EXPERT REACTION: Is paracetamol during pregnancy linked to ADHD and autism?

This new research article provides additional evidence that a woman’s use of the common drug paracetamol during pregnancy could increase the risk of two major disorders of brain development, autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD). Both disorders result from complex combinations of genetic and environmental factors which remain poorly understood. This new evidence raises important public health issues. Can pregnant women avoid taking this drug without compromising their own health? Much more research into the causes of ASD and ADHD is required. Furthermore, if women are advised to not take paracetamol during pregnancy, then alternative treatments to this drug need to be explored, which will also require more research. 

There is considerable evidence from a variety of studies that suggests there is a link between paracetamol (Acetaminophen) exposure in pregnancy and an increased risk of neurodevelopmental/behavioural problems, such as attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorders (ASD). However, due to the nature of the evidence and potential uncertainties in its interpretation there is still considerable debate as to whether or not there is sufficient evidence to formally advise women against taking this medication during pregnancy.
 
The study by Yuelong et al in JAMA Psychiatry adds to this debate, although in my view it does not clarify the exposure-risk relationship – it just adds more questions than answers. The study team measured levels of paracetamol and its metabolites in cord blood samples in 996 mother-infant pairs from the Boston Birth Cohort, 365 of whom had diagnosable ASD or ADHD, and looked at associations with the neurodevelopmental outcomes in the offspring. They found a significant, dose-dependent relationship between cord blood levels of either unaltered paracetamol or several of its metabolites and rates of both ADHD and ASD, adjusting for major confounders such as: smoking, alcohol and drug taking in pregnancy; maternal stress, BMI, education status, marital status and ethnicity; gestational age and birthweight at delivery; intrauterine inflammation/infection and maternal fever during pregnancy.
 
The strengths of the study are that, for the first time, we have actual blood paracetamol measurements to link exposure and dose with outcome, rather than indirect evidence of exposure at some point in pregnancy. The study also is quite large, adding to its power. There are, however, some significant weaknesses:
 
Firstly, there is no information on the indications for maternal paracetamol use in the hours before delivery, so we don’t know whether or not it is the reason for taking paracetamol that is the important causal link. Secondly, there is no other information on prior paracetamol taking during pregnancy, so we don’t know whether or not the infants with high levels in their systems also had repeated exposures earlier on in pregnancy because of a pattern of regular maternal paracetamol use. Thirdly, most previous studies have focussed on mid-pregnancy exposure as a likely time for potential disruption of important neurodevelopmental processes in utero, not the peripartum period.
 
Finally, and most significantly, every sample in the cohort had detectable levels of paracetamol, raising the possibility that either a) the paracetamol assay lacked specificity and there was some sort of interference present, or b) all mothers in the cohort had taken paracetamol prior to or during labour and delivery, raising concerns around the nature of the cohort and the ability to compare exposure with no exposure.

In this carefully conducted, prospective pregnancy cohort study a dose-dependent association was demonstrated between paracetamol use in pregnancy and developmental disorders in offspring, including ADHD and ASD. 

Exposure (paracetamol use) was confirmed by measuring cord blood acetaminophen in a sample of nearly 1,000 children, and statistical adjustment was made for potential confounders including preterm birth and substance use. 

Rates of ASD and ADHD were more than doubled in children exposed to higher doses of paracetamol. 

If the relationship between paracetamol use and developmental disability is causal this is reason for concern, because paracetamol is readily available ‘over the counter’ and is likely widely used during pregnancy.  

There is biological plausibility for these findings – paracetamol readily crosses the placenta, in animal models has been shown to damage nerves in the brain, and in humans has been associated with ADHD and other disorders.  
 
A number of questions spring to mind.  First, more than two thirds of children in this large cohort had one or more developmental disabilities, a high prevalence which is not adequately explained. Second, ADHD and ASD – and indeed other forms of DD - may have a genetic component, yet there is no mention of family history of developmental disorders and no indication that genetic testing was performed. Third, ADHD, ADS and other developmental problems are frequently seen in children with prenatal alcohol exposure, including children with fetal alcohol spectrum disorder. 

Alcohol use is common in young women before and during pregnancy, especially before pregnancy recognition (50-60% in low risk Australian pregnancy cohorts and 10% worldwide). As in this study, which records only a yes or no response, clinicians often fail to ask about or document alcohol use during pregnancy in a meaningful way and women often underestimate consumption. Considering the paper reported alcohol use before or during pregnancy, the reported prevalence was low. 

