] were excluded from the final analysis as they did not report which years of data collection had been analysed, therefore, possible participant overlap with other studies could not be determined. No additional information apart from that published was available from authors. Where participant overlap was possible, the study with the larger participant number was included in the final analysis. Of the 27 DONALD studies initially identified, six studies [25][26][27][28][29][30] were considered for the final analyses as they captured the full range of data collection years and included the largest number of participants whilst minimising possible participant overlap.

Data extraction and synthesis
Data extraction was completed using a data extraction template ( Table 2). The template was initially piloted on five eligible studies and modifications made where necessary. As 24-h urinary creatinine excretion either alone, in relation to expected creatinine based on sex and/or weight, is often used as a marker for complete urine collection under the assumption that urinary creatinine excretion, as an indicator of body mass is stable within individuals from day to day [31][32][33], data pertaining to 24-h excretion of creatinine was also extracted where reported.

Quality assessment
The quality of the studies included in this review was assessed using a modified version of the Newcastle-Ottawa scale (NOS) for cohort studies [34], as all studies included in the final synthesis were of a cross-sectional study design. The NOS was modified to suit the context of the studies included in the review and particular consideration was made towards the 24-h urine collection methods used in each study. This scale assigns stars to indicate higher quality based on three broad criteria specific to the design of the study: (1) selection (representativeness of the study sample); (2) comparability of the findings (normalisation of the results to a 24-h period); and (3) assessment of outcome (quality of the reported 24-h urine collection methodology) (Online Supporting Material). Studies were categorised as 'high' 'medium' or 'low' quality according to the number of stars they received (out of a maximum of 10 stars: low: 0-3; medium: 4-7; high 8-10). As only three studies provided sufficient detail on their urine collection protocol to be classified as "high" quality [35][36][37], we included a second category of quality assessment, based on studies which had reported at least one criterion for the assessment of urine collection completeness.

Initial analysis
The overall mean [95% confidence interval (CI)] estimate of urine volume for all 44 studies was determined using a random effects model and presented for the group as a whole (i.e., 2-19 year olds), as well as broken down by age group.

Primary analysis: studies with ≥ 1 criterion for assessing the completeness of urine samples
As the inclusion of criteria for assessing the completeness of urine collections can result in the exclusion of over/under collectors, the primary analysis was limited to only those studies which reported at least one criterion for assessing the completeness of the included urine samples (n = 27, "Primary Analysis, Fig. 2). The overall mean (95% CI) estimate of daily urine volume was determined using a random effects model and displayed in forest plots, broken down by age group.
A one-way ANOVA was used to assess the differences in urine volume across the three age groups. In addition, a one-way ANOVA was used to assess differences in the urine volume between those studies which did not report any criteria for assessing the completeness of included urine samples and those which reported > 1 and > 2 criteria. Tukey's post hoc tests were performed to determine significant differences between age subgroups. Heterogeneity was analysed using the I 2 and Q statistics. The coefficient of variation (CV) of the mean volume for each age group was derived from the mean and SD random effects analysis and was calculated by dividing the SD for each age group's urine volume by the mean and multiplying by 100. A two-sample t test used to assess the difference in volumes determined for the initial analysis compared to the primary analysis (i.e., limited to studies with > 1 criterion for assessing the completeness of the urine samples) across the three age groups. As 11 studies had presented the results broken down by gender [26, 28-30, 37, 53, 65, 66, 68, 70, 81], a two-sample t test was used to evaluate the difference in 24-h urine volume between genders.
To examine whether climate had an impact on overall urine volume, studies were classified according to climate based on their proximity to the equator. Studies which were conducted in countries which lie between the Tropic of Cancer and Tropic of Capricorn (23.5° north and south) (2 studies) [54,55] were classified as having a "warm" climate, whereas studies which fell outside of this area were classified as having a "cold" climate (25 studies) [25, 26, 28-30, 35-37, 43, 46, 48, 52, 56, 57, 60, 61, 64, 66, 67, 70, 73-75, 81, 82]. A two sample t test was used to assess the difference in urine volume between climates.
When comparing the mean urine volume between those studies which reported at least one criterion for assessing the completeness of urine samples and those which reported none, the only difference in mean urine volume was amongst the 2-5 year old age group. In this age group, the mean urine volume in the initial analysis (i.e., all studies were included) was 27% lower compared to the primary analysis (i.e., when the analysis was limited to only those studies with ≥ 1 criterion for assessing the completeness of urine samples (386 vs. 529 mL/24-h, P = 0.001). Of the 26 studies with at least one criterion for assessing the completeness of the urine samples, only twelve [25, 28, 29, 35-37, 43, 48, 52, 55, 61, 64] had at least two urine criteria (n = 2867, 3191 urine samples). In these twelve studies, the mean (95% CI urine volume estimate was 742 (639, 844) mL/24-h. There was no difference in the mean (95% CI) urine volume estimate of these studies, compared to those with only one assessment criterion [n = 15, 798 (664, 932) mL/24-h, P = 0.33] or those with none [n = 18, 635 (577, 692) mL/24-hs P = 0.28].
For those studies with at least one criterion for assessing the completeness of urine collections there was less variation in daily urinary volume for those 6-12 years compared to 2-5 and 13-19 year olds. The co-efficient of variation (CV) for 6-12 year olds was 13% compared to 20% in 2-5 and 13-19 year olds. There was no difference in CVs between those studies with no reported criterion for completeness and those reporting at least one criterion (overall CV: 32% vs. 27%). In contrast, the CV for those aged between 13 and 19 years was reduced to 12% in those studies utilising least two criteria for assessing completeness of urine collection, compared with 20% for those studies utilising only one criterion for assessment for completeness.

