bs_bs_banner

doi: 10.1111/ppe.12162

31

The Association of Maternal Age with Birthweight and

Gestational Age: A Cross-Cohort Comparison
Mara Clara Restrepo-Mndez,a Debbie A. Lawlor,b,c Bernardo L. Horta,a Alicia Matijasevich,a In S. Santos,a
Ana M. B. Menezes,a Fernando C. Barros,d Cesar G. Victoraa
a

Department of Epidemiology, Federal University of Pelotas

Department of Health and Behavior, Catholic University of Pelotas, Pelotas, Brazil

Medical Research Council (MRC) Integrative Epidemiology Unit, University of Bristol, Bristol, UK
c

School of Social and Community Medicine, University of Bristol, Bristol, UK

Abstract
Background: We examined the associations of maternal age with low birthweight (LBW) and preterm birth in four
cohorts from a middle- and a high-income country, where the patterning of maternal age by socio-economic
position (SEP) is likely to differ.
Methods: Population-based birth cohort studies were carried out in the city of Pelotas, Brazil in 1982, 1993, and
2004, and in Avon, UK in 1991 [Avon Longitudinal Study of Parents and Children (ALSPAC)]. Adjustment for
multiple indicators of SEP were applied.
Results: Low SEP was associated with younger age at childbearing in all cohorts, but the magnitudes of these
associations were stronger in ALSPAC. Inverse associations of SEP with LBW and preterm birth were observed in
all cohorts. U-shaped associations were observed between maternal age and odds of LBW in all cohorts. After
adjustment for SEP, increased odds of LBW for young mothers (<20 years) attenuated to the null but remained or
increased for older mothers (35 years). Very young (<16 years) maternal age was also associated with both
outcomes even after full SEP adjustment. SEP adjusted odds ratio of having a LBW infant in women <16 years and
35 years, compared with 2529 years, were 1.48 [95% confidence interval (CI) 1.00, 2.20] and 1.66 [95% CI 1.36,
2.02], respectively. The corresponding results for preterm birth were 1.80 [95% CI 1.23, 2.64)] and 1.38 [95% CI 1.15,
1.67], respectively.
Conclusion: Confounding by SEP explains much of the excess risk of LBW and preterm among babies born to
teenage mothers as a whole, but not for mothers aged <16 or 35 years. Given that the proportion of women
becoming pregnant at <16 years is smaller than for those 35 years, the population burden is greater for older age.
Keywords: pregnancy in adolescence, maternal age, low birthweight infant, preterm birth, socio-economic factors.

Several studies have reported increased risks of low

birthweight (LBW) among offspring of adolescent
mothers19 (generally defined as women <20 years).17
More recently, the concern about adverse perinatal
outcomes has also shifted towards older mothers as
the number of births to women 35 years and older is
increasing in both high-income countries (HIC) and
middle-income countries (MIC).1013 Several mechanisms have been suggested to explain these associations. With respect to adolescent mothers, it has been
Correspondence:
Mara Clara Restrepo-Mndez, Department of Epidemiology,
Federal University of Pelotas, Rua Marechal Deodoro, 1160 3
Piso, Pelotas CEP 96020-220, Brazil.
E-mail: [email protected]

suggested that they are still developing and growing,

and therefore, mother and offspring may compete for
the supply of nutrients.14 At older ages, women are
more likely to have pre-existing, possibly undiagnosed diseases or poor health, including reduced cardiovascular reserve, which could result in poor
placentation and LBW.11,12,15,16 Furthermore, adverse
perinatal outcomes in older mothers might be related
to relative infertility, which could influence the likelihood of preterm births and LBW.16 At both ends of the
age spectrum, the relationship between maternal age
and adverse offspring outcomes may be strongly
confounded by socio-economic position (SEP).17
However, this is likely to be in opposite directions,
with the association between having a baby as a

2014 The Authors. Paediatric and Perinatal Epidemiology Published by John Wiley & Sons Ltd.
Paediatric and Perinatal Epidemiology, 2015, 29, 3140
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and
reproduction in any medium, provided the original work is properly cited.

32

M. C. Restrepo-Mndez et al.

teenager and adverse perinatal outcomes possibly

being explained by low SEP,17 and the generally
higher SEP of older mothers concealing a biological
effect of older maternal age on poor offspring
outcomes.17
Given the difficulty of thoroughly controlling for all
known confounders in observational studies, in particular regarding SEP, novel methods for exploring
the association between maternal age and perinatal
outcomes have been proposed.17,18 In the current
paper, we have used an alternative method to explore
the extent to which associations of maternal age with
LBW and preterm birth might be confounded by SEP.
We have focused on LBW and preterm birth because
of the robust associations of both of these with infant
morbidity and mortality19,20 and with lifelong adverse
health outcomes,21 and also because these are the two
commonest outcomes examined in previous studies
of maternal age with adverse perinatal outcomes.8 We
compared results across four cohorts in which we
anticipate that SEP may relate differently to maternal
age. The four cohorts are three birth cohorts
from different time periods from Pelotas, Brazil (an
MIC) and one pregnancy cohort from the South West
of England (an HIC). Our assumption is that SEP will
relate less strongly to maternal age at birth in the Brazilian compared with the English cohort, because
young age at birth is more common in Brazil and
carries less social stigma. Over the time period
between the first (1982) and last (2004) of the Pelotas
cohorts, Brazil has become more affluent, and with
that, it adopted more Western attitudes and social
behaviours. Therefore, there may be differences in the
relationship of SEP to maternal age across these three
Pelotas cohorts. If our assumptions are true that
there are these differences in confounding structure
between these four cohorts, for which some of our
previous research provides evidence,2224 then any
consistency of association across the cohorts after
adjusting for observed confounders is unlikely to be
due to residual confounding. We are not aware
of any previous studies that have used a cross-cohort
comparison to explore maternal age and LBW or
other perinatal outcomes.

