When I use a word . . . Paracetamol/acetaminophen—autism and asthma

Serendipity

Serendip was an ancient name for the island of Ceylon, now Sri Lanka. In a letter written on 28 January 1754 to his friend and distant cousin Horace Mann, the author Horace Walpole cited “a silly fairy tale,” “The Travels and Adventures of Three Princes of Serendip,” the heroes of which, he wrote, “were always making discoveries, by accidents and sagacity, of things they were not in quest of.” He therefore coined “a very expressive word,” serendipity, which he summarised as “accidental sagacity,” to describe the princes’ facility. In fact, in the original tale the princes made no discoveries at all; what they did was to show their powers of observation and interpretation of events.1

Nevertheless, the word has entered the English language, meaning “the faculty of making happy and unexpected discoveries by accident” and, by extension, “the fact or an instance of such a discovery.”2

A hundred years later, in December 1854, on the occasion of the inauguration of the Faculté des Sciences in Lille, Louis Pasteur asserted that “dans les champs d’observation le hasard ne favorise que les esprits prepares,” “Where observation is concerned, chance favours only minds that are prepared.”3 Pasteur’s adage is usually translated as “chance favours the prepared mind,” and in coining it he added the faculty of mental receptivity to the qualities needed for achieving serendipity.

Serendipity in drug discovery

The history of drug discovery is rich in instances of serendipitous findings, often depending on observations of outcomes, such as adverse effects, that were not predicted at the start.

For example, when an analogueue of isoniazid was studied as an antituberculosis drug, it was noticed that the mood of those taking it improved, leading to the introduction of iproniazid as an antidepressant. Of course, you might feel better simply because your tuberculosis was being adequately treated. But the antidepressant action of iproniazid was unrelated to an antibacterial effect, being mediated instead by inhibition of monoamine oxidase.

Similarly, as a result of the observation that individuals enjoyed penile erection when they were given an inhibitor of phosphodiesterase type 5, in a study of its potential use as a vasodilator in angina pectoris, sildenafil emerged as a treatment for erectile impotence.

Sometimes serendipitous discoveries are made because of errors. In the 1880s, when naphthalene was being used as a treatment for intestinal worms, it seemed to have an antipyretic effect in a study of people who also had a fever. However, it was then discovered that acetanilide had been supplied instead of naphthalene.4 Acetanilide was an effective febrifuge, but it tended to cause methaemoglobinaemia and so analogues were made. The ethyl ether derivative, acetophenetidin, was found to be particularly effective and safer and was then marketed as Phenacetin, a name that became so popular that it was registered as its International Nonproprietary Name (INN) in 1959, having been on the list of proposed names since 1956. When it was discovered that phenacetin caused renal papillary necrosis,5 it was replaced by an active metabolite, paracetamol.

Paracetamol or acetaminophen?

All medicines have three different types of names. First is the chemical name, which specifies its structure. Then there is a shorter nonproprietary or generic name. Finally, drug companies that market the medicine can give it their own brand name, an even simpler name than the generic name, designed to be more easily remembered.

Different generic names may be chosen in different countries. These include British Approved Names (BANs), United States Adopted Names (USANs), Japanese Accepted Names (JANs), and Dénomination Commune Française (DCF).

In the 1950s the World Health Organisation (WHO) developed the International Nonproprietary Name (INN) system to give generic names to pharmacological substances intended for marketing as pharmaceutical products. The names were intended to be distinctive in sound and spelling, not too long, and unlikely to be confused with other names in common use. Furthermore, although in the early years INNs were often based on the chemical names of the substances, it later became more usual, when possible, for part of the name to indicate the group of pharmacologically related substances to which the compound belongs. Drugs whose names end in –vastatin, for example, are all inhibitors of HMG Co-A reductase. The first list of INNs was published in 1955.

According to the International Union of Pure and Applied Chemistry (IUPAC) the full chemical name of paracetamol is N-acetyl-para-aminophenol, although it was previously known as para-acetamidophenol.6 The name “acetaminophen” was derived from a contraction of the former, the name “paracetamol” from the latter.

The original application for an INN, addressed to the WHO, specified “paracetamol” as the desired name, and that was the name that was chosen. The name “paracetamol” was first included in the WHO’s 8th list of proposed names (pINNs) on 9 March 1959 and in the 4th list of recommended names (rINNs) on 4 March 1962, in both of which “acetaminophen” was listed as a synonym. Both names, paracetamol6 and acetaminophen,7 were already in use by 1957 at least, and the respective brand names Panadol and Tylenol, both contractions of the chemical names, the latter from acetyl+phenol, were also already in use. Paracetamol, however, was approved as a generic name by the British Pharmacopoeia Commission6 before the USAN programme began in 1961. Had the applicant suggested “acetaminophen,” which became the USAN, that might have been chosen for the INN, rather than paracetamol.

