Temperament in Adolescents with Anxiety and Depressive Disorders and in Their Families Gabriele Masi, MD Maria Mucci, MD Letizia Favilla, MD Paola Brovedani, PhD Stefania Millepiedi, MD Scientific Institute of Child Neurology and Psychiatry, IRCCS Stella Maris, Calambrone (Pisa), Italy

Giulio Perugi, MD Department of Psychiatry, University of Pisa, Italy, and Institute of Behavioural Sciences, Carrara-Pisa, Italy.

ABSTRACT: Aim of this study was to investigate whether specific temperamental features were associated with anxiety and depressive disorders in adolescents, in their siblings and in their parents. Thirty adolescents with Anxiety disorders and 25 with both Anxiety and Depressive disorders were compared to 25 adolescents with learning disorders and to 28 normal subjects. Temperament in subjects and relatives was assessed by their parents with the EAS questionnaire. Subjects with Anxiety and AnxietyDepression and their siblings showed higher scores on Emotionality and Shyness than Learning Disability and Normal subjects. Mothers and fathers of subjects from the Anxiety-Depression group had the highest Emotionality score. These findings suggest that both Emotionality and Shyness are prominent temperamental features in adolescents with anxiety with or without depression, and in their parents and siblings. KEY WORDS: temperament; anxiety; depression; adolescence.

Temperament is a long-standing behavioral style that is evident early in life, stable in time, observable in a variety of situations, and considered to affect personality development.1,2 One point of agreement among different conceptual models is that at some level and to some agree temperament is biologically based, even though the degree of adherence to a biological model varies considerably.3 A high stabilAddress correspondence to Gabriele Masi, MD, Division of Child Neurology and Psychiatry—IRCCS Stella Maris, Via dei Giacinti 2 56018 Calambrone (Pi) Italy; e-mail: [email protected]. Child Psychiatry and Human Development, Vol. 33(3), Spring 2003  2003 Human Sciences Press, Inc.

245

246

Child Psychiatry and Human Development

ity of temperament expression in time has been confirmed by many studies.4,5 Different models have described a variety of structures or dimensions of temperament, supported by factor analytic studies, with considerable conceptual overlaps.6 Clusters of related groups of dimensions have been studied, which may be more directly related with personality development.7 The most frequently researched cluster of dimensions has been described as the “difficult temperament,”8 which has been shown to correlate with behavioral maladjustment.4 Even if the concept of “difficult temperament” is still controversial, common in all definitions is a negative emotionality together with management problems for caretakers in social interactions.3 A major limiting factor is that particular temperament characteristics cannot be considered to be universally difficult in different places or times.9 However, although there are persisting problems with the concept, its value to clinicians is so high (i.e., in exploring relationship with later development and psychiatric risk) that it is unlikely to be abandoned.3 Many studies have suggested a relationship between certain maladaptive temperamental traits and emotional and/or behavioral problems in children and adolescents.3,10–12 Individuals at the extremes of the different temperamental dimensions are more likely to suffer from emotional problems.1 It is still debated if a maladaptive temperament is non-specifically related to psychopathology, or if there is a specific relationship that characterizes particular disorders or subtypes of disorders.12–18 Rutter2 revised various mechanisms which may mediate the relationship between temperament and emotional-behavioral problems. The first is that certain temperamental traits directly increase vulnerability to certain psychopathological disorders. Another possibility is that specific temperamental profiles increase the susceptibility to psychosocial adversities. Alternatively, psychopathology may arise when temperamental needs are not satisfied by specific environmental conditions (“goodness of fit”). A related mechanism is that temperament may affect children’s choice of activities and environments. Another interactional model is that children’s temperamental traits can influence other people’s responses to them. If some aspects of temperament have a critical role in the development of psychopathology, one would expect to find high rates of these temperamental traits among subjects with psychiatric disorders. Furthermore, as temperament may be viewed as an inherited response disposition, if a difficult temperament is associated with psychopathology, it could be one of the factors by which psychiatric disorders aggre-

