Association of age at menarche with cardiovascular risk factors,
vascular structure, and function in adulthood: the Cardiovascular
Risk in Young Finns study1–3
Mika Kivimäki, Debbie A Lawlor, George Davey Smith, Marko Elovainio, Markus Jokela, Liisa Keltikangas-Järvinen,
Jussi Vahtera, Leena Taittonen, Markus Juonala, Jorma SA Viikari, and Olli T Raitakari
ABSTRACT
Background: It is unclear whether age at menarche is an indepen-
dent determinant of future cardiovascular risk.
Objective: We aimed to determine whether menarcheal age is an
independent predictor of body mass index (BMI) and a wide range of
cardiovascular risk factors in adolescence and adulthood.
Design: We examined the associations of menarcheal age with BMI
(in kg/m2) and other cardiovascular risk factors in adolescence and
adulthood in a population-based sample of 794 female adolescents
aged 9 –18 y at baseline. Their age at first menstruation was re-
quested at baseline and again 3 and 6 y later. Cardiovascular risk
factors were assessed at baseline and at age 30 –39 y.
Results: A 1-y decrease in menarcheal age was associated with 0.81
(95% CI: 0.53, 1.08) higher adult BMI as well as greater waist
circumference and waist-to-hip ratio, elevated systolic blood pres-
sure, higher insulin resistance, and greater risk of metabolic syn-
drome (P 쏝 0.05 for all). In multivariable analysis in which these
adult risk factors were mutually adjusted for, only the inverse asso-
ciation between age at menarche and adult BMI remained. However,
this inverse association was lost after adjustment for premenarcheal
BMI (␤: Ҁ0.16; 95% CI Ҁ0.55, 0.23; P ҃ 0.42). Higher premen-
archeal BMI predicted earlier menarche, and the strong association
between premenarcheal BMI and adult BMI was robust to adjust-
ment for age at menarche.
Conclusions: These findings suggest that early menarche is only a
risk marker. Greater childhood BMI seems to contribute to earlier
age at menarche and, because of tracking, greater adult BMI and
associated cardiovascular risk. An independent effect of early men-
arche on adult adiposity cannot be excluded, but it is likely to be
small at best. Am J Clin Nutr 2008;87:1876 – 82.
INTRODUCTION
Many studies show that girls with earlier age at menarche tend
to have worse cardiovascular risk factor levels in adulthood than
those who underwent menarche at a later age (1–11). The most
consistent evidence relates to higher adult body mass index
(BMI; in kg/m2) in women who had on average earlier menarche
(1, 2, 6 – 8, 10, 12), but an inverse association of menarcheal age
has also been reported with blood pressure (6), glucose intoler-
ance (6), and risk of ischemic heart disease and stroke (4, 5, 13,
14). Whether early menarche is an independent predictor of ad-
verse adult outcomes or only a marker of other risk factors re-
mains unclear.
1876 Am J Clin Nutr 2008;87:1876 – 82. Printed in USA. © 2008 American Society for Nutrition
It has been hypothesized that having an earlier age at menarche
causes important differences in postpubertal growth and devel-
opment of greater adiposity (1, 12, 15). This view emphasizes the
importance of targeting girls with earlier menarche for antiobe-
Downloaded from ajcn.nutrition.org by guest on February 8, 2017
sity interventions (1, 6). However, if the associations are largely
driven by prepubertal risk factors, such targeting would not result
in public health benefit. The latter is plausible because it is known
that greater adiposity in childhood and factors that promote the
accumulation of body fat are associated with an earlier age at
menarche (8, 16 –19) and that greater adiposity tracks from child-
hood to adulthood (11). Thus, children with greater BMI would
be expected to have on average earlier menarche and would be
expected to have greater BMI and its associated risk factors in
adulthood. Other exposures occurring before puberty that are
associated both with younger age at menarche and greater
adult adiposity (eg, familiar socioeconomic adversity and intra-
uterine influences) may also result in such spurious associations
(20 –23).
1
From the Department of Epidemiology and Public Health, University
College London, London, United Kingdom (MK); MRC Centre for Causal
Analysis in Transitional Epidemiology, the Department of Social Medicine,
University of Bristol, Bristol, United Kingdom (DAL and GDS); the Depart-
ment of Psychology, University of Helsinki, Helsinki, Finland (ME, MJo,
and LK-J); the Finnish Institute of Occupational Health, Helsinki, Finland
(JV); the Department of Paediatrics, University of Oulu, Oulu, Finland (LT);
the Department of Paediatrics, Central Hospital of Vaasa, Vaasa, Finland and
Research Centre of Applied and Preventive Cardiovascular Medicine, Uni-
versity of Turku, Turku, Finland (MJu); the Department of Medicine, Uni-
versity of Turku, Turku, Finland (JSAV); and the Department of Clinical
Physiology, University of Turku, Finland (OTR).
2
The Cardiovascular Risks in Young Finns study has been supported by the
Academy of Finland (grants 77841 and 210283), the Social Insurance Institution
of Finland, the Finnish Work Environment Foundation, Turku University Foun-
dation, Juho Vainio Foundation, the Finnish Foundation of Cardiovascular Re-
search, the Finnish Cultural Foundation, and Turku University Central Hospital
Research Funds, Finland. MK and LK-J are supported by the Academy of
Finland (grants 117604, 105195, and 1209514). DAL is funded by a United
Kingdom Department of Health Career Scientist Award.
3
Address reprint requests and correspondence to M Kivimäki, Depart-
ment of Epidemiology and Public Health, University College London, 1-19
Torrington Place, London WC1E 6BT, United Kingdom. E-mail:
m.kivimaki@ucl.ac.uk.
Received October 24, 2007.
Accepted for publication February 13, 2008.
 AGE AT MENARCHE AND CARDIOVASCULAR RISK FACTORS
To date at least 7 studies have taken account of potential early
risk factors in assessing this association. The Northern Finland
birth cohort for 1966 (2), a cohort of individuals born in Aber-
deen, Scotland, in the 1950s (7), and the Fels longitudinal study
of white girls aged 8 –21 y (6) showed that the association of
earlier age at menarche with greater BMI in adulthood was not
explained by other factors measured. In contrast, in a study of
Philadelphia high school girls, the inverse association of age at
menarche with variations in BMI 1 y after menarche was mostly
explained by earlier BMI (24), and similar results were found for
the association of age at menarche with adult BMI in the Boga-
lusa Heart study (8), the 1958 British Birth Cohort (11), and the
Newton Girls study (10) (the association was largely driven by
childhood BMI in these studies). Methodologic problems in all of
