Content uploaded by Kathleen M Rasmussen
Author content
All content in this area was uploaded by Kathleen M Rasmussen
Content may be subject to copyright.
.........................................................................................................
European Journal of Public Health, Vol. 23, No. 2, 316–322
The Author 2012. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved.
doi:10.1093/eurpub/cks135 Advance Access published on 29 September 2012
.........................................................................................................
Obesity and early cessation of breastfeeding
in Denmark
Hanne Kronborg
1
, Michael Vaeth
2
, Kathleen M. Rasmussen
3
1 Department of Nursing Science, Institute of Public Health, University of Aarhus, Aarhus, Denmark
2 Department of Biostatistics, Institute of Public Health, University of Aarhus, Aarhus, Denmark
3 Division of Nutritional Sciences, Cornell University, Ithaca, NY, USA
Correspondence: Hanne Kronborg, Department of Nursing Science, Institute of Public Health, University of
Aarhus, Høegh-Guldbergsgade 6A, 8000 Aarhus C, Denmark, tel: +45 8942 4854, fax: +45 8942 5500,
e-mail: hk@nursingscience.au.dk
Background: Obesity is associated with early cessation of breastfeeding. Breastfeeding is multi-factorial, and
several factors contribute to this association. Our aim was to investigate to what extent socio-demographic
and psychosocial characteristics, parity and prenatal conditions could explain the association between high BMI
and early cessation of breastfeeding. Methods: We used data from a randomized trial of 1597 Danish mothers of
singleton infants. Self-reported maternal postnatal weight and height were available from 1375 (86%). High BMI
was defined as body mass index 32 kg/m
2
at 17 days after delivery. Outcome was cessation of exclusive breast-
feeding by 17 weeks post-partum used in proportional hazards regression models. Results: In the unadjusted
analysis, mothers with high post-partum BMI compared with other mothers had a significantly higher rate of
cessation of exclusive breastfeeding, and were more frequently characterized by socio-demographic, psychosocial,
perinatal and behavioural factors known to increase the risk of early breastfeeding cessation. In the adjusted
analyses (n= 1226), the association between BMI and duration of exclusive breastfeeding depended on parity
(P= 0.03). Among primiparae, high BMI was associated with nearly double the risk of early cessation of exclusive
breastfeeding (HR = 1.74, 95% CI 1.21–2.50). Among multiparae, the association disappeared when adjusted for
socio-demographic factors and previous breastfeeding experience (HR = 0.89, 95% CI 0.62–1.28). Conclusion: Parity
and previous breastfeeding experience are important factors to include when studying the association between
BMI and breastfeeding duration. Intervention to extend the duration of lactation among obese mothers should
focus on those with no or little previous breastfeeding experience.
.........................................................................................................
Introduction
Evidence is accumulating that maternal obesity at the time of
conception negatively affects the duration of breastfeeding
among mothers from the United States
1–4
and other countries.
5–10
Breastfeeding is multi-factorial,
11
and several factors may contribute
to this association. The consistency of findings across countries with
and without strong support for breastfeeding points to a biological
explanation.
12
Being heavier before pregnancy is associated with
both a reduced prolactin response to suckling in the first week
post-partum
13
and a delay in the onset of copious milk secretion,
lactogenesis II.
4,14
Mechanical difficulties with proper positioning of
the infant and latching on
15
and other aspects of poor early breast-
feeding behaviour, e.g. nursing frequency and use of formula
supplement,
4
could also be involved.
Among mothers in general, socio-demographic factors (such as
age, parity, education, racial/ethnic group and income) are
well-known determinants of both the choice to breastfeed and the
duration of breastfeeding.
16
Parity strongly influences the early
lactation success.
4
Psychosocial factors are also known determinants
of the choice to breastfeed
17,18
as well as the duration of breastfeed-
ing.
11,19–21
In western societies, obesity seems associated with low
self-confidence, which suggests that socio-demographic and psycho-
social factors may confound the association between BMI and
breastfeeding duration.
22
In a small study, Hilson et al.
14
found
that the association between maternal pre-pregnancy obesity and
early discontinuation of breastfeeding was partially explained by
psychosocial factors. Krause et al.
23
pointed to socio-demographic
factors and, among those, parity as additional correlates of
importance for the breastfeeding process among overweight and
obese mothers. No study to date includes all relevant factors.
Thus, it remains unknown whether obesity in pre-pregnancy is a
significant determinant of the duration of breastfeeding among
both primiparae and multiparae independent of the mothers’
previous breastfeeding experience, socio-demographic or psychoso-
cial characteristics.
The present study was based on data from 1375 mothers in a
cluster-randomized, community-based trial in Denmark.
24
Our
aims were to investigate to what extent breastfeeding mother’s
socio-demographic and psychosocial characteristics, parity,
previous breastfeeding experience and prenatal conditions could
explain the association between high BMI and early cessation of
exclusive breastfeeding. In addition, we aimed to examine whether
any of these factors modified the association of BMI with the
duration of breastfeeding.
