Page 1

Mental Disorders in Obese Children and Adolescents GILBERT VILA, MD, PHD, EWA ZIPPER, MD, MYRIAM DABBAS, MD, CATHERINE BERTRAND, PH, JEAN JACQUES ROBERT, MD, PHD, CLAUDE RICOUR, MD, PHD, MARIE CHRISTINE MOUREN-SIME´ ONI, MD, PHD Objective: To evaluate the type and frequency of psychiatric disorders in obese children and adolescents; to assess the correlation between psychopathology and severity of obesity; to explore the relationship between psychiatric disorders in obese children and obesity and psychopathology in their parents. Methods: One hundred fifty-five children referred and followed for obesity were evaluated (98 girls and 57 boys; age, 5 to 17 years). Psychiatric disorders were assessed through a standardized diagnostic interview schedule (K-SADS R) and self-report questionnaires completed by the child (STAIC Trait-anxiety and CDI for depression) or his (her) parents (CBCL or GHQ). These obese children were compared with insulin-dependent diabetic (IDDM) outpatient children (N ⫽ 171) on questionnaire data. Results: Eighty-eight obese children obtained a DSM-IV diagnosis, most often an anxiety disorder (N ⫽ 63). Psychological disorders were particularly pronounced in those obese children whose parents were disturbed. There was no correlation between severity of obesity in the child or his (her) parents and frequency of psychiatric disorders. Compared with diabetic children, they displayed significantly higher internalized and externalized questionnaire scores and poorer social skills. Conclusion: These results highlight the importance of including a child psychiatric component in the treatment of obesity, which must engage the whole family. Key words: child, adolescent, obesity, mental disorders, psychopathology, parents. BMI ⫽ body mass index; CBCL ⫽ Child Behavior Check-List; CDI ⫽ Child Depression Inventory; GHQ-28 ⫽ General Health Questionnaire; K-SADS ⫽ Kiddie Schedule for Affective Disorders and Schizophrenia, Present Episode Version; SES ⫽ socio-economic status; STAIC ⫽ State and Trait Anxiety Inventory for Children.


ruch and Touraine (1) were the first authors to show an interest in the psychological functioning of obese children and adolescents. Numerous authors have since attempted to describe psychopathological characteristics in these children, but few investigations have relied on standardized evaluation or control group methods. According to Dreyfus (2), 25% of obese children are free from psychological problems, 58% show no severe personality disorder, 15% have a personality disorder, and 2% display psychotic features. Renman et al. (3) found no difference between obese and normal weight adolescents on selfesteem, social skills, and mental health questionnaires (externalized and internalized symptom scores). These results confirm results from 2 previous controlled studies (4 –5). In contrast, more recent studies found more DSM-IV (6) disorders in obese adolescents (7). While some authors noted more depressive symptoms (8 –10) and lower self-esteem (10) in young obese patients, others found no difference on depression (11) and self-esteem (12) between obese and control subjects. Results concerning anxiety in adolescents are divergent: 3 studies showed no difference between obese and control subjects (10,12– 13), while one revealed higher levels of anxiety in obese adolescents (8). Those adolescents with severe obesity, those who seek treatment or are in the process of being treated, and those whose weight is increasing seem to display more psychopathology, particularly higher levels of depression, than their normal weight counterparts (8,10). The relation between severity of obesity and psychopathology is not clear (13). We need more studies based on diagnostic criteria (10) rather than global or quantitative evaluations (1–5,9,11–

From the Pediatric Department, Necker-Enfants Malades Hospital, Paris, France. Address correspondence and reprint requests to G. Vila, PhD, CHU Necker-Enfants Malades, 149 rue de Se`vres, 75015 Paris, France. E-mail: Received for publication January 28, 2003; revision received December 2, 2003. Psychosomatic Medicine 66:387–394 (2004) 0033-3174/04/6603-0387 Copyright © 2004 by the American Psychosomatic Society

12). There are no such studies on prepubertal children, although some investigations suggest that children have more behavioral problems than teenagers (14). There is a remarkable lack of studies (recent or not) regarding the presence of behavioral disturbances in obese children and adolescents and whether obese children have a different psychopathological risk than obese adolescents. On a developmental point of view, adolescents and children are different, and psychosocial aspects such as body image and social regard may have more psychopathological implications for obese adolescents. The personality characteristics described by Bruch and Touraine (1) in parents of obese children and adolescents have not been confirmed by subsequent studies (15). Family patterns of eating behaviors have often been incriminated in obese children’s and adolescents’ eating disturbances (16). With regard to the prevalence of psychiatric disorders in families of young obese patients, 1 study indicated a significantly higher frequency of depressive disorders in parents of obese children and adolescents compared with parents of non-obese children (10). According to Favaro and Santonastaso (17), maternal characteristics are more relevant than paternal characteristics in predicting the course of obesity in children; mothers of obese children who suffered more severe psychiatric and/or personality disorders had children with more severe obesity. Furthermore, mothers of children with the lowest weight loss after 1 year of treatment had higher BMI and neurotic scores. Other investigations confirmed the central role of mothers’ behaviors and psychopathology in determining children’s eating habits (18). Epstein et al. (19) suggested that the severity of obesity contributed little to the variance of psychosocial problems in the child compared with the relative importance of the mother’s psychopathology and family’s socioeconomic status. These incomplete and contradictory studies call for stronger data on psychopathology in young obese children, particularly prepubertal children, using control groups and diagnostic criteria, as well as on its association with severity of obesity and parental psychopathology. Our main hypothesis was that young obese patients would display more psychopathology than non-obese patients. Our 387

