Unhealthy weight gain and paediatric obesity have been demonstrated to track into later childhood and adulthood(Reference Singh, Mulder and Twisk1, Reference Must2). Due to the many long-term adverse effects of paediatric obesity, the prevention of child obesity has been recognized as a public health priority(Reference James, Rigby and Leach3).
Breast-feeding within the first months of life may prevent excess weight gain in childhood(Reference Arenz, Ruckerl and Koletzko4). However, a recent review of the epidemiological literature in this area concluded that although the influence of breast-feeding on body mass in later life appears plausible, it has been difficult to prove empirically(Reference Beyerlein and von Kries5). That review identified several observational studies that have shown evidence of a small protective effect of breast-feeding with respect to overweight in childhood. Further, three meta-analyses reported significant protective effects of breast-feeding against overweight in later life(Reference Arenz, Ruckerl and Koletzko4, Reference Harder, Bergmann and Kallischnigg6, Reference Owen, Martin and Whincup7), although another meta-analysis(Reference Owen, Martin and Whincup8) showed no effect of breast-feeding on mean BMI after adjusting for potential confounders(Reference Beyerlein and von Kries5).
Moreover, epidemiological evidence from Australia on the association between breast-feeding and body composition among infants is limited. Burke et al. conducted one of the few large, longitudinal, population-based studies in this area and showed that infants from Western Australia breast-fed for >12 months were leaner at 1 year but not at 8 years(Reference Burke, Beilin and Simmer9). Given that the prevalence of childhood overweight/obesity remains high in Australia(10), a better understanding of the roles of modifiable factors such as exposure to breast-feeding in childhood is required if any future reductions in the prevalence of overweight/obesity are to be achieved.
Therefore, we used a large data set of infants and children of pre-school age (1–6 years) to establish the relationship between duration of breast-feeding and mean BMI and waist circumference, as well as the likelihood of being overweight/obese, during early childhood.
Methods
The Sydney Paediatric Eye Disease Study (SPEDS) was conducted during 2007–2009. Eligible children between the ages 6 and 72 months (n 3333) were identified and recruited via door-to-door census in four randomly selected postcodes in metropolitan Sydney, Australia. A total of 2461 children were examined (73·8 % response rate). The current report focuses on children aged 1–6 years. Written informed consent was obtained from parents and all examination protocols were approved by the University of Sydney's Human Research Ethics Committee and adhered to the tenets of the Declaration of Helsinki.
Two questionnaires (113 items and sixty-three items) were completed by the parents (generally by the mother, otherwise by the father of the child). These included questions on self-reported ethnicity (European Caucasian (referred to as Caucasian in the present paper), East Asian (China, Malaysia, Singapore, Indonesia, Philippines, Japan, Korea, Myanmar, Thailand, Laos, Cambodia, and Vietnam), South Asian (India, Sri Lanka and Pakistan), Middle Eastern or Others/Mixed (includes all other ethnicities)) and parental education (ranged from never having attended school to having completed a higher degree such as a Masters or PhD). Mothers were also asked: ‘At any time during the pregnancy with your child did you smoke?’ (yes/no/don't know). If they answered ‘yes’ to this question they were considered to have smoked during the pregnancy.
Data from the government-issued personal health record for children provided to the parents of all Australia-born children, containing details on birth parameters (birth weight, weeks of gestation) recorded at the time of birth and at subsequent baby health check visits, were gathered where possible.
Questions on breast-feeding included the following: (i) ‘Has your child ever been breast-fed?’ (yes/no/don't know); and (ii) ‘Including times of weaning, what is the total time that your child was breast-fed?’ (number of weeks or months/less than a week/don't know). We categorized breast-feeding duration into the following: never breast-fed, <1 month, 1–<3 months, 3–<6 months and ≥6 months.
Anthropometric parameters were measured at the site of a study clinic (which was set up in rented premises) by the same study personnel. Length was measured in children less than 24 months of age and height in children 24 months and older. Height was measured using a free-standing telescopic height rod (SECA model 220; Hamburg, Germany). Waist circumference (cm) was measured using a steel measuring tape with measurements made halfway between the lower border of the ribs and the iliac crest in a horizontal plane. Weight was measured using a SECA 4802 digital floor scale (Scale-tronix, White Plains, NY, USA). BMI was calculated as weight divided by the square of height (kg/m2). Children were classified on the basis of age- and sex-specific BMI into three groups: <85th percentile (non-overweight/obese), ≥85th and <95th percentile (overweight) and ≥95th percentile (obese)(Reference Kuczmarski, Ogden and Guo11). Due to the small number of overweight and obese children, we combined the two to form a group termed ‘overweight/obese’.
