Outcomes of an Early Feeding Practices Interventionto Prevent Childhood Obesity

WHAT’S KNOWN ON THIS SUBJECT: About one in five 2-year-oldsare overweight, with potential adverse outcomes. Early feedingpractices lay the foundation for food preferences and eatingbehavior and may contribute to future obesity risk. High-qualityobesity prevention trials commencing in infancy are rare.

WHAT THIS STUDY ADDS: In this large randomized controlledtrial, anticipatory guidance on the “when, what, and how” ofcomplementary feeding was associated with increased maternal“protective” feeding practices. Differences in anthropometricindicators were in the expected direction but did not achievestatistical significance.

abstractOBJECTIVE: The goal of this study was to evaluate outcomes of a uni-versal intervention to promote protective feeding practices that com-menced in infancy and aimed to prevent childhood obesity.

METHODS: The NOURISH randomized controlled trial enrolled 698 first-time mothers (mean 6 SD age: 30.1 6 5.3 years) with healthy terminfants (51% female) aged 4.3 6 1.0 months at baseline. Mothers wererandomly allocated to self-directed access to usual care or to attendtwo 6-session interactive group education modules that providedanticipatory guidance on early feeding practices. Outcomes wereassessed 6 months after completion of the second informationmodule, 20 months from baseline and when the children were 2 yearsold. Maternal feeding practices were self-reported by using validatedquestionnaires and study-developed items. Study-measured childheight and weight were used to calculate BMI z scores.

RESULTS: Retention at follow-up was 78%. Mothers in the interventiongroup reported using responsive feeding more frequently on 6 of 9subscales and 8 of 8 items (all, P # .03) and overall less controllingfeeding practices (P , .001). They also more frequently used feedingpractices (3 of 4 items; all, P , .01) likely to enhance food acceptance.No statistically significant differences were noted in anthropometricoutcomes (BMI z score: P = .10) nor in prevalence of overweight/obesity (control 17.9% vs intervention 13.8%; P = .23).

CONCLUSIONS: Evaluation of NOURISH data at child age 2 years foundthat anticipatory guidance on complementary feeding, tailored to de-velopmental stage, increased use by first-time mothers of “protective”feeding practices that potentially support the development of healthyeating and growth patterns in young children. Pediatrics 2013;132:e109–e118

AUTHORS: Lynne Allison Daniels, PhD,a,b,c KimberleyMargaret Mallan, PhD,a,b Jan Maree Nicholson, PhD,d,e,f

Diana Battistutta, PhD,a and Anthea Magarey, PhDc,b

aInstitute of Health and Biomedical Innovation, bSchool ofExercise and Nutrition Sciences, and fCentre for LearningInnovation, Queensland University of Technology, Queensland,Australia; cDepartment of Nutrition and Dietetics, FlindersUniversity, Adelaide, South Australia, Australia; dParentingResearch Centre, Melbourne, Victoria, Australia; and eMurdochChildrens Research Institute, Melbourne, Victoria, Australia

KEY WORDSchildhood obesity, feeding practices, infant, randomizedcontrolled trial

ABBREVIATIONSCFQ—Child Feeding QuestionnairePFSQ—Parental Feeding Style QuestionnaireRCT—randomized controlled trial

Dr Daniels conceived the study; led the design, successfulfunding application, and overall implementation of the study;and wrote the first draft of the manuscript. Dr Mallan undertookthe statistical analysis under the mentorship of Dr Battistuttaand drafted the results and methods sections. Dr Nicholsoncontributed to funding applications, intervention development,and design of outcome assessments. Dr Battistutta contributedto the design, funding applications, and implementation, andprovided mentorship to Dr Mallan. Dr Magarey contributed tothe design, funding applications, and development of theintervention, and led implementation at the Adelaide site. Allauthors contributed to interpretation of the results andpreparation of the manuscript.

This trial has been registered in the Australian and New ZealandClinical Trials Registry (ACTRN 12608000056392).

www.pediatrics.org/cgi/doi/10.1542/peds.2012-2882

doi:10.1542/peds.2012-2882

Accepted for publication Mar 27, 2013

Address correspondence to Lynne Allison Daniels, PhD, School ofExercise and Nutrition Sciences, Queensland University ofTechnology, Victoria Park Rd, Kelvin Grove, Queensland, Australia,4059. E-mail: l2.daniels@qut.edu.au

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright © 2013 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: The authors have indicated they haveno financial relationships relevant to this article to disclose.

FUNDING: NOURISH was funded from 2008 to 2010 by theAustralian National Health and Medical Research Council (grant426704). Additional funding was provided by HJ Heinz(postdoctoral fellowship, Dr Mallan), Meat & Livestock Australia,Department of Health South Australia, Food Standards AustraliaNew Zealand, Queensland University of Technology, and NationalHealth and Medical Research Council Career DevelopmentAward (390136, Dr Nicholson).

PEDIATRICS Volume 132, Number 1, July 2013 e109

ARTICLE

by guest on March 3, 2021www.aappublications.org/newsDownloaded from

The rationale for early interventionsthat target feeding practices as an ef-fective approach to obesity preventionis strong.1–4 Nevertheless, there arevery few randomized controlled trials(RCTs) that have commenced in in-fancy.5,6 The NOURISH RCT7 evaluated anobesity prevention intervention thatprovided anticipatory guidance onearly feeding to first-time mothers andcommenced when the infants were 4months old. Our overarching hypothe-sis was that “protective” early feedingpractices can support the developmentof healthy child eating habits and po-tentially confer some resilience to thecontemporary “obesogenic” environ-ment. The intervention comprised 2interactive 6-session parent educationmodules with both shared and uniquecontent matched to developmentalage at commencement of each mod-ule (mean age: 4 and 13 months, re-spectively). Evaluation of the firstmodule alone (infants aged 14 months,before commencement of module 2)has been reported.8 We now reportshort-term outcomes, specifically ma-ternal feeding practices (impact eval-uation) and child anthropometric data(outcome evaluation), 6 to 8 monthsafter completion of the total plannedintervention (ie, both educationmodules) when children were 2 yearsold.

We predicted that, compared with self-directed access to usual care, receivinganticipatory guidance on early feedingpractices would demonstrate increaseduse by first-time mothers of: (1) foodexposure practices postulated to pro-mote the development of food prefer-ences consistent with healthy dietaryintake9–13; and (2) responsive feedingbehaviors that support child self-regulation of intake.4,14,15 A secondaryhypothesis was that the interventionwould result in lower anthropometricindicators of obesity risk at 2 years ofage.

