Breastfeeding and risk of child obesity Kathryn G. Dewey, PhD Program in International and Community Nutrition University of California, Davis

Breastfeeding and risk of child obesity

Kathryn G. Dewey, PhDKathryn G. Dewey, PhD

Program in International and Community NutritionProgram in International and Community Nutrition

University of California, DavisUniversity of California, Davis

Study selection criteria

N N >> 100 per feeding group 100 per feeding group Age at follow-up > 3 yAge at follow-up > 3 y Outcome = % overweight or obeseOutcome = % overweight or obese

Studies in preschool childrenAuthor, year, Author, year,

sitesite

N, ageN, age Feeding groupsFeeding groups OutcomesOutcomes ResultsResults

[*p < 0.05][*p < 0.05]

Armstrong, Armstrong, 2002, Scotland2002, Scotland

32,20032,200

3-4 y3-4 y

EBF vs. EFF at 6-8 EBF vs. EFF at 6-8 wkwk

BMI > 95BMI > 95thth


AOR 0.72*AOR 0.72*

AOR 0.70*AOR 0.70*

Bogen, 2004, Bogen, 2004, USAUSA11

73,45873,458

4 y4 y

BF < 8 wk, 8-15 wk, BF < 8 wk, 8-15 wk, 16-26 wk, > 26 wk 16-26 wk, > 26 wk (+/- concurrent FF)(+/- concurrent FF)

BMI BMI >> 95th 95th AOR 0.71 for AOR 0.71 for 16-26 wk w/o 16-26 wk w/o FF*FF*

Hediger, 2001,Hediger, 2001,

USAUSA

2,6852,685

3-5 y3-5 y

Ever BF vs. EFFEver BF vs. EFF

(+ dur. full BF)(+ dur. full BF)

BMI 85-94BMI 85-94thth

BMI BMI >> 95 95thth

AOR 0.63*AOR 0.63*

AOR 0.84AOR 0.84

Grummer-Grummer-Strawn, 2004, Strawn, 2004, USAUSA

177,304177,304

(12,587)(12,587)

4 y4 y

BF < 1, 1-3, 3-6, 6-12, BF < 1, 1-3, 3-6, 6-12, or or >> 12 mo, vs. EFF 12 mo, vs. EFF

BMI BMI >> 95 95thth AOR for White, AOR for White, non-Hispnon-Hisp

0.70* (6-12 mo)0.70* (6-12 mo)

0.49* (0.49* (>> 12 mo) 12 mo)

O’Callaghan, O’Callaghan, 1997, Australia1997, Australia

3,9093,909

5 y5 y

BF durationBF duration BMI 85-94BMI 85-94thth


NSNS

Burke, 2005, Burke, 2005, AustraliaAustralia

2,0872,087

1-8 y1-8 y

EFF vs. BF EFF vs. BF << 4 mo, 5- 4 mo, 5-8, 9-12, > 12 mo8, 9-12, > 12 mo

BMI BMI >> 95 95thth Highest risk in Highest risk in BF BF << 4 mo* 4 mo*

11Only among white children whose mothers did not smokeOnly among white children whose mothers did not smoke

Studies in school-aged childrenAuthor, year, siteAuthor, year, site N, ageN, age Feeding groupsFeeding groups OutcomesOutcomes ResultsResults

Von Kries, 1999,Von Kries, 1999,

GermanyGermany

9,3579,357

5-6 y5-6 y


(+ dur. EBF)(+ dur. EBF)



AOR 0.79*AOR 0.79*

AOR 0.75*AOR 0.75*

Wadsworth, 1999,Wadsworth, 1999,

UKUK

3,7313,731

6 y6 y

Ever BF vs. EFF Ever BF vs. EFF (+ dur. BF)(+ dur. BF)



