Weak link between grandmums' smoking and autistic grandkids

Genetics and stem cells

"Smoking in pregnancy hurts your grandkids by 'increasing their risk of autism'," The Sun brashly reports. Researchers looked at data spanning multiple generations and reported a link between girls with autism…

"Smoking in pregnancy hurts your grandkids by 'increasing their risk of autism'," The Sun brashly reports.

Researchers looked at data spanning multiple generations and reported a link between girls with autism symptoms and having a maternal grandmother who smoked.

They looked at data from more than 14,000 children, which included autism-related behavioural traits, such as poor social communication skills, and whether or not their grandmother smoked in pregnancy.

The results give quite a confusing and mixed picture. Girls whose grandmothers had smoked in pregnancy had increased likelihood of certain traits such as poor social communication skills and repetitive behaviours.

However, this link was only found if the girl's own mother had not smoked in pregnancy. And there was no such link for grandsons, although there was an increased likelihood of grandsons being diagnosed with autism if their grandmother smoked.

The study failed to look at a plethora of other factors that could potentially play a role in autism spectrum disorders. These include parent and child diet, parental alcohol consumption, exercise, weight and genetic influences.

So it's wise to interpret these results with a healthy dose of scepticism – although it remains the case that you should never smoke during pregnancy. Doing so increases the risk of stillbirth, premature birth, and the risk of the child developing asthma in later life.

Read more information about why you should stop smoking in pregnancy.

Where did the story come from?

The study was carried out by researchers from the University of Bristol and was funded by the UK Medical Research Council, the Wellcome Trust and the Escher Family Fund/Silicon Valley Community Foundation.

The study was published in the peer-reviewed journal Nature Scientific Reports, and is open-access, meaning you can read it free online.

The UK media's reporting on the story was generally accurate; making it clear that the study looked at behavioural traits linked to autism and not autism diagnoses as such.

However, it was inaccurate to report that a girl would be "67 per cent more likely to suffer poor social communication skills and repetitive behaviours" if her grandmother smoked during pregnancy. This risk was only found for poor social communication skills.

And as is so often the case, the headlines were far less subtle or precise than the actual reporting, such as The Sun's "GENERATION MAIM Smoking in pregnancy hurts your GRANDKIDS".

What kind of research was this?

This was an analysis of data from a long running UK cohort study of children. Researchers wanted to explore whether a child would be at increased risk of autism if their mother or father had been exposed to their own mother's (the child's grandmother's) smoking during pregnancy.

Autism spectrum disorders (ASD) are long-term developmental conditions characterised by difficulties with communication and social interactions and often a preference for set patterns and routines.

The cause(s) of ASD are not established. Many experts think a combination of genetic and environmental factors may be involved.

This type of research can be informative as it makes use of a very large group of people and can ask multiple questions, including about smoking, and measure multiple health outcomes, including ASD traits.

However, lots of hereditary, environmental and lifestyle factors might contribute to risk of ASD. When the causes are unknown, it is difficult to take all these factors into account and prove that a single one – in this case a grandmother's smoking – causes ASD.

What did the research involve?

The research included a large cohort of 14,062 children from the Avon Longitudinal Study of Parents and Children (ALSPAC) in the UK.

They looked at whether the child's mother or father had been exposed to their own mother's smoking during pregnancy, and whether this child would be at increased risk of ASD.

The researchers looked at certain characteristic traits of ASD, including:

  • Social communication: assessments on a 12-point scale looking at social and communication problems.
  • Repetitive behaviour: looking at mother's response to questions such as 'how often does she/he repeatedly rock his/her head or body for no reason?'
  • Speech coherence: looking at aspects of communication including speech, ability to form sentences, and whether they interpret language too literally.
  • Sociability temperament – looking at four traits: emotional reaction, social interaction with others, shyness and sociability.

They also looked at actual autism diagnosis (diagnostic criteria not described).

They adjusted for the following potential confounders:

  • the year the grandparents were born
  • grandmother's age when the study parent was born
  • number of children the grandparents had
  • sociodemographic factors (e.g. education and occupation)
  • ethnicity
  • sex of the grandchild

They also reported on whether the child's own mother smoked or not during pregnancy.

What were the basic results?

After adjusting for confounding variables, the results showed that maternal grandmother smoking in pregnancy was linked with ASD traits:

  • Social communication: Among maternal grandmothers who smoked in pregnancy, granddaughters were 67% more likely to have a high score (odds ratio (OR) 1.67, 95% confidence interval (CI) 1.25 to 2.25). This was only found when the girl's own mother did not smoke. There was no link for grandsons.
  • Repetitive behaviour: Among maternal grandmothers who smoked in pregnancy, granddaughters were 48% more likely to have a high score (OR 1.48, 95% CI 1.12 to 1.94). This was again only found when their own mother did not smoke, and not in grandsons.

No links were found for speech coherence and sociability temperament.

When combining all grandchildren whose maternal grandmother smoked, there was a 53% increased likelihood of them being diagnosed with autism (OR 1.53, 95% CI 1.06 to 2.20). However, this particular finding was only statistically significant for grandsons.

How did the researchers interpret the results?

The researchers conclude that they found "an association between maternal grandmother smoking in pregnancy and granddaughters having adverse scores in Social Communication and Repetitive Behaviour measures that are independently predictive of diagnosed autism. In line with this, we show an association with actual diagnosis of autism in her grandchildren. Paternal grandmothers smoking in pregnancy showed no associations."


This study aimed to see whether smoking in pregnancy is linked with some traits of ASD in the smoker's grandchildren.

Although this was based on a large cohort of children, the results give quite a confusing and inconclusive picture. To be frank, the study raised more questions than it answered.

Maternal grandmother smoking was linked with ASD traits only in girls (in whom ASD is less common in any case) – and then only if their own mother did not smoke. When looking at actual diagnosed cases of autism, the link was only found in boys.

The study had some important limitations to consider:

  • Most of the data was on behavioural traits, not actual diagnosed ASD, which cannot necessarily be directly linked with autism diagnoses.
  • The causes of ASD aren't known. Although the researchers attempted to adjust for some confounding variables, many other environmental and lifestyle factors could be having an influence.
  • ASD traits and autism diagnoses were only found when their own mother had not smoked in pregnancy – which indicates that it might not be smoke exposure that directly increases risk of ASD.
  • The results rely on reports from parents on their own parents, which may have been subject to recall bias if they could not remember all the facts. Some may not have known with certainty if their own parents had smoked during pregnancy.
  • Although it was a large sample, it was not very diverse with most grandparents assessed being of a white ethnic background. This may make findings less relevant to other ethnic backgrounds.

Overall, the mixed findings of this study do not provide any further answers to the causes of ASD.

What is known with certainly is that smoking in pregnancy increases the risk of stillbirth and premature birth, and later in the child's life sudden infant death syndrome and asthma.

Read more advice about quitting smoking if you are planning for a pregnancy, or have become pregnant. 

Article Metadata Date Published: Tue, 15 Aug 2017
Author: Zana Technologies GmbH
NHS Choices