Methods
Setting, design and participants
The research reported here used data from a cluster-randomized,
community-based trial designed to compare a group of women in
western Denmark who received an intervention to prolong breast-
feeding with a comparison group who received usual care from the
health visitors. The mothers were recruited at the health visitors’ first
visit, which occurred over a 6-month period that began in February
2004. All new mothers who lived in the study region, gave birth to a
single infant with a gestational age of 37 full weeks and started
breastfeeding were invited to participate. Exclusion criteria were as
follows: mothers with an ethnic background other than Danish,
316 European Journal of Public Health
at Cornell University Library on April 11, 2013http://eurpub.oxfordjournals.org/Downloaded from
preterm delivery or twin birth. Mothers were only eligible for the
present study if they were still breastfeeding at the time of the first
home visit by the health visitor. Further details on the randomized
trial have been reported elsewhere.
24
Data collection, questionnaire
Briefly, at the health visitors’ first visit, the mothers received a
self-administered questionnaire coded with study identification
number together with a reply envelope. The health visitor started
a registration form for every mother who fulfilled the inclusion
criteria and followed those who agreed to participate for 6 months
(26 weeks). The health visitors completed the registration form with
follow-up data about the duration of exclusive breastfeeding. In the
participating municipalities, 2186 mothers gave birth during the
recruitment period. Of the 1760 mothers who fulfilled the
inclusion criteria, 1597 agreed to participate and 1442 (90%)
returned the questionnaire, on average, 17 days after birth.
The self-administered questionnaire was reviewed by four experts
in the subject area and, after adjustment, was subsequently tested for
content and face validity in three rounds with a group of 24 mothers.
It covered 44 items believed to be associated with the duration of
breastfeeding, including socio-demographic characteristics, maternal
post-partum weight, previous breastfeeding experience, perinatal
conditions, breastfeeding behaviour and questions about infant
birth weight and pacifier use.
Variables
The outcome variable was the duration of exclusive breastfeeding
measured in weeks. Exclusive breastfeeding was defined as a child
being fed only the mother’s milk.
25
Early cessation was of particular
interest, so durations of longer than 17 weeks were censored at 17
weeks.
The mothers’ post-partum BMI was used in the analysis, as in-
formation on pre-pregnancy weight was not reported. Self-reported
data on height, post-partum weight and parity were obtained from
the mothers’ questionnaire. BMI was calculated as the post-partum
weight in kilograms divided by the height in metres squared. To
account for the fact that mothers who are 2–3 weeks post-partum
weigh more than other non-pregnant women,
26
we increased the
WHO cutpoints
27
by two BMI units before categorizing
the mothers’ post-partum weight-for-height values. This resulted
in the following three categories: low BMI (BMI < 27 kg/m
2
),
moderate BMI (27 BMI < 32 kg/m
2
) and high BMI (32 BMI
kg/m
2
).
Socio-demographic variables
Maternal age and schooling were measured in years and categorized:
15–24, 25–32, 33–46, and at most 10 years and >10 years, respect-
ively. Smoking referred to current smoking and was categorized in
‘yes’ or ‘no’.
Breastfeeding history
Parity was categorized as primiparous or multiparous. Previous
breastfeeding experience was measured in weeks and categorized
in none for primiparae and 0–5, 5–17 or >17 weeks for multiparae.
Psychosocial variables
Self-efficacy and confidence were both measured on a 5-point Likert
scale. Self-efficacy was categorized in high (1–2), moderate or low
(3–5), and confidence in feeling confident (1–3) or not feeling
confident (4–5). Knowledge was measured by a 10-item
knowledge test, and correct answers according to evidence-based
knowledge were categorized in high knowledge (7–10 correct
answers) or low knowledge (0–6 correct answers). Perinatal
variables included formula feeding within the first 5 days after
birth, and days from birth until onset of copious milk secretion.
Feeding formula within the first 5 days after birth was categorized
in ‘yes’ or ‘no’. Onset of lactation (lactogenesis II) was measured in
days and categorized in 0–3 days or 4+ days.
Breastfeeding behaviour
Use of nipple shield was categorized in ‘yes’ (always, now and then)
or ‘no’ (never). Use of pacifier was categorized in ‘yes’ or ‘no’.
Breastfeeding frequency was measured as number of breast
feedings during a typical 24-hour period and categorized as <7 or
7 day and night. Mothers’ perception of breastfeeding problems
was categorized as ‘yes’ or ‘no’.
Statistical methods
The association between post-partum BMI and socio-demographic
factors, breastfeeding history, psychosocial factors, perinatal condi-
tions and breastfeeding behaviour was assessed by chi-squared tests.
Cox regression analysis was used to estimate HRs to compare breast-
feeding cessation rate within categories of each of these factors. The
influence was described by a HR (95% CI) giving the cessation rate
of exclusive breastfeeding for mothers in a given category relative to
the cessation rate of exclusive breastfeeding for mothers in the
reference category. The duration of exclusive breastfeeding was
described by survival curves using the Kaplan–Meier method.
Mothers who moved out of the area were censored at the time of
change of address.
Cox regression analysis was further used to evaluate the influence
of BMI on the cessation of exclusive breastfeeding stratified on
parity. Initially, an estimate of the crude effect of post-partum
weight-for-height was obtained. Next, the estimate was adjusted
stepwise for the effect of the confounding factors identified in the
initial analysis. Finally, we included interaction terms to see whether
the association between BMI and breastfeeding cessation was
modified by any of the confounding factors. All variables were
included as covariates except for treatment allocation and previous
breastfeeding experience, which were included as stratifying
variables. Treatment allocation was included in all analyses.
