| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1
Departments of Biostatistics and Medicine and
2
Division of Endocrinology, Department of Medicine, Children's Hospital, Harvard Medical School, Boston, MA 02115;
a address correspondence to this author at: Department of Biostatistics, Children's Hospital, 300 Longwood Ave., Boston, MA 02115
Thyroid function tests provide information about hormone metabolism and thyroid dysfunction. Reference intervals enable clinicians to evaluate thyroid function. Several pediatric reference intervals for thyroid function tests have been published(1)(2)(3)(4). Laboratory tests and their nomenclature have been published (5), and the American Thyroid Association has classified thyrotropin (TSH) as the best single measurement of thyroid status because of its high sensitivity (6). However, reference intervals that are derived from small numbers of patients are not reliable for accurately evaluating test results that are dependent on covariates such as age and sex. The IFCC has recommended a minimum of 120 subjects for nonparametric methods in which subgrouping of data is performed (7). Virtanen et al. (8) have proposed that a smaller sample size may be sufficient for regression-based reference intervals.
Recently, there has been much interest in using hospital databases to extract large volumes of patient data for clinical research(9)(10)(11)(12). Hospital databases provide a sufficient number of subjects for evaluating age and sex differences and for establishing age- and sex-based reference intervals. Test results can be affected by medications or treatment received by patients for thyroid disease that alter the physiologic features of the thyroid hormone concentrations during the neonatal period (13)(14). Therefore, neonates and patients who have thyroid disease or demonstrate abnormal test results should be excluded from the analysis used to establish the reference intervals.
Here we report health-related reference intervals for serum thyroxine (T4), triiodothyronine (T3), TSH, and free T4 to be used as clinical guidelines for screening patients with suspected thyroid dysfunction. These pediatric norms are more accurate than those previously published because they are age- and sex-specific and were derived from a large population of children and adolescents.
Between January 1993 and August 1996, test results for T4, T3, TSH, and free T4 were obtained retrospectively from outpatient records at Children's Hospital in Boston for patients 1 month through 20 years of age. Anonymity of patient data was maintained. Medical record number, date of birth, gender, test code, date of test, test result, and International Classification of Diseases (ICD-9) codes were downloaded from the Oracle database on the hospital mainframe computer so that queries could be performed using Paradox (Ver. 5.0; Borland International). ICD-9 codes were used to select patients for the reference group. Excluded diagnoses were hypothyroidism (congenital hypothyroidism; postsurgical and postablative hypothyroidism; iodine hypothyroidism; and iatrogenic, acquired, or unspecified hypothyroidism), hyperthyroidism (toxic diffuse goiter; toxic nodular, uninodular, and multinodular goiter; thyrotoxicosis from ectopic thyroid nodule or other specified origin; thyrotoxicosis with or without storm; and neonatal thyrotoxicosis), and pituitary disorders (acromegaly and gigantism, other anterior pituitary hyperfunction, panhypopituitarism, pituitary dwarfism, diabetes insipidus, disorders of neurohypophysis, iatrogenic pituitary disorders, syndromes of diencephalohypophyseal origin, and dyspituitarism). Infants less than 1 month of age and premature infants were also excluded from analysis. Results from patients who had multiple measurements on the same thyroid function test (<8% for each analyte) were averaged to obtain a single value for each patient in the study. The final sample size consisted of 13 145 test results on 5817 patients (3221 females and 2596 males) after outliers were removed. The total number of patients for each thyroid function test was as follows: T4 (n = 4551), T3 (n = 2683), TSH (n = 5558), and free T4 (n = 353).
Measurements of T4, T3, and TSH were obtained with the DELFIA immunofluorometric system according to the manufacturer's instructions (Wallac Oy). The assay is a solid-phase fluoroimmunoassay that provides a quantitative determination of hormone concentrations in human serum (15). The interassay CV in the euthyroid ranges was <10% for T3 and TSH (16) and 11% for T4 and free T4 (17).