Discussion
This is the first study to systematically review the 24-h urine volume of children and adolescents. The overall 24-h overall urine volume estimate (95% CI) of 2-19 year olds was 778 (661, 895) mL/24-h urine. As expected, older children had higher urine volumes with children in the oldest age group (13-19 years) having a 28% higher 24-h urine volume compared to those aged 6-12 years (1067 vs. 771 mL/24h, P < 0.001) and approximately 50% higher urine volume compared to those aged 2-5 years (1067 vs. 531 mL/24-h, P < 0.001).
As approximately 90% of ingested iodine is excreted in the urine within 24-48 h [87], current recommendations for assessing the severity of iodine deficiency within a population are based on the measurement of urinary iodine concentration (UIC), expressed as µg iodine per liter of urine, in random 'spot' urine samples collected from school-aged children (i.e., 6-12 years [6]). Although these recommendations were originally based on the observation that goiter prevalence was < 10% in populations of children and adults where the average daily iodine intakes were > 100 µg, later iterations extrapolated this value to represent a concentration, expressed as µg/L [4]. However, results from this analysis indicate that the average 24-h urine volume of schoolaged children, the group commonly recommended for use in population iodine monitoring, is not 1 L and is closer to 0.8 L. Therefore, a median spot urine concentration of 100 µg/L (extrapolated to a median iodine intake of 100 µg/ day) would overestimate iodine intake by approximately 30 µg/day. Iodine excretion from spot urine samples is most often expressed as a concentration or as a ratio to creatinine excretion (I/Cr, µg iodine/g creatinine) [29,88,89]. The use of I/ Cr to estimate daily iodine intake is believed to provide an accurate estimate of daily iodine excretion from spot urine samples as creatinine, an endogenous indicator of lean body mass is relatively constant from day to day in healthy adult populations [32,90]. Whilst this is true in adults, estimating expected creatinine excretion values for children can be difficult as creatinine excretion can be affected by muscle mass, age, gender, ethnicity and onset of puberty [91]. Some equations for estimating daily creatinine excretion are able to account for these factors, whilst others provide a more crude estimate of daily creatinine excretion [92].
In addition, whilst the variation in individual iodine excretion between days is largely dependant on the iodine content of the diet, iodine excretion has also been found to vary over the course of the day in individuals [29,88,89,[92][93][94][95][96][97]. One study conducted in 42 adults and children (aged 4-60 years) found that urinary iodine excretion varied significantly by the timing of collection (P < 0.001), with lowest levels occurring in the morning and peaks observed following meals [98]. A recent systematic review of studies comparing spot and 24-h urine samples for estimating the iodine intakes of a population concluded that there is currently not enough evidence to determine whether iodine intake determined from spot urine samples provides an accurate reflection of daily iodine intake, as measured using 24-h urine samples [99].
Furthermore, whilst the WHO recommendation for assessing the iodine status of populations are primarily meant for use in school-aged children, and were derived primarily on data based on goiter prevalence estimates in school-aged children, they have also been used to define the iodine status of adult populations [100][101][102][103]. Issues concerning different urinary volume outputs among different subsets of the population and implications for iodine nutrition assessment have been previously identified by Zimmerman and Andersson [104]. They highlighted that as the urine volume of adults is closer to 1.5 L [27,105], the use of the median UIC determined using spot urine samples could result in the underestimation of the iodine intakes of adult individuals within the population [104]. This was demonstrated in a recent study in 301 adults (18-64 years) from New Zealand, which compared median UIC from 24-h urine samples to the WHO criteria, both with and without adjustment for total urine volume [101]. This sample of adults was classified as iodine deficient using the WHO criteria, based on a median UIC of 73 µg/L. However, the measured 24-h UIE, which accounts for urine volume and averaged 2 L was closer to 127 µg/day [101]. This value is in excess of the 100 µg/day originally associated with reduced goiter prevalence [8] and would indicate that the iodine intakes of this group of adults may be sufficient [8]. The New Zealand study demonstrates the potential impact of not accounting for the daily urine volume may have on the assessment of iodine deficiency in populations when UIC determined from spot urine samples is used as a surrogate index of iodine intake.