The hypothesis of no-effect modification was assessed by a Wald’s
test. Cox regression model assumes proportional cessation rates, and
a test based on the so-called Schoenfeld residuals was used to assess
the validity of this assumption. Multiple imputation was used to
evaluate the effect of the reduction in sample size due to missing
values in the confounding factors. Finally, the main analyses were
repeated using robust standard errors to investigate whether
adjustment for clustering had any impact on the results. Stata
version 12 was used for data management and statistical analyses.
28
Ethics
This study was approved by the Science Ethics Committee for the
Counties of Ringkjobing, Ribe and Sonderjylland and the Danish
Data Protection Agency.
Results
The 1375 mothers who had complete information on breastfeeding
duration and post-partum BMI formed the study population. The
67 mothers with no information on BMI had a significantly higher
cessation rate of exclusive breastfeeding than the mothers with in-
formation on BMI (data not shown). In the study population, 63%
had low BMI, 25% moderate BMI and 12% high BMI (table 1). The
high-BMI mothers were significantly younger and had shorter
duration of schooling than other mothers at baseline. They had a
later onset of milk secretion, expressed less self-efficacy with respect
to breastfeeding, breastfed less frequently and had more frequently
early breastfeeding problems than other mothers. Their babies
received more formula within the first days after birth (table 1).
Obesity and early cessation of breastfeeding 317
at Cornell University Library on April 11, 2013http://eurpub.oxfordjournals.org/Downloaded from
The previous breastfeeding experiences among high-BMI multiparae
were, on average, 15 weeks (SD 10.9 weeks), compared with 18
weeks (SD 9.1 weeks) among the low-BMI mothers and 16 weeks
(SD 10.6 weeks) among the moderate-BMI mothers. When looking
at the entire study population, significant associations with breast-
feeding duration were seen for all characteristics within
socio-demographic and psychosocial factors, perinatal conditions
and breastfeeding behaviour except for smoking habits (table 1).
Table 1 Characteristics of the study population: Distribution of post-pregnancy BMI and association with breastfeeding
duration (HRs) in categories of each variable
Characteristics BMI Breastfeeding duration
Number of women (n) Low Moderate High
871 344 160 1375
n(%) n(%) n(%) Pvalue HR (95% CI)
Socio-demographic factors
Age in years
15–24 83 (60) 38 (27) 18 (13) 1
25–32 546 (61) 226 (25) 118 (13) 0.7 (0.56–0.88)
33–46 239 (71) 74 (22) 23 (7) 0.006 0.4 (0.30–0.53)
Duration of schooling
7–10 years 226 (56) 109 (27) 68 (17) 1
>10 years 640 (67) 230 (24) 90 (9) <0.001 0.54 (0.46–0.64)
Smoking habits
Non smoker 720 (63) 287 (25) 132 (12) 1
Smoker 151 (64) 57 (24) 28 (12) 0.945 1.81 (1.51–2.17)
Parity
Primiparous 336 (61) 144 (26) 67 (12) 1
Multiparous 535 (65) 200 (24) 93 (11) 0.501 0.74 (0.64–0.87)
Breastfeeding history
Previous breastfeeding experience
None, primipara 336 (61) 144 (26) 67 (12) 1
0–5 weeks 60 (51) 31(27) 26 (22) 2.95 (2.34–3.71)
6–17 weeks 196 (64) 80 (26) 31 (10) 0.93 (0.77–1.12)
>17 weeks 278 (69) 88 (22) 36 (9) 0.001 0.36 (0.29–0.45)
Psychosocial factors
Intention
0–6 months 523 (64) 209 (25) 90 (11) 1
>6 months 283 (66) 100 (23) 47 (11) 0.673 0.43 (0.35–0.53)
Self efficacy
High 661 (66) 240 (24) 102 (10) 1
Medium to low 163 (58) 78 (28) 42 (15) 0.019 3.17 (2.67–3.76)
Confidence in ability to produce milk
Feeling confident 565 (65) 212 (24) 95 (11) 1
Not feeling confident 286 (61) 122 (26) 61 (13) 0.322 2.14 (1.82–2.50)
Knowledge
0–6 right answers 156 (61) 66 (26) 35 (14) 1
7–10 right answers 715 (64) 278 (25) 125 (11) 0.474 0.84 (0.69–1.01)
Perinatal conditions
Hospitalization
Hospitalized 523 (62) 215 (25) 108 (13) 1
Ambulant 318 (66) 116 (24) 47 (10) 0.178 0.67 (0.57–0.80)
Formula supplement within 5 days
Babies not given formula 617 (67) 213 (23) 87 (10) 1
Babies given formula 235 (55) 127 (29) 67 (16) <0.001 1.75 (1.50–2.05)
Onset of lactation
0–3 days 645 (66) 232(24) 99 (10) 1
>3 days 216 (57) 104 (28) 57 (15) 0.004 1.69 (1.44–1.98)
Breastfeeding behaviour
Use of nipple shield
Yes 115 (59) 60 (31) 21 (11) 1
No 710 (65) 259 (24) 119 (11) 0.119 0.66 (0.54–0.81)
Pacifier use
Yes 579 (62) 241 (26) 114 (12) 1
No 281 (66) 101 (24) 46 (11) 0.43 0.61 (0.51–0.73)
Nursing frequency
<7 day and night 128 (57) 63 (28) 35 (15) 1
7 day and night 687 (65) 257 (25) 105 (10) 0.016 0.76 (0.62–0.93)
Early problems breastfeeding
Yes 328 (59) 146 (26) 78 (14) 1 (0.48–0.65)
No 538 (66) 197 (24) 81 (10) 0.018 0.56
BMI
Low 1
Moderate 1.25 (1.04–1.49)
High 1.68 (1.34–2.10)
Low BMI (<27.0 kg/m
2
), moderate BMI (between 27.0 and 32.0 kg/m
2
), high BMI (32.0 kg/m
2
).
