Appropriate Range of Median Urinary Iodine Concentration in 8- to 10-Year-Old Children Based on Generalized Additive Model

Abstract

Background:

The appropriate range of median urinary iodine concentration (MUI) in children has always been controversial. To prevent the occurrence of a goiter epidemic in Shanghai, we explored the appropriate range of MUI by integrating multiple monitoring results.

Methods:

This study summarized and analyzed the monitoring data from 1997, 1999, 2011, 2014, and 2017 of children living in Shanghai. In each monitoring year, the probability-proportional-to-size sampling technique was used to select 30 sampling units. In each sampling unit, one primary school was randomly selected. From each selected school, 40 children 8- to 10-year-old were randomly recruited to measure thyroid volume (Tvol) and their household salt iodine intake.

Results:

In 1997, 1999, 2011, 2014, and 2017, MUI of 8- to 10-year-old children was 228, 214, 182, 171, and 183 μg/L, and median Tvol (MTvol) was 2.9, 1.2, 1.0, 1.8, and 2.8 mL, respectively. There was a linear correlation between goiter rate and MTvol (r = 0.95, p = 0.014; 100 × goiter rate = 1.314 × MTvol −1.287). Generalized additive model (GAM) was used to predict MTvol as follows, MTvol = 0.60689 + 0.00302 MUI +0.999928 s (MUI) −0.05172 mean salt iodized concentrations (MSIs) +0.03481 × 100 × iodized salt coverage rate +0.00000969 per capita disposable income +0.271422 s (per capita disposable income) −0.38772 × monitoring year gap. The results revealed that the average relative error between predicted and actual value was 15.2%. GAM results showed that at 27–277 μg/L MUI, the goiter rate was <5%.

Conclusions:

Iodine status is appropriate in Shanghai. Under the existing economy and MSI, the optimal range of MUI should be 70–277 μg/L in 8- to 10-year-old children living in Shanghai.

Introduction

According to a 2001 report by the World Health Organization (WHO), ∼285 million school-age children live in countries with significant iodine deficiency (ID) and are at risk of iodine deficiency disorders (IDDs) (1). Iodine is indispensable for thyroid hormone synthesis. Long-term ID can lead to IDD, which has irreversible effects on physical and mental development in subjects of all ages, and contributes to endemic goiter and cretinism (2 –4). A meta-analysis conducted by Chinese researchers reported that ID resulted in an IQ loss of ∼10 points in children, and that the IQ of children supplemented with iodine during the critical period of brain development was ∼12 points higher than that of children without iodine supplementation (5). Another meta-analysis that evaluated the relationship between iodine and mental development in children five years of age and under revealed that ID had a substantial impact on mental development (6). Before the implementation of universal salt iodization (USI) in 1994, China had a high prevalence of ID and IDDs (7).

Despite the prevalence of ID, Shanghai is the only provincial-level unit in China that has never experienced an IDD epidemic. Therefore, the iodine status of Shanghai residents has been studied. Since 1994, Shanghai has implemented monitoring programs to assess the iodine status of the population. Based on the monitoring data, the iodine concentration in edible salt has been adjusted four times to maintain the iodine status in the population (8). Currently, the status indicators and the recommended dietary intakes of iodine are established by WHO/United Nations International Children's Emergency Fund (UNICEF)/International Council for Control of Iodine Deficiency Disorders (ICCIDD) (9) with no prior systematic evaluations. Therefore, the applicability and accuracy of the status indicators and dietary intake recommendations have been questioned (10,11).

Median urinary iodine concentration (MUI) is used to evaluate the adequacy of recent iodine intake. MUI from a representative child's spot urine sample can be used to assess the iodine status of a population (12). Thyroid volume (Tvol) of children is the most sensitive index of long-term iodine intake, and goiter rate among children is the most important indicator of IDD (13,14).

Urinary iodine concentration (UIC) has a U-shaped curve relationship with Tvol (15). The general statistical model cannot explain specific changes of this relationship. Some studies have shown that the generalized additive model (GAM), a nonparametric regression method that is suitable for the analysis of various distribution types and nonlinear relationships, could be used for fitting and estimation (16,17). In this study, we evaluated the monitoring data from 1997, 1999, 2005, 2011, 2014, and 2017 using GAM in 8- to 10-year-old children living in Shanghai to determine the factors influencing Tvol and the appropriate range to maintain a lower goiter rate in the population.

