Here is the penultimate section of Chapter Two’s exposition on prenatal genetic testing. The previous installments of this serialized book can be accessed at this link. In this second-to-last section, I detail how, for all the marketing claims that tests like MaterniT21, Harmony, et al. are “99% accurate”, for moms 30 and younger, they really are as accurate as a coin flip.
“99% Accurate”
All of the cell free DNA screening labs emphasize that their respective tests are “99%” accurate. From comments on my blog, this marketing choice has successfully convinced health care providers on the accuracy of MaterniT21, Harmony, etc., with expectant moms leaving comments confused about whether its is a true positive or if there is a chance of being a false positive:
- “My Dr. said the tests are 99% accurate and if there’s an abnormal test, it means an abnormal pregnancy.”–OliviaR
- “I am a little confused, talking with ob and prenatal specialist they have assured me that my baby has DS and that the test is 99% accurate?” — Sarah
- “my dr fully backed the test as infallible with a 99% accuracy.”–Stephanie
Though the “99%” representation is technically correct, it is misleading.
If an expectant mom is offered prenatal screening, and, as many if not most do, they review a pamphlet from the cell free DNA screen their obstetrics clinic uses, and that pamphlet highlights that the testing is “99%” accurate, and their OB or genetics counselor tells them it’s “99% accurate” when the test result comes back for a condition, like Down syndrome, what should be expected that the mom believes are her chances for having a child with Down syndrome? Ninety-nine percent, right?
Sequenom goes one step further. Cell free DNA testing is a screening test and always has the chance of being a false positive or a false negative. Nevertheless, Sequenom highlights in its marketing that it provides a definitive “positive” or “negative” test result. Coupled with the advice that it’s 99% and that a patient’s MaterniT21 report from the laboratory says “positive” for Down syndrome, it should be expected that the expectant mom believes the child she’s carrying actually has Down syndrome.
In truth, however, it is almost never the case that any of the cell free DNA screening tests, not MaterniT21, not Harmony, not verifi, not Panorama, not LabCorp’s InformSeq, not any of them are 99% positive for Down syndrome. Here’s why (and hold on, because here comes some math).
Hmm. Not really “99% Accurate”.
All screening tests do is recalculate the baseline chances for a tested-for condition being present in a pregnancy. As this book focuses on Down syndrome, we’ll use that as our example. The incidence of Down syndrome has a direct association with the mother’s age: the older the mother, the higher the odds are that she will have a pregnancy positive for Down syndrome. But, those odds still remain in the single digits even into the forties.
For purposes of using round numbers, let’s use as an example a 30-year old mother. At that age, her baseline chance is approximately 1-in-1,000, meaning, out of 1,000 pregnancies, one will actually be carrying a child with Down syndrome. What “99% accurate” means is that of those pregnancies actually carrying a child with Down syndrome, the cell free DNA screen will detect 99% of them. This is referred to as the test’s “sensitivity” (why, I’m not sure. It used to be the “detection rate”). Back to the 30-year old mom, let’s enlarge the number of 30-year olds that are pregnant to 100,000. With a chance of 1 out of 1,000 actually having a child with Down syndrome, that means of the 100,000 30-year old pregnant moms, 100 of them will actually be carrying a child with Down syndrome. “99% accurate” means that when those 100,000 30-year old moms are screened, MaterniT21, etc., will correctly detect 99 of the 100 actually carrying a child with Down syndrome. This also means that 1 of the pregnancies actually carrying a child with Down syndrome will not be detected and will be reported as a “negative,” but it will be a false negative.
What about the other 99,900 30-year old moms who are not carrying a child with Down syndrome? Cell free DNA screening is even more accurate in screening out those pregnancies that are not carrying a child with Down syndrome, with that rate being as high as 99.99%. This is known as the screening test’s “specificity” (again, why, I’m not sure; it used to be referred to as the “false positive rate,” which we’ll soon see why). Back to the 99,900 30-year old pregnant moms who are not carrying a child with Down syndrome. The specificity rate will correctly rule out 99,800 of those as “negative,” i.e. not carrying a child with Down syndrome. But that also means that 100 of those not carrying a child with Down syndrome will receive a false positive result.
Graphically, this Bayesian analysis (named after the English mathematician Thomas Bayes who developed the analysis), is represented like this:
| 100,000 30-year old moms | 99% Detection Rate | 99.99% Specificity |
| + Positive | – Negative | |
| 100 30 y/o moms carrying a child with Down syndrome | 99 | (1) |
| 99,900 30 y/o moms not carrying a child with Down syndrome | (100) | 99,800 |
The parenthesis indicate false positives and false negatives.