G. VILA et al. objectives were: to study the type and frequency of DSM-IV mental disorders in a pediatric population of obese patients, using a standardized diagnostic tool; to compare their psychiatric problems with those of non-obese patients suffering from a metabolic illness; to assess the correlation between child psychopathology and severity of obesity; to evaluate the influence of socio-demographic variables on psychiatric disorders in the study sample; finally, to study the relationships between psychopathology and severity of obesity in obese children and their parents. MATERIALS AND METHODS Subject Samples The study population comprised 155 obese children and adolescents (57 boys and 98 girls) who were consecutive outpatients in the Department of Pediatric Nutrition at Necker-Enfants-Malades Hospital in Paris (public consultations for overweight children). Their ages ranged from 5 to 17 years, with a mean age of 11.3 (⫾2.5). All children had a physical examination on an outpatient basis and subjects with secondary obesity (ie, caused by a somatic disease as for example Cushing syndrome), mental retardation (WISC-III: IQ ⬍70), pervasive developmental disorders, or psychotic disorders were excluded from the study. The weight of the child was computed with z-scores (20), subjects’ body mass index (BMI) compared with the ideal BMI of the general population of the same gender and age (21): BMI z-scores ranged from ⫹2 to ⫹14 (BMI mean z-score: ⫹6.1 ⫾ 2.2). Fifty-seven percent of the subjects came from families with low socioeconomic status (SES) 27% from middle SES, and 16% from high SES. Mother’s mean BMI was 26.8 ⫾ 6.3; father’s mean BMI was 26.7 ⫾ 4.1. The obese sample was compared with a sample of young IDDM outpatients, followed on a regular basis in the same hospital facility (N ⫽ 171). Eighty-three girls and 88 boys were included, aged 5 to 18 years (mean 13.2 ⫾ 3.4). The SES distribution was as follows: 19% low SES, 48% middle SES, and 32% high SES. Their BMI mean z-score was ⫹0.4 ⫾ 0.9. Their glycosylated hemoglobin level was 9.3 ⫾ 1.5% (mean HBA1c). The mean duration of IDDM was 5.8 ⫾ 3.7 years. This IDDM population contained a significantly higher proportion of boys (Pearson chi-squared ⫽ 7.1; p ⫽ .008), was older (Student t test ⫽ 5.4; p ⬍ .001) and had a higher SES (Pearson chi-squared ⫽ 45.1; p ⬍ .001) than obese outpatients.

Instruments The Child Behavior Checklist (CBCL) and the General Health Questionnaire (GHQ-28) were administered to parents (22–23). Children were assessed with 2 self-reports independently of the presence of the parents: the STAIC, trait-version (24), and the CDI (25). Obese children were also assessed with a semi-structured interview, the Kiddie Schedule for Affective Disorders and Schizophrenia, Present Episode Version (26), modified for DSM-IV diagnoses (K-SADS-R). Collection of data were conducted by an experienced child psychologist trained in the use of these techniques and blind to both medical treatment and severity of illness. The State and Trait Anxiety Inventory for Children (STAIC) (24) was translated into French (27). It is composed of 2 scales of 20 items each, describing psychological manifestations of anxiety; the State scale describes the current state of anxiety, and the Trait scale describes a general propensity to react with anxiety to stressful events. We only used the Trait version to objectify the more lasting manifestations of anxiety. Each item is scored from 1 to 3, the total score for each scale thus ranging from 20 to 60. The cut-off score for pathological anxiety in children (DSM-IV anxiety disorders for nonreferred children), as validated in the French version of the STAIC-Trait Inventory, is 34 with a sensitivity of 0.63 and a specificity of 0.75 (27). The Child Depression Inventory (CDI) is a 27-item self-administered depression questionnaire for children (25). It was initially developed for 8- to 13-year-old children, but its current version was validated on large populations of 10- to 18-year-old youngsters, with a cut-off score of 13 for depression. It was translated into French (28) and is widely used in France. 388

The CBCL is a 138-item questionnaire, to be completed by the parents (22). It provides a standardized description of skills and emotional and behavioral problems in 4- to 16-year-old children. The first part concerns social skills, participation in organizations, contact with friends, participation and skills in sports, and academic performance ratings. The second part deals with emotional and behavioral problems classified as internalized or externalized. Internalized behavior problems are those that essentially affect the child himself; they include anxious/depressed, withdrawn, schizoid, and somatic behavioral problems. Externalized problems are overt in nature, with direct effects on others: they include delinquency as well as cruel and aggressive behavior. The French version of the CBCL has been demonstrated to be valid and reliable, using a community sample of 2480 children (28 –29). The GHQ-28 is a 28-item self-administered questionnaire with scores of 0 or 1 on each item, completed by parents on their own mental health status (23). Higher scores reflect more severe pathological states. This instrument has been translated into French, validated on a French population (30), and used in numerous studies. It was designed to screen subjects with psychopathological states, and its ease of administration and good sensitivity (testretest reliability, 0.90; internal consistency, 0.90; sensitivity, 94%) make it a very useful tool. It does not detect specific categories of psychiatric disorders (screening specificity: 73%). The K-SADS-R was a semi-structured interview designed by Chambers et al. (31) to record information about the mental state of children and to determine whether they were suffering from any psychiatric disorders as assessed by the DSM-III. A French version has been used for psychiatric populations (32). Kiddie Schedule for Affective Disorders and Schizophrenia (K-SADS) had several limitations (26) and was revised to produce the lifetime version (K-SADS-L; R. Klein et al., 1993, New York State Psychiatric Institute, unpublished). It meets DSM-IV criteria and assesses current and past psychopathology. Most affective symptoms are only rated for the current episode and past 2 weeks. It has been translated into French (K-SADS-R) in collaboration with R. Klein, at Necker-Enfants-Malades Hospital (Paris), where it is currently used for evaluations. For this study, we only considered current episodes of mental disorders because prospective data are more valid than retrospective data. In a preliminary validation study, we found that the French version performed satisfactorily and had good concurrent validity for diagnoses of DSM-IV anxiety disorders, depression (major depressive disorder, dysthymic disorder) against standard self-report scales (STAIC-trait, ECAP, CDI, CBCL). Compared with normal subjects (N ⫽ 72; Mann-Whitney-test), K-SADS-R anxiety disorders (N ⫽ 52) had significantly higher STAIC-Trait (37.0 vs. 29.1), ECAP (127.1 vs. 103.4), CBCL-Internalizing (65.6 vs. 50.8), and CBCL-anxiety/ depression (66.3 vs. 53.4) mean scores; K-SADS-R affective disorders (N ⫽ 18) had significantly higher CDI (17.7 vs. 8.1), CBCL-Internalizing (67.8 vs. 54.3), and CBCL-anxiety/depression (69.7 vs. 56.2) mean scores; disruptive behavior disorders (N ⫽ 8) had significantly higher CBCLExternalizing (66.0 vs. 48.4) mean scores (27).

