Content uploaded by Shanna H Swan
Author content
All content in this area was uploaded by Shanna H Swan on Oct 16, 2014
Content may be subject to copyright.
DOI: 10.1542/peds.2007-1813D
2008;121;S172Pediatrics
Shanna H. Swan
andAnders Juul, Thorkild I.A. Sørensen, Leo Dunkel, John H. Himes, Grete Teilmann
Susan Y. Euling, Marcia E. Herman-Giddens, Peter A. Lee, Sherry G. Selevan,
Panel Findings
Examination of US Puberty-Timing Data from 1940 to 1994 for Secular Trends:
http://pediatrics.aappublications.org/content/121/Supplement_3/S172.full.html
located on the World Wide Web at:
The online version of this article, along with updated information and services, is
of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.
Boulevard, Elk Grove Village, Illinois, 60007. Copyright © 2008 by the American Academy
published, and trademarked by the American Academy of Pediatrics, 141 Northwest Point
publication, it has been published continuously since 1948. PEDIATRICS is owned,
PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
SUPPLEMENT ARTICLE
Examination of US Puberty-Timing Data from 1940 to
1994 for Secular Trends: Panel Findings
Susan Y. Euling, PhD
a
, Marcia E. Herman-Giddens, PA, DrPH
b
, Peter A. Lee, MD, PhD
c
, Sherry G. Selevan, PhD
a
, Anders Juul, MD, DMSc
d
,
Thorkild I. A. Sørensen, MD, DMSc
e
, Leo Dunkel, PhD
f
, John H. Himes, PhD, MPH
g
, Grete Teilmann, MD, PhD
d
, Shanna H. Swan, PhD
h
a
National Center for Environmental Assessment, Office of Research and Development, US Environmental Protection Agency, Washington, DC;
b
Department of Maternal
and Child Health, School of Public Health, University of North Carolina, Chapel Hill, North Carolina;
c
Department of Pediatrics, Penn State College of Medicine, Milton S.
Hershey Medical Center, Hershey, Pennsylvania;
d
University Department of Growth and Reproduction, Rigshospitalet, Blegdamsvej, Copenhagen, Denmark;
e
Danish
Epidemiology Science Centre, Institute of Preventive Medicine, Copenhagen University Hospital, Copenhagen, Denmark;
f
Division of Endocrinology DIabetes and
Metabolic Diseases, Hospital for Children and Adolescents, Biomedicum, University of Helsinki;
g
Division of Epidemiology, University of Minnesota School of Public
Health, Minneapolis, Minnesota;
h
Department of Family & Communty Medicine, University of Missouri, Columbia, Missouri
The authors have indicated they have no financial relationships relevant to this article to disclose.
ABSTRACT
Whether children, especially girls, are entering and progressing through puberty
earlier today than in the mid-1900s has been debated. Secular trend analysis, based
on available data, is limited by data comparability among studies in different popu-
lations, in different periods of time, and using different methods. As a result,
conclusions from data comparisons have not been consistent. An expert panel was
asked to evaluate the weight of evidence for whether the data, collected from 1940
to 1994, are sufficient to suggest or establish a secular trend in the timing of puberty
markers in US boys or girls. A majority of the panelists agreed that data are sufficient
to suggest a trend toward an earlier breast development onset and menarche in girls
but not for other female pubertal markers. A minority of panelists concluded that the
current data on girls’ puberty timing for any marker are insufficient. Almost all
panelists concluded, on the basis of few studies and reliability issues of some male
puberty markers, that current data for boys are insufficient to evaluate secular trends
in male pubertal development. The panel agreed that altered puberty timing should
be considered an adverse effect, although the magnitude of change considered
adverse was not assessed. The panel recommended (1) additional analyses of existing
puberty-timing data to examine secular trends and trends in the temporal sequence
of pubertal events; (2) the development of biomarkers for pubertal timing and
methods to discriminate fat versus breast tissue, and (3) establishment of cohorts to
examine pubertal markers longitudinally within the same individuals.
C
HANGING TRENDS IN the timing of pubertal development may identify public
health concerns. Before the mid-1900s, secular trend analyses of age at female
puberty relied on studies of age at menarche among specific populations, and these
studies suggest that the average age of menarche declined in the United States
between the late 1800s and the mid-1900s.
1,2
This decline was attributed to improve-
ments in general health, nutrition, and other living conditions during this time
frame. From the mid-1900s to the present, some studies have reported that the age
at menarche and breast and pubic hair development have declined in US girls
3–5
;
however, individual study methods and study comparisons have been challenged,
and different studies have drawn different conclusions, leading to an unresolved
debate regarding a secular trend in female puberty timing during this time frame.
Although there are fewer published studies of puberty timing for US boys, some
studies have reported an earlier puberty onset in boys as well.
6,7
An expert panel was asked to resolve the debate by performing a thorough
weight-of-the-evidence evaluation of the data during and after “The Role of Envi-
ronmental Factors on the Onset and Progression of Puberty” Workshop (see intro-
duction article
8
for details). This is the first effort, to our knowledge, to develop
consensus on whether there has been a secular trend in the timing of puberty in the
www.pediatrics.org/cgi/doi/10.1542/
peds.2007-1813D
doi:10.1542/peds.2007-1813D
The views expressed in this article are
those of the authors and do not necessarily
reflect the views or policies of the US
Environmental Protection Agency. Mention
of trade names of commercial products
does not constitute endorsement or
recommendation for use.
Dr Dunkel’s current affiliation is
Department of Pediatrics, Kuopio
University Hospital, Kuopio, Finland. Dr
Swan’s current affiliation is Department of
Obstetrics and Gynecology, University of
Rochester School of Medicine and
Dentistry, Rochester, New York.
Key Words
puberty, secular trends, puberty,
menarche, Tanner stage, precocious
puberty, growth and development
Abbreviations
Bn— breast development Tanner stage n
PHn—pubic hair development Tanner stage n
PROS—Pediatric Research in Office Settings
NHANES—National Health and Nutrition
Examination Survey
HPA— hypothalamic-pituitary-adrenal
Gn— genital development Tanner stage n
HPG— hypothalamic-pituitary-gonadal
SES—socioeconomic status
NHES—National Health Examination Survey
HHANES—Hispanic Health and Nutrition
Examination Survey
Accepted for publication Sep 5, 2007
Address correspondence to Susan Y. Euling,
PhD, US Environmental Protection Agency,
National Center for Environmental
Assessment, Office of Research and
Development, Mail Code 8623P, US
Environmental Protection Agency, 1200
Pennsylvania Ave, NW, Washington, DC 20460.
E-mail: euling.susan@epa.gov
PEDIATRICS (ISSN Numbers: Print, 0031-4005;
Online, 1098-4275). Copyright © 2008 by the
American Academy of Pediatrics
S172 EULING et al
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
United States from the mid-1900s to the present. This
article presents the panel conclusions on the question,
“Are data sufficient to suggest or establish a secular trend
in the timing of puberty onset and/or progression (as
measured by puberty markers) in boys and girls from US
data over the time period 1940 to 1994.” This time frame
was selected because (1) it is at the center of the debate
about a secular trend in puberty timing and (2) many
studies that were performed during this interval assessed
breast and pubic hair staging as well as menarche.
PUBERTY-TIMING MARKERS
Human puberty is a complex temporal sequence of bio-
logical events leading to the maturation of secondary sex
characteristics, accelerated linear growth, and attain-
ment of reproductive capacity. During normal puberty,
the sequence of events requires gonadal and adrenal
activation. Breast development, pubic hair development,
and age at menarche have been most frequently assessed
in female puberty-timing studies in humans. The male
puberty markers included in this discussion are genital
and pubic hair development. Breast, genital, and pubic
hair development have been assessed using a progressive
scale, usually the Tanner staging. Tanner staging is based
on a scale of 5 progressive stages described and depicted
in photographs by Marshall and Tanner in which stage 1
is prepubertal, stage 2 is onset, and stage 5 is adult.
9–12
Earlier studies that assessed breast and pubic hair staging
used similar definitions.
13,14
They are based on visual
observation alone, either from photographs or in person
by a trained professional or a child (ie, self-assessment).
The first sign of puberty in girls is either the initiation
of breast development, designated as Tanner stage 2 (B2;
also called breast budding), or pubic hair development
(PH2).
10,15
Their relative order of appearance is subject to
interindividual and interracial variability.
5,16
Black girls
often begin pubic hair development before or close to
the time of breast development onset, whereas white
girls typically begin breast development before pubic
hair development.
5
B2 can be difficult to assess using
visual inspection alone, either in person or from photo-
graphs. Discerning breast from fat tissue is a key con-
cern, especially in overweight girls with excessive sub-
cutaneous fat in the chest. Palpation under the areolar
area by a trained professional can distinguish breast
(glandular) from fat tissue, improving data quality, and
is used in clinical diagnosis. The Lee
17
study and a subset
of the Pediatric Research in Office Settings (PROS) study
were the only U.S. studies to use this technique.
18,19
Comparing data from the US National Health and Nu-
trition Examination Surveys (NHANES), NHANES III
20
and NHANES 1999 –2000, Jolliffe
21
found an increased
prevalence and extent of overweight US children aged 2
to 19 years between 1971 and 2000. With an increased
percentage of overweight children, the use of palpation
is particularly useful in identifying B2. Visual inspection
remains the only method to stage breast development
progression (Tanner stages 3–5).
15
PH1 to PH5 in boys and girls have been evaluated by
visual inspection by a clinician or self-assessment. Male
pubic hair development occurs in response to dihy-
drotestosterone, produced by the testes, binding to an-
drogen receptors and activating androgen-dependent
gene transcription in genital skin regions. Female pubic
hair development is thought to occur in response to
androgens produced from the activation of the hypotha-
lamic-pituitary-adrenal (HPA) axis. Racial differences in
patterns and progressive stages of pubic hair develop-
ment increase the variability in pubic hair staging mea-
sures that use a scale (Tanner staging) developed for
white individuals.
Age at menarche can be determined by asking a girl
or her parent her “current status” (also called “status
quo”), whether she has had her first menses by the time
of assessment and her birth date, and/or by asking post-
menarcheal females (or their mothers) to “recall” their
age at first menses. Recall may be less valid and de-
creases in accuracy with greater time elapsed between
menarche and asking for the date.
22,23
Height and weight
were measured in a number of studies, some longitudi-
nal. In girls, the onset of the pubertal growth spurt
usually precedes or is concomitant with the onset of
breast development, at least in white girls.
10
In boys, the first external signs of puberty are an
increase in testicular volume above 3 mL
24,25
and Tan-
ner’s genital growth stage 2 (G2).
11
The increase in tes-
ticular volume is attributed primarily to the growth of
testicular seminiferous tubules stimulated by follicle-
stimulating hormone. A volume of ⬎3 mL is considered
the most reliable and valid male pubertal onset marker
developed to date.
26–28
Testicular volume can be mea-
sured by an examiner, using an orchidometer (first de-
veloped by Prader
29
) or a ruler or by ultrasonography.
The most common method is using an orchidometer,
either by comparing the testis size by palpation to the
orchidometer beads or by using a punched-out (ie, card
with holes) orchidometer. Measurements of testicular
volume using either orchidometer method are well cor-
related to those using ultrasound.
30,31
Although testicular
volume is often measured in the clinical setting, it has
not been used in human studies of puberty timing in the
US population. Instead, most studies used visual assess-
ment of genital development using Tanner staging. Stage
2 corresponds to the first visible signs of testicular and
scrotal growth and changes in the scrotal skin.
9,11
Addi-
tional male pubertal markers include axillary hair devel-
opment, age at first ejaculation, acne, and voice break,
all relatively late male pubertal events. The timing of
these markers is subject to higher interindividual vari-
ability, and the relationship to the timing of hypotha-
lamic-pituitary-gonadal (HPG) axis and HPA axis activa-
tion is not as well understood. Age at spermarche can be
determined by the presence of spermatozoa in the first
morning urine void (spermaturia) but is not considered
a reliable marker because of a high number of false-
negative results.
32,33
In boys, the onset of the pubertal
growth spurt is a relatively late pubertal event, usually
corresponding to the time of late G3 or early G4.
11
Hormone and receptor measurements would be very
useful for determining the timing of the activation of the
HPG and HPA axes underlying the physical manifesta-
tions of puberty. Endogenous hormone levels and other
PEDIATRICS Volume 121, Supplement 3, February 2008 S173
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
biochemical assays to evaluate the physiologic pubertal
status are costly for puberty studies with a large partic-
ipant number, and, although the technology is improv-
ing, they have validity issues such as detection limits and
interindividual hormone level variability
34
(reviewed by
Rockett et al
35
). This discussion focuses on the physically
manifested puberty markers because they were assessed
in the studies from 1940 to 1994.
ISSUES IN SECULAR TREND ANALYSIS FOR PUBERTAL TIMING
STUDIES
A secular trend is a change in the distribution of an
outcome in a population during a specified time frame,
usually “a long period of time, generally years or de-
cades.”
36
Secular trend analysis of health measures is
frequently used to track public health concerns or initi-
atives. If an adverse trend is detected, then studies are
often performed to understand better the possible causes
or determinants, and interventions may result. Secular
trend analysis is, by its nature, historic, and projections
to the future are inherently uncertain. Confidence in
predicting future trends is greater when the past trend is
stable or represents a steady change, but even smooth
trends can change. Understanding the secular trend and
the trend of its determinants should improve prediction
of future trends.
