Journal of Psychosomatic Research 76 (2014) 105–112

Contents lists available at ScienceDirect

Journal of Psychosomatic Research

Anxiety and depression predicted by medically unexplained symptomsin Pakistani children: A case-control study☆

Nazish Imran a,⁎, Cornelius Ani b,c, Zahid Mahmood d, Khawaja Amjad Hassan e, Muhammad Riaz Bhatti f

a Child & Family Psychiatry Department, King Edward Medical University/Mayo Hospital, Lahore, Pakistanb Academic Unit of Child and Adolescent Psychiatry, Imperial College London, St Mary's Hospital, Paddington, London, UKc Consultant Child and Adolescent Psychiatrist, Bracknell CAMHS, Berkshire Healthcare NHS Foundation Trust, Berkshire, UKd Department of Clinical Psychology, University of Management and Technology, Lahore, Pakistane Department of Paediatrics, Paediatrics Unit II, King Edward Medical University/Mayo Hospital, Lahore, Pakistanf Department of Psychiatry& Behavioural Sciences, King Edward Medical University/Mayo Hospital, Lahore, Pakistan

☆ Name of department where the work was conductDepartment, King Edward Medical University/Mayo Hosp⁎ Corresponding author at: 6-C, Phase 1, Defence Hous

Tel.: +92 321 4201955.E-mail addresses: nazishimrandr@gmail.com (N. Imran

zmahm54084@aol.com (Z. Mahmood), dramjadkh@gmaiprofmrbhatti@hotmail.com (M.R. Bhatti).

0022-3999/$ – see front matter © 2013 Elsevier Inc. All rihttp://dx.doi.org/10.1016/j.jpsychores.2013.11.016

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 15 April 2013Received in revised form 26 November 2013Accepted 29 November 2013

Keywords:AnxietyChildrenDepressionFunctional disabilityPsychopathologySomatisation

Objective: To explore association between medically unexplained symptoms in children in Pakistan withemotional difficulties and functional impairments.Methods: We conducted a matched three-group case-control study of 186 children aged 8–16 years in Lahore,Pakistan. Cases were 62 children with chronic somatic symptoms for which no organic cause was identifiedafter investigations. Two control groups of 62 childrenwith chronicmedical paediatric conditions, and 62healthychildren were identified. Cases and controls were matched for gender, age, and school class. Somatisation wasmeasured with the Children's Somatisation Inventory (CSI-24) while anxiety and depression were measuredwith the Spencer Children's Anxiety Scale and the Short Mood and Feelings Questionnaire respectively. Allquestionnaires were translated into Urdu.Results:Mean agewas 11.7 years (SD = 2.1). Cases scored significantly higher on somatisation (CSI-24), anxietyand depression than both control groups. Paediatric controls scored significantly higher than healthy controls on

all three measures. Two hierarchical linear regression models were used to explore if somatisation predicteddepression and anxiety while controlling for several confounders. Somatisation (higher CSI-24 scores) inde-pendently and significantly predicted higher anxiety (β = .37, p = .0001) and depression (β = .41,p = .0001) scores.Conclusion: This is the first study to show an association between medically unexplained symptoms and anxietyand depression in Pakistani children. This highlights the importance of screening for emotional difficulties inchildren presenting with unexplained somatic symptoms in this region.

© 2013 Elsevier Inc. All rights reserved.

Introduction

Medically unexplained symptoms (MUS) are physical symptomsthat cannot be explained by a conventionally defined medical diseaseafter appropriatemedical assessment [1]. These symptoms are commonin paediatric populations [2,3]. They vary from minor pain and acheswhich improve without treatment to severe multiple somatic symp-toms which leads to significant functional impairment and disability.

Multiple risk factors including individual factors such as increasedfocus on physical sensations, personal experiences of physical illness,and family factors such as physical and mental health problems,

ed: Child & Family Psychiatryital, Lahore, Pakistan.ing Authority, Lahore, Pakistan.

), c.ani@imperial.ac.uk (C. Ani),l.com (K.A. Hassan),

ghts reserved.

parental somatisation, emotional over involvement are associatedwith MUS. Environmental circumstances including life events, relation-ship difficulties, school stressors such as bullying, and academic pres-sure have also been implicated in somatisation [4].

MUS in children are also linked with a variety of adverse psychoso-cial outcomes including psychiatric difficulties, and poor school atten-dance [2,3]. Several studies in developed countries have found up tohalf of children with somatisation to have associated psychiatric disor-ders, especially anxiety and depression [5–9]. Other studies haveshown longitudinal association between somatisation and later emo-tional difficulties. For example, children with recurrent abdominalpain at age 6 years had higher rates of anxiety disorders at age 7 yearsin the Avon Longitudinal study of Parents and children [10]. A greaterlikelihood of developing psychiatric disorders in adulthood has alsobeen reported in children with somatisation [11]. In addition to mentalhealth and functional impairment for affected children, parents andsiblings can also become impaired due to the increased burden of carefor the index child [12].

106 N. Imran et al. / Journal of Psychosomatic Research 76 (2014) 105–112

In addition to adverse mental health outcomes, associationsbetween childhood somatisation and poor functional outcomes arealso well established in developed countries. One of the most commonfunctional impairments is school absence [13]. Weitzman andcolleagues found that adolescents with somatisation experiencedmore school absence than those with chronic medical conditions [14].Similarly, Smith and colleagues found that adolescents with chronicfatigue missed significantly more days of school than children withmigraine or healthy controls [15].

Although several studies have now explored somatisation in chil-dren, most were from developed countries. Incidentally, the findingsmay not be generalizable in the different socio-cultural circumstancesof developing countries; hence the importance of cross cultural perspec-tives. Evidence from low andmiddle income countries (LMICs) indicatethat adults are more likely to present with somatic symptoms whenseeking help even when they are aware of having psychological symp-toms [16–19]. A strong association between somatoform disordersand anxiety and depression in women (with odds ratio ranging from2.5 to 3.5) was observed in a systematic review covering studies fromeight LMICs [20]. In relation to children, Weiss and colleagues demon-strated in a study of American and Thai children that among clinic sam-ples, Thai children reported higher somatic versus depressivesymptoms compared with their American peers [21].

