Paracetamol (acetaminophen) use in infants and children was never shown to be safe for neurodevelopment: a systematic review with citation tracking

Most parents and pediatricians currently believe that use of paracetamol in infants and children is safe, promulgating widespread use of the drug in that population. Use of paracetamol in the pediatric populations now exceeds 90% in some studies [1] and persists even under circumstances in which the drug may have no benefits, such as prophylaxis prior to some vaccinations [2] and treatment of mild fevers [3]. Such beliefs and practices are strengthened and supported by a medical literature which repeatedly asserts without reservation that, when used as directed, the drug is safe in the pediatric population. However, mounting evidence points toward the view that paracetamol exposure during early development can have an adverse effect on neurodevelopment, even when used as directed. For example, in a recent review [4], eight studies supporting a link between prenatal paracetamol exposure and neurodevelopmental problems were identified [5‐12]. In the 3 years since that review, at least six additional studies have confirmed this same relationship, three of which have used data from the Norwegian Mother and Child Cohort Study [13‐18]. Although exposure to paracetamol in utero is associated with neurodevelopmental problems, even after consideration of potentially confounding factors, the effects are typically small, and the amount of paracetamol required to yield the effect is greater than the amount typically used by average individuals. For example, after adjusting for potential confounders such as parental education level, use of vitamin supplements, parental BMI, smoking, and use of other drugs, Skovlund and colleagues found a weak yet significant association between prenatal exposure to paracetamol and mother-reported communication skills: the chances of being in a lower development category increased with increasing periods of prenatal paracetamol use but not prenatal opioid use [13]. In another example, using propensity score matching, Vlenterie and colleagues found that 28 or more days of paracetamol use during pregnancy was associated with a modestly increased risk of delayed motor milestone attainment (OR: 1.35, 95% CI 1.07–1.70) by children at 18 months [14].

Evidence points toward a higher risk of paracetamol-induced neurodevelopmental disorders when exposure occurs after birth as compared to in utero. Studies using laboratory rodents demonstrate that exposure to near therapeutic doses of paracetamol during the first days of life induces profound, long-term neurological changes [19, 20], whereas somewhat higher doses are required to induce permanent neurological damage during pregnancy [21]. These laboratory studies demonstrate that the target organ for toxicity in neonates is the central nervous system, not the liver, and demonstrate that if paracetamol had been tested using current guidelines, it would never have been approved for use in children. More concerning are observations in children indicating that paracetamol is not safe for neurodevelopment. The 2008 study which first raised a red flag regarding the safety of paracetamol during neurodevelopment found a greater than 20-fold risk of regressive autism with paracetamol use during childhood [22]. Although this relatively small study did not attract enough interest to promote larger studies, other lines of evidence support the view that paracetamol exposure during early life can lead to neurodevelopmental disorders. For example, a startling twofold greater incidence of infantile autism in circumcised boys compared to non-circumcised boys [23] can be readily explained by potentially negative impacts of paracetamol exposure during and following the circumcision procedure [4]. Sadly, the widely held and entrenched belief that vaccines induce autism [24, 25] may be yet another result of the impact of paracetamol on neurodevelopment in combination with widespread use of the drug during vaccination [4].

With the above concerns in mind, a systematic evaluation of the peer-reviewed literature was initiated to address the question of why paracetamol is widely believed to be safe for use during early development. All papers published between 1974 and 2017 that contained the keywords “infant” and either “paracetamol” or “acetaminophen” were considered. All papers which made claims that paracetamol or acetaminophen is safe for use in infants or children were identified, and the justification for this claim was critically evaluated.

The use of paracetamol predates current safety standards used in the pharmaceutical industry, and even if current standards were applied, those standards do not mandate testing for long-term neurological development. Thus, to those aware of the inner workings of the drug approval process, especially as it has been applied to paracetamol, the results obtained in this study will not be surprising and may even be considered by some to be a foregone conclusion that need not be evaluated. Nevertheless, the widespread belief that paracetamol is safe and the resulting widespread use of the drug in the pediatric population is built on the assumption that it is safe for neurodevelopment. With this in mind, this review focuses on unqualified claims of safety in the medical literature that, reasonably, are taken by parents and even many physicians to indicate that the drug is safe for neurodevelopment.

As a first step in understanding why paracetamol is thought to be safe during early development, all titles and abstracts in the PubMed^® Database with keywords “infant” and “acetaminophen or paracetamol” published between 1974 and 2017 were identified. The term “infant” rather than “child” was selected because (a) the number of papers with the term “child” was prohibitively large, and (b) the focus of the study was intended to be on drug exposure during early development, from birth to age approximately 6 years, not individuals up to the age of 17 years. In all cases, the terms paracetamol and acetaminophen were taken to be synonymous, and no distinctions were made.

