For a large share of pregnant women, testing for Down syndrome and other chromosomal abnormalities is a standard part of prenatal care. This testing has gotten quite a bit of negative press in the last few days. In particular, news reports have questioned the value and accuracy of new technologies that claim to detect problems at a high rate with a simple blood test.
As a first shot, the New England Center for Investigative Reporting detailed several cases where this test resulted in either a false negative (the test said the fetus had the normal set of chromosomes when it did not) or a false positive (the test said there was a problem and there was not). NECIR gave several examples of women who terminated healthy pregnancies as a result of these inaccurate tests. Other outlets have piled on with concerns about the “alarming inaccuracy” of prenatal testing.
These new tests aren’t perfect, but the current discussion dramatically undervalues them, and they represent a huge leap in accuracy over what was previously available.
Without any intervention, chromosomal abnormalities (Down syndrome is the most common, others include Edwards syndrome and Patau syndrome) would occur in something like 1 in 700 live births, with the rate increasing significantly as the mother’s age increases.
If you want to know for sure whether a fetus has a chromosome imbalance, you have to look at its genome. This requires either an amniocentesis around 16 to 18 weeks of pregnancy or a more complex procedure performed earlier in pregnancy called a chorionic villus sampling (CVS) test. The amniocentesis involves sampling some of the amniotic fluid with a large needle inserted into the uterus; the CVS test involves surgically removing a small piece of the placenta. These tests are accurate (essentially, perfect), but carry a small risk of miscarriage due to their invasive nature. For this reason, there has long been a push to develop technologies that will detect fetal problems without this risk.
Until recently, the best procedure was a combination of an ultrasound and a blood test at 11 to 12 weeks of pregnancy. It’s able to generate a much more accurate measure of risk than the woman’s age alone would indicate, but the test is far from perfect. For younger women — say, those under 35 — this test detects about 80 percent of fetuses with Down syndrome. This leaves a lot undetected. For older women, the detection rate is much better — about 98 percent of fetuses with Down syndrome will be identified with this test — but there are many false positives. Up to 25 percent of pregnant women at age 40 will be told their fetuses are at high risk based on this test, even though only about 1 in 100 of them will have a fetus with Down syndrome.
The new blood tests — first available about three years ago — are basically a much, much better version of these screening tests. They check the fetal DNA circulating within the mother’s blood for chromosomal imbalances. Put very simply, they look for too many of one type of chromosome in the mother’s blood. Assuming mom has two copies of each chromosome, observing that there are a lot more copies of, say, chromosome 21 (the marker for Down syndrome) floating around than you would expect is indicative of an imbalance in the fetal genes.
These tests are wildly accurate relative to the existing ultrasound option. Rather than detecting 80 percent of cases, they detect about 99 percent of cases. And their false positive rates are also way, way lower. This advance is enormous.
But they’re still not as accurate as the invasive screening tests. If you want to detect not 99 of 100 cases but 100 of 100 cases of Down syndrome, you still need a amniocentesis or a CVS test.
Some of the complaints about the new blood tests say the manufacturers overstate their accuracy. This isn’t true; the accuracy data is supported in larger research trials. If anything, 99 percent is probably an understatement.
But 99 percent accuracy is not 100 percent accuracy. The problem may not lie in the claims made by the companies who make the tests, but in the interpretation of these results by doctors and patients. The earlier versions of the screening tests were so inaccurate that no one would think of acting on their results by terminating a pregnancy. The enhanced accuracy here may, perversely, encourage acting on this information when it is still not certain. But that problem can’t be fixed by the test manufacturers; it requires greater statistical literacy among doctors and patients.