Clin Chem Lab Med 2014; 52(12): 1807–1813
Ville L. Langén*, Teemu J. Niiranen, Juhani Mäki, Jouko Sundvall and Antti M. Jula
Thyroid-stimulating hormone reference range and factors affecting it in a nationwide random sample Abstract Background: Previous studies with mainly selected populations have proposed contradicting reference ranges for thyroid-stimulating hormone (TSH) and have disagreed on how screening, age and gender affect them. This study aimed to determine a TSH reference range on the Abbott Architect ci8200 integrated system in a large, nationwide, stratified random sample. To our knowledge this is the only study apart from the NHANES III that has addressed this issue in a similar nationwide setting. The effects of age, gender, thyroid peroxidase antibody (TPOAb)-positivity and medications on TSH reference range were also assessed. Methods: TSH was measured from 6247 participants randomly drawn from the population register to represent the Finnish adult population. TSH reference ranges were established of a thyroid-healthy population and its subpopulations with increasing and cumulative rigour of screening: screening for overt thyroid disease (thyroid-healthy population, n = 5709); screening for TPOAbpositivity (risk factor-free subpopulation, n = 4586); and screening for use of any medications (reference subpopulation, n = 1849). Results: The TSH reference ranges of the thyroid-healthy population, and the risk factor-free and reference subpopulations were 0.4–4.4, 0.4–3.7 and 0.4–3.4 mU/L (2.5th–97.5th percentiles), respectively. Although the differences in TSH between subgroups for age (p = 0.002) and gender (p = 0.005) reached statistical significance, the TSH distribution curves of the subgroups were practically superimposed. Conclusions: We propose 0.4–3.4 mU/L as a TSH reference range for adults for this platform, which is lower than *Corresponding author: Ville L. Langén, MD, Population Studies Unit, National Institute for Health and Welfare, Peltolantie 3, 20720 Turku, Finland, Phone: +358 29 524 6000, E-mail: [email protected] and Operational Division of Medicine, Turku University Hospital, Turku, Finland Teemu J. Niiranen, Juhani Mäki and Antti M. Jula: Population Studies Unit, National Institute for Health and Welfare, Turku, Finland Jouko Sundvall: Disease Risk Unit, National Institute for Health and Welfare, Helsinki, Finland
those presently used in most laboratories. Our findings suggest that intensive screening for thyroid risk factors, especially for TPOAb-positivity, decreases the TSH upper reference limit. Keywords: adult population; reference interval; reference range; thyroid dysfunction; thyroid peroxidase antibody (TPOAb); thyroid-stimulating hormone (TSH). DOI 10.1515/cclm-2014-0287 Received March 14, 2014; accepted May 19, 2014; previously published online June 20, 2014
Introduction The biochemical definition of thyroid dysfunctions relies on the reference ranges of thyroid-stimulating hormone (TSH) and thyroid hormones – thyroxine and triiodothyronine. The definition of the TSH reference range is therefore of paramount importance especially for diagnosing subclinical thyroid disorders. In 2002 the National Academy of Clinical Biochemistry (NACB) published a comprehensive guideline for diagnosing and monitoring of thyroid disease [1]. The NACB guideline instructed that TSH reference ranges should be established of rigorously screened euthyroid volunteers and held it likely that by this approach the upper limit of TSH reference range would be over time reduced from the current 4.0–4.5 mU/L to 2.5 mU/L. NACB considered that the current TSH reference range has been based on populations containing individuals with occult mild hypothyroidism that even the sensitive thyroid antibody immunoassays cannot always detect and that the failure to exclude these volunteers has resulted in the skewed upper limit of TSH distribution. In contrast, Surks et al. have suggested [2, 3] that this skew could be reduced by dividing the TSH distribution to age-specific curves as TSH has been shown to increase with age in several [3–7], though not in all studies [8–10]. In addition to age-specificity, it has also been suggested that TSH reference limits should be race[2] and assay-specific [11].