The ability to compare adequately 2 studies, per-
formed at different points in time, is affected by the
similarities in the study designs. A number of study
design choices, including general study design (eg, longi-
tudinal), population characteristics (eg, race), age(s) of
measurement, puberty measures assessed, and method(s)
of puberty measurement affect the quality of the data
comparisons. Study differences among selected studies for
the age of menarche are illustrated in Table 1. Some re-
ports are based on the same study (eg, NHANES III,
Bogalusa Heart Study) but with a different subset of
participants, age definition, or analytic approach. Al-
though no 2 studies performed at different times can
replicate definitions and analyses precisely, some studies
are judged to be similar enough to allow for comparison.
Comparability is somewhat limited by the methodologic
choices of the historic studies. For example, newer stud-
ies may have used more reliable methods (eg, palpation
for breast development onset) that are less similar to the
methods used in an older study (eg, photographs for
breast development onset).
Most of the published studies on puberty timing were
TABLE 1 Examples of Study Design Choices That Can Affect Data Comparability Among Selected US Studies of Menarche
Study Study Design Type Questionnaire Data Used
in Analysis (Respondent)
Calculation/Analysis Age Increment
Reported
Age Range of
Population, y
a
Fels Institute Study;
Reynolds and Wines
13
(1948)
Longitudinal “Menstrual record” (NR) Mean Midpoint between visits
(ie, half-year)
8–18
Guidance Study; Nicolson
and Hanley
14
(1953)
Longitudinal NR (NR) Mean Midpoint between visits 8–17
NHES I, II, and III;
MacMahon
63
(1973),
Lee et al173 (1980)
CS; nationally
representative,
longitudinal
I: recall (self); II and III:
current status and recall
(parent); “menstrual
history” (self)
I: mean; II and III: median/
fitted logistic curve;
mean
I: year; II and III:
midpoint of the year
(added 0.5 years);
half-year
I: 18–79; II and III:
11–14; 8.6–
17.8
Bogalusa Heart Study;
Foster et al
61
(1977)
CS (1 cohort)
b
Recall (self) Median Year 5–14
Pediatric Research in
Office Settings (PROS)
Network; Herman-
Giddens et al
5
(1997)
CS; well-child or problem
visit at volunteer
physicians office
Current status (self) Means/probit analysis Year 3–12
Bogalusa Heart Study;
Wattigney et al
93
(1999)
Mixed longitudinal/CS
(2 nonoverlapping CS
cohorts)
b
Limited to postmenarcheal
girls; recall (self)
Mean and analysis of
variance to assess
differences between 2
cohorts
Early (8–11 years);
intermediate (12–13
years); late (14–17
years)
8–17
Bogalusa Heart Study;
Freedman et al
4
(2002)
Mixed longitudinal/CS (7
overlapping CS cohorts)
b
Current status (self) Median/logistic regression Midpoint of year Third grade (⬃8)
to 17
NHANES III; Wu et al
47
(2002)
CS; nationally representative Current status (self); recall
(self)
Current status data: mean/
probit analysis; recall
data: mean/failure time
model
Current status: year;
postmenarcheal girls:
midpoint of year
(added 0.5 years)
10–16
NHANES III; Chumlea
et al
46
(2003)
CS; nationally representative Current status (parent for
8- to 9-year-olds; self for
10- to 19-year-olds)
Median/probit analysis Midpoint of a 3-mo
age group
8–19
NR indicates not reported; CS, cross-sectional.
a
Year ranges described in whole numbers indicate the beginning of the first age to the end of the second age (eg, 3–12 is 3.0 years to 12.99 years of age).
b
The Bogalusa Heart Study used an overlapping panel design. There were 7 CS examinations of children (or cohorts) in a specific age range, which were timed closely together so that some of the
participants were examined multiple times (ie, longitudinally). Freedman et al
4
analyzed data from all 7 CS cohorts; Wattigney et al
93
analyzed data from 2 nonoverlapping CS cohorts; Foster et al
61
analyzed data from 1 CS cohort.
S174 EULING et al
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
either cross-sectional or longitudinal. In a cross-sectional
study, all measurements are made at a single point in
time; pubertal development is assessed only once per
individual. For Tanner stage (breast, pubic hair, and
genital development), current developmental stage and
current age are recorded. For menarche, both current
status (menstruating: yes or no) and age are obtained,
and recalled age at menarche can be asked of those who
report menstruating. The age at menarche may have
greater measurement error than for Tanner staging for
reports that provided menarche in whole year of age.
Conversely, determining menarche by the current status
method is more reliable compared with Tanner staging
measurements, which are somewhat subjective. In a
longitudinal study, each boy or girl is examined at sev-
eral points in time, with age and status recorded at each
examination. Both cross-sectional and longitudinal ap-
proaches give information about the distribution of age
at onset. In a cross-sectional study, onset for any indi-
vidual is known to have occurred either before or after
the current age at the time of assessment. In a longitu-
dinal study, onset is known to have occurred before the
first age of assessment, between the ages at 2 different
assessments, or after the age at the last assessment.
10
From cross-sectional study data, mean age of onset of
successive stages can be estimated; however, this ap-
proach, based on single observations of an individual,
does not provide the highest quality information on
individual variability in the progression through the
stages. Longitudinal studies can observe progression
through Tanner stages within the same individuals. A
longitudinal study provides information on the time
each individual spends in intermediate stages, with the
quality of the information improving as the number of
examinations increases and the time between examina-
tions decreases.
Timing of breast, genital, and pubic hair stages in a
population may be described by several characteristics:
(1) the average age of onset or attainment of the mile-
stone; (2) the proportion of children who have attained
the milestone by a given age; and (3) the average age of
all children in a specific stage. Estimates of the first 2
characteristics may be obtained using censored data
(from “current status”) techniques. The common ap-
proach in puberty studies is to smooth the distribution
by assuming that it is either normal or logistic (using
probit or logistic regression, respectively). Both distribu-
tions are determined by 2 parameters, a mean and a
variance; estimates that are based on the 2 assumptions
will be similar, because the shapes of the normal and
logistic distributions differ little.
37
For longitudinal stud-
ies, techniques similar to these standard methods can be
used to study progression, producing estimates of the
distribution of duration of time within a stage.
The age range of the children studied is another im-
portant factor. To get the best estimates for age at attain-
ment, the ages at which children are examined should
span the range during which most children attain the
milestone. If few of the oldest children included have
attained the milestone at the end of the study (or, alter-
natively, most of the youngest included have attained
it), then estimates of mean age of attainment will be less
precise. A related problem is the possibility of biased
estimates of mean age in a stage in a study with a
restricted age range. Children should not be included or
excluded in analyses on the basis of attainment of the
milestone. For example, in a longitudinal study, elimi-
nation of children who have not attained the milestone
will bias the mean age at attainment downward. Be-
cause children change rapidly in the pubertal period,
information improves with more precise age measure-
ments. Creating broad groupings by current age (eg,
quarter year, whole years) is less desirable than use of
exact age. In a longitudinal study, reporting the age of
first observation of a stage as the “age of attainment”
overestimates the true age of attainment. Finally, round-
ing of recalled age at attainment, which can occur if age
is reported in whole year only or date is reported as a
calendar year only, most likely decreases the accuracy of
the estimate. It is important to cope with these problems
by applying statistical tools that account for the structure
of the data (eg, left truncation, right censoring, interval
censoring). Such tools are available (eg, life time analysis
or analysis of time to event), and models that incorpo-
rate continuous traits that vary over time may be ap-
plied. Rather than simply estimating possibly biased ob-
served mean ages, such analyses may provide age
distributions and allow for estimation of median or quar-
tiles of ages.
Differences in methods to measure puberty markers
can affect findings and, thus, comparability of findings
between studies. For breast stage, studies may use pal-
pation and/or observation to determine breast stage in
girls. In addition, the rater (eg, self, trained professional)
and his or her training as well as the setting where the
measurement takes place can affect comparability. In-
trarater and interrater variability is important informa-
tion for comparing across studies. Self-assessment is
based on showing drawings or photographs to a child
and having him or her mark the stage most like himself
or herself.
38,39
The quality of self-assessment compared
with that of a trained examiner has been examined in a
number of studies in varying populations, using a variety
of instruments in different settings (reviewed by Rockett
et al
35
). Self-assessment in boys and girls for Tanner
staging tended to overestimate stage during the early
puberty stages and underestimate in the later stages
compared with the staging by a physician.
39
In a study
of visual self-assessment by children aged 6 to 12
years, assessment of breast development by girls and
pubic hair development by boys was often different
from that of the trained examiner, and obese girls
were more likely to overestimate breast stage than
nonobese girls.
40
POPULATION VARIABLES THAT AFFECT PUBERTY TIMING
Studies of higher quality examine a defined group of
people with known characteristics (eg, gender, age, so-
cioeconomic status [SES], geographic location) during
the time period of interest. Information about genetic
and environmental factors that affect puberty timing
improves the quality of a study of puberty timing.
PEDIATRICS Volume 121, Supplement 3, February 2008 S175
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
Knowledge of genetic factors that could affect puberty
timing would be useful. Polymorphisms in genes that reg-
ulate puberty timing have been identified, but these ge-
netic markers have not been available until recently.
41–44
Racial and ethnic differences reflect a combination of
genetic, social, and environmental factors that are dif-
ferent. Race/ethnicity, defined as racial and/or ethnic
subpopulations (eg, Hispanic black, non-Hispanic black),
was not often recorded in the historic studies but has
been recorded in more recent studies. Analyses of the US
PROS and NHANES III data revealed racial differences in
the timing of female puberty; black girls were younger
than white girls at the same stage of breast development,
pubic hair development,
5,45
and menarche.
5,46
Ages for
Mexican American girls were between those for the 2
other groups.
45–47
For boys, the onset of genital and pubic
hair development occurs earlier in black than white
boys.
6,7
Information on environmental factors including pre-
vious nutrition, body fat or size, and race are important
in studies of puberty timing. Early severe protein-energy
malnutrition presenting as marasmus (severe wasting
and growth retardation as a result of insufficient intake)
but not kwashiorkor (edematous protein-deficiency
malnutrition) delays timing of puberty, at least in
girls,
48–50
and chronic malnutrition during childhood
through to puberty retards pubertal development in
boys and girls.
51
Among chronically undernourished
children, age at menarche was delayed by ⬃2 years
among the girls and sexual maturity measures were
delayed by ⬃3 years in boys.
Higher subcutaneous fat levels and BMI at prepubertal
ages (5–9 years) are associated with increased likelihood of
relatively early (⬍11 years) menarche.
4
A dose-response
association between a higher BMI and increased relative
hazards for earlier menarche in girls (8 –13 years at base-
line) has been observed.
52
The relationship of body fat and
puberty timing is reviewed by Kaplowitz.
53
Some studies found that children who had a low birth
weight (defined as ⬍2500 g) or were small for gesta-
tional age (typically defined as birth weight ⬍10th per-
centile for gestational age and gender) are more likely to
have advanced timing of puberty (girls and boys
54
; girls
only
55
); however, other results are mixed. For exam-
ple, 1 study found an association between small size at
birth and early puberty in girls but not boys,
56
and
another found no differences between pubertal timing
in children who were born at very low birth weight
(⬍1251 g) and control subjects of both genders.
57
An
additional issue is that definitions for small birth size
and puberty timing outcomes have not been consis-
tent across studies. One hypothesis is that children
who are born small and subsequently go through very
rapid catchup growth may start puberty earlier.
58–60
EXAMINATION OF US PUBERTY-TIMING DATA (1940 –1994)
FOR SECULAR TRENDS
Girls’ Data
The earliest published US data on female puberty timing
that included breast and pubic hair development as well
as menarche were collected in the 1930s and 1940s
(Tables 2 and 3). These 2 small, longitudinal studies
included mainly high-SES, white populations.
13,14
Al-
though the 2 studies used different Tanner stage defini-
tions, they reported similar mean ages of breast devel-
opment onset and menarche, but pubic hair
development onset ages were more divergent. Studying
a small, largely white population in the early 1970s,
Lee
17
reported older ages of breast development onset
and menarche than were observed in either the 2 earlier
studies or the Bogalusa Heart Study, a study of a similar
time period.
61
Analyzing the Bogalusa Heart Study se-
quential cohort data, Freedman et al
4
found that the
mean age of menarche in white girls declined 0.2 years
and in black girls declined 0.8 years from 1973–1974 to
1992–1994.
The first pubertal timing data on a representative US
sample were from the National Health Examination Sur-
vey (NHES) cycles I to III conducted by the Centers for
Disease Control (later the Centers for Disease Control
and Prevention) from 1963 to 1970 (Tables 2 and 3).
62,63
For NHES cycles II and III, the mean age at menarche
was calculated.
63
Tanner stage data were collected only
in NHES III, and Harlan et al
62
reported the percentage of
children of specific ages who were in each stage. Com-
parisons with other studies were difficult because mean
age at attainment for Tanner stages was not calculated.
From 1982 through 1984, pubertal data on menarche,
breast development, and pubic hair development were
collected in the Hispanic Health and Nutrition Examina-
tion Survey (HHANES). Villarreal et al
64
analyzed the
Mexican American data only from HHANES and re-
ported mean ages of B2 and PH2. Because neither NHES
III nor HHANES collected pubic hair and breast devel-
opment stage data on children who were younger than
12 and 10 years, respectively, the results are less useful
for examining the age of onset (Tanner stage 2) than
studies that assessed earlier ages (eg, 37% of 10-year-old
Mexican American girls in HHANES had already begun
breast development
64
; in NHES III, few participants who
were 12–17 years of age were in stages 1 or 2). Advan-
tages of the NHES II and, HHANES, and NHANES III data
for assessing secular trends are the similar sampling
strategies, which include oversampling of black and
other racial/ethnic groups, and statistical weighting de-
signed to provide a representative US sample. A possible
disadvantage is the inclusion of multiple participants
from the same household (ie, nonindependent partici-
pants who may be genetically related and have similar
environmental exposures).
