Clinical Effectiveness and Cost-Effectiveness of the Use of the Thyroxine/Thyroxine-Binding Globulin Ratio to Detect Congenital Hypothyroidi
http://www.100md.com
《小儿科》
Department of Social Pediatrics and Child and Youth Health Care, Netherlands Organization of Applied Scientific Research Prevention and Health, Leiden, Netherlands
Department of Pediatric Endocrinology, Emma Children's Hospital, Academic Medical Center, University of Amsterdam, Netherlands
National Institute for Public Health, Bilthoven, Netherlands
ABSTRACT
Context. Since the introduction of screening for congenital hypothyroidism (CH) in 1974, the optimal laboratory strategy has been the subject of debate.
Objective. To assess the clinical effectiveness and cost-effectiveness of various types of thyroxine (T4)-based strategies to screen for CH.
Design, Setting, and Participants. In the Netherlands, since January 1, 1995, a primary T4 determination with supplemental thyroid-stimulating hormone (TSH) and T4-binding globulin (TBG) measurements has been used. Results were calculated from cumulative findings for 1181079 children screened between January 1, 1995, and December 31, 2000.
Main Outcome Measures. Rates of detection of patients with CH of thyroidal origin (CH-T) or CH of central origin (CH-C), false-positive rates, laboratory costs, and costs of initial diagnostic evaluations.
Results. All known infants (n = 393) with CH-T and 92% (n = 66) of infants with CH-C were detected on the basis of low T4 levels, TSH elevation, and/or low T4/TBG ratios. If the decision to refer had been based solely on TSH elevation, then 94% of patients with CH-T and none of the patients with CH-C would have been detected. If low T4 levels (–3.0 SD) and TSH elevation had been used as the criteria for referral, then the rates of detection would have been 96% for CH-T and 31% for CH-C. The false-positive rates for the 3 approaches were 0.5, 3.3, and 4.7 cases per case detected, respectively. The introduction of the T4/TBG ratio into a program using a primary T4 with supplemental TSH approach generates an extra cost of $11206 per additional case detected. The average costs to detect 1 patient are comparable for the 3 approaches. In addition, our data revealed a substantially greater prevalence of CH-C than reported previously (1 case per 16404 children, compared with earlier estimates of 1 case per 26000 infants to 1 case per 29000 infants).
Conclusions. The T4 plus TSH plus TBG approach is a recommendable strategy for neonatal CH screening. It offers outstanding detection of patients with CH-C, in addition to those with CH-T, with acceptable costs.
Key Words: average cost cost-effectiveness congenital hypothyroidism laboratory approach marginal cost screening thyroid-stimulating hormone thyroxine thyroxine-binding globulin
Abbreviations: CH, congenital hypothyroidism TSH, thyroid-stimulating hormone TBG, thyroxine-binding globulin T4, thyroxine CH-C, congenital hypothyroidism of central origin CH-T, congenital hypothyroidism of thyroidal origin
Congenital hypothyroidism (CH) refers to a heterogeneous group of disorders that have in common a lack of thyroid hormone already present in the neonatal period. Because thyroid hormone is essential for brain development, the major risk of CH, if left untreated, is irreversible cognitive and motor impairment.1 Although CH was once a major cause of mental retardation, almost all developed countries have now instituted a neonatal mass screening program to allow for early identification and treatment of affected children.
Since the introduction of screening for CH in 1974, the optimal laboratory strategy has been subject of debate.2 Japan, Australia, most European countries, and many North American states use an approach in which only thyroid-stimulating hormone (TSH) is used to discriminate between affected and unaffected infants. In this way, "classic" CH cases with elevated TSH concentrations would be detected with high specificity. However, mild cases of CH of thyroidal origin (CH-T) with only slightly elevated TSH levels and/or delayed TSH elevation would remain undetected, as would cases of CH of central origin (CH-C). Another strategy is to first measure thyroxine (T4) in all samples, followed by TSH measurement for samples with low T4 values. Several North American states use this strategy.
In the Netherlands, the latter approach was extended with the determination of T4-binding globulin (TBG) levels for the lowest 5% of T4 values. The T4/TBG ratio serves as an indirect measure of the free T4 concentration (which cannot be determined directly in dried blood spots). In contrast to most screening programs, in which TSH levels are determined for the lowest 10% of T4 readings, TSH levels are measured for the lowest 20% of T4 values. In this way, the Dutch screening program provides unique information about the prevalence of CH-C and mild cases of CH-T. Also, data derived from the program may allow calculation of the outcomes of other, less-sensitive strategies. This information may be used for evaluation of concurrent screening programs, as well as our own. In this article, we report on the clinical effectiveness and cost-effectiveness of the Dutch T4 plus TSH plus TBG screening strategy, in comparison with 2 alternative, primary T4 strategies.
METHODS
Screening Procedures
In the Netherlands, nationwide neonatal screening for CH, based on primary T4 and supplemental TSH measurements, started in 1981. In 1995, after a 1-year pilot period, TBG measurement was added to the screening procedure. Blood samples are collected on filter paper through heel puncture. Until the middle of 1999, blood was sampled at day 5 to day 7 after birth. From that time onward, this was advanced to day 4 to day 7 after birth. T4 concentrations are normalized for each assay run and are expressed as SDs relative to the daily mean. If the T4 concentration is among the lowest 20% of values for the series for that day (ie, –0.8 SD), then TSH (in microunits per milliliter of serum) is measured in the same sample. If the T4 concentration is among the lowest 5% (ie, –1.6 SD), then TBG (in nanomoles per liter; 1 μmol/L = 54 μg/dL) is also measured. The T4/TBG ratio is calculated as follows: [T4 (in SD) plus 5.1]/[TBG (in micromoles per liter of serum; 1 μmol/L = 54 μg/dL)], where the addition of 5.1 serves to keep ratios in the positive range. Screening results are kept by the Dutch Health Administrations. Their medical advisors decide whether a second heel puncture or referral is indicated. Medical evaluation of patients is performed by general pediatricians. To prevent delayed diagnosis of the most severe cases, immediate referral occurs if the T4 level is –3.0 SD and/or the TSH level is 50 μIU/mL serum. A second heel-stick blood sample is requested in cases with a borderline T4 concentration (–2.9 SD T4 –1.6 SD) in combination with a low T4/TBG ratio (8.5) and/or a borderline TSH concentration (20 μIU/mL TSH < 50 μIU/mL). If, in a second heel puncture, the T4 level is –1.6 SD in combination with a low T4/TBG ratio and/or the TSH level is 20 μIU/mL, then the infant is referred to a general pediatrician. Newborns with a birth body mass of 2500 g and a gestational age of 36 weeks are referred exclusively on the basis of their TSH concentrations.
