Does the treatment of anxiety in children with Attention-Deficit/Hyperactivity Disorder (ADHD) using cognitive behavioral therapy improve child and family outcomes? Protocol for a randomized controlled trial

A number of studies have demonstrated that children with ADHD are at increased risk of anxiety disorders compared to their peers [3]. For example, in a large Australian study of children with ADHD (N = 389), 64% of children with ADHD aged 5 to 13 years met diagnostic criteria for at least one anxiety disorder, including Social (48%), Generalized (34%), and Separation (32%) Anxiety Disorders [5]. In children with ADHD, anxiety emerges early in development with 25% of 6–8 year old children with ADHD meeting criteria for an anxiety disorder (assessed via diagnostic interview) compared with 8% of children without ADHD [6].

It is clear that anxiety has an independent impact on functioning for children with ADHD. Anxiety in children with ADHD is associated with poorer attentional and executive functioning, [7, 8] social functioning, [9] and more strained family functioning (e.g., poorer parental mental health, less positive parenting behaviors) [10]. There are also established links between internalizing difficulties, including anxiety, and sleep problems in children with ADHD [11‐13]. In a previous study, our group found that children with ADHD who met criteria for multiple anxiety disorders also had poorer quality of life (QoL) (effect size [ES]: − 0.8), behavior (ES: 0.4), and daily functioning (ES: 0.3) compared to children with ADHD alone [5]. Children with ADHD and multiple anxiety disorders also had higher levels of school absenteeism [5].

There appears to be a significant under-identification of anxiety in children with ADHD in clinical practice. In a large study examining pediatric consultations for ADHD (n = 1528), only 8% of children with ADHD were reported to have comorbid anxiety, [14] which is well below estimates of prevalence from previous studies [3]. When identified, anxiety in children with ADHD may be treated with anxiolytic medications [15]. However, psychological therapies may be preferable given the potential side effects of medications, as well as the risk of drug interactions with multiple psychotropic medications [16]. In addition, parents may be reluctant to commence pharmacological treatment for ADHD and associated comorbidities [17].

In the general population, there is strong empirical support for a variety of skills-based strategies and packages (generally referred to as CBT) for the management of pediatric anxiety [18]. A meta-analysis found moderate effects from such interventions over waitlist conditions with anxiety remitting in 59% of children following treatment with CBT compared with 18% in waitlist controls [19]. A recent study of children aged 5 to 18 years (N = 842) with an anxiety disorder, examined whether comorbid ADHD (n = 94) influenced treatment response to CBT [20]. The study found that ADHD did not predict treatment response or rates of remission for anxiety disorders and furthermore there were small improvements in ADHD symptoms for the ADHD group following CBT [20].

Few studies have examined whether CBT is efficacious in treating anxiety in children with ADHD and comorbid anxiety, with most of them being uncontrolled and including small sample sizes. Jarrett and Ollendick [21] reported improvements in anxiety and ADHD symptom severity immediately following a 10-week CBT program for children aged 8–12 years with comorbid ADHD and anxiety (n = 8). Similarly, two other small studies (n = 7–10), found improvements in anxiety following a CBT program for children and adolescents with ADHD [22, 23]. Although Costin et al. [24] reported high levels of parent and child satisfaction following an 8-week CBT program for 7 boys with ADHD, oppositional defiant disorder and anxiety, there was no measurable improvement in anxiety, which is likely due to the high level of comorbidity in this sample.

To our knowledge only one randomized controlled trial (RCT) has examined the treatment of anxiety in children with a primary diagnosis of ADHD using CBT [4]. This small pilot study (n = 12, aged 8–12 years) from our research team aimed to test the acceptability and feasibility of treating anxiety in children with ADHD using CBT (a program known as “Cool Kids”) compared with usual care. The trial had good uptake (67%) and the majority of families attended all 10 intervention sessions, with no drop-outs. At 5-months post-randomization, 3/6 of intervention children were free of an anxiety diagnosis (assessed via diagnostic interview administered by researchers blinded to intervention status) compared with 0/6 of controls. Intervention children had marked improvements in child and family wellbeing by parent- and teacher-report including anxiety symptoms (effect size (ES) = − 0.6), inattention symptoms (ES = − 1.1), QoL (ES = 0.6), behavior (ES = − 0.5) and parent mental health (ES = − 1.0) [4].

This RCT aims to test the efficacy of an existing CBT intervention adapted for children aged 8–12 years with ADHD and comorbid anxiety compared with usual care. Eligible children will be randomized to the intervention group (10 CBT sessions over 12 weeks) versus usual clinical care. We hypothesize that, compared with the usual care group, the intervention group will have the following outcomes at 5 and 12 months post-randomization:

We will also assess the cost-effectiveness of this intervention compared with usual care.

This is a RCT of CBT versus usual clinical care conducted in the state of Victoria, Australia. The trial will be reported in accordance with CONSORT guidelines for RCTs. We have prepared this study protocol adhering to the SPIRIT checklist. This trial has been prospectively registered: Current Controlled Trials ISRCTN59518816. See Fig. 1 for the participant flowchart.