In the second step, two coauthors (JCH and JTS) independently screened all titles and abstracts. In this step, articles that could not be obtained in English and all articles not describing use of paracetamol in humans were eliminated from the study. Based on titles and abstracts (if available), articles were tagged which were deemed likely to make claims regarding the safety of paracetamol use in infants and children between birth and age 6 years.

In the third step, two coauthors (JCH and JTS), continuing to work independently, examined full texts of all tagged titles and abstracts. Texts were examined for the following three assertions:

In cases where the terms “drug of choice” or “recommended” were used, the context was considered. In some cases, particularly in manuscripts expressing caution regarding the use of paracetamol, these terms were not taken to imply safety, but rather were taken to be an indicator of the common acceptance of the drug. These articles were excluded from the study. Based on this approach, articles were tagged that were considered to have made safely claims regarding the use of paracetamol in infants or children younger than 6 years old.

Still working independently, two coauthors (JCH and JTS) evaluated each manuscript making a claim of safety, determining the source of authority for the stated claim. If no literature was cited to support the claim, this was documented. In cases where the source that was cited contained another citation, that secondary reference was obtained and evaluated. This process continued as needed until an original source or sources describing an actual demonstration of safety was identified. An example of the results of this process is shown in Fig. 1.

Bild vergrößern

In the fourth step, any discrepancies between the analyses provided by coauthors JCH and JTS were arbitrated by coauthor WP. In the fifth step, articles upon which safety claims were based were compiled. Finally, articles which made safety claims and articles upon which safety claims were based were evaluated for actual experiments designed to assess safety. For each experiment described, the study group, endpoints measured, and follow-up time were evaluated. Data were graphed and descriptive statistics calculated using GraphPad Prism 8 software. The review was not registered, and the protocol is as described in this Methods section.

An overview of results from a systematic search for studies demonstrating safety of paracetamol use in infants and children is shown in Table 1. The initial Medline search provided 3096 articles that contained the terms infant and either paracetamol or acetaminophen that were published between 1974 and 2017. From these articles, 467 were selected for assessment based on likelihood of safety claims regarding use of paracetamol in infants or children. Of these 467 articles, 218 made safety claims regarding the use of paracetamol in infants or children. During this phase of the study, numerous articles were identified which either claimed or demonstrated that paracetamol use, even at doses beyond the recommended dose, does not generally cause long-term liver damage in infants or children. Any claims of safety for liver function were not evaluated in detail and were not considered in this study. Only general claims of safety were assessed.

Of the 218 articles making claims that paracetamol use in infants or children is safe, half (114) provided no citation. The other half (114) of the articles cited additional articles as evidence that paracetamol is safe in infants or children. Articles making safety claims as well as articles cited as sources of authority for safety claims were evaluated as described in the Methods. In some cases, the “primary” cited articles did not make original claims of safety, but rather cited additional (“secondary”) articles. In cases where a primary article cited another article, the primary article was not considered to have made an original claim of safety, and was not evaluated further. An example of the results of this process is shown in Fig. 1. Both primary articles and secondary (and tertiary, etc.) articles attributed with claims of the safety of paracetamol use in infants or children were compiled and are shown in Table 2. In total, 103 articles were identified which were cited as containing original claims that paracetamol use in infants or children is safe when used as directed. In addition, 16 of the 218 articles with safety claims (a) made those claims based on original experimental evidence and (b) were not cited by other papers. These articles are also included in Table 2, listed at the bottom of the table with zero citations.

Several studies emerged as popular citations for the claim that paracetamol use in infants or children is safe when used as directed. Only 19 articles were cited more than twice, and the most popular article [26] was cited a total of 13 times (Table 2) by the 218 articles we identified. However, in some cases, well cited articles did not make original claims of safety, and are therefore not included in Table 2. For example, an article by Perrott and colleagues in 2004 [27] was cited a total of 7 total times by the 218 articles we identified. However, Perrott’s article, being a review, does not make original claims of safety, but rather cites additional articles as the authority for assurance of safety (Fig. 1). Thus, Perrott’s article is not included in Table 2 as an original source for the claim that use of paracetamol is safe for infants and children when used as directed.