these previous studies may have contributed to the conflicting
results: in 5 of the studies, measurement of adult BMI only
extended up to age 21 (6, 24) or was completely or partially based
on self-reported height and weight (2, 7, 10), in 2 of the studies
there was large loss to follow-up (ie, 쏜40%) (7, 8), in 1 study the
baseline measurement was assessed over a 50-y period and there-
fore included many birth cohorts in whom age at menarche and
adiposity distributions are likely to be heterogeneous (6), and in
1 study age at menarche was assessed retrospectively in late
adulthood (11). None of these studies examined adult cardiovas-
cular outcomes other than adiposity.
We have measurements of age at menarche in adolescence and
prospective information on childhood and adult BMI determined
by standard research measurements and a range of adult cardio-
vascular risk factors [fasting glucose, insulin, lipids, blood pres-
sure, and carotid intima-media thickness (IMT)] for a subsample
of the Young Finns study. The aim of the present study was to
determine whether menarcheal age is an independent predictor of
BMI and a wide range of cardiovascular risk factors in adoles-
cence and adulthood.
SUBJECTS AND METHODS
Study population
The first clinical examination in the population-based Cardio-
vascular Risk in Young Finns Study (25, 26) was conducted for
age cohorts of 3, 6, 9, 12, 15, and 18 y in 1980, with the partic-
ipation rate being 83%, ie, 3596 (1832 female and 1764 male) of
those invited (25, 26). Reexaminations included follow-ups in
1983, 1986, and 2001. At the latest follow-up, the participants
had reached 24 –39 y of age. Participants (or their parents) were
asked about their age at first menstruation at baseline and again
at 1983 and 1986 but have not been asked this since 1986. By
1986 the cohort were aged 9 (those aged 3 at baseline) to 24 (those
aged 18 at baseline). The 3- and 6-y-old cohorts at baseline were
excluded from this study because they were only 9 or 12 y when
menarche was last measured: for those aged 3 y at baseline
virtually none reported experiencing their first menses and for
those aged 6 y at baseline 쏝30% reported experiencing their first
menses. Thus, the eligible cohort for this study was all women
who were aged 9 –18 y at study baseline (and 15–24 y the last time
age at menarche was assessed): n ҃ 1247. Of these eligible
women 119 (9.5% of those eligible) had missing data on age at
menarche and an additional 334 (26.8%) did not attend the
follow-up clinic in 2001–2002 and therefore did not have infor-
mation on outcome measures at age 30-39 y. Thus, our present
1877
study includes 794 women (63.7% of those eligible) with data on
age at menarche and at least one outcome measure.
The 794 women included were older (age at baseline 13.6
compared with 12.8 y; P 쏝 0.0001) than the 453 excluded women
from the same age cohorts, but no differences were found in
baseline BMI (P ҃ 0.43), systolic blood pressure (P ҃ 0.62),
total cholesterol (P ҃ 0.84), HDL and LDL cholesterol concen-
trations (P ҃ 23 and P ҃ 0.63), or triglyceride concentrations (P
҃ 0.73) between these groups. There were no difference in mean
age of menarche between the participants and the 334 women
who had data on menarcheal age but were lost to follow-up (P ҃
0.69).
Assessment of age at menarche
Age at menarche was assessed by an interview on 3 occa-
sions— baseline (1980), 1983, and 1986. Participants were asked
age at their first menses (in y and mo). The 9- and 12-y-olds were
interviewed with their parents. Among the 514 girls who partic-
ipated in more than one examination after menarche, 83% of the
recalled ages at menarche differed by 울1 y and for those women
Downloaded from ajcn.nutrition.org by guest on February 8, 2017
who reported their age at first menstruation on all 3 occasions
different reports were highly correlated (Cronbach ␣ ҃ 0.92 for
all assessments), indicating high reproducibility of our measure.
Mean difference in recalled menarcheal age between the first
and second measurements was 쏝 0.03 (SD: 0.81) y and Bland-
Altman plots (27) did not show any evidence that the differences
varied in any systematic way over the range of measurement
(correlation between difference and mean scores for reported age
at menarche over 2 examinations: r ҃ 0.01, P ҃ 0.72). The
differences in reported menarcheal age between the examina-
tions were not associated with premenarcheal BMI (Pearson r ҃
0.05, P ҃ 0.53), postmenarcheal BMI before age 18 (r ҃ 0.08, P
҃ 0.12), or BMI at adulthood (r ҃ Ҁ0.03, P ҃ 0.96), and the
plots for difference by average menarcheal age over the 2 exam-
inations were similar for high and low BMI at baseline and
follow-up. In all analyses we used the first report of the date of
menarche for each woman, because errors in recalling age at
menarche are likely to increase with the duration of time.
Assessment of risk factors
For those included in our analyses, birth weight (g) and birth
height (cm) were reported by the mothers who were asked to
bring with them the booklet from the well baby center in which
these information was recorded in 1983. Other childhood and
adolescence risk factors were assessed in 1980 at the age of 9 –18
y and adulthood risk factors were assessed in 2001–2002 at the
age of 30 –39 y (26). Several cardiovascular risk factors were
assessed at both time points. Socioeconomic position was mea-
sured by parental occupational status as classified by Statistics
Finland (28) and categorized as manual versus nonmanual (29).
If socioeconomic position differed between parents, data on the
parent with the higher occupational status were used. The par-
ticipant’s own adult socioeconomic position was measured by
occupational status and categorized as for parental socioeco-
nomic position. Physical measurements of weight (kg) and
height (mm) and in 2001 also waist circumference (mm, mea-
sured in duplicate at the level of the twelfth rib or level with the
navel in thin subjects) and hip circumference (mm) were ob-
tained to calculate BMI and waist-to-hip ratio. Blood pressure
was measured with a standard mercury sphygmomanometer in
1878 KIVIMÄKI ET AL
1980 and with a random-zero sphygmomanometer in 2001 (30).
The average of 3 measurements was used in statistical analysis.
Fasting blood samples were taken in both 1980 and 2001. All
measurements of lipid concentrations were performed in dupli-
cate in the same laboratory. Standard enzymatic methods were
used for measuring concentrations of serum total cholesterol,
HDL cholesterol, and triglyceride. LDL cholesterol was calcu-
lated by the Friedewald formula (31). Interassay CVs were 2.0%
in 1980 and 2.2% in 2001 for serum cholesterol, 2.0% and 2.3%