Participants comprise families of children aged 8–12 years with diagnostically-confirmed ADHD (see inclusion criteria) who also meet diagnostic criteria for one or more of Separation, Generalized, and Social Anxiety Disorder.

Participants are recruited from two sources: 1. an existing database of families who had expressed interest in hearing about ADHD research; and 2. paediatric outpatient clinics in public hospitals or private pediatrician rooms in metropolitan and regional Victoria, since ADHD is commonly managed by pediatricians in Australia [14]. Across both recruitment sources families of children aged 8–12 years with ADHD (irrespective of whether anxiety has been identified given that this may be under-recognized) are mailed study letters and information booklets. An ‘opt out’ approach is used for initial contact about the study, whereby parents opt out if they do not wish to be telephoned by the research team to learn more about the study. If parents do not opt out within 2 weeks, the research team then telephones families to explain the study, answer any questions, and ascertain if they are interested in participating. If they are interested in hearing more, the inclusion/exclusion criteria listed below are checked.

Eligible families complete a baseline assessment visit at The Royal Children’s Hospital, their pediatrician’s office, or family home depending on preference. At this visit the parent and child each complete an electronic survey, and the child completes a direct assessment of cognitive functioning. Parents complete a study consent form and are asked to also provide consents for: data collection from their child’s teacher; data linkage access their child’s National Assessment Program-Literacy and Numeracy (NAPLAN) scores (years 3, 5, & 7); and data linkage access to their child’s Medicare/Pharmaceutical Benefits Scheme data (Australian universal subsidized health care insurance scheme) for the economic evaluation. Medicare records include services that qualify for Medicare benefits (e.g., appointments with psychologists and pediatricians) and for which claims have been processed, whereas the Pharmaceutical Benefits Scheme collects information on the prescription medications that participants have filled at pharmacies. Parents are also asked whether they consent for their de-identified data to be used for use in future studies. If parental consent is provided, the school principal is notified and teachers are invited (via email) to complete baseline and follow-up surveys assessing the child’s behavior at school.

To be included in the trial, the child needs to meet Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria 5 [25] criteria for:

The following exclusion criteria are also applied: non-English speaking, major child illness or disability (e.g., intellectual disability), child currently receiving CBT for anxiety, or commenced an anxiolytic/anti-depressant medication within the preceding 6 weeks. Children will be excluded if they have high levels of Autism Spectrum Disorder symptoms as indicated by scores of 15 or higher on the Social Communication Questionnaire [28].

A Cochrane review reported on the efficacy of CBT for children and adolescents with anxiety disorders in the general population and found remission of anxiety diagnoses in 59% of CBT participants (n = 808) immediately post-intervention compared with 18% in controls (n = 542) [19]. Taking a conservative approach, we based our sample size on observing remission in 40% of the intervention group and 20% in the control group (odds ratio of 2.7) at 5 months post-randomization (approximates post-intervention point). In order to provide 80% power to detect this clinically meaningful difference, using a two sided test with alpha = 0.05 and accounting for 20% drop-out, we need to enrol 114 children in each arm (228 total children). This sample size will enable us to detect differences of 0.42 standard deviations (SDs) between the two groups with 80% power in the continuous secondary outcomes at 5 and 12 months (using a two-sided test with alpha = 0.05). Assuming that 10% of families opt out (based on our previous ADHD trials), 47% of those contacted are eligible, and 70% of those eligible consent to participate (based on our pilot study), we need to approach approximately 771 families in order to recruit the required sample of 228 children.

Upon receipt of the completed consent form and completion of baseline parent and child measures, families are randomized to either the CBT or the usual care (control) group. A randomization schedule was generated using block randomization with variable block sizes by an independent statistician in the Clinical Epidemiology and Biostatistics Unit at the Murdoch Children’s Research Institute. Randomization is stratified by child sex to ensure similar proportions of males and females across groups.

Research assistants telephone families randomized to the intervention group to schedule intervention sessions at The Royal Children’s Hospital, their paediatrician’s office, or another community-based setting (e.g., school) depending on the family’s preference. Control children continue to access usual care for ADHD and anxiety from their pediatrician, which may include some simple behavioral management but is unlikely to include formal CBT [29]. All families can access other services, which may include CBT, while enrolled in the study.

Due to the nature of the intervention, it is not possible to blind participants to their allocation. Our primary outcome is a researcher-blinded diagnostic interview and we also conduct direct assessments of cognitive functioning. The comparison to usual care is essential, as we aim to evaluate whether our intervention is more beneficial than care currently received in the community. Given the importance of demonstrating longer-term benefits associated with interventions, we have opted for a usual care comparison group over a waitlist control group. Based on our previous intervention trials and pilot RCT of this intervention we do not foresee any risks or adverse effects of the intervention. Participants will be free to cease the intervention and/or withdraw from the study at any time.

