June 9, 2022

expert reaction to observational study of mothers who tested positive for SARS-CoV-2 during pregnancy and neurodevelopmental outcomes in their children at 1 year

A study, published in JAMA Network Open, looked at mothers who tested positive for SARS-CoV-2 during pregnancy and neurodevelopmental outcomes in their children at 1 year.

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

How does this work fit with the existing evidence? 

“I’m not aware of prior evidence – it’s not really possible to study this question until larger samples of older children are available (see below).

Is this good quality research?  Are the conclusions backed up by solid data? Have the authors accounted for confounders?  Are there important limitations to be aware of?

“The authors describe this as preliminary work and note a number of limitations. A major limitation is potential for bias that could arise if mothers who had had Covid were more likely to seek medical evaluation of their infant than mothers who did not have Covid. This could induce a spurious association between Covid status and having a diagnosis. The authors note that we really need a prospective cohort study with neurocognitive phenotyping to address this question – i.e. a study where you identify groups of exposed and unexposed infants and then aim to test all of them as they develop with suitable measures to detect neurocognitive problems.

“There are also quite substantial differences between the group of mothers who had Covid and those who did not – most are controlled for, but the hospital was not – data came from 2 academic medical centres and 6 community hospitals. Table 1 does not show the distribution of cases by hospital, but it is clear that there were substantial differences between the Covid and non-Covid mothers, with more of the Covid-exposed cases attending an academic medical centre. Because the only measure of neurodevelopmental outcome was a physician diagnosis, it is entirely possible that the differences observed here are due to differences in diagnostic practices/biases by physicians in different hospitals.

“This concern is amplified by the fact that some of the diagnoses don’t make much sense in such young babies. For instance, we have over 100 diagnoses of expressive language disorder and developmental disorder of speech and language. At 12 months of age (which is the oldest age studied here), most babies are producing at most their first words, and many don’t start to talk until 15 months or later.  As an expert in children’s speech and language disorders, I find it hard to see how these conditions could be diagnosed in such young children, because even if a child is producing no words at all, that would not be abnormal.  The authors used ICD-10-CM codes, which I don’t have to hand, but I have the more recent ICD-11 definitions: this is what is stated re Developmental Disorder of Speech and Language:

“Children vary widely in the age at which they first acquire spoken language and in the pace at which language skills become firmly established. The majority of pre-school children who acquire speech later than expected go on to develop normal language abilities. Very early delays in language acquisition are therefore not indicative of Developmental Language Disorder. However, the absence of single words (or word approximations) by the age of 2 years, the failure to generate simple two-word phrases by 3 years of age, and language impairments that are persistent over time are more likely to indicate Developmental Language Disorder, especially in the context of a known family history of language or literacy learning problems.”

So it seems that at least some of the diagnosing physicians are departing from usual use of these definitions in making this diagnosis in children under 12 months of age.

“We also are given no information about the age at which diagnoses were made. The reports of disorders of motor function are more plausible at this young age: These affected 99/7550 (1.3%) of non-Covid cases, and 6/222 (2.7%) of Covid cases.  But given the other, implausible, diagnoses, I find myself unsure what one can make of these results. They rely on numeric codes for diagnoses being accurately assigned by physicians administering routine care, and the data themselves suggest that some errors in coding may have crept in.

What are the implications in the real world?  Is there any overspeculation? 

“It is worth noting that 97% of infants with no Covid exposure vs 94% of those with Covid exposure received no neurodevelopmental diagnosis. This result is suggestive of additional risk to those with Covid exposure, but for the reasons noted above, it is far from conclusive. Clearly pregnant women should do their best to avoid Covid, but if they do get it, the odds are high that their infant will not have evident neurodevelopmental problems in the first year of life.

“As the authors note, the main implication is that it would be worth doing a large prospective study of older children using standard, quantitative measures of neurodevelopment to investigate the question of whether maternal Covid in pregnancy affects the offspring.”

Prof Dimitrios Siassakos, , Professor in Obstetrics and Gynaecology, University College London (UCL), said:

“This study provides some evidence that women who tested positive for COVID-19 had babies with neurodevelopmental problems. There is no evidence that the association is causative: that COVID-19 directly led to these problems.

“It is equally, if not more, likely that the women in these cohorts who tested positive also have some characteristics associated with later problems in their offspring, independent of COVID-19.

“For example, whereas the jury is still out whether COVID-19 can harm babies directly, we know for certain that diabetes in pregnancy does, particularly if not diagnosed and treated. It is possible that un-diagnosed diabetes is the hidden culprit; it could mean women in these cohorts studied are more likely to test positive for COVID, and to have babies with later problems.”

Prof Marian Knight, Professor of Maternal and Child Population Health, University of Oxford, said:

“This observational study used coded hospital data from six hospitals in the US. The authors report, based on only 14 infants with a diagnosis code, that infants whose mothers had a recorded positive test for SARS-CoV-2, were more likely to have a code for a neurodevelopmental disorder. There are a number of potential explanations for this observation. We do not know the reason women were tested for SARS-CoV-2 – many hospitals routinely test women on admission to hospital, therefore women who already had a pregnancy problem are more likely to be tested. This means SARS-CoV-2 infection is more likely to be detected in these women, even if it has no impact on their pregnancy. The authors only present information on a limited range of pregnancy complications (diabetes, pre-eclampsia and hemorrhage) and differences in incidence of other pregnancy complications may explain their results, rather than SARS-CoV-2 infection. Women who had a positive SARS-CoV-2 test were more likely to be cared for at an academic medical centre, which may be a marker indicating that they were a higher risk group of women, also a potential explanation of the findings.

“It is unclear what a code for a “Developmental disorder of speech and language, unspecified”, represents in infants aged 12 months or less. This is the condition reported in half the infants of mothers with SARS-CoV-2. As the authors themselves note, presence of these codes may be a reflection of closer monitoring of the infants of mothers who were unwell during pregnancy. This closer monitoring, rather than a true difference in developmental disorders, may be an explanation for the observed results.

“It is not possible on the basis of this paper, to conclude that covid-19 in pregnancy causes developmental problems for children. Different studies are needed using validated measures of development at older ages to determine whether this is a genuine concern.

“It is clear that many of the babies in this study were born prematurely, which may be part of the explanation for the findings. We know that vaccination protects pregnant women against severe covid-19 and also means their babies are less likely to be born prematurely. Ensuring they are fully vaccinated is the best thing women can do to protect their baby from adverse effects of them becoming ill with covid-19.”

‘Neurodevelopmental Outcomes at 1 Year in Infants of Mothers Who Tested Positive for SARS-CoV-2 During Pregnancy’ by Andrea G. Edlow et al. was published in JAMA Network Open at 16:00 UK time on Thursday 9th June.

DOI: 10.1001/jamanetworkopen.2022.15787  

Declared interests

Prof Dorothy Bishop: “I confirm I have no conflict of interest.”

Prof Marian Knight: “I led national surveillance of covid-19 in pregnancy and am a co-investigator on the SINEPOST study (https://www.bristol.ac.uk/translational-health-sciences/research/neurosciences/research/neurology/sinepost-study/) investigating neurodevelopmental outcomes in children after maternal covid-19 infection.”

No others received