Objective. To evaluate whether congenital adrenal hyperplasia (CAH) patients can be detected by newborn screening before the occurrence of life-threatening salt wasting and whether the prevalence, specificity, and sensitivity are adequate enough for a routine screening procedure.
Design. From 1998, a 2-year regional pilot screening for CAH was performed. In 1998, cutoff levels for 17OHP were primarily based on birth weight, and in 1999 on gestational age. In addition, nationwide, all newly diagnosed patients with CAH were reported to the Dutch Pediatric Surveillance Unit to compare screened CAH patients with CAH patients in the area without screening.
Results. In 2 years, 176 684 newborns were screened and 15 CAH patients (7 males/8 females) were detected. Therapy was started at the median age of 7 days. In the area without screening, 223 307 infants were born and 19 CAH patients (10 males/9 females) were reported to the Dutch Pediatric Surveillance Unit. Therapy was started at the median age of 14 days. The mean (standard deviation) serum sodium concentration was 134.5 (3.4) mmol/L in the area of screening versus 124.5 (10.8) mmol/L in the area without screening. The overall prevalence was 1:11 764. In 1998 and 1999, the specificity was 99.76% and 99.97%, respectively. The positive predictive value was 4.5% and 16%, respectively. To date, no false-negative cases have been detected.
Conclusion. Severe salt wasting can be prevented by neonatal screening. The prevalence, specificity, and sensitivity allowed addition of screening for CAH to the routinely performed national neonatal screening program.
Congenital adrenal hyperplasia (CAH) is, in >90% of the patients, attributable to a 21-hydroxylation defect in the adrenal cortex. Biochemically, this results in low serum concentrations of aldosterone and cortisol and elevated 17-hydroxyprogesterone (17OHP) and androstenedione concentrations. Clinically, early onset classical CAH consists of 2 variants; the salt-wasting (SW) form (75%) and the nonsalt-wasting (NSW) form (25%).1 In affected individuals, SW predisposes to “adrenal crises” with dehydration, which becomes life-threatening in the first 2 to 3 weeks of life. In NSW, no electrolyte disturbances are found. Female newborns with classical CAH are virilized at birth, ranging from slight clitoromegaly to complete masculinisation. Male infants have no physical signs at birth; therefore, they are particularly at risk for dehydration and death.
Newborns with CAH can be detected by screening.1 The elevated 17α-hydroxyprogesterone level in 21-hydroxylase deficiency is used in screening to indicate newborns at risk for having CAH.
To evaluate whether CAH patients can be detected by newborn screening before the occurrence of life-threatening SW and whether the prevalence, sensitivity, and the specificity were high enough for a national implementation, a regional pilot study was performed in the Netherlands.
PARTICIPANTS AND METHODS
Annually, approximately 200 000 infants are born in the Netherlands; 35% of the deliveries take place at home. In 1998, in 4 out of 12 provinces, in the middle and southwest of the Netherlands, a 2-year pilot screening for CAH was added to the routinely performed newborn screening program for phenylketonuria and congenital hypothyroidism. This area covered 44% of the Dutch newborn population.
Data Collection in Area of Screening
Heel-puncture blood samples were collected on filter paper Schleicher and Schull #2992 (Dassel, Germany) in term and preterm infants on day 5 to 7 after birth, the day of birth being day 0. In June 1999, the day of sampling was advanced to day 4 to 7. The blood sample absorbed on filter paper was sent by mail to the regional laboratory in a preprinted and prepaid envelope. Time for postal services was, on average, 2 days. 17OHP was determined on working days and the results were known at the end of the day. Abnormal results in CAH were directly reported by the laboratories involved, to the coordinator assigned to this pilot screening (H.J.vdK). The coordinator traced the suspected newborn and decided whether the child should have a second heel puncture or had to be referred to a pediatric endocrinologist immediately. This decision was made on the basis of the information from the parents about the condition (weight, alertness, and ambiguity) of the child and on the degree of elevation of the 17OHP concentration.
