There is a new study related to fetal testing making the news today. It is a great chance to track base rate misconceptions in action.
"A pregnant woman of 29 can get just as much benefit (without added cost) from testing as a 35-year-old woman." says the study's author. Parsing words, the study's author claims she is not recommending mandatory genetic testing, but rather pointing out that some of the current guidelines are arbitrary. The press is unlikely to understand the subtle points:
- "Doctors Say Scrap Age Guidelines for Down's Tests" reports Patricia Reaney at Reuters
- "An American study has recommended that accurate testing for Down’s syndrome babies should be extended to all pregnant women, not just those over the age of 35." leads the Scotsman
- "Headline: 'Making Better Babies' . . . now a new study suggests all women, regardless of age, could benefit from tests such as amniocentesis. " - - Joanna Schaffhausen, ABCNews.com
I think they are missing the "not a recommendation" point pretty clearly. Instead they play up "Doctors recommend", "benefit", and "all women, regardless of age". That in itself is a funny example of our human inability to track the subtle background when something stands out in the foreground - in this case, medical authority, benefits, and applicability to all pregnant women. Moreover, the press seems to miss that at the root of the study is a "cost benefit analysis", not a medical recommendation. Essentially, the study says that pregnant women are so concerned about Down's syndrome they are willing to overlook the complications involved with diagnosis. It's a classic case of human's inability to track the base rate when staring at an immediate concern. Rather than having medical professionals explain the principles behind the Hippocratic oath, give the patients what they want!
It is true, when you do read the coverage carefully, you can make out that the researchers are neither advocating additional testing, or claiming it would be medically wise to test all pregnant women for genetic defects. They are claiming that women are so concerned about having a baby with Down's syndrome that they'd prefer additional genetic testing, even when the odds of a complication from the testing are higher than the odds of a defect. They don't really go into educating vulnerable pregnant women about base rate errors. Instead, they describe the issue in terms of a woman's right to health care and her "informed consent".
How this study gets covered in the press is a great chance to see base rate misconceptions illustrated. Joanna Schaffhausen at ABC News points out that:
Both amniocentesis and CVS have small associated risks for the mother and her baby. For example, the risk that an amniocentesis could cause a miscarriage is between 1 in 200 and 1 in 400. Injury to the baby or mother, infection or early labor are other potential complications, but all are extremely rare.
CVS may carry a slightly higher risk of miscarriage than amniocentesis because the procedure is done in earlier in pregnancy (sic). Infection is also a risk. Rare cases of defects in baby's fingers or toes have been reported, especially when CVS was done before nine weeks.
What this reporter doesn't explain is that Down syndrome has an incidence of about 1.3/1000 births, far less frequent than the rate of complications. She posits the complications related to testing are "extremely rare" yet the chance of an "accidental abortion" of a healthy fetus is 3-5 times more likely than the rate of Down's syndrome. That rate is even lower for the younger women who may be persuaded to stick an enormous needle into their womb in light of this news report cheerily headlined "Making Better Babies".
This strikes me as a classic case of counter-productive intervention in a pregnancy. The fact that most women in the study can't track the base rate (in the face of a terrifying potential outcome) and therefore make astoundingly bad wagers with their healthcare doesn't really strike me as sound basis for setting policy. It seems anyone could examine this matter with "cost benefit analysis" when the issue involved is literally hysterical, and even someone with a solid understanding of conditional probability could misjudge the risks involved.
It wouldn't be hard to write a study that shows "cost benefit analysis shows millions prefer profligate living now to chance of poverty later". But we don't formulate social security policy on cost benefit analysis. Naively, I'd expect the press to cover this with a bit more care. The number one hit on this blog is the post talking about quad screen tests and the terrible scare given to pregnant women about the health of their children by a medical profession untutored in statistics. Its a pity, that when the stakes are so high, with people in a fragile emotional state making complicated health decisions, the press covers something like this so carelessly.