pediatrics 2004 chatoor e440 7

Failure to Thrive and Cognitive Development in Toddlers WithInfantile Anorexia

Irene Chatoor, MD*; Jaclyn Surles, BA*; Jody Ganiban, PhD; Leila Beker, PhD*;Laura McWade Paez, CPNP*; and Benny Kerzner, MD*

ABSTRACT. Objective. The goal of this study was toexamine the relative contributions of growth deficiencyand psychosocial factors to cognitive development intoddlers with infantile anorexia.

Methods. Eighty-eight toddlers, ranging in age from12 to 33 months, were enrolled in this study. Toddlerswere evaluated by 2 child psychiatrists and placed into 1of 3 groups: infantile anorexia, picky eater, and healthyeater. All 3 groups were matched for age, race, gender,and socioeconomic status (SES). Toddlers underwent nu-tritional evaluations and cognitive assessments with theBayley Scales of Infant Development. Toddlers and theirmothers were also videotaped during feeding and playinteractions, which later were rated independently by 2observers.

Results. On average, toddlers with infantile anorexiaperformed within the normal range of cognitive devel-opment. However, the Mental Developmental Index(MDI) scores of the healthy eater group (MDI 110)were significantly higher than those of the infantile an-orexia (MDI 99) and picky eater (MDI 96) groups.Within the infantile anorexia group, correlations be-tween MDI scores and the toddlers percentage of idealbody weight approached statistical significance (r .32).Across all groups, the toddlers MDI scores were associ-ated with the quality of motherchild interactions, SESlevel, and maternal education level. Collectively, thesevariables explained 22% of the variance in MDI scores.

Conclusions. This study demonstrated that psychoso-cial factors, such as mothertoddler interactions, mater-nal education level, and SES level, are related to thecognitive development of toddlers with feeding prob-lems and explain more unique variance in MDI scoresthan nutritional status. Pediatrics 2004;113:e440e447.URL: http://www.pediatrics.org/cgi/content/full/113/5/e440; failure to thrive, feeding disorder, infantile an-orexia, cognitive development, growth deficiency, mother-toddler interactions.

ABBREVIATIONS. FTT, failure to thrive; SES, socioeconomic sta-tus; MDI, Mental Developmental Index; BSID, Bayley Scales ofInfant Development.

Failure to thrive (FTT) describes children whoexhibit growth deficiency, as indexed by falter-ing or stunted growth. Several studies suggestthat FTT is associated with poorer cognitive devel-opment, learning disabilities, and long-term behav-ioral problems.13 More recently, Corbett et al4 de-tected a significant association between the severityof growth deficiency and IQ, whereas Raynor andRudolf5 found that 55% of the infants who werefailing to thrive exhibited developmental delay. Inaddition, a study by Reif et al6 reported that childrenwith a history of FTT were found to have morelearning difficulties and evidenced developmentaldelay at follow-up 5 years after the initial presenta-tion.

These findings from the pediatric literature haveled many to believe that FTT alone is sufficient tocause developmental delays. However, a criticalproblem with many previous studies is that FTT isfrequently confounded with psychosocial risk factors(including low socioeconomic status [SES], maternaleducation levels, and maternal deprivation) that areindependently related to lower Mental Developmen-tal Index (MDI) scores.1,4 As a result, psychosocialfactors may contribute to the apparent associationbetween FTT and cognitive delay. Consequently, theconclusion that FTT is sufficient to cause significantcognitive delay requires additional exploration.

The tendency to confound FTT and psychosocialrisk factors grew from early studies that used non-organic FTT and maternal deprivation as synony-mous terms.7,8 Whereas several authors have pro-posed that FTT should be considered a singlesymptom that describes growth deficiency,912 othershave used nonorganic FTT as a clinical syndromethat encompasses children who exhibit FTT in addi-tion to psychosocial risk factors.1,3,4 Consequently,several authors have argued strongly for disentan-gling FTT (growth deficiency) from psychosocial fac-tors and examining FTT as a single symptom of afeeding disorder, rather than a clinical syn-drome.11,13,14 Such a distinction is critically impor-tant for identifying the developmental consequencesspecifically related to growth deficiency, as well asthe multiple pathways that can lead to growth defi-ciency.11,13,14 Although many factors, genetic and en-vironmental, can contribute to cognitive develop-ment in young children, the goal of this article is totease apart the effects of growth deficiency and psy-chosocial risk on cognitive development.

From the *Childrens National Medical Center, Washington, DC; andGeorge Washington University, Washington, DC.Received for publication Aug 3, 2003; accepted Dec 1, 2003.Reprint requests to (I.C.) Department of Psychiatry, Childrens NationalMedical Center, 111 Michigan Ave, NW, Washington, DC 20010. E-mail:[email protected] (ISSN 0031 4005). Copyright 2004 by the American Acad-emy of Pediatrics.

e440 PEDIATRICS Vol. 113 No. 5 May 2004 http://www.pediatrics.org/cgi/content/full/113/5/e440 at Charles R Drew Univ on July 23, 2015pediatrics.aappublications.orgDownloaded from

The study described in this article focused on in-fantile anorexia, a feeding disorder that is charac-terized by extreme food refusal, growth deficiency,and an apparent lack of appetite.15,16 Importantly,infantile anorexia is not associated with maternaldeprivation or neglect, and most children with thisfeeding disorder come from middle- to upper-mid-dle-class families.15,17 Therefore, studying this pop-ulation affords the opportunity to disentangle thecontributions of growth deficiency and psychosocialrisk factors to cognitive outcomes.