Paracetamol and autism

On 22 September 2025, the US Food and Drug Administration (FDA) issued a notice about an intended change to the label for acetaminophen, “to reflect evidence suggesting that the use of acetaminophen by pregnant women may be associated with an increased risk of neurological conditions such as autism and ADHD in children.” The UK regulator, on the other hand, the Medicines and Healthcare products Regulatory Agency (MHRA) issued a statement on the same day, saying that “there is no evidence that taking paracetamol during pregnancy causes autism in children.”

The truth lies somewhere between these two statements, but considerably closer to the MHRA’s statement than to the FDA’s. There is evidence of an association, but the evidence is poor, and even if there is an association one cannot conclude that it is one of cause and effect. Indeed, it probably is not.

There has been an enormous amount of discussion published about this, mostly citing the studies that have previously been published, before the latest study precipitated the precipitate statements about it from the White House. I do not propose to go over the evidence itself. I simply want to mention two features that have not been much mentioned, if at all, in the coverage in the UK: conflicts of interest and biases.

Conflicts of interest

Conflicts of interest when dealing with therapeutic interventions come in various forms. They include: employment, including consultancy or directorship, by a company that makes or markets an intervention, or a rival company; possession of shares in such a company or a patent related to the intervention; payment by a company to an individual to give evidence or act in some other way on its behalf; authorship of a study of the intervention, especially, although not only, if the study was funded by a relevant company; a belief, for whatever reason, e.g. ideological, ethical, or religious, in the efficacy, harmfulness, or other feature of the intervention, including having oneself benefited from the intervention or suffered an adverse reaction to it; and a personal relationship with anyone who might have such a conflict, e.g. a spouse or a child.

Most of these types of conflicts are linked to financial gain, direct or indirect, and that is the criterion that is usually used in judging them. However, not all of them are so linked, and any conflict that may be a cause of bias in the views of the individual expressing an opinion should be taken into account when judging the quality of the opinion.

Conflicts of these kinds have not been mentioned in discussions about paracetamol and autism that I have read in the UK media. However, US commentators have reported the case of a university dean who received around $150 000 to provide expert testimony in a lawsuit against the manufacturer of Tylenol two years before publishing research that seemed to link the drug to autism.8 When the dean testified as an expert witness, the judge excluded the dean’s evidence. This raises the question of whether any other conflicts of interest might have been involved in other published studies.

Biases

The Catalogue of Bias (https://catalogofbias.org) currently includes monographs describing 65 different types of bias that can affect academic research, and the complete list of biases that might be included in the catalogue is much longer than that.

Studies of the association between an intervention taken during pregnancy and subsequent outcomes in the child are potentially subject to a range of biases. For example, a study in which a history of the use of the intervention is obtained by self-reporting, as in case-control studies and retrospective cohort studies, is likely to be subject to recall bias, a distortion that occurs when participants do not remember previous events or experiences accurately or omit details.9

In selection bias, those with the condition of interest may differ from those without it in ways that are related to the cause. This can happen in a case-control study, for example, when the controls are selected inappropriately.10

Confounding is a distortion that modifies an association between an exposure and an outcome, because a factor associated with the exposure is also independently associated with the outcome.11 It is a common problem in observational studies of all kinds. For example, if the risk of a developmental abnormality such as autism or ADHD in a child is increased when the mother suffers a severe fever, for which paracetamol would be indicated, the apparent causative association between the paracetamol and the abnormality would be confounded by the actual causative role of the fever. Indeed, if that were the case, it would be likely that the early use of paracetamol at the start of the fever would prevent or at least mitigate the subsequent abnormality, rather than causing it. Other indications that might cause confounding could include migraine and infection (as a cause of fever). This is a specific form of confounding, confounding by indication.12 Other types of confounding include genetic and environmental factors, socioeconomic status, and other interventions.

The largest study of the question so far illustrates how these problems can be overcome. In a Swedish cohort of nearly 2.5 million children the use of sibling controls eliminated the marginal effect that paracetamol exposure seemed to have as a cause of autism or ADHD.13

Paracetamol and asthma

All this is highly reminiscent of a similar story that exposure of children to paracetamol might be causatively associated with a risk of asthma and other allergic disorders in later life. Despite evidence from several studies, suggesting an association, evidence of confounding by indication came from a prospective birth cohort study of 620 children with a first-degree family history of allergic disease, who were followed until age 7 years.14 In summing up all the evidence, the author of a review suggested that the association of exposure to paracetamol during childhood with apparent increases in the risks of allergic disorders was probably due to confounding by indication and other biases, taking account, for example, of antibiotic use and viral infections in childhood, and recall bias in questionnaire studies.15

A final thought

The news that paracetamol is suspected of causing autism in children exposed to it in utero is no news at all. There are studies going back to 2013.16

Nor is the news accurate. It is highly likely, given the quality of the evidence and the possibility of biases, that the apparent association, when it has been found, was not indicative of a causal association. The previous case of a supposed association of paracetamol with asthma is a salutary example of a similar case.