Materials and Methods

Population and study design

The data were obtained from the national IDD monitoring program of 1997, 1999, 2011, 2014, and 2017. In each monitoring year, the probability-proportional-to-size sampling technique was used to select 30 sampling units. In each sampling unit, one primary school was randomly selected. From each selected school, 40 children 8- to 10-year old were randomly recruited to measure Tvol and their household salt iodine intake. In addition, >12 students were randomly recruited to measure UIC.

The medical ethics committee from the Center for Endemic Disease Control, Harbin Medical University, approved the study. Written informed consent was obtained from all parents of the participating children.

Tvol estimation and quality control

Tvol was measured by professionals trained by a national expert group, using a portable ultrasound machine. Prof. Liu, who is the corresponding author of this article, has been the leader of the national expert group since 1997. The test was carried out by him in 1997, and the subsequent tests were carried out under his guidance. Under the supervision of Prof. Liu, Z.W. conducted tests in 2011, 2014, and 2017. The thyroid lobe volume was calculated by measuring the depth (d), width (w), and length (l) of each lobe according to the following formula, V (mL) = 0.479 × d × w × l (mm)/1000. Tvol was calculated as the sum of the volumes of both lobes, but the volume of the isthmus was not included (18). Goiter was diagnosed based on the following Chinese criteria (No. WS 276—2007), Tvol >4.5 mL (in children 8 years of age), Tvol >5.0 mL (in children 9 years of age), and Tvol >6.0 mL (in children 10 years of age) (19).

Based on standards reported by the WHO, when the prevalence rate of goiter is >5%, IDD can be suspected (11,12). Therefore, we explored the appropriate range of median UIC for children based on goiter rate ≤5%.

The collection, test, and quality control of household salt sample and urine sample

More than 100 g of household salt sample and 5 mL of urine sample were collected from each participant. Household salt iodized concentrations (SICs) were analyzed by titration except in 1997. UIC was measured by the acid digestion method (20) at the Central Laboratory of Shanghai Municipal Center for Disease Control and Prevention and 17 Districts Center for Disease Control and Prevention in Shanghai. The internal quality control samples for the salt iodine and urinary iodine were provided by the China National Iodine Deficiency Disorders Reference Laboratory. In 1997, household SIC was semiquantitatively determined to distinguish iodized salt from noniodized salt. Iodized salt was defined as salt iodine >5 mg/kg. Iodine intake in children was determined according to the recommended WHO/UNICEF/ICCIDD criteria: insufficient iodine intake, MUI <100 μg/L; adequate iodine intake, MUI 100–199 μg/L; iodine intake above the requirement, MUI 200–299 μg/L; and excessive iodine intake, MUI ≥300 μg/L (1).

Household coverage with iodized salt was defined as the number of households consuming iodized salt divided by the total number of households surveyed.

Statistical analysis

Data processing and statistical analyses were performed using Excel (2010 edition; Microsoft, China) and SAS (9.2 Edition; SAS Institute, Cary, NC). Normally distributed data are expressed as mean ± standard deviation and nonparametric data are expressed as median (25th percentile, 75th percentile). The correlation of two continuous variables was determined by scatter plot and linear correlation. GAM was used to predict median Tvol (MTvol). The degrees of freedom were not specified in a smooth function, and generalized cross-validation was added to automatically determine the optimal degree of freedom of each variable. p < 0.05 was considered statistically significant.

Results

Household SIC, UIC, and Tvol in 8- to 10-year-old children

Mean SIC (MSI) values were 43.8 ± 16.3, 28.9 ± 5.8, 25.8 ± 9.2, and 24.3 ± 5.6 mg/kg in 1999, 2011, 2014, and 2017, respectively. The results showed that MSI was in the range of the standard household SIC each monitoring year. MUI values were 228, 214, 182, 171, and 183 μg/L in 1997, 1999, 2011, 2014, and 2017, respectively. Goiter rate was 3.07%, 0.40%, 0.08%, 0.86%, and 1.90%, and MTvol was 2.9, 1.2, 1.0, 1.8, and 2.8 mL in 1997, 1999, 2011, 2014, and 2017, respectively (Table 1).

Table 1.