Design All obese children older than 5 years were consecutively recruited during a 1-year period from the Department of Pediatric Nutrition of Necker-Enfants Malades Hospital in Paris after their first consultation. One hundred fiftyeight families were asked to participate in the study. All subjects were informed of the aims and methods of the study and they gave informed consent, in accordance with French legal guidelines. One boy and two girls were excluded because their BMI was below ⫹2 z-scores. All obese subjects and their parents were interviewed with the K-SADS-R (N ⫽ 155); 10 obese children were below the age of 8 years (5–7 years) and were assessed only with the K-SADS-R and the CBCL; 141 obese children completed the STAIC-Trait, 139 the CDI; 116 families completed the CBCL, 117 mothers and 25 fathers the GHQ-28; BMI were computed for all the fathers and mothers. During a 1-year period, all IDDM children who were 5 to 18 years of age were serially recruited from the Pediatric Unit of Diabetology of NeckerEnfants Malades Hospital in Paris. Parents and children were informed of the objectives and procedure of the study and gave their consent. Two hundred five families were asked to participate. Twenty-five assessments were incomPsychosomatic Medicine 66:387–394 (2004)

MENTAL DISORDERS IN OBESE CHILDREN plete because parents lacked the time to return to the hospital during the study. Nine IDDM subjects were excluded because their BMI was greater than ⫹2 z-scores (N ⫽ 171). One hundred sixty-one IDDM subjects completed the STAIC-Trait and the CDI, and 155 families completed the CBCL, 122 mothers and 25 fathers the GHQ-28.

Statistical Analyses Results are expressed in frequencies, means, and standard deviations. Data were processed with the statistics program SPSS-PC⫹. Correlations (2-tailed) were calculated using Pearson coefficient and Spearman rho coefficient for semiquantitative variables. Relationships between variables were assessed through logistic regression analyses. For quantitative variables, group means were compared using Student t test (2-tailed). Results from the evaluations were analyzed with nonparametric statistics when the number of subjects was small, for quantitative (Mann-Whitney test) or qualitative (Fisher exact test) variables. The 2 groups (obese and IDDM) were significantly different with regard to children’s mean age, sex-ratio, and family SES. As a result, these variables were entered in the computations, and the ANOVA analyses took this difference into account by entering gender, age and SES as covariables, for all obese and IDDM comparisons involving quantitative variables. The significance threshold was set at p ⫽ .05.

RESULTS DSM-IV Diagnoses in Obese Children and Adolescents Diagnostic interviews with Kiddie-SADS-R identified 88 (58%; 54 girls and 34 boys) obese children with at least 1 DSM-IV psychiatric diagnosis. Sixty-three subjects (32%; 39 girls and 24 boys) had at least 1 DSM-IV anxiety disorder (34 social phobias, 14 generalized anxiety disorders, 11 separation anxiety disorders, 2 agoraphobias, and 2 posttraumatic stress disorders). Nineteen children (12%; 16 girls and 3 boys) had an affective disorder (9 major depressions and 10 dysthymic disorders). Twenty-five children (16%; 11 girls and 14 boys) had a disruptive behavior disorder (18 defiant oppositional disorders, 5 attention deficit hyperactivity disorders, and 2 conduct disorders). DSM-IV disorders were associated in many cases (12 children with anxiety disorders and affective disorders, 8 with anxiety disorders and conduct disorders, 1 with conduct disorder and affective disorder, and 1 with the 3 types of disorders). Screening for Mental Disorders With Questionnaires Completed by Obese Children (STAIC and CDI) or by Parents (CBCL) According to self-report questionnaires, 65 obese subjects had an anxiety score above 34 on the STAIC-Trait scale. Twenty-four obese children had depression scores above 13 TABLE 1.

on the CDI (Table 1). Self-assessment questionnaires identified psychiatric problems in 70 patients (50% of those who completed these questionnaires) (with a STAIC-Trait or a CDI score above the cutoff point). Fifty obese children obtained a CBCL internalization score that was above 60. Twenty-nine had a CBCL externalization score that was above 60 and could thus be assumed to have behavioral problems. Forty-five had a CBCL total-score of psychiatric symptoms greater than 60 (Table 1). Parental assessment identified psychiatric problems in 58 young patients (50% of those who completed the CBCL) (with a CBCL Total-score, or a CBCL internalization score, or a CBCL externalization score greater than 60). Positive results were obtained for 90 obese children on at least 1 of the parent or child assessment (64% of those who completed these questionnaires). Diagnostic Agreement for Obese Children Between Different Sources of Information: Clinician, Child, and Parents Table 1 shows the number of cases for which there was agreement on the different DSM-IV diagnoses (N ⫽ 88) made by the clinician (Kiddie-SADS-R) and derived from the screening questionnaires (STAIC-Trait, CDI, and CBCL). Among the 88 subjects with a DSM-IV clinician diagnosis, 66 (75%) scored positive on at least 1 child questionnaire (with a STAIC-Trait score greater than 34 or a CDI score greater than 13) or on 1 parent questionnaire (CBCL Total-score or CBCL internalization score or CBCL externalization score greater than 60); 52 (59%) subjects with a DSM-IV clinician diagnosis scored positive on at least 1 child questionnaire (STAIC-Trait or CDI), and 46 (52%) on at least 1 parent scale (CBCL Total-score, internalization score, or externalization score). STAIC-Trait scores significantly distinguished obese children suffering from DSM-IV anxiety disorders (STAIC-T ⫽ 37.5 ⫾ 6.8) from those without anxiety disorders (STAIC-T ⫽ 30.2 ⫾ 7.4; t ⫽ 5955; p ⬍ .001). The subgroup suffering from anxiety had internalization-CBCL mean scores of 61.8 ⫾ 10.4, which were significantly different from those of other obese children (int-CBCL ⫽ 56.3 ⫾ 10.8; t ⫽ 2.780; p ⫽ .006). STAIC-Trait scores of obese children were significantly correlated with scores from the parent questionnaire (CBCL) on anxiety in their children (internalization-CBCL: r ⫽ 0.39, p⬍.001; anxiety-depression CBCL subscore: r ⫽ 0.35,

