Abstract
Attention deficit hyperactivity disorder, combined type (ADHD-CT) in primary school-age children is a common presenting condition in public child mental health services [1, 2]. Anxiety is a frequent comorbid condition in these referred primary school-age children with ADHD-CT [3–5], yet there has been relatively little systematic research of the nature of this comorbid anxiety [1, 5, 6], while comorbid oppositional defiant disorder (ODD)/conduct disorder (CD) and language-based learning disorders have been more thoroughly investigated [1, 5, 6]. This relative paucity of research is incongruent with the average 25% prevalence of ADHD-CT and anxiety in clinical and epidemiological samples [3] and its clinical importance, given the replicated association of its relatively poor response to conventional psychological and psychostimulant medication treatments, in both the short- and the longer-terms [7–9]. In addition, ADHD-CT has been associated with increased rates of comorbid anxiety and this anxiety does not appear to decrease with increasing age [10–12]. In particular, increased rates of SAD have been reported with ADHDCT, compared to ADHD, inattentive type (ADHD-IA) and control subjects [13]. Kashani et al. reported that adolescents aged 17 years with ADHD (DSM-IV [15] ADHD-CT equivalent) and anxiety had significantly more ODD behaviours than those with ADHD without anxiety [14]. This association was not apparent when children aged eight years were assessed. Therefore, there is emerging evidence for a subgroup of primary schoolage children with ADHD-CT and anxiety who continue to experience ADHD-CT, ODD/CD, and anxiety symptoms into adolescence. This subgroup is different from the more well-known association of primary school-age children with ADHD (primarily ADHD-IA) and anxiety with decreased severity of ADHD-CT symptoms, decreased comorbidity rates of ODD/CD, and a better prognosis in terms of educational achievement, occupational record, and established interpersonal relationships [1, 5].
Both the child and the parent report have been noted to be important in determining comorbid anxiety associated with ADHD-CT, for significant clinical correlates, such as levels of self-confidence and impairments in activities of daily living, may be associated with the child report alone [16]. In addition, only approximately 50% of children with self-reported anxiety have also been reported with anxiety by their parents [17]. Recently, March et al. have noted that the parent report of their given child's anxiety disorder(s), from the Multimodal Treatment Study of children with ADHD-CT, may represent their child's ‘negative affectivity and associated behavioural problems’ rather than ‘neurotic anxiety suffered by children with anxiety disorders alone [sic]’ [18]. These findings are consistent with the known usefulness of the child's self-report of anxiety in identifying internalizing symptoms [19–21].
Epidemiological samples of children and adolescents and clinical samples of children and adolescents with various diagnoses have reported higher prevalence rates of one or more anxiety disorder diagnoses of generalized anxiety disorder (GAD), separation anxiety disorder (SAD), specific phobia (SpPh) and social phobia (SoPh) [22–35]. In particular, ADHD-CT has also been associated with GAD, SAD, SpPh and SoPh in clinical samples of children [24, 26]. Factors such as the origin of the sample (for example, community derived, ADHD/anxiety/depression clinic derived, child and/or adolescent population[s]) and the informant(s) interviewed affects the prevalence rates reported. Previous studies have reported data from parent derived [24, 27], parent and child (in combination) derived [25, 26, 28–33], and child derived [34, 35] anxiety disorder diagnoses. However, to date, parent and child derived anxiety disorder diagnoses have not been reported in a clinical sample of primary school-age children with ADHD-CT.
The first aim of this study was to define two clinically referred groups of primary school-age children with ADHD-CT and comorbid anxiety disorders from (i) a parent and (ii) a child perspective. These two groups of ADHD-CT and anxiety disorders were matched for age, gender, ADHD-CT symptoms, intelligence quotients (IQ), language-based learning abilities (spelling and arithmetic), and family and social adversity factors, to compare comorbid anxiety disorder diagnoses. The first hypothesis tested was that the rates of comorbid anxiety disorder will be similar in the two groups and that the most frequent anxiety disorder diagnoses will be GAD, SAD, SpPh, and SoPh in both groups. The second aim was to follow up, in the longer-term (at two years), a representative subsample of these two groups to compare comorbid anxiety disorder diagnoses. The second hypothesis tested was that the rates of anxiety disorder diagnoses will not decrease in both these groups over the longer term.