Figures are numbers (percentage) of observations, missing values excluded.
318 European Journal of Public Health
at Cornell University Library on April 11, 2013http://eurpub.oxfordjournals.org/Downloaded from
This suggests that all four groups of factors included variables that
could confound the association between BMI with breastfeeding
duration.
All mothers included in the study started breastfeeding after birth.
The duration of exclusive breastfeeding was significantly shorter for
mothers with high BMI than for mothers with moderate and low
BMI values (P= 0.002) (figure 1). At 4 months (17 weeks) after
delivery, 74 mothers (46%) in high-BMI group were still exclusively
breastfeeding compared with 190 (55%) in the moderate-BMI group
and 554 (64%) in the low-BMI group.
In the unadjusted analyses, mothers with high and moderate BMI
had a significantly higher cessation rate of exclusive breastfeeding
during the first 17 weeks after birth (HR 1.67, 95% CI 1.34–2.09, and
HR 1.24, 95% CI 1.04–1.49, respectively) relative to mothers with
low BMI. Among both primiparous and multiparous mothers with
high BMI, the cessation rate was significantly increased (HR 1.77,
95% CI 1.28–2.46, and HR 1.59, 95% CI 1.17–2.15, respectively).
The adjusted Cox regression analyses were based on the 1226
mothers with information on all confounding factors included in
the analysis (table 2). Adjustment for socio-demographic factors
(mothers’ age and duration of schooling) reduced the cessation
rate in the high BMI group among primiparae a little. The minor
change in the estimate for high BMI in primiparous women when
adjusting for previous breastfeeding experience reflects that this
variable was not included as a covariate, but as a stratifying factor.
Adding adjustment for psychosocial factors (self-efficacy) and
perinatal conditions (formula supplement within 5 days and
timing of the onset of lactation) did not change the HRs. among
primiparae, HRs in the unadjusted and adjusted analyses were
similar, and the high-BMI mothers in this group had a significantly
higher risk of early breastfeeding cessation. Among multiparae,
adjustment for first socio-demographic factors and then previous
breastfeeding experience led to insignificant HRs for high-BMI
mothers. The dependency on parity of the association between
BMI and duration of exclusive breastfeeding differed significantly
(P= 0.03) between primiparous and multiparous mothers in the
final adjusted analysis. Additional adjustment for factors
describing breastfeeding behaviour (nursing frequency and early
problems; n= 1203) did not influence HRs for early breastfeeding
cessation for either high BMI primiparae or multiparae (HR 1.68,
95% CI 1.16–2.43, and HR 0.91, 95% CI 0.63–1.31, respectively).
The sensitivity of the results reported in table 2 was evaluated by
multiple imputation of missing values of self-efficacy, formula
supplement and onset of lactation. A completed sample size of
1330 was obtained. Estimates based on 20 imputations were
similar to those seen in table 2 (results not shown). Finally, a
series of analyses investigated the consequences of a correction for
the cluster randomization used in the trial. The robust standard
errors used in these analyses were very similar, but slightly smaller
(5% on average), to those used to derive the confidence intervals
shown in table 2. Thus, in the final analysis (adjusted 4 in table 2),
the interaction between the effect of parity and BMI on duration of
exclusive breastfeeding was here strongly significant (P= 0.002).
Potential effect modification of each of the confounding factors
included in the analyses was also investigated. For all variables except
onset of lactation, no effect modification was identified. For onset of
lactation, a significant interaction with the effect of BMI was found
(P= 0.04). Further analysis identified a complex second-order inter-
action with parity and BMI. Compared with primiparae of low BMI,
those of high BMI had a higher hazard of early cessation of breast-
feeding at either 0–3 or 4+ days. Compared with multiparae of low
BMI, those of high BMI did not have a higher hazard of cessation of
breastfeeding, but mothers of low BMI had a higher risk of cessation
of breastfeeding at 4+ days (figure 2).
