Skip to main content
NCBI home page
The BMJ logoLink to The BMJ
. 2002 Jan 12;324(7329):110.

Screening for Down's syndrome

Antenatal screening has human costs

Josephine Venn-Treloar 1
PMCID: PMC1122000  PMID: 11786462

Editor—In their study Gilbert et al made no provision for the cost of counselling.1 I still hear from women who go for antenatal care and are screened without consent or counselling.1,2 Some of these women regret entering a conveyor belt process that ends up with an invasive procedure that causes a high rate of fetal loss. Gilbert et al, by highlighting the economic implications of screening, remind us that mothers and their unborn babies continue to be used in what is primarily a cost saving procedure.

Doubt has already been cast upon the public health benefits of screening and the ethical component of such programmes.3 No pregnancy is replaceable. The cost of screening in human terms has not been evaluated. By excluding the costs of counselling and disregarding mothers' rights and the worth and value to society of children with Down's syndrome, the study by Gilbert et al has severe limitations that should prevent us from embracing such screening strategies.

References

  • 1.Gilbert RE, Augood C, Gupta R, Ades AE, Logan S, Sculpher M, et al. Screening for Down's syndrome: effects, safety, and cost effectiveness of first and second trimester strategies. BMJ. 2001;323:423–425. doi: 10.1136/bmj.323.7310.423. . (25 August.) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.J Venn-Treloar J. Nuchal translucency-screening without consent. BMJ. 1998;316:1026. [PMC free article] [PubMed] [Google Scholar]
  • 3.Ford C. The value of screening for Down's Syndrome in a socio-economically deprived area with a high ethnic population. Br J Obstet Gynaecol. 1998;105:855–859. doi: 10.1111/j.1471-0528.1998.tb10229.x. [DOI] [PubMed] [Google Scholar]
BMJ. 2002 Jan 12;324(7329):110.

Costings are crucial to model

Martin Whittle 1

Editor—Gilbert et al in their paper provide an analysis of cost and effectiveness of screening for Down's syndrome using mathematical modelling.1-1 Costing of procedures undertaken in the NHS is difficult, and Gilbert et al conclude that four screening strategies are cost effective and efficient.

Nuchal translucency screening is apparently the cheapest procedure, at £4.40 per test, but there is no indication how this figure was derived. In a cost analysis undertaken for ultrasound scanning for the Royal College of Obstetrics and Gynaecology in 1995, the cheapest scan cost about £35.1-2 This figure was probably unreliable, but the discrepancy is large, and it is important to resolve it since the costs Gilbert et al quote for nuchal translucency screening are crucial to their model. If the costing is inaccurate then the conclusions of the study are likely to be incorrect.

The literature generally agrees that nuchal translucency screening together with serum testing seems an efficient screening method, which is confirmed by this model. But the practical problems that surround implementation of such a programme nationally are formidable. Only about 7% of units are offering nuchal translucency screening. There are well recognised shortages of radiography staff and possibly inadequate facilities in some circumstances. Early screening demands effective counselling before a woman gets to the hospital, so she has time to consider her options.

One of the benefits of an early diagnosis is a surgical termination, but in this model only about one third of women had their diagnosis in time. Early diagnosis also increases the demand for chorionic villus sampling, which is generally available only in fetal medicine centres, carries a higher miscarriage rate than amniocentesis and incurs higher laboratory costs. Currently about 60% of women in the United Kingdom are offered at least a double test.

Gilbert et al indicate the weakness of that strategy, but it is a starting point, and the addition of other analytes would not necessarily be difficult. Screening is likely to be more feasible to support during the second trimester than during the first, mainly because of the difficulties of implementing nuchal translucency measurements. This model suggests that a quadruple test may prove effective, and, although there would be an increased cost, the test entails automated, laboratory techniques and not those dependent on a skill—such as nuchal translucency screening. The challenge is to find a technique that is not only cost effective and safe but that is also feasible to implement nationally.

References

  • 1-1.Gilbert RE, Augood C, Gupta R, Ades AE, Logan S, Sculpher M, et al. Screening for Down's syndrome: effects, safety, and cost effectiveness of first and second trimester strategies. BMJ. 2001;323:423–425. doi: 10.1136/bmj.323.7310.423. . (25 August.) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 1-2.Royal College of Obstetrics and Gynaecology. Ultrasound screening for fetal abnormalities. Report of the RCOG Working Party. London: RCOG; 1995. [Google Scholar]
BMJ. 2002 Jan 12;324(7329):110.