G. Masi, M. Mucci, L. Favilla, P. Brovedani, S. Millepiedi, and G. Perugi

247

gate in families.19–20 The study of temperament in parents and siblings of probands, as well as in offspring of affected adults, may outline the role of temperament in the familial transmission of mental disorders.15,21–24 A model assessing temperament is by interviews or questionnaires with parents. These diagnostic instruments can generate data which are amenable for group comparisons, are inexpensive and convenient to administer. Even if they are open to criticism of bias, because they rely on parental ratings, it is true that parents are usually the best informants on their child’s behavior.25 A widely used questionnaire for assessing temperament is the EAS Scale.1 According to the model underlying this questionnaire, temperament can be resolved in four main dimensions: Emotionality, Activity, Sociability, Shyness.7,26–27 Emotionality refers to the tendency to become upset easily; Sociability refers to the tendency to prefer the presence of others; Activity refers to the tendency to be restless; Shyness refers to the uneasiness in social situations that are novel or unfamiliar. The EAS approach is age-continuous, because it relies upon more general expressions of temperament, which are assumed to be stable across a wide age span.1 For this reason items are common across different age levels. Several studies have explored temperamental features in adolescents with anxiety and or depression using the EAS. Goodyer and coworkers have found that high Emotionality was associated with depression in a small sample of non referred adolescents compared with normal controls.15 The risk of depression was greater in subjects with high Emotionality scores, but not with high scores in other temperamental dimensions, alone or associated with Emotionality. Kelvin and coworkers reported that Emotionality was the only temperamental dimension that distinguished 29 adolescents with depression and/or anxiety (21 with depressive disorder and 8 with anxiety disorders) and normal controls.23 Rende showed that in 167 male adolescents from a community sample Emotionality only was related with the AnxietyDepression scale of the Child Behavior Checklist (CBCL),28 while in females high Emotionality and low Sociability were related with this scale.14 Kelvin et al. investigated the temperamental profile in siblings of children and adolescents with depression and/or anxiety disorders, comparing them with those of normal individuals.23 Higher levels of Emotionality, but not other temperamental dimensions, were observed in siblings of an affected subject. In summary, these findings seem to suggest that Emotionality is the temperamental trait that is most frequently related to depression and anxiety. More uncertain is

248

Child Psychiatry and Human Development

the specificity of this relationship, because these studies did not use control groups with other disorders. Methodological issues (referred versus community samples, young children versus adolescents, type of illness) and socio-cultural variables (socioeconomic status or family characteristics) may account for possible discrepancies in different studies. The aim of this preliminary investigation was twofold: • To investigate whether parentally perceived temperaments, assessed with the EAS questionnaire, were associated to anxiety with or without mood disorders in adolescents; • To determine whether these temperament profiles were present in siblings and in parents of probands with anxiety with or without depressive disorders. In order to determine specificity of the findings, a “normal” control group of adolescents from the community and a pathological control group with learning disorders were selected. Method Subjects All the adolescents aged between 12 and 18 years referred to our Division as outpatients during 18 months were screened for psychiatric disorders, using historical information, a clinical interview, the Diagnostic Interview for Children and Adolescents–Revised (DICA-R),29 symptoms ratings according to the DSM-IV criteria,30 and evaluation of temperament (EAS questionnaire). Patients with psychosis and/or mental retardation and/or pervasive developmental disorder did not participate in this study. The comprehension of the questions was carefully assessed; if necessary, they were repeated to clarify the subject’s response. All subjects participating in the study were considered competent to undergoe the diagnostic interview. Structured interview diagnoses were considered positive only if DSM-IV criteria were unequivocally met. Our previous analyses on children and adolescents assessed with DICA-R revealed a good inter-rater reliability for the diagnosis of mood and anxiety disorders (k > .75).24,31–32 Out of these patients, a consecutive sample of 58 adolescents (33 males and 25 females) were diagnosed as affected by an anxiety disorder with or without comorbid depressive disorder. The subjects were divided in two groups, according to the presence or absence of a comorbid depressive disorder: Anxiety Group: 33 subjects with one or more anxiety disorder without comorbid mood disorders, 17 males and 16 females, age range 12 to 16 years, mean age 13.5 ± 2.9 years; the anxiety disorders represented in this group (frequently comorbid) were: Generalized Anxiety Disorder (25 subjects), Simple Phobias (10 subjects), Panic Disorder (10 subjects), Social Phobia (8 subjects), Separation Anxiety Disorder (8 subjects).

