Nineteen controlled observational studies were included. The number of cases ranged from 23 to 9,175 and the number of controls from 59 to 1,690,577.
The quality scores ranged from 13 to 21 out of a possible 25. None of the studies had the specified ideal control group. Other quality problems included a lack of matching for maternal age and the timing of the infant assessment for malformations.
MM rates varied between 0 and 9.5% with IVF, between 1.1 and 9.7% with ICSI, and between 0 and 6.9% for control groups. For IVF/ICSI and singleton and multiple births combined (19 studies), there was an increased risk of MM (OR 1.29, 95% CI: 1.01, 1.67) and this was statistically significant. None of the subgroup analyses were statistically significant. The OR for MM ranged from 0.92 (95% CI: 0.75, 1.12; based on 7 studies) for IVF multiple births to 1.51 (95% CI: 0.85, 2.7; based on 8 studies) for IVF single births. There was evidence of statistically significant heterogeneity in the main meta-analysis and in all the subgroup analyses (p<0.0001).