Of the 103 articles cited as authority for the safety of paracetamol use in infants or children, 27 did not make claims of safety and did not address safety experimentally (Table 2). Thus, 76 of the 103 articles did address safety, and 48 of these 76 articles (63%) had already been identified in the original 218 articles gleaned from the Medline search. Of the 103 articles, 36 articles described experimental studies which involved paracetamol use in infants or children. As described above, from the original 218 articles making claims of safety, 16 uncited articles described experimental studies that were used to support claims of safety. Thus, 52 studies in total (36 cited plus 16 uncited) provided experimental evidence supporting claims of safety. Although several of those 52 studies provided measures of liver function (Table 2), none of the studies provided any assessment of neuropsychiatric function. Furthermore, the median follow-up time of all 52 studies was 48 h (Fig. 2), far too short to identify any long-term effects of drug exposure on neuropsychiatric function. Six studies had follow-up times of longer than 10 days, although only one study [28] evaluated patients beyond 6 weeks. However, all experimental studies were blind to any potential effects of drug exposure on long-term neuropsychiatric function. For example, although patients were followed for a full year in one study [28], the only endpoint measured was re-admission for surgery. As another example, a study following patients for up to 6 weeks measured only recurrence of typhoid fever beyond the initial treatment period of the study [29].

Bild vergrößern

The path from more recent papers to the original research addressing the safety of paracetamol in infants and children was sometimes convoluted. In one notable case, a popular citation did not did appear in the literature (Table 2). Not only did the volume and journal number not match, but the title could not be found elsewhere. As another example, the citations reporting safety of paracetamol use in children reported by Temple and colleagues in 2017 [30] are illustrated in Fig. 1. This article provides a detailed description of three prior reports to the European Medicines Agency (reports 24,570, 24,571, and 47,402) which, together, according to the authors, “confirm that the recommended standard paracetamol dose of 10 to 15 mg/kg is a safe and effective dose for use in pediatric patients when administered as a single dose or as multiple doses for up to 72 h.” However, the only safety measure used in the three studies was ALT levels as a marker for liver function, assessed for a maximum of 72 h. In addition to the three reports described in their publication, Temple and colleagues cite 10 additional articles as sources for safety, including the claim that paracetamol has a “well-established efficacy and favorable safety profile” (Fig. 1). Among these 10 papers is a clinical trial [31] that addresses efficacy but not safety, and refers to two other papers that address safety, one by Lesko [26]. The paper by Lesko contradicts the view that paracetamol is safe, finding that paracetamol is significantly worse than ibuprofen in terms of risk for outpatient visits following treatment of children with asthma. Another of the articles cited by Temple in 2017, a review written by Temple more than 30 years before [32], cites a paper in the Federal Register [33] as the source for the statement that “Paracetamol is relatively free of side effects and has a wide margin of safety between therapeutic doses and toxic doses.” The document in the Federal Register [33], a lengthy treatise primarily focused on determination of the appropriate dose for adults of salicylates in general and aspirin in particular, in turn cites two papers involving safety studies of paracetamol in the human pediatric population. One of those studies [34] evaluated 98 children using a blinded approach comparing aspirin and paracetamol, and monitored the children for only 6 h. The other study [35] monitored 20 children following administration of both aspirin and paracetamol. In that study, monitoring occurred for 6 h or less, and no information was provided regarding particular side effects that were being assessed. Importantly, the Federal Register [33] attributed their view that paracetamol has “a wide range of safety” to laboratory animal studies showing that the lethal dose of paracetamol is significantly greater than the dose administered to humans. Unfortunately, studies had not been conducted at that time showing that paracetamol induces permanent neurodevelopmental injury in laboratory animals at far lower doses than the lethal dose [19, 20], similar to doses administered to infants and children.

Our initial search of the PubMed^® Database and review of more than 3000 titles and abstracts yielded 218 papers making claims that paracetamol is safe for infants and children when used as directed. Claims of safety in those 218 papers were traced back to 103 articles shown in Table 2, but less than 20 of those were cited more than twice, indicating that a limited number of studies are considered key or cornerstone to the view that paracetamol is safe for use in infants or children.

Finding more than 200 articles making claims that paracetamol is safe and/or well tolerated for infants and children when used as directed, this study confirms the view that the drug is widely thought to be safe, despite the absence of any study demonstrating that it is safe for neurodevelopment. The fact that 27 out of 103 articles citated as authority for the safety of the drug did not, in fact, demonstrate safety or make safety claims might suggest that the safety of paracetamol is taken for granted, and is not carefully considered. This view is supported by the observation that one popular citation for safety does not exist in the literature.

This study does not in any way suggest that the effects of early life exposure to paracetamol on neurodevelopment have never been examined. Indeed, the first study to address the issue was published in 2008 by several now-prominent scientists, then at the University of California San Diego and at San Diego State University [22]. This case-controlled, survey-based study raised substantial concerns, as mentioned in the Introduction. Further, studies in animal models evaluating the issue have been conducted [19‐21, 36], all indicating that the drug is not safe for neurodevelopment despite a wide range of study designs. In addition, as described in the Introduction, at least 14 cohort analyses [5‐18] have indicated that exposure to paracetamol during pregnancy is not safe for neurodevelopment of the fetus. Thus, the present study does not demonstrate that the safety of paracetamol for neurodevelopment has never been examined, but rather demonstrates that assertions that paracetamol is safe during early development when used as recommended are based on a lack of knowledge regarding the effects of paracetamol on neurodevelopment.