for HDL cholesterol, and 4.7% and 3.8% for serum triglyceride
concentrations (32, 33).
In 2001, serum insulin was measured by a microparticle en-
zyme immunoassay kit (Abbott Laboratories, Diagnostic Divi-
sion, Dainabot, Tokyo, Japan). Details of these methods have
been described previously (34, 35). Insulin resistance was esti-
mated according to the homeostasis model (36) as the product of
fasting glucose and insulin divided by the constant 22.5. The
metabolic syndrome was defined according to the International
Diabetes Federation criteria (37) as abdominal obesity and 욷2 of
the following criteria: waist 욷94 cm in men and 욷80 cm in
women, fasting plasma glucose 욷5.6 mmol/L, hypertriglyceri-
demia 욷1.695 mmol/L and HDL cholesterol concentrations
쏝1.036 mmol/L in men and 쏝1.295 mmol/L in women, and
blood pressure 욷130/욷85 mm Hg or treatment. Information on
number of children, use of contraceptives, adult smoking, and
alcohol consumption (units/wk) was obtained by questionnaire
in 2001. One unit of alcohol (12 g) was equal to a glass of wine,
a single 4-cl shot of spirits, or a 33-cl bottle of beer (29).
In 2001–2002, ultrasound studies were performed with use of
Sequoia 512 ultrasound mainframes (Acuson, Mountain View,
CA) to measure carotid IMT and brachial artery flow-mediated
dilation (26, 38). In brief, the image was focused on the posterior
(far) wall of the left carotid artery. A minimum of 4 measure-
ments of the common carotid far wall were taken 앒10 mm
proximal to the carotid bifurcation to derive mean carotid IMT.
The between-visit CV for the IMT measurements was 6.4% (26).
To assess brachial artery flow-mediated dilation, the left brachial
artery diameter was measured both at rest and during reactive
hyperemia. The vessel diameter from scans after reactive hyper-
emia was expressed as the percentage relative to the diameter
from the resting scan. The 3-mo between-visit CV was 3.2% for
the brachial artery diameter measurements and 26.0% for the
flow-mediated dilation measurements (38).
Statistical analysis
We examined the association between age at menarche and
risk factors in childhood and adulthood with linear regression
models including menarcheal age as a continuous variable. To
determine primary adult risk outcomes, we tested linear associ-
ations of menarcheal age with each adult risk factor in a model
adjusting for age at the time of the adult clinical examination and
other adult risk factors.
For childhood and adolescent risk factors, we constructed z
scores (x៮ : 0, SD: 1) of these risk factors, standardizing for age at
the time of clinical examination to take account of the association
of age with risk factor distributions (12, 39). We determined
whether the assessment of risk factors between the ages of 9 and
18 y was before or after menarche and tested whether age-at-
examination-standardized risk factor z scores after menarche
were more strongly associated with menarcheal age than age-at-
examination-standardized risk factor z scores before menarche
by including an interaction term, age at menarche ҂ timing of
risk factor measurement, in a model including main effects. A
stronger association between menarcheal age and postmenar-
cheal risk factor levels than between menarcheal age and pre-
menarcheal risk factor levels would be consistent with the hy-
pothesis that elevated risk factor levels are a consequence rather
than a determinant of menarcheal age. We used multivariable
linear regression models to further examine whether age at men-
arche is an independent predictor of premenarcheal risk factors.
The first step involved a series of models in which we tested
age-adjusted associations of menarcheal age and premenarcheal
risk factor z scores with adult risk factors. The second step en-
tered these predictors together into the same model.
In each analysis, participants with missing values were ex-
cluded. All of the data analysis was performed by using SAS
software (version 9.1; SAS Institute Inc, Cary, NC). In all tests,
statistical significance was inferred at a 2-tailed P 쏝 0.05.
RESULTS
Downloaded from ajcn.nutrition.org by guest on February 8, 2017
Sample characteristics are presented in Table 1 and Table 2.
We assessed menarcheal age on average at the age of 14.4 y for
girls who had their first menses before age 12 y and at the age of
16.1 y for those with menarche at age 14 y or older (Table 1).
Higher BMI in childhood and adolescence was associated with
early menarche with a slightly stronger association for BMI
measured before rather than after menarche. Other baseline risk
factors were not associated with menarcheal age.
Early menarche predicted greater BMI, waist circumference,
and waist-to-hip ratio, elevated systolic blood pressure, higher
insulin concentrations and insulin resistance, and marginally
greater prevalence of metabolic syndrome in adulthood (Table
2). In multivariable analysis in which we mutually adjusted adult
risk factors for each other, only the association between age at
menarche and adult BMI remained, suggesting that age-adjusted
associations with other risk factors were fully explained by the
association of age at menarche with adult BMI (Table 3). This
association remained after additional adjustment for use of
contraceptives and the number of pregnancies and when ad-
justed for years after menarche instead of age. Analyses with
age-at-examination–standardized z scores replicated these
findings (data not shown). Thus, subsequent analyses relate to
menarcheal age and BMI only.
An independent effect of age at menarche on subsequent BMI
would be supported if age at menarche was more strongly asso-
ciated with subsequent BMI than premenarche BMI was associ-
ated with age at menarche. Analysis of BMI z scores between the
ages 9 and 18 y provided no such evidence (Table 4). If anything,
the association between premenarcheal BMI and age at men-
arche was stronger than the association between age at menarche
and postmenarcheal BMI, although the formal test failed to con-
firm a difference in the magnitude of these associations (P for
interaction age at menarche ҂ timing of BMI measurement ҃
0.08).
BMI in childhood and adolescence predicted adult BMI, waist
circumference, waist-to-hip ratio, systolic and diastolic blood
pressure, HDL cholesterol concentrations, glucose and insulin
concentrations, insulin resistance and metabolic syndrome, ca-
rotid IMT, and brachial flow-mediated dilation (all P 쏝 0.05). If
there was an independent association between age at menarche
and adult BMI, this association should be robust to controlling for
 AGE AT MENARCHE AND CARDIOVASCULAR RISK FACTORS 1879
TABLE 1
Levels of childhood and adolescence risk factors in all participants and by age at menarche1

