American Journal of Epidemiology Vol. 188, No.

1
© The Author(s) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. DOI: 10.1093/aje/kwy193
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.
org/licenses/by/4.0), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is Advance Access publication:
properly cited. September 7, 2018

Original Contribution

Acetaminophen (Paracetamol) Exposure During Pregnancy and Pubertal

Development in Boys and Girls From a Nationwide Puberty Cohort

Andreas Ernst*, Nis Brix, Lea L. B. Lauridsen, Jørn Olsen, Erik T. Parner, Zeyan Liew,
Lars H. Olsen, and Cecilia H. Ramlau-Hansen
* Correspondence to Dr. Andreas Ernst, Department of Public Health, Section for Epidemiology, Aarhus University, Bartholins Allé 2,
8000, Aarhus C, Denmark. (e-mail: [email protected]).

Initially submitted February 5, 2018; accepted for publication August 21, 2018.

This study explored the association between exposure to acetaminophen during pregnancy and pubertal devel-
opment using data from 15,822 boys and girls in the longitudinal Puberty Cohort, nested within the Danish National
Birth Cohort. Use of acetaminophen was reported 3 times during pregnancy and 6 months postpartum. In total,
54% of mothers indicated use at least once during pregnancy. Between 2012 and 2017, sons and daughters pro-
vided information on a wide range of pubertal milestones—including Tanner stages, axillary hair growth, and age
at menarche or voice break and first ejaculation—every 6 months from 11 years of age until full sexual maturation.
Data were analyzed using a regression model for interval-censored data, providing adjusted mean monthly differ-
ences in age at attaining the pubertal milestones according to intrauterine cumulative (weeks) and trimester-spe-
cific acetaminophen exposure. Our results suggested a tendency towards slightly earlier attainment of almost all
studied markers of female pubertal development with increasing number of weeks of exposure (i.e., about 1.5–3
months earlier age at pubic hair, axillary hair, and acne development comparing unexposed with those prenatally
exposed for more than 12 weeks). Male pubertal development had no strong association with acetaminophen
exposure.
endocrine disrupters; prenatal exposure delayed effects; puberty; sex characteristics; Tanner stages

Abbreviations: CI, confidence interval; DNBC, Danish National Birth Cohort.

Use of over-the-counter analgesics, especially acetaminophen, acetaminophen with alterations in fetal germ cell development
has been steadily increasing in Western countries (1). Acetamin- and differentiation, reducing later reproductive capacity in
ophen, also known as paracetamol, is now being used at least female—but not in male—offspring (14). Cohort studies on
once by more than 50% of pregnant women in some countries the associations between prenatal exposure to acetaminophen
(2–4). It is considered safe by the general public and remains a and anogenital distance support evidence from animal studies
first choice for treatment for fever and pain during pregnancy (5). of antiandrogenic disruption in males but do not demonstrate
Acetaminophen is, however, capable of crossing the placental a similar association in females (15, 16). In parallel, studies on
barrier (6, 7) and might therefore interfere with organ develop- male genital malformations show no associations with hypo-
ment affecting health in the offspring, including reproductive spadias and inconclusive results concerning cryptorchidism
health (8). Studies in rodents suggest that administration of (9, 17–22). It remains largely unknown to what extent a poten-
acetaminophen equivalent to human therapeutic doses exerts tial endocrine programming disruption of intrauterine expo-
antiandrogenic disruption, inhibiting early masculinization of sure to acetaminophen might affect long-term reproductive
male rats (9–12). On the other hand, an in vitro study of human health, such as age at puberty timing. To our knowledge,
fetal testicular tissue found no association between short- this is the first study on puberty timing, and we hypothesize
term exposure to acetaminophen and testosterone production that intrauterine exposure to acetaminophen alters the tim-
(13). A recent study in rodents linked intrauterine exposure to ing of markers of male and female pubertal development.

34 Am J Epidemiol. 2019;188(1):34–46
 Acetaminophen and Timing of Puberty 35

METHODS ensuring adequate samples sizes. We oversampled participants

from subgroups of 12 pre- and perinatal exposures (27 sam-
Study population pling frames) considered to be of relevance for timing of
Our study used data from the large nationwide Puberty Cohort puberty and paired these with a randomly selected reference
that is nested within the Danish National Birth Cohort (DNBC) group of 8,000 children (1 sampling frame), making up the
(23). The pregnant women in DNBC were recruited from 1996 Puberty Cohort of 22,439 children. The relevant pre- and perinatal
to 2002. Half of all general practitioners in Denmark partici- exposures included maternal gestational use of acetaminophen.
pated in the recruitment process, and about 60% of the preg- At 11.5 years of age, participating children were invited to
nant women accepted the invitation handed out at their first provide information on their current stage of puberty every 6
antenatal visit. A self-administrated enrollment form and 4 months until they reached full sexual maturation or turned 18
computer-assisted telephone interviews on lifestyle during preg- years of age, whichever came first. Full sexual maturation was
nancy and health of their children were carried out at approxi- defined as self-reported Tanner stage 5 for both pubic hair
mately gestational weeks 17 and 32 as well as 6 and 18 months growth and breast or genital development (24, 25).
after birth. To complement the information on puberty timing in the
A total of 56,641 live-born singletons born to mothers in Puberty Cohort, we added data from an 11-year follow-up of all
DNBC from 2000 to 2003 were eligible for participation in the children in the DNBC. Altogether 10,668 children, subse-
Puberty Cohort. The mothers had responded to the first ques- quently sampled for the Puberty Cohort, responded to a com-
tionnaire in DNBC and not withdrawn their consent before prehensive questionnaire that also included similar questions
2012 (Figure 1). An oversampling strategy was used to create on pubertal development. In total, when data from the 11-year
the Puberty Cohort, aiming to increase statistical efficiency by questionnaire was added, information on pubertal development
of 15,822 children from the Puberty Cohort was available.

Singletons Born in DNBC Between 2000–2003

(n = 61,071)

Excluded

Withdrawn from DNBC before start of the

Puberty Cohort in 2012 (n = 969)
Mothers did not complete the first
questionnaire in DNBC (n = 3,461)

Participants Eligible for the Puberty Cohort

(n = 56,641)

Not Sampled for the Puberty Cohort

(n = 34,202)

Participants Sampled for the Puberty Cohort

(n = 22,439)

Puberty Cohort Questionnaire 11-Year Questionnaire

No reply to any questionnaire in No reply to 11-year questionnaire

the Puberty Cohort (n = 7,680) (n = 10,964)
No information on puberty in 11-year
questionnaire (n = 807)

Information on Puberty in the Information on Puberty in 11-Year

Puberty Cohort (n = 14,759) Questionnaire (n = 10,688)

Participants With Information on Puberty (n = 15,822)

Puberty Cohort (n = 5,154)
11-year questionnaire (n = 1,063)
Both (n = 9,605)

Figure 1. Flow diagram of participants in the Puberty Cohort nested within the Danish National Birth Cohort (DNBC) (n = 15,822), Denmark,
2000–2017.

