Objective. To describe intellectual, motor, and school-associated outcome in young adults with early treated congenital hypothyroidism (CH) and to study the association between long-term outcome and CH variables acting at different points in time during early development (CH severity and early l-thyroxine treatment levels [0–6 years]).
Methods. Neuropsychological tests were administered to all 49 subjects with CH identified during the first 3 years of the Norwegian neonatal screening program (1979–1981) at a mean age of 20 years and to 41 sibling control subjects (mean age: 21 years).
Results. The CH group attained significantly lower scores than control subjects on intellectual, motor, and school-associated tests (total IQ: 102.4 [standard deviation: 13] vs 111.4 [standard deviation: 13]). Twelve (24%) of the 49 CH subjects had not completed senior high school, in contrast to 6% of the control subjects. CH severity (pretreatment serum thyroxine [T4]) correlated primarily with motor tests, whereas early l-thyroxine treatment levels were related to verbal IQ and school-associated tests. In multiple regression analysis, initial l-thyroxine dose (β = 0.32) and mean serum T4 level during the second year (β = 0.48) predicted Verbal IQ, whereas mean serum T4 level during the second year (β = 0.44) predicted Arithmetic.
Conclusions. Long-term outcome revealed enduring cognitive and motor deficits in young adults with CH relative to control subjects. Verbal functions and Arithmetic were associated with l-thyroxine treatment variables, suggesting that more optimal treatment might be possible. Motor outcome was associated with CH severity, indicating a prenatal effect.
- congenital hypothyroidism
- thyroid hormone
- thyroxine treatment
- adult outcome
- motor function
The prognosis for children with congenital hypothyroidism (CH) is greatly improved with neonatal screening. Although screening ensures early treatment, developmental problems in relation to IQ, motor function, and school-associated outcome are still reported in follow-up studies. Most studies have shown lower intelligence (IQ) in early treated children with CH, relative to control subjects,1 although some studies have not found significant group differences.2–4 Motor deficits are prevalent and most pronounced in fine motor function.1,3,5 Psychoeducational outcome studies show inconsistent results in that some reported normal psychoeducational function,2,6 whereas others found some form of learning difficulties, including mathematics.7–11 Previous studies have addressed outcome in adolescence,2,6,8,11–13 and adult outcome has not been studied. The understanding of why children with CH show developmental delay is still a controversial issue. Derksen-Lubsen and Verkerk1 postulated in a 1996 review article that severity of CH seems to be the most important independent risk factor for outcome, and the effect of severe CH is assumed to be attributable to prenatal hypothyroidism. They did not find that treatment variables have an important effect on cognitive development. In contrast to this, some studies advocate that developmental delay is attributable to nonoptimal treatment.14–16 Both American and European treatment recommendations propose higher initial l-thyroxine dose than previously recommended.17,18 Promising results are reported in children who are treated early with a high initial l-thyroxine dose, with which even children with severe hypothyroidism show normal intelligence.14–16 However, the high-dose treatment groups were small, and the children studied were 4 years of age or younger. However, others emphasize that the optimal l-thyroxine dose remains unclear.19,20 Hrytsiuk et al19 summarized findings on the effect of l-thyroxine starting dose on developmental outcome, concluding that the evidence for such an effect is weak. Research on the importance of the l-thyroxine treatment level in relation to developmental outcome thus has yielded inconsistent results. One problem is that most outcome studies do not report systematic treatment data and often do not use multivariate statistics to obtain estimates for the independent effects of different CH variables on outcome. In a previous study,21 we found treatment variables during the first 2 years of life to account for a significant portion of the variance in verbal IQ at age 6.
The objectives of the present study were to 1) investigate long-term outcome in young adults with CH by studying whether all subjects with CH identified during the first 3 years of the national Norwegian screening program differ from sibling control subjects on intellectual, motor, and school-associated tests and 2) study how long-term outcome is affected by severity of hypothyroidism and serum T4 levels during the first years of life. We also wanted to study whether prenatal hypothyroidism and thyroxine treatment in different time periods have effects on different outcome variables in young adults.