Infantile anorexia was first described in a series ofcase studies by Chatoor and Egan,18 and at that timeit was referred to as a separation disorder. Infantileanorexia arises in the first 3 years of life, most com-monly between the ages of 9 and 18 months, asinfants become more autonomous and make thetransition to spoon- and self-feeding. Children withinfantile anorexia fail to communicate signals of hun-ger, but they show a strong interest in exploration,play, and/or interaction with their caregivers. Theyexhibit extreme food refusal and frequently fail totake in sufficient calories to sustain growth. As aresult, these children display acute and/or chronicmalnutrition.16

Drawing from the rich literature on growth defi-ciency and the multiple factors that can have animpact on the cognitive development of young chil-dren, this article examines the relationship of cogni-tive development to physical growth, mothertod-dler interactions during feeding and play, maternaleducation, and SES. We examine these relationshipsin a group of children who have infantile anorexiaand exhibit growth deficiency, a control group ofpicky eaters with normal weight, and a second con-trol group of healthy eaters with normal weight.

The primary questions addressed by this studywere as follows: 1) Is infantile anorexia associatedwith lower scores on the MDI? Although toddlerswith infantile anorexia do exhibit growth deficiency,they typically do not experience maternal neglectand tend to be from middle- to upper-middle-classfamilies.15,17 Consequently, we hypothesized thatthey would not demonstrate significant developmen-tal delays. 2) Do psychosocial variables (SES, mater-nal education, quality of motherchild interactions)and growth deficiency make independent contribu-tions to MDI scores? We hypothesized that SES, ma-ternal education, and the quality of mothertoddlerinteractions would be stronger predictors of cogni-tive development than growth deficiency.

METHODS

SubjectsToddlers with infantile anorexia (n 34), picky eaters (n 34),

and healthy eaters (n 34) were recruited for a diagnostic studyof infantile anorexia. Previous publications from this data set havefocused on the diagnosis of infantile anorexia and attachmentpatterns, temperament characteristics, and parental characteristicsassociated with this feeding disorder.15,19,20 The current reportfocuses on the cognitive development of toddlers with infantileanorexia in relation to picky and healthy eaters, and it included asubset of toddlers from the original study (n 88) who werebetween the ages of 12 and 33 months.

After Institutional Review Board approval was obtained, par-

ents of toddlers who had infantile anorexia and were referred tothe study by pediatricians and gastroenterologists at the hospitaland in the community were asked to participate in the study.None of the parents refused consent. The diagnosis of infantileanorexia was made independently by 2 child psychiatrists, whohad excellent interrater agreement ( .89). The diagnosis wasbased on the following criteria: 1) refusal to eat adequate amountsof food for at least 1 month; 2) onset of the food refusal under 3years of age, most commonly during the transition to spoon- andself-feeding; 3) failure to communicate hunger signals, lack ofinterest in food, but strong interest in exploration and/or interac-tion with caregivers; 4) significant growth deficiency; and 5) noevidence that the food refusal followed a traumatic event or isassociated with an underlying medical illness.

Picky eaters and healthy eaters were recruited from an urbanambulatory care center. Parents of toddlers who ranged in agebetween 12 and 36 months were asked to complete a brief ques-tionnaire on their childrens feeding habits. When the parentsdescribed their toddlers as often or always healthy eaters,they were considered for recruitment to the healthy eater group.When the parents described their children as often or alwayspicky eaters, they were considered for recruitment to the pickyeater group. Assignment to the picky eater group also dependedon additional screening for medical and growth problems. Specif-ically, toddlers were assigned to the picky eater group when theydemonstrated 1) persistent refusal (for at least 1 month) to eat alltypes of food or certain types of food to cause concern to theparents and 2) no evidence of growth deficiency. The require-ments for placement in the group of healthy eaters were 1) no foodrefusal of concern for at least 1 month and 2) no evidence ofgrowth deficiency. Only parents whose toddlers matched thestudy subjects by age, gender, race, and SES were invited toparticipate in the study. Ten toddlers from the original samplewere excluded from this study because they were33 months oldat the time of testing. The final sample included 34 children in thehealthy eater group (age: 23 months; standard deviation [SD]: 5months), 26 children in the picky eater group (age: 24 months; SD:5 months), and 32 children in the infantile anorexia group (age: 21months; SD: 6 months). The 3 groups did not differ in regard togender (2 4.54, P .11), or race (2 2.21, P .90). The specificdemographic characteristics of the sample (excluding SES) aresummarized in Table 1. Because SES was used as a predictorvariable, data pertaining to SES are included in Table 2.

ProceduresAll children underwent a medical evaluation before the study,

and they were excluded from the study when they had anymedical, neurologic, or genetic illness or when they demonstrateda psychiatric disorder associated with developmental delays (eg,autism spectrum disorders). All mothers were asked to completeseveral questionnaires, and each mother and toddler completed 2laboratory sessions. During the first session, all mothers and tod-dlers were videotaped during feeding and play interactions. Afterthese interactions, a child psychiatrist administered a diagnosticinterview to assess the toddler, and a nutritionist evaluated thetoddlers height and weight.

Two child psychiatrists independently evaluated each child viamaternal interview, observing mothertoddler interactions duringfeeding and play, and the nutritional assessment. One psychiatrist

TABLE 1. Demographic Characteristics of the Study Sampleby Diagnostic Group

Diagnostic Group

HealthyEaters

PickyEater

InfantileAnorexia

% f % f % f

Gender (female) 53 18 35 9 62 20Ethnicity

Black 6 2 12 3 16 5Asian 29 10 27 7 25 8White 59 20 54 14 56 18Hispanic 6 2 8 2 3 1

f indicates frequency.

http://www.pediatrics.org/cgi/content/full/113/5/e440 e441 at Charles R Drew Univ on July 23, 2015pediatrics.aappublications.orgDownloaded from

observed the mother and the toddler during feeding and playfrom behind a 1-way mirror and interviewed the mother, whereasthe second psychiatrist evaluated the toddler through the writteninterview and by observation of the videotape of the mother andthe toddler during feeding and play. Each psychiatrist had accessto the toddlers nutritional assessment.