Differences in cultural styles of expressing distress, influenced bycultural beliefs and practices, pathways to care and differences inhealthcare systemsmay explain differences in somatisation in differentsocieties. Also stigma associatedwith emotional problems and psychiat-ric treatment compared with expression of physical problems as well asconsideration of depression and anxiety as moral weakness may alsoincrease tendency towards somatisation in LMICs [22–24]. Whilesomatisation is prevalent among adults in Pakistani society, the preva-lence in children may be higher because they are less verbally expres-sive and may tend to selectively manifest physical symptoms asidioms of psychological distress. There may also be biological reasonsto predict higher prevalence of somatisation in Pakistani children.Increasing evidence suggest that activation of the immune system caninduce somatisation [25]. For example, Lekander et al (2004) haveshown a correlation between blood cytokines and self rated physicalsymptoms [26]. These findings may be particularly relevant to childrenin developing countries like Pakistan who have higher exposure toinfections and pathogens producing immunological challenges.

Our clinical experiences suggest that somatisation in children iscommon in Pakistan. However, any associations with mental disordersand functional impairments have not been explored in this setting.We present data on childhood somatisation and child psychopathologyin Pakistan. We hypothesized that children with medically unexplainedsomatic symptoms in this settingwill havemore anxiety and depressivesymptoms and more school absence than children with diagnosedphysical illnesses and healthy controls. To our knowledge, this is thefirst study in South Asia to use a robust three-group case–control designto directly explore the emotional and functional impact of somatisationon children controlling for a wide range of psychosocial confounders.

Methods

The study was conducted in a child psychiatric clinic of a tertiarycare hospital in Lahore, Pakistan. Lahore is the capital of Punjab, whichis the largest Province in Pakistan. Approval was obtained from theEthical Review Board of King Edward Medical University (KEMU).Informed consent was obtained from the parent/legal guardian of theeligible children prior to administration of the questionnaires. Childrenbetween the ages of 8 and 16 were eligible to participate. A matchedcase control study design was used. Cases were consecutive childrenpresenting to the outpatient clinic of the Department of Child andFamily Psychiatry in Mayo Hospital, Lahore with chronic somatic com-plaints for which no organic cause has been found following physical

examination and appropriate physical investigations. Two controlgroups were identified. The first controls were children attendingpaediatric out-patient clinic in the same hospital due to an establisheddiagnosis of a physical disorder such as asthma, diabetes and nephroticsyndrome. Childrenwhomet the study criteriawere identified in liaisonwith the paediatricians and then approached by a research teammem-ber to obtain their parent's consent for participation. The second controlgroup were healthy children recruited from a local school who had nocurrent diagnosis of a medical or psychiatric disorder. The cases andboth control groups were matched for age, gender, and school class oryear.

A sample of 186 (62 in each group) was calculated as adequate toidentify 0.5 (half) standard deviation difference in depressive symp-toms (Short Mood and Feelings Questionnaire FQ score) with 80%power and 5% level of significance between childrenwithmedically un-explained somatic symptoms and each of the two control groups [27].

Following written informed consent, the questionnaires were ad-ministered to the children and their parents by trained interviewers.In total, 229 parents were approached to get the required sample of186 children (70 in psychiatric clinics where cases were recruited, 83in paediatric clinic for first control group and 76 from the school thatprovided healthy controls). The questionnaires were read out to theparents and children and the children provided a response. Some ofthe information such as socio demographic data was obtained fromparents. Good inter-rater reliability across interviewers was establishedbefore actual data collection. All the scales used for childrenwere trans-lated to Urdu by multiple forward translation method and pilotedamong children in Pakistan to ensure reliability in the local settingprior to the study. This practice is established in previous studiesin Pakistan such as the Urdu translation & cultural adaptation ofKIDI-SADS [28]. The interviewers were blind to the study hypothesisto avoid information bias.

Measurements

Socio-demographic details

Standard socio-demographic information was obtained as isrelevant to Pakistan. This includes family characteristics such as livingarrangements, number of siblings, rural or urban dwelling, school enrol-ment (which is not mandatory in the country) as well as school atten-dance, life events/stressors in Pakistani context including maritalengagement against wishes, and parent's level of education. In orderto obtain a reliable index of socio-economic status, information wassought on household incomes, and ownership of high value possessionssuch as house, car, and coloured TV. A socioeconomic scale was createdby combining income categories and the ownership of possessions(withmore expensive items such as car and house, weighted by a factorof two and three respectively).

Somatisation

This was assessed with the revised 24-item Children's SomatisationInventory (CSI-24) [29]. The CSI-24 is a reliable and psychometricallyrefined version of the original CSI [30]. Items are scored 0–4 and totalscores range from 0 to 96. The CSI-24 showed excellent internal consis-tency in our sample (Cronbach alpha 0.91).

Depressive symptoms

Depressive symptomswere assessed with the Short Mood and Feel-ingsQuestionnaire (SMFQ) [31]. This is a brief 13-itemmeasure of child-hood depressive symptoms scored on a 3-point Likert scale. The SMFQhas been found to correlate well with the Children's Depression Inven-tory (r = .67) and to discriminate well between clinical and nonclinicalsamples [31,32]. It has been used successfully in other developing

107N. Imran et al. / Journal of Psychosomatic Research 76 (2014) 105–112

countries [33,34]. The SMFQ also demonstrated very good internal con-sistency in our sample (Cronbach alpha 0.87).

Anxiety symptoms

Anxiety was assessed with the Spencer Children's Anxiety Scale(SCAS) [35]. The SCAS is well validated and assesses anxiety symptomsbroadly in line with the dimensions of anxiety disorders in DSM-IV. Thescale assesses six domains of anxiety separation anxiety, social phobia,obsessive compulsive, panic/agoraphobia, physical injury fears, andgeneralised anxiety. Both the total score and the subscales wereanalysed in this study. The SCAS total score showed very good internalconsistency in our sample (Cronbach alpha 0.87). The internal consis-tency of the subscales ranged from 0.55 (for panic and agoraphobia)to 0.75 (for generalised anxiety).

Stressful life events

The childrenwere asked if they had ever experienced one ormore ofthe following ten stressful life events in the past year (e.g. parental sep-aration, death of a relative or friend, bullying in school, hospital admis-sion for the child, hospital admission for siblings and or parents, examfailure, change of school, forced marriage, physical abuse, and breakupwith a friend). The total number of stressors experienced was summedinto a scale. These items were selected asmost relevant based on a pilotwith a larger number of stressors.