The PROS study was a cross-sectional study on the
timing of US female pubic hair and breast development
and menarche on a large, racially diverse population of
3- to 12-year-old girls conducted from 1992 to 1993 by
Herman-Giddens et al.
5
It is the largest study (n ⫽
17 077) using Tanner staging and menarche, with a
much larger sample size than NHES II and III (combined,
n ⫽ 3000) or NHANES III data sets (n ⫽ 2000 –2500;
Table 2). The youngest age assessed was 3 years, which
is optimal for estimating the onset of breast and pubic
hair development; however, the age truncation at 12
S176 EULING et al
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
TABLE 2 Description of Puberty Timing Studies for US Girls
Study Years Data
Collected
Geographic Region; SES Race/Ethnicity of
Participants Included in
Analysis
Study Type; Interval
Assessed
No. of
Participants
Markers Assessed
(Method)
Age Range, y
a
Fels Study; Reynolds and
Wines
13
(1948)
1940s Yellow Springs, OH;
mid-upper SES
NR but known to be 100%
white
L; 6 mo (birthday and
half birthday)
49 Menarche, breast and
pubic hair staging
b
(photos)
8–18
Guidance Study; Nicolson
and Hanley
14
(1953)
1930s–1940s Berkeley, CA: high
educational level
NR but known to be 100%
white
L; 6 mo 95 Menarche, breast and
pubic hair staging
c
(photos)
8–17
NHES I, II, and III;
MacMahon
63
(1973)
1960–1970 Nationally representative;
mixed SES
5635 white; 1043 black;
32 other
d
CS I: 3581; II and III
e
:
2242
Menarche I: 18–79; II and III
e
:
11–14
NHES cycle III
Harlan et al
62
(1980)
f
1966–1970 Nationally representative;
mixed SES
2688 white; 500 black CS 7514 Breast and pubic hair
Tanner staging (visual
examination)
12–17
Lee
17
(1980) 1969–1974 Baltimore, MD; upper
middle SES
94% white; 6% black
and Asian
L; 6 mo 18 Menarche and breast
(palpation) and pubic
hair Tanner staging
(visual examination)
8.6–17.8
Bogalusa Heart Study
Foster et al
61
(1977)
g
1973–1974 Bogalusa, LA;
low-average SES
1059 white
h
; 621 black CS 1680 Menarche, breast and
pubic hair Tanner
staging (visual
examination)
5–14
Freedman et al
4
(2002) 1973–1994 Bogalusa, LA;
low-average SES
61% white
h
; 39% black ML/CS 9158 Menarche 7–16.9
Wattigney et al
93
(1999)
i
1978–1979;
1992–1994
Bogalusa, LA;
low-average SES
78–79: 332 white
h
;
238 black
92–94: 348 white
h
;
305 black
ML/CS 78–79: 570
92–94: 653
Menarche 8–17
HHANES; Villarreal et al
64
(1989)
1982–1984 Mexican Americans from
the Southwest
100% Mexican American
j
CS 699 Breast and pubic hair
Tanner staging (visual
examination)
10–17
PROS; Herman-Giddens et
al
5
(1997)
1992–1993 Volunteer physician
practices in United
States (34 states and
Puerto Rico); NR
90.4% white; 9.6% black CS 17 077 Menarche, breast and
pubic hair Tanner
staging (visual
examination for all;
palpation for subset
k
)
3–12
NHANES III
Wu et al
47
(2002) 1988–1994 Nationally representative;
mixed SES
466 white; 589 black; 568
Mexican American
(Tanner stages); 330
white; 419 black; 419
Mexican American
(menarche)
l
CS 1623 (Tanner stages);
1168 (menarche)
Menarche, pubic and
breast development
Tanner staging (visual
examination)
8–16 (Tanner stages);
10–16 (menarche)
PEDIATRICS Volume 121, Supplement 3, February 2008 S177
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
TABLE 2 Continued
Study Years Data
Collected
Geographic Region; SES Race/Ethnicity of
Participants Included
in Analysis
Study Type; Interval
Assessed
No. of Participants Markers Assessed
(Method)
Age Range, y
a
Sun et al
45
(2002) 1988–1994 Nationally representative;
mixed SES
594 white; 788 black; 763
Mexican American
l
CS 2145 Breast and pubic hair
development Tanner
staging (visual
examination)
8–18
Chumlea et al
46
(2003) 1988–1994 Nationally representative;
mixed SES
710 white; 917 black; 883
Mexican American
l
CS 2510 Menarche 8–19
NHES II/III; NHANES III;
Anderson et al
3
(2003)
1963–1965 (NHES II);
1966–1970 (NHES
III); 1988–1994
(NHANES)
Nationally representative;
mixed SES
White, black, other; NHES
II and III
e
: 77.8% white;
NHANES III: 86.0%
white
l
CS 3272 NHES II/III
m
and
1326 NHANES III
Menarche 10–15
NHES III; HHANES; NHANES
III; Sun et al
65
(2005)
1966–1970 (NHES
III); 1982–1984
(HHANES); 1988–
1994 (NHANES III)
Nationally representative;
mixed SES (NHES and
NHANES); Hispanic
(HHANES)
NHES III: white (83.9%);
black; HHANES: 100%
Mexican American;
NHANES III: white
(23.9%); black; Mexican
American
l
CS 3130 NHES III; 712
HHANES; 1218
NHANES III
Breast and pubic hair
development Tanner
staging (visual
examination)
12–17 (NHES III–
NHANES III
comparison);
10–17 (HHANES–
NHANES III
comparison)
L indicates longitudinal; CS, cross-sectional; ML/CS, mixed longitudinal and cross-sectional components or overlapping panel design; NR, not reported by authors.
a
Years described in whole numbers indicate the range from the beginning of the first age to the end of the second age (eg, 3–12 is 3.0 years to 12.99 years of age).
b
Reynolds and Wines
13
used a 5-stage rating system for breast and pubic hair development that had similar definitions to Tanner staging.
c
Nicolson and Hanley
14
used a 4-stage system for breast development. They defined stage 2 as onset (similar to Tanner stage 2), stage 3 as similar to Tanner stage 3, and stage 4 as similar to Tanner stage 5. A 5-stage system similar to Tanner staging was used for pubic
hair development.
d
Hispanic white was included in white; Hispanic black was included in black.
e
Data from NHES cycles II and III were combined, and the age range was limited to 11 to 14 years.
f
Data from Harlan et al
62
are not included because neither means nor medians were reported.
g
Foster et al
61
analyzed 1 CS data set.
h
White was defined as nonblack.
i
The analysis of Wattigney et al
93
grouped ages as 8 to 11 (“early”), 12 to 13 (“intermediate”), or 14 to 17 (“late”) years of age. Data from this study are not presented in Table 3 because of lack of comparisons with other studies.
j
Mexican Americans from the Southwest United States from HHANES were included.
k
Data for breast staging using palpation were reported by Kaplowitz and Oberfield.
18
l
Non-Hispanic white and non-Hispanic black.
m
Anderson et al
3
combined data from NHES cycles II and III and limited the age range to 10 to 15 years.
S178 EULING et al
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
TABLE 3 Ages of Pubertal Events From Studies of US Girls
Study Mean/Median
a
Age of Attainment of a Stage, y (SD, SE, or FL/CL
b
)
B2 B3 B4 B5 PH2 PH3 PH4 PH5 Menarche
Fels Study; Reynolds and
Wines
13
(1948)
10.8 (1.1) 11.4 (NR) 12.2 (NR) 13.7 (NR) 11.0 (1.1) 11.9 (NR) 12.5 (NR) 13.9 (NR) 12.9 (1.4) white
Guidance Study;
Nicolson and
Hanley
14
(1953)
10.6 (1.2) 11.2 (1.1) ND
c
13.9 (0.9) 11.6 (0.9) 12.5 (1.0) 13.2 (0.9) NR
d
12.8 (1.1)
NHES I, II, and III;
MacMahon
63
(1973)
ND ND ND ND ND ND ND ND I
e
;: 12.73 (1.386) white;
12.85 (1.534) black;
II and III
f
: 12.80
(0.41) white; 12.52
(0.111) black
NHES II and III; Anderson
et al
3
(2003)
ND ND ND ND ND ND ND ND 12.75 (12.69–12.82)
all; 12.80 (12.73–
12.87) white;
12.48 (12.28–12.67)
black
NHES III
Sun et al
65
(2005) ND 11.52 (9.98–12.15)
white; NS for
black
f
12.78 (12.49–13.02)
white; 11.46
(10.09–12.21)
black
15.47 (15.26–15.69)
white; 14.29
(13.80–14.71)
black
ND NS for white
g
;NSfor
black
g
12.74 (12.54–12.92)
white; 11.85
(10.97–12.39)
black
14.02 (13.58–14.39)
white; 14.02
(13.58–14.39)
black
ND
Lee
17
(1980) 11.2 (1.6) 12.4 (1.2) 13.1 (0.7) 14.5 (1.6) 11.9 (1.5) 12.7 (0.5) 13.4 (1.2) 14.6 (1.1) 13.3 (1.3)All
Bogalusa Heart Study;
Foster et al
61
(1977)
10.37 (0.105) white;
10.22 (0.125) black
11.60 (0.09) white;
10.91 (0.12) black
NR
g
NR
g
10.86 (0.095) white;
10.13 (0.125) black
11.66 (0.09) white;
10.91 (0.12) black
NR
h
NR
h
1973–1974: 12.69
(0.085) white; 12.83
(0.11) black
HHANES; Villarreal et al
64
(1989)
i
10.95 (10.93–11.31)
Mexican American
12.16 (12.00–12.44)
Mexican American
13.86 (13.57–14.13)
Mexican American
15.10 (14.78–15.36)
Mexican American
11.32 (11.19–11.61)
Mexican American
12.44 (12.25–12.65)
Mexican American
14.07 (13.74–14.26)
Mexican American
15.50 (15.11–15.65)
Mexican American
ND
HHANES; Sun et al
65
(2005)
ND 11.98 (11.52–12.4)
Mexican American
13.72 (13.46–13.98)
Mexican American
14.97 (14.70–15.25)
Mexican American
ND 12.24 (11.67–12.76)
Mexican American
13.92 (13.68–14.16)
Mexican American
15.27 (15.01–15.54)
Mexican American
ND
Bogalusa Heart Study;
Freedman et al
4
(2002)
ND ND ND ND ND ND ND ND 1973–1974
j
: 12.7
(12.6–12.9) white;
12.9 (12.7–13.1)
black; 1992–1994
j
:
12.5 (12.4–12.8)
white; 12.1
(11.9–12.3) black
PROS; Herman-Giddens
et al
5
(1997)
9.96 (1.82) white;
8.87 (1.93) black
11.30 (1.42) white;
10.19 (1.42) black
ND ND 10.51 (1.67) white;
8.78 (2.00) black
11.53 (1.21) white;
10.35 (1.63) black
ND ND 12.88 (1.20) white;
12.16 (1.21) black
NHANES III; Wu et al
47
(2002)
10.3 (10.0–10.5)
white; 9.5 (9.3–9.8)
black; 9.7 (9.4–9.9)
Mexican American
ND ND ND 10.6 (10.4– 10.9)
white; 9.5 (9.2–9.8)
black; 10.3
(10.1–10.6)
Mexican American
ND ND ND 12.6 (12.4–12.8) white;
12.2 (12.0–12.4)
black; 12.2
(12.0–12.5)
Mexican American
k
NHANES III; Sun et al
45
(2002)
l
,
m
10.38 (10.11–10.65)
white; 9.48 (9.14–
9.76) black; 9.80
(0–11.78
n
)
Mexican American
11.75 (11.49–12.02)
white; 10.79
(10.50–11.08)
black; 11.43 (8.64–
14.50) Mexican
American
13.29 (12.97–13.61)
white; 12.24
(11.87–12.61)
black; 13.07
(12.79–13.36)
Mexican American
15.47 (15.04–15.94)
white; 13.92
(13.57–14.29)
black; 14.70
(14.37–15.04)
Mexican American
10.57 (10.29–10.85)
white; 9.43 (9.05–
9.74) black; NS for
Mexican
American
o
11.80 (11.54–12.07)
white; 10.57
(10.30–10.83)
black; 11.70
(11.14–12.27)
Mexican American
13.00 (12.71–13.30)
white; 11.90
(11.38–12.42)
black; 13.19
(12.88–13.52)
Mexican American
16.33 (15.86–16.88)
white; 14.70
(14.32–15.11)
black; 16.30
(15.90–16.76)
Mexican American
ND
PEDIATRICS Volume 121, Supplement 3, February 2008 S179
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
TABLE 3 Continued
Study Mean/Median
a
Age of Attainment of a Stage, y (SD, SE, or FL/CL
b
)
B2 B3 B4 B5 PH2 PH3 PH4 PH5 Menarche
NHANES III; Chumlea
et al
46
(2003)
ND ND ND ND ND ND ND ND 12.43 (12.33–12.53)
all; 12.55 (12.31–
12.79) white; 12.06
(11.84–12.28) black;
12.25 (12.04–12.46)
Mexican American
NHANES III; Anderson
et al
3
(2003)
ND ND ND ND ND ND ND ND 12.54 (12.44–12.64)
all; 12.60 (12.48–
12.71) white; 12.14
(11.87–12.39) black
NHANES III; Sun et al
65
(2005)
ND 11.82 (10.49–12.32)
white; NS for
black
g
;NSfor
Mexican
American
g
12.90 (12.18–13.39)
white; 11.46
(9.73–12.31) black;
13.04
(12.70–13.37)
Mexican American
15.16 (14.61–15.71)
white; 13.54
(12.77–14.10)
black; 14.74
(14.42–15.09)
Mexican American
ND NS for white
g
; 11.48
(7.54–12.24) black;
11.65
(11.01–12.15)
Mexican American
12.73 (12.08–13.16)
white; NS for
black
g
; 13.15
(12.77–13.53)
Mexican American
16.13 (15.55–16.92)
white; 14.29
(13.62–14.84)
black; 16.24
(15.83–16.73)
Mexican American
ND
ND, not determined; NS, not shown in this table; CL, 95% confidence limits; FL, fiducial limits.
