Screening for Down syndrome based on maternal age or fetal nuchal translucency: a randomized controlled trial in 39 572 pregnancies

Authors


Abstract

Objectives

Nuchal translucency (NT) screening increases antenatal detection of Down syndrome (DS) compared to maternal age-based screening. We wanted to determine if a change in policy for prenatal diagnosis would result in fewer babies born with DS.

Methods

A total of 39 572 pregnant women were randomized to a scan at 12–14 gestational weeks including NT screening for DS (12-week group) or to a scan at 15–20 weeks with screening for DS based on maternal age (18-week group). Fetal karyotyping was offered if risk according to NT was ≥ 1:250 in the 12-week group and if maternal age was ≥ 35 years in the 18-week group. Both policies included the offer of karyotyping in cases of fetal anomaly detected at any scan during pregnancy or when there was a history of fetal chromosomal anomaly. The number of babies born with DS and the number of invasive tests for fetal karyotyping were compared.

Results

Ten babies with DS were born alive with the 12-week policy vs. 16 with the 18-week policy (P = 0.25). More fetuses with DS were spontaneously lost or terminated in the 12-week group (45/19 796) than in the 18-week group (27/19 776; P = 0.04). All women except one with an antenatal diagnosis of DS at < 22 weeks terminated the pregnancy. For each case of DS detected at < 22 weeks in a living fetus there were 16 invasive tests in the 12-week group vs. 89 in the 18-week group. NT screening detected 71% of cases of DS for a 3.5% test-positive rate whereas maternal age had the potential of detecting 58% for a test-positive rate of 18%.

Conclusions

The number of newborns with DS differed less than expected between pregnancies that had been screened at 12–14 weeks' gestation by NT compared with those screened at 15–20 weeks by maternal age. One explanation could be that NT screening—because it is performed early in pregnancy—results in the detection and termination of many pregnancies with a fetus with DS that would have resulted in miscarriage without intervention, and also by many cases of DS being detected because of a fetal anomaly seen on an 18-week scan. The major advantage of the 12-week scan policy is that many fewer invasive tests for fetal karyotyping are needed per antenatally detected case of DS. Copyright © 2005 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

In many Western countries, women are offered one routine ultrasound scan during pregnancy, the purpose being to date the pregnancy and to detect multiple pregnancy and fetal malformations. Usually this scan is carried out in mid-gestation. Traditionally, pregnant women are counseled about their risk of carrying a fetus with Down syndrome (DS) on the basis of their age, and in many countries women are offered amniocentesis if they are 35 years or older. The rationale behind this is that the risk of giving birth to a baby with DS increases with maternal age1, 2. During the last decade a strong association between increased nuchal translucency (NT) at ultrasound examination at 11–14 gestational weeks and DS has been established3, 4. Therefore, in many countries a routine offer of an ultrasound scan at 11–14 weeks has been introduced in addition to a routine offer of a mid-trimester scan. The main purpose of the NT scan is to identify women at high risk of carrying a fetus with DS and to offer them fetal karyotyping.

Because the introduction of an additional routine scan is costly, we wondered if it would be possible to replace the policy of offering one routine scan in mid-gestation supplemented with maternal age-based screening for DS with another policy, namely the offer of one routine scan at 12 gestational weeks including NT screening for DS. To this end we designed a randomized controlled trial (RCT)—the NUPP-trial (NackUPPklarning, i.e. Swedish for nuchal translucency)—with the aim of comparing the medical, psychological and economical effects of the two policies outlined above. Only an RCT allows correct comparison of the effects on pregnancy outcome of two different policies of offering prenatal diagnosis, and to the best of our knowledge there are no published RCTs comparing the effects of different prenatal screening policies.

We report here on the number of babies born with DS, the number of pregnancy terminations for DS, and the number of invasive tests for fetal karyotyping associated with each of two screening policies, namely ultrasound screening at 12–14 weeks by NT or ultrasound screening at 15–20 weeks by maternal age. The prenatal detection rate of malformations other than chromosomal anomalies, the accuracy of pregnancy dating, and the psychological and economical aspects will be or have already been reported separately5, 6.