0.00
0.25
0.50
0.75
1.00
Proportion
0 4 8 12 16
Weeks
Low BMI
Moderate BMI
High BMI
Breastfeeding duration according to BMI groups
Figure 1 Proportion of exclusive breastfeeding according to
post-partum BMI as a function of the age of the child. Post-partum
BMI is categorized as low BMI (<27.0 kg/m
2
), moderate BMI
(between 27.0 and 32.0 kg/m
2
) or high BMI (32.0 kg/m
2
)
Table 2 Unadjusted and adjusted HRs for breast feeding cessation before week 17 by post-pregnancy BMI categories for all women and
stratified by parity, N= 1226
BMI Unadjusted Adjusted 1
a
Adjusted 2
b
Adjusted 3
c
Adjusted 4
d
HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI)
Primiparous
LowBMI11111
Moderate BMI 1.25 (0.95–1.66) 1.22 (0.92–1.62) 1.22 (0.92–1.62) 1.22 (0.92–1.62) 1.14 (0.86–1.51)
High BMI 2.01 (1.41–2.86) 1.69 (1.18–2.41) 1.71 (1.19–2.45) 1.86 (1.30–2.67) 1.74 (1.21–2.50)
Multiparous
LowBMI11111
Moderate BMI 1.20 (0.92–1.56) 1.10 (0.85–1.44) 0.96 (0.74–1.26) 0.92 (0.71–1.21) 0.92 (0.71–1.21)
High BMI 1.48 (1.05–2.09) 1.24 (0.88–1.76) 1.01 (0.70–1.44) 0.99 (0.69–1.41) 0.89 (0.62–1.28)
All women
LowBMI11111
Moderate BMI 1.23 (1.02–1.49) 1.16 (0.95–1.40) 1.08 (0.89–1.31) 1.05 (0.87–1.28) 1.02 (0.84–1.24)
High BMI 1.70 (1.32–2.17) 1.42 (1.10–1.83) 1.29 (1.00–1.66) 1.32 (1.02–1.71) 1.21 (0.93–1.57)
BMI estimates adjusted for:
a: Mothers’ age and duration of schooling
b: Previous breastfeeding experience
c: Self-efficacy
d: Formula supplement within 5 days and onset of lactation
Randomization group was included as a stratifying factor in all these analyses.
Obesity and early cessation of breastfeeding 319
at Cornell University Library on April 11, 2013http://eurpub.oxfordjournals.org/Downloaded from
Discussion
In this study, mothers with a high post-partum BMI more frequently
had socio-demographic and psychosocial factors as well as perinatal
and behavioural conditions known to increase the risk of early
breastfeeding cessation than mothers with normal post-partum
BMI. As expected, high-BMI mothers had a significantly higher
rate of cessation of exclusive breastfeeding. In the adjusted
analyses, however, we identified that the association between BMI
and duration of exclusive breastfeeding depended on parity. Among
primiparae, the high-BMI group had a significant higher risk of early
breastfeeding cessation. Among multiparae, in contrast, adjustment
for socio-demographic factors and previous breastfeeding experience
removed the excess risk for early cessation of breastfeeding among
high-BMI mothers. These findings confirm the excess risk among
primiparous high-BMI women noted by others
4,23
and identify mul-
tiparous high-BMI women without prior breastfeeding as individ-
uals with excess risk of early cessation of breastfeeding.
Associations between obesity and breastfeeding duration have
been observed in several other investigations.
6,22,29
This study dem-
onstrates the influence of the socio-demographic variables, which
was also considered by Krause et al.
23
and Amir and Donath,
22
who concluded that obese mothers are more likely to belong to
social groups who are less likely to breastfeed. Socio-demographic
factors can point to groups of mothers who may need extra support
breastfeeding, but otherwise these factors are difficult to influence
and are less valuable to health professionals as potential targets for
intervention.
Our results showed that the association between high BMI and
duration of exclusive breastfeeding depended on parity. These
findings demonstrate the importance of parity and previous breast-
feeding experience in breastfeeding obesity research and support
earlier findings from Dewey et al.,
4
who found that primiparity
was associated with risk factors for suboptimal breastfeeding, and
Krause et al.,
23
who found parity was related to lactation score.
Previous experience with breastfeeding was crucial for multiparous
mothers. This may indicate that heavier women who have breastfed
previously have overcome the biological factors associated with
shorter duration of breastfeeding. Research in experimental
animals shows that maternal obesity is associated with abnormal
development of the mammary gland, which delays the onset of
copious milk secretion.
30,31
However, once this development is
completed, lactation can proceed normally. Inasmuch as previous
behaviour usually is a strong determinant of current behaviour,
mothers who have had positive breastfeeding experiences may be
more likely to repeat this behaviour.
21,32
Short and perhaps
negative experiences with breastfeeding may reduce a woman’s
interest in breastfeeding or her confidence in her ability to succeed
the next time.
33,34
As a result, she may make less effort to overcome
the challenges of breastfeeding associated with having a high BMI.
These findings suggest that previous breastfeeding experience is also
an important factor to consider when studying the association
between BMI and breastfeeding duration. Interventions to
improve the duration of lactation among heavier mothers should
focus on both primiparous mothers and multiparous mothers with
no or short previous breastfeeding experience.
Adjustment for self-efficacy had only a modest impact on the
association between BMI and duration of exclusive breastfeeding.
This is partly supported by a newly published study by Metha
et al.,
29
who found that psychological factors did not mediate the
association between BMI and breastfeeding duration. In contrast,
Hilson et al.
14
found that knowledge of breastfeeding modified the
relationship between higher BMI and breastfeeding duration.
Additional analyses of the present data showed that psychosocial
characteristics were independently associated with the duration of
exclusive breastfeeding among these Danish mothers (results not
shown). Future research is needed to clarify the role of psychosocial
factors in the association between high BMI and breastfeeding
duration.
Among primiparous women with high BMI, the duration of
breastfeeding was modified by the timing of the onset of copious
milk secretion. This finding is in accord with the results of
Nommsen-Rivers et al.,
35
who found that maternal obesity was
related to delayed onset of lactation among first-time mothers,
and Dewey et al.,
4
who found that mothers with high BMI were
more likely to have delayed onset of lactation. Among the factors
that could contribute to delayed onset of milk secretion is a
reduction in the prolactin response to suckling in the first week
observed among heavier women post-partum by Rasmussen and
Kjolhede.