Data and Data Sources
The Netherlands Organization of Applied Scientific Research Prevention and Health documents the screening results and diagnostic findings for all children screened for CH in the Netherlands, for the purposes of monitoring and quality improvement of the screening program. For this, information is gathered from all Dutch Health Administrations, which are responsible for the execution of the neonatal screening program for CH, and from pediatricians. Screening outcomes are kept and, for each referral, screening and diagnostic T4, TSH, and TBG concentrations are recorded. The pediatrician's diagnosis is recorded as no CH, transient CH, CH-T, or CH-C. Permanence of hypothyroidism is assumed if the pediatrician reports that an ectopic/dystopic gland or absent thyroid tissue was revealed. For all other cases of CH, the permanence of the hypothyroidism is assessed when the child reaches the age of 4 years. To attempt to identify false-negative cases, the Netherlands Organization of Applied Scientific Research Prevention and Health stays in close contact with the Dutch Health Administrators and pediatricians, mainly through the attendance of advisory board meetings on all aspects of neonatal screening and correspondence regarding screening and diagnostic outcomes. For the current analyses, we used data for the period from January 1, 1995 (the day TBG measurement was incorporated into the national screening program), to December 31, 2000.
Cost-Effectiveness Analysis
In this study, we set out to compare the results of the Dutch T4 plus TSH plus TBG approach with 2 alternative, primary T4 with supplemental TSH approaches. The first is the (T4) plus TSH approach in which T4 readings are not reported and are used only to determine whether TSH measurements are indicated. More specifically, referral is indicated for all infants with TSH concentrations of 50 μIU/mL and for infants with TSH readings between 20 and 50 μIU/mL if these findings are confirmed with a repeat heel-puncture blood sample.
The second approach is the T4 plus TSH approach, in which both T4 and TSH readings are reported and are used to determine whether referral is indicated. Referral is indicated for all subjects with T4 levels of –3.0 SD and/or TSH levels of 50 μIU/mL and for subjects with TSH readings between 20 and 50 μIU/mL if these findings are confirmed with a repeat heel-puncture blood sample.
First, we calculated the detection, second heel puncture, and false-positive rates for the (T4) plus TSH and T4 plus TSH approaches from the frequency distributions of screening T4 and TSH concentrations obtained from the Dutch neonatal CH screening program. In accordance with the T4 plus TSH plus TBG approach, we used a cutoff point for TSH measurement of 20%. The next step involved linkage of effectiveness and effort. Effectiveness was defined as the number of cases of permanent CH-T and CH-C detected, with the pediatrician's diagnosis being taken as the definite diagnosis. Effort was defined as the second heel puncture rate, false-positive rate, laboratory costs, and costs of the initial diagnostic evaluations (in US dollars; at the time of these calculations, 1 US dollar = 1 euro). In the Netherlands, laboratory costs are $1.25 for T4 measurements, $3.41 for TSH measurements, and $8.64 for TBG measurements. The sampling and laboratory costs of a second heel puncture average $11.50.3 Cost of wages and salaries for laboratory personnel, housing, and laboratory equipment were assumed to be constant for each of the 3 approaches and therefore were not taken into account. The cost of initial diagnostic evaluations for infants with abnormal screening results was estimated at $1123 (including costs of 1 consultation with a general practitioner, 1 pediatric consultation, and the initial diagnostic laboratory testing). Cost-effectiveness is presented as the average cost and the marginal incremental cost per case detected. Marginal cost refers to the extra cost required for each additional case that is detected.4,5
RESULTS
Population
Between January 1, 1995, and December 31, 2000, 1185670 children were born in the Netherlands. Of these infants, 1181079 (99.6%) were screened for CH. On the basis of their abnormal screening results, 2604 (0.2%) were referred to a general pediatrician. Of these, 459 infants (18%) were diagnosed with permanent CH, 393 (86%) of whom had CH-T and 66 (14%) of whom had CH-C. Of the infants with CH-T, 41% were male. Their mean ± SD gestational age was 39.5 ± 2.4 weeks, with a mean ± SD birth body mass of 3368 ± 716 g. For infants with CH-C, these figures were 72%, 38.9 ± 2.8 weeks, and 3244 ± 648 g, respectively.
The screening program failed to detect 6 patients diagnosed as having CH-C; no missed cases of CH-T came to our attention (see Table 1 for screening results). From this, it follows that the sensitivity of the screening was 98.5% (459 of 465 patients with CH showed abnormal screening results). The specificity was 99.9% (1180004 of 1180463 children without CH tested negative), and the positive predictive value was 18% (459 cases of permanent CH were found among 2604 referrals).
our data, we calculated that the prevalence of permanent CH-C was 1 case per 16404 live-born infants (95% confidence interval: 1 case per 13174 infants to 1 case per 21173 infants), which is clearly higher than earlier reports indicated (see Discussion). The prevalence of permanent CH-T was found to be 1 case per 3017 live-born infants (95% confidence interval: 1 case per 2721 infants to 1 case per 3318 infants). Table 2 summarizes the screening data and additional clinical results for infants diagnosed as having CH-T or CH-C.