During the second session, the toddlers cognitive developmentwas assessed with the Bayley Scales of Infant Development(BSID). A psychologist who was blind to the psychiatrists diag-nostic assessment of the toddlers performed the developmentalassessment. Two trained observers who were blind to the diag-nostic group assignment of the toddlers rated mothertoddlerinteractions during feeding and play with the Feeding Scale andthe Play Scale described below.

All toddlers completed the first laboratory session; however, 2toddlers with infantile anorexia and 2 picky eaters could not bescheduled for the developmental assessment because of technicaldifficulties or parents time constraints. Therefore, complete datawere collected for 88 of the 92 toddlers.

Measures

SES and Maternal EducationEach mother completed a demographics questionnaire that

asked her to report her highest level of education and occupationand her husbands highest level of education and occupation. SESwas based on Hollingsheads Four-Factor Index,21 which takesboth parents level of education and occupational status into con-sideration. The Hollingshead index yields 5 SES groupings, withgroup 1 reflecting the highest SES status and group 5 reflecting thelowest. Maternal education was determined from the same ques-tionnaire and was categorized according to the highest level ofeducation that the mother completed (high school, college, grad-uate school, or postgraduate degree).

Growth DeficiencyThe Waterlow criteria were used to assess childrens degree of

growth deficiency.22 These criteria compare the childrens actualweights and heights with the 50th percentile (median) referencestandard on the National Center for Health Statistics growthcharts.23 The childs weight for height at the 50th percentile isconsidered to be the ideal body weight for that length/height.Because this study was conducted before the release of the Centersfor Disease Control and Prevention growth charts, the NationalCenter for Health Statistics charts were used.24 Percentage of idealbody weight was used for the correlation analyses to determine

associations between nutritional status and cognitive develop-ment.

The Waterlow criteria for acute and chronic malnutrition wereused for the diagnostic assessment of each toddlers nutritionalstatus. To meet diagnostic criteria for infantile anorexia, the tod-dlers had to meet the following Waterlow criteria for acute and/orchronic malnutrition. Acute malnutrition is an index of muscledepletion or wasting. It is determined by dividing the childscurrent weight by his or her weight for height at the 50th percen-tile and multiplied by 100. The remaining value represents thepercentage of ideal body weight. Mild, moderate, and severemalnutrition corresponds with 80% to 89%, 70% to 79%, and70% of ideal body weight, respectively.22 Chronic malnutrition,an index of faltering linear growth or stunting, is determinedusing the 50th percentile for height for age, or ideal length orheight. The childs current height is divided by his or her idealheight and multiplied by 100. Mild, moderate, and severe chronicmalnutrition corresponds with 90% to 95%, 85% to 89%, and85% of ideal height, respectively.22

Cognitive DevelopmentThe BSID25 was used in the current study to provide a mea-

surement of the childrens cognitive status. The BSID is 1 of themost commonly used infant assessment tools. The BSID MentalScale consists of 163 items and is used to assess infants cognitivedevelopment between 1 and 30 months of age. Children receive anMDI score that indicates the degree to which they are functioningat an age-appropriate level. MDI scores between 85 and 115 gen-erally indicate that the child is at an age-appropriate level. TheBSID is based on a sample of the US population in regard to race,ethnicity, gender, and SES and has demonstrated good test-retestreliability.25

In the current study, 7 toddlers were between 31 and 33 monthsof age. To interpret the scores of these older toddlers, we extrap-olated the results (based on the relationship described by Bayley25)between the raw scores and the MDI scores for children between26 and 30 months of age. The algorithm for extrapolation involveda regression analysis, using a log scale that seemed to fit the listedvalues very well. The R2 values for the regression equationsranged from .841 to .983, indicating excellent agreement betweenthe regression and the actual scores. Because these toddlers camefrom all 3 diagnostic groups and none of these older toddlersreached the maximum raw score of 163, we believed that thismethod was adequate to include these children in the study.

TABLE 2. Means and Standard Deviations for Cognitive Development and Risk Variables byDiagnostic Group

Variable Diagnostic Group

Healthy Eaters Picky Eater Infantile Anorexia

Mean* SD Mean* SD Mean* SD

Cognitive developmentMDI score 109.9A 16.0 96.3B 19.5 99.2B 19.1

Growth deficiencyPercentage of ideal body weight 107.7A 9.7 101.4B 5.7 84.3C 4.9

Psychosocial risk factorsSES level 2.00A 1.0 1.9A 0.8 1.8A 0.9Maternal education level 16.0A 2.8 16.4A 2.6 16.0A 3.7

Feeding interactionsFeeding conflict 4.2C 4.1 8.6B 5.4 12.0A 5.5Feeding reciprocity 33.5A 4.4 31.3B 3.7 28.2C 4.8Feeding noncontingency 1.0B 1.3 2.0AB 2.1 2.8A 2.6Feeding talk and distraction 6.3B 2.0 7.4A 1.9 7.2AB 2.1Feeding struggle for control 2.2B 1.7 3.5B 2.0 4.6A 2.8

Play interactionsPlay reciprocity 26.8A 3.8 26.1A 4.2 24.5A 5.5Play nonresponsiveness 1.0A 1.4 0.9A 1.1 1.2A 1.6Play conflict 0.9B 1.2 1.2B 0.9 1.9A 1.5Play intrusiveness 4.7A 2.5 5.2A 2.4 5.8A 2.5

SD indicates standard deviation.* Means with common superscript letters are not statistically different ( .05).

e442 COGNITIVE DEVELOPMENT IN INFANTILE ANOREXIA at Charles R Drew Univ on July 23, 2015pediatrics.aappublications.orgDownloaded from