Method
Subjects
Seventy-five children aged from six to 12 years were identified in a specialized clinic for ADHD in the southern and eastern regions of metropolitan Melbourne, Australia. All the children had ADHD-CT, according to DSM-IV [15] criteria, defined through a semistructured clinical interview with the child's parent(s) and by both the parent and teacher self-reports of the subscale scores of the core symptom domains of ADHD-CT being greater than 1.5 standard deviations above the mean for a given child's age and gender. The children were all psychostimulant medication naïve and were referred for assessment because they were not responding to usual clinical psychological management approaches. The children met the inclusion criteria of living in a family home (and not in an institution) and attending normal primary schools. All had IQs above 70 and none had overt neurological disease or psychotic symptoms. There was no refusal. Informed consent was given by the parent(s). Two groups were formed based on the parent (34 children) and child (28 children) reports of anxiety and were matched for age, gender, ADHD-CT symptoms, full scale IQ, and language-based learning abilities (spelling and reading), general family functioning, and social adversity factors. Subject characteristics were compared across the two groups using independent t-tests (detailed tabulated information available from the Correspondence). Twelve children with a child report of anxiety and 12 children with a parent report of anxiety were followed up 2 years after their initial assessment (follow-up range: 23–26 months). Both groups of children received standard individual and group psychological interventions and medication treatments over this time.
Measures
Child behaviour checklist (CBCL) [36] consists of 118 behaviour problem items which are rated by the parent and teacher on a threepoint scale as to how well each describes the child. Parent and teacher forms are used. A total behaviour problems scaled score, including subscale scores of all the symptom domains of ADHD-CT, and an anxiety/depression subscale, are generated. All scores form continuous variables. The CBCL is well researched and has adequate psychometric properties.
Abbreviated Conners rating scale is a ten-item brief questionnaire that outlines the core symptom domains of ADHD-CT [37]. It is completed by parent(s) and teachers and yields a continuous variable of ADHD-CT symptoms. It has good validity and reliability and has been used extensively in the literature.
Anxiety disorders interview schedule for children (A-DISC) is a structured diagnostic interview schedule based on DSM-IV criteria with child and parent versions [38]. It is frequently used internationally by researchers to diagnose childhood anxiety disorders and a range of other disorders, including ADHD-CT and ODD, based on the DSM-IV criteria. It yields a categorical variable of the presence or absence of a given disorder. Research findings support the clinical utility, reliability and validity of the A-DISC.
Revised child manifest anxiety scale (R-CMAS) [39] consists of 37 items designed to assess anxiety. The R-CMAS also contains a scale to assess social desirability or lying. All scores form continuous variables. Good test-retest reliability and concurrent validity have been demonstrated.
Parental account of childhood symptoms (PACS) [40] is a semistructured clinical interview developed as an instrument for the measurement of children's behaviour problems as experienced at home. It is administered by a trained interviewer. Parents are asked for detailed descriptions of what their child has done in specified situations over the previous week and over the preceding 12-month period. In all, 44 items of behaviour are enquired about in this way. Three subscales are generated as continuous variables; hyperactivity, emotional disorder and defiance. It also yields demographic data. A social adversity scale (range 0–5) is formed from the summary of family income level (0–1), mother's educational level (0–1), single parent status (0–1), sibling size (0–1), and broken home status (0–1). It has adequate validity and reliability and has been used extensively in published research studies.
Hopkins symptom checklist (HSCL) [41] comprises 58 items that measure five symptom dimensions: anxiety, depression, obsessivecompulsive symptoms, somatization, and interpersonal sensitivity, using a 1–4 Likert scale. A total score provides a general measure of psychopathology. Adequate validity and reliability have been demonstrated in published research findings.