RR 0.95RR 0.95

RR 0.88RR 0.88

Bergmann, 2003, Bergmann, 2003, GermanyGermany

480480

6 y6 y

BF BF >> or < 3 mo or < 3 mo BMI > 90BMI > 90thth


AOR 0.53*AOR 0.53*

AOR 0.46*AOR 0.46*

Reilly, 2005,Reilly, 2005,

UKUK

8,2348,234

7 y7 y

EBF EBF >> 2 mo or < 2 mo or < 2 mo, vs. EFF2 mo, vs. EFF

BMI BMI >> 95 95thth Lower risk if Lower risk if EBF*, if non-EBF*, if non-smoking mothersmoking mother

Toschke, 2002,Toschke, 2002,

Czech RepublicCzech Republic

33,76833,768

6-14 y6-14 y

Ever BF vs. EFF, Ever BF vs. EFF, (+ duration BF)(+ duration BF)



AOR 0.80*AOR 0.80*

AOR 0.80*AOR 0.80*

Liese, 2001,Liese, 2001,

GermanyGermany

2,1082,108

9-10 y9-10 y


(+dur. BF, EBF)(+dur. BF, EBF)

BMI > 90thBMI > 90th AOR 0.66*AOR 0.66*

Gillman, 2001, Gillman, 2001,

USAUSA

15,34115,341

9-14 y9-14 y

Pred BF vs. Pred Pred BF vs. Pred FF 0-6 moFF 0-6 mo

BMI > 95BMI > 95thth AOR 0.78*AOR 0.78*

Studies in older adolescentsAuthor, year, siteAuthor, year, site N, ageN, age Feeding groupsFeeding groups OutcomesOutcomes ResultsResults

Kvaavik, 2005,Kvaavik, 2005,

NorwayNorway

635635

13 y13 y

BF > 3 mo vs. BF > 3 mo vs. EFFEFF

BMI > 95BMI > 95thth AOR 0.15*AOR 0.15*

Kramer, 1981,Kramer, 1981,

CanadaCanada

427427

12-18y12-18y


(+dur. full BF)(+dur. full BF)

> 120% median > 120% median weight for htweight for ht

RR 0.31*RR 0.31*

Tulldahl, 1999,Tulldahl, 1999,

SwedenSweden

781781

17-18y17-18y

EBF > vs. EBF > vs. << 2 mo 2 mo BMI > 85BMI > 85thth RR 0.70*RR 0.70*

Poulton, 2001,Poulton, 2001,

New ZealandNew Zealand

1,0371,037

3-26 y3-26 y

BF > 6 mo vs. BF > 6 mo vs. EFFEFF

BMI > 25 kg/mBMI > 25 kg/m22 AOR 0.25-1.01AOR 0.25-1.01

Li, 2003, Li, 2003,

UKUK

2,6312,631

4-18 y4-18 y

Duration BF vs. Duration BF vs. BF < 1 wkBF < 1 wk

BMI > 95BMI > 95thth AOR 0.68-2.02AOR 0.68-2.02

Victora, 2003, Victora, 2003,

BrazilBrazil

2,2502,250

18 y (18 y (♂♂))

Duration Pred BFDuration Pred BF BMI > 85BMI > 85thth

skinfolds > 90skinfolds > 90thth

Significant Significant linear trendlinear trend

Nelson, 2005, Nelson, 2005,

USAUSA

11,99811,998

850 sib 850 sib pairspairs

12-21 y12-21 y

BF BF >> 9 mo vs. 9 mo vs. EFFEFF

BMI BMI >> 85 85thth AOR 0.78* (F)AOR 0.78* (F)

AOR 0.83 (M)AOR 0.83 (M)

NS for sib pairsNS for sib pairs

Additive interactions of maternal prepregnancy BMI and breastfeeding on childhood overweight. Li et al. Obesity Res 2005;13:362-371 (2-14 years of age)

0

5

10

15

20

25

30

35

% Overweight

never <4 mo >4 mo

Breast-feeding (months)

< 2525-29

> 30

Maternal BMI

31.5

6.0

Breastfeeding and childhood obesity – a systematic review. Arenz et al. Intl J Obes, 2004; 28:1247-56.