METHODS

Study Design

The NOURISH RCT commenced in 2008across2Australiancities (BrisbaneandAdelaide). The protocol has been pub-lished elsewhere.7,16 The outcome as-sessment at 20 months from baselinewhen the children were 2 years old(range: 21–27 months) is reportedhere. The trial was approved by theQueensland University of TechnologyHuman Research Ethics Committee.

Recruitment, Participants, andAllocation

The 2-stage recruitment strategy hasbeen described elsewhere.7,16 A con-secutive sample of first-time mothers($18 years old) of healthy term infants(.35 weeks’ gestation, $2500 g birthweight) was approached at 7 maternityhospitals across both cities. Additionalinclusion criteria subsequently as-sessed were facility with written andspoken English and no documented orself-reported history of intravenoussubstance abuse, domestic violence, oreating disorders. We conceptualizedthis study as an efficacy trial, and theintent was to optimize the potentialintervention effect. Mothers who con-sented to later contact provided de-mographic and other data (Table 1) andwere recontacted for full enrollment

when their infant was 4 months old(range: 2–7 months). Assessments andintervention delivery occurred in com-munity child health clinics locatedacross each city. After baseline assess-ment, participants were randomly allo-cated to an intervention or a controlgroup by an external statistician. Apermutated-block schedule with blocksof 4 within each assessment clinic lo-cation was used to optimize the balanceof participant socioeconomic charac-teristics across the groups. Participantswere not compensated for time ortravel to attend intervention or assess-ment sessions.

Treatment Components

Modules 1 and 2 commenced immedi-ately after baseline, when the childrenwere aged 4 to 7 months and 13 to 16months, respectively. Each modulecomprised 6 interactive group sessionsof1 to1.5hoursduration, deliveredover12 weeks (40 groups across bothmodules and sites). Sessions werecofacilitated by a dietitian (n = 13) anda psychologist (n = 13). Facilitatorsreceived standardized training, used acomprehensive facilitator manual andstandard presentation materials, andparticipated in fortnightly supervisionteleconferences to promote interventionquality and integrity.

TABLE 1 Characteristics of 2094 First-Time Mothers who Consented at Stage 1 and Were Allocatedor not Allocated at Stage 2

Variablea Allocated(n = 698)

Not Allocatedb (n = 1396)

Consented Did not Consent(n = 885)

Could not Recontact(n = 511)

Maternal age at delivery, yc (n = 2087) 30.1 6 5.3 28.0 6 5.5 26.2 6 5.5Maternal education (university degree)d (n = 2078) 58 (406) 36 (311) 27 (137)Born in Australia/New Zealandd 78 (542) 77 (667) 75 (376)Married/de factod (n = 2062) 95 (659) 90 (778) 83 (421)Intend to breastfeed exclusivelyd (n = 2088) 93 (652) 90 (794) 87 (441)Smoked during pregnancyd (n = 2081) 12 (85) 21 (185) 32 (164)

n values given in parentheses reflect missing data. Stage 1, when participants were first approached in hospital postdelivery;Stage 2, second contact, infants aged 2 to 7 months.a Based on data provided at stage 1.b Excluding an additional 75 participants who became ineligible.c Mean 6 SD.d Proportion % (count) reported.

e110 DANIELS et al by guest on March 3, 2021www.aappublications.org/newsDownloaded from

The intervention was informed by 3theoretical models. Attachment theorycenters on maternal sensitivity toinfant cues14 and hence provides aframework for responsive feeding.4

Anticipatory guidance is a proactiveapproach that provides parents withinformation about behaviors they canexpect and constructive ways tomanage these behaviors, rather thanwaiting until parents seek advice onestablished problems.15 A social cog-nitive approach17 informed interven-tion activities to promote maternalself-efficacy, competence, and confi-dence to adopt program recommen-dations. Strategies included progressivegoal setting, identification of facilitators/barriers to implementation, self-monitoring and review, and individ-ualized problem solving. It was pilottested in 25 mothers.

Content provided anticipatory guid-ance, targeted to developmental stage,on 3 aspects of early feeding associatedwith positive outcomes in children’seating behavior and weight status: (1)exposure to a wide range of texturesand tastes to promote development ofhealthy food preferences9–13 (module1); (2) responsive feeding that recog-nizes and responds appropriately toinfant cues of hunger and satiety topromote self-regulation of energy in-take to need4,18,19 (module 2); and (3)positive parenting (warmth, encourage-ment of autonomy, and self-efficacy)20–22

(modules 1 and 2). Content as presentedto mothers focused on healthy eatingpatterns and growth, rather than obesityprevention. All intervention participantswere provided with detailed written in-formation that covered session content.The control group had standard accessto universal community child healthservices, which, at the mothers’ initia-tive, could include child weighing andinformation via the Web or the telephonehelp line. An important distinction wasthat controls sought advice in response

to an existing need/problem and hencedid not receive anticipatory guidance. Nodata were collected on the frequencywith which mothers accessed standardcare.

Outcome Measures

Demographic and behavioral datawerecollected by using self-completedquestionnaires at first contact (face-to-face), baseline, and follow-up (bymail).

Feeding Practices

TheChild FeedingQuestionnaire (CFQ)23

is the most widely used tool to assessparents’ attitudes and practices re-lated to feeding children aged 2 to11 years. Five of the 7 subscales wereincluded to assess controlling feed-ing practices (restriction, pressure toeat, and monitoring) and perceivedresponsibility and concern about childweight. The Parental Feeding StyleQuestionnaire (PFSQ)24 was also usedbecause it assesses additional feedingpractice constructs that reflect parentuse of (non-)responsive feeding. Foursubscales were included: instrumentalfeeding, encouragement, emotionalfeeding, and control over eating. In-ternal consistency estimates in oursample were equivalent or higher thanthose reported in the original valida-tion studies. Items based on clinicalexperience of the investigators andused previously25 examined mothers’overall perceptions of their child’seating behavior and specific strategiesthey used in response to refusal offamiliar foods (cues of satiety) or un-familiar foods (neophobia). Twoadditionalitems assessed Satter’s conceptuali-zation of responsive feeding (“parentprovide, child decide”)26: (1) Whodecides what your child eats—you oryour child?; and (2) Who decides howmuch food your child eats—you oryour child? Responses were scored as1 = you only to 5 = your child only.