Methods

Study design

The NUPP trial was approved by the ethics committees at the Karolinska Institute, Stockholm, the Medical Faculties of Lund and Uppsala University. It was a multicenter RCT involving eight Swedish hospitals. Women were recruited between March 1999 and October 2002 from an unselected population of pregnant women cared for at the maternity care units affiliated to the hospitals involved. Eligibility criteria were: ability to understand the information about the trial, and gestational age at booking ≤ 13 + 2 weeks according to the last menstrual period. Women who agreed to participate were randomized either to have a routine scan at 12 or 18 gestational weeks. Second-trimester maternal serum screening was not a routine offer, but was infrequently performed at the request of the mother-to-be.

Patient information

After booking, written information about the study was sent to each study participant. Oral information was then given at the local maternity care unit. Informed consent was noted in the record.

Randomization

Randomization was performed at the ultrasound units using an internet-based software (Medscinet AB, Stockholm, Sweden). Demographic data, results of ultrasound examinations and invasive tests, and information on pregnancy outcome were registered in this database. Randomization was done block-wise and stratified for maternal age (< 35 or ≥ 35 years).

The screening procedures

The ultrasound examinations were scheduled at 12–14 weeks (12-week group) or at 15–20 weeks (18-week group). The examinations were performed by 46 midwives with median 11 (interquartile range, 5–17) years' experience of mid-gestation routine ultrasound examinations. Before the study started, all the midwives had qualified for a certificate of competence in the theory and practice of the 11–14-week scan including NT measurement from The Fetal Medicine Foundation (FMF), London, UK. During the trial all midwives performed both 12- and 18-week scans. The quality of our NT measurements was regularly checked by The FMF.

Twenty-six obstetricians—also holding the 11–14-week scan certificate of The FMF—were involved in the study. Their role was to confirm/refute any suspicion of fetal structural anomaly, and to provide counseling in cases of increased risk of chromosomal anomaly. The examinations were performed transabdominally using any of the following ultrasound systems: Aloka 1400, SSD 2000, 4000 or 5000 (Aloka Co. Ltd, Tokyo, Japan), Acuson XP 10, Aspen or Sequoia (Siemens Acuson Inc., Mountain View, CA, USA) with 3.5–5-MHz curvilinear transducers. For NT measurements a 5-MHz transducer was always used. Only on rare occasions was a transvaginal examination added. Thirty minutes were allocated for the ultrasound examinations. Pregnancy dating and screening for fetal malformations were performed at both the 12- and 18-week scans.

12-week scan policy

The 12-week scan policy implied the offer of fetal karyotyping to women with increased risk of fetal chromosomal anomaly according to NT screening, a fetal structural anomaly detected at any scan during pregnancy, or a history suggesting an increased risk.

NT was measured in accordance with the technical guidelines published by The FMF7. The presence of a cystic hygroma was considered equivalent to increased NT and was measured accordingly. The risk of DS was calculated using software developed by The FMF. A risk of 1 : 250 or more was regarded as increased (i.e. test-positive). Women who desired fetal karyotyping because of worry despite their risk being considered low were not denied this. Women at increased risk but with normal results of fetal karyotyping were offered an extended anomaly scan at 18 weeks. If the NT was ≥ 3.5 mm they were also offered fetal echocardiography. If gestational age was beyond 14 weeks according to fetometry, or if it was impossible to obtain a reliable NT measurement, the woman was counseled about her risk of DS on the basis of her age, history or any fetal anomaly found at scan.

18-week scan policy

The 18-week scan policy implied the offer of fetal karyotyping to women ≥ 35 years old on the estimated day of delivery and to women with a fetus with a structural anomaly detected at any scan during pregnancy, or a history implying an increased risk. Women < 35 years who desired fetal karyotyping because of worry were not denied this. Soft markers of chromosomal anomalies were not systematically searched for, but during the study period the awareness of their significance increased and the finding of a soft marker (choroid plexus cyst, echogenic focus in the heart, echogenic bowel, mild pyelectasis, short femur) often resulted in fetal karyotyping.

Fetal karyotyping

Because of the traditional practice in our centers, fetal karyotyping was most often performed following amniocentesis at 15 completed gestational weeks. In women at very high risk according to NT screening and in those in whom a fetal structural anomaly was seen, a fluorescent in-situ hybridization analysis for trisomy 21, 18 and 13, X and Y was often added to fetal karyotyping to obtain rapid results.