At 5 and 12 months post-randomization, the research team telephones parents to complete a researcher-blinded diagnostic interview [27] and to book a face-to-face assessment visit (same locations as described above). During these 60–90 min visits, parents and children complete surveys and a direct child assessment of cognitive functioning. The research team also e emails follow-up surveys to teachers at both time points.

This project uses REDCap to collect survey and interview data from parents, children and teachers. Study measures are summarized in Table 2. The primary outcome is whether the child meets diagnostic criteria for one or more anxiety disorders (Separation, Social or Generalized Anxiety Disorder) measured by the Anxiety Disorders Interview Schedule for Children V at 5 months post-randomization [27]. This measure has been selected as it is rigorous, maps onto DSM criteria, and takes into account both anxiety symptom severity and impairment. This interview is administered by an assessor unaware of group allocation via telephone with a parent/carer parents [42]. The diagnostic interview is repeated at 12 months post-randomization.

We also assess a range of other outcomes via parent, child and teacher reports (see Table 2). In addition, the baseline parent survey collects family and child demographic and medical data (i.e., medication, diagnosed comorbidities). Service use and treatments accessed are also assessed at each time point. At all time points, we conduct a direct assessment of cognitive functioning (NIH Toolbox Cognition Battery) [39] administered by a researcher.

We use rigorous data management procedures to ensure confidentiality is maintained. The majority of the data is collected online using REDCap and can only be accessed by the research team. Hardcopy data is securely stores in locked cabinets, only accessible by the research team. All hardcopy data is entered into REDCap and subjected to data checking to ensure accuracy of data entry.

Primary analyses will follow the ‘intention to treat’ principle at the child level. To account for loss to follow-up, analysis of the primary outcome (presence of Separation, Social and/or Generalised Anxiety Disorder) and all secondary outcomes will be carried out using mixed effects regression. For each outcome, a single mixed effects model will be fitted incorporating baseline, 5 month and 12 month data, adjusted for gender (randomization stratification factor), using random effects to allow for the repeated measures within an individual. These models will fit separate random effects and error terms at each time point, and a separate fixed effect of treatment at each of the two post-randomization time points. This approach implicitly deals with missing data as it enables all participants with baseline data to be included in the analysis.

For the primary outcome and other binary outcomes, groups will be compared using mixed effects logistic regression, with results presented as odds ratios and their 95% confidence intervals [CI] at 5 and 12 months. For continuous and count data secondary outcomes, groups will be compared using linear or Poisson mixed effects regression, as appropriate, with results presented as mean differences (and 95% CIs) in outcomes at 5 and 12 months. In addition, we will repeat the analysis adjusted for baseline covariates as appropriate: for example, child sex (male, female), child age, medication use (yes/no), comorbidities (yes/no), therapist, and family socioeconomic status). For our primary outcome, whether the child meets criteria for an anxiety disorder (separation, social or generalised anxiety diagnosis) at 5 months, a sensitivity analysis will be conducted using multiple imputation to handle the missing data with analysis via logistic regression applied to the 5 month data only.

We will determine the cost-effectiveness of the Calm Kids intervention compared with usual care ‘within-trial’ from a health care perspective and a broader societal perspective. The collection of data on the Child Health Utilities Index Nine Dimension (CHU-9D) [36, 37] will enable the calculation of quality-adjusted life years (QALYs) that will form a cost-utility analysis. Health state preference values for the CHU-9D will be based on the value set derived by Ratcliffe et al., reflecting Australian adolescents’ preferences [43, 44], which will be combined with the period of the trial to derive QALYs. The cost-utility analysis will be complemented with a cost-effectiveness analysis, using the primary outcome measure of whether the child meets criteria for Separation, Social and/or Generalised Anxiety Disorder at 5 months post-randomization. This approach has been found to be useful to decision-makers [45]. Key costs and their measure (s) include: i) intervention costs: project team financial records; ii) costs accrued outside the intervention: parent-reported resource use questionnaire (baseline, 5 months and 12 months), adapted to reflect services used by children with ADHD (including community-based services, hospitalizations, out-of-pocket costs and parental productivity costs); iii) accurate and detailed health service usage for medical, pathology, diagnostic and pharmaceutical services: linked Medicare/Pharmaceutical Benefits Scheme data. Resource use and unit costs will be combined to derive a total cost for each participant. For both groups, mean values of costs and outcomes will be reported, as well as mean differences between the groups. The analyses will adopt both a health sector costing perspective (all health related costs will be included) as well as a broader partial societal perspective (parental work productivity also included). Generalized linear models will be used to assess mean difference in costs and outcomes in the two groups, using the same approach to confounding as described above. An incremental cost-effectiveness ratio (ICER) will be calculated as the difference in average cost between the groups, divided by the difference in average QALYs. Uncertainty in the data will be dealt with using nonparametric bootstrapping from the distribution of the observed cost/QALY pairs (1000 simulated replications) to determine CIs and presented in a cost-effectiveness plane along with a cost-effectiveness acceptability curve. To explore the robustness of the results, sensitivity analyses will be carried out (e.g., complete case analysis and unit cost variation).