Anthropometric measurements weretaken by trained study staff blinded toparticipant group and independent ofintervention delivery. Duplicate childnaked weight and recumbent length(baseline) or standing height (follow-up) and single maternal height andweight were measured by using stan-dard clinic equipment. Standardizedweight-for-age, length/height for age,and BMI-for-age z scores were calcu-lated by using the World Health Orga-nization Anthro version 3.0.1 andmacros program. As recommended,0.7 cm was added to the follow-upstanding height of children aged ,2years to correct for use of recumbentlength in the reference sample.27

Classification as overweight (includingobese) was based on InternationalObesity Task Force gender-specific2-year-old BMI cutoffs.28 Birth weightwas obtained from hospital records.

Covariates

Covariate data were collected at firstcontact in the maternity hospitals from2094 mothers who consented to latercontact (Table 1). Birth weights werecollected from hospital records. Socio-economic status was determined byusing the Socio-Economic Indexes forAreas score for the Index of RelativeAdvantage and Disadvantage. Scoresbelow the seventh decile (sample me-dian) indicated relative disadvantage.29

Statistical Analysis

Sample size calculationswere based onexpected meaningful differences at the2-year-old follow-up in selected impactoutcomes that included a subset ofprotective feeding practices reportedhere. Details of the anticipated differ-ences based on our pilot study of chil-dren aged 12 to 36 months25 areprovided in the protocol article.7 As-suming 80% power and a type I error of5% (2-tailed), we sought to have 265subjects per group at follow-up and to

ARTICLE

PEDIATRICS Volume 132, Number 1, July 2013 e111 by guest on March 3, 2021www.aappublications.org/newsDownloaded from

enroll 830 based on an anticipated 35%attrition rate. Anthropometric vari-ables were considered as secondaryoutcomes in the original protocol andwere excluded from sample size cal-culations.7 At the time the study wasplanned (2006), there were no datafrom relevant interventions commenc-ing in infancy to enable effect sizeestimations as a basis for sample sizecalculation.

An intention-to-treat approach to anal-ysis was used as far as missing datapermitted (no imputationsweremade).There was no evidence of differencesby group (intervention versus control)at baseline (Table 2). Accordingly, noadjustment for covariates was un-dertaken, and comparisons betweengroups on continuous and dichoto-mous outcome variables used inde-pendent sample t tests and likelihoodratio x2 tests, respectively. The CFQ23

restriction, pressure to eat, and moni-toring subscale means were examined

by using a multivariate analysis ofvariance to test for an overall inter-vention effect on controlling feedingpractices and to statistically accountfor the close theoretical relatedness ofthese subscales.

All statistical tests were computed byusing SPSS version 19 (IBM SPSSStatistics, IBM Corporation, Armonk,NY). A P value of .05 (2-tailed) wasused throughout to indicate statisti-cal significance.

RESULTS

Participant flow is shown in Fig 1, andTable 1 displays the characteristics ofmothers who participated in the trialversus those who agreed to a secondcontact but either could not be recon-tacted or declined enrollment. Therewere no differences according togroup allocation at baseline (Table 2).

At follow-up (20months from baseline),total attrition was 22% (n = 157). Rea-

sons for withdrawal were no longerinterested (n = 14), returned to work(n = 11), poor health of child or family(n = 11), or did not need feeding advice(n = 4). Fifteen participants moved outof the region, and 1 child was de-ceased. Withdrawal was higher amongyounger and less educated mothers(all, P , .001) (Table 3) and in the in-tervention group (26%, n = 92) than inthe control group (19%, n = 65; P = .01),but it did not differ according tomothers’ BMI (P = .51). However, therewere no differences by group in thebaseline characteristics of those whowithdrew (data not shown). Atten-dance at$2 sessions formodule 1 wasn = 229 (65%) and module 2 wasn = 130 (45% of those retained atmodule commencement). At follow-up,the mean age of children (52% female)was 24.16 0.7 months, 8% of motherswere still breastfeeding (control: n = 19;intervention: n = 16), and 25% hada second child (control: n = 50; in-tervention: n = 65).

Maternal Feeding Practices

Across the measures of maternalfeeding, there were significant differ-ences by group in the expected direc-tions on 6 of 9 subscales from the CFQ23

and PFSQ24 (Table 4) and on 11 of 12individual items assessing responsesto food refusal (Table 5). For example,interventionmothers usednonresponsivefeeding practices significantly less of-ten and responsive feeding practicesmore often (P values, .033 to ,.001).There was no difference according togroup in the proportion of mothersreporting that they were “mostly/only”responsible for what their child eats(intervention: 72%, n = 159; control:76%, n = 187 [P = .29]). However, moreintervention mothers than controlmothers (82% [n = 183] vs 49% [n =120]) reported that their child wasmostly/only responsible for howmuch toeat (P, .001). Specific feeding practicesin response to child neophobia are

TABLE 2 Characteristics of Mothers and Children Allocated to the Control Group Compared Withthe Intervention Group

Variable Control(n = 346)

Intervention(n = 352)

Total(N = 698)

MotherEducation (university degree) 58 (199) 59 (207) 58 (406)Smoked during pregnancy 11 (40) 13 (45) 12 (85)Born in Australia 79 (270) 78 (272) 78 (542)Married/de facto 95 (327) 95 (332) 95 (659)SEIFA for the Index of Relative Advantage and

Disadvantage (relative disadvantage#seventh decile)34 (117) 32 (113) 33 (230)

Age at delivery, y 29.9 6 5.3 30.2 6 5.3 30.1 6 5.3BMI 26.2 6 5.5 25.8 6 5.1 26.0 6 5.3

InfantGender (female) 50 (173) 51 (181) 51 (354)Birth weight, kg 3.5 6 0.4 3.5 6 0.4 3.5 6 0.4Birth weight z scorea 0.38 6 0.87 0.39 6 0.88 0.38 6 0.87Age (months) at baseline assessment 4.3 6 1.0 4.3 6 1.0 4.3 6 1.0Current feeding modeb

Fully/exclusively breast-fed 55 (170) 60 (191) 57 (361)Formula only 27 (83) 26 (84) 27 (167)Combination (formula + breast-fed) 19 (59) 14 (44) 16 (103)Ever breast-fedb 96 (266) 98 (250) 97 (516)Ever given solidsb 34 (114) 34 (115) 34 (229)

Age solids introduced (weeks)c 22.7 6 4.9 22.8 6 4.4 22.8 6 4.7

% within group (count) reported for dichotomous variables and mean 6 SD reported for continuous variables.SEIFA, Socio-Economic Indexes for Areas.28a World Health Organization standards.36b Data collected from questionnaire administered at baseline.c Data collected from questionnaire administered at first follow-up visit when infants were aged 14 months (n = 529).

e112 DANIELS et al by guest on March 3, 2021www.aappublications.org/newsDownloaded from

shown in Table 5. There were no differ-ences in the proportion of toddlers re-portedly often/very often refusing food(22% [n = 102]). However, interventionmothers were more likely than controlmothers to “strongly agree/agree” thattheir child was easy to feed (83% [n =184] vs 75% [n = 183]; P = .03).