Conclusions do not reflect reality

Tim Reynolds 1

Editor—Gilbert et al have tried to identify the best approach for screening for Down's syndrome.2-1 They have, however, fallen into the trap of uncritically evaluating the published literature. There is no statistical proof that the detection rate is significantly better for any of triple screening versus double screening, quadruple screening versus double or triple screening, or screening in the first trimester versus the second trimester, or that the integrated test is an improvement.2-2

All studies of Down's syndrome screening have had small numbers of examples of pregnancies positive for Down's. Consequently, although false positive rates can be accurately assessed, the detection rate cannot. The confidence intervals about the detection rate are so wide that no one has yet proved that the superiority of one method over another is not entirely due to chance.2-2 It has also been known for a long time that where small numbers have been used to derive a population model that is used to estimate false positive and detection rates, large errors can result from an error in the estimate of the population variance for a single analyte giving rise to a hopelessly optimistic estimate of the effectiveness of a procedure.2-3

Studies of the efficacy of nuchal translucency screening have all been interventional, and the quoted detection rate (of the order of 80%) therefore needs to be deflated to account for natural fetal losses. The effect of natural fetal loss is to reduce the detection rate for screening during the first trimester to less than 60%, making it comparable to screening during the second trimester.2-4 In addition, for nuchal translucency measurement, there is a far greater incidence of failure to be able to make a measurement than there is failure of biochemical testing to be able to produce a result, which further depresses the detection rate.

Finally, there are serious statistical doubts about the integrated test and worries about the ethics of withholding early positive results until a second test has been performed.2-5 The unit costs estimated for the different tests are also interesting. How can it be that an ultrasound test requiring 15-20 minutes of hands on technical work is cheaper than a biochemical test, when one person can analyse hundreds of samples per day?

Since the detection rates for nuchal translucency screening are grossly overestimated and the integrated test is expensive and unproved, the conclusion that these two modalities of screening represent the boundary of cost effectiveness cannot be left unchallenged.

References

  • 2-1.Gilbert RE, Augood C, Gupta R, Ades AE, Logan S, Sculpher M, et al. Screening for Down's syndrome: effects, safety, and cost effectiveness of first and second trimester strategies. BMJ. 2001;323:423–425. doi: 10.1136/bmj.323.7310.423. . (25 August.) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2-2.Reynolds TM. Downs syndrome screening: A controversial test with more controversy to come! J Clin Path. 2000;53:893–898. doi: 10.1136/jcp.53.12.893. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2-3.Reynolds TM. Chapter 4: Screening by test combination: a statistical overview. In: Grudzinskas JG, Chard T, Chapman M, Cuckle H, editors. Screening for Down's syndrome. Cambridge: Cambridge University Press; 1994. pp. 47–71. [Google Scholar]
  • 2-4.Reynolds T. Antenatal screening for Down's syndrome. Lancet. 1998;352:1145. doi: 10.1016/s0140-6736(05)79785-3. [DOI] [PubMed] [Google Scholar]
  • 2-5.Reynolds T, Zimmermann R, Wright E. Integrated screening for Down's syndrome. New Engl J Med. 1999;341:1395–1397. [PubMed] [Google Scholar]
BMJ. 2002 Jan 12;324(7329):110.

Modelling does not predict reality accurately

David Howe 1

Editor—Gilbert et al conclude that, of the antenatal screening tests for Down's syndrome, the integrated test is the most effective and safest strategy and that all other strategies result in more liveborn babies or miscarriages of unaffected pregnancies.3-1 Their conclusions are derived from mathematical modelling rather than trial data. This is particularly true when evaluating the integrated test, which has never been tested in a controlled trial nor even in a large uncontrolled demonstration project.

It is difficult to think of any other area of medicine where health policy is so driven by modelling without the support of properly gathered clinical evidence. There are no controlled studies showing that serum or nuchal translucency screening is more effective than maternal age or reduces the rates at which invasive procedures are performed. By contrast, there is clear evidence that modelling does not accurately predict the effectiveness of screening programmes in practice. We reported a 68% antenatal detection rate over six years, screening by maternal age where routine anomaly scanning was also offered.3-2 Ford et al found that serum screening detected only 31% of cases of Down's syndrome antenatally.3-3 The modelling relied on by Gilbert predicts that in the first study the detection rate should have been 32% and in the second 60%.

Modelling does not predict reality accurately for two reasons.

Firstly, women do not behave as predicted. In Ford's study only 71% accepted screening, so only 13 of 19 cases occurred in those screened.3-3 Only 11 of these 13 were screen positive, of which only six had invasive testing. Only four of the six chose termination. In other studies many older mothers decline screening in favour of a definitive diagnostic test, increasing the amniocentesis rate above that predicted.

