February 12, 2018

expert reaction to prenatal ultrasonography association with autism

A new study, published in JAMA Pediatrics, investigates whether prenatal ultransonography frequency, timing, duration, or strength are associated with later diagnosis of autism spectrum disorder.

Dr Michael Craig, Reader in Forensic & Neurodevelopmental Psychiatry, King’s College London’s Institute of Psychiatry, Psychology & Neuroscience, said:

“Autism is a heterogeneous condition for which there are likely to be multiple causes. The publication of an association between the MMR vaccine and ASD in the Lancet on February 28th 1998 (almost twenty years ago) is perhaps a cautionary tale in making inferences about the relationship between association and causation in this highly complex condition.

“Rosman et al are careful to point out that ‘a large prospective cohort study will be a more definitive way to evaluate possible associations between prenatal ultrasound exposure and later risk of ASD’. It is hoped in the interim this study will not lead to a premature change in the public’s perception of the relative safety and value of ultrasound scans in foetal medicine.

“However, if the findings of Rosman et al are supported by an appropriately conducted prospective study, they may further our understanding into the aetiology of Autism Spectrum Disorders.”

Prof. Dorothy Bishop, Professor of Developmental Neuropsychology, University of Oxford, said:

“I do not have confidence that the reported findings are of clinical importance because of the following. The authors explored a very large number of variables when looking for associations. I can find no evidence that they did any corrections to take into account the play of chance. Any finding that is obtained this way needs to be replicated, preferably in a pre-registered design, before it is reported in the scientific literature, especially when it has such high potential to cause unnecessary alarm.

“Even if we draw the questionable conclusion that the effect is real, it is very small in magnitude and no different for Autism Spectrum Disorder (ASD) and developmental delay. Unfortunately the effect size is not easy to deduce from the way the statistics are reported, but as far as I can tell, the effect size is so small that one could not predict with any accuracy which children would be autistic or have developmental delay from the depth measure – particularly when one bears in mind that such prediction should take into account the relatively low prevalence of autism.

“The authors appear to jump to the conclusion that the (very weak and possibly chance) association is causal, without pausing to consider why some children have ultrasounds at deeper depths than others. I don’t know much about ultrasound, but I wonder if factors such as head size play a role. When correlations are observed it is essential to consider non-causal explanations before jumping to a causal account. This is particularly important in the case of autism, where parents are bombarded with explanations based on inadequate data almost daily.”

Prof. Basky Thilaganathan, Consultant obstetrician and spokesperson for the Royal College of Obstetricians and Gynaecologists, said:

“This small study reports reassuring findings that children with autism spectrum disorder (ASD) were less exposed to prenatal ultrasound and that the ultrasound energy used during the examination was no different compared to children without ASD.

“However, it also reports that the mean depth of penetration of the ultrasound beam is greater in those that developed ASD compared to healthy children. This is difficult to interpret, as the ‘mean depth of penetration’ is not a standard metric, but rather one that was created to overcome the significant limitations of the retrospective nature of this study. In short, ‘mean depth of penetration’ as calculated in this study is unlikely to be a good metric for exposure to ultrasound. For example, the depth of penetration would be greater in more obese women – as was the trend in the ASD group in this study (although the data was not provided). An additional and important statistical limitation of the study, is that the authors did not correct for multiple comparisons – a process that often throws up spurious statistical associations.

“This paper does not suggest that ultrasound exposure is associated with ASD. On the contrary, its findings are reassuring in that children with ASD had shorter scan duration and exposure.

“Currently, no evidence exists to question the safety of ultrasound and it is important that women are not alarmed by the results of this study and should continue to attend scheduled ultrasound appointments with her doctor to monitor the health of her baby.”

Dr Abigail Thompson, Research Leader, Autistica, said:

“This study looks at links between ultrasound scans in pregnancy and autism, which is a relatively under-investigated area. The researchers report that in children who were later diagnosed as autistic, the ultrasound waves penetrated deeper in the mother’s stomach compared to children who developed typically or were diagnosed with development delay.

“As the scientists note, several factors make it difficult to draw any definitive conclusions from this finding. The mothers of children later diagnosed as autistic were more likely to be obese in comparison to the mothers whose children were diagnosed with developmental delay, which may account for a greater depth of ultrasound wave used between these groups.

“In addition, a number of factors which have separately been linked to autism were not well characterised in the study. The mothers of autistic children were more likely to be older and more likely to seek prenatal care later, which has been linked to autism. Other important factors such as maternal smoking, which has been linked to autism, were not included within the study. For these reasons, although the study was good quality, it is not possible to draw firm conclusions about any risks from ultrasound scans. Future studies with larger sample sizes are needed.

“There are already guidelines in place around the use of ultrasound in pregnancy which doctors should continue to follow. This study alone is not sufficient to suggest that pregnant women should be concerned about ultrasounds.”

Dr Christoph Lees, Consultant in fetal maternal medicine, head of a London fetal medicine unit and Chair of the International Society of Ultrasound in Obstetrics and Gynecology Safety Committee, said:

“This modest sized retrospective study purports to show a relationship between prenatal ultrasound and autism. Women should be reassured that it shows no link at all between duration of ultrasound exposure and autism, nor between ultrasound energy used and autism.