Anthropometric Outcomes

Child anthropometric data at baselineand follow-upareshown inTable6.Usinggender-specific international BMI cut-offs,30 13.8% (34 of 246) of interventionchildren versus 17.9% (49 of 274) ofcontrol children were classified asoverweight or obese (P = .23).

DISCUSSION

NOURISH is one of the first and largestRCTs to report outcomes of an in-tervention to prevent childhood obe-sity that commenced in infancy andexplicitly targets maternal feedingpractices. We found that anticipatoryguidance on early feeding, tailored todevelopmental stage, increased use byfirst-time mothers of protective feed-ing practices that potentially supportthe development of healthy eating andgrowth patterns in their 2-year-oldchildren. Anthropometric differenceswere consistently in the expected di-rection; however, effect sizes did notachieve statistical significance.

In healthy toddlers, refusal of familiar,usually accepted foods generally sig-nals satiety.8,25 The responsive feedingresponse is to trust the child’s appetiteand to interpret refusal as satiety.18

Nonresponsive feeding fails to acceptthat the child has eaten enough and ischaracterized by excess overt mater-nal control.2,4 Specific nonresponsivefeeding practices include pressure andactive encouragement to eat moreby using a range of strategies suchas coaxing (eg, game playing), usingrewards (food or nonfood), or offeringfavorite foods as alternatives. Praise

FIGURE 1CONSORT diagram showing flow of all participants. aSome provided multiple reasons.

TABLE 3 Characteristics of Allocated Mothers (N = 698) who Completed the Study orDiscontinued

Variablea Completed (n = 541) Discontinuedb (n = 157)

Maternal age at delivery, yc 30.6 6 5.2 28.0 6 5.5Maternal education (university degree)d 63 (343) 40 (63)Born in Australia/New Zealandd 81 (438) 82 (129)Married/de factod (n = 696) 92 (516) 92 (143)Intend to breastfeed exclusivelyd (n = 694) 94 (509) 92 (143)Smoked during pregnancyd (n = 696) 12 (62) 15 (23)BMI (n = 691)c 25.9 6 5.3 26.3 6 5.4

n values given in parentheses reflect missing data.a Based on data provided at stage 1 except BMI, which was measured at baseline when infants were aged 4 months.b Discontinued due to either active withdrawal or could not be contacted.c Mean 6 SD.d % within group (count).

ARTICLE

PEDIATRICS Volume 132, Number 1, July 2013 e113 by guest on March 3, 2021www.aappublications.org/newsDownloaded from

for eating and using food to comfort,distract, or reward good behavior(emotional feeding) are also non-responsive feeding strategies in thatthey encourage the child to eat forreasons unrelated to appetite.25 Over-all, these nonresponsive feeding prac-tices are postulated to result indiscordant feeding in which childrenare habitually encouraged to eat morethan they want and need.4 As such, theyundermine intrinsic regulation of in-take and in the longer term result inpositive energy balance and excessweight gain.4,19,25,31,32 On the PFSQ,24

intervention mothers used lower levelsof instrumental and emotional feedingand prompting encouragement. In ad-dition, differences according to studygroup for all 8 items assessing re-sponses to refusal of familiar foodsindicated that intervention mothers

were more likely to interpret foodrefusal as a signal of satiety and lesslikely to use nonresponsive or coercivepractices such as insisting their childeat or offer a reward for eating. Theseresults indicate that maternal behav-iors aremodifiable. Overall, interventionmothers more frequently used a rangeof responsive feeding practices.

Intervention mothers overall used lesscontrolling feeding practices. On theCFQ,23 intervention mothers had lowerscores on “pressure” and “restriction”subscales, which, at an item level,predominantly assess controlling ma-ternal behaviors in relation to what orhow much the child eats. These resultswere consistent with a higher pro-portion of intervention mothers (82%)compared with control mothers (49%)reporting their child was mostly/onlyresponsible for deciding how much to

eat. Three of the feeding subscales didnot show significant differences bygroup at follow-up: CFQ perceived re-sponsibility and monitoring23 and PFSQcontrol.24 Inspection of the items inthese scales provides some insight intothis apparent inconsistency. In thePFSQ control subscale, 9 of 10 itemsrefer to maternal versus child roles indetermining the timing and location ofeating. Similarly, the CFQ perceivedresponsibility subscale comprises 3items assessing mothers’ responsibilityfor feeding at home, the kind of foodoffered, and portion size. Thus, thesescales seem to assess the “where” and“when” and responsibility for feeding,rather than recognition of and responseto child satiety cues. The CFQmonitoringsubscale assesses maternal tracking oftheir child’s consumption of unhealthyfoods, which is plausibly an antecedentto, rather than a feeding practice. Theabsence of group differences on these 3feeding subscales is consistent withthese “practices” not being a focus ofthe intervention. The high mean scalescores for monitoring and responsi-bility indicate high levels of these prac-tices and are not unexpected inmotherswho volunteer for a feeding trial. Thesevariable findings highlight the defini-tional and measurement issues existingin this field of research, which not onlymake it challenging to interpret find-ings within studies but also confoundbetween-study comparisons.21

Refusal of novel or unfamiliar foods(neophobia) is very common in tod-dlers.11 A key intervention message wasthat children need to be explicitly taughtto like new foods through repeatedneutral exposure. Intervention motherswere more likely to respond to refusalof new foods with strategies likely toincrease familiarity, acceptance, and in-take, thus supporting increased dietaryvariety and quality.9,11,33 We found nogroup differences in the frequency oftoddler food refusal; however, mothers

TABLE 4 Child-Feeding Practices at Follow-up Visits of Mothers Enrolled in the NOURISH Trial

Maternal Feeding Practicesa Control(n = 245)

Intervention(n = 222)

P

CFQPerceived responsibility (3 items, a = 0.80); for example, How

often are you responsible for deciding if your child haseaten the right kind of foods?