Detection Rates
Second Heel Puncture Rates
The T4 plus TSH plus TBG approach gave rise to 5310 second heel punctures during the 6-year study period. For the (T4) plus TSH and T4 plus TSH approaches, this would have been 969. These findings indicate second heel puncture rates of 11.6 punctures per case detected for the T4 plus TSH plus TBG approach, 2.4 punctures per case detected for the T4 plus TSH approach, and 2.6 punctures per case detected for the (T4) plus TSH approach (Table 3).
False-Positive Rates
During the 6-year study period, the T4 plus TSH plus TBG approach yielded 2604 referrals. For the T4 plus TSH and (T4) plus TSH approaches, the numbers of referrals would have been 1702 and 572, respectively. The false-positive rates for the 3 approaches were 4.7 cases per case detected for the T4 plus TSH plus TBG approach, 3.3 cases for the T4 plus TSH approach, and 0.5 case for the (T4) plus TSH approach (Table 3).
Cost-Effectiveness
DISCUSSION
Since 1995, the Dutch neonatal screening program for CH has been based on a T4 plus TSH plus TBG laboratory approach. We studied the cost-effectiveness of this approach in comparison with 2 other approaches, which were based on T4 and TSH determinations but not TBG measurements. Our main finding is that incorporation of TBG measurements into a T4 plus TSH approach leads to a threefold increase in the rate of detected cases of permanent CH-C.
Introduction of TBG measurements into a primary T4 with supplemental TSH approach can be undertaken with acceptable costs. Because almost all developed countries have already instituted CH screening programs, an important question addresses the extra cost of one approach over the cost of another. We therefore calculated the marginal incremental cost. Incorporation of TBG measurements into a T4 plus TSH approach generated an extra cost of $11206 for each additional patient with permanent CH identified (Table 4). Because the addition of TBG measurements resulted in greater numbers of patients detected, the average cost to detect 1 patient differed only slightly for the 3 approaches [$6353, $6209, and $6851 for the (T4) plus TSH, T4 plus TSH, and T4 plus TSH plus TBG approaches, respectively]. Patients with CH-C benefit most from inclusion of TBG measurements. In the Netherlands, from 1995 through 2000, 91.6% of known cases with permanent CH-C were detected by means of the T4 plus TSH plus TBG approach. The T4 plus TSH approach would have yielded a detection rate of only 30.6% for CH-C (Table 3). Although there is no estimate of the lifetime health care costs and costs of loss of productivity associated with non-timely treated CH-C, it can be expected that these costs far outweigh the additional screening costs.
The major advantage of the T4 plus TSH plus TBG approach is its excellent rate of detection of CH-C cases. Although the figures initially seem less impressive than those for CH-C, it should be noted that introduction of TBG measurements into the screening program for CH also leads to improved identification of patients with CH-T. From Table 3, it can be read that changing from a T4 plus TSH approach to a T4 plus TSH plus TBG approach would lead to a 3.8% increase in the detection rate for CH-T, over an already very high detection rate of 96.2% for CH-T. Although these patients are likely to have milder forms of CH-T, they do deserve medical attention, because of the disadvantageous effect that even slight hypothyroidism may have on brain development.6
Regarding the effectiveness of the different screening strategies, first we point out that we did not take into account the number of patients with transient CH, although many of these infants did need (temporary) T4 supplementation and therefore had true positive screening results. The main reason for leaving these cases out of the analyses is the occurrence of great variation in the severity and duration of hypothyroidism in this group. Also, in a substantial number of cases, the cause of the hypothyroidism lies outside the neonate's thyroid gland and hypothalamic-hypophyseal system (for example, transient thyroidal hypothyroidism may be attributable to abundant postnatal use of iodide-containing disinfectants, maternal use of antithyroid or iodide-containing drugs, or the presence of maternal antithyroid antibodies, and transient central hypothyroidism may be attributable to inadequately treated maternal Graves' disease7). Therefore, transient CH is an entity that is usually not included in CH statistics.
Second, it should be noted that we took the pediatrician's diagnosis as the final diagnosis. In some cases, it may be difficult to establish whether mild permanent or transient CH is present.8 Also, CH is a very heterogeneous disease with respect to the underlying cause.1 It is therefore not surprising that reassessment in some cases results in revision of the diagnosis. Even reexamination during the first year of life, in our experience, leads to a change of cause in some cases. However, this situation does not differ from that in other parts of the world. One advantage of our study is that we used the revised diagnoses established at 4 years of age whenever possible. Also, we calculated the outcomes of the 3 laboratory approaches by using the same study group. It is therefore unlikely that this undermines our conclusion.
A less attractive feature of the T4 plus TSH plus TBG approach, compared with the T4 plus TSH and (T4) plus TSH approaches, is its relatively high false-positive rate. This leads not only to higher laboratory costs and diagnostic evaluation costs but also to more parental anxiety.9 Theoretically, the number of false-positive test results for the T4 plus TSH plus TBG approach could be decreased if the T4/TBG ratio was measured not only for the group of infants with borderline T4 levels (–3.0 SD < T4 < –1.5 SD) but also for the group with low T4 concentrations (–3 SD); however, such an approach would lead to a delay of 1 to 3 days before an infant could be referred to a pediatrician, because of the increased workload for the laboratory. Because a group of patients with potentially severe CH is involved, such a delay is considered unacceptable.
Several North American states have considered switching from a (T4) plus TSH approach to a primary TSH strategy (ie, without initial T4 measurements). From our data, we can estimate the cost-effectiveness of such a change. The detection rate and the rates of second heel punctures and false-positive screening results are comparable for the 2 strategies (Table 3). However, because T4 measurements are inexpensive, in comparison with TSH measurements, and because TSH is measured for all screened infants and not just a preselected group with decreased T4 concentrations, the laboratory costs would increase sharply. This is especially important because laboratory costs make up a substantial part of the total cost of maintaining a screening program for CH. We estimated that the total laboratory costs would be nearly twice as high for a program using a primary TSH approach, compared with the (T4) plus TSH approach (ie, $4027479 vs $2292990) (Table 4). Similarly, the average laboratory cost per case detected is high (ie, $10196) when the primary TSH approach is used, compared with the (T4) plus TSH approach ($6353).
