Comorbidity Between Obesity and AttentionDeficit/Hyperactivity Disorder: Population

Study with 13–15-Year-Olds

Luis Rojo, MD1,2*Elıas Ruiz, Pharm D3

Jose Alberto Domınguez, MD2

Myralys Calaf, MS4

Lorenzo Livianos, MD1,2

ABSTRACT

Objective: This study analyzes whether

obese children have a higher risk of

attention deficit/hyperactivity ‘‘character-

istics’’ (AD/HD) than do children with

other nutritional states.

Method: This study included 35,403

participants from 486 community schools.

They completed the AD/HD scale of the

Strengths and Difficulties Question-

naire (SDQ) and were weighed and

measured. 2879 of the participants

were obese and 78 were morbidly

obese (BMI >40).

Results: A discrete, nonsignificant,

increment was found in the AD/HD char-

acteristics of male participants with mor-

bid obesity, as compared with the other

nutritional states. Among morbidly obese

females, the prevalence of AD/HD charac-

teristics was slightly superior, although

not significantly, to that found in partici-

pants with normal weight, overweight or

obese (BMI <40).

Conclusion: Among nonclinical popu-

lations with a communitarian origin, pre-

vious findings reporting high rates of AD/

HD in obese children are not replicated.

This increment in the prevalence of AD/

HD among hospitalized obese children

could be the result of selection bias.VVC 2006 by Wiley Periodicals, Inc.

Keywords: obesity; morbid obesity;

attention deficit/hyperactivity disorder;

AD/HD; adolescents; SDQ; community

sample; comorbidity

(Int J Eat Disord 2006; 39:519–522)

Introduction

In a recent article, Agranat-Meged et al.1 depictedthe comorbidity, not previously described, betweenchildhood obesity and attention deficit/hyperactiv-ity disorder (AD/HD) in children hospitalized forsevere obesity. AD/HD is a prevalent disorder inchildhood, with an estimated rate of 4% in Spain.2,3

In obese populations, the AD/HD diagnosis tendsto go unnoticed. In fact, the above-mentionedinvestigators observed that the AD/HD diagnosiswas missed in 60% of their obese patients.

The importance of this finding is manifold. Forstarters, the diagnosis, treatment and, therefore,

the prognosis of patients with both diagnoses couldbe improved. In addition, it is possible that a frac-tion of the conduct problems and poor perform-ance of this population of obese children is attrib-utable to their comorbid pathology, rather than tocharacter difficulties. Their adequate detection andtreatment could then contribute to their improve-ment, and, consequently, to a reduction in the fre-quent social stigmatization to which they are sub-ject, and to a significant improvement in their qual-ity of life.

Agranat-Meged et al.1 assessed a sample of 26obese children (Body Mass Index (BMI) percentile>85), aged 6 through 18, hospitalized for obesitytreatment between 1997 and 2002. Of that sample,57.7% (n ¼ 15) received an AD/HD diagnosis, anumber that is highly above the 10% reported inthe general population of that same age. Their sam-ple consisted of a very selective subgroup of obesepatients, which brings us to the question of whathappens in ambulatory and nonclinical samples.

The purpose of the present study is precisely toapproximate an answer to the second part of thisquestion: to analyze the prevalence of AD/HD in acommunity sample of 13–15-year-olds, as a functionof their nutritional state (BMI). Our population con-sisted of more than 35,000 students, which consti-

1 Department of Medicine, University of Valencia, Spain2 Eating Disorders Unit, Psychiatric Service, H.U. La Fe, Valencia.

Spain3 Direccion General de Salud Publica, Consellerıa de Sanidad.

Valencia, Spain4 Departamento de Psicologıa Clınica, Universidad Carlos Albizu,

San Juan, Puerto Rico

Accepted 18 December 2005

*Correspondence to: Luis Rojo, MD, UTCA, Planta 11 Pabellon

Maternal, Hospital Universitario La Fe, Avda de Campanar 21,

46009 Valencia, Spain. E-mail: luis.rojo@uv.es

Published online 11 April 2006 in Wiley InterScience

(www.interscience.wiley.com). DOI: 10.1002/eat.20284

VVC 2006 Wiley Periodicals, Inc.