4.46 0.5 4.3 6 0.6 .61

Concern child is overweight (3 items, a = 0.76); for example,How concerned are you about your child eating too muchwhen you are not around her?

1.36 0.6 1.2 6 0.4 .016

Controlling feeding practicesb ,.001Restriction (8 items, a = 0.75); for example, I have to be

sure my child does not eat too many sweets3.06 0.7 2.9 6 0.8 .055

Pressure to eat (4 items, a = 0.77); for example, If my childsays “ I am not hungry’ I try to get her to eat anyway

2.36 1.0 1.8 6 0.9 ,.001

Monitoring (3 items, a = 0.93); for example, How much doyou keep track of the snack foods that your child eats?

4.36 0.9 4.3 6 1.0 .76

PFSQInstrumental feeding (4 items, a = 0.77); for example, I

reward my child with something to eat when she is wellbehaved

1.66 0.5 1.4 6 0.5 ,.001

Encouragement (8 items; a = 0.75); for example, I praise mychild if she eats what I give her

4.06 0.5 3.9 6 0.5 .005

Emotional feeding (5 items, a = 0.81); for example, I give mychild something to eat to make him feel better when he isupset

1.66 0.5 1.5 6 0.5 .039

Control over eating (10 items, a = 0.72); for example, I let mychild decide when he would like to have his meal

3.96 0.4 4.0 6 0.4 .56

Mean child age overall was 24.16 0.7 months; 52% were female. n values given in parentheses indicate missing data; a isCronbach’s a. CFQ20 response options: perceived responsibility and monitoring, 1= never to 5 = always; pressure andrestriction, 1 = disagree to 5 = agree; concern, 1 = unconcerned to 5 = very concerned. PFSQ21 response options: 1 = never to5 = always.a Continuous variables based independent samples t tests; mean 6 SD reported.b Between-group difference on related constructs tested via multivariate analysis of variance, F(3,463) = 9.023, P , .001,hp

2=0.055.

e114 DANIELS et al by guest on March 3, 2021www.aappublications.org/newsDownloaded from

in the intervention group were lesslikely to interpret their child as beinga difficult eater. This finding suggeststhat the intervention anticipatoryguidance assisted mothers to expectand understand neophobia as normalbehavior and hence to be less con-cerned and to respond appropriately.This finding is supported by themothers in the intervention groupalso reporting less food fussiness intheir children (data not shown).

Very few RCTs evaluating interventionsto reduce childhood obesity risk have

commenced before 12months of age.5,6

Only 2 comparable studies have re-ported outcomes of early feeding obe-sity prevention interventions, both ofwhich were delivered via nurse-ledhome visits. Paul et al34 reported out-comes at 12 months of age (n = 110;69% retention) of a 3-visit interventionthat targeted nonfeeding soothingstrategies and/or the timing and pro-cess of solid introduction. Wen et al,35

in another large Australian study (HBT[Healthy Beginnings Trial]), reportedoutcomes at 2 years of age (n = 497;

75% retention) after 6 visits. Bothstudies34,36 reported improvements inbreastfeeding and timing of solid in-troduction, but no comparable feed-ing practices data were provided. Usinga single-item measure, Wen et al35

reported that intervention motherswere less likely to use food as a reward(62% vs 72%) with their 2-year-oldchildren.

Despite strong and consistent inter-vention effects on maternal feedingpractices, we failed to find statisticallysignificant group differences in growthor weight status. Because we wereunable to postulate a priori theexpected meaningful differences forthese measures, a type II error is pos-sible. Nevertheless, differences in an-thropometric data were all in theexpected directions. Paul et al34 re-ported lower mean weight-for-lengthpercentiles (33rd vs 50th; n = 22 vsn = 30) at 12 months of age (69% re-tention) for the combined intervention.The HBT35 showed a small (2.3%) butstatistically significant standardizedmean difference (intervention – control)of –0.23 BMI units (n = 249 and 234;P = .01) at 2 years compared with ourresult of –0.14 (n = 279 and 251; P = .12).As in NOURISH, HBT found no significantintervention– control difference inweight(–0.17 vs –0.16 kg, respectively) orlength (0.31 vs –0.15 cm, respectively).

TABLE 5 Child-Feeding Strategies Used in Response to Food Refusal at Follow-up of MothersEnrolled in the NOURISH Trial

Variablea Control(n = 245)

Intervention(n = 222)

P

Response to refusal of familiar foodsb

Nonresponsive feeding strategies: override satiety cuesInsist child eats it (n = 466) 37 (90) 18 (39) ,.001Offer milk drink instead (n = 464) 22 (53) 14 (30) .022Offer liked food instead (n = 465) 78 (189) 63 (140) .001Encourage to eat: turn mealtime into game (n = 466) 57 (139) 21 (47) ,.001Offer food reward (n = 465) 31 (75) 9 (19) ,.001Encourage to eat: offer nonfood reward (n = 464) 27 (65) 18 (39) .026

Responsive feeding strategies: respond appropriately to satiety cuesOffer no food until next usual meal/snack time (n = 463) 48 (116) 67 (147) ,.001Accept that child may not be hungry; take food away (n = 464) 91 (222) 96 (213) .033

Response to refusal of unfamiliar foods (neophobia)c

Assume child dislikes; do not offer again (n = 457) 13 (32) 5 (11) .003Disguise food (n = 456) 65 (156) 45 (98) ,.001Offer with liked food (n = 462) 94 (229) 94 (206) .99Times offered a food before deciding whether liked ($ 6 times) (n = 465) 35 (87) 72 (159) ,.001

Mean child age overall was 24.1 6 0.7 months; 52% were female. n values given in parentheses indicate missing data.P values based on likelihood ratio chi-square test.a Dichotomous variables; % within group (count) reported.b Response options: 1 = never to 5 = most of the time, dichotomized to 1, 2 versus 3, 4, 5 and % reported.c Response options: 1 = never to 4 = often, dichotomized to 1, 2 versus 3, 4 and % reported.

