Age- and gender-specific reference intervals of TSH and free T4 in an iodine-replete area: Data from Korean National Health and Nutrition Examination Survey IV (2013–2015)

Background Establishment of the reference interval of thyroid-stimulating hormone (TSH) is critical in the diagnosis of thyroid dysfunction and is affected by age, gender, iodine nutrition, and ethnicity. The aim of this study was to determine the reference intervals of TSH and free thyroxin (FT4) from a large, nationwide data of Korea where iodine intake is more than adequate. Methods We analyzed data from the Korea National Health and Nutrition Examination Survey VI that measured serum TSH, FT4, and thyroid peroxidase antibody from 7,061 individuals (urinary iodine measurement in 6,565). Age- and gender-specific reference intervals were established from 95% confidence limits from the 2.5 to 97.5 percentile of TSH (log-transformed) and FT4 in reference populations. Results The geometric mean of TSH was 2.16 ± 0.01 mIU/L, with the lowest value found in the middle aged group (2.04 ± 0.02 mIU/L) and higher values noted in age groups of 10–19 and over 70 years (2.38 ± 0.02 and 2.32 ± 0.07 mIU/L, respectively). The association of TSH and age was U-shaped. The overall reference interval of TSH was 0.59–7.03 mIU/L. Mean FT4 was 1.25 ± 0.003 ng/dL (16.09 ± 0.039 pmol/L), and it showed a small but continuous decrease after 20 years of age (P < 0.001). There was a significant positive correlation between TSH and urine iodine concentration (r = 0.154, P < 0.001). Conclusions The reference interval of TSH in Korea, where iodine intake is above the requirement, was 0.59–7.03 mIU/L and showed U-shaped change with age, which was a similar pattern to iodine intake. The reference interval of FT4 was 0.92–1.60 ng/dL. The geometric mean and upper limit of TSH were higher than those of Western populations, reflecting the paramount importance of iodine intake on thyroid function.


Study subjects
The KNHANES surveys were designed to obtain national estimates of the health and nutritional status of Koreans from 1998 and are an ongoing surveillance system. KNHANES VI (2013-2015) was conducted using a stratified, multistage, clustered probability sampling design. In this data set, approximately 2,400 individuals (about one-third of the total sample) were selected each year between 2013 and 2015 using stratified subsampling and underwent measurements for serum TSH, FT4, and TPOAb. This subsample of KNHANES VI consisted of 7,061 individuals aged 10 years and older, weighted to represent the total Korean population. In this sample, urinary iodine (UI) in a spot urine sample was measured in 6,564 individuals. The participants responded to a questionnaire regarding family history of thyroid disease, personal history of thyroid disease, and drugs that could affect thyroid hormone level.
To define precise reference intervals of thyroid hormones, disease-free and reference populations were established. The disease-free population was selected from the total population after exclusion of individuals with known thyroid disease, family history of thyroid dysfunction, and current pregnancy. The reference population was selected from the disease-free population after exclusion of those with positive TPOAb, defined as ! 34.0 IU/ml, as provided by the kit manufacturer.
The study protocol was approved by the Institutional Review Board of the Korea Centers for Disease Control and Prevention. Written informed consent was obtained from all participants or parents/guardians.

Data collection and laboratory methods
Data on demographic characteristics, personal medical history, and family history of thyroid diseases were collected by interview during the survey. Life style risk factors (i.e., smoking status) were based on self-reporting. Serum TSH, FT4, and TPOAb concentrations were measured by an electrochemiluminescence immunoassay. Serum TSH was measured using an E-TSH kit (Roche Diagnostics, Mannheim, Germany). Serum FT4 was measured by E-Free T4 kit (Roche Diagnostics, Mannheim, Germany). Serum TPOAb was measured by E-Anti-TPO kit (Roche Diagnostics, Mannheim, Germany). The results of TSH, FT4, and TPOAb met the specifications regarding accuracy, general chemistry, special immunology, and ligand of the quality control and quality assurance program of the College of American Pathologists. UI was measured using an inductively coupled plasma mass spectrometry device (ICP-MS; Perkin Elmer ICP-MS, Waltham, MA, USA). The laboratory that measured UI is enrolled in the "Ensuring the Quality of Urinary Iodine Procedures (EQUIP)" quality assurance program run by the Centers for Disease Control of the United States of America [17].

Statistical analyses
We used SAS version 9.4 (SAS Institute, Cary, NC) to analyze complex sample survey data using the weights assigned to the individuals sampled in order to represent the nationwide population. The geometric mean ± standard error (SE) for TSH (log-transformed value), arithmetic mean ± SE for FT4, and median (interquartile range [IQR]) for UI were calculated as descriptive statistics using Proc Surveymeans. The 95% confidence limits from the 2.5 percentile to 97.5 percentile of TSH (log-transformed) and FT4 were obtained to establish reference intervals. Regression analysis using Proc Surveyreg was used to compare the differences in TSH and FT4 between the age groups and genders. Correlation coefficients using Proc Corr were adjusted for age, gender, BMI, and smoking status and were calculated to estimate the relationship between TSH and UI concentration. A p-value < 0.05 was considered significant.