By age at menarche (y)

울11.9 12–12.9 13–13.9 욷14

Risk factor Subjects Value (n ҃ 108) (n ҃ 231) (n ҃ 266) (n ҃ 189) P for trend

n
2
Age (y) 794 13.6 앐 3.2 14.43 15.13 15.93 16.13 N/A
Age at menarche (y) 794 13.1 앐 1.2 11.3 12.4 13.3 14.7 N/A
Parental SEP (% of manual) 766 40.9 37.5 42.6 40.0 41.8 0.56
Birth weight (g) 687 3428 앐 514 3479 3433 3414 3409 0.43
Birth height (cm) 672 50.0 앐 2.3 50.2 50.1 50.0 49.8 0.31
BMI (in kg/m2) 792 19.0 앐 2.9 19.7 19.4 18.9 18.3 0.0001
Premenarcheal 349 17.1 앐 2.4 17.7 17.9 16.8 16.5 0.0008
Postmenarcheal 443 20.4 앐 2.4 20.7 20.5 20.5 19.9 0.06
Height (cm) 792 155.5 앐 12.7 155.2 155.6 155.9 155.0 0.40
Systolic blood pressure (mm Hg) 794 115.1 앐 9.9 115.6 115.6 114.8 114.5 0.80
Diastolic blood pressure (mm Hg) 794 69.4 앐 9.2 69.5 69.2 69.5 69.2 0.88
Total cholesterol (mmol/L) 789 5.13 앐 0.8 5.17 5.13 5.08 5.18 0.52
HDL cholesterol (mmol/L) 793 1.52 앐 0.28 1.50 1.51 1.51 1.54 0.11
LDL cholesterol (mmol/L) 793 3.32 앐 0.74 3.37 3.32 3.27 3.35 0.88

Downloaded from ajcn.nutrition.org by guest on February 8, 2017

Triglycerides (mmol/L) 793 0.65 앐 0.33 0.66 0.65 0.66 0.63 0.42
1
N/A, not applicable; SEP, socioeconomic position. All risk factors were measured in 1980. Age at menarche was measured in 1980, 1983, and 1986 (the
first report of the date of menarche for each woman was used).
2
x 앐 SD (all such values).
3
Participant’s age when menarcheal age was assessed.

BMI before menarche. However, among the 341 women with the
baseline measurement of BMI before menarche, the age-adjusted
association between age at menarche and adult BMI was almost
completely lost after adjustment for premenarcheal BMI, indi-
cating that age at menarche had no independent effect on adult
BMI (Table 5). In contrast, the strong effect of premenarcheal
BMI on adult BMI remained largely unchanged when adjusted
for age at menarche. The same was true for the association be-
tween postmenarcheal BMI assessed at age 18 y or earlier and
adult BMI.

