Birth defects after IVF, and after ICSI
A real but modest increase. The IVF signal mostly disappears after adjustment. The ICSI signal does not.
The cleanest study linked 308,974 South Australian births to a defects register, with diagnoses counted up to the child's fifth birthday:
| Group | Any birth defect | Adjusted odds ratio |
|---|---|---|
| No assisted conception | 5.8% | reference |
| IVF | 7.2% | 1.07 (0.90 to 1.26), not significant |
| ICSI | 9.9% | 1.57 (1.30 to 1.90), remains significant |
| Any assisted conception | 8.3% | 1.28 (1.16 to 1.41) |
Put in plain numbers: about 5 to 6 babies in every 100 conceived without help are born with a birth defect. After IVF that is around 7 in 100, and after ICSI around 10 in 100.
The adjustment column is the interesting part. Once the parents' age, health and the infertility itself were accounted for, the IVF excess was no longer statistically significant. The ICSI excess remained. The authors add that residual confounding cannot be excluded, which is the correct caution: couples needing ICSI differ from couples who do not.
A separate pooling of 45 cohort studies covering 92,671 assisted reproduction infants found any birth defect at relative risk 1.32 (95% CI 1.24 to 1.42), and 1.36 (95% CI 1.30 to 1.43) when restricted to single babies, so this is not an artefact of twins.