Family assessment device (FAD) [42] was developed from the McMaster model of family functioning to measure the structural and organizational characteristics of families. It consists of 60 items that measure six dimensions of family functioning: problem solving, communication, affective responsiveness, affective involvement, roles and behavioural control. Each dimension forms a subscale with an additional general functioning subscale that measures the overall health/dysfunction in the family unit. Adequate validity and reliability have been demonstrated in published research findings.
The third edition of the Wechsler intelligence scale for children (WISC-3) [43] was used. This provides verbal, performance and full scale scores of measured intelligence. The third edition of the wide range achievement test (WRAT-3) [44] provides a measure of spelling and arithmetic abilities. The revised version of the Neale analysis of reading [45] provides a measure of reading accuracy and comprehension. All of the above tests are well established with Australian normative data.
Procedure
The CBCL and ACRS parent and teacher forms were completed prior to the child's assessment. The child was initially interviewed by a fellow in child and adolescent psychiatry who completed the A-DISC-child form. The child subsequently completed the R-CMAS. At a separate interview, a registered psychologist administered the WISC-3, the WRAT-3, and the Neale analysis of reading ability, revised, to the child, while a fellow in child and adolescent psychiatry concurrently administered the A-DISC-parent form and the parents completed the PACS, HSCL and the FAD.
Variable definition and data analysis
Continuous variables of anxiety from the child's self-report and anxiety/depression from the parent report were also defined and then point-biserial correlations made with the respective child and parent categorical variables, to ensure the categorical variables were an adequate measure of anxiety, from the child and the parent perspectives. The parent and child reports of comorbid anxiety were described by their number, and the percentage of each diagnosis within the total sample of 75 children. ODD and CD diagnoses were also obtained from the parent report. Separate categorical variables were formed from the child and parent reports of anxiety disorder(s). Each had a value of ‘1’ if one or more anxiety disorder(s) was/were present or ‘0’ if not present, from the respective parent and child reports. The Chi-Squared test was used to compare these two categories of anxiety in the sample of 75 children. The α level was set at 0.01 to guard against an inflated type 1 error rate due to multiple comparisons. The parent and child reports of comorbid anxiety and the parent report of ODD and CD, at baseline and at two-year follow-up, were described by their number and the percentage of each diagnosis within the total, respective, parent and child reports of anxiety in the 24 children followed-up. The paired samples t-test was used to compare the baseline and twoyear follow up (i) parent reports, and (ii) child reports of anxiety disorders separately.
Results
The presence of one or more anxiety disorders from the child's perspective had a moderate correlation with the total anxiety score from the child's self-report of anxiety (r = 0.49; p < 0.001). The presence of one or more anxiety disorders from the parent's perspective had an adequate correlation with the anxiety/depression subscale score from the parent report (r = 0.40; p < 0.05).
The parental semistructured clinical interview revealed an ODD rate of 48% (n = 36) and CD rate of 8% (n = 6). Both the parent and the child report data suggested relatively increased rates of OCD, aside from GAD, SAD, SpPh and SoPh being the most commonly reported anxiety disorders. Differences between parent and child reports were evident. The parent report revealed a slightly higher rate of GAD and SoPh, and a significantly higher rate of SpPh (χ 2 [1] = 5.65, p = 0.02). In contrast, the child report resulted in significantly higher rates of SAD (χ 2 [1] = 4.10, p = 0.04) and slightly higher rates of OCD. There was one child and one parent report of PTSD (Table 1). A Chi-Squared test of the separate categorical variables formed from the child and parent reports of anxiety (parent and child anxiety present: 16; parent and child anxiety absent: 29; parent anxiety present and child anxiety absent: 18; parent anxiety absent and child anxiety present: 12) revealed no significant association between the parent and child reports of anxiety (χ 2 [1] = 0.003, p = 0.96).
