The association between breastfeeding and childhood obesity: a meta-analysis

Abstract

Background

Obesity has become a major public health crisis around the world. Since overweight and obesity are strongly correlated with diseases such as diabetes, hypertension, high cholesterol, cardiovascular diseases, stroke, arthritis, and certain types of cancers, the number of obesity-related death is estimated at 2.8 million per year [1, 2]. Childhood obesity has also dramatically worsened and is now considered a major threat to human health [3–5]. According to the International Obesity Taskforce (IOTF) [1] and the World Health Organization (WHO) [6, 7], there are more than 200 million overweight or obese children throughout the world and, in 2010, over 42 million children under age of 5 were classified as overweight. Unfortunately, childhood obesity is linked to several physical and mental health conditions, including orthopedic problems, menstruation problems, sleep trouble, depression, anxiety and diabetes in childhood [8, 9]. Moreover, overweight/obese children are more likely to be obese in adulthood [10].

Childhood obesity has multiple causes, including genetic factors, personal behaviors (e.g., exercise, sleep duration, and TV viewing), dietary habits, and their interactions [11–14], and many researchers have investigated various possible interventions to prevent childhood obesity. Among these, breastfeeding has been associated with a decreased risk of obesity, along with other health benefits for the child and mother. According to the WHO recommendations, infants should be exclusively breastfed for the first 6 months, and breastfeeding should be supplemented with additional foods for the first 2 years (or beyond) [15]. Breast milk is considered the ideal food for infants, as it provides adequate energy and nutrients to meet the infants’ needs. In addition, as breast milk is safe and contains antibodies, breastfeeding could reduce the risk of neonatal infection, gastrointestinal infection, and pneumonia during infancy [9, 10]. It has been indicated that approximately 45% of neonatal infectious deaths, 30% of diarrhoeal death and 18% of respiratory death among children <5 years old are associated with suboptimal breastfeeding [15]. Moreover, breastfeeding has long-term benefits throughout a child’s lifetime. Children and adults who were breastfed have lower rates of overweight/obese, type-2 diabetes, hypertension, and are known score higher on intelligence tests than persons who were formula-fed [12, 15]. Based on the WHO report, if every child in the world was exclusively breastfed for the first 6 months, followed by breastfeeding until 2 years, the lives of 800,000 children would be saved each year [16]. Additionally, breastfeeding protects mothers against breast cancer, ovarian cancer, and obesity. Exclusive breastfeeding also is an effective, natural method of birth control, providing 98% protection between birth and 6 months postpartum [15].

In 2012, approximately 38% of infants who were under 6 months old were exclusively breastfed worldwide [15], with additional feeding methods including partial breastfeeding and exclusively formula feeding [9, 12]. The major factors that affect prevalence and duration of breastfeeding include maternal race/ethnicity, education, breast diseases, inadequate breast milk production, employment, length of maternity leave, inadequate knowledge regarding breastfeeding, lack of familial and societal support, and lack of guidance and encouragement from health care professionals [9, 15]. To strengthen breastfeeding practices, families, employers, professional workers and society as a whole should fully support to breastfeeding mothers.

Over the previous decade, a growing body of research has indicated an association between breastfeeding and childhood obesity. Breastfeeding has been identified as a protective factor for childhood obesity in many studies [16–20], while other studies failed to establish an association between the breastfeeding and childhood obesity [21, 22]. Three previous reviews have addressed this issue, the latest one being published about 9 years ago in 2005 [23–25]. Arenz et al. [23] included 9 studies that focused on the relationship between breastfeeding and childhood obesity with adjusted odds ratio (AOR). Owen et al. [24] presented the pooled odds ratio (OR) for 28 studies reporting the effects of infant feeding on the risk of obesity, as well as the pooled AOR for 6 studies that adjusted for socioeconomic status, parental body mass index (BMI), and maternal smoking. Lastly, Harder et al. analyzed the relationship between the duration of the breastfeeding and overweight based on 17 studies [25].

Given the lack of recent reviews, we conducted a systematic review to assess the association between breastfeeding and childhood obesity, and to explore the nature of the association. Compared to the previous reviews [23–25], this analysis is different in both eligibility criteria and study methodologies. We reviewed all studies that published until 1^st August 2014 to generate an updated and extended data set and used AOR and 95% confidence interval (CI) to control for potential confounders. Moreover, the prevalence of breastfeeding is changing, and there are unique trends emerging in different countries. For example, the prevalence of breastfeeding is increasing in the UK [26], while the prevalence of exclusive breastfeeding is declining in China [27]. Thus, this review provides important updated data to reflect the changing of breastfeeding throughout the world.