Statistical analyses were performed using the SAS statistical software package version 8·2 (SAS Institute Inc., Cary, NC, USA). Regression models were used to examine possible linear relationships between breast-feeding and BMI and waist circumference. BMI and waist circumference were analysed as continuous (per kg/m2 or cm increase) and categorical variables (overweight/obese). Breast-feeding duration was also examined as a continuous variable (per month increase) and a categorical variable (never breast-fed, <1 month, 1–<3 months, etc.). Logistic regression analysis was also used to calculate adjusted odds ratios and 95 % confidence intervals. Logistic regression models were constructed to assess significant associations between breast-feeding and overweight/obesity among 1–6-year-old children defined using the age- and sex-specific BMI percentiles. Multivariable modelling was conducted with BMI or waist circumference as the study outcome, and breast-feeding as the independent variable. Linear and logistic regression models were adjusted first for age and sex, and then further adjusted for ethnicity, height, birth weight, parental education and maternal smoking during pregnancy. Significance was taken as P < 0·05.
Results
There were 2092 children aged 1–6 years who were examined and Table 1 shows the characteristics of the study sample. Children in the oldest (5–6 years) v. the youngest (1–2 years) age group were more likely to have higher BMI and waist circumference, but were less likely to have been breast-fed for a longer duration (>1 month). We also compared study characteristics between mothers who provided information on breast-feeding duration (n 1349) compared with those who did not (n 743). Mothers who responded to the question on breast-feeding duration v. those who did not were more likely to be Caucasian (49·3 % v. 40·3 %, P < 0·0001), tertiary qualified (53·0 % v. 42·2 %, P = 0·01) and have children with higher birth weight (3·38 v. 3·27 kg, P = 0·001). Significant differences were not observed between the two groups of respondents in relation to any of the other study characteristics.
Table 1 Characteristics of the study sample by age group: infants and pre-school children (1–6 years), Sydney Paediatric Eye Disease Study, 2007–2009
*Unadjusted P value from test for heterogeneity across age groups.
†These data presented as mean and sd; other data as n and %.
Relationship between breast-feeding (as a categorical variable) and anthropometric measures (as continuous and categorical variables)
A significant correlation was observed between breast-feeding duration and BMI (ρ = −0·12, P < 0·0001). After adjusting for age, sex, ethnicity, height, birth weight, parental education and maternal smoking during pregnancy, 1–6-year-old children who were breast-fed compared with those who were not had a significantly lower mean BMI, 16·7 (se 0·1) kg/m2v. 17·1 (se 0·2) kg/m2 (P = 0·01; Table 2). A significant trend for decreasing BMI was observed with increasing duration of breast-feeding (multivariable-adjusted P trend = 0·002). Significant associations were not observed between breast-feeding and waist circumference among this group of children (Table 2).
Table 2 Adjusted mean BMI and waist circumference with their standard errors according to breast-feeding status and duration in infants and pre-school children (1–6 years), Sydney Paediatric Eye Disease Study, 2007–2009
*Adjusted for age, sex, ethnicity, height, birth weight, parental education and maternal smoking during pregnancy.
Table 3 shows that children who were breast-fed v. those never breast-fed were 46 % less likely to be overweight or obese (multivariable-adjusted OR = 0·54; 95 % CI 0·36, 0·83). Children who were breast-fed for <1 month or ≥6 months compared with those never breast-fed had a 58 % and 44 % reduced odds of being overweight/obese, respectively (Table 3). However, a significant trend for likelihood of overweight/obesity with increasing duration of breast-feeding was not observed in children of pre-school age.
Table 3 Association between breast-feeding and prevalence of overweight/obesity (BMI ≥ 85th percentile) in infants and pre-school children (1–6 years), Sydney Paediatric Eye Disease Study, 2007–2009
*Adjusted for age, sex, ethnicity, height, birth weight, parental education and maternal smoking during pregnancy.
†The reference group is the ‘never’ breast-fed group for all comparisons concerning duration category.