Retrieving information on pregnancy outcome

Information on pregnancy outcome was retrieved from delivery records and from the neonatal departments at the hospitals involved. We also retrieved information on chromosomal anomalies detected during the study period from the national registry of congenital anomalies and from the genetic laboratories. If termination, miscarriage or stillbirth occurred after the scan, the results were retrieved from autopsy and karyotyping. To facilitate follow-up, all women were given a questionnaire at their scan where they were asked to report pregnancy outcome.

Sample size

The primary outcome measure was the number of babies born alive at ≥ 22 weeks with DS. Secondary outcome measures were the total number of babies born with DS, the number of babies born with other chromosomal anomalies, the number of pregnancy terminations for DS, and the rate of invasive tests for fetal karyotyping.

Sample size was calculated on the basis of the assumption that the prevalence of DS at 12 weeks is 2.7/1000. This estimation was based on the registered numbers of terminations of pregnancies affected by DS and the registered numbers of liveborns with DS in Denmark in the period 1992–1993 (Danish Cytogenetic Central Register), corrected for the estimated number of spontaneous fetal losses8. If 75% of the cases of DS are detected at NT screening, as suggested by Snijders et al.4, and if most of these pregnancies are terminated, the prevalence of DS at birth would be approximately 0.4/1000, assuming a 40% loss rate of fetuses with DS in pregnancies not terminated between 12 and 40 weeks8. The prevalence of DS births in Sweden using maternal age screening for DS was 1.4/1000 at the time of planning the study9. To show a true difference in prevalence of babies born alive with DS as described above (1.4/1000 vs. 0.4/1000) as statistically significant at the 5% level with 90% power, the sample size needed would be 18 000 women in each group. To compensate for fetal loss and other dropouts, we decided to randomize 40 000 pregnant women.

Statistical analysis

We define the antenatal detection rate of chromosomal anomalies as detection before 22 weeks in a living fetus. The primary and secondary outcomes were evaluated by intention-to-treat analysis. Statistical analyses were carried out using the Statistical Package for the Social Sciences (SPSS Inc., Chicago, IL, USA, 2003). The statistical significance of differences in proportions was determined using Fisher's exact test or Chi-square test. The statistical significance of differences in continuous data was calculated using Student's two-sample test or the Mann–Whitney U-test. A two-sided value for P < 0.05 was considered statistically significant.

Interim analysis

An interim analysis including the first 20 000 randomized women was performed by two external reviewers. They found no differences between the randomization groups that justified an interruption of the trial.

Results

General outcome

Figure 1 shows the number of participants at each stage of the trial. Demographic characteristics of randomized women and those who declined are shown in Table 1. The women who declined participation were slightly younger (P < 0.001), more of them were parous (P < 0.001) and more of them had become pregnant during in-vitro fertilization treatment (P < 0.01). The two randomized groups did not differ in any aspect.

Figure 1.

Scheme showing the flow of participants in the trial.

Table 1. Demographic characteristics of women in the 12- and 18-week scan groups and of those women who declined to participate
Demographic characteristic12-week scan group (n = 19 796)18-week scan group (n = 19 776)Declined to participate (n = 10 061)
  1. IVF, in-vitro fertilization.

Age in years (mean ± SD)30.1 ± 4.930.2 ± 4.929.7 ± 5.2
Age ≥ 35 years (n (%))3831 (19.3)3838 (19.4)1919 (19.0)
Parity (mean (range))0.7 (0–9)0.7 (0–12)0.8 (0–11)
Nulliparous (n (%))9892 (50.0)9808 (49.6)4595 (45.7)
History of chromosomal anomaly (n (%))33 (0.2)40 (0.2)17 (0.2)
History of structural anomaly (n (%))99 (0.5)116 (0.6)46 (0.5)
IVF pregnancies (n (%))349 (1.8)347 (1.8)229 (2.3)
Current pregnancy twin (n (%))286 (1.4)257 (1.3)
Current pregnancy triplet (n)74

Pregnancy outcome for the 39 572 women randomized is presented in Table 2. In both the 12- and 18-week scan groups a similar proportion of women, 8% (1579/19 796) and 9% (1753/19 776; P < 0.005), respectively, failed to attend for ultrasound examination. In most cases the explanation was miscarriage, but in some cases the reason was unknown. The women in the 12-week scan group who failed to attend for unknown reason were slightly younger than those in the 18-week scan group (30.1 vs. 30.8 years; P = 0.008). There were no other differences between the randomization groups in demographic background data of the women who failed to attend for unknown reason or of the women lost to follow-up.