Baseline characteristics of the study subjects
Among 7,061 individuals who underwent thyroid function evaluation as a part of KNHANES VI, we excluded 705 individuals who had a prior history of thyroid disease (n = 179, 2.2%) or thyroid cancer (n = 44, 0.5%), family history (n = 346, 4.8%), or were pregnant (n = 22, 0.3%) to establish a disease-free population. Among the disease-free population of 6,356 individuals, we excluded individuals who had a positive TPO Ab result (n = 370, 5.7%). Finally, the reference population consisted of 5,987 individuals. The mean age of the reference population was 37.0 ± 0.2 years, and the proportion of males was 51.8%. The detailed characteristics of each population group are shown in Table 1.

TSH by age and gender
The geometric mean ± SE of TSH in the total population was 2.14 ± 0.01 mIU/L (Table 2), with the lowest value found in the age group of 40-49 years (2.04 ± 0.03 mIU/L) and higher values noted in age groups of 10-19 and over 70 years (2.33 ± 0.02 and 2.37 ± 0.07 mIU/L, respectively). In the reference population, the geometric mean TSH was 2.16 ± 0.01 mIU/L. The geometric mean of TSH in the reference population was lower in the middle aged group (2.04 ± 0.02 mIU/L) and higher in age groups of 10-19 and over 70 years (2.38 ± 0.02 and 2.32 ± 0.07 mIU/L, respectively). Thus, a U-shaped association was observed between age and TSH concentration.
The geometric mean of TSH was significantly higher in females (2.24 ± 0.02 mIU/L) than males (2.09 ± 0.01 mIU/L) in the reference population (P < 0.001), but showed marginal

Age-and gender-specific reference intervals of TSH
The overall reference interval of TSH in the reference population was 0.59-7.03 mIU/L and was wider in females (0.56-7.43 mIU/L) than males (0.62-6.57 mIU/L) ( Table 2). The reference interval of TSH had no significant association with age (P for trend = 0.09). The upper reference limit of TSH was the lowest in those aged 20-29 years (6.04 mIU/L) and highest in those aged 50-59 years (8.11 mIU/L). In males, it was lowest in the age group of 40-49 years (5.96 mIU/L) and highest in the group 60-69 years (7.35 mIU/L); in females, the lowest upper reference TSH level was found in 20-29-year-olds (5.97 mIU/L) and the highest was seen in 50-59-year-olds (9.19 mIU/L). The lower reference limit of TSH was the lowest in the age group over 70 years (0.42 mIU/L) and highest in the group 20-29 years (0.67 mIU/L).

Age-and gender-specific reference intervals of FT4
In the reference population, mean FT4 was 1.25 ± 0.003 ng/dL (16.09 ± 0.039 pmol/L) (  (Table 4). After adjusting for age, gender, BMI, and smoking status, there was a significant positive correlation between TSH and UI concentration (r = 0.154, P < 0.001). We also analyzed the association between TSH and UI by dividing the reference population into 3 groups according to age: younger than 30 years, between 30 and 50 years, and older than 50 years. The correlation coefficient (r) between TSH and UI in the groups was 0.185 (p = 0.008), 0.150 (P = 0.031), and 0.182 (P < 0.001), respectively. Accordingly, the curve of UI (median) by age showed very similar U-shaped appearance to that of TSH by age, suggesting iodine's effect on TSH in this population (Fig 2).