13
These results substantiate the conclusion from Dewey
et al.
4
that primiparity or lack of previous breastfeeding experience
are risk factors for delayed onset of lactation.
The maternal height and weight were both self-reported, so mis-
classification of mothers into the appropriate post-partum BMI
categories is possible. However, only a modest misclassification in
self-reported BMI in a contemporary population of Danish mothers
was observed in a previous study,
36
so it seems unlikely that mis-
classification could have materially affected the results.
Unfortunately, information on pre-pregnancy BMI was not
available for this sample. Instead the information on weight and
height was obtained, on average, 17 days post-partum, so a modifi-
cation of the usual cutpoints was necessary.
26
This procedure led to a
high-BMI group consisting of 11% of the mothers. This is
comparable with the proportion of women with pre-pregnancy
BMI 30 kg/m
2
found in Denmark of that time.
36
It is possible
that the heaviest mothers were over-represented among the
mothers who fulfilled the inclusion criteria but then did not enrol
in the study. It is also possible that the heaviest mothers are under-
represented among mothers from whom we have data on weight.
Consequently, the strength of the observed association may have
been underestimated.
The large number of subjects and the availability of information
about psychosocial and a wide range of other characteristics of the
women as well as their breastfeeding behaviour are strengths of the
present study. Mothers who entered this trial came from both rural
and urban areas in 11 municipalities of western Denmark, and the
study population represented approximately 75% of all births in
these municipalities. Because of missing values in the confounding
factors, the study population in the adjusted analyses was reduced to
1226 mothers (89%), but the crude association between BMI and
duration of breastfeeding was similar in this subgroup.
This study was performed in Denmark in a Caucasian population
where the proportion of mothers who breastfeed is high (58%
Figure 2 HRs of exclusive breastfeeding according to onset of
lactation stratified on post-partum BMI among primiparous and
multiparous mothers. BMI is categorized as low BMI (<27.0 kg/m
2
),
moderate BMI (between 27.0 and 32.0 kg/m
2
) or high BMI
(32.0 kg/m
2
)
320 European Journal of Public Health
at Cornell University Library on April 11, 2013http://eurpub.oxfordjournals.org/Downloaded from
exclusive breastfeeding until 17 weeks post-partum
21
) compared
with other western countries. The mothers who were recruited
into this study delivered in hospitals that had adopted the
Baby-Friendly Hospital Initiative. As a result, the subjects had
excellent support for the decision to breastfeed and also for
initiation of breastfeeding immediately after delivery. At the time
of the study, primiparae usually stayed in the hospital for 4 days,
but multiparae stayed for less than a full day. Moreover, Scandinavia
is known for its high social support for breastfeeding, including a
24-week paid maternity leave in Denmark. These cultural aspects
may mean that the associations with the duration of breastfeeding
would be less than might be found in other countries where
hospital-based support for breastfeeding and social support for
breastfeeding in general are much lower.
This study contributes to on-going research of the association
between obesity and breastfeeding duration by showing that parity
and previous breastfeeding experience play a significant role. Our
results indicate that primiparae as well as multiparae without
previous breastfeeding experience are at the highest risk of early
cessation of exclusive breastfeeding. As a result, these subgroups of
women with a high post-partum BMI should be the primary targets
for interventions. Unfortunately, successful interventions to improve
the duration of breastfeeding among high BMI mothers have yet to
be developed and are urgently needed.
Acknowledgements
We are indebted to the mothers who participated in the study.
Trial Registration: Clinical Trial.gov (ClinicalTrials.govIdentifier:
00145834).
Funding
This work was supported by the Danish Health Insurance
Foundation, the Lundbeck Foundation and the Counties of Ribe
and Ringkjobing in Denmark.
Conflicts of interest: None declared.
Key points
Parity and previous breastfeeding experience play a signifi-
cant role in the association between high BMI and breast-
feeding duration.
Among high BMI mothers, primiparae and multiparae
without previous breastfeeding experience are at the
highest risk of early cessation of exclusive breastfeeding.
Interventions to improve the duration of lactation among
high BMI mothers should focus on both primiparous
mothers and multiparous mothers with no or short
previous breastfeeding experience.
Reference
1 Hilson JA, Rasmussen KM, Kjolhede CL. Maternal obesity and breastfeeding success
in a rural population of white women. Am J Clin Nutr 1997;66:1371–8.
2 Kugyelka JG, Rasmussen KM, Frongillo EA Jr. Maternal obesity negatively affects
breastfeeding success among Hispanic but not Black women. J Nutr 2004;134:
1746–53.
3 Li R, Jewell S, Grummer-Strawn LM. Maternal obesity and breast-feeding practices.
Am J Clin Nutr 2003;77:931–6.
4 Dewey KG, Nommsen-Rivers LA, Heinig MJ, Cohen RJ. Risk factors for suboptimal
infant breastfeeding behavior, delayed onset of lactation, and excess neonatal weight
loss. Pediatrics 2003;112:607–19.
5 Baker JL, Michaelsen KF, Sørensen TIA, Rasmussen KM. High prepregnant body
mass index is associated with early termination of full and any breastfeeding among
Danish women. Am J Clin Nutr 2007;86:404–11.