The Norwegian national neonatal screening program was started in 1979 at the Pediatric Research Institute at the National Hospital in Oslo. Forty-nine children (29 girls) with CH were identified during the first 3 years of the screening program (November 1978 through 1981). This 3-year cohort was included in the present follow-up study and has participated in a previous study at the age of 2 and 6 years.21,22 All subjects agreed to participate in the present follow-up study at a mean age of 20.2 years (18.3–21.7 years). Forty-one siblings (16 girls) were used as control subjects (mean age: 21.4 years; 12.3–30.0 years). The sibling closest in age, preferably of the same sex, was included. One sibling refused. Exclusion criteria for all subjects were other disorders known to influence cerebral development or function. None of the CH subjects had central nervous system disorders or a history of head trauma. One sibling was excluded because of a pervasive developmental disorder. One CH-sibling pair was of foreign origin but spoke Norwegian well enough to be assessed. Four CH subjects were not treated early and continuously (3 had delayed start of treatment because of suspected transitory hypothyroid conditions; 1 was without drug therapy for approximately 1 year). Mean initial serum T4 level in these subjects was 87.0 nmol/L (±26.0 nmol/L), compared with 42.8 nmol/L (± 31.5 nmol/L) in the total cohort. We present mean neuropsychological test results for the total CH group (N = 49) and their sibling control subjects. However, in the CH-sibling comparisons, results were similar even when these 4 children were excluded (data not shown). In analyses regarding the impact of CH variables on developmental outcome, only the Norwegian CH subjects with early and continuous l-thyroxine treatment were included (N = 44). In analyzing the group difference in school completion, the siblings who were not old enough to have finished high school were excluded (N = 10).
Informed consent was obtained from the young adults with CH and their siblings and parents. The Medical Research Ethics Committee approved the study. The first author, without knowledge of earlier test results and CH characteristics (severity and treatment), conducted the neuropsychological assessments. However, she was not blind to whether a CH or a control subject was tested. A pretrained test assistant participated in test administering, blind to both CH parameters and subject status, but the first author scored and interpreted all test results.
Parental socioeconomic status (SES) was rated blindly on a 5-point scale based on the profession and education of head of household; SES-1 corresponded to university degree or head of own business; SES-5 indicated unemployed or receiving medical or social security.22 Mean SES score was 2.3 (±0.9), and there were no families in the lowest SES level.
Biomedical diagnostic and early treatment data were obtained previously from the medical records22 and are presented in Table 1. CH severity measures include serum T4 concentration at diagnosis, skeletal maturity at diagnosis (knee epiphyses score [range: 0–4; score 0–1, absent or incipient knee epiphyses, to 4, both femoral and tibial epiphyseal diameters >3 mm]),23,24 and scintigraphic classification of the type of congenital hypothyroidism. In the correlations analyses, all CH severity measures were included, whereas in the regression analyses, we used serum T4 at diagnosis as measure of CH severity.25l-Thyroxine treatment measures include l-thyroxine starting dose, mean serum T4 values calculated for each child from all serum T4 values during defined periods (first year of life [from after 14 days of treatment], 1.001–2 years, 2.001–4 years, 4.001–6 years) and serum hormone measures at 20 years of age (T4 and thyroid-stimulating hormone [TSH]).
IQ, motor function, and school-associated performance are reported. IQ was measured with the Wechsler Abbreviated Scale of Intelligence (WASI),26 consisting of 2 subtests (Vocabulary, Similarities) assessing verbal IQ and 2 subtests (Block Design, Matrix Reasoning) assessing performance IQ. US standards were used. A study of the psychometric properties of the Norwegian WASI translation found that mean T scores and IQ results, as well as intercorrelations of subtests and IQ values, closely resemble published results in the US population.27 Motor speed and coordination were evaluated using the Finger Tapping and the Grooved Pegboard tests.28 The Grooved Pegboard scores represent seconds to complete the task; thus, high scores on this test indicate more problems than low scores. The short form of Bruininks-Oseretsky Test of Motor Proficiency29 was used as an indication of general motor proficiency. School-associated performance was assessed in relation to language and arithmetic and via an interview. The Norwegian Observational Test of Reading and Writing30 is a simple test made to screen for dyslexia. It includes an evaluation of reading speed (words per minute), comprehension, knowledge of the alphabet, and dictation. The results are used together with IQ scores to clinically diagnose whether the subject has a generalized learning disorder (reduced IQ in addition to specific language difficulties) or dyslexia (normal IQ and specific language difficulties). Naming difficulties and verbal fluency were tested with the Boston Naming Test and the Controlled Oral Word Association Test.28 The Arithmetic subtest from the Wechsler Intelligence Scales (WAIS-R31; Wechsler Intelligence Scale for Children-Revised32 for the younger siblings) was used as a screening of arithmetic abilities. Results are expressed as raw scores for all tests except the WASI and the Arithmetic subtest, for which age-adjusted T scores were adapted from US norms. All subjects provided information from their school record and filled out the Botting self-rating scale of school performance33 (1- to 5-point self-rating of reading, writing, spelling, arithmetic, physical education, and overall cognitive function).
