Treatment of childhood anxiety disorder in the context of maternal anxiety disorder: a randomised controlled trial and economic analysis

Article history

This issue of Health Technology Assessment contains a project originally commissioned by the MRC but managed by the Efficacy and Mechanism Evaluation Programme. The EME programme was created as part of the National Institute for Health Research (NIHR) and the Medical Research Council (MRC) coordinated strategy for clinical trials. The EME programme is funded by the MRC and NIHR, with contributions from the CSO in Scotland and NISCHR in Wales and the HSC R&D, Public Health Agency in Northern Ireland. It is managed by the NIHR Evaluation, Trials and Studies Coordinating Centre (NETSCC) based at the University of Southampton.

The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from the material published in this report.

Declared competing interests of authors

Cathy Creswell was supported by a Medical Research Council Clinician Scientist Fellowship (G0601874).

Copyright statement

© Queen’s Printer and Controller of HMSO 2015. This work was produced by Creswell et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Scientific background

Anxiety disorders are among the most common psychological disorders in childhood, affecting 2.6–5.2% of children under the age of 12 years. 1,2 These disorders adversely affect children’s functioning in personal, social and academic domains,3,4 raise the risk for disorders in adolescence and adulthood,5 and carry a substantial health and social cost. 6 Following advances in the development of successful cognitive–behavioural therapies (CBTs) for adult anxiety disorders,7 CBT for child anxiety disorders has now been developed. Although there is still some uncertainty over the optimal form of such an intervention, recent systematic reviews of outcome research indicate that the general CBT approach produces significant therapeutic benefit in this patient group, with, on average, 59% of anxious children no longer meeting criteria for their primary anxiety disorder following CBT. 8 However, it is clear from these reviews, and from the individual treatment trials, that the outcome is highly variable, with a significant proportion (40.6%) of patients retaining their anxiety diagnoses following treatment. 8

Rationale for the research

The outcome from CBT for children with anxiety disorders is highly variable. Major factors contributing to this are likely to be the presence of parental anxiety and associated disturbances in how parents respond to their children when they are faced with challenges. Where parental anxiety has been addressed in treatment research,26–29 for several methodological reasons, it has been difficult to assess its contribution to child outcome. Two studies have systematically targeted parental anxiety in the treatment of child anxiety disorders. In one,9 child anxiety outcome was better where therapeutic measures to address parental anxiety symptoms were included, and in the other19 children’s outcomes were not improved. In both cases, as the treatment did not significantly alter levels of parental anxiety, it remains unclear what aspect of the treatment effected the clinical improvement in the children. Similarly, where therapeutic measures to address parenting responses have been included,30 it has not been possible to determine the specific role of such measures in the complex treatment package employed. A controlled trial in which both factors – treatment of parental anxiety and measures to alter parenting responses – are systematically varied, would produce data of both clinical utility and scientific importance. The study was determined on this basis, and there was no patient or public input at this stage.

Although paternal behaviours are likely to contribute to the maintenance of child anxiety disorder, this study focused on mothers for the following reasons: (i) it has been suggested that the parental responses that may promote anxiety among children differ for mothers and fathers;31 (ii) anxiety disorders are more common among women than men32 and, also, more common among mothers of children with anxiety disorders than fathers;17 (iii) mothers are most commonly the primary caregiving parent in the study region and are more likely to attend treatment sessions for their child (e.g. in a recent study in the same region, 98% of parents nominated as primary caregivers in order to attend treatment were mothers33).

Aims

The aim of the trial was to establish the relative clinical effectiveness and cost-effectiveness of treatments of (i) maternal anxiety and (ii) key maternal parenting responses for children with anxiety disorders who have a mother with current anxiety disorder.

Research questions

In a randomised controlled trial (RCT) for child anxiety occurring in the context of maternal anxiety, the principal questions are:

1. Is the impact of CCBT enhanced by first providing CBT to the mother for her own anxiety?

2. Is the impact of CCBT enhanced by the addition of therapeutic measures designed to address potentially anxiogenic maternal parenting responses?

Secondary questions are:

1. Is sustained improvement in child anxiety significantly associated with a reduction in maternal anxiety?

2. Is sustained improvement in child anxiety significantly associated with improvements in maternal modelling, encouragement, overcontrolling/overprotective behaviour and associated cognitions?

