October 27, 2026

expert reaction to study looking at night shifts during pregnancy and children’s neurodevelopmental abnormalities, as published in Occupational & Environmental Medicine

Scientists comment on study looking at night shifts during pregnancy and children’s neurodevelopmental abnormalities, as published in Occupational & Environmental Medicine.

Prof Debra Skene, Professor of Neuroendocrinology, University of Surrey, said:

“This is an important topic and indeed we need more research to look at the effects of shift work on working pregnant women.

“There are limitations to this study, and as the press release notes, this is an observational study so it cannot establish cause and effect.  The authors seem to be aware of other limitations too, as noted in the press release: ‘the researchers acknowledge that they weren’t able to account for shift rotation systems, individual chronotype, workload, exposure to light, or the timing of night shift discontinuation, all of which might have influenced the findings’.

“For me the question is, can we still consider the findings valid given these limitations?  Not having information on possible confounders such as shift rotation systems, workload, or when the women in the study stopped doing night shifts, in my view seriously undermines the validity of the findings.

“Some other points: The association of their findings with prenatal sleep length, seems like an afterthought.Was this hypothesised initially?  Or is this a post hoc finding that was not part of study hypotheses?  And secondly, they asked about mood as a proxy for sleep quality, which is not necessarily the same thing.

“Shift work involves both chronic circadian misalignment and chronic sleep deprivation.  The present study is not able to pinpoint what aspect of shift work (e.g. the misaligned circadian clocks, sleep deprivation or with abnormal sleep and meal timing etc.) could explain their findings.

“As the authors themselves say in the paper “Further research should clarify the underlying mechanisms and evaluate the roles of exposure timing, specific night work conditions and maternal sleep characteristics after accounting for potential selection effects and unmeasured confounding”.

“As it stands, we can’t know what aspects of shift work are driving their reported association”  

Prof Asma Khalil, Professor of Obstetrics and Maternal Medicine and Consultant Obstetrician, City St George’s, University of London, said:

“This large prospective Japanese study adds useful evidence on a relatively understudied question.  Its strengths include recording working patterns during pregnancy, following children to age four, adjusting for many potential confounding factors and accounting for multiple comparisons in the main analyses.

“However, the outcomes were parent-reported developmental screening results, not clinically diagnosed neurodevelopmental disorders.  Screening identifies children who may need further assessment; it does not establish a diagnosis.

“The researchers found modest associations, but this observational study cannot establish that night shifts caused developmental difficulties.  Important differences in workload, shift schedules, family circumstances and the reasons women stopped night work could still influence the findings.  Missing data and loss to follow-up also introduce potential selection bias.

“The comparison between women who continued and discontinued night shifts does not establish that stopping night shifts protects children’s development.  Finding a statistically significant association in one group and not another does not itself demonstrate that the groups differ.

“The sleep findings are exploratory and require particular caution: only 80 women both continued night shifts and reported sleeping at least ten hours.  The study does not establish that longer sleep is harmful or that changing sleep duration would prevent developmental difficulties.

“These findings warrant further investigation, but they should not be presented as proof that pregnant women working nights are harming their children.

“The press release describes an 80% higher risk associated with sleeping at least ten hours compared with six to nine hours.  In Table 3, however, 1.80 is labelled as a multiplicative interaction estimate between continued night shifts and long sleep.  It should not be presented as a straightforward comparison of long versus moderate sleep duration.”

Prof Kevin McConway, Emeritus Professor of Applied Statistics, Open University, said:

“This study is of interest, because previous research has found associations between mothers working night shifts during pregnancy and unfavourable health outcomes in their children.  That previous research did not clearly establish what might be happening.  Most of the previous research is observational, so there are problems in knowing about cause and effect.  Some of the previous work described in the new research paper is actually based on experiments in animals, and as always it’s not easy to tell the extent to which the findings in animals might apply to humans.  So there are reasons to do more research, as in the new study.

“However, I think there are several reasons for serious scepticism about the new findings.  I’ll outline them.  These reasons for scepticism don’t mean that the overall finding, that mothers who work night shifts have children with a higher chance of certain abnormalities in neurodevelopment, is definitely incorrect.  Instead they mean that I just can’t be anywhere close to sure that the finding is correct.

“That’s why I’m happy that the recommendations for changes in practice given in the paper [in the box at the top of the second page] are rather vague: flexibility in night shift schedules maywarrant further consideration, and greater awareness that there might be implications of shift work in pregnant women may facilitate workplace discussions.  It would have been wrong to be more definite on these recommendations.

“One important reason for scepticism is that the study is observational.  Therefore the reasons for the higher rate of some neurodevelopmental anomalies could be other factors, which happen to be associated with working night shifts, and not the night shifts themselves.  The researchers did take into account as many as 24 potential other factors in their statistical analysis, but we still can’t be sure that every relevant factor was taken into account, or that the way the 24 factors were included in the analysis is the most appropriate.  So, as press release and the paper make clear, we just can’t be at all sure about cause and effect.

“I think it’s important to be clear about the numbers of pregnancies involved.  The press release reports, correctly, that the data come from a study of around 100,000 pregnancies in Japan.  But not all those pregnancies contribute to the data available for the study.  Actually it could use data from only about 31,500 of the pregnancies.  The reasons for so many omissions are shown in Figure1 in the paper.  The biggest numbers of omissions are because the mothers were not working during the first 3 months of pregnancy (31,000 pregnancies) or because data were not available on the screening tests for developmental anomalies (16,000 pregnancies).