Household Salt Iodized Concentration, Urinary Iodine Concentration, and Thyroid Volume in 8- to 10-Year-Old Children

Year	Household SIC (mg/kg)		MUI (μg/L)		Tvol (mL)		Goiter rate (%)
Year	N	$\bar{x}$ ± s	N	Median (P25, P75)	N	Median (P25, P75)	Goiter rate (%)
1997	—	—	1206	228 (132, 357)	1206	2.9 (2.3, 3.6)	3.07
1999	1320	43.8 ± 16.3	1291	214 (130, 308)	1255	1.2 (0.9, 1.5)	0.40
2011	1234	28.9 ± 5.8	360	182 (125, 255)	1234	1.0 (0.8, 1.2)	0.08
2014	1510	25.8 ± 9.2	1510	171 (124, 236)	1510	1.8 (1.1, 2.5)	0.86
2017	1215	24.3 ± 5.6	1215	183 (114, 267)	1209	2.8 (2.0, 3.6)	1.90

MUI, median urinary iodine concentration; SIC, salt iodized concentration; Tvol, thyroid volume.

Relationship between goiter rate and MTvol in 8- to 10-year-old children

The data of goiter rate and MTvol were used for correlation analysis (r = 0.95, p < 0.05), which showed that there was a linear relationship between goiter rate and MTvol (goiter rate × 100 = [1.314 × MTvol] −1.287).

GAM for the prediction of MTvol in 8- to 10-year-old children

The first part of the model fitting consisted of parametric regression analysis (Table 2). MUI, MSI, iodized salt coverage rate, per capita disposable income, and monitoring year gap were statistically significant (p < 0.05). The second part of the model fitting consisted of smoothing components (Supplementary Table S1). The third part of the model consisted of the deviation analysis of the smoothing model, and F suggestions were provided for each smoothing effect in the model, which was used to compare the deviation of the whole model and the nonvariable model (Supplementary Table S2). The deviation results showed that MUI and per capita disposable income were statistically significant (p < 0.05). The prediction model was the following, MTvol = 0.60689 + 0.00302 MUI +0.999928 s (MUI) −0.05172 MSI +0.03481 × 100 × iodized salt coverage rate +0.00000969 per capita disposable income +0.271422 s (per capita disposable income) −0.38772 × monitoring year gap.

Table 2.

Parameter Estimates of Model Fitting

Parameter	Parameter estimate	Standard error	t	p
Intercept	0.60689	0.56603	1.07	0.2883
Linear (MUI)	0.00302	0.00094388	3.20	0.0022
Linear (MSI)	−0.05172	0.00981	−5.27	<0.0001
Linear (iodized_salt_coverage_rate)	0.03481	0.00662	5.26	<0.0001
Linear (per_capita_income)	0.00000969	0.00000386	2.51	0.0149
Linear (district)	−0.00381	0.00619	−0.62	0.5405
Linear (year_gap)	−0.38772	0.02868	−13.52	<0.0001

MSI, mean salt iodized concentration.

Based on the nonlinear effect of the GAM fitting (Fig. 1), MTvol and MUI in 8- to 10-year-old children had a quadratic curve relationship. When MUI ranged between 150 and 250 μg/L, MTvol was the lowest. The relationship between the per capita disposable income and MTvol in 8- to 10-year-old children was cubic regression spline. When the per capita disposable income was ∼3000–15,000 yuan, MTvol was the highest.

FIG. 1.

The nonlinear diagram of MTvol and the related index by GAM. GAM, generalized additive model; MTvol, median thyroid volume.

Validation of the GAM fitting

According to the established model, the data that were from 16 primary schools in 2015 and 80 primary schools in 2018 were predicted by backward substitution. Forty students from each school were selected for the test. The monitoring contents and methods were the same as the national IDD monitoring program in 2017. The results showed that the average relative error between the predicted and the actual value was 15.2% (Fig. 2), suggesting the model can be used to predict MTvol.

FIG. 2.

The predicted and actual MTvol distribution.

Appropriate range of MUI in 8- to 10-year-old children

According to the relationship between MTvol and goiter rate, when goiter rate was <5%, the maximum MTvol was 4.77 mL. Considering a 15% potential error of GAM, the maximum MTvol was set to 4.157 mL. On the premise of no goiter epidemic, the appropriate range of MUI was predicted. According to the existing monitoring data, household MSI was set to 23 mg/kg, the monitoring gap was 3 years, the iodized salt coverage rate was 80, and the per capita disposable income was 54,300 Chinese Yuan. The findings showed that at 27–277 μg/L MUI, MTvol was not higher than 4.157 mL.