TABLE 2
Adult risk factors in all participants and by age at menarche1

By age at menarche (y)

Risk factor in adulthood Subjects Value 울11.9 12–12.9 13–13.9 욷14 P for trend

n
Socioeconomic position (% of manual) 715 22.5 24.0 22.3 22.4 22.2 0.49
Smoking (% of smokers) 771 18.5 19.4 17.9 18.2 19.3 0.96
Alcohol consumption (units/wk)2 782 3.4 앐 4.53 3.1 3.3 3.6 3.7 0.37
BMI (in kg/m2) 772 24.7 앐 4.5 26.7 25.2 24.4 23.5 쏝0.0001
Waist circumference (cm) 772 80.4 앐 11.3 85.0 81.2 79.7 78.0 쏝0.0001
Waist-to-hip ratio 772 0.80 앐 0.06 0.82 0.80 0.80 0.79 0.02
Height (cm) 772 165.9 앐 5.8 165.8 165.9 165.6 166.4 0.89
Systolic blood pressure (mm Hg) 794 116.3 앐 12.8 119.4 116.7 115.3 115.5 0.02
Diastolic blood pressure (mm Hg) 785 69.4 앐 10.1 70.9 69.9 68.9 68.6 0.19
Total cholesterol (mmol/L) 791 5.14 앐 0.93 5.20 5.14 5.11 5.15 0.99
HDL cholesterol (mmol/L) 791 1.39 앐 0.30 1.37 1.40 1.38 1.42 0.36
LDL cholesterol (mmol/L) 788 3.24 앐 0.77 3.29 3.23 3.21 3.26 0.83
Triglycerides (mmol/L) 791 1.15 앐 0.71 1.21 1.18 1.15 1.08 0.10
Glucose (mmol/L) 791 4.94 앐 0.73 5.16 4.89 4.91 4.93 0.31
Insulin (mU/L) 791 7.54 앐 5.25 9.45 7.77 7.12 6.76 쏝0.0001
Insulin resistance (HOMA index) 789 1.69 앐 1.34 2.19 1.73 1.58 1.50 쏝0.0001
Metabolic syndrome4 741 13.3 16.7 14.6 15.3 6.8 쏝0.05
Intima-media thickness (mm) 793 0.587 앐 0.086 0.585 0.593 0.588 0.579 0.54
Brachial artery flow–mediated dilation (%) 746 9.14 앐 4.66 8.75 9.16 8.99 9.53 0.71
1
Measured at mean (앐SD) age of 34.6 앐 3.2 y in 2001. HOMA, homeostatic model assessment.
2
One unit (12 g) was equal to 1 glass of wine, a single 4-cL shot of spirits, or a 33-cL bottle of beer.
3
x 앐 SD (all such values).
4
According to the International Diabetes Federation criteria.
1880 KIVIMÄKI ET AL

TABLE 3
Age-adjusted and multivariate models on the association between age at menarche and adult risk factors1

Change in outcome per 1-y increase in menarcheal age

Age-adjusted model Mutually adjusted model2

Adulthood outcome ␤2 95% CI P ␤ 95% CI P

BMI (in kg/m )2

Ҁ0.81 Ҁ1.08, Ҁ0.53 쏝0.0001 Ҁ0.10 Ҁ0.19, Ҁ0.01 0.033
Waist circumference (cm) Ҁ17.77 Ҁ24.77, Ҁ10.77 쏝0.0001 0.52 Ҁ1.24, 2.29 0.56
Waist-to-hip ratio Ҁ0.006 Ҁ0.010, Ҁ0.002 0.003 Ҁ0.000 Ҁ0.002, 0.002 0.86
Systolic blood pressure (mm Hg) Ҁ1.07 Ҁ1.87, Ҁ0.28 0.008 Ҁ0.24 Ҁ0.99, 0.50 0.52
Insulin (mU/L) Ҁ0.600 Ҁ0.925, Ҁ0.275 0.0003 0.001 Ҁ0.059, 0.061 0.96
Insulin resistance (HOMA index) Ҁ0.155 Ҁ0.238, Ҁ0.072 0.0003 Ҁ0.002 Ҁ0.017, 0.013 0.78
Metabolic syndrome4 Ҁ0.022 Ҁ0.043, Ҁ0.001 0.04 0.008 Ҁ0.010, 0.026 0.37
1
n ҃ 742. HOMA, homeostatic model assessment.
2
Adjusted for age and all other outcomes presented in the table.
3
This association remained after additional adjustment for use of contraceptives and the number of pregnancies and when adjusted for years after menarche
instead of age (P ҃ 0.03).
4
Binary variable.