Am J Epidemiol. 2019;188(1):34–46
36 Ernst et al.

Exposure assessment Statistical analysis

Information on use of acetaminophen was retrieved through The participants in the Puberty Cohort received the ques-
the maternal enrollment form and 3 of the computer-assisted tele- tionnaires half-yearly, which makes data on pubertal mile-
phone interviews, covering use from 4 weeks before pregnancy stones censored. The statistical analyses were therefore carried
until delivery. The enrollment form included questions on use of out using a regression model for censored and normally distrib-
painkillers and specifications on timing of use until the 14th uted time-to-event data (intreg in Stata (StataCorp LLC, Col-
week of pregnancy. In the 3 telephone interviews, women lege Station, Texas)) to estimate mean monthly differences
were asked whether any kind of painkillers, available over-the- between exposure groups in age at attaining the pubertal mile-
counter or via prescription, had been taken. If women answered stones. The regression model is able to account for the left,
“yes,” they were asked to identify the type of drug(s) from a right, or interval censoring of data assuming that the underly-
list of the 44 most common types, including acetaminophen ing distribution of the timing of a pubertal milestone is normal.
as a mono or combination drug. Further questions gave re- The assumption of normality was assessed by comparison
spondents the option to report use of painkillers or drugs not of the stepwise cumulative incidence function and the cumula-
specified in the list, as well as to record intake or drugs used tive incidence function based on the normal distribution using
to treat rheumatic diseases, infections, fever, or inflammation. the R package icenReg (R Foundation for Statistical Comput-
Further, the women were asked to specify in which gestational ing, Vienna, Austria). The normal distribution fit the time of
weeks they used each of the listed drugs. Because the women attained pubertal milestones very well. We also compared the
did not indicate the dose of the painkillers used, cumulative nonparametric distribution to other waiting time distributions
weeks of use indicated exposure level in this study. (Weibull, logistic, and log-normal), but the normal distribu-
tion gave the best fit. Residuals were calculated by subtracting
Outcome measures
the model mean to the left and right interval endpoints. The
assumption of normality of these residuals was checked by a
Information on pubertal development was collected by a similar comparison further stratified by levels of included
translated version of a questionnaire developed by the British covariates.
Avon Longitudinal Study of Parents and Children (ALSPAC) We analyzed associations between intake of acetaminophen
(26). The questionnaire is available at http://www.bsig.dk. at least once during pregnancy and pubertal milestones using
The participants were asked every sixth month to report whether data from children with no recorded exposure during pregnancy
they had experienced the following pubertal milestones: axillary as the unexposed referent. In the next analysis, age at attaining
hair growth (yes/no), acne (yes/no), pubic hair growth and breast the pubertal milestones among children exposed in a cumulative
or genital development according to Tanner staging (1–5), first number of weeks (1–2, 3–12, >12 weeks) was compared with
ejaculation (yes/no), first menstrual bleeding (yes/no), voice age at attaining the milestones among the unexposed, and test for
break (yes (sometimes), yes (definitive change), no, don’t trend was conducted by fitting the number of weeks of exposure
know). If they answered “yes” to first ejaculation (boys) or as a continuous variable. We further assessed timing of exposure
first menstrual bleeding (girls), they were asked to provide comparing exposure only during the first (weeks 1–12 of gesta-
age by year and months. Questions on pubic hair growth and tion), second (weeks 13–24 of gestation), or third (week 25 of
breast or genital growth were guided by illustrations and gestation until delivery) trimester and exposure in any 2 or all 3
explanatory texts for each of the Tanner stages (24, 25). trimesters with the unexposed.
The models were fitted with robust standard errors to account
for clustering of siblings (n = 464 mothers) and sample as well
Covariates
as selection weights to address the sampling strategy and selec-
Directed acyclic graphs were used to identify potential con- tive participation in the Puberty Cohort. The sampling weights
founding factors a priori (27). The following potential confoun- estimate the inverse probability of being sampled for each
ders were prepregnancy body mass index, alcohol consumption individual in the Puberty Cohort according to the 28 different
in the first trimester, smoking in the first trimester, time to preg- sampling frames. The selection weights estimate the inverse
nancy, highest social class of parents, maternal age at menarche, probability of participation according to a directed acyclic graph
maternal age at delivery, and parity. In order to partly control used to identify potential factors associated with participation
for confounding by indication, 3 variables describing conditions in the Puberty Cohort a priori. Information on these factors,
that might induce use of acetaminophen during pregnancy were primarily related to socioeconomic conditions, were available
further included in our models. These were fever during as described above.
pregnancy, muscle or joint disease during pregnancy, and We conducted 2 subanalyses to further address confounding
inflammation or infection during pregnancy. The categorization by indication. First, the analyses were restricted to mothers
of the included potential confounders can be found in Table 1. who did not report any of the 3 mentioned indications. Second,
Information on parity and maternal age was retrieved from we performed analyses with the exclusion of mothers who took
the Danish Medical Birth Register, and socioeconomic status, acetaminophen for more than 24 weeks of gestation (3%),
classified based on the International Standard Class of Occupation because heavy use of acetaminophen during pregnancy
and Education codes (ISCO-88 and ISCED), was retrieved might be more common in women suffering from chronic dis-
from Statistics Denmark. Information on the remaining con- eases causing pain. Because gestational weight gain might
founders was available through the telephone interviews in act as a potential confounder by altering infant size at birth
DNBC. (28), we repeated our analyses including gestational weight

Am J Epidemiol. 2019;188(1):34–46
 Acetaminophen and Timing of Puberty 37

Table 1. Maternal Characteristics According to Sons’ and Daughters’ Exposure to Acetaminophen at Least Once During Gestation, the Puberty
Cohort (n = 15,822), Denmark, 2012–2017