6 Donath SM, Amir LH. Does maternal obesity adversely affect breastfeeding
initiation and duration? J Paediatr Child Health 2000;36:482–6.
7 Oddy WH, Li J, Landsborough L, et al. The association of maternal overweight and
obesity with breastfeeding duration. J Pediatr 2006;149:185–91.
8 Forster DA, McLachlan HL, Lumley J. Factors associated with breastfeeding at six
months postpartum in a group of Australian women. Int Breastfeed J 2006;1:18.
9 Donath SM, Amir LH. Maternal obesity and initiation and duration of breast-
feeding: data from the longitudinal study of Australian children. Mat Child Nutr
2008;4:163–70.
10 Mok E, Multon C, Piguel L, et al. Decreased full breastfeeding, altered practices,
perceptions, and infant weight change of prepregnant obese women: a need for extra
support. Pediatrics 2008;121:e1319–24.
11 Scott JA, Binns CW, Oddy WH, Graham KI. Predictors of breastfeeding duration:
evidence from a cohort study. Pediatrics 2006;117:646–55.
12 Rasmussen KM. Association of maternal obesity before conception with poor
lactational performance. Annu Rev Nutr 2007;27:103–21.
13 Rasmussen KM, Kjolhede CL. Prepregnant overweight and obesity diminish the
prolactin response to suckling in the first week postpartum. Pediatrics 2004;113:
e465–71.
14 Hilson JA, Rasmussen KM, Kjolhede CL. High prepregnant body mass index is
associated with poor lactation outcomes among white, rural women independent of
psychosocial and demographic correlates. J Hum Lact 2004;20:18–29.
15 Lovelady CA. Is maternal obesity a cause of poor lactation performance? Nutr Rev
2005;63:352–5.
16 Institute of Medicine (Subcommittee on Nutrition During Lactation,
Committee on Nutritional Status During Pregnancy and Lactation, Food and
Nutrition Board). Nutrition during lactation. Washington, DC: National Academy
Press, 1991.
17 Losch M, Dungy CI, Russel D, Dusdieker LB. Impact of attitudes on maternal
decisions regarding infant feeding. J Pediatr 1995;126:507–14.
18 Barnes J, Stein A, Smith T, Pollock JI. ALSPAC Study Team. Extreme attitudes to
body shape, social and psychological factors and a reluctance to breast feed. . J Royal
Soc Med 1997;90:551–9.
19 O’Campo P, Faden RR, Gielen AC, Wang MC. Prenatal factors associated with
breastfeeding duration: recommendations for prenatal interventions. Birth 1992;19:
195–201.
20 Kools EJ, Thijs C, Kester ADM, de Vries H. The motivational determinants of
breast-feeding: predictors for the continuation of breast-feeding. Prev Med 2006;43:
394–401.
21 Kronborg H, Væth M. The influence of psychosocial factors on the duration of
breastfeeding. Scand J Public Health 2004;32:210–6.
22 Amir LH, Donath S. A systematic review of maternal obesity and breastfeeding
intention, initiation and duration. BMC Pregnancy Childbirth 2007;7:9.
23 Krause KM, Lovelady CA, Ostbye T. Predictors of breastfeeding in overweight and
obese women: data from Active Mothers Postpartum (AMP). Matern Child Health J
2011;15:367–75.
24 Kronborg H, Væth M, Olsen J, et al. Effect of early postnatal breastfeeding
support: a cluster-randomized community based trial. Acta Paediatr 2007;96:
1096–70.
25 World Health Organisation. Indicators for assessing breastfeeding
practices. WHO/CDD/SER/91.14. Geneva: Division of Child Health and
Development, 1991.
26 Butte NF, Ellis KJ, Wong WW, et al. Composition of gestational weight gain impacts
maternal fat retention and infant birth weight. Am J Obstet Gynecol 2003;189:
1423–32.
27 WHO Expert Committee on Physical Status. Physical status: the use and inter-
pretation of anthropometry. WHO Technical Report Series No. 854. Geneva: World
Health Organization, 1995.
28 StataCorp 2011. Stata statistical software: release 12. College Station, TX: StataCorp
LP, 2011.
29 Mehta UJ, Siega-Riz AM, Herring AH, et al. Pregravid body mass index, psycho-
logical factors during pregnancy and breastfeeding duration: is there a link? Matern
Child Nutr. in press.
Obesity and early cessation of breastfeeding 321
at Cornell University Library on April 11, 2013http://eurpub.oxfordjournals.org/Downloaded from
30 Flint DJ, Travers MT, Barber MC, et al. Diet-induced obesity impairs mammary
development and lactogenesis in murine mammary gland. Am J Physiol Endocrinol
Metab 2005;288:E1179–87.
31 Kamikawa A, Ichii O, Yamaji D, et al. Diet-induced obesity disrupts ductal
development in mammary glands of nonpregnant mice. Dev Dyn 2009;238:1092–9.
32 Ertem IO, Votto N, Leventhal JM. The timing and predictors of the early termin-
ation of breastfeeding. Pediatrics 2001;107:543–8.
33 Hill P, Humenick SS, Argubright TM, Aldag JC. Effects of parity and weaning
practices on breastfeeding duration. Public Health Nurs 1997;14:227–34.