Methods
Pelotas is a city located in Southern Brazil where three
birth cohort studies were conducted in 1982, 1993,
and 2004. Avon Longitudinal Study of Parents and

Children (ALSPAC) is an English birth cohort of participants born in the early 1990s. Some methodological
details of the four cohorts are provided as Supplementary Information S1. More details are given
elsewhere.2529
The present analyses were restricted to singleton
births, whether live or stillborn. Multiple deliveries
were excluded because they are regarded as a potentially higher-risk group for LBW. Consequently, the
sample size was reduced to 98.3%, 98.5%, and 98.0%
from the original Pelotas cohort members, respectively, and 98.0% from the initial ALSPAC cohort
participants.
Birthweight was dichotomised and those under
2500 g were classified as LBW newborns. Births before
the 37th week of pregnancy were classified as preterm.
Maternal age at the time of delivery was initially categorised as follows: <16, 1619, 2024, 2529, 3034
and >34 years. The number of mothers <16 years were
small in all four cohorts [n = 65 (1.1%) in 1982; 108
(2.1%) in 1993; 114 (2.7%) in 2004, in Pelotas and 10
(0.07%), in ALSPAC], and therefore, we conducted
within each cohort analyses with the two lower age
groups combined into one category of <20 years. After
exploring heterogeneity between the cohorts we a
priori decided that if this was small (P > 0.05), we
would pool results across cohorts in order to be able to
specifically look at the very young (<16 years) age
group, which although rare, may be at particularly
high risk and is a group relatively under-researched in
this area.17
Considering that using a single measure
of SEP may not encompass the entirety of the
effect of SEP on health, we decided to adjust for
multiple SEP indicators to avoid as much residual
confounding as possible because of unmeasured
socio-economic circumstances. Three indicators of
SEP family income, maternal education, and paternal education and maternal skin colour/ethnic
origin, marital status, and parity were regarded as
potential confounders. A detailed description of these
covariates is provided as Supplementary Information S1.
In Pelotas, the study protocols were approved by
the Medical Ethics Committee of the Federal University of Pelotas, affiliated with the Brazilian Federal
Medical Council. In ALSPAC, ethical approval was
obtained from the ALSPAC Law and Ethics Committee and the National Health Service Local Research
Ethics Committee.

2014 The Authors. Paediatric and Perinatal Epidemiology Published by John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2015, 29, 3140

The association of maternal age with birthweight and gestational age

Statistical analysis
To compare associations of SEP with maternal age and
birthweight in the four cohorts, index of inequality
were calculated [slope index of inequality (SII) for
continuous variable (maternal age) and relative index
of inequality (RII) for binary outcomes]. Through
these indices, it is possible to relate outcomes with
SEP indicators taking into account the different proportions in each SEP category.30 Each indicator of SEP
is converted into a variable represented by scores
from 0 (lowest SEP) to 1 (highest SEP), with each category corresponding to a score calculated as the midpoint for the proportion of participants in that
category based on the cumulative distribution. The
SII/RII is then obtained by regressing each outcome
measure on these 0 to 1 scores.24 For continuous outcomes, the coefficient represents the mean difference
in the outcome measure between the extremes of SEP
spectra (i.e. the most deprived compared with the
most affluent). For binary outcomes, the results are
the odds ratio (OR) of the outcome between the
extremes. Associations of maternal age with LBW
were analysed using logistic regression. We present
results from two models from each cohort one in
which the association of maternal age with LBW is
examined without adjustment for any potential confounders and one in which we adjust for all potential
confounders.
We tested between-cohort differences for the associations of SEP variables with maternal age and with
LBW, and also for the main association of maternal
age with LBW, using individual participant data. We
combined all data from each of the four cohorts and
added an indicator variable for cohort in analyses
using these pooled data. Heterogeneity in associations
between the four cohorts was explored by including
an interaction term for cohort. As there was no
strong evidence of heterogeneity among the four
cohorts, we assessed the association between maternal
age and LBW based on all participants from the four
cohorts combined and using cohort as a covariate.
This pooled individual participant analyses provided
more precise estimates of associations, and, importantly, provided adequate statistical power for us to
examine whether very young maternal age (<16) was
associated with LBW and preterm birth.
In order to compare our results with most other
published studies, we repeated the analyses adjusting
for maternal education as the only marker of