Exposure to Acetaminophen
Missing
Covariatesa No Yes Total No.
No. % No. % No. %

Total 7,216 45.6 8,606 54.4 15,822

Prepregnancy BMIb,c 23.4 (4.4) 24.2 (4.7) 23.8 (4.6) 217 1.4
Alcohol consumption per week in first trimester, units 22 0.1
0 3,853 53.3 4,310 50.1 8,163
>0–1 2,179 30.3 2,751 32.0 4,930
>1–3 818 11.4 1,084 12.6 1,902
>3 352 4.9 453 5.3 805
No. of cigarettes daily in first trimester
c
1.8 (4.2) 2.2 (4.5) 2.0 (4.3) 53 0.3
Time to pregnancy, including ART, months 44 0.3
0 1,380 19.2 1,656 19.3 3,036
1–2 1,331 18.5 1,621 18.9 2,952
3–5 1,158 16.1 1,398 16.3 2,556
6–12 97 13.5 1,067 12.4 2,038
>12 539 7.5 703 8.2 1,242
Use of ART 702 9.8 762 8.9 1,464
Not planned 1,113 15.5 1,377 16.0 2,490
Highest social class of parents 31 0.2
High-grade professional 1,706 23.7 1,984 23.1 3,690
Low-grade professional 2,375 33.0 2,821 32.8 5,196
Skilled worker 2,011 27.9 2,342 27.3 4,353
Unskilled worker 924 12.8 1,225 14.3 2,149
Student 145 2.0 167 1.9 312
Economically inactive 42 0.6 49 0.6 91
Maternal age at menarche 123 0.8
Earlier than peers 1,724 24.1 2,288 26.7 4,012
Same time as peers 4,099 57.4 4,891 57.2 8,990
Later than peers 1,322 18.5 1,375 16.1 2,697
Maternal age at delivery, yearsc 30.7 (4.4) 30.6 (4.4) 30.6 (4.4) 6 0.1
Parity 0 0.0
First child 3,829 53.1 4,138 48.1 7,967
Second or later child 3,387 46.9 4,468 51.9 7,855
Fever during pregnancy 23 0.1
No 5,657 78.5 5,758 67.0 11,415
Yes 1,551 21.5 2,833 33.0 4,384
Muscle or joint disease during pregnancy 7 0.1
No 6,247 86.6 7,038 81.8 13,285
Yes 965 13.4 1,565 18.2 2,530
Inflammation or infection during pregnancy 261 1.6
No 6,205 88.3 7,048 82.6 13,253
Yes 825 11.7 1,483 17.4 2,308

Abbreviations: ART, assisted reproductive techniques; BMI, body mass index.

a
Numbers within each categorized covariate might not necessarily add up to the total number of participants due to missing data.
b
Weight (kg)/height (m)2.
c
Values are expressed as mean (standard deviation).

Am J Epidemiol. 2019;188(1):34–46
38 Ernst et al.

gain as an additional continuous covariate. To address the risk studied markers of pubertal development, but with wide confi-
of type-1 errors due to multiple testing, we performed a test for dence intervals. The magnitudes of the observed associations
the overall association with all puberty markers using the Huber- were largest for stage 2 of Tanner pubic hair (difference,
White robust variance estimation applied on the univariate months = −1.7, 95% CI: −3.2, −0.1), axillary hair growth
marker models, which accounts for the correlations structure (difference, months = −3.0, 95% CI: −5.1, −0.9), and occur-
between the pubertal markers (29, 30). These subanalyses were rence of acne (difference, months = −2.8, 95% CI: −5.0,
performed for cumulative weeks of exposure. −0.5) in the group of girls exposed to acetaminophen for more
Statistical analyses were conducted using Stata/MP, version than 12 weeks.
13.1 (StataCorp LLC), and R x64, version 3.3.1 (R Foundation The analyses on timing of exposure by trimester showed no
for Statistical Computing). specific pattern in boys (Figure 4 and Web Table 2). In girls,
breast development was independent of the specific timing of
Ethical approval exposure, whereas pubic hair development and to a lesser degree
axillary hair growth and occurrence of acne were skewed towards
The Committee for Biomedical Research Ethics in Denmark earlier age at attainment in girls with exposure in one of the tri-
has approved data collection in DNBC ((KF) 01-471/94). A mesters. These tendencies were most pronounced in girls exposed
written informed consent was obtained from mothers upon only in the third trimester (Figure 5 and Web Table 2).
recruitment covering both mother’s and offspring’s participation In the analyses restricted to mothers who did not report any
until the children turned 18 years of age. The present study was of the 3 conditions associated with acetaminophen intake, results
approved by the steering committee of DNBC (2012-04 and were similar to those presented for the binary and trimester-
2015-47) and the Danish Data Protection Agency (2012-41- specific exposure to acetaminophen. The tendency towards
0379 and 2015-57-0002). earlier age at female pubic hair development, however, strength-
ened in the analyses with cumulative weeks of exposure (data
RESULTS not shown). Analyses with exclusion of heavy acetaminophen
users did not change our results substantially. In the analyses
Of the 22,439 children invited to participate in the Puberty with cumulative weeks of exposure, however, female axil-
cohort, 15,822 (approximately 71%), 7,697 boys and 8,125 girls, lary hair growth and occurrence of acne were further shifted
have returned at least 1 and up to 11 questionnaires each towards earlier age at attainment among those who were highly
(Figure 1). Use of acetaminophen was not related to participa- exposed (more than 12 weeks of gestation) (data not shown).
tion in the Puberty Cohort. The responders were, by March 2017, Our subanalyses with additional adjustment for gestational
between 14 and 17 years of age, and 4,317 had reached full sexual weight gain did not change the results (data not shown). Using
maturity (27%). Among the 15,822 participants, 8,606 (54%) robust variance estimation to perform a single test, cumulative
were exposed to acetaminophen at least once during gestation. number of weeks with exposure was associated with earlier
Table 1 shows that prepregnancy body mass index, parity, and female puberty timing (P = 0.04).
reports of the 3 potential indications were higher in mothers
with any gestational intake of acetaminophen. In addition,
maternal smoking and alcohol intake during the first trimester DISCUSSION
were more frequent in this group.
Table 2 presents the mean ages in years at attaining various Results from our population-based cohort study suggest that
male and female pubertal milestones for an unexposed refer- intrauterine exposure to acetaminophen might be associated
ence person as well as the mean monthly differences between with slightly earlier attainment of almost all studied female
those who were and who were not exposed to acetaminophen markers of pubertal development, although most pronounced
at least once during fetal life. The crude estimates suggested a for pubic hair growth, axillary hair growth, and occurrence of
tendency towards earlier age at pubertal timing in exposed acne, but with wide confidence intervals. On the other hand,
boys, which was attenuated in the adjusted analyses. In exposed estimates for boys were randomly scattered around the null
girls, a more consistent tendency towards slightly earlier age at hypothesis, indicating limited if any evidence for an associa-
pubertal timing was observed. The tendency was, however, tion between intrauterine exposure to acetaminophen and male
most pronounced for axillary hair growth, with −1.2 (95% pubertal development. These findings were supported by a test
confidence interval (CI): −2.2, −0.2), and occurrence of acne, of the overall association between exposure and all puberty
with −1.7 (95% CI: −2.7, −0.6) months earlier age at attain- markers. We were not able to identify specific vulnerable expo-
ment. Analyses of cumulative number of weeks with exposure sure windows. Our results suggest that those exposed to acet-
to acetaminophen during pregnancy are presented in Figures 2 aminophen during fetal life might exhibit sex-specific endocrine
and 3 (and Web Table 1, available at https://academic.oup. disruptive effects in relation to pubertal development.
com/aje) as adjusted monthly differences between exposure Prior studies have mainly addressed whether in utero expo-
groups. We observed no strong indication of a pattern with sure to acetaminophen interferes with neurodevelopment and
increasing number of weeks of exposure in boys, although all cognition (2, 4, 31–37), asthma (38), markers of short-term
of the Tanner pubic hair stages were to some extent shifted reproductive health, such as risk of some congenital anomalies
towards later age at attainment in boys exposed for more than (9, 17–22), and differences in anogenital distance (15, 16).
12 weeks. A somewhat consistent pattern was observed Intrauterine exposure to acetaminophen might interfere with
among girls, associating longer duration of intrauterine acet- important fetal developmental processes regulated by prosta-
aminophen exposure with earlier age at attaining almost all glandins (12, 14, 39) and/or exert direct inhibitory effects on