International Journal of Eating Disorders 39:6 519–522 2006—DOI 10.1002/eat 519

BRIEF REPORT

tutes more than one half of the student populationof this age range in Valencia, Spain, making this aclearly representative sample of our juvenile popula-tion. Our goal was to verify whether the completeobese population is specially vulnerable to thiscomorbidity (i.e., whether this population has a sig-nificantly higher risk of having AD/HD); whetherthe AD/HD comorbidity is specially concentratedwithin morbid obesities, which is the populationspecifically assessed by the above-cited study1 or,alternatively, whether the findings reported by theseinvestigators can be attributed to a sample selectionbias.

Method

During the 2003–2004 academic course, the sanitary and

educational administrations of Valencia (Spain) (Consel-

lerıas de Sanidad y Educacion, Generalitat Valenciana)

offered the opportunity to participate in a study of risk

factors, early detection, and prevention of eating disor-

ders (ED) (DITCA-CV Program) to all of the schools in

Valencia (Spain) with mandatory secondary education. A

total of 486 schools and 35,403 students (aged 13–15)

participated in this study. Of these, 49.1% were females

and 50.1% were males. The schools obtained informed

consent from the parents.

Assessment packages for each student were sent to

every participating school. Among others, the assessment

measures included the Strengths and Difficulties Ques-

tionnaire (SDQ). The questionnaires were handed out by

teachers during school hours, and pupils were weighed

and measured at school, using reliable instruments that

were specially provided. The data obtained were statisti-

cally analyzed using the SPSS computer program. As we

are dealing with rates and proportions, we will use the v2

for the testing of statistical significance and odds-ratio as

a mean of evaluation of the risk.

Underweight was defined as a body mass index (BMI)

percentile for age <10; overweight was defined as a BMI

percentile for age 90–97 and obesity was defined as a

BMI percentile for age >97.4

The SDQ is a very brief self-administration question-

naire that can be used for the screening of psychopathol-

ogy in 11–16-year-old subjects.5 Goodman et al.6 have

demonstrated that, when applied to an ample commun-

ity sample, its specificity for the detection of psychiatric

disorders is equal to 95%, and its sensitivity is equal to

63% when various informants are used. Used in this

manner, it is a fine instrument for the detection of con-

duct disorders, hyperactivity, depression, and some anxi-

ety disorders. Specifically, its sensitivity for the detection

of AD/HD is equal to 75.4%. Its authors propose the fol-

lowing as application areas: initial clinical evaluation, fol-

low-up studies, epidemiology studies, screening of men-

tal disorders, and related research.6 Its sensitivity, how-

ever, is significantly reduced when only self-administered

data are gathered. For AD/HD (according to DSM-IV crite-

ria), this sensitivity is reduced to 12.8%. In other words, of

100 subjects who actually suffer from this disorder, only

13 score above the cutoff point and are therefore picked

out.6 This topic is a serious limitation of our study and we

will address it in the discussion.

The SDQ is composed of 25 questions, phrased both

positively and negatively, that are grouped into 5 sub-

scales, one of which is the Hyperactivity-Inattention

Scale. The five items comprising the SDQ’s Hyperactivity-

Inattention Scale were deliberately selected to tap inat-

tention (items 15 and 25), hyperactivity (items 2 and 10),

and impulsiveness (item 21), because these are the three

key symptom domains for a DSM-IV diagnosis of AD/

HD.7

There is a validated Spanish version of this instru-

ment.8 Scores in the Hyperactivity-Inattention Scale

range from 0 to 10. Subjects can be classified into one of

two categories according to their scores: those with nor-

mal or borderline scores (�7), and those with abnormal

scores (>7). Nevertheless, given that full diagnostics

assessments were not completed, we will not consider

abnormal scores as an indication of an AD/HD diagnosis.