Maternal anxiety and depression

Thiswas assessedwith the Aga KhanUniversityHospital anxiety anddepression rating scale (AKUADS). This is a 25 item screening instru-ment, developed indigenously in the primary health care and psychiat-ric setting of Pakistan, for screening depression and anxiety disorder[36,37]. It incorporates culturally pertinent somatic metaphors of de-pressive disorder. The questionnaire has 25 items, 13 psychologicaland 12 somatic. At cut-off of 20, it has 66% sensitivity, 79% specificity,a positive predictive value of 83% and a negative predictive value of 60%.

Table 1Socio-demographic and family characteristics

Cases N = 62 N (%)

Age (mean, SD) 11.6 (2.2)Gender (male) 31 (50.0)Residence:Urban 44 (71)Rural 18 (29)

Living with both parentsYes 53 (85.5)No 9 (14.5)

Number of siblings (mean, 95% CI)a 4.9a (4.4,5.4)Socio-economic rating scale (mean, SD)a 8.3a (7.7,8.9)Child attending schoolYes 48 (77.4)No 14 (22.6)

School absence in past year mean ranks median(interquartile range) 7 (0,20)a

N = 5690.17a

Maternal anxiety and depression scores means (95% CI)a 27.3a (24.4,30.2)Maternal educationNo educationUp to 11 years of education12 or more years of education

25 (41.0)27 (44.3)9(14.8)

Paternal educationNo educationUp to 11 years of education12 or more years of education

17 (27.4)38 (61.3)7 (11.3)

a Mean ranks (or means) that do not share same subscript differ significantly p b 0.017 (or T

Analysis

The data was entered and analysed by SPSS version 17. Most contin-uous variables were normally distributed, thus summarized with meansand standard deviations. Non-normally distributed continuous datawere presented as medians and interquartile ranges while categoricaldata was describedwith numbers and percentages. Comparisons of con-tinuous data were conducted with t-tests, Mann–Whitney U, ANOVA,Kruskal–Wallis ANOVA or Pearson & Spearman rho correlations depend-ing on the distribution and number of categories. Specifically, one-wayANOVAs were used to compare children with medically unexplainedsymptoms and the two control groups on continuous normally distribut-ed outcome and predictor variables (e.g. depression and anxiety symp-toms, and somatisation). Post-hoc pairwise comparisons used TukeyHSD. Kruskal–Wallis ANOVAwas used to compare non-normally distrib-uted continuous measures and number of days absent from school andtwo SCAS subscales (separation anxiety and panic/agoraphobia) subse-quent pair-wise analyses applied, Bonferroni correction to constrainthe level of significance to p b .017 in view of multiple comparisons.Linear hierarchical regression techniques were used to assess if med-ically unexplained symptoms independently predicted anxiety anddepressive symptomswhile controlling for demographic and other con-founding variables including study group allocation. As study group istrichotomous, this variable was dummy-coded before inclusion in theregression model. The data was explored further with a priori plannedanalysis to identify predictors of SCAS subscales. Within-group analysisexplored association between somatisation, anxiety and depressionseparately for cases, paediatric controls, and student controls.

Results

Socio-demographic and family characteristics (Table 1)

A total of 186 children participated in the study (62 in each of the three groups).Table 1 shows some characteristics of the group. The average age of the cohort was11.7 years (SD = 2.1) and did not differ across the three groups. The gender ratio wasequal (50% each). These figures confirm the accurate matching of the three groups forage and gender. Age was not significantly associated with any of the outcome variables.Female respondents had significantly more siblings (t = −2.65 df 184, p = .009) andscored significantly higher on the physical injury subscale of the SCAS than males(t = −3.66, df 184, p b .0001). No other variable differed significantly across gender.

Paediatric controlsN = 62 N (%)

School controlsN = 62 N (%)

Tests

11.6 (2.1) 11.7 (2.3) na31 (50.0) 31 (50.0) na

49 (80.3) 57 (91.9) Χ = 8.9, df = 2 P = 0.0112 (19.7) 5 (8.1)

56 (90.3) 62 (100.0) X = 9.1, df = 2, P = 0.016 (9.7)4.3ab (3.9,4.7) 3.7b (3.4,4.1) F(2,183) = 8.5 P = 0.00017.3a (6.6,8.0) 11.5b (11.0,11.9) F(2,183) = 55.6 P = 0.0001

Χ = 17.7. df = 2 P = 0.000157 (91.9) 62 (100)5 (8.1) 0 (0)N = 55 N = 62 Χ = 44.93, df = 2 p = 0.0001117.59b 57.0c18.9b (15.8,21.9), 18.2b (15.5,20.8) F(2,183) = 12.45, p = 0.0001

27 (45.8) 2 (3.2) X = 51.8, df = 4 P = 0.000127 (45.8) 26 (41.9)5(8.5) 34(54.8)

18 (30.5) 2 (3.2) X = 82.7 df = 4 P = 0.000136 (61.0) 13 (21.0)5 (8.5) 47 (75.8)

ukey HSD p b 0.05) (i.e. those that share subscript do not differ significantly).

Table 2Comparison of scores on somatization, depression, anxiety, school days missed, and number of stressors between children with medically unexplained symptoms, paediatric cliniccontrols, and school controls

Cases N = 62 N (%) Paediatric controlsN = 62 N (%)

School controlsN = 62 N (%)

Tests

Somatisation (CSI-24) means (95% CI)a 36.6a (32.5,40.7) 16.3b (13.0,19.6) 14.5b (10.9,18.1) F(2,183) = 44.3, p = 0.0001Short mood and feelings questionnaire. means (95% CI)a 9.1a (7.7,10.6) 4.9b (3.7,7.3) 5.6b (4.4,6.8) F(2,183) = 11.6, p = 0.0001Spencer Children's Anxiety Scale means (95% CI)a 43.2a (38.0,48.4) 39.4b (26.1,34.9) 25.3b (20.8,29.8) F(2,183) = 15.4, p = 0.0001Separation anxiety aMedian Ranks, Median (interquartile range) = 5(7) N = 62 N = 62 N = 62 X = 8.9, df = 2, P = 0.01

110.1a 85.6b 84.8bSocial phobia means (95% CI)a 6.6a (5.6,7.6) 3.6b (2.8,4.4) 4.0b (3.0,5.0) F(2,183) = 11.8, p = 0.0001Obsessive Compulsive aMeans (95% CI) 7.1a (6.0,8.3) 8.1a (7.2,8.9) 3.7b (2.8,4.6) F(2,183) = 22.1, p = 0.0001Panic/agoraphobia mean ranks median (interquartile range) = 4(8)a N = 62 N = 62 N = 62 X = 45.2, df = 2, P = 0.0001

128.2a 88.0b 64.2cPhysical Injury Fears Means (95% CI)a 6.3a (5.2,7.3) 4.5b (3.6,5.4) 4.6b (3.6,5.6) F(2,183) = 4.0, p = 0.02Generalised anxiety means (95% CI)a 7.1a (6.0,8.1) 4.0b (3.1,4.9) 4.6b (3.7,5.5) F(2,183) = 12.0, p = 0.0001Number of stresses means (95% CI)a 2.2a (1.8,2.7) 1.7ab (1.3,2.0) 1.6b (1.1,2.0) F(2,183) = 3.5 P = 0.03

a Mean ranks (or means) that do not share same subscript differ significantly p b 0.017 (or Tukey HSD p b 0.05) (i.e. those that share subscript do not differ significantly).

