"One in four premature babies 'faces the risk of autism'," reports the Daily Mail. The Daily Express also covers the story, saying that those who are the smallest at birth are the most vulnerable. Both newspapers go on to say that this could explain the rise in the number of children with autism in recent years. They add that the estimated amount of children with autism has risen concurrently with the numbers of babies born prematurely who survive into adulthood and attribute these higher survival rates to advances in medicine. The Daily Mail also says there are more preterm births as there is a "trend for older women, for whom premature babies are more likely, to give birth". This story is based on a study of ill, premature babies. It showed that of the 91 toddlers (aged between 18 and 24 months) 23 (26%) had social and behavioural dysfunctions that were similar to those seen with autistic spectrum disorders. It is worth noting that the researchers did not perform any actual autism diagnoses.

The babies in this study were a high-risk group who had been selected using specific criteria, so these results are not generally applicable to the wider population of preterm infants. This, in addition to the fact that the researchers were screening for autism and not diagnosing it, means that more research is needed before we can understand the true level of risk associated with prematurity. The majority of women who have preterm babies have a healthy, happy pregnancy and birth. This research does not change that picture.

Where did the story come from?

Dr Catherine Limperopoulos and colleagues from McGill University and Harvard Medical School, Boston carried out the research. The study was funded by a grant from the National Institutes of Health. It was published in Paediatrics, a peer-reviewed medical journal.

What kind of scientific study was this?

The study was a retrospective cohort study that aimed to screen preterm babies for early autistic features and to identify the clinical risk factors associated with a positive screening result. The babies included in the research were originally part of a previously published study by some of the same authors.

From the original research, 103 preterm babies who weighed less than 1500g at birth were selected for possible inclusion in this study. The babies had no chromosomal disorders, known impairments or obvious physical problems. By the time of the second study, some of these babies had died or the parents could not be reached. In total, 91 toddlers aged between 18 and 24 months were included in a round of standardised developmental outcome testing.

The babies were not a healthy population. A third showed evidence of chorioamnionitis and the group had a high average SNAP-II score (which indicates poor health after birth). The follow up tests for the toddlers included the Modified Checklist for Autism in Toddlers (M-CHAT). This is a 23-point, yes/no questionnaire completed by parents. It assesses sensory responsiveness (reaction to sound and touch), early language and communication, social relatedness (imitating parents) and whether the infant can follow a pointed finger to an object across the room.

Other questionnaires at follow up included the Child Behaviour Checklist and the Vineland Adaptive Behaviour Scale. Information about their demographics and medical history, including maternal data, were collected through reviewing medical charts.

As part of the original study, the babies were given an MRI scan before they were discharged from the intensive care unit.

The people who carried out the tests were unaware of the child's medical history and their MRI findings. When they had all the information, the researchers used statistical techniques to compare those babies who had positive screening tests for autism with those who did not. They did the same with measures of functioning and motor skills.

What were the results of the study?

Of the 91 toddlers who had been born prematurely, 23 (26%) had positive autism screening scores. Twenty nine per cent of toddlers had functional delays in motor abilities, 19% had delayed daily living skills and 23% had communication problems.

Further analysis revealed that gestational age, birth weight, male gender, placental inflammation (chorioamnionitis) and severity of illness on admission were all associated with abnormal M-CHAT scores. There was no association between abnormal MRI and M-CHAT scores.

What interpretations did the researchers draw from these results?

The researchers conclude that their study has described a high prevalence of features of autistic spectrum disorders among "survivors of extremely premature birth".

What does the NHS Knowledge Service make of this study?

This cohort study provides evidence that toddlers who are born prematurely experience some developmental delays and other impairments that may be similar to those seen with autism spectrum disorders. There are three major points to keep in mind when interpreting these results:

- The researchers do not diagnose the babies as having autism. The study found that premature babies were at higher risk of having autism-like features, not autism itself. Further follow up of these children with tools that are used to diagnose autism (e.g. Autism Diagnostic Interview) is needed to see how many of them actually develop autism. The researchers say that the presence of developmental delay in their sample (which is expected in very preterm infants) may have contributed to the high prevalence of positive M-CHAT scores.

- The population of preterm babies in this study were a 'selected high-risk' group. As the researchers acknowledge, their findings may not apply to healthier preterm populations.

- The researchers say that because the M-CHAT is designed primarily for screening toddlers at about 18 months of age, the use in an 'older' sample here may not be appropriate. They add: "It is possible that the sociobehavioural deficits identified in this study are transient or, conversely, may emerge or increase over time."

Overall, this descriptive study provides little information that can be generalised to the wider population. It is unclear how positive screen tests in a high-risk population translate into actual diagnoses of autism. Early identification of autism is an important area, but further studies are needed to evaluate the risk of autism in all preterm babies and to decide which screening tests or tools are likely to be the best predictors of a diagnosis of autism.

Links to the headlines

One in four premature babies 'faces the risk of autism'. Daily Mail, 07 April 2008

Links to the science

Positive Screening for Autism in Ex-preterm Infants: Prevalence and Risk Factors.
Study Limperopoulos C, Bassan H, Sullivan NR, et al.
Pediatrics 2008; 758-765

This news comes from NHS Choices