Therefore, participants with abnormal scores in the AD/

HD scale of the SDQ will be referred to as participants

with AD/HD ‘‘characteristics’’.

Results

In this study, 21.4% of the students, 23.3% of themales and 19.4% of the females, obtained scoresabove the cutoff point in the Hyperactivity-Inatten-tion Scale. We analyzed the distribution of partici-pants who obtained abnormal scores in this scale,according to two nutritional state categories.

Distribution According to Nutritional Status A

Obesity was defined as a BMI percentile for age>97. Of the total sample, 8.4% of the students (n ¼2947; 1312 females and 1567 males) were obese.

Table 1 shows the distribution of hyperactivity asa function of the nutritional state. Of the obese stu-dents, 18.2% obtained elevated scores in the Hyper-activity-Inattention Scale. This proportion is infe-rior that that found in the other nutritional states(v2 ¼ 19.1, df ¼ 1, p < 0.001; odds ratio [OR] ¼ 1.24,95% confidence interval [CI] ¼ 1.13–1.39). Thisholds true when analyzed by gender, as 16.6% ofthe females and 19.6% of the males scored above

ROJO ET AL.

520 International Journal of Eating Disorders 39:6 519–522 2006—DOI 10.1002/eat

the cutoff point in this scale, percentages in bothcases significantly inferior to those found in theother nutritional states (males v2 ¼ 13.8, df ¼ 1, p <0.001; OR ¼ 1.22, 95% CI ¼ -1.31; females v2 ¼ 7.2,df ¼ 1, p < 0.01; OR ¼ 1.05, 95% CI ¼ 1.05–1.3).However, as we will explain next, this is not thecase when morbid obesity is analyzed separately.

Distribution According to Nutritional Status B

The ‘‘obese’’ category is subdivided into twogroups: obesity (with BMI <40) and morbid obesity(BMI >40).9 A total of 2869 students (8.1% of thesample) were obese and 78 (0.22%) suffered frommorbid obesity (Table 1). Among the obese group,18.1% obtained abnormal scores in the Hyperactiv-ity-Inattention Scale. Analyzed by gender, this cor-responds to 16.5% of the females and 19.3% of themales (Table 1). Among the morbid obesity group,25.6% obtained abnormal scores in the Hyperactiv-ity-Inattention Scale. There were no significant dif-ferences in the prevalence of AD/HD characteris-tics between the two groups, nor when comparedwith the prevalence found in every other nutri-tional state. Analyzed by gender, this correspondsto 22.9% of the females and 27.9% of the males. It isamong males of this morbid obesity category that agreater proportion of participants with AD/HD‘‘characteristics’’ can be found, even though signifi-cant differences are not found with respect to theother nutritional states. The AD/HD characteristicsprevalence is superior among males in every nutri-tional state, although this predominance is onlysignificant among normal weight, overweight andnon-morbid obese participants. The male/femaleOR of having AD/HD characteristics is 1.09 in theunderweight category (95% CI ¼ 0.8–1.5); 1.29 inthe normal weight category (95% CI ¼ 1.22–1.36),1.22 in the overweight category (95% CI ¼ 1.0–1.47), 1.17 in the obese category with BMI <40

(95% CI ¼ 1.0–1.47) and 1.3 in morbid obesity cate-gory 1.3 (95%IC: 0.46–3.67).