108 N. Imran et al. / Journal of Psychosomatic Research 76 (2014) 105–112

Significantlymore cases lived in rural areas comparedwith the two control groups. For thecombined cohort, rural dwellers scored higher on somatisation (M = 28.1, SD = 17.6 vsM = 21.3, SD = 17.4, t(183) = 2.1, p = .038), anxiety (M = 40.6, SD = 21.6 vsM = 31.3, SD = 19.2, t(183) = 2.5, p = .012) and depression scores (M = 9.3,SD = 6.7 vs M = 5.9, SD = 4.9, t(183) = 3.4, p .001). Although majority of the childrenlived with both parents (92%), this was significantly more for the student control group(100%). Cases had more siblings than both control groups but this was only statisticallysignificant compared with the school controls. The school controls scored significantlyhigher on the socio-economic scale than both the cases and paediatric controls, but the lat-ter two groups did not differ significantly. The children with unexplained somatic symp-toms were significantly less likely to be currently enrolled in a school compared withboth control groups. Those enrolled also missed significantly more days of school thanboth control groups. Bothmothers and fathers of the school control groupweremore like-ly to have had 12 or more years of education compared with mothers of cases and paedi-atric controls. Mothers of the children with somatic symptoms scored significantly higheron the maternal anxiety and depression scale compared with both control groups.

Children's somatisation, anxiety depression, school absence, and stressors

Table 2 shows that the children with medically unexplained symptoms scored signif-icantly higher on the CSI-24 somatisation scale than the two control groups. The paediatriccontrols also scored significantly higher than the school controls on this measure. Thisobservation provides validity to the selection strategy for the cases and controls. CSI-24scores correlated significantly and positively with children's depression and anxietyscores, number of siblings, number of stressors, and maternal anxiety and depression.CSI-24 correlated negatively and significantly with SES (Table 3).

It is evident from Table 2 that the cases scored significantly higher than both controlsgroups on the SMFQ (depression) and on the SCAS (anxiety). SMFQ correlated significant-ly and positively with anxiety scores, number of siblings, number of days absent fromschool, and number of stressors. Anxiety scores correlated significantly and positivelywith number of stressors, socioeconomic scale, number of siblings and school absence.Exploration of the SCAS subscales—(separation anxiety, social phobia, obsessive compul-sive, panic/agoraphobia, physical injury fears, and generalised anxiety) showed that thecases scored significantly higher than both control groups in all subscales except obsessivecompulsive where cases differed only with student controls. Also all the subscales corre-lated similarly as the total SCAS score (details not shown).

The cases also had more days off school than both control groups and school absencecorrelated significantly and positivelywith somatisation, anxiety, depression, and numberof siblings. Cases and paediatric controls both scored significantly higher than schoolcontrols on number of stressors. The stressors that differed most between cases and

Table 3Correlations between anxiety, depression, somatisation and other predictor variables

Variable 1 2 3 4 5 6 7 8

1. Anxiety – .61⁎⁎ .62⁎⁎ .23⁎⁎ .27⁎⁎ .36⁎⁎ −.15⁎ .34⁎⁎

2. Depression .62⁎⁎ .27⁎⁎ .23⁎⁎ .28⁎⁎ −.01 .50⁎⁎

3. Somatisation .20⁎⁎ .23⁎⁎ .33⁎⁎ −.19⁎ .53⁎⁎

4. Number of siblings .17⁎ −.01 .17⁎ .28⁎⁎

5. Absence from school .06 −.11 .136. Number of stressors −.08 .23⁎⁎

7. Socioeconomic rating −.17⁎

8. Maternal anxiety anddepression

⁎ p b 0.05.⁎⁎ p b 0.01.

both control groups were physical abuse (cases = 18(29%), paediatric controls 6(9.7%),student controls 10(16.1%), X = 18.1, df = 2, p = .018), and change in school(cases = 31 (50%), paediatric controls 19(30.6%)%, student controls 4(6.5%), X = 28.6,df = 2, p = .0001).

Multivariate analyses

In order to test the study hypotheses, the independent variables significantly associat-ed with anxiety or depression in bivariate analyses were entered sequentially into twoseparate hierarchical linear regression models each with three blocks. We enteredsomatisation in the last block so that its unique contribution to the variance in anxietyand depression controlling for other predictors can be clearly identified [38]. Studygroup assignment was controlled for by dummy coding using the student controls as ref-erence group. The dummy codes for cases and paediatric controls were entered alone inthe second block. Extensive regression diagnostics including plots of residuals, Durbin–Watson and Collinearity statistics showed that both models met the underlying assump-tions for linear regression.

For the regression model on anxiety (Table 4), socio-demographic and psychosocialpredictors (e.g. age, gender, socio-economic rating etc) were entered in the first block.The model explained 27.7% of the variance in anxiety and was statistically significant(F(9,172) = 7.3, p = .0001). Entry of dummy coded group assignment (cases and paedi-atric controls) in the second block explained only 3% additional variance but the modelwas statistically significant (F(11,170) = 6.9, p = .0001). Entry of somatisation alone inthe third block further increased the variance explained by 14.8%. Thefinalmodel was sta-tistically significant (F(12,169) = 11.8, p = .0001) and showed that somatisation was asignificant predictor of anxiety. In fact somatisation by far had the highest beta (.54) com-pared with the other significant predictors: gender (beta = .13), and number of stressors(beta = .21). The final model explained 45.6% of the variance in anxiety scores.