A variety of statistical techniques have been proposed to obtain reference limits(7)(8)(11)(18). Some authors have recommended a nonparametric approach for removing possible abnormal laboratory test results by discarding the top and bottom 10% of the results before calculating 2.5% and 97.5% limits on the remaining 80% of the data (19). However, this method can distort the shape of the underlying distribution by cutting off the tails. If outliers that should be removed are not removed, the resulting intervals are too wide and can lead to failure to detect patients with thyroid dysfunction. In this study, we applied a parametric approach to preserve the shape of the distribution. For all four analytes, least-squares regression was performed using age as the predictor. Outliers were detected by the Tukey interquartile range (IQR) procedure (20) and removed from analysis. Briefly, the IQR is calculated as the difference between the 75th and 25th percentiles. Lower and upper limits are calculated as the 25th percentile - (1.5 x IQR) and the 75th percentile + (1.5 x IQR), respectively. This procedure led to the removal of 320 outliers (2.4%) before reference intervals were determined. The Kolmogorov–Smirnov test (21) was used to assess normality of the data for each thyroid test, and scatter plots were inspected to evaluate the degree of skewness. Because of the extreme right-tailed skew of the TSH and free T4 distributions, natural log transformations were applied before analysis (22). Analysis of covariance was used to compare slopes and y-intercepts between females and males. Because of significant gender differences, separate regression equations were calculated for females and males. Residual diagnostics included visual checks of normal probability plots and Kolmogorov–Smirnov tests.
Linear equations were used to predict mean values, using the midpoint for each of five age groups: 1–11 months, 1–5 years, 6–10 years, 11–15 years, and 16–20 years. Reference intervals were defined as the 95% prediction intervals around the estimated value corresponding to each midpoint. Intervals for TSH and free T4 were converted back to the original units by calculating the antilog of the logarithmically transformed values. Although there is a theoretical distinction between prediction and reference intervals, we used prediction intervals derived from least-squares regression. This method takes into account both the residual standard error as well as the uncertainty of the slope estimate of the regression line. For large sample sizes, using prediction intervals as reference intervals is equivalent to the method recommended by Virtanen et al.(23). In our study, we have results on 5817 children and adolescents. Two-tailed values of P <0.05 were considered statistically significant. SAS, Ver. 6.12 (SAS Institute), and SPSS, Ver. 8.0 (SPSS), were used for statistical analysis.
Table 1
presents the pediatric reference intervals, the mean thyroid
hormone concentrations, and the number of children analyzed for each
age group and sex. Similar numbers of females and males were analyzed
at all ages except the 16–20 year category, in which females
outnumbered males. The Kolmogorov–Smirnov test revealed no significant
departures from a gaussian distribution for T4 or
T3. Serum T4 concentrations decreased significantly
with chronological age in both females and males (P
<0.0001). There was no significant difference in slope between the
sexes for T4 (P = 0.18), indicating that
females and males shared a common rate of decline. Euthyroid mean
values for T4 were estimated to be higher for females than
males in the same age group (P <0.0001). Earlier studies(1)(24) found a similar inverse relationship
between T4 and age; however, no sex differences were reported.
|
In this investigation, euthyroid mean values for T3, TSH, and
free T4 were also inversely correlated with age (P
<0.0001). These results are consistent with those of other
investigators who have reported age-related changes for T3(25) and TSH (26). In addition, we found
significant slope differences between females and males for
T3 (P = 0.022) and TSH
(P = 0.003). Although T3 and TSH
decreased with age for both sexes, this decline was significantly
faster for females. Scatter plots showing the empirical data for
T4, T3, and TSH with
superimposed upper and lower reference limits are provided in Fig. 1
. No gender differences were detected for free
T4 (P = 0.57), and both sexes
demonstrated similar rates of decline with age (P =
0.88). Our reference intervals for free T4 are
broader than those reported by others (4).
|
Acquisition of clinical data to establish pediatric reference intervals for thyroid function testing is challenging because of the difficulty in obtaining blood samples from a large number of healthy children. Therefore, the number of tests used to define pediatric norms is limited. Hospital databases contain large stores of clinical data that can be used to establish reference intervals if the appropriate selection criteria are applied(9)(10)(26).
The large number of patients in this study permitted evaluation of age and sex as possible factors influencing the concentrations of T4, T3, TSH, and free T4. In our viewpoint, the determination of age and sex effects is essential for establishing reliable age- and sex-based pediatric reference intervals. Significant age-related declines were found for all three thyroid function tests, suggesting that during childhood, the requirements for these thyroid hormones decrease. Among postpubertal adolescents, males exhibited lower serum T4 concentrations and higher T3 concentrations compared with their female counterparts. Differences in T4 are consistent with stimulatory and suppressive effects of estrogen and androgen on thyroid-binding globulin serum concentration (26).