Brought to you by | Simon Fraser University Authenticated Download Date | 6/5/15 4:22 PM
1808 Langén et al.: TSH reference range and factors affecting it In a 2005 German study that adhered very strictly to the NACB criteria – and even used ultrasonographic assessment to exclude occult thyroid disease – the upper limit of TSH reference range was comparable between the whole population and the rigorously screened population (3.63 mU/L vs. 3.77 mU/L) [12]. Another German study, conducted in a region of mild iodine deficiency, provided the same conclusion [13]. This has raised the question of whether or not a stringent screening is needed to select a reference population for establishing a TSH reference range [14]. However, some studies have shown clear differences between the upper limits of the unscreened and screened populations, especially when participants with thyroid peroxidase antibodies (TPOAb) are excluded [4, 8]. The debate continues on how and where to set the upper limit of the TSH reference range. Previous studies with mainly selected populations have proposed contradicting reference ranges and have disagreed on how population screening, age and gender affect them. Our study aimed to determine a TSH reference range in a large, nationwide, stratified random adult sample on the Abbott Architect ci8200 integrated system. We also assessed the effect of age, gender, TPOAb-positivity and medications on TSH reference range. To our knowledge, apart from the NHANES III [6], this is the only study that has addressed these issues in a similar setting.
Laboratory analyses Plasma TSH was measured from all 6247 blood samples. Additionally, plasma TPOAb was measured from all participants with even slightly abnormal TSH values (TSH 2.5 mU/L). TPOAb was also measured from 480 (9.2%) randomly selected participants with TSH between 0.4 and 2.5 mU/L (n = 5221). All samples were stored in –70 °C and later analysed with an Abbot Architect ci8200 Analyzer (Abbott Laboratories, IL, USA). During the course of measurements the between-batch coefficient of variation in control samples was 12.1 mU/L) or with one or more exclusion factors, the thyroid-healthy population consisted of 5709 participants.
Risk factor-free subpopulation (subset of the thyroidhealthy population) In accordance with the NACB guideline, we excluded participants who had a positive TPOAb result (n = 364) from the thyroid-healthy population. As TPOAb values were not available for all participants who had a TSH value between 0.4 and 2.5 mU/L, we excluded 759 randomly selected participants with a TSH value between 0.4 and 2.5 mU/L to prevent oversampling. Thus, the total share of excluded participants within the TSH range 0.4–2.5 mU/L was exactly equal to the true prevalence of TPOAb-positivity in it (16.3%), determined from the available data. Finally, this subpopulation consisted of 4586 participants. We also formed a subgroup of this subpopulation from which we excluded participants with use of medications that have a potential effect on thyroid function tests. These medications have been described in a previous review [18] and are listed online (see Supplemental Data, Table 1, which accompanies the article at http://www. degruyter.com/view/j/cclm.2014.52.issue-12/issue-files/cclm.2014.52. issue-12.xml). This subgroup consisted of 3453 participants.
Reference subpopulation (subset of the risk factor-free subpopulation) We excluded participants with any medications (n = 2737) from the risk factor-free subpopulation. As opposed to the NACB guidelines, we excluded even users of oestrogen, as some oestrogen medications seem to affect TSH levels [19]. The final reference subpopulation consisted of 1849 participants.
Ethics The Health 2000 Survey protocol was approved by the Epidemiology Ethics Committee of the Helsinki and Uusimaa hospital region and all the participants signed informed consent according to the Declaration of Helsinki.