The need for early identification of patients with CH-C is not recognized generally.10 One commonly used argument against screening for CH-C is that it is usually mild, in terms of T4 concentrations. In our study, however, 39% of patients with CH-C had a screening T4 value 3.0 SD below the mean (Table 2). Also, it is known that three quarters of children with CH-C are found to have multiple pituitary hormone deficiencies, which is a potentially life-threatening condition.1 Delayed detection and incomplete diagnosis of the pituitary hormone deficiencies result in significant morbidity, such as severe hypoglycemia and neonatal hepatitis, and a mortality rate as high as 14%.1 Because all deficient hormones can be supplemented readily, timely diagnosis improves the outcomes of patients with CH-C significantly.
Another frequently heard argument against screening for CH-C is that its prevalence is too low to make screening worthwhile. Initial estimates of the prevalence of CH-C among live-born children in the United States and Canada ranged from 1 case per 110000 infants to 1 case per 68200 infants.11,12 Evaluation of the first 10-year screening experience in the northwest region of the United States in 1986 resulted in a prevalence estimate of 1 case per 29000 infants.13 Less than one half of these patients were detected with neonatal screening.13 A similar retrospective study of the 1981–1982 screening population in the Netherlands yielded a prevalence of 1 case per 26000 infants for CH-C.1 In the present study, 66 infants with CH-C were detected in a 6-year period. During this period, 6 additional cases (9%) were found on the basis of clinical symptoms. This adds up to a prevalence of 1 case per 16404 live newborns (95% confidence interval: 1 case per 13174 infants to 1 case per 21173 infants). Therefore, it seems that CH-C is a more commonly encountered disorder than thought previously. The prevalence of CH-C is on the same order as that of, for example, phenylketonuria (1 case per 18000 infants),14 a disease for which most developed countries have instituted a screening program, or congenital adrenal hyperplasia (1 case per 11764 infants).15 Because it is unlikely that the prevalence of CH-C in the Netherlands is significantly different from that in North America, we presume that in previous studies both the mildest and most severe cases were missed, similar to the situation for congenital adrenal hyperplasia in the prescreening era.16
CONCLUSIONS
The T4 plus TSH plus TBG approach is a recommendable strategy for neonatal CH screening. The main advantage is its outstanding ability to identify patients with CH-C. Early detection of CH-C prevents severe morbidity and death, because it is often associated with other pituitary hormone deficiencies. In addition, CH-C seems to be a more common disorder than assumed generally, which makes early detection even more worthwhile. Introduction of TBG measurements into a primary T4 with supplemental TSH approach can be done with costs that seem fairly acceptable, especially when considered against the anticipated long-term health care costs and the costs of loss of productivity for infants whose disease is not detected early.
ACKNOWLEDGMENTS
We thank M. Elske van den Akker van Marle, PhD, health economist, for valuable advice and review of the manuscript.
FOOTNOTES
Accepted Nov 16, 2004.
No conflict of interest declared.
REFERENCES
Vulsma T, de Vijlder JJM. Thyroid disease in newborns, infants, and children. In: Wass JAH, Shalet SM, eds. Oxford Textbook of Endocrinology and Diabetes. Oxford, United Kingdom: Oxford University Press; 2002: 532–544
Dussault JH. The anecdotical history of screening for congenital hypothyroidism. J Clin Endocrinol Metab. 1999;84 :4332 –4334
Oostenbrink JB, Koopmanschap MA, Rutten FFH. Handleiding voor Kostenonderzoek, Methoden en Richtlijnen voor Economische Evaluaties in de Gezondheidszorg [in Dutch]. Amstelveen, Netherlands: College voor Zorgverzekeringen; 2000
Neuhauser D, Lewicki AM. What do we gain from the sixth stool guaiac N Engl J Med. 1975;293 :226 –228
Drummond MF, O'Brien B, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. 2nd ed. Oxford, United Kingdom: Oxford University Press; 1997
Léger J, Larroque J, Norton J. Influence of severity of congenital hypothyroidism and adequacy of treatment on school achievement in young adolescents: a population-based cohort study. Acta Paediatr. 2001;90 :1249 –1256
Kempers MJ, van Tijn DA, van Trotsenburg AS, de Vijlder JJ, Wiedijk BM, Vulsma T. Central congenital hypothyroidism due to gestational hyperthyroidism: detection where prevention failed. J Clin Endocrinol Metab. 2003;88 :5851 –5857
Moreno JC, Bikker H, Kempers MJ, et al. Inactivating mutations in the gene for thyroid oxidase 2 (THOX2) and congenital hypothyroidism. N Engl J Med. 2002;347 :95 –102
Tymstra T. False positive results in screening tests: experiences of parents of children screened for congenital hypothyroidism. Fam Pract. 1986;3 :92 –96
Toublanc JE. Guidelines for neonatal screening programs for congenital hypothyroidism. Acta Paediatr Suppl. 1999;432 :13 –14
Fisher DA, Dussault JH, Foley TP, et al. Screening for congenital hypothyroidism: results of screening one million North American infants. J Pediatr. 1979;94 :700 –705
Hanna CE, Krainz PL, Skeels MR, Miyahira RS, Sesser DE, LaFranchi SH. Detection of congenital hypopituitary hypothyroidism: ten-year experience in the Northwest Regional Screening Program. J Pediatr. 1986;109 :959 –964
Fisher DA. Second International Conference on Neonatal Screening: progress report. J Pediatr. 1983;102 :653 –654
Verkerk PH. 20-year national screening for phenylketonuria in the Netherlands: National Guidance Commission PKU [in Dutch]. Ned Tijdschr Geneeskd. 1995;139 :2302 –2305
Van der Kamp HJ, Noordam K, Elvers B, van Baarle M, Otten BJ, Verkerk PH. Newborn screening for congenital adrenal hyperplasia in the Netherlands. Pediatrics. 2001;108 :1320 –1324
Thompson R, Seargeant L, Winter JS. Screening for congenital adrenal hyperplasia: distribution of 17-hydroxyprogesterone concentrations in neonatal blood spot specimens. J Pediatr. 1989;114 :400 –404(Caren I. Lanting, MD, PhD)
Department of Pediatric Endocrinology, Emma Children's Hospital, Academic Medical Center, University of Amsterdam, Netherlands
National Institute for Public Health, Bilthoven, Netherlands
ABSTRACT
Context. Since the introduction of screening for congenital hypothyroidism (CH) in 1974, the optimal laboratory strategy has been the subject of debate.