Reference intervals for thyroid function tests depend somewhat on the particular laboratory methodology used. All assays vary in sensitivity(27). Our results are in fairly close agreement with other published data. For example, some authors (2) provide a T4 reference interval of 58–161 nmol/L (4.5–12.5 µg/dL), encompassing all children. Others (3) propose an interval of 100–212 nmol/L (7.8–16.5 µg/dL) for children 4–12 months of age, and 82–171 nmol/L (6.4–13.3 µg/dL) for children 5–10 years of age. Their reference intervals for TSH are 0.4–4.2 mIU/L, which correspond closely to our intervals for older children. Becker (2) gives an upper limit for TSH of 5.0 mIU/L, and others (4)(11) report upper limits of 7.1 mIU/L for males and 8.1 mIU/L for females 1 month to 5 years of age and 6.0 mIU/L for both sexes 6–18 years of age. With respect to T3, our reference intervals show a progressive decline with age. Recently published intervals by Soldin et al.(4) demonstrate an age effect, although the limits are lower and do not show a consistent trend across the age groups(28). A plausible explanation is the application of a nonparametric percentile-based technique after cutting off 20% of the data (10% from each tail).
A limitation of clinical studies that use hospital databases is that the information is retrospective and obtained from a population of patients with a variety of illnesses, some of which may affect the thyroid gland. For example, patients with inherited deficiency of thyroid-binding globulin may have been included in the population, producing slightly lower T4 values, especially in males (29). Ideally, reference intervals should be determined using healthy subjects. Because endocrine function tests involve drawing blood, practical and ethical considerations require that we rely on the hospital database to ensure a sufficient number of subjects. We addressed this concern by excluding all tests from patients whose ICD-9 codes indicated a condition that may have affected their thyroid hormone concentrations. Another limitation of this study is that immunofluorometric systems can yield slightly different results than other laboratory assay techniques (15). However, interassay differences of this magnitude should not limit the applicability of our reference intervals for most children. We recommend that patients with borderline values be re-tested and that those with values falling outside the reference intervals should be further evaluated using ultrasensitive TSH measurements.
Our pediatric reference intervals for serum T4, T3, TSH, and free T4 are more accurate than those reported previously because they take into account the significant effects of both age and sex and are based on a very large number of children. The other advantage of these intervals is that they are easy to use and provide an improved clinical tool for assessing thyroid function in children and adolescents.
Acknowledgments
We thank Nader Rifai for valuable comments on this manuscript and Shawn F. O'Brien for technical expertise in database management.
Footnotes
fax 617-278-9770, e-mail zurakowski{at}a1.tch.harvard.edu
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  R. C Gordon, M. C Rose, S. A Skeaff, A. R Gray, K. M. Morgan, and T. Ruffman Iodine supplementation improves cognition in mildly iodine-deficient children Am. J. Clinical Nutrition, November 1, 2009; 90(5): 1264 - 1271. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Oken, L. E. Braverman, D. Platek, M. L. Mitchell, S. L. Lee, and E. N. Pearce Neonatal Thyroxine, Maternal Thyroid Function, and Child Cognition J. Clin. Endocrinol. Metab., February 1, 2009; 94(2): 497 - 503. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. G. Hollowell, N. W. Staehling, W. D. Flanders, W. H. Hannon, E. W. Gunter, C. A. Spencer, and L. E. Braverman Serum TSH, T4, and Thyroid Antibodies in the United States Population (1988 to 1994): National Health and Nutrition Examination Survey (NHANES III) J. Clin. Endocrinol. Metab., February 1, 2002; 87(2): 489 - 499. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. A. Huang, H. M. Tu, J. W. Harney, M. Venihaki, A. J. Butte, H. P.W. Kozakewich, S. J. Fishman, and P. R. Larsen Severe Hypothyroidism Caused by Type 3 Iodothyronine Deiodinase in Infantile Hemangiomas N. Engl. J. Med., July 20, 2000; 343(3): 185 - 189. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