34 Bandura A. Reflections on nonability determinants of competence. In:
Sternberg RJ, editor. Competence Considered. New York: Yale University Press,
1990:315–63.
35 Nommsen-Rivers LA, Chantry CJ, Peerson JM, et al. Delayed onset of lactogenesis
among first-time mothers is related to maternal obesity and factors associated with
ineffective breastfeeding. Am J Clin Nutr 2010;92:574–84.
36 Nøhr EA. Obesity in pregnancy [PhD thesis, Report no. 54]. Danish Epidemiology
Science Centre, Institute of Public Health, Faculty of Health Sciences. Aarhus:
University of Aarhus, 2005.
.........................................................................................................
European Journal of Public Health, Vol. 23, No. 2, 322–327
The Author 2012. Published by Oxford University Press on behalf of the European Public Health Association.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://
creativecommons.org/licenses/by-nc/3.0/), which permits unrestricted non-commercial use, distribution, and reproduction in any medium,
provided the original work is properly cited.
doi:10.1093/eurpub/cks065 Advance Access published on 29 May 2012
.........................................................................................................
Lifecourse SEP and tobacco and cannabis use
Lucy Bowes
1,4
, Aude Chollet
1,4
, Eric Fombonne
2
,Ce
´dric Gale
´ra
3
, Maria Melchior
1,4
1 CESP, Inserm U1018, Epidemiology of Occupational and Social Determinants of Health, Villejuif, France
2 Montreal Children’s Hospital, Department of Child Psychiatry, McGill University, Montreal, Canada
3 Child Psychiatry Department, Charles–Perrens Hospital, University Victor Segalen Bordeaux 2, Bordeaux, France
4 Universite
´de Versailles Saint-Quentin, Villejuif, France
Correspondence: Maria Melchior, Epidemiology of Occupational and Social Determinants of Health,
Inserm U1018 (ex U687), CESP, Ho
ˆpital Paul-Brousse, Ba
ˆtiment 15, 16 Avenue Paul Vaillant Couturier, 94807 Villejuif
Cedex, France, tel: +33 (0)1 77 74 74 27, fax: +33 (0)1 77 74 74 03, e-mail: maria.melchior@inserm.fr
Background: Social inequalities in substance use have been well-documented; however, the impact of changes in
socio-economic position from childhood to adulthood is unclear. We examined the relationship between
intergenerational trajectories of social position and tobacco and cannabis use among young adults. Methods:
Data come from 1103 participants (mean age: 28.9 years) of the Trajectoires Epide
´miologiques en Population
(TEMPO) study and their parents, participants of the GAZEL study, France. Multinomial regression analyses were
used to examine associations between lifecourse socio-economic position (SEP) assessed using the parent’s reports
of family income (1989 and 2002) and the participant’s educational attainment, occupational grade and job
stability in 2009, with self-reported tobacco and cannabis use in 2009. Results: Compared with participants with
stable intermediate/high SEP, those with stable low SEP and those with declining SEP were more likely to use
tobacco (age- and sex-adjusted ORs = 2.03 and 2.26). Participants who experienced declining SEP were also dis-
proportionately likely to use and abuse cannabis (adjusted ORs = 2.22 and 2.73). Associations remained significant
after adjusting for family (parental smoking, alcohol use, ill health, unemployment, depression and divorce) and
individual (early tobacco and cannabis use, academic difficulties, juvenile internalizing and externalizing
problems) risk factors. Conclusions: Cross-sectional studies indicate social inequalities in substance use. Our longi-
tudinal findings suggest that individuals who experienced declining SEP from childhood to adulthood may be
twice as likely to use tobacco and cannabis compared with individuals with a stable/high trajectory. Interventions
targeting substance abuse should take into account lifecourse determinants including the interplay between
individuals’ socio-economic origins and later attainment.
.........................................................................................................
Introduction
Each year, tobacco smoking accounts for approximately 5 million
deaths globally,
1
while cannabis is the most commonly used illicit
drug across industrialized countries.
2
Psychoactive substance use
generally begins in adolescence and while tobacco use largely
persists after the transition to adulthood, cannabis use tends to
decrease. However, recent evidence suggesting that a growing
proportion of individuals maintain high levels of use into their
20 s and 30 s calls for research on factors associated with such
substance use in this age group.
3,4
Tobacco and cannabis use are
disproportionately frequent in adults who belong to disadvantaged
social groups.
5,6
Additionally, childhood socio-economic disadvan-
tage may be independently associated with later substance use,
7,8
and
there is suggestion that lifecourse socio-economic characteristics
may be more precise than adult socio-economic position
(SEP).
9,10
In particular, declines in SEP (i.e. downward
socio-economic trajectories) may be associated with elevated rates
of tobacco smoking,
11,12
but little is known regarding the association
with cannabis use.
Childhood socio-economic disadvantage is associated with
specific family and individual risk factors.
13
Children who grow
up in socially disadvantaged families may be more likely to display
emotional and behavioral problems early on, to experience school
difficulties, and have problems with their peers, which may further
contribute to risk of substance use.
14
Substance use in adolescence
may, in turn, be related to poor educational and occupational
outcomes in adulthood.
15,16
The use of lifecourse socio-economic
measures offers the possibility of investigating patterns of
inequalities from childhood to adulthood.
17
322 European Journal of Public Health
at Cornell University Library on April 11, 2013http://eurpub.oxfordjournals.org/Downloaded from