Downloaded from ajcn.nutrition.org by guest on February 8, 2017

These results suggest that premenarcheal BMI largely ex-
plains the association of age at menarche with adult BMI. A
multivariable model including all premenarcheal cardiovascular
factors as predictors of adult BMI provided little evidence for
additional confounding by premenarcheal blood pressure, lipid
concentrations, or parental socioeconomic position (␤ for age at
menarche: Ҁ0.16; P ҃ 0.42 in a model adjusted for age and
premenarcheal BMI and Ҁ0.15; P ҃ 0.47 in a model additionally
adjusted for other premenarcheal cardiovascular risk factors
among the 334 women with no missing data on premenarcheal
risk factors).
DISCUSSION
This population-based study suggests that age at menarche is
associated with adult BMI and associated risk factors, but we
obtained no support for the interpretation that these associations
would reflect an independent effect of early menarche on worse
cardiovascular risk factor levels in adulthood. We found that
greater premenarcheal BMI was associated with earlier age at
menarche and that they both were associated with greater adult
BMI. In multivariate models the association of age at menarche
with adult BMI was almost completely lost after adjustment for
BMI assessed before menarche and its association with other
adult cardiovascular risk factors, such as blood pressure, glucose,
insulin resistance, and metabolic syndrome and attenuated to the
TABLE 4
Age-adjusted association between menarcheal age and age-at-exam–
standardized BMI z score before and after menarche at age 9 –18 y1
Change in BMI per 1-y increase in
menarcheal age
BMI at age 9 –18 y Subjects ␤ 95% CI
n
z Score before menarche 349 Ҁ0.30 Ҁ0.39, Ҁ0.22
z Score after menarche 443 Ҁ0.17 Ҁ0.25, Ҁ0.09
1
Interaction between BMI z score and timing of BMI assessment (be-
fore vs after menarche), P ҃ 0.08.
null after adjustment for adult BMI. Rather than supporting in-
dependent effects of earlier menarche on adult adiposity, these
findings are in agreement with the hypothesis that menarcheal
age is only a risk marker: greater childhood BMI may contribute
to earlier age at menarche and, because of tracking, greater adult
BMI. Adult BMI itself is related to other adult cardiovascular risk
factors explaining the association of early menarche with these
factors.
The Cardiovascular Risk in Young Finns Study is based on 6
birth cohorts with each randomly chosen in 5 areas of Finland
from the national register (25). We excluded the 2 youngest birth
cohorts aged 3 and 6 y at baseline as few of them reported
experiencing their first menses during the 6 subsequent years
when menarcheal age was assessed, but exclusion of these whole
birth cohorts should not introduce any selection bias. In addition,
we had to exclude 9% of those eligible (ie, at age 욷9 y at baseline)
because of missing data on menarcheal age and a further 27%
because of lack of adulthood outcomes assessed 21 y after the
baseline. It is not possible to determine whether this moderate
loss of data caused selection bias but to do so and reverse our
findings would have to mean that those excluded had a strong
association of age at menarche with adult BMI that was inde-
pendent of childhood BMI; ie, the excluded women would have
to have one or more of following: weaker association between
childhood and adult BMI than those included; weaker associa-
tions of age at menarche with childhood BMI than those in-
cluded; or stronger association of age at menarche with adult
BMI than those included. Because the likelihood of responding
to a question about age at menarche is unlikely to be affected by
future BMI and there was no evidence that attending a follow-up
clinic was determined by age at menarche, a major selection bias
because of these missing data seems unlikely.
This study was based on a smaller sample than the largest
P studies in the field (7, 11), but it had the advantage of reports of
age at menarche close to time of menarche in adolescence and
objective measures of height and weight both in childhood or
쏝0.0001 adolescence and in adulthood. This data collection reduces the
쏝0.0001 risk for artificial inflation of associations between age at men-
arche and adult BMI attributable to common method bias that
may arise if age at menarche was reported many years later in
 AGE AT MENARCHE AND CARDIOVASCULAR RISK FACTORS
TABLE 5
Associations of BMI z score at age 9 –18 y and menarcheal age with BMI at age 30 –39 y by timing of age-at-exam—standardized BMI z score measured
at age 9 –18 y1
Age-adjusted model
␤2 95% CI
BMI before menarche (n ҃ 341)
BMI z score 2.18 1.76, 2.61
Menarcheal age (y) Ҁ0.80 Ҁ1.21, Ҁ0.39
BMI after menarche (n ҃ 129)
BMI z score 2.33 1.95, 2.70
Menarcheal age (y) Ҁ0.69 Ҁ1.08, Ҁ0.30
1
BMI was measured in kg/m2. Interaction between BMI z score and timing of the BMI assessment (before vs after menarche) on adult BMI, P ҃ 0.31.
2
Model includes age, BMI z score between ages 9 and 18 y, and menarcheal age as predictors of adult BMI.
older adulthood and BMI was determined on the basis of self-
reported height and weight. Imprecision in reports of age at
menarche can lead to underestimation of associations. In analy-
sis, we used the first report of the date of menarche for each
woman to minimize recall bias, but data on 3 repeated measure-
ments enabled tests regarding the validity of the measure.
Cronbach ␣ reliability 쏜0.9 across the measurements, a mean