Agreement Between Kiddie-SADS-R Diagnoses and Significant Questionnaire Scores for Obese Children and Adolescents

N Anxiety disorder Affective disorders DBDa

Diagnoses K-SADS-R


CDI ⱖ13

Tot-CBCL ⱖ60

Int-CBCL ⱖ60

Ext-CBCL ⱖ60

88 63 19 25

65 44/61 (0.72)b 17/18 (0.94) —

24 15/58 (0.26) 12/19 (0.63) —

45 25/54 (0.46) 11/12 (0.92) 14/21 (0.67)

50 31/54 (0.57) 12/12 (1) —

29 — — 10/21 (0.48)


DBD, Disruptive Behavior Disorders. Rate of agreement between DSM-IV diagnosis and questionnaire; for example, 72% of K-SADS-R anxiety disorders (61/63 patients with K-SADS-R anxiety disorders completed the STAIC) were positive to the screening by STAIC with a score above the cutoff. b

Psychosomatic Medicine 66:387–394 (2004)


G. VILA et al. p⬍.001). CBCL total scores correlated with STAIC-Trait (CBCL Total-score: r ⫽ 0.38, p⬍.001). The 19 patients with DSM-IV depression had CDI mean scores of 13.8 ⫾ 5.2 and Internalization-CBCL mean scores of 69.4 ⫾ 5.9, which were significantly different from those obtained in other obese children (CDI ⫽ 7.3 ⫾ 4.5, t ⫽ 5.658, p⬍.001; Internalization-CBCL ⫽ 57.6 ⫾ 10.8, t ⫽ 3.688, p⬍.001). CDI scores of obese children were significantly correlated with scores from the parent questionnaire (CBCL) on depression in their children (internalization-CBCL: r ⫽ 0.30, p ⫽ .002; anxiety-depression CBCL subscore: r ⫽ 0.22, p ⫽ .023). CBCL total scores correlated with CDI (CBCL Total-score: r ⫽ 0.28, p ⫽ .004). The 25 patients with a DSM-IV Disruptive Behavior Disorder had externalization-CBCL mean scores of 59.0 ⫾ 11.0 and Total-CBCL mean scores of 65.0 ⫾ 11.6, which were significantly different from those obtained in other obese children (ext-CBCL ⫽ 51.1 ⫾ 10.3, t ⫽ 3.140, p ⫽ .002; Total-CBCL ⫽ 56.0 ⫾ 11.0, t ⫽ 3.352, p ⫽ .001). Association Between Psychiatric Variables and Each Socio-demographic Factor There was no significant correlation between the age of obese children and scores on any of the questionnaires (STAIC, CDI, or CBCL). Adolescents (age ⬎ 12 years, N ⫽ 56) did not show a greater prevalence of mental disorders (37 [65%] vs. 51 [52%] cases), nor did they have significantly higher questionnaire mean scores (except CDI) than children (age ⬍ 13 years, N ⫽ 99) (Table 2). Obese teenagers had more depressive disorders (11 vs. 8) and obese children had more disruptive behavior disorders (17 vs. 8) and anxiety disorders (37 vs. 26). They had also poorer CBCL activities scores than children. There was no significant difference between obese boys and girls (Fisher’s exact test ⫽ 0.611) with regard to the overall prevalence of mental disorders. Girls had significantly TABLE 2. Mean Scores (ⴞ SD) in Self-Report Questionnaires for Obese Children (age <13 yr) and adolescents (age >13 yr) Children N ⫽ 99 STAIC-trait CDIa CBCL Total score Internalization Externalization CBCL-anxiety/depression CBCL social competences Activities* Social School GHQ-28 Mother Father a

Adolescents N ⫽ 56

33.3 (8.0) 7.5 (4.6)

33.4 (8.0) 9.3 (5.8)

57.4 (12.8) 59.4 (11.5) 52.0 (11.7) 60.2 (10.7)

58.3 (8.8) 58.1 (9.9) 53.7 (9.0) 58.2 (7.8)

43.8 (7.9) 37.7 (10.3) 45.0 (9.4)

38.2 (7.2) 38.2 (7.2) 43.0 (8.8)

4.4 (7.0) 1.1 (3.1)

4.5 (5.7) 4.4 (8.0)

p ⬍ .05: comparison between groups with Student’s t test.