Comorbid diagnoses
The number of cases of ODD and CD in the baseline data of primary school-age children with ADHD-CT was consistent with the larger sample from which it was taken (ODD = 33%, n = 4; CD = 17%, n = 2). In addition, the 2-year follow-up data also revealed similar numbers of cases of ODD (42%, n = 5) and CD (25%, n = 3). The parent report revealed that case numbers of GAD, SoPh, OCD, and SAD were similar, while the number of cases of SpPh decreased (Table 2). In contrast, the child report revealed a notable decrease in the number of cases of GAD and SAD and similar numbers of cases of SpPh, OCD, and SoPh (Table 2). A paired samples t-test of the baseline and 2-year follow up summed categorical variables formed from the child report of anxiety revealed a significant difference (mean [standard deviation (SD)] difference of −0.42 [0.51]; t [11] = 2.80, p = 0.02). In contrast, a paired samples t-test of the baseline and 2-year follow-up summed categorical variables formed from the parent report of anxiety revealed no significant difference (mean [standard deviation (SD)] difference of: −0.14 [0.53]; t [11] = 1.00, p = 0.34).
Two-year follow up of comorbid parent-reported and child-reported anxiety diagnoses
Discussion
The dimensional measures of anxiety, from parent and child perspectives, were adequately correlated with the categorical anxiety disorder diagnoses, which supported the content validity of these diagnoses.
The ODD rate was within the expected range of 30–50% [3], while the CD rate of 8% was less than that expected. Again, an expected rate of 30%–50% [3] with a minimum rate of 22% [24], have been previously reported. There did not appear to be an obvious methodological reason for this difference, except for the common practice of reporting ODD/CD as a single entity, rather than considering the two diagnoses as separate, albeit related, disorders.
Both the parent and child reports of anxiety disorders confirmed the relatively higher rates of GAD, SAD, SpPh, and SoPh. However, increased rates of OCD were supported by both parent (9%) and child (12%) reports.
These rates were above those previously noted in association with ADHD-CT [3, 26], and approached those reported in specialist anxiety disorders samples [25, 30]. There was no significant association between the parent and child reports of anxiety disorders. In particular, the parent report suggested higher rates of GAD, SpPh and SoPh, while the child report revealed a considerably higher rate of SAD. The parent report of anxiety disorders was different from the previously published rates of comorbid anxiety disorders. A considerably lower rate of OCD and a higher rate of SAD was reported by Biederman et al. [3], and Lahey et al. noted considerably lower rates of all anxiety disorders [26]. Methodological differences, such as the age and the size of this sample, did not seem to account for these differences, while the definition of comorbid anxiety in this study was more careful, using categorical and dimensional measures of both parent and child reports. However, a sampling factor associated with the clinical referral process may account for these differences. No previous equivalent child report data have been published. Taken together, these parent and child data suggest that the anxiety identified by parent and child may have both aspects in common and different features. For example, the elevated child report compared to the parent report of SAD may indicate the child's increased awareness of their emotional state shifts, particularly at times of change in their environmental routine when separated from their parent(s) and/or home, with the subsequent changes in the attendant behavioural, emotional and cognitive cues that each environmental routine represents. In contrast, the child's parent(s) may be less aware of these emotional state shifts, while being more consciously aware of readily observed behavioural patterns associated with SpPh and SoPh.
The rates of ODD/CD in the subsample followed up were equivalent to those of the larger sample, which supported the representativeness of this subsample. Interestingly, the rates did not substantially change over the two-year follow-up period. This initially appears inconsistent with previously published follow-up data that support the increase in rates of ODD/CD associated with ADHD-CT from childhood to adolescence. However, the mean age of the follow-up subsample did not reach beyond ten years, and is therefore still preadolescent. The parent report of anxiety disorder did not decrease over the follow-up period, with rates of SoPh and OCD not changing, as expected [27, 30]. However, SpPh decreased. In contrast, the child report of anxiety disorders was notable by its decrease over the two-year follow-up period, particularly in GAD and SAD, the anxiety disorders with the highest rates; however, SoPh and SpPh did not decrease. These results have to be interpreted cautiously, because of the small numbers involved in this follow-up study and the preadolescent age of the subjects at baseline and at follow up, which prevent meaningful comparison with adolescent followup data. Nevertheless, again these parent and child follow-up data suggest that the anxiety identified by parent and child may have both aspects in common and different features. Hence, the consideration of both the child and the parent reports may aid the determination of clinical treatments and enhance the scope of research reports.