Objective. To assess the clinical effectiveness and cost-effectiveness of various types of thyroxine (T4)-based strategies to screen for CH.
Design, Setting, and Participants. In the Netherlands, since January 1, 1995, a primary T4 determination with supplemental thyroid-stimulating hormone (TSH) and T4-binding globulin (TBG) measurements has been used. Results were calculated from cumulative findings for 1181079 children screened between January 1, 1995, and December 31, 2000.
Main Outcome Measures. Rates of detection of patients with CH of thyroidal origin (CH-T) or CH of central origin (CH-C), false-positive rates, laboratory costs, and costs of initial diagnostic evaluations.
Results. All known infants (n = 393) with CH-T and 92% (n = 66) of infants with CH-C were detected on the basis of low T4 levels, TSH elevation, and/or low T4/TBG ratios. If the decision to refer had been based solely on TSH elevation, then 94% of patients with CH-T and none of the patients with CH-C would have been detected. If low T4 levels (–3.0 SD) and TSH elevation had been used as the criteria for referral, then the rates of detection would have been 96% for CH-T and 31% for CH-C. The false-positive rates for the 3 approaches were 0.5, 3.3, and 4.7 cases per case detected, respectively. The introduction of the T4/TBG ratio into a program using a primary T4 with supplemental TSH approach generates an extra cost of $11206 per additional case detected. The average costs to detect 1 patient are comparable for the 3 approaches. In addition, our data revealed a substantially greater prevalence of CH-C than reported previously (1 case per 16404 children, compared with earlier estimates of 1 case per 26000 infants to 1 case per 29000 infants).
Conclusions. The T4 plus TSH plus TBG approach is a recommendable strategy for neonatal CH screening. It offers outstanding detection of patients with CH-C, in addition to those with CH-T, with acceptable costs.
Key Words: average cost cost-effectiveness congenital hypothyroidism laboratory approach marginal cost screening thyroid-stimulating hormone thyroxine thyroxine-binding globulin
Abbreviations: CH, congenital hypothyroidism TSH, thyroid-stimulating hormone TBG, thyroxine-binding globulin T4, thyroxine CH-C, congenital hypothyroidism of central origin CH-T, congenital hypothyroidism of thyroidal origin
Congenital hypothyroidism (CH) refers to a heterogeneous group of disorders that have in common a lack of thyroid hormone already present in the neonatal period. Because thyroid hormone is essential for brain development, the major risk of CH, if left untreated, is irreversible cognitive and motor impairment.1 Although CH was once a major cause of mental retardation, almost all developed countries have now instituted a neonatal mass screening program to allow for early identification and treatment of affected children.
Since the introduction of screening for CH in 1974, the optimal laboratory strategy has been subject of debate.2 Japan, Australia, most European countries, and many North American states use an approach in which only thyroid-stimulating hormone (TSH) is used to discriminate between affected and unaffected infants. In this way, "classic" CH cases with elevated TSH concentrations would be detected with high specificity. However, mild cases of CH of thyroidal origin (CH-T) with only slightly elevated TSH levels and/or delayed TSH elevation would remain undetected, as would cases of CH of central origin (CH-C). Another strategy is to first measure thyroxine (T4) in all samples, followed by TSH measurement for samples with low T4 values. Several North American states use this strategy.
In the Netherlands, the latter approach was extended with the determination of T4-binding globulin (TBG) levels for the lowest 5% of T4 values. The T4/TBG ratio serves as an indirect measure of the free T4 concentration (which cannot be determined directly in dried blood spots). In contrast to most screening programs, in which TSH levels are determined for the lowest 10% of T4 readings, TSH levels are measured for the lowest 20% of T4 values. In this way, the Dutch screening program provides unique information about the prevalence of CH-C and mild cases of CH-T. Also, data derived from the program may allow calculation of the outcomes of other, less-sensitive strategies. This information may be used for evaluation of concurrent screening programs, as well as our own. In this article, we report on the clinical effectiveness and cost-effectiveness of the Dutch T4 plus TSH plus TBG screening strategy, in comparison with 2 alternative, primary T4 strategies.
METHODS
Screening Procedures
In the Netherlands, nationwide neonatal screening for CH, based on primary T4 and supplemental TSH measurements, started in 1981. In 1995, after a 1-year pilot period, TBG measurement was added to the screening procedure. Blood samples are collected on filter paper through heel puncture. Until the middle of 1999, blood was sampled at day 5 to day 7 after birth. From that time onward, this was advanced to day 4 to day 7 after birth. T4 concentrations are normalized for each assay run and are expressed as SDs relative to the daily mean. If the T4 concentration is among the lowest 20% of values for the series for that day (ie, –0.8 SD), then TSH (in microunits per milliliter of serum) is measured in the same sample. If the T4 concentration is among the lowest 5% (ie, –1.6 SD), then TBG (in nanomoles per liter; 1 μmol/L = 54 μg/dL) is also measured. The T4/TBG ratio is calculated as follows: [T4 (in SD) plus 5.1]/[TBG (in micromoles per liter of serum; 1 μmol/L = 54 μg/dL)], where the addition of 5.1 serves to keep ratios in the positive range. Screening results are kept by the Dutch Health Administrations. Their medical advisors decide whether a second heel puncture or referral is indicated. Medical evaluation of patients is performed by general pediatricians. To prevent delayed diagnosis of the most severe cases, immediate referral occurs if the T4 level is –3.0 SD and/or the TSH level is 50 μIU/mL serum. A second heel-stick blood sample is requested in cases with a borderline T4 concentration (–2.9 SD T4 –1.6 SD) in combination with a low T4/TBG ratio (8.5) and/or a borderline TSH concentration (20 μIU/mL TSH < 50 μIU/mL). If, in a second heel puncture, the T4 level is –1.6 SD in combination with a low T4/TBG ratio and/or the TSH level is 20 μIU/mL, then the infant is referred to a general pediatrician. Newborns with a birth body mass of 2500 g and a gestational age of 36 weeks are referred exclusively on the basis of their TSH concentrations.