Mean Scores (ⴞ SD) in Self-Report Questionnaires for Obese Girls and Boys

STAIC-trait CDI CBCL Total score Internalization Externalization CBCL-anxiety/depression CBCL competences Activitiesa Social School GHQ-28 Mother Father a

Girls N ⫽ 85

Boys N ⫽ 55

33.6 (8.3) 8.7 (5.7)

33.0 (7.6) 7.4 (4.2)

56.4 (10.3) 57.8 (10.4) 51.6 (10.7) 58.5 (8.7)

60.6 (12.8) 61.3 (11.5) 54.8 (10.5) 61.4 (11.4)

44.3 (7.7) 36.8 (9.3) 44.1 (9.7)

39.9 (7.1) 39.4 (8.7) 44.6 (8.4)

4.5 (6.3) 3.6 (6.7)

4.6 (7.0) 0.3 (0.6)

t ⫽ 3.028; p ⫽ 0.003: Comparison between girls and boys with Student’s t test.

more affective disorders (16/95 vs. 3/55; Fisher’s exact test ⫽ 0.046) and boys more disruptive behavior disorders (14/57 vs. 11/95; Fisher’s exact test ⫽ 0.044). Girls had better CBCL activities scores than boys. Table 3 shows differences in mean questionnaire scores between boys and girls. There was no significant association between socioeconomic status and mean questionnaire scores (tested by ANOVA) for anxiety (STAIC), depression (CDI), and psychiatric symptoms (CBCL: Total score, internalization, and externalization). There was a significant association between SES and overall prevalence of mental disorders (chi-squared test ⫽ 8.127, p ⫽ .017) or prevalence of anxiety disorders (chi-squared test ⫽ 10.503, p ⫽ .005), with more disorders in children from lower SES. Association Between Parent and Child Psychiatric Variables Table 4 shows that the higher the frequency of psychological disturbances in mothers (GHQ-28 scores), the higher the symptom measures in children. Mothers of obese children with a diagnosis of at least 1 DSM-IV mental disorder had GHQ-28 scores that were significantly higher than in other families (GHQ-28: 5.7 ⫾ 6.5 vs. 2.8 ⫾ 4.8; Student’s t test ⫽ 2.921; p ⫽ .004). A DSM-IV diagnosis of TABLE 4. Correlations [r (p)] Between Mother or Father GHQ-28 Scores and Questionnaire Scores (Children and Parents) of Obese Patients

CBCL-total CBCL-internalization CBCL-externalization CBCL-anxiety/depression STAIC-Trait CDI

GHQ-28mother (N ⫽ 117)

GHQ-28father (N ⫽ 25)

0.47 (⬍0.001) 0.46 (⬍0.001) 0.40 (⬍0.001) 0.50 (⬍0.001) 0.20 (0.032) 0.11 (0.2)

0.08 (0.7) 0.20 (0.4) 0.02 (0.9) 0.04 (0.9) 0.12 (0.6) ⫺0.16 (0.9)

Psychosomatic Medicine 66:387–394 (2004)

MENTAL DISORDERS IN OBESE CHILDREN affective disorder in obese children was associated with higher scores on maternal psychopathology than in the absence of a diagnosis of child depression (GHQ-28: 8.2 ⫾ 82 vs. 4.0 ⫾ 6.2; Student’s t test ⫽ ⫺2.367; p ⫽ .020). We did not find any significant difference for mother GHQ on diagnoses of anxiety or disruptive behavior disorders or for father GHQ. Mother or father GHQ-28 did not significantly distinguish obese and IDDM groups (respectively mother GHQ: 4.5 ⫾ 6.6 vs. 4.4 ⫾ 6.1; respectively father GHQ: 2.1 ⫾ 5.2 vs. 2.5 ⫾ 4.1). The clinical interview revealed family problems in 60 out of 153 (39%) cases. Externalization and internalization CBCL scores were higher in obese children who lived in these problem families (CBCL-Externalization: 56.1 ⫾ 10.7 vs. 50.3 ⫾ 10.3, Student’s t test ⫽ 2.922, p ⫽ .004; CBCL-Internalization: 61.8 ⫾ 9.4 vs. 56.9 ⫾ 11.5, Student’s t test ⫽ 2.403, p ⫽ .018), and DSM-IV affective disorders (14/59 vs 4/85; Fisher exact test ⫽ 0.001) or disruptive disorders (15/60 vs 9/90; Fisher exact test ⫽ 0.022) were more frequent. Association Between Psychiatric Variables and BMI The CDI scores of obese patients were negatively correlated with their BMI (r ⫽ ⫺0.21; p ⫽ .019), but there was no significant association between a DSM-IV diagnosis of depression and lower mean BMI z-scores (⫹5.6 ⫾ 1.9 vs. ⫹6.2 ⫾ 2.2; t ⫽ 984; p ⫽ .3). “Super obese” patients (defined by a BMI z-score ⱖ ⫹8, ie, the median BMI z-score between ⫹2 and ⫹14) had significantly lower STAIC-Trait (29.1 ⫾ 5.2 vs. 34.3 ⫾ 8.3) and CDI (6.1 ⫾ 4.6 vs. 8.8 ⫾ 5.2) mean scores than other obese children (respectively Student’s t test: t ⫽ 2.795, p ⫽ .006; t ⫽ 2.180, p ⫽ .031). There was no other significant association between obese child or parent BMI (mother or father) and questionnaire scores or the presence of a DSM-IV diagnosis. Parents’ and child’s BMIs were positively correlated for mothers (r ⫽ 0.17; p ⫽ .042) but not for fathers (r ⫽ ⫺0.05; p ⫽ .6). Comparison Between Obese and IDDM Children and Adolescents on Questionnaire Scores IDDM and obese groups showed several significant differences. Obese patients had higher mean scores on STAICTrait, CBCL total T-score, CBCL internalizing T-score, CBCL externalizing T-score, CBCL anxiety/depression subscore, but not CDI (Table 5). An ANOVA using gender, age, and SES as covariables showed that these differences remained significant. Compared with IDDM children, obese children had significantly poorer social skills and the same level of activities and school performance on the CBCL (Table 4); corrected with age, sex, and SES, this difference remained significant. We compared IDDM and obese groups with regard to the number of subjects who yielded positive results on the various instruments. The obese group showed significantly more cases of externalized (29/116 vs. 19/155; p ⫽ .010) or internalized (50/116 vs. 40/155; p ⫽ .004) Psychosomatic Medicine 66:387–394 (2004)