difference between repeated measurements of 쏝0.03 y, and
Bland-Altman plots suggested that our indicator of age at men-
arche was reproducible and reliable. Any inconsistencies in re-
peated measurements of menarcheal age were independent of
pre- and postmenarcheal adiposity, providing evidence against
systematic error. In addition to adult BMI, we assessed waist
circumference and waist-to-hip ratio in adulthood, but they may
all be imperfect measures of adiposity. This would bias our
conclusion regarding the role of menarcheal age as a marker of
premenarcheal influence only if the use of an imperfect proxy
measure for adiposity led to substantially greater underestima-
tion of the association of age at menarche with postmenarcheal
adiposity than that with premenarcheal adiposity.
Our findings confirm several earlier studies with weaknesses
that were overcome in this study. For example, the Bogalusa
Heart Study found that adult obesity was more strongly related to
childhood BMI than menarcheal age, but that study was limited
by 쏜60% loss of participants to follow-up and a reproducibility
of assessment of age at menarche that was “somewhat lower than
in other studies” (8, p. 8), investigators of the British 1958 Birth
Cohort Study reported an association between early menarche
and adult obesity but did not provide an estimate of the associ-
ation after adjustment for premenarcheal BMI (40), the Newton
Girls study suggested an association of childhood risk factors
with early menarche, but the findings were based on a small
nonrepresentative sample (10), and the Philadelphia high school
studies were limited to only 1-y follow-up (24).
In contrast to our findings, the UK Aberdeen study found that
early menarche was an independent predictor of adult BMI (7).
In that study, childhood BMI was measured but adult BMI was
derived from self-reported weight and height and age at men-
arche recalled at 42-52 y of age. Common method bias in the
cross-sectional association between age at menarche and adult
BMI combined with a low tracking correlation between child-
hood BMI at age 4-7 and adult BMI (r ҃ 0.19) may have con-
tributed to a false impression of independent effects of early
menarche. In our study, for example, the tracking correlation is
1881
Adult BMI
Mutually adjusted model2
P ␤ 95% CI P
쏝0.0001 2.12 1.66, 2.58 쏝0.0001
0.0001 Ҁ0.16 Ҁ0.55, 0.23 0.42
쏝0.0001 2.27 1.89, 2.65 쏝0.0001
0.0006 Ҁ0.30 Ҁ0.64, 0.05 0.09
much higher (r ҃ 0.52) among those whose BMI was measured
at 9 y of age with the use of BMI z scores similar to those in the
Aberdeen study. The investigators of the Fels Longitudinal
Downloaded from ajcn.nutrition.org by guest on February 8, 2017
Study reported that early menarcheal age was related to increases
in both adiposity and lean tissue and adversely affected cardio-
vascular risk factor changes, such as elevated blood pressure and
glucose intolerance, independent of body composition (6). How-
ever, the serial analyses of repeated measures from 8 –18 y of age
did not explicitly test reverse causality because risk factor trends
before and after menarche were not distinguished and a cohort
effect due to extensive variation in timing of baseline measure-
ments that occurred between 1929 and 1983 may also have in-
troduced bias.
In conclusion, although the possibility of an independent ef-
fect of early menarche on adult adiposity cannot be excluded, our
study combined with other evidence suggests that the effect is at
best small and of minor importance compared with the effects of
premenarcheal BMI that tracks on adult BMI. The present study
suggests that preventing obesity in girls before puberty rather
than focusing on early menarche would result in greater public
health benefits.
The authors’ responsibilities were as follows—MK with DAL, GDS, ME,
MJo, LK-J, JV, LT, MJu, JSAV, and OTR: designed the hypothesis, analyzed
the data, and wrote the manuscript. None of the authors had a personal or
financial conflict of interest.
REFERENCES
1. van Lenthe FJ, Kemper CG, van Mechelen W. Rapid maturation in
adolescence results in greater obesity in adulthood: the Amsterdam
Growth and Health Study. Am J Clin Nutr 1996;64:18 –24.
2. Laitinen J, Power C, Jarvelin MR. Family social class, maternal body
mass index, childhood body mass index, and age at menarche as predic-
tors of adult obesity. Am J Clin Nutr 2001;74:287–94.
3. Parsons TJ, Power C, Logan S, et al. Childhood predictors of adult
obesity: a systematic review. Int J Obes Relat Metab Disord 1999;
23(suppl 8):S1–107.
4. Okamoto K, Horisawa R, Kawamura T, et al. Menstrual and reproduc-
tive factors for subarachnoid hemorrhage risk in women: a case-control
study in Nagoya, Japan. Stroke 2001;32:2841– 4.
5. Cooper GS, Ephross SA, Weinberg CR, et al. Menstrual and repro-
ductive risk factors for ischemic heart disease. Epidemiology 1999;10:
255–9.
6. Remsberg KE, Demerath EW, Schubert CM, et al. Early menarche and
the development of cardiovascular disease risk factors in adolescent
girls: the Fels Longitudinal Study. J Clin Endocrinol Metab 2005;90:
2718 –24.
1882 KIVIMÄKI ET AL
7. Pierce MB, Leon DA. Age at menarche and adult BMI in the Aberdeen
children of the 1950s cohort study. Am J Clin Nutr 2005;82:733–9.
8. Freedman DS, Khan LK, Serdula MK, et al. The relation of menarcheal
age to obesity in childhood and adulthood: the Bogalusa Heart Study.