“One problem with the study design is that the authors don’t tell us ‘up front’ what the key comparisons they are looking for are. The authors choose many comparisons including an invented and previously unknown index: time weighted mean depth of ultrasound. They don’t describe this time-weighted mean depth of ultrasound index – as a group of clinicians and scientists closely involved in all aspects of ultrasound safety we have no idea what they are describing. In short, this ultrasound index has no obvious scientific or biological justification.

“Further, the authors have not corrected for their many comparisons: you don’t have to be a statistician to know that if you toss a coin enough times you’re bound to get a result you are looking for.

“In summary, autism and prenatal ultrasound are both really contentious areas that easily lead to unwarranted alarm for women and false messages. This study, if read carefully is in fact very reassuring in showing no apparent link between pregnancy ultrasound and risk of autism. It’s a shame that the paper as it is presented is open to some misinterpretation which we feel sure neither the authors nor journal would have wanted to see.”

Prof. Andrew Shennan, Professor of Obstetrics, Clinical Director South London CRN, King’s College London, said:

“Ultrasound scans on babies have been used extensively for over 30 years, with no apparent ill effects. This type of retrospective study can often miss important factors. Women with problems, including obesity will require more and longer scans. These basic factors have not been taken into account and the reason for the scan is potentially more likely to be related to the autism. Scans are often needed to help pregnancies and this evidence should not change this. Prospective studies are needed.”

Prof. Joyce Harper, Professor in Genetics and Human Embryology, Institute for Women’s Health, UCL, said:

“Unfortunately there are relatively few studies looking at the effects of ultrasound on the developing fetus. In this study of only 107 patients with autism spectrum disorder, the researchers did see a significant difference in the depth of ultrasonographic penetration.  But since this is a preliminary study with a relatively small number of patients, I think it is too early to draw firm conclusions.  The use of ultrasound is key to detect numerous abnormalities of the fetus.

“I do not think this study should stop women deciding to have an ultrasound, but further studies are urgently needed, especially to determine the effects of ultrasound on the fetal brain.  With the increase in the prevalence of autism spectrum disorder over the last decades, research is looking into the exposure of toxic environmental factors during pregnancy.”

Ass Prof Nadja Reissland, Department of Psychology, Durham University, said:

“This study claims that ultrasound scans are associated with Autism Spectrum Disorder (ASD). In fact depth of the ultrasound scans tested was associated in the same way with ASD as with developmental delayed (DD) individuals (depth: 13.4 (13.1-13.8) vs 13.4 (13.1-13.8)). Hence the claim needs to be revised.  The authors claim that since there is an increase in ASD which coincides with an increase in ASD diagnosis that there is an obvious link. They do not comment on the fact that it might be the better methods of identifying ASD which lead to the increase in diagnosis. The latest prevalence studies of autism indicate that 1.1% of the population (including adults) in the UK may be on the autism spectrum (http://www.autism.org.uk/about/what-is/myths-facts-stats.aspx.). Given that adults would not have been exposed to ultrasound scans (the authors claim it is a recent phenomenon) their claim that ultrasound is the cause is incorrect by their own argument.

“Furthermore, of note is that the indices used by Boston Medical Center to identify what would be acceptable use of ultrasound, including mechanical and thermal indices and length of exposure are in fact all lower, rather than higher, in the ASD compared to the typical group. The only difference they found was depth of exposure, which was the same for “ ASD” as “DD”  individuals. Hence the claim that autism is associated with ultrasound is a partial report and certainly should not be publicized like that. In fact the ASD group compared to the typical group had fewer number of scans (mean 5.9 vs 6.3), shorter exposure ( (3128.4 seconds (2807.5-3449.3) vs (3508.0 (3272.7-3743.4)) and the same depth of exposure  as the developmentally delayed group (ASD: depth of ultrasonography, cm 13.4 (13.1-13.8) DD: 13.4 (13.1-13.8).

Additionally, their conclusion that it could be “heating” cannot be sustained since the thermal index was the same for all three groups and the Doppler exposure was lower for the ASD group (mean 177.4) compared to the typical groups (189.0)

“Hence rather than claiming an association of depth of ultrasound with ASD the paper should more accurately state that depth of ultrasound has been associated with ASD and Developmental delay in this study and that a major factor, which the authors cite themselves as a “limitation” was that they did not report “maternal smoking, a variable associated with increased ASD risk”. A recent meta- analysis by Young et al 2017 (Scientific reports 10.1038/s41598-017-04413-1) shows that using population-level smoking metrics uncovers significant relationships between maternal smoking and ASD risk. Nor did they report alcohol consumption which has been found to be associated with neurobehavioural disorders including autism (e.g. Varadinova & Boyadjieva, 2015). Other factors not cited were the type of ultrasound scanners used as well as whether transvaginal and/or transabdominal scans were performed which significantly affect exposure.”

* ‘Association of Prenatal Ultrasonography and Autism Spectrum Disorder’ by N. P. Rosman et al. published in JAMA Pediatrics on Monday 12^th February 2018.

Declared interests

Prof. Dorothy Bishop: “No conflict of interest”

Dr Imarisio: “None”

Prof. Andrew Shennan: “No conflicts”

None received.