Statistical analyses The TSH distribution was neither normal nor log-normal in the reference subpopulation. We therefore established TSH reference ranges directly from the 2.5th and 97.5th percentiles of the TSH measurements in the thyroid-healthy population and all of its subsets, as has been done in several other notable studies [6, 8, 12, 20]. We controlled the validity of this non-parametric method by defining the TSH reference range for the reference subpopulation also from data on the 95% confidence limits of its TSH values that were transformed by using a best suitable function to obtain a Gaussian distribution. Differences in TSH between the subgroups for age and gender were compared using the Kruskal-Wallis test. p
Ville L. Langén*, Teemu J. Niiranen, Juhani Mäki, Jouko Sundvall and Antti M. Jula
Thyroid-stimulating hormone reference range and factors affecting it in a nationwide random sample Abstract Background: Previous studies with mainly selected populations have proposed contradicting reference ranges for thyroid-stimulating hormone (TSH) and have disagreed on how screening, age and gender affect them. This study aimed to determine a TSH reference range on the Abbott Architect ci8200 integrated system in a large, nationwide, stratified random sample. To our knowledge this is the only study apart from the NHANES III that has addressed this issue in a similar nationwide setting. The effects of age, gender, thyroid peroxidase antibody (TPOAb)-positivity and medications on TSH reference range were also assessed. Methods: TSH was measured from 6247 participants randomly drawn from the population register to represent the Finnish adult population. TSH reference ranges were established of a thyroid-healthy population and its subpopulations with increasing and cumulative rigour of screening: screening for overt thyroid disease (thyroid-healthy population, n = 5709); screening for TPOAbpositivity (risk factor-free subpopulation, n = 4586); and screening for use of any medications (reference subpopulation, n = 1849). Results: The TSH reference ranges of the thyroid-healthy population, and the risk factor-free and reference subpopulations were 0.4–4.4, 0.4–3.7 and 0.4–3.4 mU/L (2.5th–97.5th percentiles), respectively. Although the differences in TSH between subgroups for age (p = 0.002) and gender (p = 0.005) reached statistical significance, the TSH distribution curves of the subgroups were practically superimposed. Conclusions: We propose 0.4–3.4 mU/L as a TSH reference range for adults for this platform, which is lower than *Corresponding author: Ville L. Langén, MD, Population Studies Unit, National Institute for Health and Welfare, Peltolantie 3, 20720 Turku, Finland, Phone: +358 29 524 6000, E-mail: [email protected] and Operational Division of Medicine, Turku University Hospital, Turku, Finland Teemu J. Niiranen, Juhani Mäki and Antti M. Jula: Population Studies Unit, National Institute for Health and Welfare, Turku, Finland Jouko Sundvall: Disease Risk Unit, National Institute for Health and Welfare, Helsinki, Finland
those presently used in most laboratories. Our findings suggest that intensive screening for thyroid risk factors, especially for TPOAb-positivity, decreases the TSH upper reference limit. Keywords: adult population; reference interval; reference range; thyroid dysfunction; thyroid peroxidase antibody (TPOAb); thyroid-stimulating hormone (TSH). DOI 10.1515/cclm-2014-0287 Received March 14, 2014; accepted May 19, 2014; previously published online June 20, 2014
Introduction The biochemical definition of thyroid dysfunctions relies on the reference ranges of thyroid-stimulating hormone (TSH) and thyroid hormones – thyroxine and triiodothyronine. The definition of the TSH reference range is therefore of paramount importance especially for diagnosing subclinical thyroid disorders. In 2002 the National Academy of Clinical Biochemistry (NACB) published a comprehensive guideline for diagnosing and monitoring of thyroid disease [1]. The NACB guideline instructed that TSH reference ranges should be established of rigorously screened euthyroid volunteers and held it likely that by this approach the upper limit of TSH reference range would be over time reduced from the current 4.0–4.5 mU/L to 2.5 mU/L. NACB considered that the current TSH reference range has been based on populations containing individuals with occult mild hypothyroidism that even the sensitive thyroid antibody immunoassays cannot always detect and that the failure to exclude these volunteers has resulted in the skewed upper limit of TSH distribution. In contrast, Surks et al. have suggested [2, 3] that this skew could be reduced by dividing the TSH distribution to age-specific curves as TSH has been shown to increase with age in several [3–7], though not in all studies [8–10]. In addition to age-specificity, it has also been suggested that TSH reference limits should be race[2] and assay-specific [11].