Secondly, modelling also fails because of inaccurate assumptions about the age structure of the antenatal population. The models assume that 5% of pregnant women are aged over 35 whereas nationally the proportion is 16.5% (launch of National Sentinel Caesarean Section Audit). In Southampton this resulted in over 65% of cases of Down's arising in older women.3-2

Finally, Gilbert et al do not consider some of the negative impact of screening. Women increasingly complain that midwives and doctors are not interested in their pregnancies until they have ruled out Down's syndrome. Age based screening need be discussed with only a small segment of the population, who are often aware when planning their pregnancy of the higher risk of Down's. If we are to retain scientific credibility the choice of the best screening policy should be made on the basis of evidence from controlled clinical trials and not models that are demonstrably inaccurate.

References

  • 3-1.Gilbert RE, Augood C, Gupta R, Ades AE, Logan S, Sculpher M, et al. Screening for Down's syndrome: effects, safety, and cost effectiveness of first and second trimester strategies. BMJ. 2001;323:423–425. doi: 10.1136/bmj.323.7310.423. . (25 August.) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3-2.Howe DT, Gornall R, Wellesley D, Boyle T, Barber J. Six year survey of screening for Down's syndrome by maternal age and mid-trimester ultrasound scans. BMJ. 2000;320:606–610. doi: 10.1136/bmj.320.7235.606. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3-3.Ford C, Moore AJ, Jordan PA, Bartlett WA, Wyldes MP, Jones AF, et al. The value of screening for Down's syndrome in a socioeconomically deprived area with a high ethnic population. Br J Obstetr Gynaecol. 1998;105:855–859. doi: 10.1111/j.1471-0528.1998.tb10229.x. [DOI] [PubMed] [Google Scholar]
BMJ. 2002 Jan 12;324(7329):110.

Authors' reply

R E Gilbert 1,2,3,4, C Augood 1,2,3,4, R Gupta 1,2,3,4, S Logan 1,2,3,4, A E Ades 1,2,3,4, M Sculpher 1,2,3,4, J H P van der Meulen 1,2,3,4

Editor—The questions about costs raised by Venn-Treloar, Whittle, and Reynolds were addressed in the full text version of the report on bmj.com, with further details in the technical report (www.ich.ucl.ac.uk/srtu/frampubs.htm).4-1 We included the costs of counselling before amniocentesis, chorionic villus sampling, or termination, but we assumed that screening options were discussed with all women at booking. As all women were assumed to have had a dating ultrasound scan, the cost of the nuchal fold translucency test relates to the additional time to take measurements, explain the results, and train ultrasonographers.

Reynolds seems to have missed the section in the methods that explains that the nuchal fold measurement was adjusted for verification bias. Howe makes the case for a modelling exercise. Differences between our detection rates and those from studies based on routine care will be strongly affected by uptake rates, referral practices, and verification bias. Modelling takes account of these factors to allow comparison of test performance and would still be required even if trials were feasible.

Finally, Reynolds raises an important point about the poor precision of the detection rate. One approach is to look for consistency of the characteristics of test performance. Meta-analyses of the results for biochemical markers produce comparatively precise results, which are consistent with the characteristics used in the analysis. But to take account of the correlation between markers we used test characteristics from a single large, archived dataset. Other archived datasets have given similar results.4-2 We believe that this approach gives the best estimates of test performance but accept that random error is not represented.

References

  • 4-1.Gilbert RE, Augood C, Gupta R, Ades AE, Logan S, Sculpher M, et al. Screening for Down's syndrome: effects, safety, and cost effectiveness of first and second trimester strategies. BMJ. 2001;323:423–425. doi: 10.1136/bmj.323.7310.423. . (25 August.) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4-2.Wald NJ, Kennard A, Hackshaw A, McGuire A. Antenatal screening for Down's syndrome. J Med Screen. 1997;4:181–246. doi: 10.1177/096914139700400402. [DOI] [PubMed] [Google Scholar]
BMJ. 2002 Jan 12;324(7329):110.

Ratio of femoral length to tibial length needs to be evaluated extensively

Pooja Sachdev 1, Shubhra Bahl 1, Jacob M Puliyel 1

Editor—Gilbert et al and Howe et al in their articles rely heavily on maternal age to screen for Down's syndrome.5-1,5-2 But maternal age is not so useful in India and other countries where early marriage is the norm and the social pressures for early motherhood are enormous. In our series, which included 3000 deliveries and seven babies with Down's syndrome, we saw that all babies with the syndrome were born to mothers younger 35 years.