“Now the remaining 31,500 pregnancies sound like a lot, for statistical purposes, though there must be doubts about whether they remain typical of Japanese pregnancies, given all the omissions.  However, the total number of pregnancies is not the most important number for assessing the accuracy of the estimates that come from the data analysis.  We have to consider how many of the mothers actually did night shifts at all.  That’s a lot smaller – about 1,800 did night shifts only during the first three months of pregnancy (described in the paper as ‘discontinued nightshifts’), and about 2,100 did night shifts in the final six months as well (‘continued nightshifts’).  (There seem to be no women who did night shifts during the final six months but not the first three months.  The paper doesn’t explain why not.)

“Those numbers are importantly smaller than the number of pregnancies included in the statistical analysis, but it’s also important that the number of children who were classified as being at neurodevelopmental risk, on the J-ASQ-3 screening carried out at age 4, was also rather small.  In total, just over 4,000 were classified at risk on at least one of the five domains of development considered.  Because each of the domains was analysed separately, what really matters is that the numbers classified as at risk on each of the five domains was between about 900 and about 2,000.

“The fact that all these numbers were relatively low, and a lot smaller than the overall 31,500, means that there is actually a lot of statistical uncertainty in the estimates of risk.

“Thus, for instance, the press release states that the children of pregnancies where the mother worked night shifts only during the first three months had a 24% higher risk of developmental anomalies in the fine motor domain, than children where the mother worked no night shifts.  That’s indeed the figure in the paper as the central estimate of the higher risk.  But the paper also tells us that the risk difference could plausibly be somewhere else between 4% and 48%.  The study can’t actually tie down the estimates of increased risk very closely at all.

“The paper and press release make a lot of detailed statements about differences between pregnancies where the mothers worked night shifts only during the first three months and pregnancies where the mother worked night shifts in later pregnancy too.  For instance, the children of women who worked night shifts throughout pregnancy (‘continued night shifts’) showed evidence, the paper and the press release say, of abnormalities compared to women who worked no night shifts in three of the five developmental domains – communication skills and personal-social functioning as well as fine motor skills.  The children of women who worked discontinued night shifts (first three months only), showed evidence, the paper says, of abnormalities in fine motor skills.  The paper says that, in these children, the shift work was ‘not associated with a meaningful increased in the risk of developmental screening anomalies’.

“But does this hold water, if we take into account the possible statistical uncertainty in the estimates?  I don’t think it does.  For communication skills, the central estimate of the difference in risk in children whose mothers worked continued night shifts compared to mothers who worked no night shifts was 29%, but plausibly it could be somewhere else between 5% and 58%.  That’s a wide range.  For the children of mothers who worked discontinued night shifts, the central estimate of risk difference was 16%, but plausibly it could be anywhere between the risk being 11% lower in children of mothers who worked discontinued night shifts compared to those whose mothers did not work any night shifts, to the risk being 51% higher in the children of mothers who worked discontinued night shifts.

“Because, for children of mothers who worked discontinued night shifts, the risk difference could be close to zero or even smaller than the risk in no-night shift mothers, it’s certainly possible that there’s no meaningful risk increase.  But that’s just saying that we can’t rule out the possibility of no risk difference, or only a small one, not that the risk difference has to be small.  It could be quite a large difference.

“The margins of statistical uncertainty of this risk difference on communication go from 5% to 58% when comparing children of mothers who worked continued night shifts with those who worked no night shifts, and from ‑11% (that is, 11% in the opposite direction) to 51% for the comparison for children of mothers who worked discontinued night shifts.  Those two ranges of statistical uncertainty overlap a lot.

“I think that all we can really say about this is that the difference may be smaller for the women working discontinued nights shifts than those working continued night shifts, but the margins of uncertainty are so wide that we really can’t be sure whether these groups differ at all.  The same goes for other developmental domains where the paper claims there is a difference between mothers working discontinued nights and those working continued nights.And all this goes back to the fact that the number of women and the number of children showing developmental anomalies are very small.

“It’s maybe also worth considering that there is no analysis of the actual amount of night work carried out.  To count as having worked nights, the mother had to say she worked at least one night shift a month.  Just one night shift a month isn’t a lot, and in some cases it might not have gone long into the night hours.  Some women would have worked more nights than that, but this was not taken into account in the data analysis (and could not have been, really, given the relatively low number of mothers who worked any night shifts).

“I’m not going to write anything about the exploratory exploration of effect modification by the mother’s sleep duration and subjective sleep quality.  It’s interesting, but only exploratory, and complicated to pick apart (and I’ve written too much already).

“Finally, it might be worth mentioning that results from Japan, where work culture and so many aspects of culture are different from the UK, may not carry over to the UK situation anyway.”

* ‘Maternal night shift patterns during pregnancy and offspring’s development at 4 years of age: the Japan Environment and Children’s Study’ by Subaru Ikeda et al. was published in Occupational & Environmental Medicine at 23:30 UK time on Tuesday 27 October 2026

DOI: 10.1136/oemed-2026-111018

Declared interests

Prof Debra Skene: “I have no conflicts of interest.”

Prof Asma Khalil: “None related to this study.”

Prof Kevin McConway: “I have no conflicts of interest to declare.”