MotherToddler Interactions During FeedingThe Feeding Scale used in this study is a global rating scale that

includes 46 items describing mothers and toddlers behaviorsduring feeding.17 To rate the quality of mothertoddler interac-tions during feeding, observers use a 4-point Likert scale to indi-cate the frequency and/or the intensity with which specific be-haviors are displayed in a 20-minute feeding session (0 none, 1 a little, 2 pretty much, 3 very much). The items are groupedinto 5 subscales: 1) Dyadic Reciprocitypositive exchanges be-tween mother and toddler; 2) Dyadic Conflicttoddlers foodrefusal and negative affect, as well as mothers negative affect andnegative comments regarding her toddler; 3) Talk and Distrac-tionengagement in talk and play by mother and toddler thatinterferes with feeding; 4) Struggle for Controlcontrolling ma-ternal behaviors (eg, overriding the toddlers cues or forcing foodinto the toddlers mouth) and the toddlers resistance (eg, spittingout food); and 5) Maternal Noncontingencyinappropriate ma-ternal behaviors during feeding (eg, ignores toddlers cues, han-dles toddler excessively, restricts toddlers movements). In a pre-vious study, the Feeding Scale discriminated between feeding-disordered and nonfeeding-disordered populations anddemonstrated acceptable test-retest reliability over a 2-week peri-od.17 Interrater reliability was high for each subscale, with intra-class correlations ranging from .95 to .99.

MotherToddler Interactions During PlayThe Play Scale used in this study provides a global rating of the

behaviors that toddlers and their mothers display during freeplay. To rate the quality of mothertoddler interactions during a10-minute free play situation, observers assess 32 mother andtoddler behaviors by using a 4-point Likert scale (0 none, 1 alittle, 2 pretty much, 3 very much). The individual items aregrouped into 4 subscales: 1) Dyadic Reciprocitypositive ex-changes between mother and toddler; 2) Maternal Nonresponsive-ness to Toddlers Needsinappropriate maternal behaviors dur-ing play (eg, handles toddler in an abrupt manner, restrictstoddlers movements, seems detached and/or oblivious to tod-dlers activities); 3) Dyadic Conflictmother and toddler seemdistressed and/or angry, or mother makes critical remarks abouttoddler and/or toddlers play; and 4) Maternal Intrusivenessmother directs toddler or controls play without regard for tod-dlers cues. In a previous study, the Play Scale discriminatedbetween feeding-disordered and nonfeeding-disorderedgroups.26 In addition, the 2-week test-retest reliabilities for theindividual subscales ranged from .39 to .58.27

Data AnalysisAnalysis of variance was used to determine whether diagnostic

group (infantile anorexia, picky eater, healthy eater) is associatedwith MDI score. In this analysis, the diagnostic group was used asthe single factor. Duncans multiple range test was used to rankthe 3 groups and to assess the degree to which specific diagnosticgroups differed from each other. An level of .05 was used foreach analysis.

The next set of analyses focused on associations between MDIscores and previously identified risk factors for low MDI scores:growth deficiency (percentage of ideal body weight), SES level,maternal education level, and quality of parentchild interactionsduring feeding and play. First, we examined the associationsbetween the diagnostic group and the risk factors. Analysis ofvariance was used to determine whether the 3 diagnostic groupsdiffered in regard to any of the hypothesized psychosocial riskfactors. When there were statistically significant differences ofmean between the diagnostic groups, the Duncan multiple rangetest was used to determine where those differences occur. An level of .05 was used for each analysis.

Second, Pearson correlations were used to assess associationsamong the growth deficiency, the psychosocial risk factors, andthe childrens MDI scores. Because we sought to determinewhether psychosocial variables predict variance in MDI scoresindependent of growth deficiency, we also conducted additionalmultiple regression analyses that assessed the relationship be-tween feeding conflict and MDI score, controlling for the effects ofpercentage of ideal body weight. An level of .05 was used foreach analysis.

RESULTSA primary question of the current study was

whether the diagnostic group was associated withthe childrens cognitive development. As indicatedin Table 2, on average, all 3 groups seemed to func-tion within the normal range of cognitive develop-ment. However, there was a significant effect of di-agnostic group on MDI scores (F[2,85] 4.75, P .05). The healthy eater group exhibited significantlyhigher MDI scores than the infantile anorexia andpicky eater groups. The last 2 groups did not differfrom each other.

The second question addressed by this study wasthe degree to which growth deficiency and psycho-social risk variables explain variance in MDI scores.Table 2 includes the means and standard deviationsfor MDI scores and each risk factor by diagnosticgroup. The 3 groups did not differ in regard to SESlevel and maternal education level. However, the 3groups differed in regard to percentage of ideal bodyweight (F[2,89] 91.16, P .0001). Consistent withthe diagnostic criteria for infantile anorexia, the tod-dlers in the infantile anorexia group displayed sig-nificantly lower percentage of ideal body weightthan the toddlers in the remaining groups. In addi-tion, however, the picky eater group displayed sig-nificantly lower percentage of ideal body weightwhen compared with the healthy eater group. Thefeeding and play subscales were also related to di-agnostic group. The infantile anorexia group dis-played significantly higher levels of feeding conflict(F[2,88] 19.67, P .0001), struggle for controlduring feeding (F[2,88] 8.95, P .001), play con-flict (F[2,89] 5.90, P .01), and less reciprocity(F[2,88] 12.43, P .0001) than the remaininggroups. In addition, the infantile anorexia and pickyeater groups displayed more feeding noncontin-gency (F[2,88] 2.84, P .10) than the healthy eatergroup.