TABLE 6 Anthropometric Data at Birth, Baseline, and Follow-up for Children Enrolled in the NOURISH Trial

Variable Baseline Difference Follow-up Difference

Control Intervention Intervention – Control Control Intervention Intervention – Control

Mean 6 SD Mean (SE) P Mean 6 SD Mean (SE) P

N 346 352 279 251Age, mo 4.3 6 1.0 4.3 6 1.0 20.01 (0.07) .89 24.1 6 0.8 24.1 6 0.6 20.06 (0.06) .33Weight, kg 6.84 6 0.96 6.83 6 1.01 20.01 (0.07) .93 12.94 6 1.55 12.78 6 1.56 20.16 (0.14) .23Weight z score 20.03 6 0.91 20.04 6 0.93 20.01 (0.07) .95 0.69 6 0.91 0.58 6 0.98 20.11 (0.08) .20

N 345 349 274 246Length/height (cm) 64.06 6 3.04 64.26 6 3.15 0.20 (0.24) .39 87.35 6 3.24 87.20 6 3.16 20.15 (0.28) .60Length/height z score 0.27 6 0.95 0.39 6 0.98 0.11 (0.07) .12 0.27 6 0.99 0.24 6 0.96 20.02 (0.09) .78BMI 16.61 6 1.48 16.46 6 1.48 20.15 (0.11) .17 16.94 6 1.49 16.74 6 1.43 20.20 (0.13) .12BMI z score 20.26 6 0.98 20.36 6 0.98 20.10 (0.07) .18 0.75 6 0.98 0.61 6 1.01 20.14 (0.09) .10

P value for test of difference between conditions using independent sample t test; mean difference (intervention – control) and SE of difference reported. Weight available at baseline forn = 696; recumbent length available at baseline for n = 694; female 51%. Weight available at follow-up for: n = 530/541 retained; standing height available at follow up for n = 520/541 retained;female 51%.

ARTICLE

PEDIATRICS Volume 132, Number 1, July 2013 e115 by guest on March 3, 2021www.aappublications.org/newsDownloaded from

It is possible that in HBT, the relativelyhigher intervention length (as de-nominator) may have contributed tothe significant difference in BMI. HBTalso reported a 2.9% point lowerprevalence of overweight in theirintervention group compared witha 4.1% point lower prevalence foundin NOURISH. Although not statisticallysignificant (P = .23), if translated to apopulation level, this difference wouldbe important from a public healthand longer term obesity preventionperspective.

Thereareseveral possible explanationsfor the differences in BMI-for-age zscore outcomes between the NOURISHand HBT35 trials. HBT was conductedin a socially disadvantaged area, withmothers who were younger (75% ,30years old) and less well educated(24% university degree). Given thesociodemographic differentials in health-related behaviors and obesity preva-lence,37 there may have been morescope for intervention response inthe HBT sample. The intervention ap-proaches varied with respect to age atcommencement (HBT: 1 month; NOURISH:4 months), the relative intensity (dose)of intervention, the home visit versusgroup delivery format, and the extentto which there was a focus on breast-feeding maintenance and physicalactivity (both greater in HBT). The 2approaches have been evaluated withindifferent populations and service de-livery modalities. Together, they pro-vide evidence of the potential capacityof interventions commencing in in-fancy to reduce early obesity risk.Much more work is clearly required tounderstand the most effective inter-vention components for different pop-ulations and to determine the longerterm effectiveness. A planned prospec-tive meta-analysis (EPOCH [Early Pre-vention of Obesity in Children])38 willinclude data from both trials to beginto answer these questions.

Our midterm evaluation of the firstmodule (infants aged 14 months)showed positive intervention effects on5 of 12 specific feeding strategies (Ta-ble 5) relevant to responsive feedingand neophobia.8 By 2 years of age, theintervention effect on maternal feedingpractices seems to have strengthened,with group differences evident for 11 of12 of these same strategies. (The sub-scales reported in Table 4 have notbeen validated for use with childrenaged ,2 years and were not used inthe midterm follow-up at 14 months.)However, this relative strengthening ofintervention effect on maternal feedingpractices does not seem to havetranslated to commensurate enhancedeffects on anthropometric outcomes.At 2 years of age, we found a stan-dardizedmean difference (intervention– control) in BMI-for-age z scores of–0.14 (n = 274 and 246; P = .10) com-pared with –0.22 (n = 292 and 273;P, .01) at 14months of age.8 There area number of plausible explanations.The intervention focused on intrinsicdrivers of eating habits, namely foodpreferences and appetite regulation.It is possible that intervention effectson anthropometric factors may notmanifest until children become moreindependent and face the challenges ofthe obesogenic environment beyondpredominantly maternal control. Inaddition, interindividual variations inrates of weight gain are high from 0 to2 years. From ages 2 to 5 years, growthparameters show less variability, bet-ter tracking, and more directly reflectthe interaction between environmentalfactors and genetic propensity forobesity.39–42 Our planned follow-upcontacts at age 3.5 and 5 years willexamine whether significant benefitsto weight accrue over the longer term.

Strengths of NOURISH include its designand implementation according toCONSORT guidelines,43 large samplesize, and good retention. Researchers

were blinded to group allocation foroutcome assessment and analysis. Theuse of group education sessions wasconsistent with programs delivered inthe relevant community child healthsectors at the time of the study. Thelack of a true attention control groupdoes not allow us to preclude a Haw-thorn effect. However, this option wasnot feasible in terms of cost, partici-pant burden, and identification of 18hours of authentic content that wouldnot potentially affect parenting andobesity risk. Self-report behavioraldata always have the potential for ac-quiescence bias but are the only fea-sible option in large population-basedstudies. Our consecutive samplingframework, an approach rarely used inobesity prevention trials, enabled de-tailed assessment of selection and re-tention biases. This method revealedevidence of both, with the most im-portant difference being higher levelsof education among mothers whoconsented and completed. However,these biases do not compromise theinternal validity of the trial. Given thatour participants were comparativelywell-educated, first-time mothers, mostof whom were born in Australia, thewider generalizability of the interven-tion and its effectiveness are unknown.Although attrition was higher in theintervention group than in the controlgroup, the characteristics of non-completers did not vary by group.Attendance at module 2 was disap-pointing. It is possible that moduleattendance has led to an under-estimation of effect size, but theCONSORT-recommended intention-to-treat analytical approach has pro-vided the most unconfounded, albeitconservative, estimation we could ach-ieve. Although all intervention par-ticipants received detailed writteninformation, these data suggest thatdifferent delivery formats (eg, tele-phone or Web-based) which are moreconvenient for mothers faced with the

e116 DANIELS et al by guest on March 3, 2021www.aappublications.org/newsDownloaded from

demands of caring for a toddler needto be explored. We acknowledge thatmultiple comparisons do not allow usto exclude the potential of a type I error.However, given that we reported sig-nificant group differences in feedingpractices consistent with interventionfocus in 17 of 21 possible feeding sub-scales or items, this source of type Ierror is unlikely to alter our overallconclusions.