The regression analysis for anxiety was repeated for each of the six subscales of theSCAS. All six models were statistically significant. Separation anxiety was predicted bythe number of stressors and somatisation; social phobiawas predicted by gender, numberof stressors, and somatisation; obsessive compulsive subscalewas predicted by number ofstressors, being a paediatric control, and somatisation; panic/agoraphobia was predictedby younger age, number of stressors, and somatisation; physical injury was predicted byfemale gender and somatisation;while generalised anxietywas predicted by somatisationalone. Thus somatisation predicted all six SCAS subscales and was the only predictor forthe generalised anxiety subscale.

Similar regression procedure was adopted for depressive symptoms (Table 5).Socio-demographic and psychosocial predictors were entered in the first block. Themodel was statistically significant (F(9, 172) = 11.2, p = .0001) and explained 36.9% ofthe variance in depression. Groupdummy codes for cases and paediatric controlswere en-tered in the second block, which increased the total variance explained by 7%. This modelwas also statistically significant (F(11,170) = 9.3 p = .0001). Finally, somatisation alonewas entered in the third block, which increased the variance explained by 141%. This finalmodel was statistically significant (F(12, 169) = 15.2, p = .0001), and showed thatsomatisation was a significant predictor of depressive symptoms recording the highestbeta (.53) compared with the other significant predictors: number of siblings(beta = .15), school enrolment (beta = .14), maternal anxiety and depression(beta = .19) and, rural dwelling (beta = − .15). The final model explained 52% of thevariance in depression scores.

Within-group analysis

The contribution of somatisation to anxiety and depressive symp-toms was further explored in each of the study groups separately.Thus three separate hierarchical regression analysis was conductedeach for cases, paediatric controls and student controls.

Table 4Regression coefficients for independent predictors of child reported Anxiety symptoms (SCAS scores)

Variable Model 1 Model 2 Model 3

B SE B β B SE B β B SE B β

Age −0.71 0.61 −.08 −0.58 0.61 −.06 −0.23 0.54 −.03Gendera 5.81 2.68 .15⁎ 5.96 2.64 .15⁎ 5.35 2.35 .13⁎

No. of siblings 1.21 0.88 .10 0.80 0.88 .07 0.87 0.79 .07School enrolmentb 5.17 4.66 .08 2.99 4.65 .05 5.25 4.15 .08No. of stressors 4.14 0.85 .33⁎⁎⁎ 3.79 0.84 .30⁎⁎⁎ 2.68 0.77 .21⁎⁎

SES scale −0.05 0.51 −.01 0.31 0.59 .05 0.31 0.53 .05Maternal anxiety/depression 0.36 0.12 .21⁎⁎ 0.29 0.12 .17⁎ −0.05 0.12 −.03Father's educationc −4.42 3.59 −.09 −4.02 3.54 −.08 −1.91 3.16 −.04Rural vs urband −5.15 3.39 −.10 −4.07 3.36 −.08 −3.07 2.99 −.06Student control vs cases 10.16 3.90 .24⁎ 0.66 3.73 .011Student control vs paediatric controls 3.44 3.98 .08 3.22 3.54 .08Somatisation 0.61 0.09 .54⁎⁎⁎

R2 .28 .31 .46F for change in R2 7.33⁎⁎⁎ 3.87⁎ 45.93⁎⁎⁎

a Coded “male” = 1, “female” = 2.b Coded “Yes” = 1, No = 2.c Coded “No education” = 1, “Any education” = 2.d Coded “rural” = 1, “urban” = 2.⁎ p b .05.⁎⁎ p b .01.⁎⁎⁎ p b 0.001.

109N. Imran et al. / Journal of Psychosomatic Research 76 (2014) 105–112

For cases alone, thefinalmodel for anxietywas statistically significantF(10,51) = 4.2, p = .0001 and explained45% of the variance in this sub-group. Somatisation was the only statistically significant predictor ofanxiety (SCAS total score) in cases (beta = .60). The final model for de-pressive symptoms was also statistically significant for this groupF(9,52) = 9.3, p = .0001 and explained 64.5% of the variance in depres-sive symptoms. Depressive symptoms in cases were predicted by age(beta = − .24), number of siblings (beta = .22), school enrolment(beta = .21), maternal depression and anxiety (beta = .26), ruraldwelling (beta = -.29) and somatisation (beta = .43).

Similar regression analysis for paediatric controls alone showed sta-tistically significant model for anxiety (F(10,47) = 2.3, p = .028) withgender as the only statistically significant predictor (beta = .40). Thismodel explained 32.7% of the variance of anxiety in the paediatric con-trols. Depressive symptoms in this group was predicted by the numberof siblings (beta = 2.57), no paternal education (beta = .26), andsomatisation (beta = .43). The model was statistically significant

Table 5Regression coefficients for independent predictors of child reported Depressive symptoms (SM

Variable Model 1

B SE B β

Age −0.28 0.16 −.11Gendera −0.87 0.68 −.08No. of siblings 0.51 0.22 .15⁎

School enrolmentb 2.15 1.18 .12No. of stressors 0.62 0.22 .18⁎⁎

SES scale 0.25 0.13 .13Maternal anxiety/depression 0.19 0.03 .41⁎⁎⁎

Father's educationc 0.67 0.91 .05Rural vs urband −2.41 0.86 −.18⁎⁎

Student control vs casesStudent control vs paediatric controlsSomatisationR2 .37F for change in R2 11.17⁎⁎⁎

a Coded “male” = 1, “female” = 2.b Coded “Yes” = 1, No = 2.c Coded “No education” = 1, “Any education” = 2.d Coded “rural” = 1, “urban” = 2.⁎ p b .05.⁎⁎ p b .01.⁎⁎⁎ p b 0.001.

(F10,47) = 3.9, p = .001 and explained 45.5% of the variance in de-pressive symptoms.

For student controls, the model for anxiety was statistically signifi-cant (F(9,52) = 9.7, p = 0.0001) and the significant predictors arethe number of stressors (beta = .40) and somatisation (beta = .61).This model explained 62.7% variance in anxiety scores. Depressionwas predicted by somatisation (beta = − .26), and the numberof stressors (beta = .35). The model was statistically significantF(9,52) = 6.2, p = 0001 and explained 51.7% of the variance indepression.