Methods

Results

Study selection and characteristics

The comprehensive literature search of electronic databases, key journals, and cross-references yielded 718 publications, which included 428 unduplicated articles, regarding the association between obesity and breastfeeding published before 1^st August 2014 as potentially relevant articles. A total of 25 studies with 226,508 subjects [39–63] were included in the present meta-analysis. The publication dates for these studies ranged from 1997 to 2014, and involved the population of 12 countries, including 5 German studies, 5 American studies, 3 British studies, 3 Australian studies, 2 Chinese studies, 1 Japanese study, 1 Irish study, 1 Greek study, 1 Brazilian study, 1 Dutch study, 1 Czech study, and 1 Canadian study. Of these, 24 included studies published in English and 1 study in Chinese. Ten studies were cross-sectional surveys and 15 were cohort studies, which included 10 prospective cohort studies and 5 indicate historical cohort studies. The selection process is detailed in Figure 1.

Open in a separate window

Figure 1

Flowchart of literature search and selection process ^a . ^aPlease contact the corresponding author for detailed search strategy. ^bThe study did not focus on childhood obesity, breastfeeding, or the relationship between breastfeeding and childhood obesity. The study provided incomplete data. ^cThe study provided incomplete data. The weight was not expressed as BMI.

The description of the included studies is provided in Table 1. Six of the 25 studies defined overweight and obesity according to the IOTF recommendation [45, 47, 49, 52, 57, 58], 10 studies defined overweight as BMI ≥85^th percentile and obesity as BMI ≥95^th percentile according to the criteria developed by the Centers of Disease Control and Prevention (CDC) [39, 42–44, 46, 50, 53, 54, 60, 61], 5 studies defined overweight and obesity according to the German reference as BMI ≥90^th percentile and ≥97^th percentile [40, 41, 48, 59, 62], and 4 studies defined obesity alternatively as BMI ≥90^th percentile, BMI ≥94^th percentile, BMI >2 SD, and the Chinese children’s BMI cut-off for obesity [51, 55, 56, 63].