Conclusion

Our results suggest that there is a slight incrementin the comorbidity of AD/HD characteristics in thesubgroup of obese students with BMI >40, just as itwas first described by Agranat-Meged et al.1 Thereare, however, significant differences between bothstudies. Although they conducted an exhaustiveclinical assessment of their participants, the Israeliinvestigators used a very selective population withhospital origin and did not include a control group.In contrast, our population is highly representativeof the 13–15-year-old adolescent population in ourcommunity. One of the strengths of our study is thesize of the sample that was originally assessed,which allowed us to evaluate the prevalence of AD/HD characteristics in 2869 obese and 78 morbidlyobese students, and to compare this prevalencebetween these and the other nutritional states.Given the low prevalence of morbid obesity, amplesamples are hard to collect, and this is even moreproblematic when one is dealing with a nonclinicalsample that is representative of the general popula-tion. The AD/HD characteristic comorbidity in themorbid obesity category is superior, although notsignificantly, to that detected in the obese categorywith BMI <40, as well as to that detected in the restof the nutritional states. In every nutritional state, apredominance of AD/HD characteristics amongmales was found, that does not reach statistic sig-nificance in the underweight and morbid obesitycategories, probably due to the small number ofparticipants in these categories. For the total sam-ple, the AD/HD characteristics male/female ratio

TABLE 1. Distribution of hyperactivity characteristics rates (SDQ) according to gender and nutritional state

UnderweightNormalWeight Overweight

ObesityBMI < 40

Morbid ObesityBMI > 40 Total

FemaleAD/HD SDQ normal or borderline Count 470 11,459 975 1,096 27 14,027

% 76.4 80.4 82.1 83.5 77.1 80.6AD/HD SDQ abnormal Count 145 2,794 213 216 8 3,376

% 23.6 19.6 17.9 16.5 22.9 19.4Total Count 615 14,253 1,188 1,312 35 17,403

MaleAD/HD SDQ normal or borderline Count 243 11,004 1,263 1,264 31 13,805

% 74.8 76.1 79.0 80.7 72.1 76.7AD/HD SDQ abnormal Count 82 3,462 336 303 12 4,195

% 25.2 23.9 21.0 19.3 27.9 23.3Total Count 325 14,466 1,599 1,567 43 18,000

Note: SDQ ¼ Strengths and Difficulties Questionnaire; AD/HD ¼ attention deficit/hyperactivity deficit; BMI, body mass index.

OBESITY AND AD/HD

International Journal of Eating Disorders 39:6 519–522 2006—DOI 10.1002/eat 521

was equal to 1.24 (OR ¼ 1.26; 95% CI ¼ 1.2–1.33),close to the inferior limit of the ratio found in com-munity-based prevalence studies, of 1.5–3,10 andalso inferior to the 2:1 ratio detected by Agranat-Meged et al.1

Evidently, our prevalence rate of AD/HD charac-teristics, 21.4% of the total sample, is way beyondthe real prevalence rate (4%, as was previouslystated). As noted, the sensitivity for predicting adiagnosis of AD/HD in children aged 11 to 15 wasreported to be 12.8% if based only on student self-report, This finding would seem to suggest a rela-tively high rate of missed cases. This discrepancy isexpected when screening instruments are used inphase 1 of epidemiological studies.11 Indeed, themost important shortcoming of the present studyrelies in the fact that the AD/HD assessment is con-ducted exclusively by the application of a self-administered questionnaire. A clinical evaluation ofthe probable cases has not been conducted. There-fore, we cannot talk about AD/HD ‘‘cases’’ per se,but rather, as we have done, about populationswith AD/HD characteristics.

There are, however, two reasons why we considerour results deserve consideration. First, although aclinical evaluation would have obviously been pref-erable, at least in a fraction of the sample, webelieve that the methodological weakness to whichwe have referred should be conceptualized withinthe framework of the objectives of this study. Thepurpose of the present study was not to establishthe AD/HD prevalence rate in our population, forwhich our method would not be justified, butrather to investigate whether there is a significantincrement of cases in a subgroup of communitar-ian origin with a determined nutritional state. Sec-ond, we must affirm that the instrument in ques-tion is a validated questionnaire with which, afterbeing applied to various informants, a sensitivity of75.4% in the detection of AD/HD has been estab-lished in community samples. In their Spanishpopulation study with children younger than 16years of age, Garcıa et al.8 also demonstrated that,in externalization disorders like AD/HD, the sensi-tivity is better, with false-negative results muchmore infrequent than false-positives. Goodman7

also agrees in that this is a high-quality instrumentfor the detection of AD/HD.