These findings suggest that the lower MDI scoresof the infantile anorexia and picky eater groups maybe explained by percentage of ideal body weight orby variables associated with feeding conflict. To testthese possibilities, we computed Pearson correla-tions for each of the risk variables and MDI scores.The overall correlation between percentage of idealbody weight and MDI score was not statisticallysignificant (r .16). However, the correlation be-tween percentage of ideal body weight and MDIscore varied by diagnostic group (Fig 1). The corre-lation between MDI score and percentage of idealbody weight was positive for the infantile anorexiagroup and approached significance (r .32, P .10).In contrast, this association was negative within thehealthy eater group (r .31, P .10). The pickyeater group also evidenced a positive associationbetween percentage of ideal body weight and MDIscore; however, this association did not approachstatistical significance (r .25, P .10). These dif-ferences are depicted in Fig 1.

The remaining correlations between SES and ma-ternal education level and MDI scores did not varyacross the diagnostic groups. Consequently, correla-


tions for the entire sample are presented in Table 3.Consistent with previous studies, SES level was in-versely related to MDI score, indicating that childrenfrom middle- to upper-middle-class families (eg,Hollingshead scores of 1 and 2) tended to havehigher MDI scores. Higher maternal education levelwas also positively associated with MDI score. Inregard to mothertoddler interactions, the followingfeeding subscales were negatively associated withMDI score: Feeding Conflict, Feeding Noncontin-gency, and Feeding Struggle for Control. One feed-ing subscale, Feeding Reciprocity, was positively as-sociated with MDI score. In regard to the playinteractions, only maternal intrusiveness was nega-

tively associated with MDI score. In summary, moreproblematic and conflictual mothertoddler feedingand play interactions were associated with lowerMDI scores. Last, Table 3 indicates that many of thepsychosocial risk factors were related to MDI scoreas well as to each other. Specifically, SES level andmaternal education were associated with each otherand with play intrusiveness. In addition, lower SESlevel was related to more struggles for control duringfeeding, whereas lower maternal education was as-sociated with more feeding conflict, feeding noncon-tingency, and feeding talk and distraction and withlower feeding reciprocity.

In the final set of analyses, we explored the degree

Fig. 1. Relationship between percentage of ideal weight and MDI for each of the 2 diagnostic groups. , Children with infantile anorexia;, healthy eaters. The lines represent the results of an analysis of covariance with percentage of ideal weight as a continuous covariateand indicator variables for each diagnostic group. The slopes of those regression lines are different to a statistically significant degree (P.029).

TABLE 3. Correlations Between Psychosocial Risk Variables and MDI Scores

Variables 1 2 3 4 5 6 7 8 9 10 11 12

1. MDI score 2. SES .36 3. Maternal education .33 .50 4. Feeding conflict .33 .10 .20 5. Feeding reciprocity .30 .02 .23* .656. Feeding noncontingency .22* .20 .27* .67 .53 7. Feeding talk and distraction .02 .15 .25* .25* .08 .11 8. Feeding struggle for control .34 .22* .16 .64 .47 .61 .13 9. Play nonresponsiveness .02 .07 .15 .20 .22* .18 .07 .12

10. Play reciprocity .16 .09 .16 .42 .46 .35 .08 .31 .71 11. Play conflict .21 .11 .11 .45 .40 .27 .16 .43 .33 .54 12. Play intrusiveness .39 .29 .28 .40 .26* .31 .02 .43 .03 .11 .52

* P .05. P .01. P .001. P .0001.


to which psychosocial risk variables predicted vari-ance in MDI scores that was independent of growthdeficiency (ie, percentage of ideal body weight). Be-cause the feeding subscales and play subscales werehighly intercorrelated (Table 3), factor analysis usinga promax rotation was used to generate compositescores for parentchild interactions during feedingand play sessions. This analysis yielded 2 generalfactors. Factor 1 included all of the feeding subscales,as well as the Intrusiveness subscale for play. Theloadings for factor 1 were .84 for Feeding Struggle forControl, .83 for Feeding Conflict, .79 for FeedingMaternal Noncontingency, .60 for Feeding Reci-procity, and .68 for Play Maternal Intrusiveness.Overall, this factor reflected conflict and struggles forcontrol between mothers and toddlers and wasnamed Interactional Conflict. The second factor in-cluded only play scale subscales and was namedInteractional Responsiveness. The loadings for factor2 were .95 for Play Maternal Nonresponsiveness, .45for Play Conflict, and .90 for Play Reciprocity. Fac-tor scores for Interactional Conflict and InteractionalResponsiveness were generated for each subject. Al-though the Interactional Conflict Factor score wassignificantly correlated with MDI score (r .41, P .0001), SES level (r .23, P .05), and maternaleducation level (r .27, P .001), the InteractionalResponsiveness score was not associated with MDIscore (r .07), SES level (r -.02), or maternaleducation level (r .15).

Multiple regression was used to examine the de-gree to which psychosocial factors (eg, SES, maternaleducation) versus motherchild interactions were re-lated to MDI scores. In all, 2 multiple regressionanalyses were conducted. In the first regression, theassociation between SES level and maternal educa-tion level and MDI score was computed, controllingfor percentage of ideal body weight. In the secondanalysis, the association between the InteractionalConflict Factor score and MDI score was computed,controlling for percentage of ideal body weight, SESlevel, and maternal education level. The results ofthese analyses are presented in Table 4.

In the first regression, SES level, maternal educa-tion level, and percentage of ideal body weight col-

lectively explained 17% of variance in the toddlersMDI scores (F[3,79] 5.25, P .01). Given that theassociation between percentage of ideal body weightand MDI score was statistically nonsignificant ( .17), most of the explained variance could be attrib-uted to SES level and maternal education level. Ofthese 2 predictors, SES level explained the mostunique variance in MDI scores. When the Interac-tional Conflict Factor score was included in the sec-ond regression analysis, R2 increased to 22%, sug-gesting that this variable alone explained 5% of thevariance in MDI scores. SES level continued to ex-plain unique variance in MDI scores. The weights,however, indicate that the unique contributions ofSES level and Interactional Conflict score wereequivalent. Consequently, although SES level andInteractional Conflict score were significantly corre-lated, both variables explained unique variance inMDI scores.

