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

birth problems and fertility treatment

A study in PLOS ONE found children born after IVF treatment had a greater risk of birth complications. This before the headlines analysis accompanied a roundup


Title, Date of Publication & Journal

Perinatal outcomes by mode of assisted conception and sub-fertility in an Australian data linkage cohort

Jan 8 2014, PLoS One


Claim supported by evidence?

The paper does not support the claim that fertility treatment increases the risk of adverse birth outcomes.

This paper does not support the claim that there is a difference in risk between frozen embryo cycles and fresh embryo cycle based assisted conception



  • The study compares spontaneous conceptions with assisted ones: differences: differences observed could be caused by the treatment or the underlying condition that led couples to seek treatment
  • The conclusion of a difference between frozen and non-frozen embryo based methods is statistically flawed


Study conclusions, strengths & limitations

No causal claim: differences confounded by parental risk factors

Both article and press release rightly steer clear of inferring a causal effect of fertility treatment. This is because the paper’s principle analysis compares assisted conceptions to spontaneous conceptions. There could be differences between couples who seek assistance and those who do not, and those differences could be causing the differences in outcomes observed in this study.

The press release states: “the risk of serious complications such as stillbirth, preterm birth, low birth weight and neonatal death is around twice as high for babies conceived by assisted reproductive therapies compared with naturally conceived babies.”  This is a fair assessment of the data.  But it does not indicate a difference in outcomes from different treatments; nor does it demonstrate that any treatment is the cause of the higher risk.

The authors also state in their conclusions that they cannot rule out the effect of parental risk factors (e.g. lifestyle or genetic): “The contribution of maternal risk factors (other than maternal age) to the observed treatment effects is unclear, as we did not adjust for infertility aetiology.”


No evidence of difference between frozen/non-frozen

The authors make a basic statistical mistake in claiming that there is a difference between two forms of assisted conception.

They observe a difference between spontaneous conception and frozen-embryo based ICSI or IVF.

They fail to show a statistically significant difference between spontaneous conception and fresh-embryo based ICSI or IVF (partly because they have many fewer of these births). They conclude that frozen & fresh are different.

There is no statistically significant difference between frozen & fresh embryos in the risk of adverse birth outcomes. Statistical tests are available to test that hypothesis. The authors did not perform these test but a scan of the table suggests that those tests would be non-significant.

To conclude that there is a difference between these groups is akin to saying “Arsenal are >10pts of Man Utd but <10 pts ahead of Tottenham. Therefore Tottenham are doing much better than Man Utd” (At time of writing, Arsenal: 45, Tottenham 37, Utd 34).


Any specific expertise relevant to studied paper (beyond statistical)?



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