0 1

O'Callaghan 1997

Bergmann 2003

Hediger 2001

Li 2003

Poulton 2001

von Kries 1999

Liese 2001

Toschke 2001

Gillman 2001

Meta-analysis

Adjusted odds-ratioSource: Arenz et al. Intl J Obes, 2004

AOR 0.78 (0.71, 0.85)

Effect of infant feeding on the risk of obesity across the life course: a quantitative review of published evidence. Owen et al. Pediatrics 2005;115:1367-77. Reviewed 61 studies; 28 with odds ratio estimatesReviewed 61 studies; 28 with odds ratio estimates OR for any BF, all studies: 0.87 (0.85-0.89)OR for any BF, all studies: 0.87 (0.85-0.89)

For infants: For infants: OR = 0.50 (0.26-0.94)OR = 0.50 (0.26-0.94) For young children:For young children: OR = 0.90 (0.87-0.92)OR = 0.90 (0.87-0.92) For older children: For older children: OR = 0.66 (0.60-0.72)OR = 0.66 (0.60-0.72) For adults: For adults: OR = 0.80 (0.71-0.91)OR = 0.80 (0.71-0.91)

Adjusted for SES, parental BMI & maternal smoking) for any BF, all Adjusted for SES, parental BMI & maternal smoking) for any BF, all studies: AOR = 0.93 (0.88-0.99)studies: AOR = 0.93 (0.88-0.99)

Protective effect of BF stronger in 4 studies in which initial feeding Protective effect of BF stronger in 4 studies in which initial feeding groups were exclusive: OR = 0.76 (0.70-0.83)groups were exclusive: OR = 0.76 (0.70-0.83)

Stronger relationship with longer duration of BF:Stronger relationship with longer duration of BF: OR = 0.81 (0.77-0.84) for BF OR = 0.81 (0.77-0.84) for BF >> 2 mo 2 mo OR = 0.89 (0.86-0.91) for any BF durationOR = 0.89 (0.86-0.91) for any BF duration

Bergmann et al. (Germany)N=480; BMI at 0-6 y

Percentage of children > 90th percentile

Poulton & Williams (New Zealand)N=1,037, born 1972-73BMI at 3, 5, 7, 9, 11, 13, 15, 18, 21, 26 y

In those BF > 6 mo, a lower risk of obesity In those BF > 6 mo, a lower risk of obesity was observed at 9-18 years of age, but not was observed at 9-18 years of age, but not at younger ages (3-8 years) or in adulthood at younger ages (3-8 years) or in adulthood (> 18 years)(> 18 years)

Is puberty / adolescence a critical period, Is puberty / adolescence a critical period, when the influence of infant feeding mode when the influence of infant feeding mode is most clearly expressed? is most clearly expressed?

YesYes NoNoBogenBogen HedigerHedigerGrummer-StrawnGrummer-Strawn O-Callaghan O-Callaghan Von KriesVon Kries WadsworthWadsworthToschkeToschke LiLiLieseLiese VictoraVictoraGillmanGillman BurkeBurkePoulton (trend)Poulton (trend)Nelson (in girls)Nelson (in girls)

Is there a dose-response relationship between BF duration and lower risk of child obesity?

Harder et al. Duration of breastfeeding and risk of overweight: a Harder et al. Duration of breastfeeding and risk of overweight: a meta-analysis. Am J Epidemiol 2005;162:1-7.meta-analysis. Am J Epidemiol 2005;162:1-7.

Included 17 studies:Included 17 studies:Duration BFDuration BF OROR< 1 mo< 1 mo 1.001.001-3 mo1-3 mo 0.81*0.81*4-6 mo4-6 mo 0.76*0.76*7-9 mo7-9 mo 0.67*0.67*> 9 mo> 9 mo 0.68*0.68*

* Significantly different from reference group. [No control for * Significantly different from reference group. [No control for potential confounders.]potential confounders.]

Is there a dose-response relationship between BF duration and lower risk of child obesity?

Does exclusivity of breastfeeding matter?

