Skip to content

Testing choices: weighing up risks of death and Down syndrome for fetuses

In the Observer yesterday, researchers from a major disability charity
have claimed that the risks of screening for Down syndrome during
pregnancy have been underestimated. The researchers suggest that for
every 3 fetuses with Down syndrome that are detected by screening 2
unaffected fetuses miscarry as a complication of the testing process.
Should screening be stopped? If screening continues how should
prospective parents weigh up this risk?

The Observer headline is alarming. “Down’s tests are a great risk for all babies” it announces. The research that is cited in the article is unpublished, so it is impossible to scrutinise the claims that are contained within. However it appears that the researchers are not claiming (contra the sub-editors who presumably wrote the headline) that testing represents a large absolute risk to infants. Rather, it appears that they are comparing the relative risk of different outcomes of testing.

Down syndrome is a genetic disorder (involving an extra copy of chromosome 21) that leads to moderate intellectual disability, as well as a range of physical and health problems. Screening for Down syndrome during pregnancy can be done in a number of ways. In the UK it is usually done in two stages. The first stage involves a combination of a blood test and ultrasound. These tests cannot diagnose Down syndrome, they merely establish whether women are at a higher risk than usual. Mothers who are identified as being at higher risk are then offered an invasive test that samples fluid around the fetus, or cells of the placenta in order to definitively diagnose Down syndrome. These invasive tests cause miscarriage in up to 1-2% of pregnancies.

There are two problems worth noting. The first is that current first-line screening tests are imperfect. The overall risk of having a fetus with Down syndrome is 1 in 1000 (this risk increases with the age of the mother). The UK standard for antenatal testing is that the false positive rate should be less than 3%. This means that there are up to 30 women who need to have invasive (and risky) tests for every 1 fetus that actually has Down syndrome.*
There is no question that it would be better for these tests to be more accurate as well as for the risk of miscarriage with testing to be reduced.

But the second issue relates to how prospective parents make choices about antenatal testing. The choice involves a number of conflicting priorities. Parents need to weigh up the uncertainties associated with testing, the need for further invasive tests if the first test is positive, how bad it would be for their family if they were to have an affected child, their own feelings about abortion as well as the risk of miscarriage. Although information leaflets are provided to parents about screening, decisions are often made quickly along with decisions about many other ‘routine’ tests. Some people have worried that decisions about antenatal screening are often not fully informed.

One way of helping patients and parents make difficult decisions is to use decision theory, and decision-support tools. Decision theory is a way of weighing up the relative value of different outcomes, along with their probabilities in order to come up with the ‘best choice’. There has been quite a lot of interest in decision theory in relation to testing during pregnancy – though mostly with a view to helping doctors and policy makers decide which testing strategy to use. There has also been some interest in the development of tools to help individuals work out which decision would be best for them. How would they work? We might imagine a web-based program that provided information for parents about Down syndrome, testing methods and results, and risks of complications. The program would help parents to weigh up different outcomes. Then patients could put in details relevant to them (for example the mother’s age or the local characteristics of testing), and the program could give an indication of whether or not to undertake testing (‘Computer says "No”…’). For example, for some women it may be acceptable to undertake screening for Down syndrome even if the relative risk of miscarriage were higher than the research reported in the Observer indicates. The impact on mothers and families of miscarriage should not be underestimated, but the impact of looking after a child with significant disability may be substantially greater. If parents think that it would be much worse to have an affected child than to have a miscarriage it would be rational to go ahead with testing.

Although our decisions and responses to risk are not always rational and consistent with the results of decision theory, one advantage of this process is that it would help parents think through the decision. They may or may not ultimately follow a decision that the computer suggests, but their decision is likely to be better informed for having gone through the process. Whether or not the new research turns out to be significant, there is a strong case for using methods such as decision analysis to help parents make difficult decisions about antenatal screening.

*If we assume (conservatively) that there is a 3% false positive rate, and a 1% risk of miscarriage with amniocentesis (the risk is probably lower than this, but this is the figure usually quoted), there would be approximately 3 fetuses detected with Down syndrome for every 1 unaffected fetus that miscarries. The figure quoted in the Observer might result if the miscarriage rate were higher than usual, if the population risk of Down syndrome were lower than 1 in 1000, or if the false positive rate were higher.


New medical research shows Down syndrome tests are a great risk for all babies Observer 14/8/08

Prenatal screening and your baby – general information about prenatal testing and risks

Down syndrome and decision theory Radford Neal blog 7/9/08

Amniocentesis and chorionic villus sampling – information from the NHS

Fetal anomaly screening program uk

Multiple marker screening for Down syndrome Journal American Board Family Practice 1999

Bioethical decision-making in Decision making in Health Care

A review of decision support technologies for amniocentesis Human Reproduction Update 2008

Share on

3 Comment on this post

  1. Thanks for that link Professor Dawid – it is very interesting to see how a statistician might use decision theory to make difficult real-life decisions.

    The way that you resolve the problem is I think by reference to the Independence axiom (see the link to Radford Neal’s blog above), though you don’t put it in those terms in the article that you have cited (I’m not a statistician, so correct me if I am wrong). To work out the ‘utility’ placed on the different outcomes for the decision about testing could involve a variant of the standard gamble. So parents might ask themselves, if facing a choice between certain miscarriage and a risk (r) of their infant surviving with Down syndrome, how big would r have to be for them to reject the ‘gamble’.
    Here is one way of expressing the gamble in a perhaps slightly more intuitive way than the magic coin.

    Imagine that a treatment has been developed for mid-trimester miscarriage. Mothers who develop symptoms that indicate that they are about to miscarry can have a new-fangled treatment that involves irradiating the fetus. The treatment, if undertaken, will definitely stop the fetus from miscarrying, but [this is not terribly plausible, but suspend your disbelief] has a chance of causing chromosome duplication leading to Down syndrome. There is an ‘r’ chance of this occurring.

    Parents might look at this scenario, and consider that (for example) they would not undertake the treatment if there was a more than 30% chance of Down syndrome. This could then be used to calculate the relative utilities and work out whether or not to proceed with testing.

    Someone who is more familiar with decision theory maths than I am would probably be able to tell us how to convert ‘r’ into the ratio of Down syndrome:miscarriage that would be acceptable (ie testing would still be rational). For example, crunching the numbers I get a figure of 2:1 (ie 2 infants with Down syndrome per miscarriage) if r is 0.3, but I am not sure that is correct.


  2. The editorial by Drs Buckley is now available online – see
    I haven’t had a chance to do more than skim it yet, but it looks as though the main cause for the higher rate quoted by the authors is a higher false positive rate, that they derive from a 2003 paper.

    see also for an article describing the ‘collateral damage’ of screening for Down syndrome, citing the Buckleys’ article

    the news has been picked up by other media


Comments are closed.