This important study suggests the need for further research to confirm a causal relationship between paracetamol use and developmental disorders. Future studies should adequately document exposure to known neurotoxins and family history of developmental disorders and exclude potential genetic causes.

Whilst studies like this aim to establish the link between ADHD and autism they do little to decrease the anxiety that comes with being pregnant, undergoing childbirth and raising a child. 

Causation has not been established in this study. Any study such as this should be prospective and not seek answers to the question of causation by questionnaire nor a blood sample taken at the time of birth.

Indeed it is difficult to establish if the paracetamol was taken 1 day before, the day of child birth, on one occasional or multiple occasions. Furthermore (without reviewing the study ) there will no doubt be confounders such as other agents taken through pregnancy, family predisopostion or indeed social economic factors and education. 

For this very reason caution is advisable when publishing such a study, and until a prospective well-documented study is performed then claims of its adverse effects should be tempered by common sense.

This is an epidemiological study suggesting an association between taking paracetamol around the time of birth and clinical characteristics of autism and/or attention-deficit/hyperactivity disorder later in life in the offspring. As with all epidemiological studies this can only suggest an association not causation. It was not a prospective study as claimed by the authors as the patient sample was taken from a long-established patient data base; also there was no pregnant patient group who had not taken paracetamol. These limitations make it difficult to assess influence of other factors that might contribute to ASD and ADHD.

This is an improvement on previous studies because measurements of paracetamol and metabolites in placental cord blood were available, although unsatisfactory in that one major metabolite (Paracetamol sulphate) was not measured as acknowledged by the authors. Previous studies have relied on self-reporting of taking paracetamol by the patients, which is notoriously unreliable.

The results show a dose related correlation (Table 2) with the clinical problems.

An important limitation of the study is that although the blood measurements confirm whether or not the patient actually took paracetamol at around the time of childbirth, we know nothing of when else during pregnancy the paracetamol might have been taken. It is possible. but seems unlikely. that a few tablets of paracetamol taken around the time of birth would cause problems;  it seems more likely that a longer period at a critical time in brain development would do so. The subgroup analysis (Figure 2) suggests higher risks for lead and bottle feeding than for paracetamol, something the authors do not comment on. An important conclusion of the authors was that the results 'warrant additional investigations'. It is a pity that there was no discussion of what these might be. The current official advice in Australia (e.g. Australian Medicines Handbook, 2019) is that paracetamol is 'safe' to use in pregnancy and breast feeding. In the light of this and other published information it is high time for this advice to be modified.

This study examines an association between the levels of Acetaminophen (paracetamol in Australia) and its metabolites in umbilical cord blood with the risk of various neurodevelopmental conditions such as Attention Deficit/Hyperactivity Disorder (ADHD), Autistic Spectrum Disorder (ASD) and other less well defined conditions designated Developmental Disabilities (DD) by the authors of the study.

The study reports a positive association, with the odds ratio of between two and three for all these conditions, which roughly translates to doubling or tripling the risk of these neurodevelopmental disorders for children whose cord blood contained increased levels of paracetamol and its metabolites.

This was not a truly prospective study, despite the authors' claim to the contrary.

Cord blood samples were collected at the time of birth, and some, but not all, children were followed up with neurodevelopmental testing for a number of years. Only those children who had neurodevelopmental testing, which was not routine for the whole cohort, were included in the analysis. This significant selection bias is evident in extremely high prevalence of all neurodevelopmental conditions in this cohort, with only 32.8 per cent of children that were included in this study not receiving a diagnosis of either ADHD, ASD, DD or some combination of the three. Therefore the applicability of the study’s findings to general population, where the prevalence of neurodevelopmental conditions is generally accepted to be less than 5 per cent, is highly questionable.

One must also question the method used to assess paracetamol exposure. Cord blood levels would reflect maternal medication intake immediately or shortly before giving birth. It is unknown how long paracetamol and its metabolites are present in fetal circulation after maternal ingestion, but this window must be a few days long at most. Therefore, this study contributes nothing to the question of the effect of paracetamol exposure during pregnancy and not immediately before birth.