a
Means and medians are considered comparable values because the ages for various pubertal markers are typically normally distributed.
b
See references for details.
c
Nicolson and Hanley
14
used a 4-stage breast development scale in which B3 was equivalent to Tanner B3 and B4 was equivalent to Tanner B5. Thus, Tanner B4 was not determined.
d
Not reported because PH5 was considered by the authors to be unreliable as a result of the interrater agreement score.
e
Mean age data were restricted to women aged 18 to 34 years in cycle I; women were born 1926 to 1944 and thus were peripubertal ages (10 –15 years) ⬃1936 –1956.
f
Data from cycles II and III were combined, and the analysis was limited to ages 11 to 14 years.
g
Not shown in this table because the FLs were labeled “unstable” by the authors.
h
Foster et al
61
only reported median ages for transition into B2, B3, PH2, and PH3 because of the “limited age range,” 5 to 14 years, of their population.
i
Villarreal et al
64
calculated percentages of girls in a given stage by age, means for being in a stage, means (from probit analysis), and medians (from Spearman-Karber analysis) for entry into a stage. Means for entry into a stage are presented here for comparison with
the other studies.
j
Freedman et al
4
reported data from 7 cohorts assessed between 1973 and 1994. Because of space limitations, only median ages are shown for the first (1973–1974) and last (1992–1994) cohorts studied.
k
Wu et al
47
calculated mean menarcheal age separately for current status and recalled age. Mean ages from current status data are shown.
l
Sun et al
45
calculated means for “age in a stage” and medians for “age at entry” into a stage. Medians for “age at entry” are reported here for comparison with other studies that looked at age of attainment.
m
FLs were computed at 98.3% CLs to adjust for multiple comparisons between racial/ethnic groups.
n
The left side FL of 0 is likely attributable to the small sample size for Mexican Americans.
o
Not shown in this table because the authors did not report the FL because of the small sample size for Mexican Americans.
S180 EULING et al
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
years excludes observation of puberty completion (Tan-
ner stage 5 and menarche) for many girls. For example,
64.8% of the white girls and 37.9% of the black girls had
not achieved menarche by the end of their 12th year (up
to 12.99 years of age). The PROS study group is a con-
venience sample and not representative of a larger pop-
ulation; however, all of the earlier study populations
(1930 –1970), including Marshall and Tanner (1969),
10
on which US clinical standards for puberty timing were
derived, were also convenience samples. The partici-
pants included girls who were examined during pediatric
visits for either a (1) well-child visit at offices that as-
sessed Tanner staging and menarcheal status routinely
in the well-child visit (95.74%) or (2) a problem visit
that required a complete examination (4.26%). The
study by Herman-Giddens et al
5
reported younger me-
dian ages than any preceding study (Table 3). For exam-
ple, the median B2 reported was ⬃0.6 to 1.2 years earlier
and the median PH2 was ⬃0.5 to 1.4 years earlier than
the values from Reynolds and Wines,
13
Nicolson and
Hanley,
14
or Lee.
17
Several articles presented analyses of puberty-timing
data of children in 3 racial/ethnic groups (non-Hispanic
white, non-Hispanic black, and Mexican American)
from NHANES III, conducted between 1988 and 1994
(Table 3).
45–47,65
Comparing ages of attainment of puber-
tal stages between PROS and NHANES III, conducted
during similar time intervals (1980 –1990s), the median
ages from NHANES III are slightly higher than values
from PROS (Table 3); however, the different study de-
signs complicate any comparison (Tables 1 and 2) and
may explain the differences in findings. An alternative
interpretation is that the NHANES III and PROS findings
corroborate one another because both reported an ear-
lier age of onset of breast and pubic hair development
compared with the ages from data collected in the 1930s
and 1940s. Additional comparisons of the NHANES III
with data from the 1960s and 1970s were conducted.
Comparing NHANES III (1988 –1994) and NHES II/III
data,
63
Chumlea et al
46
reported a decrease of 4 months
in the median age of menarche (combining racial/ethnic
groups) between 1963–1970 and 1988 –1994; however,
the authors considered the change too small to be mean-
ingful, but statistical results were not presented. In
another analysis that compared the NHANES III with
reanalyzed NHES II/III data, Anderson et al
3
calculated
a decrease of 2.5 months in the median age of men-
arche that the authors interpreted as a meaningful
change; nonoverlapping confidence intervals for the
median ages were for whites only, suggesting statisti-
cal significance. Furthermore, Herman-Giddens et al
18
argued that there has been a statistically significant
decline in the median age of menarche from 1963–
1970 to 1988 –1994 because confidence intervals do
not overlap for most calculations from NHANES III
and NHES II/III data.
To assess whether Tanner stage age attainment has
changed between the late 1960s and early 1990s, Sun et
al
65
performed a reanalysis of NHES III, HHANES, and
NHANES III data (reviewed at the workshop in unpub-
lished form). For making better comparisons among the
data sets, the analysis was limited to stages 3 to 5 (be-
cause NHES III included children 12 to 17, considered
too old to estimate stage 2 reliably) and Mexican Amer-
ican children from HHANES. The authors concluded that
there is no evidence of an earlier puberty (as measured
by median ages of Tanner stages 3–5) during the time
spanning the 3 surveys (1960s through 1990s) for either
non-Hispanic black or white girls but “some evidence”
for Mexican American girls between 1982 and 1994. An
advantage of this study is the consistent reanalysis of
data from 3 national surveys. Disadvantages of the anal-
ysis are the exclusion of information on onset (Tanner
stage 2), which many consider to be the key puberty-
timing issue of concern, and decreased study power as a
result of limiting the sample sizes for comparisons.
A subset of the female puberty-timing studies could
assess whether the duration of the puberty, defined as
the interval from onset (as measured by B2 or PH2,
whichever occurs first) to completion (as measured by
menarche). Because the earlier studies
13,14,17
(Table 3) all
calculated approximately a 2.1- to 2.2-year interval for
largely white populations, the duration for white girls
only could be compared. Foster et al
61
and Wu et al
47
observed an increase of ⬃2.3 years in the duration of
puberty for white girls; however, Herman-Giddens et al
5
observed a longer duration of ⬃3.1 year for white girls.
One interpretation that explains the reported increased
duration is an earlier puberty onset with a relatively
unchanged or less-changed age at menarche. When
comparing studies that allowed for racial comparisons,
black girls were found to have a longer duration than
white girls.
5,47,61
Boys’ Data
Far fewer data are available on the ages and patterns of
pubertal development in US boys than girls. This difference
may indicate a lower interest in studying boys’ puberty that
may reflect less cultural awareness of male pubertal devel-
opment compared with girls (eg, breast development,
menstruation). In addition, the greater number of pub-
lished studies assessing girls’ puberty timing may reflect the
concern about the relationship between early female pu-
berty and later disease, which has been less investigated for
boys
66,67
(reviewed by Golub et al
68
).
All of the US analyses of male puberty timing have
assessed Tanner stages for genital and pubic hair devel-
opment staging (Table 4), albeit with study design dif-
ferences. Testicular volume was measured in only 1 US
study, but Tanner genital development was not also
assessed.
25
The history of US boys’ studies is similar to
the girls’ because most of the US puberty studies col-
lected data on both boys and girls (eg, Fels, Guidance,
NHES III, HHANES, NHANES, Bogalusa). For example,
the earliest data on the age of puberty in US boys were
from small, mostly high-SES white populations.
14,17,69
The first boys’ data representative of the US population
were collected from 1963 to 1970 in NHES III.
70
The
earliest ages included in NHES III or HHANES data are
not optimal for estimating the mean ages of the onset
(eg, 14.5% of 10-year-old Mexican American boys in
HHANES already had begun genital development) com-
PEDIATRICS Volume 121, Supplement 3, February 2008 S181
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
TABLE 4 Description of Puberty Timing Studies for US Boys
Study Years Data
Collected
Geographic Region; SES Race/Ethnicity of
Participants Included in
Analysis
Study Type; Interval
Assessed
N Markers Assessed (Method) Age Range, y
a
Fels Study; Reynolds and
Wines
69
(1951)
b
1940s Yellow Springs, OH; mid-
upper SES
NR but known to be 100%
white
L; 6 mo (birthday and
half birthday)
59 Pubic hair and genital development
Tanner staging
b
(photos)
9–21
Guidance Study; Nicolson
and Hanley
14
(1953)
c
1930s–1940s Berkeley, CA: high
educational level
NR but known to be 100%
white
L;6mo 92
d
Sex stage
d
(photos) 8–19
NHES III; Harlan et al
70
(1979)
e
1966–1970 Nationally representative;
mixed SES
499 black and 3047 white CS 3546 Pubic hair and genital development
Tanner staging (visual
examination)
12–17
Lee
17
(1980) 1969–1974 Baltimore, MD; upper
middle SES
94% white; 6% black and
Asian
L 36 Pubic hair and genital development
Tanner staging (visual
examination)
9.2–17.5
Bogalusa Heart Study;
Foster et al
61
(1977)
f
1973–1974 low-average SES 1153 white; 676 black
g
CS
5
1829 Pubic hair and genital development
Tanner staging (visual
examination)
5–14
HHANES; Villarreal et al
64
(1989)
1982–84 Mexican Americans of
the Southwest
100% Mexican American
h
CS 704 Pubic hair and genital development
Tanner staging (visual
examination)
10–17
NHANES III
Sun et al
45
(2002) 1988–1994 Nationally representative;
mixed SES
537 white; 798 black; 783
Mexican American
i
CS 2118 Pubic hair and genital development
Tanner staging (visual
examination)
8–18
Biro et al
25
(1995) NR Greater Cincinnati, OH
area; NR
⬃50% white; ⬃50% black
(precise value NR)
L 515 Pubic hair and genital development
Tanner staging (visual
examination) and testicular
volume
j
10–15
NHANES III; Herman-Giddens
et al
6
(2001)
1988–1994 Nationally representative;
mixed SES
536 white; 797 black; 781
Mexican American
i
CS 2114 Pubic hair and genital development
Tanner staging (visual
examination)
8–18
NHES III, and NHANES III;
Karpati et al
7
(2002)
NHES II: 1963–1965;
NHES III: 1966–
1970; NHANES III:
1988–1994
Nationally representative;
mixed SES
NHES III: white, black, other
g
;
NHANES III: white, black,
Mexican American, other
i
CS 3010 (NHES II); 3514 (NHES III);
2481 (NHANES III)
Pubic hair and genital development
Tanner staging (visual
examination)
NHES II: 8–11; NHES III:
12–17; NHANES III:
8–18
NHES III; HHANES; NHANES
III; Sun et al
65
(2005)
1966–1970 (NHES
III); 1982–1984
(HHANES);1988–
1994 (NHANES III)
Nationally representative;
mixed SES (NHES and
NHANES); Hispanic
(HHANES)
NHES III: white (86.4%);
black
7
; HHANES: 100%
Mexican American;
NHANES III: white (20.8%),
black, Mexican American
i
CS 3520 NHES III; 717 HHANES;
1246 NHANES III
Breast and pubic hair development
Tanner staging (visual
examination)
12–17 (NHES III–NHANES
III comparison); 10–17
(HHANES–NHANES III
comparison)
a
Years, described in whole numbers, indicate the range from the beginning of the first age to the end of the second age (eg, 3–12 is 3.0 years to 12.99 years of age).
b
For pubic hair development, Reynolds and Wines
69
used a 5-stage rating scale that had similar definitions to Tanner staging. For genital development, they used 2 separate scales, a 5-stage genital development scale that had some differences from Tanner GD stages,
and a 7-stage penis growth rating scale. G2 was defined as “enlargement of scrotum, first reddening and texture change.” G3 was defined as “first ’sculpturing’ and enlargement of penis.” G4 was defined as “pronounced sculpturing and darkening.” G5 was defined as
“essentially adult; reddish brown color, loose penile skin, loss of sharp sculpturing.”
c
Nicolson and Hanley
14
used a 5-stage “sex stage” system of Greulich et al
94
that combined assessment of pubic hair and genital development.
d
The number depends on the stage or outcome assessed and varied between 80 and 92 (eg, n ⫽ 92 for sex stage II).
e
Harlan et al
70
data are not included because neither means nor medians were reported.
f
The Bogalusa Heart Study used an overlapping panel design of multiple cohorts. Foster et al
61
analyzed data from 1 CS cohort.
g
Hispanic white was included in white; Hispanic black was included in black.
h
Mexican Americans from the Southwest United States from HHANES were included.
i
White is non-Hispanic white; black is non-Hispanic black.
j
Biro et al
25
assessed testicular volume using a Prader orchidometer.