SEP in the confounder-adjusted models. This is consistent with most published studies assessing these
associations.
For the main analyses presented here, we only
included women in any of the four cohorts who had
complete data on any variables included in the maximally adjusted multivariable model. However, to
assess any potential bias caused by missing data, we
also repeated the unadjusted associations using the
maximal sample available and compared these results
to the unadjusted associations in the main complete
case analyses.
Finally, we repeated all of the above analyses with
preterm birth as the outcome. In contrast to methods
for measuring gestational age, which varied among
the cohorts, birthweight was assessed in a uniform
manner. As gestational age had a high proportion of
missing values (22%) in the 1982 Pelotas cohort, hot
deck multiple imputation was employed,31 using
information from family income, maternal education,
maternal skin colour, number of antenatal care visits,
self-report of high blood pressure during pregnancy,
and parity.
The HosmerLemeshow goodness-of-fit test was
applied to examine adequacy of final models of the
association of maternal age with LBW and preterm
birth. All analyses were performed using Stata,
version 12.0 (StataCorp, College Station, Texas, USA).

Results
The prevalence of LBW tended to be highest among
adolescents in the Pelotas cohorts and among mothers
older than 34 years in ALSPAC. LBW prevalence
decreased in Pelotas from 1982 to 2004 in almost all
maternal age groups. In all cohorts, the prevalence of
preterm birth was highest among adolescents and
among mothers older than 34 years. Preterm birth
was less frequent in ALSPAC than in Pelotas, in
almost all maternal age groups; in Pelotas, preterm
birth prevalence tended to increase over time
(Table 1).
Table S1 shows the SII for maternal age, expressing
the differences between the extremes of the income
and parental education spectra and therefore allowing
a direct comparison of the magnitudes of the associations in the four cohorts. Mean maternal age tended to
be directly associated with SEP in the four cohorts.
The only exception to this pattern was the inverse
association of paternal education with maternal age in

2014 The Authors. Paediatric and Perinatal Epidemiology Published by John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2015, 29, 3140

33

34

M. C. Restrepo-Mndez et al.

Table 1. Prevalence of low birthweight (LBW) and preterm births in the Pelotas cohorts and ALSPAC

Cohort
1982

1993

2004

ALSPAC

Maternal age

LBW
% [95% CI]

<20
2024
2529
3034
>34
All
<20
2024
2529
3034
>34
All
<20
2024
2529
3034
>34
All
<20
2024
2529
3034
>34
All

674
1391
1226
800
485
4576
802
1326
1231
885
517
4761
584
904
737
574
426
3225
173
1121
3081
2442
863
7680

9.9 [7.7, 12.2]

6.5 [5.2, 7.8]
5.1 [3.9, 6.4]
6.0 [4.3, 7.6]
8.0 [5.6, 10.5]
8.2 [7.5, 8.9]
11.6 [9.4, 13.8]
8.5 [7.0, 10.0]
7.1 [5.7, 8.6]
8.4 [6.5, 10.2]
11.4 [8.7, 14.2]
9.3 [8.5, 10.1]
11.1 [8.6, 13.7]
8.5 [6.7, 10.3]
8.0 [6.0, 10.0]
7.8 [5.6, 10.0]
10.6 [7.6, 13.5]
9.8 [8.9, 10.7]
5.2 [1.9, 8.5]
3.6 [2.5, 4.7]
3.1 [2.5, 3.7]
3.2 [2.5, 3.8]
4.1 [2.7, 5.4]
3.3 [2.9, 3.7]

P-valuea
<0.001

0.003

0.16

0.37

Preterm birth
% [95% CI]
8.0 [5.7, 10.4]
4.0 [2.9, 5.2]
4.5 [3.2, 5.8]
4.5 [3.0, 6.1]
7.9 [5.2, 10.5]
5.5 [4.8, 6.2]
13.1 [10.8, 15.5]
10.1 [8.5, 11.8]
9.2 [7.6, 10.9]
10.5 [8.5, 12.6]
13.1 [10.2, 16.0]
10.8 [9.9, 11.7]
18.2 [15.0, 21.3]
13.6 [11.3, 15.8]
12.4 [10.0, 14.7]
12.2 [9.5, 14.9]
13.4 [10.2, 16.7]
13.9 [12.7, 15.0]
5.1 [1.9, 8.4]
5.0 [3.8, 6.3]
3.8 [3.1, 4.4]
3.9 [3.1, 4.6]
4.1 [2.8, 5.4]
4.0 [3.6, 4.5]

P-valuea
0.001

0.03

0.02

0.36

Chi-squared for heterogeneity.

the Pelotas 1982 cohort. Although the direction of

these associations was similar for all four cohorts, the
magnitude of associations tended to be stronger in
ALSPAC than any of the three Pelotas cohorts, especially for maternal and paternal education (P < 0.001
for interaction between SEP variables and study). In
contrast, the associations of SEP indicators with LBW
(Table S2) had similar magnitude across all four
cohorts (P-value for interaction between both income
and maternal education and study = 0.2; and P = 0.8
for interaction between paternal education and
study = 0.8).
Table 2 displays the ORs and 95% confidence interval of the associations between maternal age and LBW
for each individual study. Adjustment for confounders
tended to attenuate the ORs associated with all age
groups except that of mothers 30 years or older, for
which adjustment increased the OR. The reduction in
the risk for adolescent mothers following confounder
adjustment was largest in the Pelotas 2004 and
ALSPAC studies. The increase in risk for older
mothers following adjustment was particularly noticeable in ALSPAC.