Developmental variables were reasonably normally distributed. Means (±standard deviation) are reported. A linear mixed model34 was used to analyze group differences controlling for age and sex. For adjusting for dependence between siblings, variation between sibling pairs was introduced as a random effect in the mixed model. Significance level was set to .05. Bonferroni correction was applied to adjust for multiple comparisons within each neuropsychological domain, thus setting the critical values to P = .02 for the IQ measures and P = .01 for the motor measures and the 4 school-associated measures. The χ2 test was used to assess whether the groups differed in the presence of dyslexia. Within the CH group, independent t tests were used to evaluate outcome differences between the school completers (N = 37) and noncompleters (N = 12). Bivariate correlation analysis (Pearson) was used between CH variables (severity and treatment), background variables, and outcome at age 20. Serum hormone levels at age 20 were skewed, and Spearman rank correlations were used for these measures.
Hierarchical linear multiple regression analyses were used to analyze the effect of the l-thyroxine treatment variables on outcome measures, controlling for background (sex and SES) and CH severity (serum T4 at diagnosis). Forced entry of background and CH severity and stepwise introduction of T4 variables were performed. Missing mean substitution was used. R2 adjusted was used as an expression of explained variance, and in evaluating the significance of explained variances, we used Cohen’s criteria: 2% to 13% is small, 13% to 26% is medium, and >26% is large.35
Long-term Outcome in CH Versus Sibling Control Subjects
Neuropsychological test results are presented in Table 2.
The CH group performed two thirds of a standard deviation below sibling control subjects. The differences are statistically significant.
The groups did not differ in simple motor speed (Finger Tapping), but the CH group was significantly impaired in motor coordination (Grooved Pegboard) and global motor proficiency (Bruininks-Oseretsky) compared with control subjects.
On the basis of the performance on the administered reading and writing tests, along with IQ, 4 CH subjects were classified as having either a generalized learning disorder (2) or dyslexia (2), compared with 5 subjects showing signs of dyslexia in the sibling control group. The difference is nonsignificant (χ2 = 2.05, P = .15). The CH group showed no impairment in verbal fluency (Controlled Oral Word Association Test) compared with sibling control subjects, but significant differences were found in naming ability (Boston Naming Test) and on the arithmetic screening (Wechsler Arithmetic subtest).
On the Botting self-rating of school performance, the CH group rated themselves lower in arithmetic (P = .04) and in overall cognitive functioning at school (P = .02), whereas there were no differences in their self-rating of other school performances, such as reading, writing and physical education (data not shown). All subjects completed 9 years of compulsory education. Twelve (24%) of the 49 CH subjects started but did not graduate from senior high school, in contrast to only 2 of the 41 sibling control subjects (6% of the 31 of the siblings who were old enough to have completed senior high school). Some CH subjects failed final math examinations; some just left for no obvious reason, whereas some explicitly stated that they found school too demanding. SES did not influence school dropout: 10 of the 12 dropouts come from the highest (N = 2) and the next highest (N = 8) SES level. Verbal IQ was 94.7 (±12) in the 12 CH subjects who did not complete senior high school, in contrast to 105 (±12) in the completers, a significant difference (t = 2.53, P = .015). Analyses across other motor or school-associated tests did not reveal significant group differences.
Long-Term Outcome Related to CH Variables
Neuropsychological Test Results
The bivariate correlations in Table 3 show that the CH severity factors primarily correlated with motor outcome. The l-thyroxine treatment during the first 6 years correlated primarily with verbal IQ, other language tests, and the arithmetic screening. Regression analyses generally confirmed the results from bivariate correlations (Tables 4 and 5). Early treatment variables were significant predictors of verbal IQ, language, and the arithmetic screening. The starting dose of l-thyroxine and mean serum T4 during the second year of life explained 21% of the variance (R2 adjusted) in verbal IQ at age 20. Treatment variables did not significantly predict performance IQ. In general, severity of CH (serum T4 at diagnosis) predicted motor performance. An inverse relationship was found on 1 motor test (Bruininks-Oseretsky), in that high mean serum T4 level during 4 to 6 years was negatively related to performance. SES was not significantly related to outcome at age 20. At 20 years of age, 44.9% of the CH subjects (N = 22) had serum TSH values above upper reference range. However, neither serum fT4 nor serum TSH at 20 years of age correlated significantly with outcome measures.
The starting dose of l-thyroxine was significantly higher in the school completers (N = 37) compared with the school noncompleters (N = 12; 9.2 [±3.5] μg/kg vs 7.1 [±2.1] μg/kg; t = 1.99, P = .018).