Discussion

The main cause of ID is lack of iodine. The monitoring results of the Shanghai Municipal Center for Disease Control and Prevention in 2002 showed that the median drinking water iodine concentration was 8.0 μg/L in Shanghai (21). This result provided a scientific basis for the use of iodized salt, even though there is no prevalence of ID in Shanghai.

Adequate dietary iodine intake in children is essential for optimal physical and neurological development. According to WHO/UNICEF/ICCIDD standards (9), the monitoring data showed that MUI of Shanghai children between the ages of 8 and 10 years had an above average iodine intake, which reveals a risk of excess iodine in 1997 and 1999. In 2000, the standard of SIC in Shanghai was appropriately lowered, from 20–60 to 20–50 mg/kg. Average household SIC decreased from 43.8 mg/kg in 1999 to 33.0 mg/kg in 2005. Since 2005, MUI of 8- to 10-year-old children ranged from 171.4 to 198.1 μg/L in Shanghai (8).

Even though the average SIC of households decreased by ∼44.5% from 1999 to 2017 and the iodized salt coverage rate decreased from 94.6% in 1997 to 76.5% in 2007, MUI consistently fluctuated at 200 μg/L. The maximum goiter rate was 3.07% in 1997, which is lower than the epidemic standard for IDDs established by the WHO (14). This result provides a false sense of whether USI is necessary; iodized salt has little effect on iodine status of Shanghai residents. In fact, the reduction of household SIC should not cause excessive panic. The inconsistencies between the changes in household SIC and the changes in UIC may be related to the economic level and food consumption habits in Shanghai. Shanghai has always been the economic center of China. Studies have shown that Shanghai residents consume more animal protein especially milk, which improves the utilization of iodine (22,23). In the early 1950s, researchers discovered that casein protects against low iodine goiter and that animal protein is rich in casein (24). A study that evaluated the effect of insufficient protein intake on Tvol showed that inadequate protein intake can exacerbate the prevalence of goiter in the presence of ID (25). In addition, according to a German study, diets with high animal protein are accompanied by higher salt intake, and iodine intake from iodized salt tends to be higher (26).

Even though iodine from household iodized salt was reduced, the iodine intake from restaurant and prepackaged foods cannot be ignored. With the rapid development of social economy, dining out rate and consumption of prepackaged foods are becoming more common among Chinese residents. China implements a nationwide policy that centralizes catering facilities including restaurants and school cafeterias, and most prepackaged products incorporate iodized salt. A dietary nutrition survey conducted in Shanghai from 2012 to 2014 showed that the prevalence of eating out and at restaurants was 55.1% and 31.8% in adults, respectively (27). In Shanghai, almost all primary and secondary school students have lunch at school cafeterias. In the past 20 years, with the gradual development of food sales from modern shopping systems such as chain supermarkets, convenience stores, and online shopping, consumption of prepackaged foods has increased (28).

Based on our findings, iodine status among Shanghai residents is appropriate. Household SIC has little effect on the iodine status of residents. Future studies should analyze dietary sources of iodine, especially from eating out and prepackaged foods, in an attempt to propose accurate prevention and control strategies of ID among Shanghai residents.

The results obtained from GAM showed that at 27–277 μg/L MUI, MTvol was not higher than 4.157 mL and that goiter rate did not exceed 5%, which is the cutoff point of the international IDD prevalence standard. In 1996, USI was carried out to prevent IDDs in Shanghai. In 1995, the first monitoring results in Shanghai showed that goiter rate was 1.57% in 9- to 13-year-old urban students and 3.71% in adults, and MUI was 70 μg/L in 9- to 13-year-old urban students, which is the lowest number in the history of Shanghai. Based on comprehensive analysis of the effect of SIC and iodized salt coverage rate on MTvol by GAM, the results of the first iodine nutrition monitoring in 1995 in Shanghai, and the article by Zimmermann et al. that found that UIC in the range of 200–500 μg/L did not increase Tvol in children (29), we suggest that the appropriate range of MUI in 8- to 10-year-old children living in Shanghai should be 70–277 μg/L. And most participants have UIC in this range, the proportion is 73.4%.