Bogen et al. Obesity Research 2004;12:1527-1535

Relationship between obesity at age 4 y and duration of breastfeeding and concurrent formula use for whites (solid line) and blacks (dashed line)

Breastfeeding duration and obesity at 4 y among white children whose mothers did not smoke during pregnancy [Bogen et al., 2004]

BF durationBF duration AORAORNever 1.00< 8 wk 0.978-15 wk w/ FF 0.848-15 wk w/o FF 0.8016-26 wk w/ FF 0.8616-26 wk w/o FF 0.71*> 26 wk w/ FF 0.70*> 26 wk w/o FF 0.55** p < 0.05, adjusted for maternal age, education, parity, marital status, pregnancy

conditions, delivery method, child sex, birth weight, birth order, birth year. Inclusion of maternal BMI did not alter the results.

Breastfeeding and subsequent obesity: potential explanations

Learned self-regulation of energy intakeLearned self-regulation of energy intake Metabolic programmingMetabolic programming

InsulinInsulin LeptinLeptin Consequences of high protein intake in early Consequences of high protein intake in early

lifelife Residual confounding by attributes of mothers Residual confounding by attributes of mothers

and/or family environmentand/or family environment

Learned self-regulation of energy intake

Breastfeeding allows infant to control intake based Breastfeeding allows infant to control intake based on internal satiety cueson internal satiety cues

Bottle-fed infants may be encouraged to finish Bottle-fed infants may be encouraged to finish bottle even if they are fullbottle even if they are full

This may lead to later differences in ability to self-This may lead to later differences in ability to self-regulate energy intakeregulate energy intake

Infant self-regulation of breast milk intake K.G. Dewey & B. Lonnerdal Acta Paediatr Scand 1986; 75: 893-8

18 exclusively breastfeeding mothers stimulated 18 exclusively breastfeeding mothers stimulated milk supply by daily expression of extra milk for 2 milk supply by daily expression of extra milk for 2 wk. All but 4 increased milk volume by > 73 g/d.wk. All but 4 increased milk volume by > 73 g/d.

Among the 14 infants with access to increased Among the 14 infants with access to increased milk volume, most increased intake in the first 2 d, milk volume, most increased intake in the first 2 d, but returned to near baseline levels of intake after but returned to near baseline levels of intake after 1-2 wk1-2 wk

Intake increased more in fatter than leaner infantsIntake increased more in fatter than leaner infants Breastfed infants self-regulate milk intakeBreastfed infants self-regulate milk intake

Differences in milk intake between BF and FF infants increase between 1 and 5 mo

600700

800900

10001100

1200

1 mo 3 mo 5 mo

Age (mo)

Milk

Inta

ke (

ml/d

)

BF

FF-C

FF-MP

FF-LP

b

bb

a aa

b

b

b,c

ca

Dewey et al., EB 2004

Response to introduction of solid foods differs between breastfed and formula-fed infants

In BF infants, breast milk intake declines In BF infants, breast milk intake declines when solid foods are introducedwhen solid foods are introduced

In FF infants, formula intake does not In FF infants, formula intake does not decline when solid foods are introduceddecline when solid foods are introduced

Heinig et al., Acta Paediatr 1993;82:999-1006Heinig et al., Acta Paediatr 1993;82:999-1006

Effects of over-feeding in early life?

Animal studiesAnimal studiesIn baboons, overfeeding in infancy In baboons, overfeeding in infancy fat depot mass during puberty, especially fat depot mass during puberty, especially

in in females (Lewis et al., 1986)females (Lewis et al., 1986)

Human studiesHuman studiesRapid weight gain during infancy is Rapid weight gain during infancy is

correlated with childhood obesitycorrelated with childhood obesity(Ong et al., 2000; Stettler et al., 2002; (Ong et al., 2000; Stettler et al., 2002;

Cameron et al., 2003; Ekelund et al., 2006)Cameron et al., 2003; Ekelund et al., 2006)

Stettler et al. (U.S.)Pediatrics 2002;109:194-199

N=19,397 children born 1959-65N=19,397 children born 1959-65

Outcome: BMI > 95th percentile at age 7 yOutcome: BMI > 95th percentile at age 7 y

Rate of weight gain during the first 4 mo was Rate of weight gain during the first 4 mo was associated with risk of child obesity, even after associated with risk of child obesity, even after adjustment for weight at 1 yearadjustment for weight at 1 year

Almost 20% of obesity attributable to having a Almost 20% of obesity attributable to having a high rate of weight gain 0-4 mohigh rate of weight gain 0-4 mo

Weight gain in the first week of life and overweight in adulthood. Stettler et al. Circulation 2005;111:1897-1903.