Mean Scores (ⴞ SD) in Self-Report Questionnaires for Diabetic and Obese Children

STAIC-Traita CDI CBCL Total scoreb Internalizationb Externalizationb CBCL-anxiety/depressionb CBCL social competencies Activities Sociala School GHQ-28 Mother Father



31.1 (7.0) 8.3 (5.8)

33.3 (8.0) 8.2 (5.1)

50.8 (10.5) 53.4 (10.4) 48.1 (9.5) 55.1 (6.4)

57.7 (11.5) 58.9 (10.9) 52.6 (10.8) 59.5 (9.8)

42.6 (7.7) 40.6 (8.0) 45.9 (8.3)

42.7 (7.7) 37.9 (9.2) 44.3 (9.2)

4.4 (6.1) 2.5 (4.1)

4.5 (6.6) 2.1 (5.3)

p ⱕ .01; b p ⬍ .001: Comparison between IDDM and obese groups with Student’s t test.


disorders on the CBCL and a trend for more anxiety-positive cases (65/141 vs. 58/161; p ⫽ .048 for one-way hypothesis), on the basis of the STAIC cutoff score; 58/116 obese children vs. 50/155 IDDM patients (p ⫽ .004) were positive on at least one of the CBCL subscale (internalization or externalization or total score) and 90/128 vs. 86/153 (p ⫽ .019) were positive on at least one scale (STAIC or CDI or CBCL-internalization or -externalization or -total score). A logistic regression analysis with group status (obese or IDDM) as dependent variable and age (p ⬍ .001), sex (p ⫽ .048), SES (p ⬍ .001), mother GHQ-28 (p ⫽ .045), schoolCBCL (p ⫽ .060), CBCL-Internalization (p ⫽ .030), and CBCL-Externalization (p ⫽ .013) as independent variables, yielded a significant model (chi-squared ⫽ 68.157; p ⬍ .001); all variables remained significant in the variance of the dependent variable (obesity vs. IDDM). DISCUSSION Our objective was to evaluate the frequency of psychiatric disorders in young overweight patients referred to a specialty clinic for their obesity. Mental disorders were frequent in our population of obese children and adolescents, as measured by standardized methods (58% on the basis of K-SADS-R; between 50% and 64% as measured by questionnaire screening techniques). French epidemiological data based on a community sample of school-aged children indicated a prevalence of mental disorders of 5% (29). By far, the most frequent disorders were anxiety disorders. Separation anxiety and social phobia appeared to be more frequent in these children and adolescents (10). This observation converges with the classical portrayal of these children as being excessively dependent on their family environment, as displaying excessive attachment to parental figures associated with separation anxiety and fear of criticism, and as avoiding social contact, all of which confine them to familiar surroundings. Poor social compe391

G. VILA et al. tences of obese children and adolescents confirm this pattern, as poorer activities scores in obese adolescents than in obese children. It is nevertheless difficult to determine whether this relational pattern precedes or follows obesity, because of the social handicap imposed by our cultural environment on these overweight children. Behavior disorders, which have been rarely studied in obese children, seem to be less frequent (16% on the basis of K-SADS-R; 25% as measured by questionnaire screening procedures). Pine et al. (33) have shown in a longitudinal study that conduct disorder symptoms in adolescence predicted BMI and obesity in early adulthood. We did not find and had not to exclude any case of severe mental disorder as schizophrenia, although some authors have shown the high frequency of obesity in these patients, particularly when treated with atypical antipsychotic drugs (34); this type of severe psychiatric patient does not consult obesity centers and seems to have insufficient pediatric care. None of the obese patients had a psychiatric medication. Only 12% of our obese subjects had a DSM-IV disorder of depression. The prevalence rate was low for depression (2% to 4% with questionnaires), close to prevalence figures observed in the general population. This is in contrast with results from other studies, although our results confirm the low frequency of DSM-IV major depressive disorder (10). The high homogeneity of our recruitment population may explain these differences. All patients were assessed shortly after the beginning of treatment, most likely at a time when motivation for losing weight was high. Furthermore, our subjects were young and had not been confronted (or minimally so) with multiple experiences of failure regarding weight loss, as opposed to those obese adults who are in a depressed state of learned helplessness (35). One may surmise that dietary efforts or failures generate depression, rather than obesity itself, which children often adapt to (personally and socially) better than adults. In our results, the lack of correlation between children’s BMI and mental disorders partially supports this notion. For Goodman and Whitaker (36), depression predicts obesity, but there is no evidence for obesity predicting depression. To our knowledge, this study on obese children is the first one to use a validated diagnostic interview and DSM-IV criteria. The validity of our results was enhanced, and biases limited, by the use of standardized methods of evaluation and multiple sources of information. There is good concordance between data obtained from the diagnostic interview and from the questionnaires, as well as between parent and child questionnaires. With regard to internalized disorders, children were better evaluators of their internal state than their parents, despite the fact that the small size of our depressed sample (N ⫽ 19) limited statistical power. The congruency of DSM-IV diagnoses and questionnaire scores, which reflects clinical relevancy, nevertheless reinforces the validity of our diagnoses. When sources of information were cross-analyzed by pairs to maintain methodological rigor, the prevalence of disorders remained high in 392