Table 1

Characteristics of included studies in the meta-analysis

Reference	Origin	Study design	Age	Sample size a	Definition of feeding patterns	Definition of obesity	Confounders
Armstrong 2002 [39]	Great Britain	IHC	39-42 months	32,200	Formula-fed, BF	BMI ≥ 95th	Gender, birth weight, SES
Bergmann 2003 [40]	Germany	PC	6 years	480	Bottle-fed, BF	BMI ≥ 97th	Overweight mother, SESc
Beyerlein 2008 [41]	Germany	CS	54-88 months	9,368	Never & ever BF	BMI ≥ 97th	Weight gain from birth to years of life, maternal BMI, parents’ graduatebc
Burke 2005 [42]	Australia	PC	1-8 years	1,672	Never, BF ≤ 4 mo, BF ≤ 8 mo, BF ≤ 12 mo, BF > 12 mo	BMI ≥ 95th	Birth weight, gestational age, ethnicity, sex, maternal BMI, first child, and maternal educationc
Gillman 2001 [43]	United States	CS	9-14 years	14,377	Mostly or only fed formula, mostly or only BF	BMI ≥ 95th	Age, sex, birth weight, birth order, Tanner stage, exercises, daily energy intake, mother’s BMI, and household income, dietary restraint, weight cycling, and weight concernsbc
Grummer-Strawn 2004 [44]	United States	IHC	4 years	12,587	Never, BF < 1 mo, BF ≤ 2.9 mo, BF ≤ 5.9 mo, BF ≤ 11.9 mo, BF ≥ 12 mo	BMI ≥ 95th	Gender, race/ethnicity, birth weight, and mother’s age, education, prepregnancy BMI, and weight gain during pregnancy c
Hawkins 2009 [45]	Great Britain	PC	3 years	13,172	Never, BF < 4 mo, BF ≥ 4 mo	IOTF	Birth weight, gender, ethnicity, introduction of solid foods, maternal SES, education, lone motherhood status, age at first live birth, number of children, household income, parental overweight, country, et al.bc
Hediger 2001 [46]	United States	CS	3-5 years	2,656	Never & ever EBF	BMI ≥ 95th	Birth weight, race/ethnicity, gender, age group, mother’s BMI, timing of introduction to solid foods
Jwa 2014 [47]	Japan	PC	1.5-8 years	41,572	Formula feeding, mixed feeding, EBF	IOTF	Birth weight, having elder sibling(s), maternal educational level, age and total house-hold income at 0.5 years old c
Koletzko 2009 [48]	Germany	CS	5-6 years	9,357	Never & ever BF	BMI ≥ 97th	Parental education, low birth weight, own bedroom, frequent consumption of butterc
Kwok 2010 [49]	Hong Kong	PC	7 years	7,026	Never, partially or EBF <3 mo, EBF ≥ 3 mo	IOTF	Sex, birth weight, gestational age, birth order, mother’s age at birth, birthplace, number of hospital admissions at 0–3 months, parental education, occupation, incomec
Li 2003 [50]	Great Britain	CS	4-18 years	2,631	Never, BF ≤ 1 mo, BF ≤ 3 mo, BF ≤ 6 mo, BF ≤ 9 mo, BF > 9 mo	BMI ≥ 95th	Sex, parent’s BMI, birth weight, and social classc
Liese 2001 [51]	Germany	CS	9-10 years	2,108	Formula-fed, BF	BMI > 90th	Age, gender, city, nationality, SES, number of siblingsc
McCrory 2012 [52]	Republic of Ireland	IHC	9 years	7,798	Never, BF ≤ 4 wk, BF ≤ 8 wk, BF ≤ 12 wk, BF ≤ 25 wk, BF ≥ 26 wk	IOTF	Dietary quality, exercise, maternal education, gestational age, nationality, weight status, household incomebc
Moschonis 2008 [53]	Greek	IHC	1-5 years	2,374	Exclusive formula-fed, mixed, EBF	BMI ≥ 95th	Parental education, weight status, maternal gestational diabetes mellitus, alcohol consumption patterns during pregnancy, age at birth, birth rankc
Moss 2014 [54]	United States	PC	2-4 years	14,150	Never & ever BF	BMI ≥ 95th	Maternal education, maternal age, maternal race, family poverty, children’s birth weight and gender
Novaes 2011 [55]	Brazilian	CS	6-10 years	764	Never & ever BF	BMI > 2SD	Gender, physical education classes, siblings, maternal nutritional state during pregnancybc
O’Challaghan 1997 [56]	Australia	PC	5 years	2,034	Formula fed, BF ≥ 6 mo, exclusiveness unclear	BMI ≥ 94th	Birth weight, gander, small for gestational age, feeding problems, sleeplessness, parental BMI, maternal education, household income
Scholtens 2008 [57]	Dutch	PC	8 years	2,043	Never, BF ≤ 16 wk, BF > 16 wk	IOTF	Maternal education, overweight, diet at 7 years of agebc
Scott 2012 [58]	Australia	IHC	9-16 years	2,066	Never, BF < 2 mo, BF < 4 mo, BF < 6 mo, BF ≥ 6 mo	IOTF	Age, gender, energy intake, physical activity, sleep duration, maternal age, education, ethnicityb
Toschke 2002 [59]	Czech Republic	CS	6-14 years	33,768	Never & ever BF	BMI ≥ 97th	Parental education and obesity, birth weight, sports outside school, number of siblingsbc
Twells 2010 [60]	Canada	CS	4 years	1,026	formula feeding ≥ 3 mo, mixed feeding ≥3 mo, EBF ≥ 3 mo	BMI ≥ 95th	Birth weight, gender, age, maternal education, whether the baby was preterm or full termc
Van Rossem 2010 [61]	United States	PC	3 years	884	Never, any BF <6 mo, partial BF ≥ 6 mo, EBF ≥ 6 mo	BMI ≥ 95th	Age, sex, maternal educational, race, BMI, pregnancy weight gain, birth weight, gestational agec
Von Kries 1999 [62]	Germany	CS	5-6 years	9,206	Never, EBF ≤ 2 mo, EBF ≤ 5 mo, EBF ≤ 12 mo, BF > 12 mo	BMI ≥ 97th	Parental education, low birth weight, own bedroom, consumption of butterc
Zhao 2013 [63]	China	PC	2-8.4 years	1,189	Formula feeding, mixed feeding, EBF	Boys’ BMI ≥96.3th Girls’ BMI ≥98th	Maternal weight and BMI before pregnancy, maternal weight increase and the level of blood glucose during pregnancy, gender, birth weight, father weight

Open in a separate window

Abbreviation: IHC indicates historical cohort, PC Prospective cohort, CS Cross-sectional, BMI body mass index (calculated as weight in kilograms divided by height in meters squared), BF breastfeeding, EBF Exclusive breastfeeding, IOTF the International Obesity Task Force, SES socio-economic status.