Another interesting and somewhat surprising finding of the study is that all cord blood samples had some paracetamol and its metabolites detected. This implies that all women in this cohort took this medication around the time of giving birth. This may be standard care in the USA, where the study was conducted, but in Australia, paracetamol is not routinely given to women in labour and is not the drug of choice for pain relief.

Due to the universal exposure to the drug in the cohort, researchers did not have a group of women who were not exposed to paracetamol in late pregnancy and around the time of childbirth. This is yet another major weakness of this study as all the comparisons were done between groups with some degree of drug exposure with no true control group. There are other multiple criticisms of the study methodology as related to statistical analysis and techniques used.

What conclusions can be drawn?

Based on the profile of subjects included in the study and the methodology used to assess paracetamol exposure, it can be concluded that there may be an association between used of paracetamol around the time of birth and the risk of neurodevelopmental disorders in children in high risk populations. This may or may not translate to increased risk in general population. One must also be cognisant of the fact that a statistical association is not equivalent to causation.

Implications for prospective mothers in Australia

Nothing in this study indicates that an occasional intake of a couple of Panadol tablets while pregnant would have any effect on the risk of ADHD/ASD. Like any medication, paracetamol is not risk free and should only be taken if necessary, especially in pregnancy. Based on this study’s findings it may be best to avoid paracetamol around the time of labour and delivery. It may also be prudent to increase the time interval between doses to 6-8 hours while pregnant, rather than the usually recommended 4-6 hours. This is related to uncertainties surrounding paracetamol metabolism and the possibility that toxic metabolites may accumulate in fetal circulation.

Around the world, rates of children being diagnosed with ADHD and autism spectrum disorder (ASD) are increasing year on year and there are many reasons for this, including changes being made to the diagnostic criteria needing to be fulfilled to achieve a diagnosis. This study is an attempt to look at  an association that has previously been reported between paracetamol use and incidence of these conditions.
 
Many maternal, (BMI, smoking and alcohol history, race, and presence of stress during pregnancy) paternal (age) and indeed child (male gender, mode of delivery, preterm delivery and low birth weight) characteristics have been linked to increasing rates of ADHD and ASD but all only show association without any definitive causation. The increased risk associated with many of these factors appears to be higher than any association with taking paracetamol.
 
This particular prospective cohort study looked at one third (close to a thousand) of the babies born as part of the 'Boston birth cohort' which followed mothers and children for 10 years. Measurements were taken of paracetamol and some of its metabolites in the stored cord blood of babies and from maternal samples taken up to three days after the delivery. The authors concluded that there was a statistically significant association between the presence of paracetamol in the cord blood and in childhood ADHD and ASD. Moreover higher levels of the drug were associated with higher rates of the condition.
 
There were a number of concerns with the study making any conclusion other than more research is needed, difficult to arrive at.
 
The rates of children affected with ADHD and autism spectrum disorder in this study seemed inordinately high (up to 30 per cent) when compared with the Australian population, suggesting a possible source of bias. In order to be involved in the study, there had to be a sufficient amount of cord blood available to measure drug and metabolite levels, meaning that not all the women in the study and their children could be assessed, and because paracetamol was found in 100 per cent of cord blood samples used, it means that babies not exposed to paracetamol were unable to be studied.
 
There is no way of telling how long a particular fetus was exposed to paracetamol during the pregnancy or whether there were critical times during the pregnancy where exposure was more likely to lead to problems. The reason for the pregnant woman taking the paracetamol in the first instance was also not evaluated. All we know is that the drug had been taken in the relatively short time prior to delivery and was therefore able to be measured in cord blood.
 
Over 50 per cent of all pregnant women will require some form of pain relief during their pregnancy. There is no clear indication from this study that the taking of paracetamol causes problems. Until there is clear evidence of harm, we must not deny pregnant women effective analgesia nor direct them to take other forms of pain relief which may have their own recognised problems. No clear recommendations can be made, but as always, an evaluation of risk and benefit must be taken before taking any medication.
 
This study documents an association which may or may not be found in the future to be causative. Paracetamol exposure may simply be a marker for a truly causative agent like an inflammatory process which in turn causes pain or fever, leading the woman to take paracetamol.
 
All in all, interesting and warranting of further research into the area, but not strong enough an association to change current recommendations for pain relief when needed in pregnancy.