DISCUSSIONThis study examined the unique and combined

contributions of growth deficiency (ie, FTT) and psy-chosocial risk factors to MDI scores in 3 groups oftoddlers. The study revealed 2 important findings: 1)although toddlers with infantile anorexia exhibitgrowth deficiency, they performed within the nor-mal range of cognitive development, and 2) the MDIscores of toddlers with infantile anorexia and that ofthe normal-weight picky eaters were, respectively, 11and 14 points below that of the healthy eaters with-out feeding problems.

Although we found a positive correlation betweenpercentage of ideal body weight and the MDI scorefor the group of toddlers with infantile anorexia, thecorrelation was not strong and only approached sta-tistical significance. It is interesting that whereas thecorrelation between MDI and percentage of idealbody weight was positive for the infantile anorexiagroup, these variables were inversely related for thehealthy eater group. This is a most interesting find-ing, which indicates that more malnourished tod-dlers and heavy toddlers may perform less well cog-nitively. Although a negative effect of growthdeficiency, or FTT, on cognitive development hasbeen reported by several studies,2,46 the inverse re-lationship between weight and cognitive develop-ment for heavy toddlers was unexpected, and thisfinding should be replicated with a larger sample ofnormal and overweight toddlers. However, as statedpreviously, it should be kept in mind that withinboth groups, the correlation between percentage ofideal weight and MDI was only marginally signifi-cant, indicating that the effect size of percentage ofideal weight was small.

Because percentage of ideal body weight ac-counted for only a small portion of variance in theMDI score of toddlers within all 3 groups, otherfactors clearly contributed to the cognitive develop-ment of these young children. Several studies haveexamined the independent effects of psychosocialrisk variables on childrens cognitive development,including SES, maternal education, the quality ofparentchild interactions, and child maltreat-

TABLE 4. Predicting MDI Scores From Psychosocial RiskVariables, Controlling Toddlers Percentage of Ideal Body Weight

Risk Variable T

Multiple Regression IPercentage of ideal body weight .17 1.63SES level .32* 2.72Maternal education .11 0.97

F(3,79) 5.25R2 .17

Multiple regression IIPercentage of ideal body weight .04 0.36SES level .26* 2.18*Maternal education .08 0.73Feeding conflict .27* 2.25*

F(4,77) 5.37R2 .22

* P .05. P .01. P .0001.


ment.2838 Although most of the toddlers in ourstudy came from middle- and upper-middle-classfamilies and had well-educated mothers and none ofthe toddlers had a history of neglect or abuse, we stillfound that many of the psychosocial variables as-sessed were related to MDI scores across all 3groups. Specifically, higher SES, maternal education,and feeding reciprocity were related to higher MDIscores. In contrast, higher levels of conflict, noncon-tingency, and control struggles during feeding inter-actions and maternal intrusiveness during play in-teractions were related to lower MDI scores. Whenmothers and toddlers were in conflict with eachother over the toddlers refusal to eat and struggledfor control, with the mothers overriding the toddlerscues and/or forcing food into the toddlers mouths,these interaction behaviors correlated negativelywith the toddlers cognitive performance. It is inter-esting that maternal intrusiveness during play, whenmothers directed the toddlers and/or controlled theplay without regard for the toddlers cues, also cor-related negatively with the toddlers cognitive per-formance.

These findings are consistent with previous stud-ies. For example, in a 4-year longitudinal study, Beeet al28 found that motherinfant interactions and thequality of the environment were among the bestpredictors of child IQ and language at 24 and 36months of age. Similarly, Coates and Lewis29 re-ported that maternal responsivity was related to thechildrens reading and conversation skills, as well asto their math skills over a 6-year period. Crandelland Hobson30 also found that the degree to whichparentchild interactions were synchronous was re-lated to child IQ. Additional studies have docu-mented significant associations between child mal-treatment and deficits in cognitive functioning.31,32Importantly, Mackner et al33 reported that maternalneglect and FTT have similar but independent effectson cognitive development and that both factors rep-resent additive risks for deficits in cognitive func-tioning. This latter finding underscores the impor-tance of differentiating neglect and FTT as 2 separaterisk factors.

Previous studies also have found that maternaleducation, which is frequently used as an indicatorof SES, has been associated with childrens cognitiveoutcomes.34,35 In addition, Singer and Fagan36 re-ported that lower parental education levels wereassociated with lower levels of cognitive develop-ment in 3-year-old children with a history of FTT. Asthese studies have shown, maternal IQ and educa-tion are significant predictors of cognitive develop-ment of their children.35,37,38

In the final set of analyses, we examined the cu-mulative effects of psychosocial factors on MDIscores. Collectively, SES, maternal education, andInteractional Conflict explained 22% of the variancein MDI scores, with only negligible contributionsfrom childrens percentage of ideal body weight.These findings further underscore that previouslyfound associations between FTT and MDI score mayhave been primarily explained by the childs socialworld.