obese children (24% out of the 155 obese subjects for at least 1 mental disorder; 6% to 10% for externalized disorders; 17% to 25% for internalized disorders; 14% to 23% for anxiety disorders). Our findings must take recruitment biases into account. The sample limits generalizablity to children seeking treatment for obesity. One would expect that there might be an overrepresentation of psychiatric illness in patients seeking help for their obesity in a specialty clinic, compared with the broader population of obese children. This help-seeking population may overselect for anxiety disorders. Our results cannot be generalized to all obese children. It would be worthwhile to replicate our study on a population of overweight children followed in private practice or who are not clinically-referred. In this culture, obesity is a significant source of stress that may contribute to the development of psychological disorders and to the maintenance of obesity. But another hypothesis may be that psychopathology (and family interactions between child and parent disorders), particularly anxiety, may influence the patient’s initial level of motivation to seek professional help. We did not find any association between severity of obesity and psychopathology, in agreement with other studies (18 – 19), but contrary to the results of Favaro and Santonastaso (17) based on a younger population (4 to 13 years). Our data suggest that “super-obese” youngsters may be less anxious and depressed, perhaps because they have been obese for a long time or because they have become fatalistic regarding the large amount of weight they have to lose. “Super-obesity” could be related to lack of personal or familial attention to and anxiety about weight gain. Our results suggest that the presence of psychopathology in obese youngsters is highly correlated with parents’ psychopathology, and to a lesser extent with psychosocial problems in the family and its socioeconomic status. Age and sex play a minor role (obese adolescents do not seem to have a higher psychopathological risk than obese children; as in the general population, boys have more behavior disorders and girls more emotional disorders). One limitation of this study is associated with fathers’ poor cooperation and lesser involvement in the problems of their overweight child. The positive correlation found between the GHQ and parent (most often mother) questionnaire scores (CBCL) did not hold for child questionnaire scores (Trait-STAIC and CDI). This raises the possibility that parents’ judgment was contaminated by their own psychopathology. Nevertheless, child-questionnaire scores (Trait-STAIC and CDI) correlated with parent-questionnaire scores (CBCL), showing good correspondence between the judgment provided by the parent and the child on his (or her) psychological state. Johnson et al. (37) have shown in a community-based prospective longitudinal study that childhood adversities as parental psychopathology or maladaptative parental behavior is associated with eating and weight problems during adolescence that persist into early adulthood. Our findings tend to highlight the importance of involving the family in the treatment of childhood obesity, in both its dietary and psychological/psychiatric dimensions (10,16). Obesity is Psychosomatic Medicine 66:387–394 (2004)

MENTAL DISORDERS IN OBESE CHILDREN often familial in nature, yet the child’s psychological problems are not associated with parental obesity; they are associated with parents’ psychological and psychosocial problems. One may hypothesize that the child’s distress (his experience of obesity) is linked to the mother’s own distress, irrespective of the severity of obesity, whereas the child’s tendency to overeat is linked to the mother’s eating behaviors (as suggested by the correlation between severity of maternal and child obesity observed in most studies). Our obese patients were compared with diabetic patients. This choice of a diabetic control group, as opposed to a community sample, seemed justified by the fact that both populations were followed in specialty clinics and therefore were in the role of sick children. In addition, insulin-dependent diabetes is a metabolic disease that requires dietary management. These young subjects were followed on an outpatient basis and were not burdened by somatic complications (as opposed to diabetic adults); disregarding sociodemographic factors, obese children recruited in a specialty clinic had significantly more disorders than diabetic patients, both in terms of severity and frequency. This finding confirms the fact that obesity presents a high risk for psychopathology that is as great as or greater than a somatic disease involving significant stress. Parents from both groups did not differ in this respect. CONCLUSIONS In highlighting the high frequency of mental disorders in obese children and the importance of parental psychopathology, our findings show that it is essential to include a child psychiatric component in the treatment of obese children, which involves the whole family. It is crucial that the child’s mental disorders be identified and treated and that family dynamics be improved to enhance the quality of life of these patients, reinforce their motivation (often sorely tried over time), work on relapse prevention, and perhaps increase rates of therapeutic success. An ongoing prospective study will help us assess the predictive value of mental disorders in the course of obesity. REFERENCES 1. Bruch H, Touraine G. Obesity in childhood. Psychosom Med 1940;2: 141–206. 2. Dreyfus M. Abord psychologique de l’obe´ site´ de l’enfant et de l’adolescent dans une consultation pluridisciplinaire. Annales de Pe´ diatrie 1993;40:305–12. 3. Renman C, Engstrom I, Silfverdal SA, Aman J. Mental health and psychosocial characteristics in adolescent obesity: a population-based case-control study. Acta Paediatr 1999;88:998 –1003. 4. Quaade F. Obese children: anthropology and environment. Copenhaguen, Dansk Videnskabs Forlag; 1955. 5. Sallade J. A comparison of the psychological adjustment of obese versus non-obese children. J Psychosom Res 1973;17:89 –96. 6. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders DSM-IV. 4th ed. Washington DC: American Psychiatric Association; 1995. 7. Buddeberg-Fischer B, Klaghofer R, Reed V. Associations between body weight, psychiatric disorders and body image in female adolescents. Psychother Psychosom 1999;68:325–32. 8. Hammar SL, Campbell MM, Campbell VA, Moores NL, Saren C, Gareis FJ, Lucas B. An interdisciplinary study of adolescent obesity. J Pediatr 1972;80:373– 83. Psychosomatic Medicine 66:387–394 (2004)