Table 2. Outcome of the 39 572 pregnancies included in the present trial
Outcome (n (%))12-week scan group18-week scan groupP
  • *

    Including 10 women with one liveborn baby and one spontaneous fetal death or selective feticide.

  • Including six women with one liveborn baby and one spontaneous fetal death. NS, not significant; weeks, gestational weeks.

Randomized19 796 (100)19 776 (100) 
Miscarriage before scan846 (4.2)976 (4.9)< 0.005
Termination before scan60 (0.3)55 (0.3) NS
Failed to attend for unknown reason673 (3.4)722 (3.6) NS
Underwent ultrasound scan18 217 (92.0)18 023 (91.1) NS
Missed miscarriage at scan244 (1.2)63 (0.3)< 0.0005
Scan with living fetus17 973 (90.8)17 960 (90.8) NS
Termination after scan134 (0.7)101 (0.5)< 0.05
Spontaneous fetal loss between scan and 22 weeks112 (0.6)53 (0.3)< 0.0005
Fetal death between 22 weeks and term82 (0.4)76 (0.4) NS
Lost to follow-up118 (0.6)137 (0.7) NS
Birth of at least one liveborn baby (n (%))17 537* (88.6)17 599 (89.0) NS
Liveborn babies (n)17 81017 850 

The performance of the two screening policies

The median gestational age at the scan was 12 + 6 weeks in the 12-week scan group and 17 + 4 weeks in the 18-week scan group. In 6% (999/17 973) of the women in the 12-week scan group with a living fetus at the scan, the fetal crown–rump length (CRL) was > 84 mm, and therefore NT screening could not be carried out. Among the remaining 16 974 women in whom the fetal CRL fell within the defined interval for NT measurement (45–84 mm), the NT was successfully measured in 96% (16 307/16 974) of cases. In 7% (1169/16 307) of these women the NT measurement was obtained at a repeat scan, either because the fetal CRL was too small at the first scan or because there were problems with obtaining an accurate measurement. In all, 9% (1666/17 973) of the women who attended for their 12-week scan and were found to have at least one living fetus at that scan never had an NT measurement carried out.

Among the women in the 12-week scan group who had NT measurement carried out, the risk of DS was increased (≥ 1:250) in 3.5% (587/16 577) of the fetuses and in 3.6% (584/16 307) of the women. NT was measured in 52/55 cases of DS in the 12-week scan group, and in 37/52 cases (71%) the risk of DS was ≥ 1:250. For a 5% test-positive rate, corresponding to a risk ≥ 1:335, 40/52 (77%) cases with DS would have been detected.

Among the women randomized to an 18-week scan with a living fetus at the scan, 18% (3269/17 960) were ≥ 35 years old. Had all these women undergone amniocentesis, 58% (24/41) of the fetuses with DS alive at the scan would have been detected prenatally.

Identification, antenatal detection and outcome of fetuses with Down syndrome

Ninety-eight cases of DS were identified during the study period (98/39 572; 2.5/1000 pregnancies randomized). There were more cases of DS identified in the 12-week scan group, but the difference was not statistically significant (55/19 796; 2.8/1000 vs. 43/19 776; 2.2/1000; P = 0.18). The mode of identification of cases with DS is shown in Table 3.

Table 3. Mode of identification of Down syndrome
Identification method12-week scan group (n = 55)18-week scan group (n = 43)
  • *

    In the 12-week scan group one fetus had hydrocephalus. In the 18-week scan group five fetuses had one or more soft markers and the sixth had bilateral clubfeet, choroid plexus cysts and pericardial effusion.

  • Chorionic villus biopsy was performed without an NT measurement because of a previous Down syndrome pregnancy.

  • All three cases in women ≥ 35 years old with a risk at NT screening of < 1:250. NT, nuchal translucency.