Brought to you by | Simon Fraser University Authenticated Download Date | 6/5/15 4:22 PM
1808 Langén et al.: TSH reference range and factors affecting it In a 2005 German study that adhered very strictly to the NACB criteria – and even used ultrasonographic assessment to exclude occult thyroid disease – the upper limit of TSH reference range was comparable between the whole population and the rigorously screened population (3.63 mU/L vs. 3.77 mU/L) [12]. Another German study, conducted in a region of mild iodine deficiency, provided the same conclusion [13]. This has raised the question of whether or not a stringent screening is needed to select a reference population for establishing a TSH reference range [14]. However, some studies have shown clear differences between the upper limits of the unscreened and screened populations, especially when participants with thyroid peroxidase antibodies (TPOAb) are excluded [4, 8]. The debate continues on how and where to set the upper limit of the TSH reference range. Previous studies with mainly selected populations have proposed contradicting reference ranges and have disagreed on how population screening, age and gender affect them. Our study aimed to determine a TSH reference range in a large, nationwide, stratified random adult sample on the Abbott Architect ci8200 integrated system. We also assessed the effect of age, gender, TPOAb-positivity and medications on TSH reference range. To our knowledge, apart from the NHANES III [6], this is the only study that has addressed these issues in a similar setting.
Laboratory analyses Plasma TSH was measured from all 6247 blood samples. Additionally, plasma TPOAb was measured from all participants with even slightly abnormal TSH values (TSH 2.5 mU/L). TPOAb was also measured from 480 (9.2%) randomly selected participants with TSH between 0.4 and 2.5 mU/L (n = 5221). All samples were stored in –70 °C and later analysed with an Abbot Architect ci8200 Analyzer (Abbott Laboratories, IL, USA). During the course of measurements the between-batch coefficient of variation in control samples was 12.1 mU/L) or with one or more exclusion factors, the thyroid-healthy population consisted of 5709 participants.
Risk factor-free subpopulation (subset of the thyroidhealthy population) In accordance with the NACB guideline, we excluded participants who had a positive TPOAb result (n = 364) from the thyroid-healthy population. As TPOAb values were not available for all participants who had a TSH value between 0.4 and 2.5 mU/L, we excluded 759 randomly selected participants with a TSH value between 0.4 and 2.5 mU/L to prevent oversampling. Thus, the total share of excluded participants within the TSH range 0.4–2.5 mU/L was exactly equal to the true prevalence of TPOAb-positivity in it (16.3%), determined from the available data. Finally, this subpopulation consisted of 4586 participants. We also formed a subgroup of this subpopulation from which we excluded participants with use of medications that have a potential effect on thyroid function tests. These medications have been described in a previous review [18] and are listed online (see Supplemental Data, Table 1, which accompanies the article at http://www. degruyter.com/view/j/cclm.2014.52.issue-12/issue-files/cclm.2014.52. issue-12.xml). This subgroup consisted of 3453 participants.
Reference subpopulation (subset of the risk factor-free subpopulation) We excluded participants with any medications (n = 2737) from the risk factor-free subpopulation. As opposed to the NACB guidelines, we excluded even users of oestrogen, as some oestrogen medications seem to affect TSH levels [19]. The final reference subpopulation consisted of 1849 participants.
Ethics The Health 2000 Survey protocol was approved by the Epidemiology Ethics Committee of the Helsinki and Uusimaa hospital region and all the participants signed informed consent according to the Declaration of Helsinki.
Statistical analyses The TSH distribution was neither normal nor log-normal in the reference subpopulation. We therefore established TSH reference ranges directly from the 2.5th and 97.5th percentiles of the TSH measurements in the thyroid-healthy population and all of its subsets, as has been done in several other notable studies [6, 8, 12, 20]. We controlled the validity of this non-parametric method by defining the TSH reference range for the reference subpopulation also from data on the 95% confidence limits of its TSH values that were transformed by using a best suitable function to obtain a Gaussian distribution. Differences in TSH between the subgroups for age and gender were compared using the Kruskal-Wallis test. p