Referring to the table, it presents a challenge to the expectant mom (and frankly to the OB and genetics counselor). They all have been told by the labs that the cell free DNA screening test is “99% accurate” and even higher for ruling out pregnancies with tested-for conditions. The lab report says “positive” (if it’s Sequenom’s) or “high risk” or “99%” and a result for Down syndrome. But here’s the thing: if it’s a 30 year old mom with the “positive” result, does that mean she’s one of the 99 true positives, or is she one of the 100 false positives? Based on her test report, her chances are essentially 50/50 whether her pregnancy is actually positive for Down syndrome. Put another way: 99% accurate, in her case, means she has a 50% chance that her child has Down syndrome, not a 99 out of 100 chance.
The “50%” is the figure that matters. It is what is referred to as the “positive predictive value” or “PPV.” The PPV is the “accuracy” that expectant couples are concerned about, not the relative sensitivity or specificity of the test. However, because in the rollout of cell free DNA screening, the labs’ marketing departments chose to highlight the screen’s 99% sensitivity and specificity, practitioners and patients alike misunderstood that figure for the one that matters: the screen’s PPV.
Because of the relation of the mother’s age and the chance for Down syndrome, this relationship further means that for every mother under the age of 30 who has no other indication of an increased chance for Down syndrome, a “positive” cell free DNA result means that odds are, their pregnancy is not positive for Down syndrome. As the age of the mother rises, so do the actual chances for having a child with Down syndrome, but still, even a 40-year old mom receiving a “positive” cell free DNA screen result would still have a 7% chance of it being a false positive, not just a 1% chance. The labs’ representations that their cell free DNA screens are “99% accurate” are misleading, and purposefully so in order to make their offerings more appealing to providers and patients.
Hi Mark! I came across your site and it is unbelievably informative, I’ll be recommending you everywhere. I’m 41 y/o and recently had my blood drawn for Panorama and I’m waiting for the results. My question to you is, what is the true value if my test comes back as “low risk”? Can I trust a “low risk” result? Or do you recommend a diagnostic tests even with a low risk result? Thanks Mark!
While false negatives remain a possibility, they remain incredibly rare, even as the baseline chance increases with maternal age. According to this online calculator, based on your age, a negative result for Down syndrome would have a greater than 99% chance of being a true negative. Even diagnostic tests like amnio have false positives and false negatives, also at the likelihood of greater than 99%. I haven’t seen research that compares the accuracy for a cell free DNA screen negative versus the chance of a false positive from amnio, but both are in excess of 99%. My point being that the risk of a false negative cfDNA screening likely approaches the chances of a false positive amnio, which could equate the two risks making an amnio no more reliable than a negative cfDNA. Of course, discuss with your practitioner and/or a genetic counselor if a greater than 99% chance that a negative cfDNA is a true negative is something you still have questions about. And, thanks for the compliments for the site.
Hi Mark! I got my results back from Natera’s panorama test (42 year old at time of EDD) and it’s all low risk. They put my risk *before test* at 1/38 for trisomy 21 and of course the other ones as well, but my risk *after test* they put all of them at <1/10,000 which seems to be their average and not very personalized.
However, my doubt comes from my fetal DNA fraction, which was 3.8% at 11 weeks 3 days, and it seems incredibly low to me as they are supposed to test from 4%. My doctor assures me the test would have come back inconclusive if the fetal fraction had been insufficient. Do you have any opinions, information or articles available about Panorama test performance with "low fetal fraction"? I am very concerned about this 3.8% number. Thanks Mark!
Per the guidelines, labs are to report the fetal fraction and no calls should be considered at higher risk. Your report included the fetal fraction, which complies with current guidelines, and it was not a no-call, but a screen negative. Your physician’s advice is consistent with the guidelines. But, these are YOUR test results and if you have questions about it, Panorama should answer them. If you have questions about the test result, you should have your doctor ask them of Panorama to get the information. He or she may learn something new in the process as well.
Thanks for your response Mark! I have another question, your site has been so helpful to me I would like to get involved by volunteering, aside from donating of course, however, I live in Mexico where this is still a stigma of the worst kind. Do you know of any trustworthy organizations in Mexico where I can offer my help, or could you connect me with someone who knows? I am a teacher by training and working on my Masters in cognitive sciences, so I feel there’s a lot I can offer. Thanks again for all your help and your responses, you truly are an outstanding and exceptional human being.
Thank you for your kind words. I am not familiar with an organization in Mexico, but Down Syndrome International lists the following organization as it’s recognized partner organization in Mexico: Fundacion John Langdon Down DSi.