9. Strauss CC, Smith K, Frame C, Forehand REX. Personal and interpersonal characteristics associated with childhood obesity. J Pediatr Psychol 1985;10:337– 43. 10. Isnard-Mugnier P, Vila G, Nollet-Cle´ menc¸ on C, Vera L, Rault G, Mouren-Simeoni MC. Etude controˆ le´ e des conduites alimentaires et des manifestations e´ motionnelles dans une population d’adolescentes obe`ses. Archives Franc¸ aises de Pe´ diatrie 1993;50:479 – 84. 11. Wadden TA, Foster GD, Stunkard AJ, Linowitz JR. Dissatisfaction with weight and figure in obese girls: discontent but not depression. Int J Obes 1989;13:89 –97. 12. Pastore DR, Fischer M, Friedman SB. Abnormalities in weight status, eating attitudes, and eating behaviors among urban high school students: correlations with self-esteem and anxiety. J Adolesc Health 1996;18:312–9. 13. Britz B, Siegfried W, Ziegler A, Lamertz C, Herpertz-Dahlmann BM, Remschmidt H, Wittchen HU, Hebebrand J. Rates of psychiatric disorders in a clinical study group of adolescents with extreme obesity and in obese adolescents ascertained via a population based study. Int J Obes Relat Metab Disord 2000;24:1707–14. 14. Stradmeijer M, Bosch J, Koops W, Seidell J. Family functioning and psychosocial adjustment in overweight youngsters. Int J Eat Disord 2000;27:110 – 4. 15. Bernfeld J, Plazanet M. Le pe`re de l’enfant obe`se. Revue de Me´ decine Psychosomatique 1977;19:51– 64. 16. Klesges RC, Malott JM, Boshee PF, Weber JM. The effects of parental influences on children’s food intake, physical activity, and relative weight. Int J Eat Disord 1983;5:335– 46. 17. Favaro A, Santonastaso P. Effects of parents’ psychological characteristics and eating behaviour on childhood obesity and dietary compliance. J Psychosom Res 1995;39:145–51. 18. Drucker RR, Hammer LD, Agras WS, Bryson S. Can mothers influence their child’s eating behavior? J Dev Behav Pediatr 1999;20:88 –92. 19. Epstein LH, Myers MD, Anderson K. The association of maternal psychopathology and family socioeconomic status with psychological problems in obese children. Obes Res 1996;4:65–74. 20. Cole TJ. The LMS method for constructing normalized growth standards. Eur J Clin Nutr 1990;44:45– 60. 21. Rolland-Cachera MF, Cole TJ, Sempe´ M, Tichet J, Rossignol C, Charraud A. Body Mass Index variations: centiles from birth to 87 years. Eur J Clin Nutr 1991;45:13–21. 22. Achenbach TM, Edelbrock CS. Manual for the Child Behavior Checklist and Revised Child Behavior Profile. Burlington, Vermont, University of Vermont, Department of Psychiatry; 1983. 23. Goldberg DP. Manual of the General Health Questionnaire. Windsor: NFER Publishing; 1978. 24. Spielberger CD. Manual for the State Trait Anxiety Inventory for Children (STAI-C). Palo Alto, California: Consulting Psychologist Press; 1973. 25. Kovacs M. The Children Depression Inventory (CDI). Psychopharmacol Bull 1985;21:995–98. 26. Kaufman J, Birmaher B, Brent D, Rao U, Flynn C, Moreci P, Williamson D, Ryan N. Schedule for Affective Disorders and Schizophrenia for School Age Children. Present and Lifetime Version (K-SADS-PL): Initial reliability and validity data. J Am Acad Child Adolesc Psychiatry 1997;36:980 – 8. 27. Vila G, Nollet-Clemencon C, De Blic J, Falissard B, Mouren-Simeoni MC, Scheinmann P. Assessment of anxiety disorders in asthmatic children. Psychosomatics 1999;40:1–10. 28. Mack C, Moor L. Versions franc¸ aises d’e´ chelles d’e´ valuation de la de´ pression. Les e´ chelles CDI et ISC de Maria Kovacs. Neuropsychiatrie de l’Enfance 1982;30:627–52. 29. Fombonne E, Chehdan F, Carradec AM, Achard S, Navarro N, Reis S. Le Child Behavior Checklist: un instrument pour la recherche en psychiatrie de l’enfant. Psychiatrie Psychobiologie 1988;3:409 –18. 30. Fombonne E. The Chartres study: Prevalence of psychiatric disorders among French school-aged children. Br J Psychiatry 1994;164:69 –79. 31. Bolognini M, Bettschart W, Zehnder-Gubler M, Rossier L. The validity of the French version of the GHQ-28 and Psydis in a community sample of 20 year olds in Switzerland. European Archives of Psychiatry and Neurological Science 1989;238:161– 8. 32. Chambers WJ, Puig-Antich J, Hirsch M, Paez P, Ambrosini PJ, Tabrizi MA, Davies M. The assessment of affective disorders in children and adolescents by semi-structured interview: test-retest reliability of the Schedule for Af393

G. VILA et al. fective Disorders and Schizophrenia for School Age Children. Present Episode Version. Arch Gen Psychiatry 1985;42:696–702. 33. Pine DS, Cohen P, Brook J, Coplan JD. Psychiatric symptoms in adolescence as predictors of obesity in early adulthood: a longitudinal study. Am J Public Health 1997;87:1303–10. 34. Theisen FM, Linden A, Geller F, Schafer H, Martin M, Remschmidt P, Hebebrand J. Prevalence of obesity in adolescent and young adult patients ith and without schizophrenia and in relationship to antipsychotic medication. J Psychiatr Res 2001;35:339 – 45.


35. Seligman ME. Helplessness: on depression, development and death. San Francisco: W. H. Freeman; 1975. 36. Goodman E, Whitaker RC. A prospective study of the role of depression in the development and persistence of adolescent obesity. Pediatrics 2002;110:497–504. 37. Johnson JG, Cohen P, Kasen S, Brook JS. Childhood adversities associated with eating disorders and weight problems during adolescence and early adulthood. Am J Psychiatry 2002;159:394 – 400.

Psychosomatic Medicine 66:387–394 (2004)

mental disorders  
mental disorders  

FromthePediatricDepartment,Necker-EnfantsMaladesHospital,Paris, France. Address correspondence and reprint requests to G. Vila, PhD, CHU Nec...