Discussion

This study aimed to explore the association between medically un-explained somatic symptoms and anxiety, depression and functionalimpairment among Pakistani children. The data supported our hypoth-esis that children with medically unexplained somatic symptoms will

FQ)

Model 2 Model 3

B SE B β B SE B β

−0.26 0.16 −.10 −0.16 0.14 −.07−0.83 0.68 −.08 −0.99 0.60 −.09

0.47 0.23 .14⁎ 0.49 0.20 .15⁎

1.88 1.20 .11 2.48 1.06 .14⁎

0.58 0.22 .17⁎⁎ 0.28 0.20 .080.24 0.15 .13 0.24 0.14 .130.18 0.03 .39⁎⁎⁎ 0.09 0.03 .19⁎⁎

0.69 0.91 .05 1.26 0.81 .09−2.31 0.87 −.17⁎⁎ −2.04 0.76 −.15⁎⁎

0.91 1.00 0.8 −1.63 0.95 −.14−.022 1.03 −.02 −0.28 0.91 −.02

0.16 0.02 .53⁎⁎⁎

.38 .520.90 50.21⁎⁎⁎

110 N. Imran et al. / Journal of Psychosomatic Research 76 (2014) 105–112

have more anxiety and depressive symptoms than children with diag-nosed medical problem and healthy children. This was shown in thewhole cohort even controlling for group assignment. The associationwas also demonstrated separately within each study group includinghealthy student controls who were non-help seeking. In within-groupanalysis, somatisation was the only predictor of anxiety among cases.It predicted all six subscales of the SCAS and was the only predictor ofthe generalised anxiety subscale.

Somatisation, anxiety and depression

These results expandand compliment existing literature fromdevel-oped countries documenting increased anxiety and depressive symp-toms in children with somatic complaints particularly recurrentabdominal pain in community based and clinically referred children[5,39–41]. For example, a prospective school based study of abdominalpain and other common somatic complaints in children found highlevels of anxiety and depression in the cohort. Presence of abdominalpain has also been independently associated with school absence [42].

Lack of comparable studies of somatisation in children fromPakistanlimits us to compare our results with adult literature on this topic fromthe country. Three community based epidemiological surveys of adultsin Pakistan have found somatic complaints to be the predominantpresentation of anxiety and depression in 66% of women and 25% ofmen [43–45]. Similar association between somatisation and emotionaldisorders has been reported in more recent studies of adults fromclinical settings in Pakistan [46,47].

The underlying mechanism between somatisation and psychopa-thology is complex and may be bi-directional. Children exposed to psy-chosocial stressors are more likely to develop medical symptoms [48].Equally, Lavigne and Faier-Routman have shown in an elegant meta-analysis that children with chronic medical problems are twice as likelyto developmental disorders as unaffected controls [49]. It is recognizedthat both scenarios involve complex interactions between risk and pro-tective factors; hence only a minority of children with psychosocialstressors develop medical symptoms and vice versa [12]. Thus under-standing the risk factors to be limited and the protective factors to bepromoted is crucial.

Recent reviews have proposed a complex underlying the biochemi-cal and immunological explanation for somatisation in children [25].The proposed mechanism involves sensitization of brain cytokines inearly life by repeated activation fromenvironmental stressors,which in-creases likelihood of the system being further activated by other envi-ronmental triggers in later life [12]. The potential role of immuneactivation is pertinent in the context of a LMIC country like Pakistanwhere poverty, malnutrition, higher prevalence of communicablediseases increase exposure to pathogens which may increase immuneactivation. Further research is required to explore whether curtailingthese adverse socio-environmental factors brings secondary benefitsin terms of lowering the prevalence of somatisation and associatedimpairments in LMICs.

This study supports previous findings associating somatisation withdepression and anxiety, and functional impairment in the formof schoolabsence. These associations are likely to be underlined by several cogni-tive, social and biological processes [25]. Cognitively, the children mayperceive somatic complaints as more tangible and by implicationmore treatable than emotional symptoms [21]. However, subsequentpoor response to treatment may result in frustration, desperation, in-creased worry and dysphoria. Somatic symptoms may limit peer inter-actions and activities to improve sense of well-being and self-efficacy.For the family, supporting a child with somatisation can be economical-ly challenging especially in a country like Pakistan. The reasons includeout-of-pocket payments for health-care, which means a huge financialdemand for poor families. Secondly, the burden of care on parentsmay mean one or both parents relinquishing income generating activi-ties —setting up a potentially vicious spiral into more poverty. Our

finding of a negative correlation between somatisation and socio-economic ratings supports these hypotheses. Parents may become frus-trated by persistence of the child's somatic symptoms despite theirinvestments in the affected child's care. This can lead to scape-goatingof the affected child and may explain our finding of higher experienceof physical abuse among the children with unexplained somaticcomplaints.

The potentially reinforcing association between somatisation anddepression and anxiety could be a particularly challenging “triplewhammy” in that each is individually associatedwith increased person-al distress in their own rights and can independently or synergisticallylead to functional impairment such as school absence. Educationaldisruption can be a long-term impairment by predisposing to fu-ture under-achievement and socio-economic disadvantage. Just likesomatisation, childhood anxiety and depression also cause distress tofamily members. The increased care needs of the index childmay resultin the child's siblings feeling neglected as parents devotemore attentionto the ill child.

Fortunately there is increasing evidence, albeitmainly from adults indeveloped countries that depression and anxiety can be successfullytreated in people with somatisation syndromes [50–53]. Appropriateeducational support is important to limit the educational disadvantagesof school absence in this illness group [54]. It is therefore important toscreen children with somatisation syndromes for anxiety, depressionand functional impairment so that appropriate interventions could beoffered to reduce these adverse effects.

Other associations with childhood anxiety and depression

Although the main focus of this study is to explore associations be-tween unexplained somatic symptoms and anxiety and depression,the study nonetheless identified other significant predictors of anxietyand depression which are worth further consideration. The regressionanalysis found that total number of stressors, higher number of siblings,non-school enrolment increased maternal anxiety and depression andrural dwelling all predicted higher child depression scores. Total num-ber of stressors and female gender predicted more anxiety scores.

These associations are consistentwith previous studies on childhoodanxiety and depression [55]. They go to demonstrate that emotional dif-ficulties in children havemultifaceted associations and that comprehen-sive socio-economic and psychosocial approaches are often required toprevent their onset and to support those already affected.