Am J Epidemiol. 2019;188(1):34–46
 Acetaminophen and Timing of Puberty 39

Table 2. Mean Differences in Age at Attaining Various Pubertal Milestones According to Exposure to Acetaminophen at Least Once During
Gestation, the Puberty Cohort, Denmark, 2012–2017

Exposure to Acetaminophen
a Nob Yesc
Pubertal Milestone No. of Persons
Crude Adjustedd
Mean Age, years 95% CI
Mean Difference Mean Difference 95% CI

Boys
Tanner stage, genitals
Stage 2 7,469 11.2 10.5, 11.9 −0.3 0.0 −1.0, 1.0
Stage 3 7,469 12.7 11.9, 13.5 −0.7 −0.2 −1.2, 0.8
Stage 4 7,469 13.7 12.0, 14.4 −0.4 0.1 −0.9, 1.0
Stage 5 7,469 15.2 14.3, 16.0 −0.4 0.3 −1.2, 1.8
Tanner stage, pubic hair
Stage 2 7,473 11.7 11.0, 12.4 −0.7 −0.3 −1.2, 0.7
Stage 3 7,473 12.7 12.0, 13.4 −0.5 0.1 −0.7, 1.0
Stage 4 7,473 13.7 13.2, 14.3 0.1 0.6 −0.2, 1.4
Stage 5 7,473 14.9 14.4, 15.4 −0.3 0.2 −0.8, 1.3
Axillary hair 7,478 13.5 12.7, 14.3 −0.2 0.4 −0.7, 1.4
Acne 7,478 11.5 10.9, 12.2 −0.8 −0.3 −1.3, 0.6
Voice break 7,274 13.1 12.4, 13.8 −0.5 0.1 −0.9, 1.1
Adult voice 7,274 14.4 13.1, 15.6 −0.4 0.5 −1.2, 2.2
First ejaculation 7,465 13.5 12.9, 14.2 −0.7 −0.3 −1.2, 0.7
Girls
Tanner stage, breast
Stage 2 7,888 10.7 9.9, 11.5 −1.3 −0.7 −2.0, 0.7
Stage 3 7,888 12.1 11.5, 12.7 −1.2 −0.6 −1.4, 0.3
Stage 4 7,888 13.4 12.7, 14.1 −1.0 −0.3 −1.3, 0.6
Stage 5 7,888 16.0 15.1, 17.0 −1.5 −0.4 −2.1, 1.2
Tanner stage, pubic hair
Stage 2 7,889 11.4 10.9, 11.9 −1.0 −0.7 −1.5, 1.2
Stage 3 7,889 12.9 12.5, 13.3 −0.9 −0.6 −1.3, 0.1
Stage 4 7,889 13.8 13.4, 14.3 −1.0 −0.7 −1.6, 0.2
Stage 5 7,889 15.7 14.9, 16.5 −1.3 −0.8 −2.1, 0.6
Axillary hair 7,894 12.2 11.6, 12.9 −1.7 −1.2 −2.2, −0.2
Acne 7,894 11.9 11.4, 12.5 −2.1 −1.7 −2.7, −0.6
Menarche 7,886 13.4 12.9, 13.8 −0.7 −0.1 −0.8, 0.7

Abbreviations: BMI, body mass index; CI, confidence interval.

a
Number of persons included in analysis for each milestone.
b
Mean age in years at attaining pubertal milestones for a given reference person with no exposure to acetaminophen: prepregnancy BMI of 18.5
(calculated as weight (kg)/height (m)2), 0 alcohol units per week in first trimester, 0 cigarettes daily in first trimester, 0 months time to pregnancy,
highest social class of parents was high-grade professional, maternal age of menarche same as peers, maternal age at delivery of 30.7 years, par-
ity: first child, no fever during pregnancy, no muscle or joint disease during pregnancy, and no inflammation or infection during pregnancy.
c
Mean difference (months) at attaining milestone between unexposed and exposed.
d
Adjusted for prepregnancy BMI, alcohol units per week in first trimester, daily number of cigarettes in first trimester, time to pregnancy, highest
social class of parents, maternal age of menarche, maternal age at delivery, parity, fever during pregnancy, muscle or joint disease during preg-
nancy, and inflammation or infection during pregnancy.

fetal testosterone production (9, 10). However, it remains Puberty is initiated by a reactivation of the hypothalamic-
unknown to what extent such potential effects are translat- pituitary-gonadal axis that stimulates androgen production,
able to humans and might induce long-lasting effects on the causing testicular and pubic hair growth in boys, and stimu-
complex neuroendocrine system that regulates pubertal lates ovarian estrogen production, causing breast enlarge-
development and which parts of the regulatory system might ment and menarche in girls. Girls’ pubic and axillary hair
be susceptible to acetaminophen exposure. growth are initially regulated by androgens from the adrenal

Am J Epidemiol. 2019;188(1):34–46
40 Ernst et al.

Milestone and Cumulative

Exposure, weeks Difference (95% CI)

Tanner genital
Stage 2
1−2 Weeks 0.7 (−0.6, 1.9)
3−12 Weeks −0.1 (−1.7, 1.5)
>12 Weeks 0.9 (−1.4, 3.3)
Stage 3
1−2 Weeks −0.1 (−1.4, 1.1)
3−12 Weeks −0.1 (−1.6, 1.5)
>12 Weeks −0.4 (−2.6, 1.8)
Stage 4
1−2 Weeks 0.3 (−0.9, 1.5)
3−12 Weeks 0.3 (−1.3, 1.8)
>12 Weeks 0.2 (−2.1, 2.4)
Stage 5
1−2 Weeks −0.1 (−2.0, 1.8)
3−12 Weeks 0.3 (−2.1, 2.7)
>12 Weeks 0.1 (−3.3, 3.5)
Tanner pubic hair
Stage 2
1−2 Weeks 0.1 (−1.1, 1.3)
3−12 Weeks −1.1 (−2.7, 0.4)
>12 Weeks 1.2 (−1.0, 3.4)
Stage 3
1−2 Weeks −0.1 (−1.2, 0.9)
3−12 Weeks 0.5 (−0.9, 1.9)
>12 Weeks 0.9 (−1.0, 2.8)
Stage 4
1−2 Weeks 0.5 (−0.5, 1.6)
3−12 Weeks 0.7 (−0.6, 1.9)
>12 Weeks 1.2 (−0.6, 2.9)
Stage 5
1−2 Weeks 0.4 (−1.0, 1.7)
3−12 Weeks −0.1 (−1.7, 1.6)
>12 Weeks 1.4 (−0.9, 3.6)
Axillary hair
1−2 Weeks 0.1 (−1.3, 1.4)
3−12 Weeks 0.6 (−1.0, 2.3)
>12 Weeks 0.5 (−1.8, 2.8)
Acne
1−2 Weeks −0.8 (−2.0, 0.4)
3−12 Weeks −0.3 (−1.8, 1.2)
>12 Weeks −0.4 (−2.5, 1.7)
Voice break
1−2 Weeks −0.4 (−1.7, 0.8)
3−12 Weeks 0.3 (−1.3, 1.9)
>12 Weeks 0.6 (−1.7, 2.9)
Adult voice
1−2 Weeks 0.3 (−1.9, 2.6)
3−12 Weeks 0.6 (−2.1, 3.3)
>12 Weeks 0.4 (−3.0, 3.8)
First ejaculation
1−2 Weeks 0.1 (−1.2, 1.3)
3−12 Weeks −0.1 (−1.6, 1.5)
>12 Weeks −1.5 (−3.7, 0.7)