The attenuation in the odds of preterm birth in adolescent mothers following confounder adjustment was
also largest in the Pelotas 2004 and ALSPAC studies.
The increase in odds for older mothers following
adjustment was particularly noticeable in the 1982
Pelotas cohort. Further, older mothers had consistently higher odds for preterm birth in comparison
with women aged 2529 years, but the magnitude of
the difference decreased from the 1982 to the 2004
cohort (Table 3).
There were too few adolescent mothers aged <16
years in each individual cohort to reliably estimate the
odds of outcomes compared with the reference group.
Since there was no strong evidence for the interaction
between maternal age categories and LBW by cohort
study (unadjusted and adjusted model had P = 0.8
and P = 0.9, respectively), we combined the cohorts.
Figure 1 shows that with all four studies combined,
there are important increases in the odds of LBW in
those <16 years and >34 years, compared with the reference group of 2529 years (see also Table S3). In
addition, adjustment for confounders tended to
attenuate the ORs associated with all age groups

2014 The Authors. Paediatric and Perinatal Epidemiology Published by John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2015, 29, 3140

The association of maternal age with birthweight and gestational age

35

Table 2. Associations of maternal age with low birthweight (LBW) in the Pelotas cohorts and ALSPAC
Cohort

Maternal age

1982

<20
2024
2529
3034
>34

1993

<20
2024
2529
3034
>34

2004

<20
2024
2529
3034
>34

ALSPAC

<20
2024
2529
3034
>34

Unadjusted OR [95% CI]

Adjusted ORa [95% CI]

Adjusted ORb [95% CI]

P = 0.002
2.04 [1.43, 2.92]
1.30 [0.93, 1.80]
1.00
1.18 [0.80, 1.74]
1.62 [1.07, 2.45]
P = 0.003
1.70 [1.25, 2.31]
1.21 [0.91, 1.62]
1.00
1.19 [0.86, 1.64]
1.67 [1.18, 2.37]
P = 0.17
1.44 [0.99, 2.08]
1.07 [0.75, 1.52]
1.00
0.98 [0.65, 1.46]
1.36 [0.90, 2.04]
P = 0.28
1.93 [1.02, 3.66]
1.17 [0.82, 1.68]
1.00
1.01 [0.75, 1.36]
1.28 [0.87, 1.89]

P = 0.14
1.42 [0.95, 2.13]
1.15 [0.82, 1.61]
1.00
1.21 [0.82, 1.79]
1.65 [1.08, 2.52]
P = 0.02
1.30 [0.93, 1.84]
1.11 [0.82, 1.49]
1.00
1.31 [0.95, 1.81]
1.77 [1.25, 2.52]
P = 0.29
0.94 [0.61, 1.44]
0.91 [0.64, 1.31]
1.00
1.00 [0.66, 1.51]
1.45 [0.95, 2.20]
P = 0.03
0.82 [0.39, 1.75]
0.83 [0.56, 1.23]
1.00
1.29 [0.94, 1.76]
1.75 [1.16, 2.65]

P = 0.08
1.63 [1.10, 2.41]
1.20 [0.86, 1.68]
1.00
1.19 [0.81, 1.76]
1.57 [1.03, 2.39]
P = 0.03
1.36 [0.97, 1.91]
1.12 [0.83, 1.50]
1.00
1.27 [0.92, 1.75]
1.73 [1.22, 2.46]
P = 0.40
1.01 [0.67, 1.53]
0.97 [0.68, 1.39]
1.00
0.99 [0.66, 1.49]
1.43 [0.94, 2.17]
P = 0.22
0.98 [0.46, 2.08]
0.96 [0.66, 1.42]
1.00
1.20 [0.88, 1.64]
1.61 [1.07, 2.43]

Adjusted for confounding factors: family income, maternal education, paternal education, skin colour/ethnic group, parity, and living
with a partner.
b
Adjusted for maternal education, skin colour/ethnic group, parity, and living with a partner.

except that of mothers 30 years or older, for which

adjustment increased the OR. In a similar manner,
associations of maternal age with preterm birth did
not differ between the four studies. Results with
preterm birth as the outcome were generally similar
to those for LBW (Figure 2). However after adjustment for confounders, both adolescent and older
mothers continued to show a strong evidence of
increased odds for preterm birth (see also Table S3).
In comparison with adjusting for all potential SEP
indicators in our studies, when we adjusted only for
confounding by maternal education, the association of
young maternal age with LBW attenuated less and
that of older maternal age with LBW increased less in
the Pelotas 1982 cohort, but for all of the other cohorts,
there was no marked difference between the two
models (Table 2). For preterm birth, adjusting only for
maternal education produced broadly similar findings
to adjusting for all SEP confounders in all four
cohorts.
Analyses with maximum sample showed similar
magnitudes of association between LBW and preterm

birth with maternal age for all four cohorts (Table S4).
In addition, when comparing results with and
without imputation data for gestational age for the
1982 Pelotas cohort, we observed similar magnitudes
of association between preterm birth and maternal
age.