^aThe number of participants included in the analysis with different feeding patterns at last follow-up.

^bAdjustment for amount of watching TV for children.

^cAdjustment for maternal smoking in pregnancy.

Main meta-analysis results

The pooled AOR derived from all 25 studies was 0.78 (95% CI: 0.74, 0.81) and the details are presented in a Forest plot (Figure 2). There was no significant heterogeneity across studies (I² = 45.28%, P < 0.01), and the studies met fixed effects model. Additionally, we analyzed unadjusted data of results. Sixteen studies [39, 40, 43, 45–47, 51, 52, 55–60, 62, 63] presented information on the crude ORs, and the pooled crude OR and the corresponding 95% CIs are shown in Figure 3 with a forest plot. The pooled OR was 0.61 (95% CI: 0.55, 0.68). The homogeneity hypothesis was rejected (I² = 65.40%, P < 0.01), and hence the studies met random effected meta-analysis criteria.

Open in a separate window

Figure 2

Forest plot of the associations between breastfeeding (never & ever) and obesity in all 25 studies. Adjusted odds ratios and 95% confidence intervals.

Open in a separate window

Figure 3

Forest plot of the associations between breastfeeding (never & ever) and obesity in 16 studies. Unadjusted odds ratios (UOR) and 95% confidence intervals (CI).

Publication bias

The funnel plot presented an asymmetric pattern (Figure 4), and the results of Egger test (p = 0.16) and Begg test (p = 0.25) were not statistical significant. The result of Fail-safe N (N_fs0.05 = 939.78) indicated high reliability of results. Therefore, we cannot completely exclude potential publication bias.

Open in a separate window

Figure 4

Funnel plot of standard error by log odds ratio for meta-analysis of studies.

Discussion

The association between breastfeeding and childhood obesity has long been under debate; however, no agreement has been achieved with respect to this issue. This meta-analysis suggests that breastfeeding protects against obesity in childhood. The risk of childhood obesity was lower in breastfed children by 22% compared with those who were never breastfed. Meanwhile, a stronger link was found between breastfeeding and obesity based on the crude OR. Two previous meta-analyses have demonstrated that breastfeeding is a protective factor for obesity, but reported inconsistent pooled ORs [23, 24]. A change to dose-response effect between breastfeeding duration and childhood obesity was confirmed in our study, in agreement with Harder et al. [25]. In particular, breastfeeding for more than 7 months significantly decreased the risk of obesity [64].

This meta-analysis shows that breastfeeding is a protective factor of adiposity, in contrast to results from several studies, which have shown no link between breastfeeding and obesity [21, 22, 49]. There are several possible explanations for this finding. First, breast milk provides a moderate amount of calories and nutrients for infant, such as sugar, water, protein and fat [9]. Moreover, the composition of breast milk changes with time and the mother’s diet [65]. In contrast, formula feeding provides higher levels of fat and protein than the baby’s needs [66]. Higher protein and fat intake in early childhood have been associated with adiposity [67]. Moreover, breast milk rather than other milk contains bioactive substances such as leptin and ghrelin, which can influence the proliferation and differentiation of the infant’s adipocytes [52, 67]. Thus, breast milk is rich in effective ingredient with higher nutritional value.

Interestingly, many investigations have indicated that breast milk can be influenced by environmental and behavioral factors [66, 67]. However, several researchers have suggested that unadjusted results should not be ignored either irrespective of the adjusted results [68]. In the present analysis, we compared the pooled ORs before and after combined adjustment for confounders, and observed a decreasing trend after the adjustment for potential confounders. These results support the hypothesis that maternal and individual factors, including genetic, environmental, and behavioral factors, can influence the body shape of children [45, 69]. Nevertheless, the interaction among multiple factors is still unclear. Factors such as physical activity and screen time (e.g., computer use or TV viewing) are known to significantly affect on the risk of obesity among children [52, 55]. The subgroup analyses of our study show that breastfeeding was associated with a lower risk of obesity, and the association was independent of TV viewing and exercise level [24]. Therefore, the mechanism(s) by which breastfeeding influences the later life influences the risk of obesity appear to be complex [70].