Fetal karyotyping at < 22 weeks because of:
 Missed miscarriage at scan 0 2
 Fetal loss after scan 0 1
 Increased risk according to NT screening37 0
 Maternal age ≥ 35 years15
 Abnormal ultrasound scan at < 22 weeks 1* 6*
 Previous Down syndrome 1 0
 Worry 3 1
 Increased risk at second-trimester serum screening 0 3
Fetal karyotyping at ≥ 22 weeks because of:
 Abnormal ultrasound scan 0 1
 Karyotyping of stillborn baby 3 0
 Karyotyping of liveborn baby1014

In the 12-week scan group there were 10 liveborn babies with DS. All were undetected before birth. In two of these cases NT measurement had not been carried out because the CRL was > 84 mm at the scan (one woman had no risk factor, the other was 41 years old and her fetus had two soft markers at the scan but she declined fetal karyotyping), the remaining eight cases all had a risk of DS < 1:250. In the 18-week scan group, there were 16 liveborn babies with DS. In one case the diagnosis was made by amniocentesis at < 22 weeks but the mother decided to continue the pregnancy, in one case the diagnosis was made at 35 weeks, and in 14 cases the diagnosis was unknown before birth. Three of the 14 mothers where the diagnosis was unknown before birth were ≥ 35 years old and had declined the offer of fetal karyotyping. The rate of liveborn babies with DS was 0.5/1000 (10/19 796) in the 12-week scan group vs. 0.8/1000 (16/19 776) in the 18-week scan group (P = 0.25). The total antenatal detection rate (< 22 weeks in a living fetus) was 76% (42/55) in the 12-week scan group vs. 61% (25/41) in the 18-week scan group (P = 0.12). Had there been no fetal karyotyping performed for indications other than those defined by each policy (i.e. no fetal karyotyping because of anxiety or increased risk according to serum screening) then the corresponding detection rates would have been 71% (39/55) vs. 51% (21/41) (P = 0.06).

All 42 women in the 12-week scan group in whom DS was diagnosed at < 22 weeks in a living fetus chose to terminate the pregnancy or to undergo selective feticide. Three of these women miscarried before the termination. Of the 13 cases with DS in the 12-week scan group that remained undiagnosed until after birth three were stillborn, and ten were born alive. Of the 25 women in the 18-week scan group who had DS diagnosed in a living fetus at < 22 weeks, 24 terminated the pregnancy whereas one continued her pregnancy and gave birth to a liveborn baby.

There were more terminations for DS in the 12-week scan group than in the 18-week scan group, but the difference was not statistically significant (39/19 796 vs. 24/19 776; P = 0.08). The total fetal loss rate (terminations and miscarriages) in pregnancies affected with DS was also higher in the 12-week scan group (45/19 796 vs. 27/19 776; P = 0.04). The median gestational age at termination of pregnancies with DS was 16 + 3 weeks in the 12-week scan group vs. 17 + 6 weeks in the 18-week scan group (P < 0.001).

Invasive tests for fetal karyotyping

In all, 8% (1593/19 796) of the women in the 12-week scan group and 11% (2118/19 776) of those in the 18-week scan group underwent fetal karyotyping (P < 0.001). Of the women who were at increased risk according to NT screening in the 12-week scan group, 73% (429/584) chose to undergo fetal karyotyping (Table 4). Of the women aged ≥ 35 years in the 18-week scan group, 52% (1689/3269) chose to undergo fetal karyotyping. Indications for testing are shown in Table 4. Some 88% (15/17) of the women in the 12-week scan group and 62% (41/66) of those in the 18-week scan group who underwent fetal karyotyping because of increased risk at serum screening were ≥ 35 years old.

Table 4. Indications for fetal karyotyping
Indication (n (%))12-week scan group (n = 1593)18-week scan group (n = 2118)
  • *

    No NT measurement carried out; fetal karyotyping because of age ≥ 35 years.

  • Booked for a 12-week scan by mistake.

  • Two women came too late for NT screening; 15 had a low risk according to NT screening.

  • §

    Twenty-four women had no NT screening; 142 had a low risk according to NT screening.

  • All had a low risk according to NT screening. NT, nuchal translucency.