Comment
We found slightly different associations of maternal
age with odds of LBW and preterm birth in three
cohorts from different time periods (early 80s, 90s,
and 2000s) based in Brazil and one from the early 90s
based in England. In the 1982 and 1993 Pelotas
cohorts, maternal age younger than 20 years and older
than 34 years, compared with the reference group of
2529 years, had increased odds of LBW. After adjustment for SEP, the increased odds of LBW for younger
mothers (<20 years) attenuated to the null, and the
higher odds for older mothers increased for the 1993
Pelotas cohort as well as for ALSPAC. This confirms
our hypothesis that the association of young maternal

2014 The Authors. Paediatric and Perinatal Epidemiology Published by John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2015, 29, 3140

36

M. C. Restrepo-Mndez et al.

Table 3. Associations of maternal age with preterm births in the Pelotas cohorts and ALSPAC
Cohort

Maternal age

1982

<20
2024
2529
3034
>34

1993

<20
2024
2529
3034
>34

2004

<20
2024
2529
3034
>34

ALSPAC

<20
2024
2529
3034
>34

Unadjusted OR [95% CI]

Adjusted ORa [95% CI]

Adjusted ORb [95% CI]

P = 0.001
1.75 [1.21, 2.53]
0.90 [0.63, 1.28]
1.00
1.03 [0.69, 1.53]
1.66 [1.09, 2.52]
P = 0.03
1.43 [1.10, 1.86]
1.09 [0.85, 1.40]
1.00
1.14 [0.86, 1.50]
1.46 [1.08, 1.98]
P = 0.006
1.56 [1.20, 2.03]
1.15 [0.89, 1.48]
1.00
1.00 [0.75, 1.34]
1.06 [0.78, 1.45]
P = 0.007
1.37 [0.98, 1.92]
1.30 [1.06, 1.59]
1.00
0.89 [0.73, 1.09]
1.09 [0.84, 1.43]

P = 0.006
1.64 [1.00, 2.69]
0.85 [0.55, 1.31]
1.00
1.01 [0.63, 1.63]
1.81 [1.11, 2.93]
P = 0.08
1.32 [0.96, 1.81]
1.05 [0.80, 1.38]
1.00
1.19 [0.89, 1.60]
1.49 [1.08, 2.07]
P = 0.70
1.24 [0.87, 1.76]
1.02 [0.76, 1.38]
1.00
0.98 [0.70, 1.37]
1.10 [0.76, 1.58]
P = 0.74
1.01 [0.48, 2.13]
1.17 [0.83, 1.64]
1.00
1.14 [0.86, 1.52]
1.26 [0.85, 1.88]

P = 0.003
1.70 [1.05, 2.77]
0.85 [0.55, 1.30]
1.00
1.03 [0.64, 1.66]
1.87 [1.16, 3.01]
P = 0.07
1.35 [0.99, 1.85]
1.06 [0.81, 1.39]
1.00
1.19 [0.89, 1.60]
1.49 [1.08, 2.07]
P = 0.56
1.30 [0.92, 1.84]
1.06 [0.79, 1.42]
1.00
0.98 [0.70, 1.38]
1.11 [0.78, 1.60]
P = 0.22
1.10 [0.52, 2.30]
1.26 [0.90, 1.75]
1.00
1.12 [0.84, 1.48]
1.22 [0.83, 1.81]

Adjusted for confounding factors: family income, maternal education, paternal education, skin colour/ethnic group, parity, and living
with a partner.
b
Adjusted for maternal education, skin colour/ethnic group, parity, and living with a partner.

Figure 1. Combined unadjusted and adjusted odds ratio (OR) of

low birthweight (LBW) by maternal age (including the three
Pelotas cohorts and ALSPAC). P-values for interaction between
associations of age and study with LBW were 0.8 for the unadjusted analyses and 0.9 for the adjusted analyses (i.e. suggesting
that associations of age with LBW were similar in all four
studies).

Figure 2. Combined unadjusted and adjusted odds ratio (OR) of

preterm birth by maternal age (including the three Pelotas
cohorts and ALSPAC). Unadjusted and adjusted P-values for
interaction between maternal age and study were 0.5 and 0.7,
respectively.

2014 The Authors. Paediatric and Perinatal Epidemiology Published by John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2015, 29, 3140