N= 653 formula-fed infants, measured frequently N= 653 formula-fed infants, measured frequently during infancy & again at 20-32 y of ageduring infancy & again at 20-32 y of age

32% were overweight as adults32% were overweight as adults Weight gain during the first week of life was identified Weight gain during the first week of life was identified

as the most sensitive period regarding the association as the most sensitive period regarding the association with adult overweight: AOR for each 100-g increase with adult overweight: AOR for each 100-g increase was 1.28 (1.08-1.52) [adjusted for sex, birth weight, was 1.28 (1.08-1.52) [adjusted for sex, birth weight, type of formula, age at follow-up, maternal & paternal type of formula, age at follow-up, maternal & paternal weight status, income]weight status, income]

Weight gain during the first week of life ranged from 0 Weight gain during the first week of life ranged from 0 to 400 gto 400 g

Infant feeding, plasma insulin & weight gain

Formula-fed infants have higher plasma insulin Formula-fed infants have higher plasma insulin levels and prolonged insulin response at 6 d of age levels and prolonged insulin response at 6 d of age (Lucas et al., 1981)(Lucas et al., 1981)

Higher insulin levels stimulate greater fat Higher insulin levels stimulate greater fat deposition, and have been associated with deposition, and have been associated with subsequent subsequent weight gain & obesity in Pima Indian weight gain & obesity in Pima Indian children 5-9 y of age (Odeleye et al., 1997)children 5-9 y of age (Odeleye et al., 1997)

Infant feeding and plasma leptin

Plasma leptin is a key regulator of appetite and Plasma leptin is a key regulator of appetite and body fatnessbody fatness

Breastfeeding may affect leptin levels during Breastfeeding may affect leptin levels during infancy and later in lifeinfancy and later in life

Early diet of preterm infants is associated with Early diet of preterm infants is associated with leptin concentration at 13-16 y of age (Singhal et leptin concentration at 13-16 y of age (Singhal et al., 2002)al., 2002)

Ratio of leptin concentration to fat mass at 13-16 y of age, by tertile of human milk intake by preterm infants in early life (median + 95% CI, n=191, p = 0.006; Singhal et al., 2002)

Infant feeding and plasma leptin:postulated mechanism (Singhal et al., 2002)

Greater body fatness during infancy “programs” Greater body fatness during infancy “programs” the leptin-dependent feedback loop to be less the leptin-dependent feedback loop to be less sensitive to leptin later in life (i.e. greater leptin sensitive to leptin later in life (i.e. greater leptin resistance)resistance)

Greater leptin resistance contributes to overeating Greater leptin resistance contributes to overeating and obesityand obesity

In rats, overfeeding before weaning leads to In rats, overfeeding before weaning leads to overweight and leptin resistance in later life overweight and leptin resistance in later life (Plagemann et al., 1999)(Plagemann et al., 1999)

Early protein intake and subsequent body fatness Formula-fed infants consume 66-70% more protein than Formula-fed infants consume 66-70% more protein than

breastfed infants at 3-6 mo; by 12 mo, intakes may be 5-breastfed infants at 3-6 mo; by 12 mo, intakes may be 5-6 times the requirement6 times the requirement

High protein intake stimulates higher insulin secretion High protein intake stimulates higher insulin secretion adipose tissue depositionadipose tissue deposition

Association between high protein intake in early life and Association between high protein intake in early life and overweight in childhood reported by Rolland-Cachera et overweight in childhood reported by Rolland-Cachera et al. (1995) and Scaglioni et al. (2000), but not by Dorosty al. (1995) and Scaglioni et al. (2000), but not by Dorosty et al. (2000)et al. (2000)

Residual confounding? Child feeding practices & parental control over feedingChild feeding practices & parental control over feeding

Mothers who breastfed for Mothers who breastfed for >> 12 mo reported lower levels 12 mo reported lower levels of control over feeding at 18 mo (Fisher et al., 2000)of control over feeding at 18 mo (Fisher et al., 2000)