The difficulty in moving forward into the clinical arena based on current scientific knowledge is perhaps reflected in the debate surrounding a recent consensus statement supported by almost 100 clinicians and scientists [37] urging caution with the use of paracetamol during pregnancy. This consensus statement of 2021 was met with some skepticism, including an announcement by the American College of Obstetricians and Gynecologists (ACOG) asserting that “Most importantly, patients should not be frightened away from the many benefits of acetaminophen (paracetamol)” [38]. Furthermore, the ACOG asserts that “This consensus statement, and studies that have been conducted in the past, show no clear evidence that proves a direct relationship between the prudent use of acetaminophen (paracetamol) during any trimester and fetal developmental issues” [38]. This latter assertion by the ACOG could technically be considered correct if studies in animal models are ignored, but it demands a level of proof that is not met by the over one dozen studies of cohort data with a wide range of controls for confounding factors via multivariate analysis.

In their response to the 2021 consensus statement, the ACOG clarified their demands for proof, stating that “ACOG’s clinical guidance remains the same and physicians should not change clinical practice until definitive prospective research is done” [38]. However, it is difficult to rationalize the need for such a high level of certainty regarding a drug never demonstrated to be safe or life-saving, where judgment should presumably err on the side of caution and avoidance of harm. Indeed, the drug would not meet current safety standards during preclinical testing due to adverse, long-term neurological effects in laboratory animals, and thus would never reach phase I testing under the current regulatory system. Furthermore, the potential difficulty in obtaining the prospective, controlled study demanded by the ACOG is of concern. Although a study during pregnancy might be envisioned, exposures after birth are likely also important (see Introduction), and therefore must be taken into account in any long-term study. The magnitude and difficulty of a sufficiently powered study, starting from conception and extending into early childhood, is considerable. For example, a group at the University of Oulu conducted a 5-year prospective, placebo-controlled study on 49 children following exposure to paracetamol (n = 19) or to saline control (n = 20) [39]. However, as the authors point out, their study is underpowered to test the impact of paracetamol on neurodevelopment. In addition, the authors did not control for exposure during all 5 years of the study, but rather only for exposure during a single, 4-day period. Furthermore, it is difficult to imagine a placebo control for treating fevers in babies and children, since withholding paracetamol may need to be accompanied by non-medicinal methods of treating some fevers [40]. Even more importantly, the University of Oulu study used the intravenous formulation of paracetamol rather than the much more commonly used oral formulation. The intravenous formula contains an antidote for paracetamol toxicity (cysteine, a glutathione precursor), which should, hypothetically, block much of the adverse effects of paracetamol. Since this antidote is not present in the commonly used oral formulation, the University of Oulu study, even if it had been much larger and controlled for drug exposure over a period of years, would still not apply to most cases of paracetamol use. It should be noted that, in laboratory animals, exposure during the postpartum period to currently accepted levels of the intravenous formulation of paracetamol with the antidote present causes dramatic increases in asocial behavior later in life [36]. Thus, the argument is not that use of paracetamol with the antidote is safe, but rather that, hypothetically, some of the more serious adverse effects might be prevented by inclusion of the antidote with the drug.

The difficulty in obtaining prospective, controlled studies evaluating the safety of paracetamol in humans is, as outlined above, a complex problem involving large numbers of patients and years of study time, difficulty in establishing controls, and the variable presence of an antidote for paracetamol toxicity in paracetamol formulations. These issues point toward the importance of careful examination of presently available evidence or, as the case may be, the lack of evidence regarding the safety of paracetamol for neurodevelopment.

Although not the intended purpose of this systematic review with citation tracking, it demonstrated that paracetamol has been proven safe for liver function in infants and in small children, even at doses higher than those currently recommended. During the course of this review with citation tracking, an assumption was repeatedly encountered: because the target of paracetamol toxicity in adults is the liver, demonstration of safety in infants and children need only be tested in the liver. This assumption was/is held despite the fact that the target tissue for drug function is in the central nervous system, not the liver. A similar assumption has proven tragically fatal in the past, when it was assumed that metabolism of the antibiotic chloramphenicol was the same in infants as in adults. In that case, administration of the drug in infants led to a number of deaths [41‐43] before the problem was identified.

Despite apparently being taken for granted, this study demonstrates that paracetamol was never shown to be safe for neurodevelopment. This conclusion is consistent with emerging studies showing a connection between paracetamol use during development and long-term neuropsychiatric disfunction as described in the Introduction. This conclusion is also consistent with emerging studies in animal models showing exquisite sensitivity of long-term behavior to early life exposure to paracetamol at near-therapeutic doses.