In summary, we found that although toddlers withinfantile anorexia displayed growth deficiency (ie,low percentage of ideal body weight), they still werewithin the normal range of cognitive developmentand received MDI scores that were similar to those ofnormal-weight picky eaters. Subsequent analyses in-dicated that psychosocial variables were more potentpredictors of MDI than nutritional status. Althoughseveral studies have related cognitive developmentto motherinfant interactions and the quality of thehome environment,29,30,39 this is the first study thatrevealed a significant correlation between mothertoddler interactions during feeding and play sessionsand the toddlers cognitive performance. In addition,SES and maternal education were significantly re-lated to MDI scores. Together, the quality of motherchild feeding and play interactions, SES, and mater-nal education explained 22% of variance in MDI.Although this finding is significant, other factorsclearly contribute to MDI scores. For example, chil-drens temperament characteristics and attention lev-els also may have influenced childrens scores. Fu-ture studies are needed to explore the impact of thesevariables on MDI as well as on the quality of parentchild interactions. Last, it should be noted that ourfindings rely on correlational data. Consequently, thedirection of effects between MDI and parentchildinteractions cannot be established conclusively. It ispossible that lower MDI scores may contribute tofeeding problems. However, given that most chil-dren in the infantile anorexia group had MDI scoreswithin the normal range, it is unlikely that MDIcontributed significantly to feeding problems withinthis particular group.

The findings from this study also emphasize theimportance of distinguishing between nonorganicforms of growth deficiency related to maternal ne-glect and growth deficiency that is related to dyadicconflict during feeding. The concern over the effectof poor weight gain on the cognitive development ofinfants and young children often overrides the man-agement of their feeding difficulties. Because parentsbecome worried about the future cognitive develop-ment of their infants, they resort to coercive feeding,which ultimately intensifies parentchild conflictduring feeding. Although correlations do not allowfor the interpretation of causality, the findings fromthis study suggest that the concern for the nutritionalneeds of young children needs to be balanced withthe management of their feeding difficulties.

ACKNOWLEDGMENTSThe research and the preparation of the manuscript were sup-

ported by a grant from the National Institute of Mental Healthawarded to Dr Chatoor (R01-MH58219) and a grant from theNational Center for Research Resources awarded to the ChildrensClinical Research Center (RR13297).

REFERENCES1. Drotar D, Sturm L. Prediction of intellectual development in young

children with early histories of nonorganic failure-to-thrive. J PediatrPsychol. 1988;13:281296

2. Galler JR, Ramsey F, Solimano G, Lowell WE, Mason E. The influence ofearly malnutrition on subsequent behavioral development: I. Degree ofimpairment in intellectual performance. J Am Acad Child Psychiatry.1983;22:815


3. Oates RK, Peacock A, Forrest D. Long-term effects of nonorganic failureto thrive. Pediatrics. 1985;75:3640

4. Corbett SS, Drewett RF, Wright CM. Does a fall down a centile chartmatter? The growth and developmental sequelae of mild failure tothrive. Acta Paediatr. 1996;85:12781283

5. Raynor P, Rudolf MCJ. What do we know about children who fail tothrive? Child Care Health Dev. 1996;22:241250

6. Reif S, Beler B, Villa Y, Spirer Z. Long-term follow-up and outcome ofinfants with non-organic failure to thrive. Isr J Med Sci. 1995;31:483489

7. Patton RG, Gardner LI. Influence of family environment on growth: thesyndrome of maternal deprivation. Pediatrics. 1962;12:957962

8. Reinhart JB. Failure to thrive. In: Noshpitz JD, ed. Basic Handbook of ChildPsychiatry. New York, NY: Basic Books; 1979:593599

9. Goldbloom RB. Failure to thrive. Pediatr Clin N Am. 1982;29:15116610. Bithoney WG, McJunkin J, Michalek J, Snyder J, Egan H, Epstein D. The

effect of multidisciplinary team approach on weight gain in nonorganicfailure to thrive children. J Dev Behav Pediatr. 1992;12:254258

11. Kessler D. Failure to thrive and pediatric undernutrition: historical andtheoretical context. In: Kessler D, Dawson P, eds. Failure to Thrive andPediatric Undernutrition. Baltimore, MD: Paul H. Brookes, Co; 1999:318

12. Winters NC. Feeding problems in infancy and early childhood. PrimPsychiatry. 2003;10:3034

13. Benoit D. Phenomenology and treatment of failure to thrive. ChildAdolesc Psychiatr Clin N Am. 1993;2:6173

14. Chatoor I. Feeding and other disorders of infancy. In: Tasman A, Kay J,Lieberman J, eds. Psychiatry. Philadelphia, PA: WB Saunders; 1997:683701

15. Chatoor I, Hirsch R, Ganiban J, Persinger M, Hamburger E. Diagnosinginfantile anorexia: the observation of mother-infant interactions. J AmAcad Child Adolesc Psychiatry. 1998;37:959967

16. Chatoor I. Feeding disorders in infants and toddlers: diagnosis andtreatment. Child Adolesc Psychiatr Clin N Am. 2002;11:163183

17. Chatoor I, Getson P, Menville E, et al. A feeding scale for research andclinical practice to assess mother-infant interactions in the first threeyears of life. Infant Ment Health J. 1997;18:7691

18. Chatoor I, Egan J. Nonorganic failure to thrive and dwarfism due tofood refusal: a separation disorder. J Am Acad Child Adolesc Psychiatry.1983;22:294301

19. Chatoor I, Ganiban J, Colin V, Plummer N, Harmon RJ. Attachment andfeeding problems: a reexamination of nonorganic failure to thrive andattachment insecurity. J Am Acad Child Adolesc Psychiatry. 1998;37:12171224

20. Chatoor I, Ganiban J, Hirsch R, Spurrell E, Reiss D. Maternal character-istics and toddler temperament in infantile anorexia. J Am Acad ChildAdolesc Psychiatry. 2000;39:743751

21. Hollingshead RJ. Four Factor Index of Social Status. New Haven, CT: YaleUniversity; 1975

22. Waterlow JC, Buzina R, Keller W, Lan JM, Nichaman MZ, Tanner JM.The presentation and use of height and weight data for comparing the

nutritional status of groups of children under the age of 10 years. BullWorld Health Organ. 1977;55:489498

23. Hamill PVV, Drizd TA, Johnson CL, Reed RB, Roche AF, Moore WM.Physical growth: National Center for Health Statistics percentiles. Am JClin Nutr. 1979;32:602629