We reported in the American Journal of Perinatology our finding that the ratio of femoral length to tibial length remains remarkably constant around 1.15 (range 1.13-1.19) in fetuses after 13 weeks' gestation.5-3 Fetuses with Down's syndrome had this ratio greater than 1.2 (standard deviation 4.5) compared with norms. The youngest fetus with Down's syndrome in our sample was 22 weeks old at the time of measuring. We hope that this ratio will be evaluated more extensively and earlier in pregnancy, to see if these findings are valid in the early second trimester and across ethnic groups.

References

  • 5-1.Gilbert RE, Augood C, Gupta R, Ades AE, Logan S, Sculpher M, et al. Screening for Down's syndrome: effects, safety, and cost effectiveness of first and second trimester strategies. BMJ. 2001;323:423–425. doi: 10.1136/bmj.323.7310.423. . (25 August.) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5-2.Howe DT, Gornall R, Wellesley D, Boyle T, Barber J. Six years survey of screening for Down's syndrome by maternal age and mid trimester ultrasound scans. BMJ. 2000;320:606–610. doi: 10.1136/bmj.320.7235.606. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5-3.Gupta R, Thomas RD, Sreenivas V, Walter S, Puliyel JM. Ultrasonographic femur-tibial length ratio: a marker of Down syndrome from the late second trimester. Am J Perinatology. 2001;18:217–223. doi: 10.1055/s-2001-15500. [DOI] [PubMed] [Google Scholar]
BMJ. 2002 Jan 12;324(7329):110.

Nuchal translucency screening may help detect congenital heart disease

John M Simpson 1

Editor—Screening for Down's syndrome during fetal life has received much attention, with over 350 articles in the medical literature over the past five years alone. The data on cost effectiveness provided by Gilbert et al are welcome, given the variable screening strategies offered by different NHS providers.6-1 Increased nuchal translucency may be a marker of many fetal diseases, and a discussion of its cost effectiveness should not be restricted to Down's syndrome alone.

Congenital heart disease is the most common type of fetal malformation (at least 8/1000 during fetal life), but only 25% of clinically significant malformations are actually detected prenatally in the United Kingdom,6-2 which is worse than any other type of fetal anomaly. Increased nuchal translucency is associated with congenital heart disease, independent of karyotypic abnormalities, with a sensitivity of 15-56%.6-3–6-5 Although far from perfect as a screening tool for congenital heart disease, nuchal scanning makes some contribution to prenatal detection of congenital heart disease and chromosomal abnormalities. This is highly relevant when the relative merits and cost effectiveness of prenatal screening strategies are evaluated.

References

  • 6-1.Gilbert RE, Augood C, Gupta R, Ades AE, Logan S, Sculpher M, et al. Screening for Down's syndrome: effects, safety, and cost effectiveness of first and second trimester strategies. BMJ. 2001;323:423–425. doi: 10.1136/bmj.323.7310.423. . (25 August.) [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6-2.Bull C. Current and potential impact of fetal diagnosis on prevalence and spectrum of serious congenital heart disease at term in the UK. British Paediatric Cardiac Association. Lancet. 1999;354:1242–1247. doi: 10.1016/s0140-6736(99)01167-8. [DOI] [PubMed] [Google Scholar]
  • 6-3.Mavrides E, Cobian-Sanchez F, Tekay A, Moscoso G, Campbell S, Thilaganathan B, et al. Limitations of using first-trimester nuchal translucency measurement in routine screening for major congenital heart defects. Ultrasound Obstet Gynecol. 2001;17:106–110. doi: 10.1046/j.1469-0705.2001.00342.x. [DOI] [PubMed] [Google Scholar]
  • 6-4.Michailidis GD, Economides DL. Nuchal translucency measurement and pregnancy outcome in karyotypically normal fetuses. Ultrasound Obstet Gynecol. 2001;17:102–105. doi: 10.1046/j.1469-0705.2001.00341.x. [DOI] [PubMed] [Google Scholar]
  • 6-5.Hyett J, Perdu M, Sharland G, Snijders R, Nicolaides KH. Using fetal nuchal translucency to screen for major congenital defects at 10-14 weeks of gestation: population based cohort study. BMJ. 1999;318:81–85. doi: 10.1136/bmj.318.7176.81. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from BMJ : British Medical Journal are provided here courtesy of BMJ Publishing Group

RESOURCES