S182 EULING et al
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
pared with the earliest ages assessed in the Bogalusa and
NHANES III studies.
Comparison of the timing of genital development for
white boys between the studies conducted in the 1940s
to 1970s
14,17,61,69
and the 1980s to 1990s
6,7,45
suggests that
the mean or median age of G2 declined by 1 to 2 years
(Table 5). For the age of G3, the difference is less dra-
matic but is also earlier in 1988–1994. Comparing NHES
III and NHANES III data, Sun et al
65
calculated a decrease
in the median ages for G3 and G4 for non-Hispanic
white boys but not for non-Hispanic black boys during
this time span. Comparing HHANES and NHANES III
data for Mexican Americans, Sun et al
65
reported a de-
crease in the ages of G3 and G4 as well. For the age of
G4, differences are not consistently found when studies
are compared (eg, G4 is much later according to Lee
17
than to Reynolds and Wines
69
). The age of G5 varies
greatly, even among studies conducted within the same
decade, calling rater or marker reliability into question
(eg, G5 is earlier or later when comparing the 1940s
studies with various 1970 –1990 studies).
Comparison of the age of PH2 onset for white boys
between the earlier studies
17,61,69
and NHANES III (1988 –
1994)
6,7,45
suggests that the mean or median age for pubic
hair development onset has declined by ⬃0.2 to 0.5 years;
however, Foster et al
61
reported a higher PH2 mean age for
white boys than previous or subsequent studies (from the
1940s to 1950s and the 1988 –1994 NHANES III; Table 5),
suggesting that pubic hair staging is not a reliable marker.
Sun et al
65
reported that PH3 and PH4 were earlier for
white boys in NHANES III compared with NHES III. Mean
ages for PH4 and PH5 exhibit variability among studies,
even among those performed in the same decade. Looking
at the percentages of 12-year-old white boys in PH2 or
higher between NHES III and NHANES III data, similar
percentages (73.5% for NHES III; 71.5% for NHANES III)
were reported, suggesting no change in the timing of pro-
gression of pubic hair development overall between 1966 –
1970 and 1988 –1994.
For white boys’ data, comparisons of the duration of
male puberty are limited to the studies of Reynolds and
Wines,
69
Lee,
17
and NHANES III, from which the puber-
tal interval for white boys can be calculated as ⬃4.6
years (1940s), ⬃3.4 years (1960s/1970s), and ⬃5.5 to
5.6 years (1980s/990s). Excluding the study by Lee,
which found a shorter pubertal interval than the earlier
study (perhaps reflecting pubertal measure variability
issues), the other studies suggest that the pubertal inter-
val has increased between the 1940s and the 1990s.
Given the caveats of low reliability of male puberty
markers and the small number of studies, the reported
increased pubertal duration may reflect an earlier age of
onset (stage 2) with no significant change in the age of
completion (stage 5), as with the girls’ study findings.
DISCUSSION
Panel Consensus: Identified Areas of Agreement and
Disagreement
The panel agreed that these data indicate racial/ethnic
differences in the timing and order of pubertal events for
boys and girls and recommended that studies present
results by race/ethnic group. Racial/ethnic distinctions
may reflect genetic or environmental factors or a com-
bination of both. For example, an individual identified as
“non-Hispanic black” in NHANES III might have more in
common, genetically and environmentally, with others
identified differently. Even with the problems of defin-
ing race/ethnicity, the panel preferred and recom-
mended that studies present results by race/ethnic
group. Stratification by body size, particularly body fat,
was also considered appropriate because body fat and
growth have a profound impact on puberty timing.
All panel members questioned whether studies span-
ning the time frame 1940 –1994 were similar enough to
compare. Many of the existing studies differ in study
size, participant characteristics (eg, race/ethnicity, SES),
study group selection method, study type (eg, longitu-
dinal), marker assessment methods, and statistical tech-
niques.
Girls’ Data
Breast Development
Onset (B2)
The majority of panelists concluded that there are
sufficient data to suggest (most) or establish (fewer) a
secular trend toward an earlier age at onset of breast
development on the basis of B2 as a fairly reliable
marker of breast development onset and the opinion
that an early onset of breast development, even if iso-
lated, was a public health concern (Table 6). B2 is con-
sidered a more reliable external puberty-timing marker
than pubic hair development because breast develop-
ment requires estrogen action from the HPG axis. The
basis for this opinion was a weight-of-evidence approach
from comparing either the recent PROS or NHANES III
data for white girls with the earlier studies that assessed
B2
13,14,17
(Tables 2 and 3). Additional corroboration
comes from comparing PROS and Bogalusa Heart Study
B2 ages for black and white girls separately.
61
Compari-
sons of B2 ages for Hispanic girls were not possible
because earlier studies before HHANES did not report
data for Hispanic girls and the comparison by Sun et al
65
of HHANES and NHANES III did not report values for
B2; however, some panelists thought that the data for
Mexican American girls from HHANES
64
could be reli-
ably compared with those from NHANES III.
A minority of panelists concluded that there are in-
sufficient data to suggest a secular trend on the bassi of
the limited quantity and quality of the data, particularly
the lack of studies that assessed girls young enough to
determine reliably the timing of puberty onset markers,
such as B2. As mentioned, girls 10 or 12 years and older
were assessed in studies before the 1970s. There was
concern about the lack of palpation to determine B2 as
well as the difference in Tanner stage assessment
method (photographs versus visual examination) across
studies. The only study to use palpation in all breast
assessments was the small study by Lee et al
17
; however,
the PROS study performed palpation and visual assess-
PEDIATRICS Volume 121, Supplement 3, February 2008 S183
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
TABLE 5 Ages of Pubertal Events From Studies of US Boys
Study Mean/Median
a
Age of Attainment of a Stage, y (SD, SE, or 95% CI
b
)
G2 G3 G4 G5 PH2 PH3 PH4 PH5
Reynolds and Wines
69
(1951)
11.5 (0.9) NS
c
13.4 (0.7) 17.3 (0.8) 12.2 (1.1) 13.3 (0.8) 13.9 (0.7) 16.1 (1.2)
Nicolson and
Hanley
14
(1953)
d
11.8 (1.0) 13.1 (1.0) 13.8 (1.0) 15.2 (1.0) ND ND ND ND
NHES III; Karpati et al
7
(2002)
NR 13.0 (13.0–13.1) 13.9 (13.8–14.0) 15.1 (14.8–15.3) NR 13.3 (13.2–13.4) 14.0 (13.9–14.1) 15.3 (15.1–15.5)
NHES III; Sun et al
65
(2005)
ND 13.05 (12.91–13.17) white;
12.74 (12.38–13.02)
black
13.88 (13.72–14.04) white;
13.72 (13.47–13.94) black
15.07 (14.96–15.18) white;
14.76 (14.42–15.11) black
ND 13.29 (13.20–13.38) white;
13.28 (13.01–13.51)
black
14.04 (13.96–14.11) white;
14.03 (13.80–14.24)
black
15.39 (15.30–15.49) white;
15.20 (14.95–15.46)
black
Lee
17
(1980) 11.9 (1.1) 13.2 (0.8) 14.3 (0.8) 15.1 (1.1) 12.3 (0.8) 13.9 (0.9) 14.7 (0.9) 15.3 (0.8)
Bogalusa; Foster et al
61
(1977)
e
11.82 (0.11) white;
11.16 (0.185) black
NR NR NR 12.52 (0.085) white;
11.72 (0.125) black
NR NR NR
HHANES; Villarreal et al
64
(1989)
f
12.39 (12.14–12.58)
Mexican American
13.50 (13.27–13.73)
Mexican American
14.59 (14.35–14.89) Mexican
American
16.26 (15.94–16.54) Mexican
American
12.81 (12.43–12.83) Mexican
American
13.62 (13.36–13.84)
Mexican American
14.64 (14.32–14.94)
Mexican American
16.06 (15.71–16.33)
Mexican American
HHANES; Sun et al
65
(2005)
ND 13.29 (12.90–13.70)
Mexican American
14.38 (14.19–14.59) Mexican
American
16.08 (15.76–16.46) Mexican
American
ND 13.41 (13.10–13.72)
Mexican American
14.42 (14.17–14.69)
Mexican American
15.86 (15.61–16.14)
Mexican American
Biro et al
25
(1995)
g
ND ND ND ND 12.79 (NR) 13.74 (NR) 14.63 (NR) 15.19 (NR)
NHANES III; Herman-
Giddens et al
6
(2001)
10.1 (9.6–10.6) white;
9.5 (8.9–10.0)
black; 10.4 (9.6–
11.1) Mexican
American
12.4 (12.0–12.7) white;
11.8 (11.3–12.3) black;
12.5 (12.2–12.8)
Mexican American
13.5 (13.2–13.8) white; 13.4
(13.1–13.6) black; 13.7
(13.4–14.1) Mexican
American
15.9 (15.3–16.4) white; 14.9
(14.4–15.5) black; 15.7
(15.3–16.2) Mexican
American
12.0 (11.7–12.3) white; 11.2
(10.9–11.4) black; 12.3
(12.1–12.5) Mexican
American
12.6 (12.3–13.0) white;
12.5 (12.3–12.8) black;
13.1 (12.9–13.3)
Mexican American
13.5 (13.2–13.8) white;
13.7 (13.5–13.9) black;
14.1 (13.8–14.4)
Mexican American
15.7 (15.3–16.0) white;
15.4 (14.9–15.9) black;
15.8 (15.5–16.2)
Mexican American
NHANES III; Karpati et al
7
(2002)
10.1 (9.6–10.6) white;
9.3 (8.7–9.9) black;
10.4 (9.6–11.2)
Mexican American
12.4 (12.1–12.7) white;
11.8 (11.3–12.3) black;
12.5 (12.2–12.8)
Mexican American
13.5 (13.2–13.8) white; 13.4
(13.2–13.7) black; 13.8
(13.4–14.1) Mexican
American
16.0 (15.4–16.6) white; 14.9
(14.3–15.5) black; 15.8
(15.4–16.3) Mexican
American
12.0 (11.7–12.3) white; 11.1
(10.8–11.5) black; 12.3
(12.1–12.6) Mexican
American
12.7 (12.3–13.0) white;
12.6 (12.3–12.8) black;
13.1 (12.9–13.3)
Mexican American
13.6 (13.3–13.9) white;
13.7 (13.6–13.9) black;
14.0 (13.8–14.3)
Mexican American
15.6 (15.2–16.0) white;
15.2 (14.8–15.7) black;
15.7 (15.4–16.1)
Mexican American
NHANES III Sun et al
45
(2002)
h
10.03 (9.61–10.40)
white; 9.20 (8.62–
9.64) black; 10.29
(9.94–10.60)
Mexican American
12.32 (12.00–12.67) white;
11.78 (11.50–12.08)
black; 12.53 (12.29–
12.79) Mexican
American
13.52 (13.22–13.83) white;
13.40 (13.15–13.66) black;
13.77 (13.51–14.03)
Mexican American
16.01 (15.57–16.50) white;
15.00 (14.70–15.32) black;
15.76 (15.39–16.14)
Mexican American
11.98 (11.69–12.29) white;
11.16 (10.89–11.43) black;
12.30 (12.06–12.56)
Mexican American
12.65 (12.37–12.95) white;
12.51 (12.26–12.77)
black; 13.06 (12.79–
13.36) Mexican
American
13.56 (13.27–13.86) white;
13.73 (13.49–13.99)
black; 14.08
(13.83–14.32)
Mexican American
15.67 (15.30–16.05) white;
15.32 (14.99–15.67)
black; 15.75
(15.46–16.03)
Mexican American
NHANES III; Sun et al
65
(2005)
ND 12.44 (11.73–12.84) white;
12.37 (11.82–12.70)
black; 12.56 (12.34–
12.80) Mexican
American
13.49 (13.12–13.81) white;
13.46 (13.10–13.75) black;
13.76 (13.52–14.01)
Mexican American
15.83 (15.38–16.36) white;
14.86 (14.38–15.31) black;
15.60 (15.23–16.02)
Mexican American
ND 12.55 (11.95–12.93) white;
12.69 (12.27–12.99)
black; 13.08 (12.84–
13.33) Mexican
American
13.48 (13.12–13.79) white;
13.79 (13.45–14.09)
black; 14.08
(13.85–14.31)
Mexican American
15.59 (15.23–15.96) white;
15.21 (14.71–15.71)
black; 15.79
(15.49–16.11)
Mexican American
CI, confidence interval.
a
Means and medians are considered comparable values because the ages for various pubertal markers are typically normally distributed.
b
See references for details.
c
Reynolds and Wines
69
definitions of G stages differ from Tanner G stages (see Table 4 footnote b). Their G3 mean age is not shown here because it was defined differently from the other studies.
d
Nicolson and Hanley
14
used a 5-stage “sex stage” system with assessment of pubic hair and genital development; genital stage descriptions were similar to those of Tanner and are reported here as G2 to G5, but comparability with other studies that used the Tanner
stage definitions is a concern.
e
Foster et al
61
only presented median ages for transition into G2 and PH2 because of the “limited age range,” 5 to 14 years, of their population.
f
Villarreal et al
64
calculated percentages of boys in a given stage by age, means for being in a stage, means (from probit analysis), and medians (from Spearman-Karber analysis) for entry into a stage. Means for entry into a stage are presented in this table.
g
Biro et al
25
defined a PS1: “PS1 was defined as the absence of pubic hair and a testicular volume of the larger testis ⬍3 mL.” Their PS2 was similar to Tanner’s PH2. Data for comparable Tanner PH stages are presented here. PS1 data are not presented here. Data were
not presented by race.
h
Sun et al
45
calculated means for “age in a stage” and medians for “age at entry” into a stage. Medians are reported here for comparison with other studies that looked at age of attainment.