Our data do not support the highly elevated pro-portion of AD/HD in persons with morbid obesity

as reported by the Israeli group. In their sample of26 hospitalized children, 57.7% met clinical criteriafor AD/HD, which is way above the 25.6% of mor-bid obesity cases with AD/HD characteristics de-tected by the Hyperactivity-Inattention Scale ofthe SDQ. We believe that the most plausibleexplanation for this discrepancy is one of sampleselection bias, to which Agranat-Meged et al.1 alsoreferred, in which obese children with comorbidAD/HD may tend to be hospitalized more fre-quently than obese children without this comor-bidity.

References

1. Agranat-Meged AN, Deitcher C, Goldzweig G, Leibenson L, Stein

M, Galili-Weisstub E. Childhood obesity and attention deficit/

hyperactivity disorder: a newly described comorbidity in obese

hospitalized children. Int J Eat Disord 2005;37:357–359.

2. Andres A, Catala MA, Gomez-Beneyto M. Estudio de la prevalen-

cia del trastorno por deficit de atencion con hiperactividad en

ninos de 10 anos residentes en el municipio de Valencia. Actas

Luso-Esp Neurol Psiquiatr Cienc Afin 1995;23:184–188.

3. Benjumea P, Mojarro MA. Trastornos hipercineticos: estudio

epidemiologico en doble fase de una poblacion sevillana. Ann

Psiquiatr 1993;9:306–311.

4. Sobradillo B, Aguirre A, Aresti U, Bilbao A, Fernandez-Ramos C,

Lizarraga A, et al. Curvas y tablas de crecimiento (Estudios Lon-

gitudinal y Transversal). Instituto de investigacion sobre creci-

miento y desarrollo. Bilbao: Ed. Fundacion Faustino Orbegozo

Eizaguirre; 2004.

5. Goodman R, Meltzer H, Bailey V. The Strengths and Difficulties

Questionnaire: a pilot study on the validity of the self-report

version. Eur Child Adolesc Psychiatry 1998;7:125–130.

6. Goodman R, Ford T, Simmons H, Gatward R, Meltzer H. Using

the Strengths and Difficulties Questionnaire (SDQ) to screen for

child psychiatric disorders in a community sample. Br J Psy-

chiatry 2000;177:534–539.

7. Goodman R. Psychometric properties of the Strengths and Diffi-

culties Questionnaire (SDQ). J Am Acad Child Adolesc Psychiatry

2001;40:1337–1345.

8. Garcıa P, Goodman R. Mazaria J, Torres A, Rodriguez-Sacristan

J, Hervas A., et al. El cuestionario de capacidades y dificultades.

Rev Psiquiatrıa Infanto-Juvenil 2000;1:12–17.

9. Sociedad Espanola para el Estudio de la Obesidad. SEEDO’2000

consensus for the evaluation of overweight and obesity and the

establishment of criteria for therapeutic intervention. Med Clin

(Barc) 2000;115:587–597.

10. Scahill L, Schwab-Stone M. Epidemiology of attention-deficit/

hyperactivity disorder in scholl aged children. Child Adolesc

Psychiatr Clin North Am 2000;9:541–556.

11. Rojo L, Livianos L, Conesa L, Garcıa A, Domınguez A, Rodrigo G,

Sanjuan L, Vila ML. Epidemiology and risk factors of eating dis-

orders. A two stage epidemiological study in a Spanish popula-

tion aged 12–18 years. Int J Eat Disord 2003;34:281–291.

ROJO ET AL.

522 International Journal of Eating Disorders 39:6 519–522 2006—DOI 10.1002/eat