G. Masi, M. Mucci, L. Favilla, P. Brovedani, S. Millepiedi, and G. Perugi

249

Anxiety-Depression Group: 25 subjects with one or more anxiety disorders, comorbid with a depressive disorder (Major Depressive Disorder and/or Dysthymia), 16 males and 9 females, age range 12 to 16 years, mean age 14.3 ± 2.5 years. The anxiety disorders in this group were: Generalized Anxiety Disorder (16 subjects), Simple Phobias (7 subjects), Panic Disorder (7 subjects), Social Phobia (6 subjects); Separation Anxiety Disorder (5 subjects). The depressive disorders in this group were: Dysthymia (13 subjects), Major Depressive Disorder (10 subjects), Dysthymia + Major Depressive Disorder (2 subjects). To compare temperamental features of these probands, two control groups of adolescents were selected. The first consisted of subjects with Learning disorders (reading disorder, mathematics disorder, disorder of written expression) according to DSM IV diagnostic criteria, and normal IQ, assessed with the WISC-R. This type of disorder was selected because emotional impact on subjects and their families and functional impairment were considered similar to those of anxiety disorders. Twenty-five subjects with Learning Disorders without anxiety and/or depression, 15 males and 10 females, age range 12 to 16 years, mean age 13.8 ± 2.6 years, were included in the study. A Normal control group was composed of 28 non-referred students, 13 males and 15 females, age range 12–14 years, mean age 13.3 ± 0.4 years, attending the second and third year of a Junior High school from our district. These adolescents did not receive a full clinical assessment, but only an evaluation of temperament. However, they did not present previously academic or behavioral disorders according to teachers’ evaluation. The temperamental characteristics of both parents and siblings of the subjects in the four groups were examined with the EAS questionnaire. The sample included 220 parents and 77 siblings under the age of 17. Socio-demographic variables were assessed by the parents’ education and occupation, according to the Hollingshead’s two-factor index.33 All the subjects were in the categories IIIII (middle to upper-middle socioeconomic status level). Socioeconomic status distribution (highest versus middle categories) and family structure (intact versus one-parent families) did not differ between the groups. All subjects and their families participated in the study after informed consent. The study was approved by the Human Subject Committee of our hospital.

Measures As a part of a broader clinical assessment, a clinical interview, the Diagnostic Interview for Children and Adolescents-Revised (DICA-R) was administered individually to the adolescents and their parents by child psychiatrists who were highly experienced in diagnosis and treatment of mood and anxiety disorders, and properly trained in using the diagnostic interview. The DICAR is a structured interview exploring the presence or absence of each of the symptoms in different psychiatric syndromes, allowing diagnosis to be made according to DSM IV criteria. The DICA-R was entirely administered separately to parents and subjects of the Anxiety, Anxiety-Depression and Learning disorder groups. Structured interview diagnoses were considered positive only if DSM IV criteria were unequivocally met.