Age ≥ 35 years117* (7.3)1689 (79.7)
Increased risk according to NT429 (26.9)1 (0.0)
History (e.g. previous affected pregnancy)40 (2.5)54 (2.5)
Anomaly at ultrasound scan46 (2.9)121 (5.7)
Increased risk according to second-trimester biochemistry17 (1.1)66 (3.1)
Parental anxiety (low risk at NT screening or no NT screening carried out and no risk factors)
 Maternal age < 35 years166§ (10.4)163 (7.7)
 Maternal age ≥ 35 years755 (47.4) 
Fetal death15 (0.9)21 (1.0)
Other10 (0.6)4 (0.2)

There were 38 invasive tests for fetal karyotyping for each case of DS detected in a living fetus at < 22 weeks in the 12-week scan group (1593/42). In the 18-week scan group, the corresponding number was 85 (2118/25). Had there been no fetal karyotyping performed for indications other than those defined by each policy (i.e. no fetal karyotyping because of worry or increased risk according to serum screening) there would have been 16 (632/39) vs. 89 (1864/21) tests for each case of DS detected at < 22 weeks in a living fetus. Almost all women who underwent fetal karyotyping had amniocentesis (97%; 3612/3711); the others had chorionic villus sampling (CVS) (n = 97) or cordocentesis (n = 2). Median gestational age at fetal karyotyping in pregnancies with a living fetus was 15 + 0 weeks in the 12-week scan group and 15 + 1 weeks in the 18-week scan group (P = 0.01). The spontaneous fetal loss rate after amniocentesis/CVS but before 22 weeks in fetuses with no chromosomal or structural anomaly was 0.9% (14/1507) in the 12-week scan group and 0.7% (15/2041) in the 18-week scan group (P = 0.58). This means that for each case of DS diagnosed prenatally by amniocentesis/CVS in the 12-week scan group, 0.3 normal fetus was lost in miscarriage at < 22 weeks after amniocentesis/CVS. In the 18-week scan group the corresponding figure was 0.6. Had there been no fetal karyotyping performed for indications other than those defined by each policy (i.e. no fetal karyotyping because of pure worry and none because of positive serum screening test) the corresponding figures would have been 0.1 vs. 0.7.

Antenatal detection of other unbalanced chromosomal anomalies

The number of unbalanced chromosomal anomalies other than DS and the mode of ascertainment are presented in Table 5. The number of cases identified, the number of babies born alive, and the number of cases diagnosed at < 22 weeks in a living fetus were virtually identical in the two arms of the trial (35 vs. 35, 10 vs. 9 and 27 vs. 27). The detection rates of the 12-and 18-week scan policies were similar: 20/35 vs. 25/35 (P = 0.32).

Table 5. Identification and antenatal detection of unbalanced chromosomal anomalies other than trisomy 21
Unbalanced chromosomal anomalies12-week scan group18-week scan group
  • *

    Increased risk according to NT screening in the 12-week scan group; maternal age ≥ 35 years in the 18-week scan group; history implying increased risk, or fetal anomaly in both groups.

  • One woman had an increased risk according to NT screening but declined fetal karyotyping on the basis of the NT risk. NT, nuchal translucency.

Trisomy 13 and 18 (n)1520
 Identified at fetal karyotyping at < 22 weeks because of: 
  Missed miscarriage at scan 0 1
  Increased risk in living fetus*1115
  Worry, no risk factors 1 0
 Identified at fetal karyotyping ≥ 22 weeks because of: 
  Ultrasound anomaly 1 0
 Identified in liveborn baby2 4
Turners syndrome (n) 5 5
 Identified at fetal karyotyping at < 22 weeks because of: 
  Increased risk in living fetus* 4 5
  Worry, no risk factor 1 0
Other sex chromosome anomalies (n) 6 3
 Identified at fetal karyotyping at < 22 weeks because of: 
  Missed miscarriage at scan 1 0
  Increased risk in living fetus* 2 2
  Worry, no risk factor 3 1
Triploidy (n) 3 2
 Identified by fetal karyotyping at < 22 weeks because of: 
  Missed miscarriage at scan 0 1
  Increased risk in living fetus* 3 1
Other unbalanced autosomal anomalies (n) 6 5
 Identified at fetal karyotyping at < 22 weeks because of: 
  Fetal loss after scan 0 1
  Increased risk in living fetus* 0 3
  Worry, no risk factor 2 0
 Identified at fetal karyotyping at ≥ 22 weeks because of:  0
  Ultrasound anomaly3 
 Identified in liveborn baby 1 1

All pregnancies in which trisomy 18 or 13, or triploidy was diagnosed at < 22 weeks in a living fetus were terminated. Three babies with trisomy 18 or 13 in the 12-week scan group and four in the 18-week scan group were born alive.