The association of maternal age with birthweight and gestational age

age with LBW was positively and strongly confounded by SEP, whereas for older maternal age, SEP
acts as a negative (masking) confounder that attenuates the association in unadjusted analyses. In a
similar manner, the higher odds of preterm birth for
older mothers increased after adjustment for SEP in
the three Pelotas cohorts.
The comparison of confounding structures in the
four cohorts shows that SEP was positively associated
with maternal age in all cohorts, but the association
was particularly strong in ALSPAC. Therefore, one
would expect more marked changes from the unadjusted to the adjusted ORs in ALSPAC than in Pelotas,
and this was indeed the case. For instance, in
ALSPAC, the unadjusted OR of 1.93 for adolescent
mothers was reduced to 0.82, whereas for older
mothers, the corresponding values were 1.28 and 1.75.
Changes from the unadjusted to the adjusted ORs in
Pelotas were considerably less marked.
Changes in the socio-economic characteristics of the
mothers as well as the reproductive and health care
conditions over time in Pelotas may explain the differences in the association of maternal age with LBW
and preterm births among the Pelotas cohorts.
Improvements in maternal education, household conditions, and presence of consumer goods as well as
increased coverage of antenatal care from 1982 to 2004
may explain the lower risk for LBW in the 2004
Pelotas cohort.32 For instance, adolescents had lower
family income, lower parity, and were less likely to
live with a partner, compared with mothers aged
2029 years, in the three Pelotas cohorts. In 1982 and
1993, adolescents also had lower schooling, but by
2004, this trend had reversed.33 This improvement in
the education of adolescents may explain why the
associations of younger maternal age with LBW and
preterm birth were less marked in the 2004 cohort
than in the 1982 and 1993 cohorts. Furthermore,
smaller birth interval and higher parity13 may explain
the greater risk for LBW among older women in
the 1982 and 1993 Pelotas cohort. In contrast, the
excessive medicalisation of pregnancy and delivery32
especially among more affluent groups of SEP may
be responsible for the greater prevalence for preterm
birth in the 2004 Pelotas cohort and for the absent of
association of older maternal age with LBW and
preterm births.32,34 Because of the improvements of
socio-economic characteristics of the mothers in Brazil
over time, one would expect that findings from the
2004 Pelotas cohort would be more similar to those

from ALSPAC, but this was not the case. Again, the
increase of excessive medicalisation over time in
Brazil may explain the different pattern observed in
the 2004 Pelotas cohort.
To test how effective controlling for maternal education, instead of the full set of SEP measures, was in
reducing residual confounding, we ran analyses for the
four cohorts and compared both sets of estimates. The
magnitudes of odds for LBW and preterm birth were
higher in the model adjusted for maternal education
relative to the full model that also included income and
paternal education for all cohorts. This suggests that, at
least in some previous studies, controlling for maternal
education alone may not have been sufficient to eliminate socio-economic confounding and may explain
why some of the reports in the literature are in disagreement with our present findings.17,35
An additional important finding of our study is that
by combining all four cohorts, we had adequate statistical power to examine whether very young maternal
age (<16 years) was associated with adverse perinatal
outcomes. Very few studies have sufficient statistical
power to precisely estimate the association between
very young maternal age at birth and adverse perinatal outcomes. Our results show that although in
general much of the association of pregnancy aged
20 years or younger with LBW and preterm birth is
spurious and explained by SEP, confounding this
finding masks an important increased risk of adverse
outcomes in women aged 16 years or younger even
after adjustment for SEP. The suggestion that adverse
perinatal outcomes occur in women of young age
because they are still growing and are hence unable to
obtain sufficient nutrients for both their own growth
and development as well as that of the fetus is likely
to be more the case for very young mothers (<16
years) than those who are 1619 years. These findings
are consistent with the small number of other studies
that have been able to examine associations with very
young maternal age (defined as either <16 or
<14 years).1,5,36
The strengths of these analyses include the
population-based samples from four prospective
studies that allowed a comparison between cohorts
from HIC and MIC, and the detailed assessment of
maternal characteristics, which enabled us to control
for three indicators of SEP, as well as other factors
(maternal skin colour, parity) that might confound the
associations of maternal age with LBW and preterm
birth.

2014 The Authors. Paediatric and Perinatal Epidemiology Published by John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2015, 29, 3140

37

38

M. C. Restrepo-Mndez et al.

Mothers who were not included in the main analyses (restricted sample) were more likely to be from
the younger age groups (<20 and 2024 years) in
ALSPAC; therefore, we cannot rule out the possibility
that our findings can be biased towards the null
among younger mothers. However, results were
similar for the unadjusted analyses when we used the
maximum sample compared with the restricted
sample for all four cohorts.
Our findings point to different mechanisms, as well
as different social patterning, being responsible for the
associations of young maternal age with LBW compared with those of older maternal age. They suggest
that very young maternal age is associated with
higher risk of LBW and preterm birth, but that the
apparent increase in risk for mothers aged 1619 years
is explained by their socio-economic, rather than biological conditions. For this group, a focus on socioeconomic inequalities rather than younger maternal
age per se might yield greater population benefit,
whereas for older maternal age, our findings highlight
the need for further research to explore biological
mechanisms underlying these associations. Although
there is considerable research and media coverage
regarding the potential detrimental consequences of
adolescent pregnancy, the possible adverse impact of
delaying a first birth until older ages is much less discussed in the media or health journals. Given our
results, it might be time for public discussion to consider adverse outcomes of older maternal age at birth
in more detail.