S184 EULING et al
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
ment for 39% of the participants,
19
and comparisons of
findings for those participants indicated no evidence of
biased staging when visual assessment alone was per-
formed (ie, mistaking fat tissue for breast tissue).
19
Some studies did not describe race/ethnicity of the
participants, include multiple racial/ethnic groups (eg,
all white population), or adjust for race/ethnicity, yield-
ing 1 summary result for all race/ethnic groups, which
limited comparisons for separate racial/ethnic groups
between the studies. For this reason, some panelists
considered data only for white girls because the older
studies assessed only or mainly white girls. The older
studies are mainly a high-SES population, which pre-
sents another complication when comparing data with
those of the more recent studies.
Some panelists did not consider B2 timing changes in
isolation, without subsequent breast development pro-
gression, as a marker of puberty. This viewpoint is based
on the fact that B2 can result without pubertal progres-
sion from exogenous estrogen exposure and, thus, does
not necessarily indicate HPG axis activation.
71,72
How-
ever, if pubertal development continues to progress after
onset, then this suggests that the breast development
onset was attributable to HPG activation. For this reason,
some panelists considered B2 to be a less reliable marker
of breast development timing than B3 to B5.
Progression (B3–B5)
The majority of the panelists concluded that the data
are insufficient to suggest or establish a secular trend
toward an earlier achievement of B3 to B5 stages. The
reasons for this opinion include a lack of study design
comparability between the historical and more recent
studies. Some panelists argued that the NHES II/III,
HHANES, and NHANES III studies were the only set that
could be reliably compared, and they concluded that the
change in ages of B3 to B5 was not significant over this
time period (1960s–1990s); however, NHES II/III and
NHANES III studies do not span the complete interval of
time (1940s–1990s) considered by the panel.
The minority concluded that there are sufficient data
to suggest a secular trend toward an earlier B3 to B5.
They viewed B3 to B5 as more reliable markers of breast
development timing because progression indicates acti-
vation of the HPG and they are less prone to inaccuracies
in staging than B2. The basis for earlier B3 to B5 came
from comparing either the more recent NHANES III or
PROS data for white girls with the earlier studies
13,14,17
(Table 3).
Pubic Hair Development
Onset (PH2)
The majority concluded that PH2 is highly variable
(ie, interindividual and interracial/ethnicity variability of
ages for PH2), and, therefore, the data are insufficient to
suggest or establish a secular trend in the age of PH2.
Some panelists viewed pubic hair development, in gen-
eral, as less reliable and more variable than breast de-
velopment. A concern about the appropriateness of Tan-
ner staging of pubic hair development
10
across racial/
ethnic groups, because it was developed on the basis of
white children, was raised. As with breast development,
the lack of studies with similar designs was considered a
key issue that hindered conclusions about secular
changes.
A minority concluded that the data are sufficient to
suggest a secular trend toward an earlier onset of pubic
hair development because they considered PH2 a more
reliable marker than B2 (in the absence of palpation).
The more recent studies suggest that the mean age of
PH2 is ⬃6 months earlier than several decades ago. This
conclusion comes from comparing the recent PROS and
NHANES III data for white girls that showed an earlier
PH2 to any of the earlier studies
13,14,17
(Table 3).
Progression (PH3–PH5)
The panel unanimously concluded that the data are
insufficient to suggest or establish a secular trend in the
timing of pubic hair stages PH3 through PH5 on the basis
of (1) the view that PH3 to PH5 markers are relatively
subjective (particularly discerning between 4 and 5) and
variable (eg, interracial/ethnicity and interindividual);
(2) data comparability issues; and (3) exclusion of reli-
ance on the PROS PH3 to PH5 timing data, because the
oldest girls were 12 years old and, thus, many were too
young to assess these stages.
Menarche
The majority concluded that the data are sufficient to
suggest (most) or establish (fewer) a secular trend to-
ward an earlier age of menarche. Most panelists with
this opinion included only the NHES II/III and NHANES
III studies in their evaluation because these studies used
similar methods and population selection criteria. These
panelists were convinced by the finding that menarche
has decreased by 2.5
3
to 4
46
months during ⬃25 years.
Others considered the full spectrum of studies spanning
1940 –1994 on which to base their conclusions.
A minority of the panelists concluded that the cross-
sectional study data are insufficient to conclude that
there has been a biologically meaningful change in the
age of menarche. Some panelists agreed with the con-
clusions of Chumlea et al
46
that interpreted the 4-month
decrease as not significant. In addition, some panelists
TABLE 6 Panel Conclusions About Evidence of Secular Trends in the
Timing of Puberty for 1940 –1994 Data
Marker Sufficient Data:
Yes Trend
Insufficient
Data
Sufficient Data:
No Trend
Girls’ BD onset (B2) ⫹⫹ ⫹ ⫺
Girls’ BD progression (B3–B5) ⫹⫹⫹⫺
Girls’ PH onset (PH2) ⫹⫹⫹⫺
Girls’ PH progression (PH3–PH5) ⫺ ⫹⫹⫹ ⫺
Girls’ menarche ⫹⫹ ⫹ ⫺
Boys’ GD onset (G2) ⫺ ⫹⫹⫹ ⫺
Boys’ GD progression (G3–G5) ⫺ ⫹⫹⫹ ⫺
Boys’ PH onset (PH2) ⫹⫹⫹⫺
Boys’ PH progression (PH3–PH5) ⫹⫹⫹⫺
BD indicates breast development; ⫺, no one’s opinion; ⫹, minority opinion; ⫹⫹, majority
opinion; ⫹⫹⫹, unanimous opinion.
PEDIATRICS Volume 121, Supplement 3, February 2008 S185
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
excluded the menarche data from the PROS study
5
from
consideration because the age range (3–12 years) was
not optimal for calculating the mean age at menarche.
Some panelists were concerned about the impact of the
different statistical methods used in the studies that com-
pared the NHES II/III and NHANES III data.
Boys’ Data
All agreed that statistical power and data quality issues
make robust conclusions about the boys’ data difficult.
The panel discussed but did not come to conclusions on
data regarding the tempo of puberty and the pubertal
growth spurt.
Genital Development
The unanimous opinion was that the data for genital
development (G2 and G3–G5) are insufficient to suggest
or establish a trend toward an earlier puberty in boys, on
the basis of the lack of data quantity, data quality, and
marker reliability. Many boys who were included in
NHES III and NHANES III were too old for adequate G2
or PH2 assessment. In addition, the lack of validity and
reliability of genital staging, particularly for G2, was
considered problematic. Panelists agreed that testicular
volume assessment, not performed in the studies, was a
more meaningful measure of puberty onset where a
testicular volume of ⬎3 mL is conclusive evidence of
HPG activation. Data quality was a concern given the
degree of disparity, particularly for genital development,
between the NHANES III findings and NHES III find-
ings.
6
Pubic Hair Development
An overwhelming majority found the pubic hair devel-
opment (PH2 and PH3–PH5) timing data to be insuffi-
cient to suggest or establish a trend. Similar to the argu-
ment for insufficiency of genital development data, this
opinion was based on the lack of data quantity, data
comparability, and marker reliability.
A very small minority of panelists concluded that
there is suggestive evidence of an earlier pubic hair onset
(PH2) and progression (PH3–PH5) for Mexican Ameri-
can boys but not for white or black boys during this time
frame. This opinion was based on the viewpoint that
pubic hair development may be easier to stage than
genital development and, thus, more reliable. Data sup-
porting a trend toward an increase in linear height in
boys from the 1963–1970 to 1988 –1994
7
and from 1973
to 1992
73
were also considered corroborative of an ear-
lier male pubertal growth spurt.
Reasons for Differences of Opinion
Differences of opinion among the panelists were based
on judgment calls that placed more or less value on key
aspects of study comparability. Overall, some panelists
placed more weight on nationally representative studies
(NHES I–III, HHANES, and NHANES III) because of sim-
ilar designs yet performed at different times. For exam-
ple, the clinicians tended to place more value on these
data perhaps because they look to nationally represen-
tative data on which to base normative puberty-timing
clinical guidelines. In assessing the girls’ data, some of
the panelists placed more weight on the study by Her-
man-Giddens et al
5
because of its much larger sample
size and, hence, statistical power and inclusion of
younger ages for assessing Tanner staging. One impor-
tant design choice was the Tanner stage progression
definition (eg, some studies reported the mean age of
being “in a stage,” others reported the mean age of
“transition to a stage,” and some reported both). Differ-
ent statistical methods also affect comparisons among
studies; however, few panelists were experts in statistics
and none in survey statistics. For this reason, detailed
scrutiny of similar and appropriate statistical analyses,
within and among the studies, was not a focus of the
discussions.
Panel Research Recommendations
New Analyses of Existing Data
If the original data from historical puberty studies are
available, then parallel analyses of data from multiple
studies could yield more meaningful results. Thus, when
possible, reanalysis of some of the existing studies with
similar statistical approaches is recommended. For ex-
ample, multiple examinations of the same participants
should not be treated as separate cross-sectional studies
or included in a single analysis as though they were from
separate children. For the studies of largest study popu-
lations, 1 approach to examining secular trends is to
compare children at specific ages in studies that were
conducted at different times and in which each child is
only examined once.
When analyzing a probability-based study, such as 1
that uses stratified multistage sampling, analyses must
incorporate sample weights to yield accurate results. The
size of the population affects the power and, in turn, the
study quality. It is important to use statistical methods
that are appropriate for the type of data and question at
hand, as well as to have comparable statistical methods.
The data from the existing studies could be examined
for changes in temporal sequence and tempo of puberty for
different racial/ethnic groups. The relative timing of puber-
tal events may be a critical factor in separating out different
types of puberty-timing phenotypes. For example, discor-
dance between Tanner stages (eg, simultaneous PH4 and
B2) suggests a relatively high androgen-to-estrogen con-
centration.
Priority New Studies
The panel recommended longitudinal studies as the op-
timal design for examining puberty timing in future
studies. Tanner stages are best assessed for an individual
over time, because it is a relative measure. Pubertal
marker data collection may also be imbedded within
larger studies. Examples of longitudinal studies in which
pubertal timing are or will be tracked are the National
Children’s Study
74,75
(http://nationalchildrensstudy.gov)
of 100 000 children to be followed from in utero to 21
years of age, the Danish National Birth Cohort study
76
(www.ssi.dk/sw3820.asp) of 100 000 women and their
S186 EULING et al
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
offspring during the first 6 years of life, and the Norwe-
gian Mother and Child Cohort Study (www.fhi.no/
artikler/?id⫽51488). These and other studies will pro-
vide baseline information with which to compare past
studies and future cohorts of children and test hypoth-
eses regarding specific exposures and puberty timing.
A second useful study design is a short-term longitu-
dinal study with sequential cohorts (eg, the Bogalusa
Heart Study), sometimes referred to as a mixed cross-
sectional/longitudinal study. The sequential cohort de-
sign is excellent for looking at trends within 1 study
because design choices are maintained and secular
trends could be detected sooner than with the longitu-
dinal study design. For either of these study designs,
collaboration among disciplines and between countries
could enhance and encourage the use of similar study
designs, which would allow for geographic and temporal
comparisons.
For the purposes of secular trend analysis, any new
study using different methods (eg, testicular volume as-
sessment) is still limited by methods used in the earlier
studies. Thus, new studies would benefit from including
design aspects that are in common with the historical
studies. For example, the most recent NHANES surveys
(4 cycles spanning 1999 –2006) have not collected Tan-
ner stage data for boys and girls, losing comparability
with the past NHES (I–III), HHANES, and NHANES III
data. Recognizing that there are practical limitations of
conducting large studies of puberty timing, the panel
recommended a study design “wish list” for longitudinal
studies:
● individual observations repeated every 6 months;
● Tanner staging;
● breast palpation and testicular volume measurement
for puberty onset;
● hormonal measurements;
● adequately wide age ranges (eg, 6–18 years of age);
● geographic differences considered, including migra-
tion and adoption;
● SES noted;
● race/ethnicity noted and defined and data stratified
when appropriate;
● body fat and weight covariates (eg, BMI);
● inclusion (or targeted studies) of other racial/ethnic
groups (eg, Native Americans, Asians) in large enough
numbers to analyze; and
● hypothesis testing.
Tanner staging method was considered critical to the
collection of puberty-timing data, but inclusion of breast
tissue palpation and testicular volume assessment was
recommended to increase the reliability of estimating
the ages of breast and genital development onset. Of
note, testicular volume in early puberty is being assessed
in a US boys’ puberty-timing PROS study.
77
Training
standards for assessing Tanner stage were considered
necessary (for girls
15
; for boys
78
), and some suggested
that only trained clinicians should be raters; however,
the use of palpation by a trained examiner for breast and
testicular development has practical (eg, financial, tech-
nical training) and ethical (eg, consent for breast palpa-
tion or testicular measurement of study participants)
constraints.
Underlying the interest in performing a secular trend
analysis is the concern about whether environmental
factors alter the age at puberty. If so, then unless the
environmental factors are distributed uniformly, they
should result in geographic variation that is measurable
in cross-sectional studies. This reasoning led to well-
controlled, multicenter studies of semen quality con-
ducted recently in the United States, Europe, and Japan
that demonstrated significant geographic variation in
semen quality.
79–81
These studies identified areas with
unusually high and low semen quality and thus suggest
etiologic studies to identify causes of these differences.