Data and Data Sources
The Netherlands Organization of Applied Scientific Research Prevention and Health documents the screening results and diagnostic findings for all children screened for CH in the Netherlands, for the purposes of monitoring and quality improvement of the screening program. For this, information is gathered from all Dutch Health Administrations, which are responsible for the execution of the neonatal screening program for CH, and from pediatricians. Screening outcomes are kept and, for each referral, screening and diagnostic T4, TSH, and TBG concentrations are recorded. The pediatrician's diagnosis is recorded as no CH, transient CH, CH-T, or CH-C. Permanence of hypothyroidism is assumed if the pediatrician reports that an ectopic/dystopic gland or absent thyroid tissue was revealed. For all other cases of CH, the permanence of the hypothyroidism is assessed when the child reaches the age of 4 years. To attempt to identify false-negative cases, the Netherlands Organization of Applied Scientific Research Prevention and Health stays in close contact with the Dutch Health Administrators and pediatricians, mainly through the attendance of advisory board meetings on all aspects of neonatal screening and correspondence regarding screening and diagnostic outcomes. For the current analyses, we used data for the period from January 1, 1995 (the day TBG measurement was incorporated into the national screening program), to December 31, 2000.
Cost-Effectiveness Analysis
In this study, we set out to compare the results of the Dutch T4 plus TSH plus TBG approach with 2 alternative, primary T4 with supplemental TSH approaches. The first is the (T4) plus TSH approach in which T4 readings are not reported and are used only to determine whether TSH measurements are indicated. More specifically, referral is indicated for all infants with TSH concentrations of 50 μIU/mL and for infants with TSH readings between 20 and 50 μIU/mL if these findings are confirmed with a repeat heel-puncture blood sample.
The second approach is the T4 plus TSH approach, in which both T4 and TSH readings are reported and are used to determine whether referral is indicated. Referral is indicated for all subjects with T4 levels of –3.0 SD and/or TSH levels of 50 μIU/mL and for subjects with TSH readings between 20 and 50 μIU/mL if these findings are confirmed with a repeat heel-puncture blood sample.
First, we calculated the detection, second heel puncture, and false-positive rates for the (T4) plus TSH and T4 plus TSH approaches from the frequency distributions of screening T4 and TSH concentrations obtained from the Dutch neonatal CH screening program. In accordance with the T4 plus TSH plus TBG approach, we used a cutoff point for TSH measurement of 20%. The next step involved linkage of effectiveness and effort. Effectiveness was defined as the number of cases of permanent CH-T and CH-C detected, with the pediatrician's diagnosis being taken as the definite diagnosis. Effort was defined as the second heel puncture rate, false-positive rate, laboratory costs, and costs of the initial diagnostic evaluations (in US dollars; at the time of these calculations, 1 US dollar = 1 euro). In the Netherlands, laboratory costs are $1.25 for T4 measurements, $3.41 for TSH measurements, and $8.64 for TBG measurements. The sampling and laboratory costs of a second heel puncture average $11.50.3 Cost of wages and salaries for laboratory personnel, housing, and laboratory equipment were assumed to be constant for each of the 3 approaches and therefore were not taken into account. The cost of initial diagnostic evaluations for infants with abnormal screening results was estimated at $1123 (including costs of 1 consultation with a general practitioner, 1 pediatric consultation, and the initial diagnostic laboratory testing). Cost-effectiveness is presented as the average cost and the marginal incremental cost per case detected. Marginal cost refers to the extra cost required for each additional case that is detected.4,5
RESULTS
Population
Between January 1, 1995, and December 31, 2000, 1185670 children were born in the Netherlands. Of these infants, 1181079 (99.6%) were screened for CH. On the basis of their abnormal screening results, 2604 (0.2%) were referred to a general pediatrician. Of these, 459 infants (18%) were diagnosed with permanent CH, 393 (86%) of whom had CH-T and 66 (14%) of whom had CH-C. Of the infants with CH-T, 41% were male. Their mean ± SD gestational age was 39.5 ± 2.4 weeks, with a mean ± SD birth body mass of 3368 ± 716 g. For infants with CH-C, these figures were 72%, 38.9 ± 2.8 weeks, and 3244 ± 648 g, respectively.
The screening program failed to detect 6 patients diagnosed as having CH-C; no missed cases of CH-T came to our attention (see Table 1 for screening results). From this, it follows that the sensitivity of the screening was 98.5% (459 of 465 patients with CH showed abnormal screening results). The specificity was 99.9% (1180004 of 1180463 children without CH tested negative), and the positive predictive value was 18% (459 cases of permanent CH were found among 2604 referrals).
our data, we calculated that the prevalence of permanent CH-C was 1 case per 16404 live-born infants (95% confidence interval: 1 case per 13174 infants to 1 case per 21173 infants), which is clearly higher than earlier reports indicated (see Discussion). The prevalence of permanent CH-T was found to be 1 case per 3017 live-born infants (95% confidence interval: 1 case per 2721 infants to 1 case per 3318 infants). Table 2 summarizes the screening data and additional clinical results for infants diagnosed as having CH-T or CH-C.
Detection Rates
Second Heel Puncture Rates
The T4 plus TSH plus TBG approach gave rise to 5310 second heel punctures during the 6-year study period. For the (T4) plus TSH and T4 plus TSH approaches, this would have been 969. These findings indicate second heel puncture rates of 11.6 punctures per case detected for the T4 plus TSH plus TBG approach, 2.4 punctures per case detected for the T4 plus TSH approach, and 2.6 punctures per case detected for the (T4) plus TSH approach (Table 3).