Relationship between breast-feeding (as a continuous variable) and anthropometric measures (as a continuous variable)
After multivariable adjustment, each month increase in breast-feeding was associated, on average, with a BMI decrease of 0·04 kg/m2 (P = 0·002) and 0·03 kg/m2 (P = 0·03) among children aged 1–2 years and 3–4 years, respectively (Table 4). Among 1–2-year-olds only, each month increase in breast-feeding was associated with, on average, a 0·06 cm decrease in waist circumference (multivariable-adjusted P = 0·04). Significant associations between breast-feeding and anthropometry were not observed among 5–6-year-old children (Table 4).
Table 4 Regression analyses of breast-feeding and BMI and waist circumference in infants and pre-school children (1–6 years), Sydney Paediatric Eye Disease Study, 2007–2009
*Adjusted for age, sex, ethnicity, height, birth weight, parental education and maternal smoking during pregnancy.
Associations between breast-feeding and BMI stratified by sex
We also stratified the associations by sex, given that previous studies have demonstrated a sex difference in the protective effects of breast milk(Reference Klein, Bergel and Gibbons12, Reference Nelson, Gordon-Larsen and Adair13). We found that 1–6-year-old girls who were ever breast-fed (n 68, 12·4 %), compared with those never breast-fed (n 21, 22·3 %), were 50 % less likely to be overweight/obese (multivariable-adjusted OR = 0·50; 95 % CI 0·26, 0·94). A non-significant association was observed between breast-feeding and being overweight/obese among boys (multivariable-adjusted OR = 0·58; 95 % CI 0·33, 1·03). Similarly, breast-fed girls had significantly lower mean BMI than girls never breast-fed, 16·4 v. 17·0 kg/m2, respectively (multivariable-adjusted P = 0·03). Significant associations were not observed with mean BMI among boys (P = 0·10).
Discussion
In the present relatively large population-based study of 1–6-year-old Australian children, a modest inverse association between breast-feeding duration and BMI was observed. Linear associations between breast-feeding with BMI and waist circumference were more marked in the age groups 1–2 years and 3–4 years. Children who were ever breast-fed had a 46 % less likelihood of being overweight or obese compared with children who were never breast-fed. This association was stronger in girls than boys, i.e. girls who were v. those never breast-fed had a 50 % reduced odds of being overweight/obese.
Breast-feeding as an independent variable showed a modest inverse, linear association with BMI and waist circumference among these children of pre-school age, and this concurs with prior childhood studies(Reference Weyermann, Rothenbacher and Brenner14–Reference Brion, Lawlor and Matijasevich16). More recently, the Avon Longitudinal Study of Parents and Children (n 4852) found that longer breast-feeding duration was associated with lower mean BMI (P trend < 0·001)(Reference Brion, Lawlor and Matijasevich16). The observed association between breast-feeding duration and waist circumference concurs with previous reports and suggests a possible protective effect of breast-feeding on measures of central obesity(Reference Rudnicka, Owen and Strachan17). The explanation for the observed relationship with BMI and waist circumference probably lies less in inherent obesity-preventing properties of breast milk (e.g. presence of leptin in breast milk) than in the fact that breast-feeding displaces potentially more energy-dense formula feeding(Reference Hediger, Overpeck and Kuczmarski18). Additionally, there may be differences in the regulation of intake between formula-fed infants and breast-fed infants, or the higher protein/nitrogen content of infant formula compared with breast milk could cause a metabolic response of increased insulin in formula-fed infants resulting in excessive weight gain(Reference Hediger, Overpeck and Kuczmarski18). These findings could have potential public health implications, as adiposity in early childhood could influence body mass in later life given that tracking of BMI from childhood through to adulthood has been previously documented(Reference Parsons, Power and Logan19).