Acknowledgements
We are thankful to all of the families who took part in
the Pelotas Birth Cohort Studies (1982, 1993, and 2004)
and the whole Pelotas team.
We are extremely grateful to all of the families who
took part in ALSPAC, the midwives for recruiting
them, and the whole ALSPAC team, which includes
interviewers, computer and laboratory technicians,
clerical workers, research scientists, volunteers, managers, receptionists, and nurses.
This article is based on data from the 1982, 1993,
and 2004 Pelotas birth cohort studies, conducted by
Postgraduate Program in Epidemiology at Universidade Federal de Pelotas. The studies are currently
supported by the Wellcome Trust Initiative entitled
Major Awards for Latin America on Health Consequences of Population Change. Previous phases

of the studies were supported by the International

Development Research Center, The World Health
Organization, Overseas Development Administration,
European Union, National Support Program for
Centers of Excellence (PRONEX), the Brazilian
National Research Council (CNPq), Brazilian Ministry
of Health, and the Child Pastorate of Brazil.
The UK Medical Research Council and Wellcome
Trust (092731), together with the University of Bristol,
provide core support for the ALSPAC study. DALs
contribution to this work was supported by the UK
Medical Research Council (MC_UU_12013/6) and her
NIHR Senior Investigator Award (NF-SI-0611-10196).
The views expressed in this article are those of the
authors and not necessarily of any funders. The
funders had no role in data collection, analysis, or
interpretation of results. The authors declare that they
have no conflict of interest.

References
1 Chen XK, Wen SW, Fleming N, Demissie K, Rhoads GG,
Walker M. Teenage pregnancy and adverse birth outcomes:
a large population based retrospective cohort study.
International Journal of Epidemiology 2007; 36:368373.
2 Haldre K, Rahu K, Karro H, Rahu M. Is a poor pregnancy
outcome related to young maternal age? A study of
teenagers in Estonia during the period of major
socio-economic changes (from 1992 to 2002). European
Journal of Obstetrics, Gynecology, and Reproductive Biology
2007; 131:4551.
3 Schempf AH, Branum AM, Lukacs SL, Schoendorf KC.
Maternal age and parity-associated risks of preterm birth:
differences by race/ethnicity. Paediatric and Perinatal
Epidemiology 2007; 21:3443.
4 Stewart CP, Katz J, Khatry SK, LeClerq SC, Shrestha SR,
West KP Jr, et al. Preterm delivery but not intrauterine
growth retardation is associated with young maternal age
among primiparae in rural Nepal. Maternal and Child
Nutrition 2007; 3:174185.
5 Conde-Agudelo A, Belizan JM, Lammers C.
Maternal-perinatal morbidity and mortality associated with
adolescent pregnancy in Latin America: cross-sectional
study. American Journal of Obstetrics and Gynecology 2005;
192:342349.
6 Gilbert W, Jandial D, Field N, Bigelow P, Danielsen B. Birth
outcomes in teenage pregnancies. The Journal of
Maternal-fetal and Neonatal Medicine 2004; 16:265270.
7 da Silva AA, Simoes VM, Barbieri MA, Bettiol H,
Lamy-Filho F, Coimbra LC, et al. Young maternal age and
preterm birth. Paediatric and Perinatal Epidemiology 2003;
17:332339.
8 Cunnington AJ. Whats so bad about teenage pregnancy?
The Journal of Family Planning and Reproductive Health Care
2001; 27:3641.

2014 The Authors. Paediatric and Perinatal Epidemiology Published by John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2015, 29, 3140

The association of maternal age with birthweight and gestational age

9 Jolly MC, Sebire N, Harris J, Robinson S, Regan L. Obstetric
risks of pregnancy in women less than 18 years old.
Obstetrics and Gynecology 2000; 96:962966.
10 Breart G. Delayed childbearing. European Journal of
Obstetrics, Gynecology, and Reproductive Biology 1997;
75:7173.
11 Heffner LJ. Advanced maternal age how old is too old?
The New England Journal of Medicine 2004;
351:19271929.
12 Balasch J, Gratacos E. Delayed childbearing: effects on
fertility and the outcome of pregnancy. Current Opinion in
Obstetrics and Gynecology 2012; 24:187193.
13 Santos IS, Barros AJ, Matijasevich A, Tomasi E, Medeiros RS,
Domingues MR, et al. Mothers and their pregnancies:
a comparison of three population-based cohorts in
Southern Brazil. Cadernos de Saude Publica 2008;
24 (Suppl. 3):S381S389.
14 Scholl TO, Hediger ML, Schall JI, Khoo CS, Fischer RL.
Maternal growth during pregnancy and the competition for
nutrients. The American Journal of Clinical Nutrition 1994;
60:183188.
15 Stein Z, Susser M. The risks of having children in later life.
Social advantage may make up for biological disadvantage.
BMJ (Clinical Research Ed.) 2000; 320:16811682.
16 Luke B, Brown MB. Elevated risks of pregnancy
complications and adverse outcomes with increasing
maternal age. Human Reproduction 2007; 22:12641272.
17 Lawlor DA, Mortensen L, Andersen AM. Mechanisms
underlying the associations of maternal age with adverse
perinatal outcomes: a sibling study of 264 695 Danish
women and their firstborn offspring. International Journal of
Epidemiology 2011; 40:12051214.
18 Geronimus AT, Korenman S. Maternal youth or family
background? On the health disadvantages of infants with
teenage mothers. American Journal of Epidemiology 1993;
137:213225.
19 McCormick MC. The contribution of low birth weight to
infant mortality and childhood morbidity. The New England
Journal of Medicine 1985; 312:8290.
20 Victora CG, Barros FC, Vaughan JP, Teixeira AM.
Birthweight and infant mortality: a longitudinal study of
5914 Brazilian children. International Journal of Epidemiology
1987; 16:239245.
21 Barker DJ. The developmental origins of chronic adult
disease. Acta Paediatrica. Supplementum 2004; 93:2633.
22 Matijasevich A, Victora CG, Golding J, Barros FC,
Menezes AM, Araujo CL, et al. Socioeconomic position
and overweight among adolescents: data from birth cohort
studies in Brazil and the UK. BMC Public Health 2009;
9:105.
23 Brion MJ, Victora C, Matijasevich A, Horta B, Anselmi L,
Steer C, et al. Maternal smoking and child psychological
problems: disentangling causal and noncausal effects.
Pediatrics 2010; 126:e57e65.
24 Brion MJ, Lawlor DA, Matijasevich A, Horta B, Anselmi L,
Araujo CL, et al. What are the causal effects of breastfeeding
on IQ, obesity and blood pressure? Evidence from
comparing high-income with middle-income cohorts.
International Journal of Epidemiology 2011; 40:670680.

