The alarming inaccuracy of Slate’s piece on prenatal testing

iStock_000035509214Small

Americans’ public understanding of science is woefully inadequate. We are remarkably ill informed on issues as wide ranging as evolution and climate change, rendering us uniquely vulnerable to manipulation by special interest groups and startlingly gullible when it comes to the peddling of pseudoscience “cures” by quacks and charlatans.

That’s why it is especially disappointing to see a mainstream publication featuring a piece that actually increases public misunderstanding of science instead of ameliorating it. Unfortunately, Libby Copeland’s article The Alarming Inaccuracy of Prenatal Testing displays  a remarkable lack of understanding of science, explaining nothing, confusing everything, and creating unnecessary fear for pregnant women.

According to Copeland:

The New England Center for Investigative Reporting has a story out that should alarm every woman who’s recently had blood drawn for a prenatal screening. Beth Daley reports that, “sparked by the sequencing of the human genome a decade ago, a new generation of prenatal screening tests…has exploded onto the market in the past three years. The unregulated screens claim to detect with near-perfect accuracy the risk that a fetus may have Down or Edwards syndromes, and a growing list of other chromosomal abnormalities.”

The problem is, they don’t.

Copeland’s central claim is utterly false.

Copeland, who is generally an excellent writer, falls into a common trap that afflicts the unwary. She fails to recognize the difference between the colloquial meaning of words and the meaning within scientific discourse. Because she utterly misunderstands the use of words like “screening,” “accuracy” and “predictive,” she has produced a breathless piece of baseless fear mongering.

Before we can understand what’s wrong with Copeland’s piece, we need to understanding the meaning of those words in scientific discourse. Let’s start by framing the problem.

Most expectant parents want to know if their unborn child is afflicted with a serious genetic problem. With the information in hand, they may choose to terminate the pregnancy or to consult with appropriate specialists to prepare for the birth of the child. Amniocentesis (putting a needle into the uterus and drawing off amniotic fluid for analysis) can provide a nearly 100% accurate diagnosis of the major genetic problems of interest such as Down’s syndrome.

Amniocentesis is a diagnostic test, because it provides a diagnosis. The problem with amniocentesis is that it is invasive, expensive and increases the risk of miscarriage. If all 4 million pregnant women had amniocentesis each year, there would be hundreds of miscarriages and the cost would be astronomical. Obviously we don’t want to do that. What we’d like to do is restrict the use of amniocentesis only to those women at high risk of carrying a baby with a serious genetic problem.

How do we find out who is at high risk? We offer all 4 million pregnant women a screening test.

The critical difference between a diagnostic test and a screening test is this:

A screening test can never provide a diagnosis.

Never!

What do we mean then when we talk about the accuracy of a screening test if a screening test, by definition, can NEVER give us a diagnosis?

The perfect screening test would mean that everyone who has a positive test result actually is at high risk (positive predictive value = 100%) and everyone who has a negative result is actually at low risk (negative predictive value = 100%). However, and this is the critical point, even a perfect test with 100% positive predictive value and 100% negative predictive value could NEVER tells us which child has Down’s syndrome or any other major genetic problem. It can ONLY tell us who ought to have amniocentesis; that’s it, nothing more.

Copeland clearly does not understand this when she writes:

… [W]hile these non-invasive prenatal tests are more accurate than the combination of blood test and ultrasound … they nevertheless can lead to false alarms as much as half the time. The rarer the condition, the greater the risk of false alarm. And yet they promise things like “99% Accurate, Simple & Trusted.”

The test promises that is is 99% accurate AT IDENTIFYING WOMEN AT RISK, not at diagnosing major chromosomal problems.

Copeland concludes her piece with a bit of information that she believes is a scathing indictment of the test, but it actually a scathing indictment of her understanding:

Three of the labs returned results indicating that the women were pregnant with healthy female fetuses.

The catch? The women who gave their blood samples weren’t pregnant at all.

That’s only to be expected from a screening test. A false positive means that the test indicates high risk for a major genetic problem when no major genetic problem exists. Therefore, a non-pregnant woman could have a false positive test. This is not news to people who understanding screening tests.

Copeland had a perfect opportunity to educate people about what screening tests can and cannot do, and she blew it. But she could redeem herself and her piece. Copeland could write a revision explaining her error and why expectant parents shouldn’t make the same error that she did. That would increase the public understand of this specific area of science, which, presumably was the reason for writing the piece in the first place.