−6 −3 0 3 6
Mean Difference, months

Figure 2. Adjusted mean differences (months, 95% confidence intervals (CIs)) in age at attaining various pubertal milestones in boys according
to number of weeks with exposure to acetaminophen during gestation, the Puberty Cohort, Denmark, 2012–2017.

Am J Epidemiol. 2019;188(1):34–46
 Acetaminophen and Timing of Puberty 41

Milestone and Cumulative

Exposure, weeks Difference (95% CI)

Tanner breast
Stage 2
1−2 Weeks −1.0 (−2.6, 0.7)
3−12 Weeks −0.9 (−3.0, 1.2)
>12 Weeks −0.7 (−3.6, 2.3)
Stage 3
1−2 Weeks −0.6 (−1.7, 0.4)
3−12 Weeks −0.5 (−1.8, 0.9)
>12 Weeks −1.6 (−3.6, 0.5)
Stage 4
1−2 Weeks −0.9 (−2.0, 0.3)
3−12 Weeks −0.2 (−1.5, 1.1)
>12 Weeks 0.2 (−1.8, 2.2)
Stage 5
1−2 Weeks −1.4 (−3.5, 0.8)
3−12 Weeks 0.2 (−2.3, 2.8)
>12 Weeks −1.4 (−4.6, 1.9)
Tanner pubic hair
Stage 2
1−2 Weeks −0.5 (−1.4, 0.5)
3−12 Weeks −1.6 (−2.7, −0.5)
>12 Weeks −1.7 (−3.2, −0.1)
Stage 3
1−2 Weeks −0.5 (−1.4, 0.4)
3−12 Weeks −1.1 (−2.1, −0.1)
>12 Weeks −1.4 (−3.0, 0.3)
Stage 4
1−2 Weeks −1.0 (−2.2, 0.2)
3−12 Weeks −0.5 (−1.8, 0.9)
>12 Weeks −1.4 (−3.5, 0.8)
Stage 5
1−2 Weeks −1.1 (−2.8, 0.7)
3−12 Weeks −1.3 (−3.3, 0.8)
>12 Weeks −1.3 (−4.1, 1.6)
Axillary hair
1−2 Weeks −1.4 (−2.7, −0.2)
3−12 Weeks −0.6 (−2.1, 0.9)
>12 Weeks −3.0 (−5.1, −0.9)
Acne
1−2 Weeks −0.8 (−2.2, 0.6)
3−12 Weeks −2.5 (−4.1, −0.9)
>12 Weeks −2.8 (−5.0, −0.5)
Menarche
1−2 Weeks 0.2 (−0.8, 1.1)
3−12 Weeks −0.2 (−1.3, 0.9)
>12 Weeks −0.9 (−2.5, −0.7)

−6 −3 0 3 6
Mean Difference, months

Figure 3. Adjusted mean differences (months, 95% confidence intervals (CIs)) in age at attaining various pubertal milestones in girls according to
number of weeks with exposure to acetaminophen during gestation, the Puberty Cohort, Denmark, 2012–2017.

Am J Epidemiol. 2019;188(1):34–46
42 Ernst et al.

Milestone and Trimester

of Exposure Difference (95% CI)

Tanner genital
Stage 2
First −0.5 (−2.1, 1.1)
Second −0.1 (−2.1, 1.9)
Third −0.2 (−1.9, 1.4)
Any 2 or 3 0.4 (−0.9, 1.6)
Stage 3
First 0.4 (−1.2, 2.0)
Second −0.2 (−2.0, 1.7)
Third −1.2 (−2.7, 0.4)
Any 2 or 3 0.0 (−1.2, 1.2)
Stage 4
First 0.2 (−1.4, 1.8)
Second 1.6 (−0.3, 3.5)
Third −0.6 (−2.1, 0.9)
Any 2 or 3 0.0 (−1.2, 1.2)
Stage 5
First 0.9 (−1.6, 3.3)
Second 3.0 (0.1, 5.9)
Third −0.2 (−2.5, 2.0)
Any 2 or 3 −0.4 (−2.2, 1.5)
Tanner pubic hair
Stage 2
First 0.6 (−0.9, 2.2)
Second −0.9 (−2.7, 0.9)
Third −0.2 (−1.7, 1.4)
Any 2 or 3 −0.5 (−1.7, 0.7)
Stage 3
First 0.4 (−1.1, 1.8)
Second 0.4 (−1.2, 2.0)
Third −0.4 (−1.8, 0.9)
Any 2 or 3 0.2 (−0.9, 1.2)
Stage 4
First 0.9 (−0.4, 2.3)
Second 1.3 (−0.2, 2.8)
Third 0.0 (−1.3, 1.3)
Any 2 or 3 0.5 (−0.5, 1.5)
Stage 5
First 0.6 (−1.2, 2.3)
Second 1.1 (−0.9, 3.1)
Third −0.1 (−1.8, 1.6)
Any 2 or 3 0.0 (−1.3, 1.3)
Axillary hair
First 1.4 (−0.3, 3.1)
Second 0.3 (−1.8, 2.3)
Third −1.2 (−2.9, 0.5)
Any 2 or 3 0.6 (−0.7, 1.9)
Acne
First 0.0 (−1.6, 1.6)
Second −0.5 (−2.3, 1.2)
Third −0.4 (−1.9, 1.1)
Any 2 or 3 −0.4 (−1.6, 0.8)
Voice break
First −0.3 (−2.0, 1.4)
Second −1.2 (−3.1, 0.7)
Third −0.2 (−1.8, 1.4)
Any 2 or 3 0.7 (−0.6, 1.9)
Adult voice
First 0.6 (−2.2, 3.4)
Second 1.3 (−2.0, 4.5)
Third −2.0 (−4.7, 0.7)
Any 2 or 3 1.4 (−0.7, 3.5)
First ejaculation
First −0.2 (−1.7, 1.3)
Second 1.6 (−0.3, 3.5)
Third −0.2 (−1.8, 1.3)
Any 2 or 3 −0.8 (−2.0, 0.5)

−6 −3 0 3 6
Mean Difference, months

Figure 4. Adjusted mean differences (months, 95% confidence intervals (CIs)) in age at attaining various pubertal milestones in boys according
to trimester of exposure to acetaminophen during gestation, the Puberty Cohort, Denmark, 2012–2017.