CONCLUSIONS

NOURISH is an obesity prevention in-tervention that focused on maternalfeeding practices and commenced inthefirst fewmonths of life. It representsan important advance in efforts toprevent pediatric obesity.6 Across sev-eral measures, mothers in the in-tervention group consistently reportedhigher levels of responsive feeding

practices and lower levels of non-responsive feeding practices with their2-year-old children. Overall, anticipa-tory guidance on the “when, what, andhow” of infant feeding resulted inincreased use of protective feedingpractices that potentially support ex-panded food preferences and childself-regulation of intake, clearly in-dicating that maternal feeding practi-ces are modifiable. The interventioneffects on maternal behaviors did nottranslate into statistically significantdifferences in anthropometric childoutcomes at 2 years of age. The 4.1%point reduction in prevalence ofoverweight/obesity found here trans-lated to population level would rep-resent an important reduction inprevalence and longer term obesityrisk. It is plausible that the extent towhich protective feeding practices that

focus on intrinsic determinants ofeating habits, such as food preferencesand appetite regulation, can conferresilience to the contemporary obeso-genic environment may not manifestuntil the child is older. Further follow-up when the children are 3.5 and 5years old is underway to shed light onthe longer term efficacy of obesityprevention interventions that start ininfancy and target maternal feedingpractices.

ACKNOWLEDGMENTSWeacknowledge theNOURISH investiga-tors: Professors Ann Farrell, GeoffreyCleghorn, and Geoffrey Davidson. Wesincerely thank all our participants,recruiting staff, and study staff, includ-ing Dr Rebecca Perry, Dr Carla Rogers,Josephine Meedeniya, Gizelle Wilson,and Chelsea Mauch.

REFERENCES

1. Baird J, Fisher D, Lucas P, Kleijnen J,Roberts H, Law C. Being big or growingfast: systematic review of size and growthin infancy and later obesity. BMJ. 2005;331(7522):929–931

2. Anzman SL, Rollins BY, Birch LL. Parentalinfluence on children’s early eating envi-ronments and obesity risk: implications forprevention. Int J Obes (Lond). 2010;34(7):1116–1124

3. Birch LL, Davison KK. Family environmentalfactors influencing the developing behav-ioral controls of food intake and childhoodoverweight. Pediatr Clin North Am. 2001;48(4):893–907

4. DiSantis KI, Hodges EA, Johnson SL, FisherJO. The role of responsive feeding inoverweight during infancy and toddler-hood: a systematic review. Int J Obes(Lond). 2011;35(4):480–492

5. Ciampa PJ, Kumar D, Barkin SL, et al.Interventions aimed at decreasing obesityin children younger than 2 years: a sys-tematic review. Arch Pediatr Adolesc Med.2010;164(12):1098–1104

6. Yanovski JA. Intervening during infancy toprevent pediatric obesity. Obesity (SilverSpring). 2011;19(7):1321–1322

7. Daniels LA, Magarey A, Battistutta D, et al.The NOURISH randomised control trial:positive feeding practices and food pref-erences in early childhood - a primaryprevention program for childhood obesity.BMC Public Health. 2009;9:387

8. Daniels LA, Mallan KM, Battistutta D,Nicholson JM, Perry R, Magarey A. Eval-uation of an intervention to promote pro-tective infant feeding practices to preventchildhood obesity: outcomes of the NOURISHRCT at 14 months of age and 6 months postthe first of two intervention modules. Int JObes (Lond). 2012;36(10):1292–1298

9. Wardle J, Cooke L. Genetic and environ-mental determinants of children’s foodpreferences. Br J Nutr. 2008;99(suppl 1):S15–S21

10. Cooke L. The importance of exposure forhealthy eating in childhood: a review. JHum Nutr Diet. 2007;20(4):294–301

11. Dovey TM, Staples PA, Gibson EL, HalfordJCG. Food neophobia and ‘picky/fussy’ eat-ing in children: a review. Appetite. 2008;50(2–3):181–193

12. Maier A, Chabanet C, Schaal B, Issanchou S,Leathwood P. Effects of repeated exposureon acceptance of initially disliked vegetables

in 7-month old infants. Food Qual Prefer.2007;18(8):1023–1032

13. Benton D. Role of parents in the de-termination of the food preferences of chil-dren and the development of obesity. Int JObes Relat Metab Disord. 2004;28(7):858–869

14. Bakermans-Kranenburg MJ, van IJzendoornMH, Juffer F. Less is more: meta-analyses ofsensitivity and attachment interventions inearly childhood. Psychol Bull. 2003;129(2):195–215

15. Nelson CS, Wissow LS, Cheng TL. Effective-ness of anticipatory guidance: recentdevelopments. Curr Opin Pediatr. 2003;15(6):630–635

16. Daniels LA, Wilson JL, Mallan KM, et al.Recruiting and engaging new mothers innutrition research studies: lessons fromthe Australian NOURISH randomised con-trolled trial. Int J Behav Nutr Phys Act.2012;9(1):129

17. Bandura A. Health promotion by socialcognitive means. Health Educ Behav. 2004;31(2):143–164

18. Eneli IU, Crum PA, Tylka TL. The trust model:a different feeding paradigm for managingchildhood obesity. Obesity (Silver Spring).2008;16(10):2197–2204

ARTICLE

PEDIATRICS Volume 132, Number 1, July 2013 e117 by guest on March 3, 2021www.aappublications.org/newsDownloaded from

19. Hurley KM, Cross MB, Hughes SO. A sys-tematic review of responsive feeding andchild obesity in high-income countries. JNutr. 2011;141(3):495–501

20. Wake M, Nicholson JM, Hardy P, Smith K.Preschooler obesity and parenting styles ofmothers and fathers: Australian nationalpopulation study. Pediatrics. 2007;120(6).Available at: www.pediatrics.org/cgi/con-tent/full/120/6/e1520

21. Jansen E, Daniels L, Nicholson J. The dy-namics of parenting and early feeding—constructs and controversies: a viewpoint.Early Child Dev Care. 2012;182(8):967–981