Long-Term Outcome in CH Versus Sibling Control Subjects
The CH group showed a mean IQ deficit compared with sibling control subjects at 20 years of age, which was similar to the difference found at age 6.22 This indicates an enduring but not increasing IQ deficit in these CH subjects. Consistent with the literature,1,3,5 the CH group experienced motor deficits compared with control subjects, and group differences increased with the complexity of the motor task performed. There were no significant group differences in verbal fluency or in the prevalence of dyslexia. We propose that the reading difficulties reported in the CH literature11 is related more to a general verbal deficit rather than to a specific language disorder. The deficit found on the arithmetic screening is consistent with that of other studies,8,11 possibly caused by impaired concentration as well as difficulties with handling numbers in the CH group. Consistent with their test results, the CH group rated themselves lower than did sibling control subjects on the Botting scales of arithmetic and of overall cognitive functioning at school. Despite their weaker motor test results, the CH group did not rate themselves different from sibling control subjects in physical education, and this finding could suggest a successful integration of students with different degrees of motor fitness in this subject. The relatively greater percentage of CH subjects who did not complete senior high school is a matter of concern. SES did not account for significant variation in outcome at age 20. This might be because the study does not comprise subjects from the lowest SES level and that Norway is a relatively homogeneous society compared with most European and American countries. In this article we have focused on selected outcome measures (IQ, motor function, and school-associated outcome) and cannot conclude on other functions possibly affected in adult CH subjects.
Long-Term Outcome Related to CH Variables
In this study, neuropsychological outcome at 20 years of age was associated with both CH severity and early treatment factors. In multiple regression analysis, several motor tests were consistently associated with serum T4 at diagnosis, a measure of CH severity, whereas verbal functions and the arithmetic screening were associated with l-thyroxine starting dose and mean serum T4 level during the first 6 years. Treatment factors explained 4% to 29% of the variance in different IQ and school-associated measures (R2 adjusted). Specifically, l-thyroxine starting dose and mean T4 during the second year explained 21% of the variance in verbal IQ at 20 years, and this is considered a medium effect size.35
As already mentioned, a greater percentage of the CH subjects did not finish high school. Additional weight to this finding is added by the fact that the noncompleters were given a lower starting dose of l-thyroxine than the completers. This result is in line with a large population-based cohort study of young CH adolescents in France,6 where early l-thyroxine treatment level was important in relation to school delay.
That long-term outcome was associated with l-thyroxine treatment levels during childhood is consistent with some other studies,6,11,14–16,36 and supports the view that treatment might be improved by higher treatment levels.14–17 However, the importance of l-thyroxine treatment levels after infancy has not been systematically examined in most earlier studies. The CH subjects who were included in this study were born 20 years ago and were not treated according to present recommendations in infancy. Still, the children did not receive inadequate treatment, as defined by the New England Congenital Hypothyroidism Collaborative (TSH >15 mU/L after 2 weeks of treatment and T4 concentration <103 nmol/L on >1 occasion during the first year of life),37 but the initial dose was lower than recent recommendations.17,18 Several of the early treatment variables correlated with verbal IQ and school-associated abilities. We studied effects of the initial l-thyroxine starting dose and serum T4 levels from start of treatment until age 6 years and found significant correlations between mean serum T4 levels throughout these age periods and verbal outcome. Correlations were most pronounced and consistent for treatment variables during the first 2 years. However, as the treatment variables during different time periods were correlated and to control for confounding variables (in particular, severity of hypothyroidism), multiple regression analyses were conducted. The effects of treatment variables persisted in these analyses, and l-thyroxine starting dose and mean serum T4 level during the second year of life were the strongest predictors. The exact effect of each treatment variable on outcome is difficult to determine in our study. The effect of each single factor was weakened by a relatively low N and by the fact that several of the early treatment variables, which were entered into the regression analyses, correlated with each other and with outcome measures. On 1 of the motor tests (Bruininks-Oseretsky), mean serum T4 level at age 4 to 6 was inversely related to outcome in the regression analysis. The significance of this finding remains unclear: whether it is a coincidence or a negative effect of a higher l-thyroxine treatment level on some functions in CH children. There were no significant relations between serum hormone levels at age 20 and outcome, in line with our previous study at age 6 years.22 This is in contrast to some studies on children with CH, in which negative associations were reported between T4 levels at time of testing and attention measures38 and positive associations between thyroid hormone levels at time of testing and several outcome measures (memory, speed of processing, fine motor function, and language).36 Many of the CH subjects had serum TSH above upper reference range at age 20, which could suggest nonoptimal follow-up or poor compliance by the young adults.