250

Child Psychiatry and Human Development

The temperament of probands and siblings was measured using the Buss and Plomin EAS Questionnaire. This instrument explores general expressions of temperament, which are assumed to be stable across a wide age span.1 The EAS Questionnaire has been used extensively at all ages throughout childhood and adolescence, both in clinical and community samples.1,7,14,15,34 It is composed of 20 questions, each with a five-point response scale, which has three anchor points (1 = non typical; 3 = fairly typical; 5 = very typical) and two points, 2 and 4, undefined and intermediate. Questions can be grouped in four subscales that correspond to different temperamental dimensions or traits: Emotionality, Activity, Sociability, Shyness. A mean score for each dimension is obtained by summing the scores of the questions referring to the dimension and dividing the total score of the subscale by 5. The mean score indicates to what degree the subject possesses that temperamental dimension. Parents were instructed to consider lifetime temperamental features of their children, that is the temperament that preceded the onset of current psychopathological disorders. The overall reliability and validity of the instrument and its dimensions are not influenced by sex or age of the child.1,7,34–35 The longitudinal Colorado Adoption Project conducted a test-retest reliability (administration of the scales being 12 months apart) for each subscale of parent EAS report using both mothers’ and fathers’ report independently in a sample of 214 10 to 16 year olds. The correlation was 0.80 or greater for each subscale for either parent report.35 Goodyer et al. report on a study investigating teacher reports on 253 11-year-olds whilst at primary school and again by new teachers, 9 months later, following transfer to secondary school.15 Pearson correlation coefficients for each subscale were: Emotionality 0.66, Activity 0.70, Sociability 0.58, Shyness 0.65.15 These findings indicate that the EAS has sufficient reliability and stability to be used for research on adolescent populations. Assuming the stability of temperamental traits across ages, an adult form of EAS has been created.1 The items of the adult EAS (i.e., I like to be with people, I usually seem to be in a hurry, I am easily frightened, etc.) refer to the temperament of the subjects as adults. This form explores five traits, Activity, Sociability, Distress, Fearfulness and Anger. As the last three dimensions are considered by the Buss and Plomin as different expressions of Emotionality, in our study they were considered as a unique dimension (Emotionality). The Emotionality score was obtained by summing the scores of Distress, Fearfulness and Anger, and then dividing the sum by 3. The Questionnaire for assessing probands’ and siblings’ temperament was completed by each of the parents. Parents also completed the Adult form of the Questionnaire relative to their own temperament. The adult form of EAS was validated in 330 subjects, in terms of factor analyses, distributional properties (means approximately 3 and standard deviation 1), and test-retest reliability (.82).1 Furthermore, intercorrelations between traits showed that Emotionality scales were independent of Activity and Sociability.1

Statistical Analyses An analysis of variance with between subjects factors (diagnosis, gender) and within subjects factors (temperamental traits) was computed to explore

G. Masi, M. Mucci, L. Favilla, P. Brovedani, S. Millepiedi, and G. Perugi

251

main effect of gender or diagnosis. Post hoc analyses were performed with the Tukey test. Bonferroni correction was applied, to minimize Type I errors.

Results Differences among the temperamental traits in the total sample of probands are reported in Table 1. The analysis of variance with between subjects factors (diagnosis, gender) and within subjects factors (temperamental traits) yielded no main effect of gender, nor of diagnosis. There were significant simple interactions between diagnosis and temperamental traits (F = 5.20, df = 9,264, p < .001) and complex interactions between diagnosis, gender and temperamental traits (F = 1.94, df = 9,264, p = .047). After Bonferroni correction, which set alpha at 0.001, the only significant gender difference resulted in the Anxiety-Depression group, where males showed higher Shyness than females (t = 4.7, df = 23, p < .0001). Other differences did not reach statistical significance. Separate analyses of variance were computed for each temperamental trait in the diagnostic groups. Emotionality score significantly distinguished the four groups (F = 11.33, df = 3,107, p < .0001). Post-hoc analysis (Tukey) revealed that both Anxiety and Anxiety-Depression groups had higher Emotionality than Normal and Learning Disability groups. Shyness also differentiated the four groups (F = 11.12, df = 3,107 p < .0001). After the post-hoc, Anxiety and Anxiety-Depression groups revealed significantly higher scores than Normal and the Learning Disability adolescents. Activity and Sociability scores did not differentiate the four groups. Temperamental traits in the siblings of the subjects are described Table 1 Mean Scores and Standard Deviations of EAS Temperament Subscales in the Four Groups of Probands Probands

AD (n = 25)

A (n = 33)

LD (n = 25)

N (n = 28)

p

Emotionality Activity Sociability Shyness

3.34 ± 0.79 3.19 ± 0.99 3.27 ± 0.70 2.94 ± 0.91

2.93 ± 0.9 3.39 ± 0.68 3.25 ± 0.43 2.84 ± 0.68

2.34 ± 0.76 3.74 ± 0.79 3.44 ± 0.83 2.2 ± 0.77

2.10 ± 0.85 3.63 ± 0.96 3.56 ± 0.65 1.93 ± 0.75