Limitations

The findings of our study need to be considered in the context ofits limitations. First, the cross sectional design means that no causalinferences can be drawn from the reported associations. Thus the strongassociation between somatisation and depression and anxiety in ourstudymay be due to forward or reverse causality or because both disor-ders share common underlying risk factors. It has been suggested bystudies of Pakistani adults, that somatisation is a symptomof depressionand anxiety. This may also be the case for children in Pakistan. Anotherlimitation is that the studywas conducted in tertiary care clinical settingthus the sample may not be representative of children presenting withsomatisation in primary care or community settings.We usedmeasuresof depression, anxiety, and somatisation developed in Western coun-tries. Although the instruments were carefully translated, culturallyadapted, and they demonstrated very good internal consistencies inour study, they have nonetheless not been formally validated inPakistani children. In addition, self report measures may not allow fulland intended description of symptomatology. Thus the significantdifferences in symptomatology between groups do not necessarilyindicate the presence of specific disorders but rather reflect relativedifferences. We also relied on parental information about number of

111N. Imran et al. / Journal of Psychosomatic Research 76 (2014) 105–112

school days missed and did not obtain official school absence data dueto limited resources.

While acknowledging the limitations, the study also has somestrengths. To our knowledge this is the first study in South-East Asiato use a three group case–control design to systematically documentthe association of anxiety and depressive symptoms among childrenwith medically unexplained somatic symptoms in the region. Thesample size was adequate to test the hypothesis. Use of two controlgroups instead of one, strengthened the design. The questionnaireswere robustly translated into Urdu. We also blinded the participantsand interviewers to the study hypothesis to reduce information bias.

Conclusion & future directions

The results have important clinical implications for themanagementof childrenwith somatisation. It highlights the need for clinicians work-ing with children with medically unexplained symptoms in Pakistan toactively screen affected children for anxiety and depression and func-tional impairment. Children identified with these difficulties need tobe offered appropriate mental health and psychosocial interventionsto limit the additional burden, distress and impairments on them andtheir families. However, given that intervention studies in this illness-group have been conducted in developed countries treatment trialsare required in Pakistan and other developing countries.

Conflict of interest

The authors have no competing interests to report.

Acknowledgments

We thank the children and parents who participated in the study.We thank the interviewers and translators. We are grateful to thehead teacher of the school that allowed for the recruitment of thehealthy controls.

References

[1] Wessely S, Nimnuan C, Sharpe M. Functional somatic syndromes: one or many?Lancet 1999;354:936–9.

[2] Campo JV, Fritsch SL. Somatization in children and adolescents. J Am Acad ChildAdolesc Psychiatry 1994;33:1223–35.

[3] Garralda ME. Somatization in children and adolescents. J Child Psychol Psychiatry1996;37:13–33.

[4] Gledhill J, Garralda ME. The relationship between physical and mental health inchildren and adolescents. In: Gelder MG, Andreassen NC, Lopez-Ibor JJ, et al, editors.New oxford textbook of psychiatry. 2nd ed. Oxford: Oxford University Press; 2009.p. 1740–6.

[5] Egger HL, Costello E, Erkanli A, Angold A. Somatic complaints and psychopathologyin children and adolescents: stomach aches, musculoskeletal pains, and headaches. JAm Acad Child Adolesc Psychiatry 1999;38:852–60.

[6] Walker LS, Garber J, Greene JW. Psychosocial correlates of recurrent childhood pain:a comparison of pediatric patients with recurrent abdominal pain, organic illness,and psychiatric disorders. J Abnorm Psychol 1993;102:248.

[7] Walker LS, Garber J, Van Slyke DA, Greene JW. Long-term health outcomes inpatients with recurrent abdominal pain. J Pediatr Psychol 1995;20:233–45.

[8] Garber J, Zeman J, Walker LS. Recurrent abdominal pain in children: psychiatricdiagnoses and parental psychopathology. J Am Acad Child Adolesc Psychiatry1990;29:648–56.

[9] Hodges K, Kline JJ, Barbero G, Woodruff C. Anxiety in children with recurrentabdominal pain and their parents. Psychosomatics 1985;26:859–66.

[10] Ramchandani PG, Fazel M, Stein A, Wiles N, Hotopf M. The impact of recurrentabdominal pain: predictors of outcome in a large population cohort. Acta Paediatr2007;96:697–701.

[11] Hotopf M, Carr S, Mayou R, Wadsworth M,Wessely S. Why do children have chronicabdominal pain, and what happens to them when they grow up? Population basedcohort study. BMJ 1998;316:1196–200.

[12] Sandberg S, Stevenson J. Psychiatric aspects of somatic disease. In: Rutter M, BishopD, Pine D, Scott S, Stevenson J, Taylor E, Thapar A, editors. Rutter's child and adoles-cent psychiatry. 5th ed. Oxford: Wiley-Blackwell; 2008. p. 930–44.

[13] Campo JV, Jansen-McWilliams L, Comer DM, Kelleher KJ. Somatization in pediatricprimary care: association with psychopathology, functional impairment, and useof services. J Am Acad Child Adolesc Psychiatry 1999;38:1093–101.

[14] Weitzman M, Walker DK, Gortmaker S. Chronic illness, psychosocial problems, andschool absences, results of a survey of one county. Clin Pediatr 1986;25:137–41.

[15] SmithMS,Martin-Herz SP,WomackWM,Marsigan JL. Comparative study of anxiety,depression, somatization, functional disability, and illness attribution in adolescentswith chronic fatigue or migraine. Pediatr 2003;111:e376–81.

[16] Parker G, Gladstone G, Chee KT. Depression in the planet's largest ethnic group: theChinese. Am J Psychiatry 2001;158:857–64.

[17] Patel V, Gwanzura F, Simunyu E, Lloyd K, Mann A. The phenomenology and explan-atory models of common mental disorder: a study in primary care in Harare,Zimbabwe. Psychol Med 1995;25:1191–9.

[18] Patel V, Pereira J, Coutinho L, Fernandes R. Is the labelling of common mental disor-ders as psychiatric illness clinically useful in primary care? Indian J Psychiatry1997;39:239–46.

[19] Patel V, Pereira J, Mann AH. Somatic and psychological models of common mentaldisorder in primary care in India. Psychol Med 1998;28:135–43.

[20] Shidhaye R, Mendenhall E, Sumathipala K, Sumathipala A, Patel V. Association ofsomatoform disorders with anxiety and depression in women in low and middleincome countries: a systematic review. Int Rev Psychiatry 2013;25:65–76. http://dx.doi.org/10.3109/09540261.2012.748651.

[21] Weiss B, Tram JM,Weisz JR, Rescorla L, Achenbach TM. Differential symptom expres-sion and somatization in Thai vs American children. J Consult Clin Psychol2009;77:987–92.

[22] Katon W, Kleinman A, Rosen G. Depression and somatization: a review. Am J Med1982;72:127–35.