Strengths and limitations
We observed considerable between-study heterogeneity across all three age groups in the primary analysis limited to studies with > 1 indicator of urine completeness. For all three age groups the age range of participants included within each group varied considerably. For example, within the 2-5 year age group one study consisted of participants aged 1.5-3.5 years [56] whilst another consisted of participants aged 3-6 years [72]. Differences in both the age range of participants as well as the number of participants between studies, along with season of assessment and overall diet composition may have contributed to the observed heterogeneity.
In addition, there were considerable variations in the mean/median reported urine volumes, even within the three age groups, particularly for 2-5 year olds. This may represent the practical difficulties in obtaining accurate 24 h urine collections in young children, however, it is important to note that the included studies did not consistently report their 24-h urine collection protocol, nor the indicators used to assess the completeness of included urine samples. Of the 44 studies reviewed, 17 studies (40%) did not report at least one indicator for completeness of the 24-h urine samples. The only difference in the mean urine volume estimate between the total sample and those studies with ≥ 1 indicator for completeness of urine samples, was seen among the youngest age group (2-5 years) where average urine volume was 143 ml/24-h less in the studies which reported no criterion for assessing the completeness of included urine samples. A recent systematic review of methods for assessing the completeness of 24-h urine collections in adults and children (15-89 years) concluded that that the use of two or more indicators for assessing urine completeness increases the likelihood of detecting incomplete samples, thus increasing the validity of the results [33]. Our findings are contrary to this in that we found no difference in urine volume among those studies that had at least two urine assessment criteria, compared to those with only one. In the present review, only 12 studies utilised more than one criterion [25, 28, 29, 35-37, 43, 48, 52, 55, 61, 64], and there was no difference in mean 24-h urine volume estimated from these studies compared to the overall estimate from all 44 studies. However, there was less variation in daily urinary volume in the 6-12 year group (CV 13%). Overall there seemed to be little impact on urine volume variation of including a number of criteria for completeness of 24 h urine collection, except for the [13][14][15][16][17][18][19] year age group where studies that included at least two criteria for completeness appeared to have less variation (CV 16%) compared with studies with only one criterion (CV 20%).
In this analysis, only two studies collected 24-h urine samples from countries classified as having a "warm" climate", compared to 25 studies from a "cold" climate. There is also considerable within and between person daily variability in iodine excretion [92,95,96,98,106,107]. One study conducted in 42 adults and children (aged 4-60 years) observed that the lowest level of iodine excretion occurred in the morning with peaks observed following meals [98]. Furthermore, a study in adults noted that UIC determined from a fasting spot urine samples was 10% lower than that determined in a non-fasting spot urine sample [96]. Although this variation has yet to be assessed in children, this study indicates that the timing of a spot urine sample used to estimate the iodine intake of a population may also have a significant impact on the overall assessment of iodine nutrition. Therefore, it is clear that a number of factors need to be considered when making population estimates of iodine intake using spot urine collections across the age range from early childhood to adolescence.

Conclusion
This is the first systematic review to report the average 24-h urine volume of children and adolescents from 44 studies representing 7712 individuals with 3772 individuals within the 6-12 year old age group, which included at least one criterion for completeness of urine collection. The average urine volume in this group was 771 mL, which is less than 1 L. This has implications when extrapolating median iodine values (µg/L) from spot urine samples to daily iodine intakes of 6-12 year old children as the average 24-h urine volume is less than 1 L, potentially resulting in an overestimate of dietary iodine intake in the order of 30%. Future studies employing spot urine samples to determine the iodine status of children and adolescents should consider undertaking 24-h urine collections in a subset of participants, to determine total urine volume and iodine excretion. This will allow the assessment of the accuracy of utilizing UIC as a proxy measure of daily iodine intake and potentially prevent the misclassification of iodine intakes in the population.