24. Kuczmarski RJ, Ogden C, Grummer-Strawn LM, et al. CDC growthcharts: United States. Adv Data. 2000;314:127

25. Bayley N. Manual for the Bayley Scales of Infant Development. New York,NY: The Psychological Corporation; 1969

26. Chatoor I, Egan J, Getson P, Menvielle E, ODonnell R. Mother-infantinteractions in infantile anorexia nervosa. J Am Acad Child Adolesc Psy-chiatry. 1988;27:535540

27. Chatoor I, Getson P, Brassaux C, et al. How consistent are mother-infantinteractions observed in the laboratory? Scientific Proceedings of the An-nual Meeting of the American Academy of Child and Adolescent Psychiatry.1993;9:51

28. Bee HL, Barnard KE, Eyres SJ, et al. Prediction of IQ and language skillfrom perinatal status, child performance, family characteristics, andmother-infant interaction. Child Dev. 1982;53:11341156

29. Coates DL, Lewis M. Early mother-infant interaction and infant cogni-tive status as predictors of school performance and cognitive behaviorin six-year-olds. Child Dev. 1984;55:12191230

30. Crandell LE, Hobson RP. Individual differences in young childrens IQ:a social-developmental perspective. J Child Psychol Psychiatry. 1999;40:455464

31. Hoffman-Plotkin D, Twentyman CT. A multimodal assessment of be-havioral and cognitive deficits in abused and neglected preschoolers.Child Dev. 1984;55:794802

32. Sandgrund A, Gaines RW, Green AH. Child abuse and mentalretardation: a problem of cause and effect. Am J Ment Defic. 1974;7:327330

33. Mackner LM, Starr RH, Black MM. The cumulative effect of neglect andfailure to thrive on cognitive functioning. Child Abuse Negl. 1997;21:691700

34. Drewett RF, Corbett SS, Wright CM. Cognitive and educational attain-ments at school age of children who failed to thrive in infancy: apopulation-based study. J Child Psychol Psychiatry. 1999;40:551561

35. Drewett RF, Wolke D, Asefa M, Kaba M, Tessema F. Malnutrition andmental development: is there a sensitive period? A nested case-controlstudy. J Child Psychol Psychiatry. 2001;42:181187

36. Singer TW, Fagan JF. Cognitive development in the failure-to-thriveinfant: a three-year longitudinal study. J Pediatr Psychol. 1984;9:363383

37. Boddy J, Skuse D, Andrews B. The developmental sequelae of nonor-ganic failure to thrive. J Child Psychol Psychiatry. 2000;41:10031014

38. Dykman RA, Casey PH, Ackerman PT, McPherson WB. Behavioral andcognitive status in school-aged children with a history of failure tothrive during early childhood. Clin Pediatr. 2001;40:6370

39. Olson SL, Bates JE, Bayles K. Mother-infant interaction and the devel-opment of individual differences in childrens cognitive competence.Dev Psychol. 1984;20:166179


2004;113;e440PediatricsBenny Kerzner

Irene Chatoor, Jaclyn Surles, Jody Ganiban, Leila Beker, Laura McWade Paez andAnorexia

Failure to Thrive and Cognitive Development in Toddlers With Infantile

ServicesUpdated Information &

tmlhttp://pediatrics.aappublications.org/content/113/5/e440.full.hincluding high resolution figures, can be found at:

References

tml#ref-list-1http://pediatrics.aappublications.org/content/113/5/e440.full.hat:This article cites 34 articles, 4 of which can be accessed free

Citations

tml#related-urlshttp://pediatrics.aappublications.org/content/113/5/e440.full.hThis article has been cited by 6 HighWire-hosted articles:

Subspecialty Collections

orders_subhttp://pediatrics.aappublications.org/cgi/collection/eating_disEating Disorders

ubhttp://pediatrics.aappublications.org/cgi/collection/nutrition_sNutrition

anguage:learning_disorders_subhttp://pediatrics.aappublications.org/cgi/collection/cognition:lCognition/Language/Learning Disorders

nt:behavioral_issues_subhttp://pediatrics.aappublications.org/cgi/collection/developmeDevelopmental/Behavioral Issuesthe following collection(s):This article, along with others on similar topics, appears in

Permissions & Licensing

mlhttp://pediatrics.aappublications.org/site/misc/Permissions.xhttables) or in its entirety can be found online at: Information about reproducing this article in parts (figures,

Reprints http://pediatrics.aappublications.org/site/misc/reprints.xhtml

Information about ordering reprints can be found online:

rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.Grove Village, Illinois, 60007. Copyright 2004 by the American Academy of Pediatrics. All and trademarked by the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elkpublication, it has been published continuously since 1948. PEDIATRICS is owned, published, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

at Charles R Drew Univ on July 23, 2015pediatrics.aappublications.orgDownloaded from

2004;113;e440PediatricsBenny Kerzner

Irene Chatoor, Jaclyn Surles, Jody Ganiban, Leila Beker, Laura McWade Paez andAnorexia

Failure to Thrive and Cognitive Development in Toddlers With Infantile

http://pediatrics.aappublications.org/content/113/5/e440.full.htmllocated on the World Wide Web at:

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

of Pediatrics. All rights reserved. Print ISSN: 0031-4005. Online ISSN: 1098-4275.Boulevard, Elk Grove Village, Illinois, 60007. Copyright 2004 by the American Academy published, and trademarked by the American Academy of Pediatrics, 141 Northwest Pointpublication, it has been published continuously since 1948. PEDIATRICS is owned, PEDIATRICS is the official journal of the American Academy of Pediatrics. A monthly

at Charles R Drew Univ on July 23, 2015pediatrics.aappublications.orgDownloaded from

pediatrics 2004 chatoor e440 7

Documents