Study design

The trial was set up to evaluate the benefit of supplementing individual CCBT with either treatment of maternal anxiety disorder or treatment that targeted maternal responses when interacting with her child, for children with anxiety disorders whose mothers also had a current anxiety disorder. A three-arm trial was conducted in which children received individual CCBT in all three arms, supplemented by either CBT for the maternal anxiety disorder [CCBT + maternal cognitive–behavioural therapy (MCBT)] or a mother–child interaction (MCI) focused intervention. Non-specific interventions were also delivered in all treatment arms in order to balance therapist contact. The main trial was supplemented with an economic evaluation to consider the cost-effectiveness of the CCBT and MCI interventions.

Ethical approval and research governance

Ethical approval for the study was given by Berkshire Research Ethics Committee (07/H0505/156) and the University of Reading Research Ethics Committee (07/48). The trial was registered with the International Standard Randomised Controlled Trial Register under the reference number 19762288.

Participants

Participants were 211 children, aged 7–12 years [mean age 10.22 years, standard deviation (SD) 1.58], with a current anxiety disorder, together with their mothers. As noted above, the study focused on mothers as (i) intergenerational associations for anxiety disorders are most commonly found between mothers and their children;17 (ii) mothers are most commonly the primary caregivers in the study region; and (iii) paternal behaviours may have different associations with childhood anxiety. 34 Participants were all referred to Berkshire Child Anxiety Clinic, run jointly by Berkshire Healthcare NHS Foundation Trust and the University of Reading, by a health or educational professional. Participants were recruited between June 2008 and May 2011, with the last follow-up assessment in February 2013.

Recruitment procedure

The recruitment schedule is shown in Figure 1.

FIGURE 1.

Flow chart outlining recruitment schedule. ASD, autistic spectrum disorder; CAMHS, Child and Adolescent Mental Health Services.

Informed consent

Participants were given a complete description of the study orally and in writing prior to written informed consent being obtained from participating mothers and assent from participating children. As shown in Figure 2, 676 children were referred and assessed for eligibility. A total of 435 families did not meet the inclusion criteria (24 children and 311 mothers because they did not meet criteria for a current anxiety disorder). Assent/consent was not given by 30 families.

FIGURE 2.

Consolidated Standards of Reporting Trials diagram. A3, assessment 3; A4, assessment 4. DNA, did not attend; OCD, obsessive–compulsive disorder; PTSD, post-traumatic stress disorder.

Randomisation, concealment and blinding

Participants were randomised to one of three treatment conditions: (i) CCBT; (ii) CCBT plus CBT for maternal anxiety disorder (CCBT + MCBT); or (iii) CCBT plus treatment focused on the MCI (CCBT + MCI). Each of the three conditions included non-specific therapeutic interventions to balance the treatment arms for therapist contact with both children and mothers. These were non-directive counselling (NDC; for mothers not receiving MCBT, i.e. groups i and iii) and a family health (FH) intervention (for those not receiving MCI, i.e. groups i and ii), see Table 1.

Randomisation was performed externally at the Centre for Statistics in Medicine (University of Oxford, UK) on receipt of anonymised participant information by fax. Patients were randomised with a 1 : 1 : 1 ratio and minimisation was used to ensure balanced allocation across the treatment groups for child age, gender and type of child anxiety disorder, and baseline severity of the child’s and mother’s primary anxiety disorder. The trial manager was informed of randomisation and allocated participants to therapists for treatment. All assessors and coders remained blind to treatment group for the duration of the study.

Treatment group allocation

The order of treatment delivery is shown in Table 1. MCBT/NDC was delivered first, then the CCBT and MCI/FH interventions were delivered in parallel. Each phase of treatment (MCBT/NDC, CCBT, MCI/FH) was delivered by a different therapist.

Data collection and management

Trial data were entered into an International Business Machine Corporation Statistical Package for the Social Sciences database (IBM SPSS, version 17; IBM Corporation, Armonk, NY, USA) and monitoring and tracking information was entered onto a Microsoft Access 2003 database (Microsoft Corporation, Redmond, WA, USA). A range of data validation checks were carried out in Access, SPSS and Stata Software Release 12 (StataCorp LP, College Station, TX, USA) to minimise erroneous or missing data.

Assessments of maternal anxiety disorder and parenting were made before and immediately following the interventions. Assessments of child anxiety disorder status and severity were conducted before and following treatment, as well as at 6 and 12 months after treatment. All assessors were blind to treatment group allocation throughout the trial.

Baseline assessment

Baseline assessment for the trial comprised diagnostic interviews conducted with children and their mothers to ascertain whether or not both the child and his/her mother met diagnostic criteria for a current anxiety disorder. All of the follow-up measures were also administered at baseline. All baseline assessments were conducted between May 2008 and May 2011.