Surprisingly, a significantly reduced likelihood of being overweight/obese was observed only among children who were breast-fed for either <1 month (38 % reduced odds) or ≥6 months (46 % reduced odds). This observed protective effect of breast-feeding was larger than that reported in previous studies(Reference Burke, Beilin and Simmer9, Reference Weyermann, Rothenbacher and Brenner14, Reference Hediger, Overpeck and Kuczmarski18), and this could be because of the small number of cases in each monthly duration of breast-feeding stratum. We also observed a non-significant trend with increasing duration of breast-feeding. In a US study of 2685 children(Reference Hediger, Overpeck and Kuczmarski18), a similar non-significant dose-dependent effect of the duration of breast-feeding on overweight status was demonstrated. Additionally, that US study also observed no threshold effect or a reduction in risk of adiposity with breast-feeding for ≥3 months or ≥6 months(Reference Hediger, Overpeck and Kuczmarski18). Our study finding could be explained by the duration of breast-feeding being based on mother's recall. Recall bias could potentially mask a true dose-dependent association between breast-feeding and overweight/obesity in these children. Another reason could be selection bias, as we cannot rule out that significant differences observed in study characteristics (e.g. parental education) between those children's mothers who did and did not provide information on breast-feeding duration could have influenced our results. Further, we did not have data available on the exclusivity of breast-feeding; given that exclusive breast-feeding could influence infant BMI(Reference Arenz, Ruckerl and Koletzko4, Reference Harder, Bergmann and Kallischnigg6, Reference Owen, Martin and Whincup7), we cannot rule out residual confounding from this unmeasured factor. Finally, there could be potential subgroups of children with a specific genetic makeup that makes them more susceptible to the dose-dependent effects of breast-feeding than others(Reference Beyerlein and von Kries5).
Our observation of a strong linear association of breast-feeding with BMI and waist circumference in the younger age groups, but not among 5–6-year-old children, concurs with the findings reported by Burke et al. (Reference Burke, Beilin and Simmer9). In that Australian study it was observed that measures of adiposity with breast-feeding tended to converge with increasing age of children. At the age of 5–6 years, patterns of dietary intake and physical activity probably have more of an impact on weight status than infant feeding(Reference Hediger, Overpeck and Kuczmarski18), thus explaining the non-significant observation in this age group. Nevertheless, further longitudinal studies with adequate study power are needed to examine the temporal association between breast-feeding and BMI during childhood.
We observed a differential sex response to breast-feeding on weight status, i.e. breast-fed girls rather than boys were less likely to be overweight/obese. These results parallel those from the US National Longitudinal Study of Adolescent Health (n 11 998)(Reference Nelson, Gordon-Larsen and Adair13), which demonstrated that ever breast-feeding was associated with a reduced likelihood of overweight in girls (OR = 0·83; 95 % CI 0·72, 0·95), but not among boys (OR = 0·90; 95 % CI 0·76, 1·05). It is unclear as to why the association is stronger in girls than boys of pre-school age in our study sample; however, one possible explanation is that the effects of hormones potentially involved in the metabolic imprinting effects of breast-feeding in infants(Reference Geary, Pringle and Rodeck20, Reference Savino, Fissore and Liguori21) could be influenced by sex and this indirectly/directly influences weight status(Reference Klein, Bergel and Gibbons12). Nevertheless, these results could also reflect residual confounding e.g. by unmeasured behavioural attributes or a loss of power due to stratification by sex, hence further research is needed to confirm this differential gender response to breast-feeding.
The strengths of the present study include its population-derived design, a satisfactory response rate and standardized anthropometric measurement protocols. Limitations include the fact that the data come primarily from a survey of eye disease in Australian pre-school children, and not of obesity. The parental questionnaire did not target adiposity as a key study outcome; hence, exhaustive information was not collected on some variables such as maternal BMI, physical activity or sedentary pursuits. Therefore, we cannot rule out the possibility of residual confounding from these unmeasured parameters. Finally, the study is also limited by the presentation of only cross-sectional analyses and we cannot infer cause-and-effect relationships.
Conclusions
We observed a modest, inverse association between breast-feeding duration and mean BMI during early childhood. Children aged 1–6 years who were breast-fed had a 46 % reduced likelihood of being overweight/obese compared with those who were never breast-fed. Associations between breast-feeding and mean BMI were more marked in the younger age groups and among girls than boys. Our findings add to the growing evidence base that any salutary effects of breast-feeding on weight status are modest among children of pre-school age.
Acknowledgements
The Sydney Paediatric Eye Disease Study was supported by the Australian National Health and Medical Research Council, grant number 402425. The authors have no conflicts of interest to declare. B.G., I.S. and P.M. conceived and designed the research. B.G. and G.B. analysed and interpreted the data. B.G. drafted the manuscript. P.M., I.S., V.M.F., L.A.B. and N.P. critically revised the manuscript for important intellectual content.