As presented in Table 1, in 1998, cutoff levels were primarily based on birth weight, because birth weight was always reported on the filter paper forms, whereas reporting gestational age on filter paper forms was only obligatory when birth weight was ≤2500 g. So, when birth weight was below 2500 g, gestational age could be used to adjust cutoff levels. In 1999, it became obligatory to report birth weight and gestational age on the filter paper forms. From this time forward, we were able to establish cutoff levels based on gestational age. Birth weight was measured by the midwife or gynecologist involved. Gestational age was assessed as the time from the first day of the last menstrual period until birth. It was calculated using anamnestic information and cross-checked by the midwife or gynecologist, who estimated gestational age through physical investigation or by ultrasound.
Data Collection by the Dutch Pediatric Surveillance Unit (DPSU)
This unit was founded in 1992 under the auspices of the Pediatric Association of the Netherlands, following the example of the British Pediatric Surveillance Unit.2 Since 1998, all pediatricians have been requested to report monthly to the DPSU, whether they had diagnosed or suspected congenital adrenal hyperplasia by 21-hydroxylase deficiency in infancy or childhood. A questionnaire was sent each month to the pediatrician involved. He or she was asked to provide data concerning age on admission, sex, symptoms at presentation, physical examination, laboratory investigations (including DNA analysis), and age at start of therapy. Registration by the DPSU was performed nationwide to compare the clinical condition of screened CAH patients with CAH patients in the area without screening and to detect false-negative cases of the screening.
In the area without screening, informed consent was obtained from the parents of the 19 CAH patients to determine 17OHP retrospectively in heel-puncture samples, which were collected in the first week of life for screening on phenylketonuria and congenital hypothyroidism.
Prematurity was defined as a gestational age <37 weeks. The diagnosis of CAH was confirmed by an elevated serum 17OHP, or an abnormal urine steroid profile (gas chromatography-mass spectrometry), and in most cases in the area of screening by defects in CYP21 on chromosome 6 by DNA analysis.3 Patients were defined as having SW when they had a serum sodium concentration below 130 mmol/L or a plasma renin activity >50 ng/mL/hour (>14 ng/[L/s]). Patients in the area of screening were detected before SW was apparent. Patients with homozygosity for a deletion or large conversion of the CYP21 or with compound heterozygoty for Arg357Trp, E3del8bp, Gln310stop, or Intron 2-splice in DNA analysis were also recorded as SW.3
For the determination of the 17OHP concentration in dried filter paper, the AutoDELFIA Neonatal 17OHP assay (Perkin Elmer Life Sciences, previously; Wallac, Finland) was used. The interassay variance was maximally 10% and the intra-assay variance was maximally 8.5%. Results are expressed as nmol/L (μg/L) serum (1 nmol/L [0.33 μg/L] blood = 2.0 nmol/L [0.66 μg/L] serum). Heel-puncture samples were assayed in duplicate.
Differences between groups were analyzed by the Mann-Whitney test (2-tailed).
In 2 years, 176 684 newborns were screened for CAH and 15 patients (7 males/8 females) were detected by screening (Table 2). Median (range) birth weight was 3410 (2600–4055) g, median (range) gestational age was 39.4 (37.3–42.2) weeks. The specificity was 99.76% in 1998 and increased to 99.97% in 1999 because of adjustment of cutoff levels. The positive predictive values (PPV) in 1998 and 1999 were 4.5% and 16%, respectively. To date, no false-negative cases have been detected.
In 1998, 148 of a total of 214 false-positives were prematures large for gestational age (Table 3). After the introduction of primarily gestational age-related cutoff limits, the number of false-positive prematures decreased to 10 out of a total of 26 false-positives. Of the term newborns with false-positive results, 45% was hospitalized for a general nonendocrine illness in 1998 compared with 88% in 1999.
Of the approximately 430 Dutch pediatricians who were contacted by the DPSU, 92% returned the monthly report card.4 In 2 years, CAH was reported 34 times. In the area of screening, the 15 CAH patients detected were all reported to the DPSU. In the area without screening, 223 307 infants were born and 19 CAH patients (10 males/9 females) were reported to the DPSU. Median (range) birth weight was 3620 (1340–4880) g, median (range) gestational age was 40.3 (30.3–42.3) weeks. Two prematures (30.3 and 33.3 weeks) were reported. The first premature CAH patient was prenatally diagnosed and treatment started soon after birth. The 17OHP level of the second premature was 912 nmol/L.