Follow-up

As shown in Table 1, follow-up data collection was scheduled to take place ‘mid-treatment’ [i.e. after the initial maternal intervention, MCBT/NDC (assessment 1B)], then ‘post treatment’ [i.e. after the CCBT and MCI/FH intervention (assessment 2)], and 6 and 12 months from the post-treatment assessment. Diagnostic assessments were conducted to establish whether or not interventions had successfully altered maternal anxiety at the ‘mid-treatment’ (1B) and ‘post-treatment’ (2) assessments. To establish whether or not the interventions had successfully altered maternal responses, observational and parent-reported measures were administered at the ‘post-treatment’ assessment. Child diagnostic and symptom outcomes were assessed at all time points.

All follow-up data were collected between September 2008 and February 2013. A flow chart showing all recruitment and retention is given in Figure 2.

Measures

Sample size

The study was powered to provide 90% power at the 5% (two-sided) significance level to detect a 30% difference in the proportion of children who recovered from their primary anxiety disorder post treatment in the CCBT + MCI or CCBT + MCBT conditions compared with the CCBT condition, with an estimated remission rate for the CCBT group of 40%. 9 Although the effects of the non-specific treatment on child outcomes were not clear, using the 40% remission rate from Cobham and colleagues9 was considered reasonable to account for the effect of CCBT plus any non-specific intervention, given the substantially briefer form of CCBT delivered in the current trial.

A difference of 30% in the proportion of anxiety-free children following completion of the treatment was considered to be the minimum that would be clinically worthwhile taking into account the increased resources required and change to service delivery that would be required if either of these interventions were found to be effective and implemented in practice. The required sample size of 56 children per group was increased to allow for an estimated 20% loss to follow-up. The sample size was estimated as if two independent trials were conducted, with no adjustment for multiple testing, as recommended by Machin and colleagues. 72

Statistical analysis

A comprehensive statistical analysis plan was prepared before embarking on the analysis. All primary and secondary analyses, apart from the per-protocol (PP) sensitivity analyses, were conducted on the intention-to-treat (ITT) population. The primary end points (recovery from primary diagnosis and overall improvement in anxiety (CGI-I ratings) at post treatment and other binary end points were analysed using a modified Poisson regression approach with robust error variance adjusting for the minimisation factors [child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other)], baseline severity of the child’s and the mother’s primary anxiety disorder (ADIS-IV CSR). The modified Poisson regression approach described by Zou73 is an alternative to logistic regression which allows for estimation of risks ratios (RRs) rather than odds ratios. Sensitivity analyses of the primary end points included (i) no adjustment for minimisation criteria; (ii) PP population (this included those participants who had received at least half of the treatment sessions and had data for the post-treatment assessments, with the exception of one mother in the MCBT condition who also received the MCI intervention in error, rather than the FH control; data from this family was also removed for the PP analyses); and (iii) multiple imputation analysis. Missing data for the primary end points were multiply imputed by chained equations methods. 74 All results from sensitivity analyses were very similar to the primary results. Interim analyses were conducted by the trial statistician when 156 participants had been recruited following a request from the funders. The interim results were kept confidential from the trial manager, all assessors, therapists and their supervisors.

Questionnaire scores, maternal behaviours and maternal cognitions were modelled using linear regression models with the change from baseline as the dependent variable, adjusted for baseline score and minimisation factors. There were outliers present in some of the regression models; however, these were reviewed and were not considered to be due to incorrect completion of the questionnaires. Furthermore, their removal did not change the conclusions from the regression.

All analyses were conducted using Stata software.

Patient flow and numbers analysed

Patient flow is shown in Figure 2. The number of available participants for each treatment arm were as follows:

• post treatment: CCBT n = 56 (79%), CCBT + MCBT n = 60 (87%), CCBT + MCI n = 62 (87%)

• 6 months post treatment: CCBT n = 49 (69%), CCBT + MCBT n = 53 (77%), CCBT + MCI n = 51 (72%)

• 12 months post treatment: CCBT n = 43 (61%), CCBT + MCBT n = 50 (70%), CCBT + MCI n = 46 (65%).

Baseline data

Baseline characteristics were well balanced across treatment groups (Table 3).

Manipulation checks: effects of the interventions on maternal anxiety and responses

Manipulation checks were conducted to evaluate whether or not the MCBT and MCI interventions successfully altered maternal anxiety and maternal responses, respectively.

Primary outcomes

Unadjusted analyses: primary end points

As shown in Table 23 and Figure 3, 48% of the children in the CCBT arm were free of their primary diagnosis status at assessment 2 compared with 58% of children in the CCBT + MCBT and 60% of children in the CCBT + MCI arms.

FIGURE 3.

Presence of pre-treatment ADIS-C/P child primary diagnosis.

