How much of the risk do these tests actually remove?
Nobody has measured it, and there are three concrete reasons why. That absence is the most honest thing in this stage.
You have now seen a lot of numbers. Here is the one that does not exist.
No published study reports how much of a baby's remaining risk these newer tests actually remove. No study has given the full set of them to an ordinary population and then followed the children. Any figure a laboratory offers you for how much safer this makes your pregnancy is not a measurement of anything.
There are three concrete reasons, and each of them is worth understanding.
One. Nobody has assembled the stack. Prenatal sequencing in structurally normal fetuses has been studied in four small cohorts totalling under two thousand pregnancies. Newborn genome screening has been studied in fewer than six thousand babies across the two largest programmes combined. There is no group of ordinary pregnancies that received all these tests and was then followed to a defined outcome.
Two. The counting systems do not match. Birth defect registers count physical malformations. Genomic screening counts variants in DNA. Neurodevelopmental outcomes are counted by neither. You cannot score a test that finds variants against a register that counts malformations.
Three. Even the studies that exist lose track of people. Follow-up is partial almost everywhere: about half of the positives in the single-gene screening series were never followed up, and only a small fraction of parents answered the newborn screening survey.
The closest thing to an answer, and exactly what it cannot say. One 2026 study looked at 281 infants at a children's hospital in Indiana who were found in their first year to have a genetic condition, and asked backwards which prenatal tests could have found each one.
| Prenatal approach | Share of the 291 diagnoses it could have detected |
|---|---|
| ACOG-recommended cfDNA plus ACOG-recommended carrier screening | 25.4% (74) |
| Genome-wide cfDNA plus the largest commercial carrier panel | 55.3% (161) |
| No commercially available prenatal screening test at all | 44.7% (130) |
The authors' own summary: The fact that over 40% of all genetic diagnoses (and 62% of monogenic conditions) would be missed with the most comprehensive screening approach highlights the limitations of screening.
Three limits travel with that table and none can be dropped. First, the authors state it themselves: It is important to note that the yield of these tools in this population does not reflect the yield in a low-risk prenatal population.
Second, these were hospitalised infants who reached a genetics consultation, so every child whose condition appeared later or more mildly is excluded by construction. Third, every figure assumes the tests work perfectly, which the paper says in its own figure legend. So it answers one question only: of babies who turn out to have a genetic condition, how many were findable beforehand? It cannot tell you your risk.
Set the figures side by side, and do not subtract them. They come from different populations with different definitions, and the comparison is an order of magnitude, not arithmetic.
| Quantity | Figure |
|---|---|
| Babies born with a birth defect in the US, as CDC estimates it | About 1 in 33 |
| Major birth defects with no identifiable cause of any kind, in Utah statewide surveillance of 270,878 births from 2005 to 2009 | 79.8%, or 4,390 of the 5,504 major birth defects found |
| Sequencing yield when the fetus looks normal on ultrasound | 1.6%, and 0.5% for severe disease |
| Healthy newborns flagged by genome screening, after removing one common enzyme variant and one off-target gene | 0.8% |
That second row is the wall every test in this stage runs into. In a modern population study of 5,504 major birth defects, a definite cause was assigned in only 20.2%. Four out of five had no identifiable cause of any kind. No sequencing technology finds a cause that has not been identified.
None of this is an argument against testing. Sequencing a fetus with an abnormal scan can end a diagnostic odyssey before it starts. Newborn screening catches conditions where early treatment changes a life. Standard cfDNA screening does what it says for the common trisomies. Where a test is genuinely useful in the right population, take it.
The argument is about the distance between the promise and the delivery, and about which of the two you are making your decision on. ASRM wrote the closing sentence in its own document: No technology can guarantee healthy offspring.
"Give me the absolute numbers, before and after."
"What does a normal result still not rule out?"
"Has anyone published what happened to children after this test? Can I see it?"