Am J Epidemiol. 2019;188(1):34–46
 Acetaminophen and Timing of Puberty 43

Milestone and Trimester

of Exposure Difference (95% CI)

Tanner breast
Stage 2
First −0.6 (−2.7, 1.6)
Second −2.0 (−4.5, 0.6)
Third 0.3 (−1.9, 2.4)
Any 2 or 3 −0.8 (−2.5, 0.9)
Stage 3
First −0.2 (−1.6, 1.1)
Second −0.8 (−2.4, 0.8)
Third −0.4 (−1.7, 1.0)
Any 2 or 3 −0.7 (−1.8, 0.4)
Stage 4
First −0.3 (−1.7, 1.1)
Second −0.1 (−1.8, 1.7)
Third −1.0 (−2.3, 0.4)
Any 2 or 3 −0.2 (−1.3, 1.0)
Stage 5
First −0.7 (−3.4, 1.9)
Second −0.4 (−3.6, 2.9)
Third 0.2 (−2.5, 2.8)
Any 2 or 3 −0.5 (−2.6, 1.5)
Tanner pubic hair
Stage 2
First −0.4 (−1.5, 0.7)
Second 0.3 (−1.2, 1.7)
Third −1.4 (−2.6, −0.2)
Any 2 or 3 −0.9 (−1.7, 0.0)
Stage 3
First −0.3 (−1.3, 0.8)
Second 0.0 (−1.3, 1.4)
Third −1.4 (−2.5, −0.3)
Any 2 or 3 −0.5 (−1.4, 0.4)
Stage 4
First −0.7, (−2.1, 0.7)
Second −0.5 (−2.4, 1.3)
Third −1.7 (−3.1, −0.2)
Any 2 or 3 −0.3 (−1.4, 0.9)
Stage 5
First −0.9 (−3.0, 1.3)
Second −1.1 (−3.8, 1.5)
Third −2.4 (−4.6, −0.2)
Any 2 or 3 0.1 (−1.6, 1.7)
Axillary hair
First −0.9 (−2.5, 0.7)
Second −0.7 (−2.5, 1.2)
Third −1.5 (−3.1, 0.1)
Any 2 or 3 −1.4 (−2.6, −0.2)
Acne
First −1.5 (−3.2, 0.2)
Second −0.8 (−2.9, 1.3)
Third −1.6 (−3.4, 0.1)
Any 2 or 3 −1.9 (−3.2, −0.6)
Menarche
First 1.0 (−0.2, 2.1)
Second −0.7 (−2.2, 0.8)
Third −0.2 (−1.4,0.9)
Any 2 or 3 −0.3 (−1.2, 0.6)

−6 −3 0 3 6
Mean Difference, months

Figure 5. Adjusted mean differences (months, 95% confidence intervals (CIs)) in age at attaining various pubertal milestones in girls according to
trimester of exposure to acetaminophen during gestation, the Puberty Cohort, Denmark, 2012–2017.

Am J Epidemiol. 2019;188(1):34–46
44 Ernst et al.

glands, and at later stages of puberty, hair growth becomes indications, cumulative number of weeks with exposure was
hypothalamic-pituitary-gonadal-dependent (40, 41). Earlier more strongly associated with earlier age at attaining pubic hair
age at onset of pubic and axillary hair growth in girls, as development, especially for children with more than 12 weeks
seen in our study, might suggest that acetaminophen inter- of exposure. This finding could be partially explained by simi-
feres with peripheral androgen production rather than inducing lar indications among this group of mothers, because these indi-
hypothalamic-pituitary-gonadal axis dysfunction. However, if cations were slightly associated with earlier puberty timing in
acetaminophen also exerts peripheral antiandrogenic effects, our data. A prolonged use of acetaminophen during pregnancy
lack of androgen would be expected to delay rather than is more common in women suffering from chronic diseases
accelerate puberty in girls. such as rheumatic diseases, but we are not aware of studies that
Major strengths of the study are the high participation rate and have investigated whether this chronic disease is associated
the use of a variety of different markers of pubertal development. with pubertal development. However, exclusion of heavy
Further, information on exposure and pubertal development users of acetaminophen, defined as use for more than 24
was collected close to the time of the event (23, 42). The par- weeks of gestation (3%), did not change our results substan-
ticipants ranked their current pubertal stage by a wide range tially. Still, we cannot exclude the possibility that residual
of markers every 6 months beginning at age 11 years. A large confounding by indication or unmeasured confounding fac-
proportion of the boys and girls had already experienced tors might serve as alternative explanations, although that
some of the early puberty stages at inclusion. Our regression would require that these factors are also strong risk factors for
model was, however, able to account for the censoring of earlier pubertal development in girls. A potential confounding
data assuming that the underlying distribution of the timing factor could be unmeasured, poor, family-related lifestyle be-
of pubertal milestones is normal. A recent consensus paper havior that increased maternal intake of acetaminophen and
strongly recommended the use of these types of regression mod- accelerated pubertal timing by increasing the risk of high
els in longitudinal studies with repeated measures of pubertal childhood body mass index. To examine the potential impact
development (43). of residual confounding by indication from conditions for
We used self-administrated questionnaires, including Tanner long-term use and unmeasured confounding from, for exam-
rating scales, to assess pubertal development. Although studies ple, poor overall health, we conducted a multidimensional bias
on the reliability of self-assessment of pubertal development analysis using the approach suggested by VanderWeele and
using Tanner staging have been conflicting (44–47), any Arah (48). Even after applying realistic scenarios for female
potential misclassification would most likely be nondifferen- axillary hair growth, the impact of potential bias was minimal.
tial in this study. Further, self-reporting of pubertal develop- Description and results of this analysis are found in the Web
ment serves as a cost- and timing-savings tool in large material (Web Appendix 1 and Web Table 3).
longitudinal cohorts such as the Puberty Cohort that do In conclusion, our study suggests a tendency towards earlier
not pose the same strong risk of selection bias compared female pubertal development with increasing number of weeks
with clinical examinations. of intrauterine exposure to acetaminophen but seems to show
Information on use of acetaminophen was collected prior to no strong associations, if any, with male pubertal development.
outcome occurrence, including details on timing and duration of The results suggest that acetaminophen might interfere with pre-
use. Lack of information and recall bias are therefore expected to natal, sex-specific developmental processes that lead to long-
be independent of later pubertal development. We used time of term effects on pubertal development. Although the observed
interview to assign trimester-specific exposure when mothers shifts in female puberty timing were minor, these findings could
stated that they took acetaminophen but were unable to specify signal other endocrine-related changes that might have later repro-
timing of use. In addition, 2,357 interviewees (15%) were unable ductive and health disadvantages, as suggested by others (49–52).
to identify intake on a weekly basis, thereby introducing some
uncertainty to the analyses on cumulative exposure. Moreover,
the grouping of number of weeks of exposure was a crude indi-
cator for cumulative duration of exposure, because it does not ACKNOWLEDGMENTS
take into account the frequencies of use. This misclassification
would often lead to an underestimation of the effect sizes. Author affiliations: Section for Epidemiology,
Similarly, we were not able to account for the frequencies Department of Public Health, Aarhus University, Aarhus,
of use in the analyses with timing of exposure. Further, by Denmark (Andreas Ernst, Nis Brix, Lea L.B. Lauridsen,
combining intake of acetaminophen in any 2 or all 3 trimesters Cecilia H. Ramlau-Hansen); Department of Epidemiology,
into one category, effects of cumulative and time-specific ex- Fielding School of Public Health, University of California
posure to acetaminophen might be difficult to separate. We Los Angeles, Los Angeles, California (Andreas Ernst, Nis
believe, however, that by comparing children with no exposure Brix, Jørn Olsen, Zeyan Liew); Department of Clinical
to children with exposure in one specific trimester, we provide Epidemiology, Aarhus University Hospital, Aarhus,
some insight into potential vulnerable exposure windows. Denmark (Jørn Olsen); Section for Biostatistics, Department
We used inverse probability weighting to account for the of Public Health, Aarhus University, Aarhus, Denmark (Erik
sampling procedure and nonresponse and controlled for several T. Parner); and Section for Paediatric Urology, Department
potential confounders. Use of acetaminophen was, as expected, of Urology, Aarhus University Hospital, Aarhus, Denmark
more common in pregnant women who reported episodes of (Lars H. Olsen).
inflammation or infection, fever, or musculoskeletal disorders. This work was supported by the Danish Council for
In subanalyses restricted to women with no reports of these Independent Research (DFF 4183-00152 to C.H.R-.H.) and