BMC Pediatr 2003;3:3.
9. Garn SM, LaVelle M, Rosenberg KR, et al. Maturational timing as a
factor in female fatness and obesity. Am J Clin Nutr 1986;43:879 – 83.
10. Must A, Naumova EN, Phillips SM, et al. Childhood overweight and
maturational timing in the development of adult overweight and fatness:
the Newton Girls Study and its follow-up. Pediatrics 2005;116:620 –7.
11. Power C, Lake JK, Cole TJ. Body mass index and height from childhood
to adulthood in the 1958 British born cohort. Am J Clin Nutr 1997;66:
1094 –101.
12. Demerath EW, Li J, Sun SS, et al. Fifty-year trends in serial body mass
index during adolescence in girls: the Fels Longitudinal Study. Am J Clin
Nutr 2004;80:441– 6.
13. Cui R, Iso H, Toyoshima H, et al. Relationships of age at menarche and
menopause, and reproductive year with mortality from cardiovascular
disease in Japanese postmenopausal women: the JACC study. J Epide-
miol 2006;16:177– 84.
14. Colditz GA, Willett WC, Stampfer MJ, et al. Weight as a risk factor for
clinical diabetes in women. Am J Epidemiol 1990;132:501–13.
15. de Ridder CM, Thijssen JH, Bruning PF, et al. Body fat mass, body fat
distribution, and pubertal development: a longitudinal study of physical
and hormonal sexual maturation of girls. J Clin Endocrinol Metab 1992;
75:442– 6.
16. Stark O, Peckham CS, Moynihan C. Weight and age at menarche. Arch
Dis Child 1989;64:383–7.
17. Berkey CS, Gardner JD, Frazier AL, et al. Relation of childhood diet and
body size to menarche and adolescent growth in girls. Am J Epidemiol
2000;152:446 –52.
18. Law M. Dietary fat and adult diseases and the implications for childhood
nutrition: an epidemiologic approach. Am J Clin Nutr 2000;72(suppl):
1291S– 6S.
19. Lee JM, Appugliese D, Kaciroti N, et al. Weight status in young girls and
the onset of puberty. Pediatrics 2007;119:e624 –30.
20. Cooper C, Kuh D, Egger P, et al. Childhood growth and age at menarche.
Br journal of obstetrics and gynaecology 1996;103:814 –7.
21. Wronka I, Pawlinska-Chmara R. Menarcheal age and socio-economic
factors in Poland. Ann Hum Biol 2005;32:630 – 8.
22. Okasha M, McCarron P, McEwen J, et al. Age at menarche: secular
trends and association with adult anthropometric measures. Ann Hum
Biol 2001;28:68 –78.
23. Windham GC, Bottomley C, Birner C, et al. Age at menarche in relation
to maternal use of tobacco, alcohol, coffee, and tea during pregnancy.
Am J Epidemiol 2004;159:862–71.
24. Hediger ML, Scholl TO, Schall JI, et al. One-year changes in weight and
fatness in girls during late adolescence. Pediatrics 1995;96:253– 8.
25. Åkerblom HK, Uhari M, Pesonen E, et al. Cardiovascular Risk in Young
Finns. Ann Med 1991;23:35–9.
26. Raitakari OT, Juonala M, Kähönen M, et al. Cardiovascular risk factors
in childhood and carotid artery intima-media thickness in adulthood: the
Cardiovascular Risk in Young Finns Study. JAMA 2003;290:2277– 83.
27. Bland JM, Altman DG. Statistical methods for assessing agreement
between two methods of clinical measurement. Lancet 1986;1:307–10.
28. StatisticsFinland. Classification of occupations. Helsinki, Finland: Sta-
tistics Finland, 1987.
29. Kivimäki M, Davey Smith G, Juonala M, et al. Socioeconomic position
in childhood and adult cardiovascular risk factors, vascular structure and
function: the Cardiovascular Risk in Young Finns Study. Heart 2006;
92:474 – 80.
30. Kivimäki M, Lawlor DA, Davey Smith G, et al. Early socioeconomic
position and blood pressure in childhood and adulthood: the Cardiovas-
cular Risk in Young Finns Study. Hypertension 2006;47:39 – 44.
31. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concen-
tration of low-density lipoprotein cholesterol in plasma, without use of
the preparative ultracentrifuge. Clin Chem 1972;18:499 –502.
32. Viikari J, Akerblom HK, Nikkari T, et al. Atherosclerosis precursors in
Finnish children and adolescents. IV. Serum lipids in newborns, children
and adolescents Acta Paediatr Scand Suppl 1985;318:103–9.
33. Juonala M, Viikari JS, Hutri-Kahonen N, et al. The 21-year follow-up of
the Cardiovascular Risk in Young Finns Study: risk factor levels, secular
Downloaded from ajcn.nutrition.org by guest on February 8, 2017
trends and east-west difference. J Intern Med 2004;255:457– 68.
34. Viikari J, Rönnemaa T, Seppänen A, et al. Serum lipids and lipoproteins
in children, adolescents and young adults in 1980 –1986. Ann Med
1991;23:53–9.
35. Porkka KV, Raitakari OT, Leino A, et al. Trends in serum lipid levels
during 1980 –1992 in children and young adults. The Cardiovascular
Risk in Young Finns Study. Am J Epidemiol 1997;146:64 –77.
36. Matthews DR, Hosker JP, Rudenski AS, et al. Homeostasis model as-
sessment: insulin resistance and ␤-cell function from fasting plasma
glucose and insulin concentrations in man. Diabetologia 1985;28:412–9.
37. Alberti KG, Zimmet P, Shaw J. The metabolic syndrome—a new world-
wide definition. Lancet 2005;366:1059 – 62.
38. Juonala M, Viikari JS, Laitinen T, et al. Interrelations between brachial
endothelial function and carotid intima-media thickness in young adults:
the Cardiovascular Risk in Young Finns study. Circulation 2004;110:
2918 –23.
39. Cole TJ, Flegal KM, Nicholls D, et al. Body mass index cut offs to define
thinness in children and adolescents: international survey. BMJ 2007;
335:194.
40. Power C, Lake JK, Cole TJ. Measurement and long-term health risks of
child and adolescent fatness. Int J Obes Relat Metab Disord 1997;21:
507–26.