82
Such an approach may be useful in identifying factors
that affect age at puberty. One testable hypothesis with
new studies is whether the secular trend toward an
earlier puberty (at least concluded for girls) may be
attributable to subpopulation differences that may in
turn reflect differential exposures to factors that affect
puberty. If puberty timing varies by geography, then the
next question is which factor(s) distinguishes 1 geo-
graphic region from another (eg, an exposure, genetics).
In support of the hypothesis of geographic variation
in puberty-timing data, studies from several countries
have found somewhat different mean ages of menarche
(reviewed by Parent et al
58
). For example, European data
collected in the 1990s showed mean ages of menarche
that are older than those from the US data (eg, menarche
mean ages ranged from 13.0 to 13.4 years in Den-
mark,
83,84
⬃13.2 years in Norway
85
and Sweden,
86
and
12.6 –12.9 years for white girls in the United
States
3,5,46,47
). Urban–rural differences in the mean age of
menarche have also been observed within a country
87
;
however, it is not possible to conclude definitively that
mean age at menarche is different among countries
without an analysis of the study design comparability.
Racial patterns of differences in puberty timing may
also differ by geographic areas. For example, reported
racial differences from South African data on female
puberty timing have a different pattern than those re-
ported from US data. The median ages for breast devel-
opment initiation (B2) and completion (B5) were similar
for black and white girls, but black girls had a later PH2
and PH5 than white girls who lived in Johannesburg,
88
and the median age at menarche for black girls was 10
months later than for white girls. These differences in
puberty timing by geography may be a reflection of SES,
race/ethnicity, and/or environmental factors.
Recommended Methods Development
Puberty markers that are easier to measure, less inva-
sive, less expensive, and more sensitive and reliable need
to be explored, developed, and validated. New methods
to enhance and/or replace Tanner staging are recom-
mended. These fall into 2 categories: physical markers
and physiologic markers.
PEDIATRICS Volume 121, Supplement 3, February 2008 S187
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
Physical Markers
Tanner staging was recognized as an excellent system
for assessing pubertal onset and progression; however, a
less invasive examination could decrease stress to the
participant. Improvements that may achieve this goal
include using a same-gender examiner and developing
methods and markers. Methods and markers to be de-
veloped and validated include the following:
● behavioral markers of male and female puberty onset;
● muscle mass increase in boys;
● infrared photographs for Tanner stage analysis;
● computer imaging for testicular volume measure-
ments;
● race/ethnic-specific Tanner staging (eg, pubic hair
staging in Peruvian Indians is not accurate for estimat-
ing sexual maturation
89
; it was noted that PH4 for
Asians is typically the last, adult-type stage); and
● male pubertal markers not currently extensively used
(eg, spermarche, voice break).
Physiologic Markers
The development of reliable biochemical markers
would represent a major step forward for puberty-timing
studies in that biochemical changes may identify precur-
sor events and provide validation of later phenotypic
puberty changes. Important biochemical markers in-
clude measurement of hormone and hormone receptor
levels. Hormone level assays are available, but improved
assay sensitivity and reliability for urine, blood, and sa-
liva are needed for use in large-scale human studies.
Assessment of changing hormone levels during a 24-
hour period complicates measurements. For example, in
prepubertal boys and girls, gonadotropins in serum ex-
hibit large diurnal variations as a result of their pulsatile
secretion; therefore, any single measurement is unreli-
able.
34
The uncertainty can be overcome by measuring
urinary secretion of luteinizing hormone, which in-
creases before physical signs of puberty are evident.
90
During early puberty in boys, testosterone measures are
most reliable when blood sampling is done in the morn-
ing, when testosterone levels are expected to be high-
est.
91
There are a few available assays for steroid receptor
levels in humans that are limited in their validity and
still under additional development.
92
Recommended
physiologic marker methods for development and vali-
dation include the following:
● improved sensitivity and reliability and reduced cost of
hormone level measurements to define further the
associations between hormone levels and pubertal sta-
tus;
● steroid hormone receptor level measurements; and
● puberty markers that distinguish between central pre-
cocious puberty and variants of precocious puberty
(eg, nonprogressive early B2).
CONCLUSIONS
The optimum data for studying puberty timing would be
the collection of a suite of puberty-timing measures that
include physical as well as physiologic markers. The
combination of physical and physiologic measures of
human puberty timing will help in the understanding of
the progression from onset to completion and of inter-
individual variability and in the characterization of bio-
chemical precursors. For example, sensitive hormone
assay data in conjunction with Tanner staging data could
define a marker that precedes overt physical changes yet
still allow for comparison with earlier studies of puberty
timing.
ACKNOWLEDGMENTS
Support for “The Role of Environmental Factors on the
Onset and Progression of Puberty” workshop was pro-
vided by US Environmental Protection Agency cooper-
ative agreement 830774, the National Institute of Envi-
ronmental Health Sciences (this research was supported
in part by the Intramural Research Program of the Na-
tional Institutes of Health and National Institute of En-
vironmental Health Sciences), and Serono Inc.
We gratefully acknowledge David Schlumper and the
Serono Symposia International for organizing and hold-
ing “The Role of Environmental Factors on the Onset
and Progression of Puberty” Expert Panel Workshop. We
thank Tim Mealey of the Meridien Institute for skillful
moderation of the workshop and Beth Gladen for edu-
cating us about puberty study design and analytical
methods and providing detailed comments on the
manuscript. We thank the other members of the Expert
Panel for participation and contributions: David Abbott,
Carlos Bourdony, Jean-Pierre Bourguignon, Germaine
Buck Louis, Farid Chehab, Gwen Collman, Ralph Coo-
per, Paul Foster, Beth Gladen, Mari Golub, L. Earl Gray,
Mel Grumbach, Paul Kaplowitz, Carole Kimmel, Robert
Lustig, Michele Marcus, Sergio Ojeda, Ora Pescovitz,
Tony Plant, Ed Reiter, Steve Schrader, Richard Sharpe,
Wolfgang Sippell, Niels Skakkebaek, Ana Soto, Jorma
Toppari, Rochelle Tyl, Martin Wabitsch, Patricia Whit-
ten, and Selma Witchel.
REFERENCES
1. Zacharias L, Wurtman RJ. Age at menarche: genetic and envi-
ronmental influences. N Engl J Med. 1969;280(16):868 – 875
2. Wyshak G, Frisch RE. Evidence of a secular trend in age of
menarche. N Engl J Med. 1982;306(17):1033–1035
3. Anderson SE, Dallal GE, Must A. Relative weight and race
influence average age at menarche: results from two nationally
representative surveys of US girls studied 25 years apart. Pedi-
atrics. 2003;111(4 pt 1):844 – 850
4. Freedman DS, Kahn LK, Serdula MK, Dietz WH, Srinivasan SR,
Berenson GS. Relation of age at menarche to race, time period,
and anthropometric dimensions: the Bogalusa Heart Study.
Pediatrics. 2002;110(4). Available at: www.pediatrics.org/cgi/
content/full/110/4/e43
5. Herman-Giddens ME, Slora EJ, Wasserman RC, et al. Second-
ary sexual characteristics and menses in young girls seen in
office practice: a study from the Pediatric Research in Office
Settings Network. Pediatrics. 1997;99(4):505–512
6. Herman-Giddens ME, Wang L, Koch G. Secondary sexual char-
acteristics in boys: estimates from the National Health and
Nutrition Examination Survey III, 1988–1994. Arch Pediatr
Adolesc Med. 2001;155(9):1022–1028
S188 EULING et al
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
7. Karpati AM, Rubin CH, Kieszak SM, Marcus M, Troiano RP.
Stature and pubertal stage assessment in American boys: the
1988 –1994 Third National Health and Nutrition Examination
Survey. J Adolesc Health. 2002;30(3):205–212
8. Euling SY, Selevan SG, Pescovitz OH, Skakkebaek NE. The role
of environmental factors in the timing of puberty. Pediatrics.
2008;121(2 suppl):S167–S171
9. Tanner JM. Growth at Adolescence. 2nd ed. Oxford, England:
Blackwell Scientific Publications; 1962
10. Marshall WA, Tanner JM. Variations in pattern of pubertal
changes in girls. Arch Dis Child. 1969;44(235):291–303
11. Marshall WA, Tanner JM. Variations in the pattern of pubertal
changes in boys. Arch Dis Child. 1970;45(239):13–23
12. Van Wieringen JC, Wafelbakker F, Verbrugge HP, De Haas JH.
Growth diagrams. In: Yen SSC, Jaffe RB, eds. Reproductive
Endocrinology: Physiology, Pathophysiology and Clinical Manage-
ment. 2nd ed. Philadelphia, PA: WB Saunders; 1978:1–68
13. Reynolds EL, Wines JV. Individual differences in physical
changes associated with adolescence in girls. Am J Dis Child.
1948;75:329 –350
14. Nicolson AB, Hanley C. Indices of physiological maturity: der-
ivation and interrelationships. Child Dev. 1953;24(1):3–38
15. Herman-Giddens ME, Bourdony CJ. Assessment of Sexual Matu-
rity Stages in Girls. Elk Grove Village, IL: Pediatric Research in
Office Settings, American Academy of Pediatrics; 1995. Report
MA0089
16. Biro FM, Lucky AW, Simbartl LA, et al. Pubertal maturation in
girls and the relationship to anthropometric changes: pathways
through puberty. J Pediatr. 2003;142(6):643– 646
17. Lee PA. Normal ages of pubertal events among American males
and females. J Adolesc Health Care. 1980;1(1):26 –29
18. Herman-Giddens ME, Kaplowitz PB, Wasserman R. Navigating
the recent articles on girls’ puberty in Pediatrics: what do we
know and where do we go from here? Pediatrics. 2004;113(4):
911–917
19. Kaplowitz PB, Oberfield SE; Drug and Therapeutics and Exec-
utive Committees of the Lawson Wilkins Pediatric Endocrine
Society. Reexamination of the age limit for defining when
puberty is precocious in girls in the United States: implications
for evaluation and treatment. Pediatrics. 1999;104(4 pt 1):
936 –941
20. US Department of Health and Human Services. Third National
Health and Nutrition Examination Survey, 1988 –1994, NHANES III
Laboratory Data File [CD-ROM]. Hyattsville, MD: National Cen-
ter for Health Statistics, Centers for Disease Control and
Prevention; 1996. Public use data file documentation No.
76200
21. Jolliffe D. Extent of overweight among US children and ado-
lescents from 1971 to 2000. Int J Obes Relat Metab Disord.
2004;28(1):4 –9
22. Must A, Phillips SM, Naumova EN, et al. Recall of early men-
strual history and menarcheal body size: after 30 years, how
well do women remember? Am J Epidemiol. 2002;155(7):
672– 679
23. Damon A, Damon ST, Reed RB, Valadian I. Age at menarche of
mothers and daughters, with a note on accuracy of recall. Hum
Biol. 1969;41(2):160 –175
24. Zachmann M, Prader A, Kind HP, Hafliger H, Budliger H.
Testicular volume during adolescence: cross-sectional and lon-
gitudinal studies. Helv Paediatr Acta. 1974;29(1):61–72
25. Biro FM, Lucky AW, Huster GA, Morrison JA. Pubertal staging
in boys [published correction appears in J Pediatr. 1995;127(4):
674]. J Pediatr. 1995;127(1):100 –102
26. Rivkees SA, Hall DA, Boepple PA, Crawford JD. Accuracy and
reproducibility of clinical measures of testicular volume. J Pe-
diatr. 1987;110(6):914 –917
27. Chipkevitch E, Nishimura RT, To DS, Galea-Rojas M. Clinical
measurement of testicular volume in adolescents: comparison
of the reliability of 5 methods. J Urol. 1996;156(6):2050 –2053
28. Carlsen E, Anderson AG, Buchreitz L, et al. Inter-observer
variation in the results of the clinical andrological examination
including estimation of testicular size. Int J Androl. 2000;23(4):
248 –253
29. Prader A. Testicular size: assessment and clinical importance.
Triangle. 1966;7(6):240 –243
30. al Salim A, Murchison PJ, Rana A, Elton RA, Hargrave TB.
Evaluation of testicular volume by three orchidometers com-
pared with ultrasonographic measurements. Br J Urol. 1995;
76(5):632– 635
31. Shiraishi K, Takihara H, Kamiryo Y, Naito K. Usefulness and
limitation of punched-out orchidometer in testicular volume
measurement. Asian J Androl. 2005;7(1):77– 80
32. Nielsen CT, Skakkebaek NE, Richardson DW, et al. Onset of the
release of spermatozoa (spermarche) in boys in relation to age,
testicular growth, pubic hair, and height. J Clin Endocrinol
Metab. 1986;62(3):532–535
33. Jørgensen M, Keiding N, Skakkebaek NE. Estimation of sper-
marche from longitudinal spermaturia data. Biometrics. 1991;
47(1):177–193
34. Wu FC, Butler GE, Kelnar CJ, Sellar RE. Patterns of pulsatile
luteinizing hormone secretion before and during the onset of
puberty in boys: a study using an immunoradiometric assay.