25 Victora CG, Barros FC. Cohort profile: the 1982 Pelotas

(Brazil) birth cohort study. International Journal of
Epidemiology 2006; 35:237242.
26 Victora CG, Hallal PC, Araujo CL, Menezes AM, Wells JC,
Barros FC. Cohort profile: the 1993 Pelotas (Brazil) birth
cohort study. International Journal of Epidemiology 2008;
37:704709.
27 Santos IS, Barros AJ, Matijasevich A, Domingues MR,
Barros FC, Victora CG. Cohort profile: the 2004 Pelotas
(Brazil) birth cohort study. International Journal of
Epidemiology 2011; 40:14611468.
28 Fraser A, Macdonald-Wallis C, Tilling K, Boyd A, Golding J,
Davey Smith G, et al. Cohort profile: the Avon Longitudinal
Study of Parents and Children: ALSPAC mothers cohort.
International Journal of Epidemiology 2013; 42:97110.
29 Boyd A, Golding J, Macleod J, Lawlor DA, Fraser A,
Henderson J, et al. Cohort profile: the children of the 90s
the index offspring of the Avon Longitudinal Study of
Parents and Children. International Journal of Epidemiology
2013; 42:111127.
30 Mackenbach JP, Kunst AE. Measuring the magnitude of
socio-economic inequalities in health: an overview of
available measures illustrated with two examples
from Europe. Social Science and Medicine 1997; 44:
757771.
31 Mander A, Clayton D. Hotdeck imputation. Stata Technical
Bulletin 1999; 51:3234.
32 Barros FC, Victora CG. Maternal-child health in Pelotas,
Rio Grande do Sul State, Brazil: major conclusions from
comparisons of the 1982, 1993, and 2004 birth cohorts.
Cadernos de Saude Publica 2008; 24 (Suppl. 3):S461S467.
33 Restrepo-Mendez MC, Barros AJ, Santos IS, Menezes AM,
Matijasevich A, Barros FC, et al. Childbearing during
adolescence and offspring mortality: findings from three
population-based cohorts in southern Brazil. BMC Public
Health 2011; 11:781.
34 Cano F, Simon C, Remohi J, Pellicer A. Effect of aging on
the female reproductive system: evidence for a role of
uterine senescence in the decline in female fecundity.
Fertility and Sterility 1995; 64:584589.
35 Machado CJ. Impact of maternal age on birth outcomes: a
population-based study of primiparous Brazilian women in
the city of Sao Paulo. Journal of Biosocial Science 2006;
38:523535.
36 Salihu HM, Sharma PP, Ekundayo OJ, Kristensen S,
Badewa AP, Kirby RS, et al. Childhood pregnancy
(1014 years old) and risk of stillbirth in singletons and
twins. The Journal of Pediatrics 2006; 148:522526.

Supporting information
Additional Supporting Information may be found in
the online version of this article at the publishers
web-site:
Table S1. Associations of indicators of socio-economic
position (SEP) with maternal age in the Pelotas
cohorts and ALSPAC.

2014 The Authors. Paediatric and Perinatal Epidemiology Published by John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2015, 29, 3140

39

40

M. C. Restrepo-Mndez et al.

Table S2. Associations of indicators of socio-economic

position (SEP) with low birthweight (LBW) and
preterm birth in the Pelotas cohorts and ALSPAC.
Table S3. Combined unadjusted and adjusted odds
ratio (OR) and 95% CI of low birthweight (LBW) and
preterm birth by maternal age (including the three
Pelotas cohorts and ALSPAC).

Table S4. Associations of maternal age with low

birthweight (LBW) and preterm birth in the Pelotas
and ALSPAC cohorts (maximum sample).
Supplementary Information S1. Description of
covariates regarded as potential confounders in the
association of maternal age with low birthweight and
preterm birth for Pelotas cohorts and ALSPAC.

2014 The Authors. Paediatric and Perinatal Epidemiology Published by John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2015, 29, 3140