Duration of BF associated with less restrictive behavior Duration of BF associated with less restrictive behavior regarding child feeding at 1 year. Compared to FF regarding child feeding at 1 year. Compared to FF mothers, restrictive behavior much less likely among mothers, restrictive behavior much less likely among mothers who EBF for 6 mo [OR 0.27] (Taveras et al., mothers who EBF for 6 mo [OR 0.27] (Taveras et al., 2004)2004)

Highly controlling feeding practices may interfere with Highly controlling feeding practices may interfere with child’s ability to self-regulate energy intake (Birch et al., child’s ability to self-regulate energy intake (Birch et al., 2003)2003)

Residual confounding? (cont)

Physical activityPhysical activity Breastfeeding associated with healthier Breastfeeding associated with healthier

lifestyle, greater physical activity?lifestyle, greater physical activity? Some studies controlled for physical activity Some studies controlled for physical activity

and results were still significantand results were still significant

Residual confounding? Analysis of sibling pairsNelson et al. Epidemiology 2005;16:247-53. In full cohort, odds of being overweight decreased as BF duration In full cohort, odds of being overweight decreased as BF duration

increased, at least among girlsincreased, at least among girls In sibling pairs, no evidence of BF effectIn sibling pairs, no evidence of BF effect

Adjusted for age, sex, birth order and LBW statusAdjusted for age, sex, birth order and LBW status Did not have data on exclusivity of BFDid not have data on exclusivity of BF

Gillman et al. Epidemiology 2006;17:112-114. N=5614 siblings 9-14 y. Compared overweight in sibs BF longer than N=5614 siblings 9-14 y. Compared overweight in sibs BF longer than

mean for sibship with sibs BF shorter than meanmean for sibship with sibs BF shorter than mean Sibs who were BF longer (mean diff 3.7 mo) had lower OR for Sibs who were BF longer (mean diff 3.7 mo) had lower OR for

overweight: 0.94 for each 3.7 mo increase in BF durationoverweight: 0.94 for each 3.7 mo increase in BF duration OR for within-family analysis close to overall estimate, suggesting little OR for within-family analysis close to overall estimate, suggesting little

residual confoundingresidual confounding

Summary of BF & obesity studies

17 of the 21 studies showed an association 17 of the 21 studies showed an association between breastfeeding and a lower risk of obesitybetween breastfeeding and a lower risk of obesity

All of the studies that took into account the All of the studies that took into account the exclusivity of BF showed a significant associationexclusivity of BF showed a significant association

13 of the 16 studies that controlled for maternal 13 of the 16 studies that controlled for maternal BMI showed a significant associationBMI showed a significant association

In meta-analysis, duration of BF showed a dose-In meta-analysis, duration of BF showed a dose-response relationship with risk of child obesity. response relationship with risk of child obesity. Lowest risk was for > 6 mo of BFLowest risk was for > 6 mo of BF

Explanations?

Not solely due to lower fatness during first 2 yNot solely due to lower fatness during first 2 y Potential mechanisms include:Potential mechanisms include:

Learned self-regulation of energy intakeLearned self-regulation of energy intake Metabolic programming due to differences in Metabolic programming due to differences in

milk composition, protein intake, fatness and/or milk composition, protein intake, fatness and/or rate of weight gain in early liferate of weight gain in early life

Residual confounding, e.g. by child feeding Residual confounding, e.g. by child feeding practices, physical activitypractices, physical activity

Clinical & public health implications

Provides further evidence to promote Provides further evidence to promote breastfeedingbreastfeeding

However, role of breastfeeding is probably small However, role of breastfeeding is probably small compared to other factors such as parental compared to other factors such as parental overweight, dietary practices and physical activity overweight, dietary practices and physical activity

Relationship less evident in African-Americans Relationship less evident in African-Americans and Hispanics – not clear whyand Hispanics – not clear why

Documents

Breastfeeding and risk of child obesity Kathryn G. Dewey, PhD Program in International and Community Nutrition University of California, Davis