Discussion
This study presented the age-and gender-specific reference intervals of TSH and FT4 from the large, nationwide survey dataset of KNHANES VI (2013-2015). The results demonstrated that TSH has a U-shaped association with age and is higher in females. In addition, there was a strong association between UI by age and TSH by age. In contrast, FT4 decreased continuously after 20 years of age and was higher in males than in females. To our knowledge, this is the first study to establish reference intervals of TSH and FT4 from a large, nationwide study in  Korea, where iodine intake is reported to be above-requirement or excessive according to WHO epidemiological criteria. Previous studies from Korea based on two regional population-based cohorts reported similar results to our report [18,19]. In one study from an Ansung cohort (n = 3,399, adults between 40 and 70 years old), mean FT4 was 0.99 ± 0.14 ng/dL, and mean TSH was 2.53 ± 3.31 mIU/L. In the other study from the Korean Longitudinal Study on Health and Aging (KLoSHA, n = 940), which consisted of elderly subjects over 65 years old, mean FT4 was 1.22 ± 0.38 ng/dL, and mean TSH was 3.49 ± 7.12 mIU/L. Only in males of the Ansung cohort, TSH increased and FT4 decreased with age; age was not associated with TSH and FT4 in the other subjects. The other study from Korea, based on routine health examination data, also reported a higher reference interval of TSH from 0.73 to 7.06 mIU/L [20]. However, the data were all regional or institutional rather than nationwide standardized data like our study. There are some distinctive differences between our report and other published reports. First, TSH in this study was apparently higher than that of other reports from Western populations. The geometric mean and overall upper reference limit of TSH were 2.16 and 7.03 mIU/ L, respectively, in our study, whereas they were 1.40 and 4.12 mIU/L in the National Health and Nutrition Examination Survey (NHANES III) from the United States of America [10]. One possible explanation for these differences is related to the different iodine intake status of Korea, which was classified to be above-requirement or excessive in previous reports and also in this study. In other regions with sufficient iodine supply such as North America and East Asia [21,22], concentrations and reference limits of TSH tend to be higher than those in regions with iodine deficiency, such as Europe [23,24]. Similarly, in NHANES III, higher UI/ Cr excretion was significantly related to higher TSH concentration [10]. These findings are consistent with some reports showing that distribution of TSH shifted toward the right after improved iodine supply [25,26]. However, this reason is not sufficient to explain these apparent differences, because our results were much higher than those in similar iodine intake regions. Even in an iodine-excessive area in China where the median UI of schoolchildren was 650.9 ug/L, the geometric mean and reference range of TSH were 1.90 and 0.59-5.98 mIU/L, respectively [27]. Thus, as recently suggested, hereditary and genetic influences on the setpoint of thyroid hormone might be additional reasons for the high upper limits of this study. A few studies have suggested that the negative feedback set points for FT4 and TSH might be genetically determined [28]. Other reports suggested that polymorphisms in thyroid hormone pathway genes might be related with increased TSH concentration [29,30]. Further studies will be required to validate a genetic basis of thyroid hormone regulation. Second, we did not show an increase of TSH with age but a U-shaped association. This Ushaped trend of TSH with age, with values lower in the middle age group and higher younger and older ages, was very similarly seen in UI concentration in this study. There was a significant correlation between UI by age and TSH by age. Therefore, we speculate that differences of iodine intake with age group affected TSH, as previously stated, and could offset the increase in TSH according to age reported in other reports [10,12,31]. However, as UI by age and TSH by age do not show a correlation in those aged 30-39 years, the change of TSH according to UI should be interpreted with caution. Further studies will be needed to more accurately clarify the association between TSH and iodine status.
Finally, continuous decrease of FT4 with age was observed after 20 years of age. This result is in accord with other studies about FT4, showing either no change or a minimal decrease in FT4 with age [10], although some reports suggested that the differences were too small to have clinical significance [31]. Recently, there were some reports showing that increased TSH and decreased free thyroid hormone are associated with prolonged life span in some populations [12,32]. These reports support the suggestion that age-specific reference ranges of TSH should be set in order to avoid over-treatment.
Most clinicians would agree to the need to establish reference intervals of thyroid hormones with consideration of age, gender, ethnicity, and iodine intake. However, they still use the values provided by the guidelines or commercial kits manufacturers because of the lack of https://doi.org/10.1371/journal.pone.0190738.g002 regional or population data. Therefore, we hope that the reference intervals of TSH and FT4 provided in this study will be used as reference intervals in clinical practice in Korea and other areas where iodine intake is similar. Our findings show that the widely used upper limit of TSH of 4.5 or 5.0 mIU/L might be too low to be applicable where iodine intake is aboverequirement and thus could lead to overestimates of subclinical hypothyroidism.
The limitation of this study is that the presence of goiter was not obtained during the survey by ultrasound or physical examinations but only by questionnaire. However, the similarity of the reference interval of this study to a previous report from Korea where the presence of goiter was checked by ultrasound support the solidarity of our study data [20]. In the previous study (n = 5,778), TSH RI obtained using ultrasound findings was 0.73-7.06 mIU/L, which is almost identical to our result. Thyroglobulin Ab (TgAb) also could not be obtained in this survey. However, it is known that the presence of TgAb alone does not correlate with abnormal TSH level [33]. Especially in iodide sufficient areas, it is not usually necessary or cost-effective to order both TPOAb and TgAb, because TPOAb-negative patients with detectable TgAb rarely display thyroid dysfunction [9]. Therefore, we cautiously speculate that the presence or absence of goiter or TgAb would not lead to a significant impact on our overall results. In addition, the KNHANES is cross-sectional and therefore does not show individual changes over time according to iodine intake and so cannot prove the relationships between iodine intake and TSH and/or FT4. Finally, TSH and FT4 were measured by one type of kit, even though the results met the specifications of the quality assurance program of the College of American Pathologists. Further studies are needed to evaluate the compatibility of this study's results to results measured by different kits.
In summary, we present the reference intervals of TSH and FT4 from KNHANES VI data in Korea, where iodine intake is above-requirement. The reference interval of TSH was 0.59-7.03 mIU/L and showed U-shaped change with age, similar to the pattern of iodine intake. The reference interval of FT4 was 0.92-1.60 ng/dL and showed continuous decrease with age after 20 years. The geometric mean and upper limit of TSH were higher than those of Western populations, reflecting the paramount importance of iodine intake on thyroid function [34].