Discussion

The main purpose of this trial was to compare the pregnancy outcome in terms of number of babies born alive with DS between two policies of offering prenatal diagnosis, rather than comparing the prenatal detection rates of each policy. In our unselected pregnant population the number of babies born with DS did not differ substantially between the groups (10 vs. 16 born alive and 13 vs. 16 in all). In order to correctly interpret the results of our trial it is important to know that the performance of NT screening for DS in our hands was similar to that reported in other studies. Our detection rate was 77% for a 5% test-positive rate, which is similar to the detection rate of around 80% with corresponding test-positive rates of around 5% reported in review articles10–12. In our trial more cases of DS were identified (55 vs. 43), more were prenatally diagnosed at ≤ 22 weeks (42 vs. 25) and more were terminated (39 vs. 24) in the 12-week scan group than in the 18-week scan group (non-significant differences). However, the number of cases with DS diagnosed at > 22 weeks was almost identical (13 vs. 15) in the two groups, and the number of babies born with DS differed much less than we had expected: 13 in the 12-week scan group vs. 16 in the 18-week scan group, and 10 vs. 16 born alive. Our study was not designed to detect such small differences as statistically significant, even though the differences might be regarded by some as clinically important. The small difference in the number of babies born with DS between the two groups may have several explanations. One explanation could be that NT screening—because it is performed early in pregnancy—results in the detection and termination of many pregnancies with a fetus with DS that would have resulted in miscarriage without intervention. This assumption is supported by significantly more cases of DS being spontaneously lost or terminated in the 12-week scan group than in the 18-week scan group without this substantially affecting the number of babies born with DS. The small difference may also be explained by a higher than expected antenatal detection rate of DS in the 18-week scan group, mainly because many cases were identified due to fetal structural anomalies or soft markers at the 18-week scan, and because some women requested and underwent second-trimester biochemistry screening despite serum screening not being part of the 18-week policy. The prevalence of live births with DS in the 18-week scan group was lower than that on which we calculated our sample size (i.e. the prevalence in Denmark in the period 1992–1993). This may be explained by more cases of DS being detected antenatally at the 18-week scan in our study than in 1992–1993 because of a general increase in the skill of ultrasound examiners, use of better ultrasound equipment, and the systematic search for ultrasound anomalies. Also, the proportion of women ≥ 35 years old was larger in our study than it was in Denmark in 1992–1993, resulting in more cases of DS being detected because of advanced maternal age.

A very important difference between the two scan policies is that the 12-week scan policy was associated with many fewer invasive tests for fetal karyotyping per antenatally detected case of DS than the 18-week scan policy (i.e. 38 vs. 85). As a consequence, the number of normal fetuses lost in miscarriage after amniocentesis/CVS per case of DS detected by amniocentesis/CVS was much lower in the 12-week scan group. This difference would have been even more dramatic (16 vs. 89) had there been no amniocenteses/CVSs performed for indications other than those stipulated within each policy.

Our trial may be criticized for the two policies of offering prenatal diagnosis to pregnant women not being strictly adhered to. This reflects the situation in real life. Neither patients/pregnant women nor health personnel adhere blindly to recommended policies. There is no way of preventing patients from seeking a second opinion and ‘finally getting what they want’. Almost 60% of the amniocenteses in the 12-week scan group were performed because of parental worry in the absence of medical risk factors. This probably reflects the fact that the pregnant women did not trust the new NT method. In the 18-week scan group 11% of the amniocenteses/CVSs were performed because of parental worry. A more restrictive attitude towards performing amniocentesis/CVS in the absence of medical risk factors would have decreased the number of invasive tests substantially, especially in the 12-week scan group, with little effect on the number of cases with DS diagnosed at ≤ 22 weeks: only 4/1084 amniocenteses/CVSs performed because of worry showed that the fetus had DS. Some women requested—and obviously got—second-trimester serum screening, even though this did not form part of either of the policies offered. It is impossible to determine retrospectively how many women underwent serum screening, but it seems to have been more common in the 18-week scan group, since only 1% of the invasive tests in the 12-week scan group vs. 3% of those in the 18-week scan group were performed because of a positive serum screening result. The fact that serum screening was requested by some women may reflect that they were dissatisfied with the method of screening for DS that they had been offered. Many women ≥ 35 years old probably wanted serum screening to get some support for refraining from amniocentesis/CVS.