Severity of CH correlated with motor outcome, whereas early l-thyroxine treatment levels were related to verbal and mathematical abilities. We hypothesize that these contrasting findings result from different thyroid effects at different stages of development. Severe hypothyroidism, as found in children with low serum T4 at diagnosis, is assumed to be associated with prenatal hypothyroidism. Motor outcome at age 20 correlated with initial serum T4 at diagnosis, also in multivariate analyses controlling for treatment variables, indicating an irreversible effect of prenatal hypothyroidism on motor functions. This is consistent with animal studies showing that hypothyroidism leads to dendritic spread reduction in the Purkinje cells of the rat cerebellum. To prevent this reduction, l-thyroxine supplementation had to be given before postnatal week 2, which is equivalent to the prenatal period in humans. Later l-thyroxine supply had no effect.39
Strengths and Limitations
The strength of this study is the inclusion of a total 3-year cohort of CH subjects who were followed from infancy to young adulthood and a sibling control group. We had systematic data on the l-thyroxine treatment until 6 years of age. The relatively low N imposed limitations to the analyses that could be performed. That these children started their treatment 20 years ago may limit the representativeness of the study for present cohorts.
Our study highlights the importance of the l-thyroxine treatment level during both infancy and early childhood years for the adult outcome in CH subjects.
The present study, focusing on IQ, motor function, and school-associated outcome, found significant differences between young adults with CH and sibling control subjects. Although the children were treated adequately according to recommendations at that time, we found that early l-thyroxine treatment level was associated with outcome. Different thyroid variables acting at different points in early development had effects on different outcome variables. Motor outcome was not associated with treatment variables but with the severity of CH, indicating an irreversible prenatal effect on motor functions. Verbal IQ and school-associated outcome was associated with l-thyroxine treatment variables, suggesting that more optimal treatment might be possible.
This study was supported by grants from the Norwegian Research Council and The Norwegian Research Fund For Mental Deficiency.
- ↵Simons WF, Fuggle PW, Grant DB, Smith I. Intellectual development at 10 years in early treated congenital hypothyroidism. Arch Dis Child.1994;71 :232– 234
- ↵Simons WF, Fuggle PW, Grant DB, Smith I. Educational progress, behaviour, and motor skills at 10 years in early treated congenital hypothyroidism. Arch Dis Child.1997;77 :219– 222
- Bargagna S, Dinetti D, Pinchera A, et al. School attainments in children with congenital hypothyroidism detected by neonatal screening and treated early in life. Eur J Endocrinol.1999;140 :407– 413
- ↵Rovet JF, Ehrlich R. Psychoeducational outcome in children with early-treated congenital hypothyroidism. Pediatrics.2000;105 :515– 522
- ↵Rovet JF. Long-term neuropsychological sequelae of early-treated congenital hypothyroidism: effects in adolescence. Acta Pediatr.1999;88(suppl 432) :88– 95
- ↵American Academy of Pediatrics, Section on Endocrinology and Committee on Genetics, and American Thyroid Association, Committee on Public Health. Newborn screening for congenital hypothyroidism: recommended guidelines. Pediatrics.1993;91 :1203– 1209
- ↵Hindmarsh PC. Optimisation of thyroxine dose in congenital hypothyroidism. Arch Dis Child.2002;86 :73– 75
- ↵Wechsler D. Wechsler Abbreviated Scale of Intelligence Manual. San Antonio, TX: Harcourt Brace & Company; 1999
- ↵Brager-Larsen L, Sundet K, Engvik H, et al. Psychometric properties of a Norwegian research version of the Wechsler Abbreviated Scale of Intelligence (WASI) [abstract]. Nevropsykologi. Bull Norwegian Neuropsychol Assoc2001;4(suppl 1) :70
- ↵Lezak M. Neuropsychological Assessment. New York, NY: Oxford University Press; 1995
- ↵Bruininks-Oseretsky Test of Motor Proficiency: Examiners Manual. Circle Pines, MN: American Guidance Service; 1978
- ↵Dale G, Carlsten C. Norwegian Test of Reading and Writing for Secondary School. Oslo, Norway: The Norwegian University Press; 1990
- ↵Wechsler D. Wechsler Adult Scale of Intelligence Revised. San Antonio, TX: Harcourt Brace & Company; 1981
- ↵Wechsler D. Wechsler Intelligence Scale for Children Revised. San Antonio, TX: Harcourt Brace & Company; 1974
- ↵SPSS Inc. Statistical Package for the Social Sciences: SPSS 11. Chicago, IL: SPSS Inc; 2002
- ↵Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates; 1988
- ↵Rovet J, Alvarez M. Thyroid hormone and attention in school-age children with congenital hypothyroidism. J Child Psychol Psychiatry.1996;7 :579– 585
- Copyright © 2003 by the American Academy of Pediatrics