Am J Epidemiol. 2019;188(1):34–46
 Acetaminophen and Timing of Puberty 45

the Faculty of Health at Aarhus University. The Danish differentiating genital tract of the fetal mouse. Endocrinology.
National Birth Cohort was established with a significant 1989;124(1):129–133.
grant from the Danish National Research Foundation. 12. van den Driesche S, Macdonald J, Anderson RA, et al.
Additional support was obtained from the Danish Regional Prolonged exposure to acetaminophen reduces testosterone
production by the human fetal testis in a xenograft model. Sci
Committees, the Pharmacy Foundation, the Egmont
Transl Med. 2015;7(288):288ra80.
Foundation, the March of Dimes Birth Defects Foundation, 13. Mazaud-Guittot S, Nicolas Nicolaz C, Desdoits-Lethimonier
the Health Foundation, and other minor grants. The DNBC C, et al. Paracetamol, aspirin, and indomethacin induce
Biobank has been supported by the Novo Nordisk endocrine disturbances in the human fetal testis capable of
Foundation and the Lundbeck Foundation. Follow-up of interfering with testicular descent. J Clin Endocrinol Metab.
mothers and children has been supported by the Danish 2013;98(11):E1757–E1767.
Medical Research Council (grants SSVF 0646, 271-08- 14. Dean A, van den Driesche S, Wang Y, et al. Analgesic
0839/06-066023, O602-01042B, 0602-02738B), the exposure in pregnant rats affects fetal germ cell development
Lundbeck Foundation (grants 195/04 and R100-A9193), the with inter-generational reproductive consequences. Sci Rep.
Innovation Fund Denmark (grant 0603-00294B (09- 2016;6:19789.
15. Fisher BG, Thankamony A, Hughes IA, et al. Prenatal
067124)), the Nordea Foundation (grant 02-2013-2014),
paracetamol exposure is associated with shorter anogenital
Aarhus Ideas (grant AU R9-A959-13-S804). University of distance in male infants. Hum Reprod. 2016;31(11):
Copenhagen (Strategic Grant IFSV 2012), and the Danish 2642–2650.
Council for Independent Research (grants DFF–4183-00594 16. Lind DV, Main KM, Kyhl HB, et al. Maternal use of mild
and DFF–4183-00152). analgesics during pregnancy associated with reduced
Conflict of interest: none declared. anogenital distance in sons: a cohort study of 1027 mother-
child pairs. Hum Reprod. 2017;32(1):223–231.
17. Snijder CA, Kortenkamp A, Steegers EA, et al. Intrauterine
exposure to mild analgesics during pregnancy and the
occurrence of cryptorchidism and hypospadia in the offspring:
REFERENCES the Generation R Study. Hum Reprod. 2012;27(4):1191–1201.
18. Lind JN, Tinker SC, Broussard CS, et al. Maternal medication
1. Hudec R, Božeková L, Tisoňová J. Consumption of three most and herbal use and risk for hypospadias: data from the National
widely used analgesics in six European countries. J Clin Birth Defects Prevention Study, 1997–2007.
Pharm Ther. 2012;37(1):78–80. Pharmacoepidemiol Drug Saf. 2013;22(7):783–793.
2. Brandlistuen RE, Ystrom E, Nulman I, et al. Prenatal 19. Jensen MS, Henriksen TB, Rebordosa C, et al. Analgesics
paracetamol exposure and child neurodevelopment: a sibling- during pregnancy and cryptorchidism: additional analyses.
controlled cohort study. Int J Epidemiol. 2013;42(6): Epidemiology. 2011;22(4):610–612.
1702–1713. 20. Jensen MS, Rebordosa C, Thulstrup AM, et al. Maternal use of
3. Werler MM, Mitchell AA, Hernandez-Diaz S, et al. Use of acetaminophen, ibuprofen, and acetylsalicylic acid during
over-the-counter medications during pregnancy. Am J Obstet pregnancy and risk of cryptorchidism. Epidemiology. 2010;
Gynecol. 2005;193(3 pt 1):771–777. 21(6):779–785.
4. Liew Z, Ritz B, Rebordosa C, et al. Acetaminophen use during 21. Rebordosa C, Kogevinas M, Horváth-Puhó E, et al.
pregnancy, behavioral problems, and hyperkinetic disorders. Acetaminophen use during pregnancy: effects on risk for
JAMA Pediatr. 2014;168(4):313–320. congenital abnormalities. Am J Obstet Gynecol. 2008;198(2):
5. Black RA, Hill DA. Over-the-counter medications in 178.e1–178.e7.
pregnancy. Am Fam Physician. 2003;67(12):2517–2524. 22. Wagner-Mahler K, Kurzenne JY, Delattre I, et al. Prospective
6. Syme MR, Paxton JW, Keelan JA. Drug transfer and study on the prevalence and associated risk factors of
metabolism by the human placenta. Clin Pharmacokinet. 2004; cryptorchidism in 6246 newborn boys from Nice area, France.
43(8):487–514. Int J Androl. 2011;34(5 pt 2):e499–e510.
7. Rayburn W, Shukla U, Stetson P, et al. Acetaminophen 23. Olsen J, Melbye M, Olsen SF, et al. The Danish National Birth
pharmacokinetics: comparison between pregnant and Cohort—its background, structure and aim. Scand J Public
nonpregnant women. Am J Obstet Gynecol. 1986;155(6): Health. 2001;29(4):300–307.
1353–1356. 24. Marshall WA, Tanner JM. Variations in pattern of pubertal
8. Kristensen DM, Mazaud-Guittot S, Gaudriault P, et al. changes in girls. Arch Dis Child. 1969;44(235):291–303.
Analgesic use—prevalence, biomonitoring and endocrine and 25. Marshall WA, Tanner JM. Variations in the pattern of pubertal
reproductive effects. Nat Rev Endocrinol. 2016;12(7): changes in boys. Arch Dis Child. 1970;45(239):13–23.
381–393. 26. Monteilh C, Kieszak S, Flanders WD, et al. Timing of
9. Kristensen DM, Hass U, Lesné L, et al. Intrauterine exposure to maturation and predictors of Tanner stage transitions in boys
mild analgesics is a risk factor for development of male enrolled in a contemporary British cohort. Paediatr Perinat
reproductive disorders in human and rat. Hum Reprod. 2011; Epidemiol. 2011;25(1):75–87.
26(1):235–244. 27. Pearl J. Causality: Models, Reasoning and Inference. 2nd ed.
10. Kristensen DM, Lesné L, Le Fol V, et al. Paracetamol Cambridge, United Kingdom: Cambridge University Press;
(acetaminophen), aspirin (acetylsalicylic acid) and 2009.
indomethacin are anti-androgenic in the rat foetal testis. Int J 28. Nohr EA, Vaeth M, Baker JL, et al. Combined associations of
Androl. 2012;35(3):377–384. prepregnancy body mass index and gestational weight gain
11. Gupta C. The role of prostaglandins in masculine with the outcome of pregnancy. Am J Clin Nutr. 2008;87(6):
differentiation: modulation of prostaglandin levels in the 1750–1759.