In addition, stratified analyses revealed differences in obesity definition and breastfeeding definition in the studies we examined. However, it is important to note the smaller sample size in several categories, which may have reduced the statistical power [23, 24]. The subgroup analyses showed that the association between breastfeeding and childhood obesity was particularly strong in 4 studies that applied other definitions of obesity. The pooled estimates from cross-sectional studies, cohort studies, and studies defining obesity cutoff ≥ 95^th were similar to the overall pooled AOR; the pooled estimates from studies using other categories for subgroup analyses were also close to the overall pooled AOR. Moreover, the link between breastfeeding and childhood obesity was slightly different in studies that used different breastfeeding types. This may be because the original data from several studies did not distinguish between exclusive and partial breastfeeding, thereby reducing the precision of our analysis [57, 58]. Therefore, future studies should compare the results for exclusive breastfeeding, mixed feeding, and exclusive formula feeding.

A change to dose-response relationship between breastfeeding duration and childhood obesity was confirmed in this study [25]. Breastfeeding for <3 months provided a minor protective effect for childhood obesity, while breastfeeding for ≥7 months showed a significantly high protection. These results are inconsistent with the findings of several studies. For example, Toschke et al. reported no association between breastfeeding for <6 months and obesity [71]. Similarly, another study found children who were exclusively breastfed for <4 months did not have a reduced the risk of obesity [72]. In contrast, McCrory et al. performed a change to dose-response relationship in children breastfed for ≥4 weeks and lower risk of obesity [52], and another study showed breastfeeding for 1–16 weeks was associated with a significantly reduced risk of obesity [57]. These discrepancies in the findings may be related to the diverse populations (with unique genetic and environmental backgrounds) and different sample sizes.

This meta-analysis has several strengths. First, it was based on adjusted ORs and 95% CIs from each study, which largely ruled out residual confounding by other factors and improved the accuracy of the effect estimate. In addition, we analyzed the most recently data published (before 1^st August 2014) and historical data, with no lower date limit. Compared to previous reviews [23, 24], which reported the pooled AOR, this meta-analysis included a larger sample size, which improved the statistical power. Also, several limitations of this study should be considered. An important limitation is the funnel plot presented an asymmetry; thus, the publication bias cannot be fully ruled out. There are 2 possible explanations for this result. First, among the 25 included studies, 24 were published in English and 1 in Chinese, while other language were not included in this study. It is well accepted that studies showing significant results are more likely to be published, and more likely to be published in English, which is a common cause of publication bias. Although we searched several electronic databases of Chinese to increase the number of studies reported in Chinese in this analysis, none could be added due to the poor quality of reporting. Only 2 of the 25 included papers investigated the Chinese population, and the results from 1 study showed no association between breastfeeding and childhood obesity [49]. Therefore, confirmation of these findings is needed in further research conducted in the Chinese population. Second, we did not consider unpublished investigations that may lead to publication bias. Another limitation of the present analysis is the AORs of each study that were adjusted for different confounders. As well, a weaker association between breastfeeding and obesity was observed after the adjustment of known confounders. Therefore, if we adjust for the same and more relevant confounders, the protective effect of early breastfeeding might reduce largely. However, breastfeeding has been suggested to be a protective factor to childhood obesity in numerous studies [64, 73]. Moreover, because several studies did not report the breastfeeding type, we could not distinguish all the collected data between exclusive and partial breastfeeding, and hence, the precision of effect estimates may be influenced. However, the pooled estimate for exclusive breastfeeding was calculated, based on data from 8 studies [46, 47, 49, 53, 60–63]. Finally, the different BMI cut-offs using for defining obesity in different studies may also have had an impact on the overall estimate; therefore, we performed a subgroup analysis by obesity definition to minimize this limitation.

Conclusion

In summary, the aim of this systematic review was to investigate the association between breastfeeding and childhood obesity. The results indicate a protective effect of breastfeeding for childhood obesity, and prolonged breastfeeding is directly related to a decreasing risk of obesity. In particularly, children being breastfed for ≥7 are significantly less likely to be obese in later childhood. Future research should distinguish the data between exclusive breastfeeding, mixed feeding, and exclusive formula feeding, and use uniform potential confounders.

Authors’ information

PPW is a professor of epidemiology in the Faculty of Medicine of Memorial University of Newfoundland (MUN). GWH is a professor of dietetics/nutrition, and the Dean of the School of Public Health in Tianjin Medical University (TMU). LL and ZC are current Master’s students at MUN, while JY is a lecturer at TMU.

Acknowledgements

Abbreviations

Footnotes

Contributor Information

Jing Yan, Email: nc.ude.umt@gnijnay.

Lin Liu, Email: ac.num@2333ll.

Yun Zhu, Email: moc.361@10216891nuyuhz.

Guowei Huang, Email: nc.ude.umt@iewouggnauh.

Peizhong Peter Wang, Email: ac.num@gnawp.

References