False-Positive Rates
During the 6-year study period, the T4 plus TSH plus TBG approach yielded 2604 referrals. For the T4 plus TSH and (T4) plus TSH approaches, the numbers of referrals would have been 1702 and 572, respectively. The false-positive rates for the 3 approaches were 4.7 cases per case detected for the T4 plus TSH plus TBG approach, 3.3 cases for the T4 plus TSH approach, and 0.5 case for the (T4) plus TSH approach (Table 3).
Cost-Effectiveness
DISCUSSION
Since 1995, the Dutch neonatal screening program for CH has been based on a T4 plus TSH plus TBG laboratory approach. We studied the cost-effectiveness of this approach in comparison with 2 other approaches, which were based on T4 and TSH determinations but not TBG measurements. Our main finding is that incorporation of TBG measurements into a T4 plus TSH approach leads to a threefold increase in the rate of detected cases of permanent CH-C.
Introduction of TBG measurements into a primary T4 with supplemental TSH approach can be undertaken with acceptable costs. Because almost all developed countries have already instituted CH screening programs, an important question addresses the extra cost of one approach over the cost of another. We therefore calculated the marginal incremental cost. Incorporation of TBG measurements into a T4 plus TSH approach generated an extra cost of $11206 for each additional patient with permanent CH identified (Table 4). Because the addition of TBG measurements resulted in greater numbers of patients detected, the average cost to detect 1 patient differed only slightly for the 3 approaches [$6353, $6209, and $6851 for the (T4) plus TSH, T4 plus TSH, and T4 plus TSH plus TBG approaches, respectively]. Patients with CH-C benefit most from inclusion of TBG measurements. In the Netherlands, from 1995 through 2000, 91.6% of known cases with permanent CH-C were detected by means of the T4 plus TSH plus TBG approach. The T4 plus TSH approach would have yielded a detection rate of only 30.6% for CH-C (Table 3). Although there is no estimate of the lifetime health care costs and costs of loss of productivity associated with non-timely treated CH-C, it can be expected that these costs far outweigh the additional screening costs.
The major advantage of the T4 plus TSH plus TBG approach is its excellent rate of detection of CH-C cases. Although the figures initially seem less impressive than those for CH-C, it should be noted that introduction of TBG measurements into the screening program for CH also leads to improved identification of patients with CH-T. From Table 3, it can be read that changing from a T4 plus TSH approach to a T4 plus TSH plus TBG approach would lead to a 3.8% increase in the detection rate for CH-T, over an already very high detection rate of 96.2% for CH-T. Although these patients are likely to have milder forms of CH-T, they do deserve medical attention, because of the disadvantageous effect that even slight hypothyroidism may have on brain development.6
Regarding the effectiveness of the different screening strategies, first we point out that we did not take into account the number of patients with transient CH, although many of these infants did need (temporary) T4 supplementation and therefore had true positive screening results. The main reason for leaving these cases out of the analyses is the occurrence of great variation in the severity and duration of hypothyroidism in this group. Also, in a substantial number of cases, the cause of the hypothyroidism lies outside the neonate's thyroid gland and hypothalamic-hypophyseal system (for example, transient thyroidal hypothyroidism may be attributable to abundant postnatal use of iodide-containing disinfectants, maternal use of antithyroid or iodide-containing drugs, or the presence of maternal antithyroid antibodies, and transient central hypothyroidism may be attributable to inadequately treated maternal Graves' disease7). Therefore, transient CH is an entity that is usually not included in CH statistics.
Second, it should be noted that we took the pediatrician's diagnosis as the final diagnosis. In some cases, it may be difficult to establish whether mild permanent or transient CH is present.8 Also, CH is a very heterogeneous disease with respect to the underlying cause.1 It is therefore not surprising that reassessment in some cases results in revision of the diagnosis. Even reexamination during the first year of life, in our experience, leads to a change of cause in some cases. However, this situation does not differ from that in other parts of the world. One advantage of our study is that we used the revised diagnoses established at 4 years of age whenever possible. Also, we calculated the outcomes of the 3 laboratory approaches by using the same study group. It is therefore unlikely that this undermines our conclusion.
A less attractive feature of the T4 plus TSH plus TBG approach, compared with the T4 plus TSH and (T4) plus TSH approaches, is its relatively high false-positive rate. This leads not only to higher laboratory costs and diagnostic evaluation costs but also to more parental anxiety.9 Theoretically, the number of false-positive test results for the T4 plus TSH plus TBG approach could be decreased if the T4/TBG ratio was measured not only for the group of infants with borderline T4 levels (–3.0 SD < T4 < –1.5 SD) but also for the group with low T4 concentrations (–3 SD); however, such an approach would lead to a delay of 1 to 3 days before an infant could be referred to a pediatrician, because of the increased workload for the laboratory. Because a group of patients with potentially severe CH is involved, such a delay is considered unacceptable.
Several North American states have considered switching from a (T4) plus TSH approach to a primary TSH strategy (ie, without initial T4 measurements). From our data, we can estimate the cost-effectiveness of such a change. The detection rate and the rates of second heel punctures and false-positive screening results are comparable for the 2 strategies (Table 3). However, because T4 measurements are inexpensive, in comparison with TSH measurements, and because TSH is measured for all screened infants and not just a preselected group with decreased T4 concentrations, the laboratory costs would increase sharply. This is especially important because laboratory costs make up a substantial part of the total cost of maintaining a screening program for CH. We estimated that the total laboratory costs would be nearly twice as high for a program using a primary TSH approach, compared with the (T4) plus TSH approach (ie, $4027479 vs $2292990) (Table 4). Similarly, the average laboratory cost per case detected is high (ie, $10196) when the primary TSH approach is used, compared with the (T4) plus TSH approach ($6353).