The unadjusted RR for the effect of CCBT + MCBT versus CCBT on recovery from primary ADIS-C/P diagnosis by assessment 2 was 1.21 (95% CI 0.86 to 1.71; p = 0.280). This was very similar to the unadjusted estimate of the effect of CCBT + MCI versus CCBT, RR 1.24 (95% CI 0.88 to 1.74; p = 0.219).

The unadjusted RR for the effect of CCBT + MCBT versus CCBT on CGI-I by assessment 2 was 1.24 (95% CI 0.99 to 1.57; p = 0.065) and for the effect of CCBT + MCI versus CCBT the RR was 1.18 (95% CI 0.93 to 1.50; p = 0.179). Frequencies are displayed in Table 24 and Figure 4.

FIGURE 4.

Child CGI-I.

Secondary outcomes

Severity of child’s primary Anxiety Disorder Interview Schedule diagnosis at assessment 2

By assessment 2, 59% of the CCBT arm, 73% of the CCBT + MCBT arm and 73% of the CCBT + MCI arm children had seen an improvement of at least 2 points (Figure 5 and Tables 30 and 31).

There were no significant differences between CCBT + MCBT and CCBT or between CCBT + MCI and CCBT (p = 0.101 and 0.118, respectively).

FIGURE 5.

Box plot of change in severity of child’s primary ADIS-C/P diagnosis at assessment 2 by treatment group.

TABLE 30 - Change in severity of child’s pre-treatment ADIS-C/P primary diagnosis at assessment 2, frequency (%)

Treatment	–7	–6	–5	–4	–3	–2	–1	0	1	Total
CCBT	1 (1.8)	9 (16.1)	11 (19.6)	2 (3.6)	4 (7.1)	6 (10.7)	13 (23.2)	9 (16.1)	1 (1.8)	56
CCBT + MCBT	2 (3.3)	14 (23.3)	9 (15.0)	4 (6.7)	4 (6.7)	11 (18.3)	7 (11.7)	8 (13.3)	1 (1.7)	60
CCBT + MCI	2 (3.2)	17 (27.4)	8 (12.9)	4 (6.5)	2 (3.2)	12 (19.4)	14 (22.6)	3 (4.8)	0 (0.0)	62
Total	5	40	28	10	10	29	34	20	2	178

TABLE 31 - Proportion of children with at least a 2-point reduction in severity of their pre-treatment ADIS-C/P primary diagnosis at assessment 2

Chi-squared test, comparison with control group.

Treatment	n	n (%) with a 2 or more point reduction	p-valuea
CCBT	56	33 (58.9)
CCBT + MCBT	60	44 (73.3)	0.101
CCBT + MCI	62	45 (72.6)	0.118

Post-treatment follow-up

Child outcomes at 6-month follow-up

Presence of child’s Anxiety Disorder Interview Schedule primary diagnosis at 6-month follow-up

Table 38 shows the proportion of missing data within each treatment arm for the assessment of the child’s primary diagnosis at 6 months post treatment. The CCBT arm has the highest proportion of missing data (31%).

TABLE 38 - Presence of pre-treatment ADIS-C/P primary diagnosis at 6 months post treatment (including missing data)

Treatment allocation	Missing, n (%)	No diagnosis, n (%)	Diagnosis, n (%)	Total, n (%)
CCBT	22 (30.99)	29 (40.85)	20 (28.17)	71
CCBT + MCBT	16 (23.19)	34 (49.28)	19 (27.54)	69
CCBT + MCI	20 (28.17)	38 (53.52)	13 (18.31)	71
Total	58	101	52	211

Table 39 shows the proportion of children who recovered from their primary ADIS-C/P anxiety diagnosis by 6 months. The CCBT + MCI arm had the highest proportion of recovered children (75%). The CCBT arm had the lowest proportion of recovered children (59%).

TABLE 39 - Presence of pre-treatment ADIS-C/P primary diagnosis at 6 months

Treatment allocation	No diagnosis, n (%)	Diagnosis, n (%)	Total, n (%)
CCBT	29 (59.18)	20 (40.82)	49
CCBT + MCBT	34 (64.15)	19 (35.85)	53
CCBT + MCI	38 (74.51)	13 (25.49)	51
Total	101	52	153

Table 40 shows the results of the adjusted linear regression of the child’s primary diagnosis. The results show no statistically significant difference between CCBT + MCBT or CCBT + MCI treatment arms in comparison with CCBT.

As in all of the analyses reporting RRs in this report, a value > 1 implies a better outcome for the index treatment arm compared with the control (CCBT) arm, if the value is < 1, the outcome is better for the CCBT arm.