Comparison Between CAH Patients in the Area With and Without Screening
From January 1998 until January 2000, 34 CAH patients (17 males/17 females) were born in the Netherlands. The prevalence was 1:11 764 (95% confidence interval: 1:8507–1:16 835). The SW form of CAH was present in 30 out of 34 cases; 4 cases had the NSW form (2 in the screening area and 2 in the area without screening).
In the area of screening, CAH was clinically not suspected in 6 out of 7 males and in 3 out of 8 females (Prader stage 1, 2, and 3).5 NSW was apparent in a boy and a girl (Prader stage 1) diagnosed by screening at day 6 and 8 with a serum sodium of 136 and 135 mmol/L, respectively. In the area without screening, all 10 males and 2 out of 9 females (Prader stage 2 and 5) were clinically not suspected at birth. Sex misassignment was reported in the child with Prader stage 5. NSW was apparent in 2 girls, diagnosed at day 11 and 115, both with a serum sodium of 137 mmol/L and with Prader stage 4 and 2, respectively.
The comparison between the screening area and the nonscreening area in day of heel puncture, day of admission, day at start of therapy, electrolytes, and 17OHP in heel-puncture blood is presented in Table 4. In the area of screening, patients were diagnosed before severe SW appeared, treatment could start 7 days earlier, and hospitalization of males was shortened by 6 days compared with the area without screening.
In Table 5, the median (range) data of CAH patients who were clinically not suspected are presented. The difference between the 2 areas was more pronounced; in the area without screening, the median serum sodium concentration in the 10 males was 117 mmol/L (normal value: 135–145 mmol/L) and treatment started 10 days later in males compared with the area with screening.
Mean (standard deviation) total weight loss (difference between birth weight and the lowest weight ever measured) was 7.7 (1.4) percent in the patients with a serum sodium ≥135 mmol/L at admission and 10.2 (3.0) percent in the patients with a serum sodium <135 mmol/L at admission (P < .01).
Newborn screening for CAH was feasible in the pilot area. Despite screening at the end of the first week of life, all patients were admitted before day 10 and severe SW was prevented. Male patients were primarily detected by screening, as were 3 out of 8 female infants. In the area without screening, most male patients were admitted with severe hyponatraemia. Failure to thrive was present in all patients, even if the serum sodium concentration was above 135 mmol/L.
A drawback of neonatal screening is that parents may be unnecessarily alarmed in case of false-positives. The number of false-positives has to be as low as possible without loss of sensitivity. In 1998, the false-positive rate was 0.24%. This was caused mainly (70%) by preterms attributable to the use of birth weight-related cutoff levels. The use of a birth weight of 3000 g to establish cutoff levels instead of 2500 g would result in a reduction of the number of premature false-positives (Table 4) from 0.24% to 0.07%. However, with a cutoff level of 3000 g, the small-for-gestational-age term neonates are at risk for false-negative results, and the false-positive rate is still higher than the 0.03% in gestational age-related cutoff levels, as was found in 1999. That gestational age is a better predictor of 17OHP in newborns than birth weight is was also concluded from a separate study (personal data). In this study, regression analysis and calculation by the LMS-method showed that gestational age adjusted cutoff levels would result in a greater sensitivity.
The results of screening in other European countries are summarized in Table 6. The 95% confidence intervals are also presented. The prevalence of 1:11 764 as found in the Netherlands, was similar to the prevalence found in most other countries. In Belgium, where screening covered the province of Antwerp, the prevalence was lower. The low PPVs that were found in most countries were attributable to the high false-positive rates (0.2%–0.9%).6,7,8 In Switzerland and Sweden, cutoff levels are only based on gestational age. The higher false-positive rate in Switzerland might be explained by the use of the 97th percentiles reference line of the screening results.9 In Sweden, the cutoff levels were set at a higher level at the cost of a lower sensitivity.10
As far as we know, this is the first prospective study reported where an area with screening was compared with an area without screening. In a Swedish retrospective study, data from patients diagnosed between 1969 and 1986 were compared with the results of neonatal screening in 1989 to 1994.11 Their average prevalence before screening was 1:11 500, but more girls were reported (M:F = 1:1.2), indicating that a number of boys probably had died undiagnosed. After the introduction of screening, the prevalence was 1:9800 with a sex ratio of 1. The severity of SW before screening was similar to that in our study in the area without screening, despite of the increase in health care and acquaintance of CAH in the last 2 decades. In an American retrospective cohort study, the results of screening in Texas were related to the unscreened population of Arkansas and Oklahoma.12 The incidence of CAH was similar in these States (1:15 974 vs 1:17 396). The median age of diagnosis in males was 14 days lower in the screening area (day 12 versus day 26), but no data were reported about the severity of SW in both groups.