J Clin Endocrinol Metab. 1990;70(3):629 – 637
35. Rockett JC, Lynch CD, Buck GM. Biomarkers for assessing
reproductive development and health: part 1—pubertal devel-
opment. Environ Health Perspect. 2004;112(1):105–112
36. Last JM, ed. A Dictionary of Epidemiology. 4th ed. New York, NY:
Oxford University Press; 2001
37. Cox DR. Analysis of Binary Data. London, England: Methuen
and Co Ltd; 1970
38. Wu Y, Schreiber GB, Klementowicz V, Biro F, Wright D. Racial
differences in accuracy of self-assessment of sexual maturation
among young black and white girls. J Adolesc Health. 2001;
28(3):197–203
39. Schlossberger NM, Turner RA, Irwin CE Jr. Validity of self-
report of pubertal maturation in early adolescents. J Adolesc
Health. 1992;13(2):109 –113
40. Bonat S, Pathomvanich A, Keil MF, Field AE, Yanovski JA.
Self-assessment of pubertal stage in overweight children. Pedi-
atrics. 2002;110(4):743–747
41. Iba´n˜ ez L, Ong KK, Mongan N, et al. Androgen receptor gene
CAG repeat polymorphism in the development of ovarian hy-
perandrogenism. J Clin Endocrinol Metab. 2003;88(7):
3333–3338
42. Kadlubar FF, Berkowitz GS, Delongchamp RR, et al. The
CYP3A4*1B variant is related to the onset of puberty, a known
risk factor for the development of breast cancer. Cancer Epide-
miol Biomarkers Prev. 2003;12(4):327–331
43. Stavrou I, Zois C, Ioannidis JP, Tsatsoulis A. Association of
polymorphisms of the oestrogen receptor alpha gene with the
age of menarche. Hum Reprod. 2002;17(4):1101–1105
44. Xita N, Tsatsoulis A, Stavrou I, Georgiou I. Association of SHBG
gene polymorphism with menarche. Mol Hum Reprod. 2005;
11(6):459 – 462
45. Sun S, Schubert CM, Chumlea WC, et al. National estimates of
the timing of sexual maturation and racial differences among
US children. Pediatrics. 2002;110(5):911–919
46. Chumlea WC, Shubert CM, Roche AF, et al. Age at menarche
and racial comparisons in US girls. Pediatrics. 2003;111(1):
110 –113
47. Wu T, Mendola P, Buck GM. Ethnic differences in the presence
of secondary sex characteristics and menarche among US girls:
the Third National Health and Nutrition Examination Survey,
1988 –1994. Pediatrics. 2002;110(4):752–757
PEDIATRICS Volume 121, Supplement 3, February 2008 S189
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
48. Galler JR, Ramsey FC, Salt P, Archer E. Long-term effects of
early kwashiorkor compared with marasmus: I—physical
growth and sexual maturation. J Pediatr Gastroenterol Nutr.
1987;6(6):841– 846
49. Cameron N, Mitchell J, Meyer D, et al. Secondary sexual
development of Cape coloured girls following kwashiorkor.
Ann Hum Biol. 1988;15(1):65–75
50. Cameron N, Mitchell J, Meyer D, et al. Secondary sexual
development of Cape coloured boys following kwashiorkor.
Ann Hum Biol. 1990;17(3):217–228
51. Kulin HE, Bwibo N, Mutie D, Santner SJ. The effect of chronic
childhood malnutrition on pubertal growth and development.
Am J Clin Nutr. 1982;36(3):527–536
52. Koprowski C, Ross RK, Mack WJ, Henderson BE, Bernstein L.
Diet, body size and menarche in a multiethnic cohort. Br J
Cancer. 1999;79(11–12):1907–1911
53. Kaplowitz PB. The link between body fat and the timing of
puberty. Pediatrics. 2008;121(2 suppl):S208 –S217
54. Lazar L, Pollak U, Kalter-Leibovici O, Pertzelan A, Phillip M.
Pubertal course of persistently short children born small for
gestational age (SGA) compared with idiopathic short children
born appropriate for gestational age (AGE). Eur J Endocrinol.
2003;149(5):425– 432
55. Iba´n˜ ez L, Ferrer A, Marcos MV, Hierro FR, de Zegher F. Early
puberty: rapid progression and reduced final height in girls
with low birthweight. Pediatrics. 2000;106(5). Available at:
www.pediatrics.org/cgi/content/full/106/5/e72
56. Veening MA, van Weissenbruch MM, Roord JJ, de Delme-
marre-van Waal HA. Pubertal development in children born
small for gestational age. J Pediatr Endocrinol Metab. 2004;
17(11):1497–1505
57. Powls A, Botting N, Cooke RWI, Pilling D, Marlow N. Growth
impairment in very low birthweight children at 12 years: cor-
relation with perinatal and outcome variables. Dis Child Fetal
Neonatal Ed. 1996;75:F152–F157
58. Parent AS, Teilmann G, Juul A, Skakkebaek NE, Toppari J,
Bourguignon. The timing of normal puberty and the age limits
of sexual precocity: variations around the world, secular
trends, and changes after migration. Endocr Rev. 2003;24(5):
668 – 693
59. Krstevska-Konstantinova M, Charlier C, Craen M, et al. Sexual
precocity after immigration from developing countries to
Belgium: evidence of previous exposure to organochlorine pes-
ticides. Hum Reprod. 2001;16(5):1020 –1026
60. dos Santos Silva I, De Stavola BL, Mann V, Kuh D, Hardy R,
Wadsworth MEJ. Prenatal factors, childhood growth trajecto-
ries and age at menarche. Int J Epidemiol. 2002;31(2):405– 412
61. Foster TA, Voors AW, Webber LS, Frerichs, Berenson GS.
Anthropometric and maturation measurements of children,
ages 5 to 14 years, in a biracial community: the Bogalusa Heart
Study. Am J Clin Nutr. 1977;30(4):582–591
62. Harlan WR, Harlan EA, Grillo GP. Secondary sex characteristics
of girls 12 to 17 years of age: the U.S. Health Examination
Survey. J Pediatr. 1980;96(6):1074 –1078
63. MacMahon B. Age at Menarche: United States. Rockville, MD:
National Center for Health Statistics; 1973. DHEW publication
(HRA)74 –1615
64. Villarreal SF, Martorell R, Mendoza F. Sexual maturation of
Mexican-American adolescents. Am J Hum Biol. 1989;1(1):
87–95
65. Sun SS, Schubert CM, Liang R, et al. Is sexual maturity occur-
ring earlier among U.S. children? J Adolesc Health. 2005;37(5):
345–355
66. Weir HK, Kreiger N, Marrett LD. Age at puberty and risk of
testicular germ cell cancer (Ontario, Canada). Cancer Causes
Control. 1998;9(3):253–258
67. United Kingdom Testicular Cancer Study Group. Aetiology of
testicular cancer: association with congenital abnormalities,
age at puberty, infertility, and exercise. BMJ. 1994;308(6941):
1393–1399
68. Golub MS, Collman GW, Foster MD, et al. Public health im-
plications of altered puberty timing. Pediatrics. 121;2(2 suppl):
S218 –S230
69. Reynolds EL, Wines JV. Physical changes associated with ado-
lescence in boys. Am J Dis Child. 1951;82(5):529–547
70. Harlan WR, Grillo GP, Cornoni-Huntley J, Leaverton P. Sec-
ondary sex characteristics of boys 12 to 17 years of age: the U.S.
Health Examination Survey. J Pediatr. 1979;95(2):293–297
71. Hertz R. Accidental ingestion of estrogens by children. Pediat-
rics. 1958;21(2):203–206
72. Tiwary CM. Premature sexual development in children follow-
ing the use of estrogen- or placenta-containing hair products.
Clin Pediatr (Phila). 1998;37(12):733–739
73. Freedman DS, Khan LK, Serdula MK, Srinivasan SR, Berenson
GS. Secular trends in height among children during 2 decades:
the Bogalusa Heart Study. Arch Pediatr Adolesc Med. 2000;
154(2):155–161
74. Branum AM, Collman GW, Correa A, et al. The National
Children’s Study of Environmental Effects on Child Health
and Development. Environ Health Perspect. 2003;111(4):
642– 646
75. Wang RY, Needham LL, Barr DB. Effects of environmental
agents on the attainment of puberty: considerations when
assessing exposure to environmental chemicals in the National
Children’s Study. Environ Health Perspect. 2005;113(8):
1100 –1107
76. Olsen J, Melbye M, Olsen SF, et al. The Danish National Birth
Cohort: its background, structure and aim. Scand J Public
Health. 2001;29(4):300 –307
77. Herman-Giddens ME. Recent data on pubertal milestones in
United States children: the secular trend toward earlier devel-
opment. Int J Androl. 2006;29(1):241–246
78. Herman-Giddens ME, Bourdony CJ, Dowshen SA. Assessment of
Sexual Maturity Stages in Boys. Elk Grove Village, IL: Pediatric
Research in Office Settings, American Academy of Pediatrics;
2005
79. Jorgensen N, Carlsen E, Nermoen I, et al. East-West gradient in
semen quality in the Nordic-Baltic area: a study of men from
the general population in Denmark, Norway, Estonia and Fin-
land. Hum Reprod. 2002;17(8):2199 –2208
80. Punab M, Zilaitiene B, Jorgensen N, et al. Regional differences
in semen qualities in the Baltic region. Int J Androl. 2002;25(4):
243–252
81. Swan SH, Brazil C, Drobnis EZ, et al. Geographic differences in
semen quality of fertile US males. Environ Health Perspect. 2003;
111(4):414 – 420
82. Swan SH, Kruse R, Liu F, et al. Biomarkers of pesticide expo-
sure in relation to semen quality. Environ Health Perspect. 2003;
111(12):1478 –1484
83. Helm P, Grolund L. A halt in the secular trend towards earlier
menarche in Denmark. Acta Obstet Gynecol Scand. 1998;77(2):
198 –200
84. Juul A, Teilmann G, Scheike T, et al. Pubertal development in
Danish children: comparison of recent European and US data.
Int J Androl. 2006;29(1):247–255; discussion 286 –290
85. Nafstad P, Stray-Pedersen B, Solvberg M, Tangen T. Menarche
and menstruation problems among teenagers in Oslo 1993 [in
Norwegian]. Tidsskr Nor Laegeforen. 1995;115(5):604 – 606
86. Lindgren GW, Degerfors IL, Fredriksson A, et al. Menarche
1990 in Stockholm schoolgirls. Acta Paediatr Scand. 1991;
80(10):953–955
87. Rimpela¨ AH, Rimpela MK. Towards an equal distribution of
health? Socioeconomic and regional differences of the secular
S190 EULING et al
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
trend of the age of menarche in Finland from 1979 to 1989.
Acta Paediatr. 1993;82(1):87–90
88. Chaning-Pearce SM, Solomon L. Pubertal development in
black and white Johannesburg girls. S Afr Med J. 1987;71(1):
22–24
89. Freyre EA, Llaza G, Rebaza-Gonzalez R, Sami D. Tanner’s
pubic hair staging is not applicable to all adolescents. J Adolesc
Health. 2002;30(3):144 –145
90. Demir A, Voutilainen R, Juul A, et al. Increase in first morning
voided urinary luteinizing hormone levels precedes the phys-
ical onset of puberty. J Clin Endocrinol Metab. 1996;81(8):
2963–2967
91. Wu FC, Brown DC, Butler GE, Stirling HF, Kelnar CJ. Early
morning plasma testosterone is an accurate predictor of imminent
pubertal development in prepubertal boys. J Clin Endocrinol Metab.
1993;76(1):26 –31
92. Mudarris A, Peck EJ Jr. Human anti-estrogen receptor
antibodies: assay, characterization, and age- and sex-related
differences. J Clin Endocrinol Metab. 1987;64(2):246 –254
93. Wattigney WA, Srinivasan SR, Chen W, Greenlund KJ, Beren-
son GS. Secular trend of earlier onset of menarche with in-
creasing obesity in black and white girls: the Bogalusa Heart
Study. Ethn Dis. 1999;9(2):181–189
94. Greulich WW, Dorfman RI, Catchpole HR, Solomon CI,
Culotta CS. Somatic and endocrine studies of pubertal and
adolescent boys. Monogr Soc Res Child Dev. 1942;7(3):1
PEDIATRICS Volume 121, Supplement 3, February 2008 S191
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
DOI: 10.1542/peds.2007-1813D
2008;121;S172Pediatrics
Shanna H. Swan
andAnders Juul, Thorkild I.A. Sørensen, Leo Dunkel, John H. Himes, Grete Teilmann
Susan Y. Euling, Marcia E. Herman-Giddens, Peter A. Lee, Sherry G. Selevan,
Panel Findings
Examination of US Puberty-Timing Data from 1940 to 1994 for Secular Trends:
Services
Updated Information &
_3/S172.full.html
http://pediatrics.aappublications.org/content/121/Supplement
including high resolution figures, can be found at:
Citations
_3/S172.full.html#related-urls
http://pediatrics.aappublications.org/content/121/Supplement
This article has been cited by 39 HighWire-hosted articles:
Subspecialty Collections
b
http://pediatrics.aappublications.org/cgi/collection/puberty_su
Puberty
ogy_sub
http://pediatrics.aappublications.org/cgi/collection/endocrinol
Endocrinology
the following collection(s):
This article, along with others on similar topics, appears in
Permissions & Licensing
ml
http://pediatrics.aappublications.org/site/misc/Permissions.xht
tables) or in its entirety can be found online at:
Information about reproducing this article in parts (figures,
Reprints
http://pediatrics.aappublications.org/site/misc/reprints.xhtml
Information about ordering reprints can be found online:
rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.
Grove Village, Illinois, 60007. Copyright © 2008 by the American Academy of Pediatrics. All
and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk
publication, it has been published continuously since 1948. PEDIATRICS is owned, published,
PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly
at Mt Sinai School Of Med on October 16, 2014pediatrics.aappublications.orgDownloaded from