The need for early identification of patients with CH-C is not recognized generally.10 One commonly used argument against screening for CH-C is that it is usually mild, in terms of T4 concentrations. In our study, however, 39% of patients with CH-C had a screening T4 value 3.0 SD below the mean (Table 2). Also, it is known that three quarters of children with CH-C are found to have multiple pituitary hormone deficiencies, which is a potentially life-threatening condition.1 Delayed detection and incomplete diagnosis of the pituitary hormone deficiencies result in significant morbidity, such as severe hypoglycemia and neonatal hepatitis, and a mortality rate as high as 14%.1 Because all deficient hormones can be supplemented readily, timely diagnosis improves the outcomes of patients with CH-C significantly.
Another frequently heard argument against screening for CH-C is that its prevalence is too low to make screening worthwhile. Initial estimates of the prevalence of CH-C among live-born children in the United States and Canada ranged from 1 case per 110000 infants to 1 case per 68200 infants.11,12 Evaluation of the first 10-year screening experience in the northwest region of the United States in 1986 resulted in a prevalence estimate of 1 case per 29000 infants.13 Less than one half of these patients were detected with neonatal screening.13 A similar retrospective study of the 1981–1982 screening population in the Netherlands yielded a prevalence of 1 case per 26000 infants for CH-C.1 In the present study, 66 infants with CH-C were detected in a 6-year period. During this period, 6 additional cases (9%) were found on the basis of clinical symptoms. This adds up to a prevalence of 1 case per 16404 live newborns (95% confidence interval: 1 case per 13174 infants to 1 case per 21173 infants). Therefore, it seems that CH-C is a more commonly encountered disorder than thought previously. The prevalence of CH-C is on the same order as that of, for example, phenylketonuria (1 case per 18000 infants),14 a disease for which most developed countries have instituted a screening program, or congenital adrenal hyperplasia (1 case per 11764 infants).15 Because it is unlikely that the prevalence of CH-C in the Netherlands is significantly different from that in North America, we presume that in previous studies both the mildest and most severe cases were missed, similar to the situation for congenital adrenal hyperplasia in the prescreening era.16
CONCLUSIONS
The T4 plus TSH plus TBG approach is a recommendable strategy for neonatal CH screening. The main advantage is its outstanding ability to identify patients with CH-C. Early detection of CH-C prevents severe morbidity and death, because it is often associated with other pituitary hormone deficiencies. In addition, CH-C seems to be a more common disorder than assumed generally, which makes early detection even more worthwhile. Introduction of TBG measurements into a primary T4 with supplemental TSH approach can be done with costs that seem fairly acceptable, especially when considered against the anticipated long-term health care costs and the costs of loss of productivity for infants whose disease is not detected early.
ACKNOWLEDGMENTS
We thank M. Elske van den Akker van Marle, PhD, health economist, for valuable advice and review of the manuscript.
FOOTNOTES
Accepted Nov 16, 2004.
No conflict of interest declared.
REFERENCES
Vulsma T, de Vijlder JJM. Thyroid disease in newborns, infants, and children. In: Wass JAH, Shalet SM, eds. Oxford Textbook of Endocrinology and Diabetes. Oxford, United Kingdom: Oxford University Press; 2002: 532–544
Dussault JH. The anecdotical history of screening for congenital hypothyroidism. J Clin Endocrinol Metab. 1999;84 :4332 –4334
Oostenbrink JB, Koopmanschap MA, Rutten FFH. Handleiding voor Kostenonderzoek, Methoden en Richtlijnen voor Economische Evaluaties in de Gezondheidszorg [in Dutch]. Amstelveen, Netherlands: College voor Zorgverzekeringen; 2000
Neuhauser D, Lewicki AM. What do we gain from the sixth stool guaiac N Engl J Med. 1975;293 :226 –228
Drummond MF, O'Brien B, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. 2nd ed. Oxford, United Kingdom: Oxford University Press; 1997
Léger J, Larroque J, Norton J. Influence of severity of congenital hypothyroidism and adequacy of treatment on school achievement in young adolescents: a population-based cohort study. Acta Paediatr. 2001;90 :1249 –1256
Kempers MJ, van Tijn DA, van Trotsenburg AS, de Vijlder JJ, Wiedijk BM, Vulsma T. Central congenital hypothyroidism due to gestational hyperthyroidism: detection where prevention failed. J Clin Endocrinol Metab. 2003;88 :5851 –5857
Moreno JC, Bikker H, Kempers MJ, et al. Inactivating mutations in the gene for thyroid oxidase 2 (THOX2) and congenital hypothyroidism. N Engl J Med. 2002;347 :95 –102
Tymstra T. False positive results in screening tests: experiences of parents of children screened for congenital hypothyroidism. Fam Pract. 1986;3 :92 –96
Toublanc JE. Guidelines for neonatal screening programs for congenital hypothyroidism. Acta Paediatr Suppl. 1999;432 :13 –14
Fisher DA, Dussault JH, Foley TP, et al. Screening for congenital hypothyroidism: results of screening one million North American infants. J Pediatr. 1979;94 :700 –705
Hanna CE, Krainz PL, Skeels MR, Miyahira RS, Sesser DE, LaFranchi SH. Detection of congenital hypopituitary hypothyroidism: ten-year experience in the Northwest Regional Screening Program. J Pediatr. 1986;109 :959 –964
Fisher DA. Second International Conference on Neonatal Screening: progress report. J Pediatr. 1983;102 :653 –654
Verkerk PH. 20-year national screening for phenylketonuria in the Netherlands: National Guidance Commission PKU [in Dutch]. Ned Tijdschr Geneeskd. 1995;139 :2302 –2305
Van der Kamp HJ, Noordam K, Elvers B, van Baarle M, Otten BJ, Verkerk PH. Newborn screening for congenital adrenal hyperplasia in the Netherlands. Pediatrics. 2001;108 :1320 –1324
Thompson R, Seargeant L, Winter JS. Screening for congenital adrenal hyperplasia: distribution of 17-hydroxyprogesterone concentrations in neonatal blood spot specimens. J Pediatr. 1989;114 :400 –404(Caren I. Lanting, MD, PhD)