The costs of the CAH screening were 45 000 Euro per CAH patient detected. This is comparable to other screening programs in the Netherlands.13,14 The costs of CH screening (prevalence 1:3 000) and PKU screening (prevalence 1:18 000) are 22 000 Euro and 47 000 Euro per child detected. Cost-effectiveness in terms of morbidity in CAH is not easy to calculate. Analysis of reduction of morbidity needs long-term follow-up to investigate whether prevention of SW will improve the outcome of patients with CAH. Minor handicaps (learning disabilities and behavioral problems) were reported in 2 retrospective studies in patients who suffered from SW; however, the severity of SW was not mentioned. Additional effects of the introduction of the neonatal screening were the increased awareness and knowledge of CAH by health workers, the achievement of a consensus for treatment, and the formation of a patients association. Nowadays, the early detection of CAH-patients by screening resulted in a reduction of hospitalization.
This study demonstrates that 1) not only males benefit from screening because CAH had not been recognized in 3 out of 8 female neonates, 2) newborns with CAH suffer from severe SW, which could be prevented by neonatal screening, and 3) a change in cutoff levels resulted in a high specificity and PPV without loss of sensitivity. Based on the results of this pilot study, newborn screening for CAH was implemented throughout the Netherlands as of July 1, 2000.
Financial assistance was given by “Zorg Onderzoek Nederland”(Health Research and Development Council).
We thank Dr J. M. Wit for his critical reading of the manuscript.
- Received November 28, 2000.
- Accepted June 22, 2001.
Reprint requests to (H.J.V.d.K.) Department of Pediatrics, J-6S, Leiden University Medical Center, Box 9600, 2300 RC Leiden, the Netherlands. E-mail:
- 4.↵Hirasing RA. Jaarverslag Nederlands Signalerings-Centrum Kindergeneeskunde. TNO-rapport PG/JGD/g8.050. TNO Prevention and Health Leiden. 1999
- 5.↵Prader A. Der Genitalbefund beim Pseudohermaphroditismus femininus des kongenitalen adrenogenitalen Syndroms. Helv Paediatr Acta.1954 ;3:231–248
- 6.↵Eyskens F. Screening for inborn errors of metabolism; the experience in the province of Antwerp [thesis]. Antwerp, Belgium: University of Antwerp; 1997
- ↵Balsamo A, Cacciari E, Piazzi S, et al. Congenital adrenal hyperplasia: neonatal mass screening compared with clinical diagnosis only in the Emilia-Romagna region of Italy, 1980–1995. Pediatrics.1996 ;98:362–367
- 9.↵Torresani T, Gruters A, Scherz R, Burckhardt JJ, Harras A, Zachmann M. Improving the efficacy of newborn screening for congenital adrenal hyperplasia by adjusting the cut-off level of 17α-hydroxyprogesterone to gestational age. Screen.1994 ;3:77–84
- 10.↵Larsson A, Thil’en A, Hagenfeldt L, von Dobeln U, Guthenberg C. Screening of half a million Swedish newborn infants for congenital adrenal hyperplasia. Screen.1992 ;1:159–166
- 11.↵Thilen A, Nordenstrom A, Hagenfeldt L, von Dobeln U, Guthenberg C, Larsson A. Benefits of neonatal screening for congenital adrenal hyperplasia (21-hydroxylase deficiency) in Sweden. Pediatrics.1998 ;101(4):. Available at: http://www.pediatrics.org/cgi/content/full/101/4/e11
- 14.↵Donaldson MD, Thomas PH, Love JG, Murray GD, McNinch AW, Savage DC. Presentation, acute illness, and learning difficulties in salt wasting 21-hydroxylase deficiency. Arch Dis Child.1994 ;70:214–218
- American Academy of Pediatrics