select search filters
roundups & rapid reactions
factsheets & briefing notes
before the headlines
Fiona fox's blog

IVF and autism

In a JAMA study of more than 2.5 million children born in Sweden, in vitro fertilization (IVF) treatment was associated with a small but statistically significantly increased risk of mental retardation, though not associated with autistic disorder. A roundup and this before the headlines analysis accompanied an SMC briefing.


Title, Date of Publication & Journal

Autism and Mental Retardation Among Offspring Born After In Vitro Fertilization

1 July 2013



Claim supported by evidence?

This large population-based, prospective cohort study provides evidence that supports the safety of IVF in terms of potential for intellectual disability and susceptibility to autism, particularly for single embryo transfers. However, it also provides some evidence for a link between a treatment for male infertility and a small but significant increased risk of intellectual disability and autism.



  • This is a large study with sufficient power to identify any major risk for IVF in terms of potential mental retardation in the offspring.
  • Several possible risk associations with differing procedures were investigated and as expected a few were statistically significant. Several associations disappeared when multiple births were excluded.
  • The use of intracytoplasmic sperm injection (ICSI) is worthy of further investigation. But the small numbers make interpretation difficult at this stage.


Study Conclusions

  • The very large size of the study allows the investigation of two relatively rare events: IVF and autism/intellectual disability. In the IVF cohort 103 autism cases and 80 cases of intellectual disability provide sufficient numbers to give sufficient statistical power.
  • The prospective cohort nature of the study makes relative comparison between IVF and spontaneous conception well supported.
  • However, any differences in rates for deleterious traits in the offspring may result from the types of individual requiring IVF, rather than the impact of the procedure. The major ones, mother’s and father’s age and their psychiatric history at the time of the birth, have been controlled for in the figures but not the tables.



  • The study presents rates per 100,000 person-years. These are diagnosis rates and used to statistically allow for subjects entering and leaving the cohort study (Table 2). The most robust conclusions come from the resulting relative risks. The crude percentages are less reliable as the denominator is not adjusted for the length of time observed.
  • They describe the results of an analysis that just looks at singleton births. In many cases this means that previously observed differences became not statistically significant. This may be because the sample size is reduced, but it may be because the added risk is caused by the increased risk of multiple births with IVF.
  • As expected parental age is very different between IVF and spontaneous births. Controlling for these factors is very important. Studies have suggested autism is more prevalent for older fathers.
  • Autism is described as “narrow diagnosis of infantile and childhood autism (diagnostic codes ICD-9 299A or ICD-10 F84.0)” but the rate in the text 1.5% (103/6959) is high whereas the figure in table 2 is a much more typical 0.33%, compared to 0.27% in spontaneous conceptions.


‘Before the headlines’ is a service provided to the SMC by volunteer statisticians: members of the Royal Statistical Society (RSS), Statisticians in the Pharmaceutical Industry(PSI) and experienced statisticians in academia and research.  A list of contributors, including affiliations, is available here

in this section

filter Headlines by year

search by tag