[23] Mubbashar MH, Saeed K. Development of mental health services in Pakistan. EastMediterr Health J 2001;7:392–6.

[24] Minhas FA, Nizami AT. Somatoform disorders: perspectives from Pakistan. Int RevPsychiatry 2006;18:55–60.

[25] Rief W, Barsky A. Psychobiological perspectives on somatoform disorders.Psychoneuroendocrinology 2005;30:996–1002.

[26] Lekander M, Elofsson S, Neve IM, Hansson LO, Unden AL. Self-rated health is relatedto levels of circulating cytokines. Psychosom Med 2004;66:559–63.

[27] Wade A. Statistics and researchmethodology. London: Institute of Child Health; 1997.[28] Hussein SA, Vostanis P. Urdu translation and cultural adaptation of schedule for

affective disorders & schizophrenia for school age children (6–18 yrs) K-SADS-IVR JPPS. 2008;5:81.

[29] Walker LS, Beck JE, Garber J, Lambert W. Children's somatization inventory: psycho-metric properties of the revised form (CSI-24). J Pediatr Psychol 2009;34:430–40.

[30] Walker LS, Garber J, Green JW. Somatization symptoms in pediatric abdominal painpatients: relation to chronicity of abdominal pain and parent somatization. J AbnormChild Psychol 1991;19:379–94.

[31] Angold A, Costello EJ, Messer SC, Pickles A, Winder F, Silver D. The development of ashort questionnaire for use in epidemiological studies of depression in children andadolescents. Int J Methods Psychiatry Res 1995;5:237–49.

[32] Costello EJ, Angold A. Scale to assess child & adolescent depression: checklists,screens and nets. J Am Acad Child Adolesc Psychiatry 1988;27:726–37.

[33] Walker S, Chang SM, Powell CA, Simonoff E, Grantham-McGregor S. Early childhoodstunting is associated with poor psychological functioning in late adolescence andeffects are reduced by psychosocial stimulation. J Nutr 2007;137:2464–9.

[34] Thabet AAM, AbedYand Vostanis P. Comorbidity of PTSD and depression amongrefugee children during war conflict. J Child Psychol Psychiatry 2004;45:533–42.

[35] Spence SH, Barrett PM, Turner CM. Psychometric properties of the Spence Children'sAnxiety Scale with young adolescents. J Anxiety Disord 2003;17:605–25.

[36] Ali BS, Reza H, Khan MM, et al. Development of an indigenous screening instrumentin Pakistan: the Aga Khan University Anxiety and Depression Scale. J Pak Med Assoc1998;48:261–5.

[37] Ali BS. Validation of an indigenous screening questionnaire for anxiety anddepression in an urban squatter settlement of Karachi. J Coll Physicians Surg Pak1998;8:207–11.

[38] Pallant J. SPSS survival manual. 3rd ed. Maidenhead: Open University Press; 2007.[39] Zuckerman B, Stevenson J, Bailey V. Stomachaches and headaches in a community

sample of preschool children. Pediatrics 1987;79:677–82.[40] Hyams JS, Burke G, Davis PM, Rzepski B, Andrulonis PA. Abdominal pain and irritable

bowel syndrome in adolescents: a community-based study. J Pediatr 1996;129:220–6.[41] Campo JV, Bridge J, Ehmann M, Altman S, Lucas A, Birmaher B, et al. Recurrent ab-

dominal pain, anxiety, and depression in primary care. Pediatrics 2004;113:817–24.[42] Saps M, Seshadri R, Sztainberg M, Schaffer G, Marshall BM, Di Lorenzo C. A prospec-

tive school-based study of abdominal pain and other common somatic complaints inchildren. J Pediatr 2009;154:322–6.

[43] Mumford DB, Minhas FA, Akhtar I, Akhter S, Mubbashar MH. Stress and psychiatricdisorder in urban Rawalpindi. Community survey. Br J Psychiatry 2000;177:557–62.

[44] Mumford DB, Nazir M, Jilani FU, Baig IY. Stress and psychiatric disorder in the HinduKush: a community survey of mountain villages in Chitral, Pakistan. Br J Psychiatry1996;168:299–307.

[45] Mumford DB, Saeed K, Ahmad I, Latif S, Mubbashar MH. Stress and psychiatricdisorder in rural Punjab. A community survey. Br J Psychiatry 1997;170:473–8.

[46] Afridi MI, Siddiqui MA, Ansari A. Gastrointestinal somatization in males and femaleswith depressive disorders. J Pak Med Assoc 2009;59:675–9.

[47] Mian MD, Rubeena H. The somatization of depression in native Pakistani women.Jefferson J Psychiatry 2012;14:19–23.

[48] Haavet OR, Grünfeld B. Are life experiences of children significant for the developmentof somatic disease? A literature review. Tidsskr Nor Laegeforen 1997;117:3644–7.

[49] Lavigne JV, Faier-Routman J. Psychological adjustment to pediatric physical disor-ders: a meta-analytic review. J Pediatr Psychol 1992;17:133–57.

[50] Woolfolk RL, Allen LA, Tiu JE. New directions in the treatment of somatisation.Psychiatr Clin North Am 2007;30:621–44.

112 N. Imran et al. / Journal of Psychosomatic Research 76 (2014) 105–112

[51] Campo JV, Fritz G. A management model for pediatric somatization. Psychosomatics2001;42:467–76.

[52] Hicks CL, Von Baeyer CL, McGrath PJ. Online psychological treatment for pediatric re-current pain: a randomized evaluation. J Pediatr Psychol 2006;31:724–36.

[53] Robins PM, Smith SM, Glutting JJ, Bishop CT. A randomized controlled trial of acognitive-behavioral family intervention for pediatric recurrent abdominal pain. JPediatr Psychol 2005;30:397–408 [28].

[54] Fiertag O, Taylor S, Tareen A, GarraldaM E. Somatoform disorders. In: Rey JM, editor.IACAPAP e-textbook of child and adolescentmental health. Publisher IACAPAP; 2012[http://hdl.handle.net/10044/1/10361 Publisher URL http://iacapap.org/iacapap-textbook-of-child-and-adolescent-mental-health.Pine D and Klein R. Anxiety disor-ders. In: Rutter M, Bishop D.

[55] Pine D, Scott S, Stevenson J, Taylor E, Thapar A, editors. Rutter's Child and AdolescentPsychiatry, Fifth Edition. Oxford: Wiley-Blackwell; 2008. p. 628–47.