What would have been the pregnancy outcome if the policies offered had been strictly adhered to, namely if no amniocenteses/CVSs had been performed for indications other than those stipulated within the respective policy? One extreme would have been 10 babies born alive with DS in the 12-week scan group vs. 20 in the 18-week scan group (P = 0.07). This would have happened if all three cases of DS detected prenatally by ‘inappropriate’ indications and terminated in the 12-week scan group had not survived (fetal loss or termination for some other reason), whilst the four cases detected by ‘inappropriate’ indications and terminated in the 18-week scan group had been born alive. The other extreme would have been 13 babies born alive in the 12-week scan group vs. 16 in the 18-week scan group. This would have happened if all three cases of DS detected prenatally by ‘inappropriate’ indications and terminated in the 12-week scan group had been born alive, whilst the four cases detected by ‘inappropriate’ indications and terminated in the 18-week scan group had not survived (fetal loss or termination for some other reason). Anything between these extremes is possible and perhaps more likely than either of the two extremes.

The 12-week scan policy allows earlier detection of DS and earlier pregnancy termination than the 18-week scan policy. This potential advantage was not made full use of in our trial, because most procedures for fetal karyotyping were performed as amniocenteses at 15 weeks. This is explained by local traditions. Early termination is supposed to be both medically and psychologically advantageous. Reviews comparing modern methods of abortion show that the rate of medical complications is low after both first- second-trimester abortions, but that the complications are potentially more serious with second-trimester abortion13. Even though we have found no scientific evidence supporting the contention that early termination is psychologically advantageous, this is supported by clinical experience. It is possible that the 12-week scan will result in more ‘unnecessary’ pregnancy terminations because more pregnancies with DS destined to miscarry are likely to be detected and terminated with the 12-week scan policy. Whether this is a disadvantage is debatable. The long-term psychological effects of pregnancy termination performed because of fetal anomalies seem to be similar to those of spontaneous fetal loss of unknown cause14. Several studies suggest that pregnant women prefer early screening for DS15–17.

The antenatal detection rate and live born rate of fetuses with unbalanced chromosomal anomalies other than trisomy 21 were almost identical in the 12- and 18-week scan groups. This illustrates that the question of a 12- or an 18-week scan to screen for fetal chromosomal anomalies is a question merely of screening for DS, not of screening for other chromosomal anomalies, because these seem to be equally detectable with both policies.

It is important to realize that the results of our trial are applicable only to pregnant women with ‘risk profiles’ (e.g. age) and attitudes towards invasive testing for fetal karyotyping and pregnancy termination similar to those of the women in our trial. All women except one in our trial who were informed that their fetus had DS at < 22 weeks chose to terminate the pregnancy. The results may also be affected by the attitudes of those providing counseling (e.g. whether they have a generous or restrictive attitude towards fetal karyotyping because of worry). The effect of a screening method for fetal chromosomal anomaly offered to an entire population may be different from that in a mathematical model or demonstration project. This is because the performance of a screening method depends on the age of the women screened, screening uptake, women's preferences with regard to screening methods, their choices of invasive testing and their attitudes towards pregnancy termination. Smith-Bindman et al.18 have discussed the efficacy of population-based screening programs for DS. They found a detection rate of DS of 37% in geographical areas where maternal age was the most common indication for fetal karyotyping and a detection rate of about 50% in areas where serum screening or NT screening were the most common indications for fetal karyotyping18.

In summary, our results indicate that from the point of view of screening for DS, the 12-week scan policy is superior to the 18-week scan policy, its major and unequivocal advantage being the much lower number of amniocenteses/CVSs needed to be performed per case of DS detected at ≤ 22 weeks. A possible effect of the 12-week scan policy on the number of babies born with DS is likely to be small.

Acknowledgements

This trial was supported by grants from the Stockholm County Council Public Health and Medical Services Committee Research and Development departments; Landstingsfinansierad regional forskning, Region Skåne (a governmental grant); Uppsala County Council and Karolinska Institute South Hospital. The authors want to thank all the midwives and colleagues at the antenatal and ultrasound clinics who contributed to the NUPP trial with loyalty and commitment. Drs Sverker Ek, Christina Pilo, Alf Maesel and Gudmundur Gunnarsson are particularly acknowledged. Finally, The Fetal Medicine Foundation and Professor Kypros Nicolaides are acknowledged for generous support in education and quality control.

Ancillary