This study integrated the results of iodine nutrition monitoring data in Shanghai from the past 20 years and analyzed the relevant influencing factors. It is the first time that GAM has been applied to predict adequate urinary iodine levels. Compared with previous studies with single factor analysis, this study used GAM, which has an unlimited parameter distribution. There are two main advantages. First, in addition to fitting the relationship between MTvol and MUI, other factors were introduced to better control for confounding factors. Second, GAM can be used to predict MTvol, which is difficult to assess from MUI when other factors are controlled, thereby significantly reducing manpower, material resources, money, and social resources. There are also some limitations to our study. We had no data on thyrotropin and free thyroxine levels that would shed some light on the children's thyroid functions.

Based on our findings, the median UIC of 8- to 10-year-old children in Shanghai should be the one established by WHO/UNICEF/ICCIDD. Household SIC has little effect on the iodine status of residents. GAM can predict the prevalence of goiter in 8- to 10-year-old children by using appropriate indicators as dependent variables, such as MUI, SIC, and per capita disposable income. Under the current economy and household SIC, to avoid the occurrence of goiter epidemic and to maintain an optimal iodine nutrition status, the recommended median range of urinary iodine for 8- to 10-year-old children in Shanghai should be 70–277 μg/L.

Footnotes

Acknowledgments

We are grateful to the children who participated in this study and to the health care professionals from the Centers for Disease Control and Prevention of the 17 districts in Shanghai.

Authors' Contributions

Z.W., X.C., Q.S., W.J., C.G., and S.L. performed the investigations; C.G. and S.L. were involved in project management and coordination; Z.W. and S.L. performed Tvol estimation; Z.W., B.L., J.Z., and Z.S. analyzed the data; and Z.W. and J.Z. wrote the article. All authors read and approved the final version of the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This study was financially supported by the National Nature Science Foundation of China (Grant No. 81602851) and the Excellent Young Talents of Health System in Shanghai (Grant No. 2017YQ043). None of the funders played a role in the study design, data analysis, or article preparation.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

References

ICCIDD/UNICEF/WHO. 2001. Assessment of IDD and Monitoring Their Elimination: A Guide for Programme Managers. Second Edition. WHO, Geneva.

Pearce

, Andersson

, Zimmermann

. 2013. Global iodine nutrition: where do we stand in 2013?. Thyroid, 23:523–528.

Zimmermann

, Aeberli

, Melse-Boonstra

, Grimci

, Bridson

, Chaouki

, Mbhenyane

, Jooste

. 2009. Iodine treatment in children with subclinical hypothyroidism due to chronic iodine deficiency decreases thyrotropin and C-peptide concentrations and improves the lipid profile. Thyroid, 19:1099–1104.

Auso

, Lavado-Autric

, Cuevas

, Del Rey

, Morreale De Escobar G, Berbel P. 2004 A moderate and transient deficiency of maternal thyroid function at the beginning of fetal neocorticogenesis alters neuronal migration. Endocrinology, 145:4037–4047.

Qian

, Yan

, Chen

, Wang

. 2002. Meta-analysis on the relationship between children's intelligence and factors as iodine deficiency, supplement iodine and excessive iodine. Chin J Epidemiol, 23:246–249.

Bougma

, Aboud

, Harding

, Marquis

. 2013. iodine and mental development of children 5 years old and under: a systematic review and meta-analysis. Nutrients, 5:1384–1416.

Wang

, Liu

, Su

, Zou

, Song

, Liu

. 2018. A comparison of iodine status in children and pregnant women after a policy change in the iodized salt standard in Shanghai, China. Biol Trace Elem Res, 185:275–281.

Wang

, Liu

, Su

, Zou

, Song

, Liu

. 2019. A comparative study of iodized salt programs: Shanghai and Switzerland. Biol Trace Elem Res, 187:59–64.

ICCIDD/UNICEF/WHO. 2007. Assessment of Iodine Deficiency Disorders and Monitoring Their Elimination: A Guide for Programme Managers. Third Edition. World Health Organization, Geneva.

10.

Zimmermann

, Aeberli

, Andersson

, Assey

, Yorg

, Jooste

, Jukic

, Kartono

, Kusic

, Pretell

, San Luis

Jr ., Untoro

, Timmer

. 2013. Thyroglobulin is a sensitive measure of both deficient and excess iodine intakes in children and indicates no adverse effects on thyroid function in the UIC range of 100–299 mug/L: a UNICEF/ICCIDD study group report. J Clin Endocrinol Metab, 98:1271–1280.