22. Patrick H, Nicklas TA, Hughes SO, MoralesM. The benefits of authoritative feedingstyle: caregiver feeding styles and child-ren’s food consumption patterns. Appetite.2005;44(2):243–249

23. Birch LL, Fisher JO, Grimm-Thomas K,Markey CN, Sawyer R, Johnson SL. Confir-matory factor analysis of the Child FeedingQuestionnaire: a measure of parental atti-tudes, beliefs and practices about childfeeding and obesity proneness. Appetite.2001;36(3):201–210

24. Wardle J, Sanderson S, Guthrie CA, RapoportL, Plomin R. Parental feeding style and theinter-generational transmission of obesityrisk. Obes Res. 2002;10(6):453–462

25. Chan L, Magarey AM, Daniels LA. Maternalfeeding practices and feeding behaviors ofAustralian children aged 12–36 months.Matern Child Health J. 2011;15(8):1363–1371

26. Satter E. The feeding relationship: prob-lems and interventions. J Pediatr. 1990;117(2 Pt 2):S181–S189

27. World Health Organization. WHO childgrowth standards: length/height-for-age,weight-for-age, weight-for-length, weight-for-height and body mass index-for-age:

Methods and development. Geneva: WorldHealth Organization;2006

28. Cole TJ, Bellizzi MC, Flegal KM, Dietz WH.Establishing a standard definition for childoverweight and obesity worldwide: in-ternational survey. BMJ. 2000;320(7244):1240–1243

29. Trewin D. Census of population and housing:Socio-economic indexes for area’s (SEIFA).Canberra; Australian Bureau of Statistics(ABS);2001

30. Cole TJ, Bellizzi MC, Flegal KM, Dietz WH.Establishing a standard definition for childoverweight and obesity worldwide: in-ternational survey. BMJ. 2000;320(7244):1240–1243

31. Anzman SL, Birch LL. Low inhibitory controland restrictive feeding practices predictweight outcomes. J Pediatr. 2009;155(5):651–656

32. Gubbels JS, Kremers SPJ, Stafleu A, et al.Diet-related restrictive parenting practices.Impact on dietary intake of 2-year-oldchildren and interactions with child char-acteristics. Appetite. 2009;52(2):423–429

33. Skinner JD, Carruth BR, Bounds W, ZieglerP, Reidy K. Do food-related experiences inthe first 2 years of life predict dietary va-riety in school-aged children? J Nutr EducBehav. 2002;34(6):310–315

34. Paul IM, Savage JS, Anzman SL, et al. Pre-venting obesity during infancy: a pilotstudy. Obesity (Silver Spring). 2011;19(2):353–361

35. Wen LM, Baur LA, Simpson JM, Rissel C,Wardle K, Flood VM. Effectiveness of homebased early intervention on children’s BMIat age 2: randomised controlled trial. BMJ.2012;344:e3732

36. Wen LM, Baur LA, Simpson JM, Rissel C,Flood VM. Effectiveness of an early intervention

on infant feeding practices and “tummytime”: a randomized controlled trial. ArchPediatr Adolesc Med. 2011;165(8):701–707

37. Wake M, Hardy P, Canterford L, Sawyer M,Carlin JB. Overweight, obesity and girth ofAustralian preschoolers: prevalence andsocio-economic correlates. Int J Obes(Lond). 2007;31(7):1044–1051

38. Askie LM, Baur LA, Campbell K, et al; EPOCHCollaboration Group. The Early Prevention ofObesity in CHildren (EPOCH) Collaboration—an individual patient data prospective meta-analysis. BMC Public Health. 2010;10(1):728

39. Magarey AM, Daniels LA, Boulton TJ,Cockington RA. Predicting obesity in earlyadulthood from childhood and parentalobesity. Int J Obes Relat Metab Disord.2003;27(4):505–513

40. Ong KK, Ahmed ML, Emmett PM, Preece MA,Dunger DB. Association between postnatalcatch-up growth and obesity in childhood:prospective cohort study. BMJ. 2000;320(7240):967–971

41. Gardner DS, Hosking J, Metcalf BS, JefferyAN, Voss LD, Wilkin TJ. Contribution of earlyweight gain to childhood overweight andmetabolic health: a longitudinal study(EarlyBird 36). Pediatrics. 2009;123(1).Available at: www.pediatrics.org/cgi/con-tent/full/123/1/e67

42. Barker DJ, Osmond C, Forsén TJ, Kajantie E,Eriksson JG. Trajectories of growth amongchildren who have coronary events asadults. N Engl J Med. 2005;353(17):1802–1809

43. Moher D, Schulz KF, Altman DG. The CON-SORT statement: revised recommendationsfor improving the quality of reports ofparallel-group randomised trials. Lancet.2001;357(9263):1191–1194

e118 DANIELS et al by guest on March 3, 2021www.aappublications.org/newsDownloaded from

originally published online June 10, 2013; Pediatrics Battistutta and Anthea Magarey

Lynne Allison Daniels, Kimberley Margaret Mallan, Jan Maree Nicholson, DianaObesity

Outcomes of an Early Feeding Practices Intervention to Prevent Childhood

ServicesUpdated Information &

012-2882http://pediatrics.aappublications.org/content/early/2013/06/05/peds.2including high resolution figures, can be found at:

Permissions & Licensing

http://www.aappublications.org/site/misc/Permissions.xhtmlin its entirety can be found online at: Information about reproducing this article in parts (figures, tables) or

Reprintshttp://www.aappublications.org/site/misc/reprints.xhtmlInformation about ordering reprints can be found online:

by guest on March 3, 2021www.aappublications.org/newsDownloaded from

originally published online June 10, 2013; Pediatrics Battistutta and Anthea Magarey

Lynne Allison Daniels, Kimberley Margaret Mallan, Jan Maree Nicholson, DianaObesity

Outcomes of an Early Feeding Practices Intervention to Prevent Childhood

http://pediatrics.aappublications.org/content/early/2013/06/05/peds.2012-2882located on the World Wide Web at:

The online version of this article, along with updated information and services, is

by the American Academy of Pediatrics. All rights reserved. Print ISSN: 1073-0397. the American Academy of Pediatrics, 345 Park Avenue, Itasca, Illinois, 60143. Copyright © 2013has been published continuously since 1948. Pediatrics is owned, published, and trademarked by Pediatrics is the official journal of the American Academy of Pediatrics. A monthly publication, it

by guest on March 3, 2021www.aappublications.org/newsDownloaded from