Am J Epidemiol. 2019;188(1):34–46
46 Ernst et al.

29. White H. A heteroskedasticity-consistent covariance matrix 41. Grumbach MM. The neuroendocrinology of human puberty
estimator and a direct test for heteroskedasticity. revisited. Horm Res. 2002;57(suppl 2):2–14.
Econometrica. 1980;48(4):817–838. 42. Rockett JC, Lynch CD, Buck GM. Biomarkers for assessing
30. Huber PJ. The behavior of maximum likelihood estimates under reproductive development and health: part 1–pubertal
nonstandard conditions. Presented at the Fifth Berkeley Symposium development. Environ Health Perspect. 2004;112(1):
on Mathematical Statistics and Probability, Berkeley, CA, June 21– 105–112.
July 18, 1965 and December 27, 1965–January 7, 1966. 43. Euling SY, Herman-Giddens ME, Lee PA. Examination of
31. Liew Z, Bach CC, Asarnow RF, et al. Paracetamol use during US puberty-timing data from 1940 to 1994 for secular
pregnancy and attention and executive function in offspring at trends: panel findings. Pediatrics. 2008;121(suppl 3):
age 5 years. Int J Epidemiol. 2016;45(6):2009–2017. S172–S191.
32. Liew Z, Ritz B, Virk J, et al. Prenatal use of acetaminophen and 44. Duke PM, Litt IF, Gross RT. Adolescents’ self-assessment of
child IQ: a Danish cohort study. Epidemiology. 2016;27(6): sexual maturation. Pediatrics. 1980;66(6):918–920.
912–918. 45. Chan NP, Sung RY, Kong AP, et al. Reliability of pubertal
33. Bornehag CG, Reichenberg A, Hallerback MU, et al. Prenatal self-assessment in Hong Kong Chinese children. J Paediatr
exposure to acetaminophen and children’s language Child Health. 2008;44(6):353–358.
development at 30 months. Eur Psychiatry. 2018;51:98–103. 46. Desmangles JC, Lappe JM, Lipaczewski G, et al. Accuracy of
34. Stergiakouli E, Thapar A, Davey Smith G. Association of pubertal Tanner staging self-reporting. J Pediatr Endocrinol
acetaminophen use during pregnancy with behavioral Metab. 2006;19(3):213–221.
problems in childhood: evidence against confounding. JAMA 47. Hergenroeder AC, Hill RB, Wong WW, et al. Validity of self-
Pediatr. 2016;170(10):964–970. assessment of pubertal maturation in African American and
35. Thompson JM, Waldie KE, Wall CR, et al. Associations European American adolescents. J Adolesc Health. 1999;
between acetaminophen use during pregnancy and ADHD 24(3):201–205.
symptoms measured at ages 7 and 11 years. PLoS One. 2014; 48. VanderWeele TJ, Arah OA. Bias formulas for sensitivity
9(9):e108210. analysis of unmeasured confounding for general outcomes,
36. Vlenterie R, Wood ME, Brandlistuen RE, et al. treatments, and confounders. Epidemiology. 2011;22(1):
Neurodevelopmental problems at 18 months among children 42–52.
exposed to paracetamol in utero: a propensity score matched 49. Berkey CS, Frazier AL, Gardner JD, et al. Adolescence and
cohort study. Int J Epidemiol. 2016;45(6):1998–2008. breast carcinoma risk. Cancer. 1999;85(11):2400–2409.
37. Ystrom E, Gustavson K, Brandlistuen RE, et al. Prenatal exposure 50. Moss AR, Osmond D, Bacchetti P, et al. Hormonal risk factors
to acetaminophen and risk of ADHD. Pediatrics. 2017;140(5). in testicular cancer. A case-control study. Am J Epidemiol.
38. Cheelo M, Lodge CJ, Dharmage SC, et al. Paracetamol 1986;124(1):39–52.
exposure in pregnancy and early childhood and development 51. Frontini MG, Srinivasan SR, Berenson GS. Longitudinal
of childhood asthma: a systematic review and meta-analysis. changes in risk variables underlying metabolic Syndrome X
Arch Dis Child. 2015;100(1):81–89. from childhood to young adulthood in female subjects with a
39. Bayne RA, Eddie SL, Collins CS, et al. Prostaglandin E2 as a history of early menarche: the Bogalusa Heart Study. Int J
regulator of germ cells during ovarian development. J Clin Obes Relat Metab Disord. 2003;27(11):1398–1404.
Endocrinol Metab. 2009;94(10):4053–4060. 52. Freedman DS, Khan LK, Serdula MK, et al. The relation of
40. Ebling FJ. The neuroendocrine timing of puberty. menarcheal age to obesity in childhood and adulthood: the
Reproduction. 2005;129(6):675–683. Bogalusa Heart Study. BMC Pediatr. 2003;3:3.

Am J Epidemiol. 2019;188(1):34–46