11.

Wang

, Su

, Liu

. 2017. Dynamic changes in the bivariable distribution of urinary iodine concentration and thyroid volume in children aged 8 to 10 years in China. Asia Pac J Public Health, 29:288–295.

12.

Melse-Boonstra

, Jaiswal

. 2010. Iodine deficiency in pregnancy, infancy and childhood and its consequences for brain development. Best Pract Res Clin Endocrinol Metab, 24:29–38.

13.

Gutekunst

, Smolarek

, Hasenpusch

, Stubbe

, Friedrich

, Wood

, Scriba

. 1986. Goitre epidemiology: thyroid volume, iodine excretion, thyroglobulin and thyrotropin in Germany and Sweden. Acta Endocrinol, 112:494–501.

14.

WHO/UNICEF/ICCIDD. 1994. Indicators for Assessing Iodine Deficiency Disorders and Their Control Program Through Salt Ionization. WHO, Geneva.

15.

Laurberg

, Bulow Pedersen

, Knudsen

, Ovesen

, Andersen

. 2001. Environmental iodine intake affects the type of nonmalignant thyroid disease. Thyroid, 11:457–469.

16.

McLean

, Hooker

, Staicu

, Scheipl

, Ruppert

. 2014. Functional generalized additive models. J Comput Graph Stat, 23:249–269.

17.

Shujie

, Carroll

, Liang

, Xu

. 2015. Estimation and inference in generalized additive coefficient models for nonlinear interactions with high-dimensional covariates. Ann Stat, 43:2102–2131.

18.

Fuse

, Saito

, Tsuchiya

, Shishiba

, Irie

. 2007. Smaller thyroid gland volume with high urinary iodine excretion in Japanese schoolchildren: normative reference values in an iodine-sufficient area and comparison with the WHO/ICCIDD reference. Thyroid, 17:145–155.

19.

WHO/UNICEF/ICCIDD. 1993. Indicators for assessing iodine deficiency disorders and their control programmes. Review Version. WHO, Geneva.

20.

Yan

, Zhang

, Liu

, Liu J

W L

. 2006. Method for Determination of Iodine in Urine by As3+-Ce4+Catalytic Spectrophotometry (WS/T 107–2006). People's Medical Publishing House, Beijing, China.

21.

Xianyi Chen

, Liu

. 2003. Iodine Deficiency Disease Surveillance in China (2002). People's Medical Publishing House, Beijing, China.

22.

Zang

, Yu

, Zhu

, Lu

, Liu

, Yao

, Bai

, Guo

, Jia

, Zou

, Wu F 2017. Does the dietary pattern of shanghai residents change across seasons and area of residence: assessing dietary quality using the Chinese diet balance index (DBI). Nutrients, 9:pii: E251.

23.

Wang

, Zhu

, Zang

, Zou

, Luo

, Jia

, Guo

, Wang

, Sun

, Wu

. 2018. [Dairy product intake and influencing factors in residents of Shanghai]. Zhonghua Liu Xing Bing Xue Za Zhi, 39:886–891.

24.

Axelrad

, Leblond

. 1954. Prevention of low iodine goitre by casein in mice. Can Med Assoc J, 70:78–79; passim.

25.

Brahmbhatt

, Brahmbhatt

, Boyages

. 2001. Impact of protein energy malnutrition on thyroid size in an iodine deficient population of Gujarat (India): is it an aetiological factor for goiter?. Eur J Endocrinol, 145:11–17.

26.

Montenegro-Bethancourt

, Johner

, Stehle

, Remer

. 2015. Dietary ratio of animal:plant protein is associated with 24-h urinary iodine excretion in healthy school children. Br J Nutr, 114:24–33.

27.

Zang

, Luo

, Wang

, Zhu

, Wang

, He

, Wang

, Guo

, Chen

, Wang

, Guo

, Zou

, Jia

, Wu

. 2018. Eating out-of-home in adult residents in shanghai and the nutritional differences among dining places. Nutrients, 10:pii: E951.

28.

, Dunford

. 2013. Complexities and opportunities in monitoring and evaluating US and global changes by the food industry. Obes Rev, 14 Suppl 2:29–41.

29.

Zimmermann

, Ito

, Hess

, Fujieda

, Molinari

. 2005. High thyroid volume in children with excess dietary iodine intakes. Am J Clin Nutr, 81:840–844.