TABLE 40 - Analysis of pre-treatment ADIS-C/P primary anxiety diagnosis at 6 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder and baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder.

Analysis conducted using the modified Poisson regression framework with robust error variance.

Parameter		Adjusted RRa	95% CI	p-valueb
Treatment	CCBT	Ref.
	CCBT + MCBT	1.09	0.81 to 1.46	0.566
	CCBT + MCI	1.26	0.97 to 1.64	0.077

Child’s Clinical Global Impression – Improvement at 6 months

The proportion of missing data for child’s CGI-I assessment at 6 months was lower in the CCBT + MCBT arm (23%) and similar in the CCBT (31%) and CCBT + MCI arms (28%) (Table 41).

TABLE 41 - Clinical Global Impression: child – Improvement at 6 months (including missing data)

Treatment allocation	Missing, n (%)	Much/very much improved, n (%)	Not much/very much improved, n (%)	Total, n (%)
CCBT	22 (30.99)	39 (54.93)	10 (14.08)	71
CCBT + MCBT	16 (23.19)	41 (59.42)	12 (17.39)	69
CCBT + MCI	20 (28.17)	45 (63.38)	6 (8.45)	71
Total	58	125	28	211

The proportion of children who were ‘much/very much improved’ at 6 months is shown in Table 42. The CCBT + MCI arm had the highest proportion (88%). CCBT and CCBT + MCBT were similar (80% and 77%, respectively).

TABLE 42 - Clinical Global Impression: child – Improvement at 6 months

Treatment allocation	Much/very much improved, n (%)	Not much/very much improved, n (%)	Total, n (%)
CCBT	39 (79.59)	10 (20.41)	49
CCBT + MCBT	41 (77.36)	12 (22.64)	53
CCBT + MCI	45 (88.24)	6 (11.76)	51
Total	125	28	153

The linear regression results in Table 43 show that there are no statistically significant differences between treatment arms.

TABLE 43 - Analysis of CGI-I at 6 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder and baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder.

Analysis conducted using the modified Poisson regression framework with robust error variance.

Parameter		Adjusted RRa	95% CI	p-valueb
Treatment	CCBT	Ref.
	CCBT + MCBT	0.97	0.79 to 1.19	0.771
	CCBT + MCI	1.16	0.94 to 1.33	0.216

Presence of any Anxiety Disorder Interview Schedule anxiety diagnosis in children at 6 months

As shown in Table 44, the proportion of missing data for any ADIS-C/P anxiety diagnosis was similar in CCBT (31%) and CCBT + MCI (28%), but lower in MCBT (23%) children.

TABLE 44 - Presence of any ADIS-C/P anxiety diagnosis at 6 months (including missing data)

Treatment allocation	Missing, n (%)	No diagnosis, n (%)	Diagnosis, n (%)	Total, n (%)
CCBT	22 (30.99)	23 (32.39)	26 (36.62)	71
MCBT	16 (23.19)	25 (36.23)	28 (40.58)	69
CCBT + MCI	20 (28.17)	24 (33.80)	27 (38.03)	71
Total	58	72	81	211

Table 45 shows the proportion of children recovering from all anxiety diagnoses by 6 months; and this is the same in each group (47%).

TABLE 45 - Presence of any ADIS-C/P anxiety diagnosis at 6 months

Treatment allocation	No diagnosis, n (%)	Diagnosis, n (%)	Total, n (%)
CCBT	23 (46.94)	26 (53.06)	49
CCBT + MCBT	25 (47.17)	28 (52.83)	53
CCBT + MCI	24 (47.06)	27 (52.94)	51
Total	72	81	153

The results of the linear regression of recovery from all ADIS-C/P anxiety diagnoses by 6 months are shown in Table 46; there are no statistically significant differences between treatment groups.

TABLE 46 - Analysis of recovery from any ADIS-C/P anxiety diagnosis at 6 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder and baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder.

Analysis conducted using the modified Poisson regression framework with robust error variance.

Parameter		Adjusted RRa	95% CI	p-valueb
Treatment	CCBT	Ref.
	CCBT + MCBT	1.04	0.70 to 1.53	0.860
	CCBT + MCI	1.04	0.71 to 1.55	0.814

Severity of child’s primary Anxiety Disorder Interview Schedule diagnosis at 6 months

Figure 6 and Table 47 show the distribution of the change scores from baseline to 6 months for the severity rating of the child’s primary diagnosis. As shown in Table 48, the proportion of children with a reduction of 2 or more points is 74% in the CCBT arm, 70% in the CCBT + MCBT arm and 92% in the CCBT + MCI arm. When comparing the CCBT + MCBT arm with CCBT there is no significant difference. However, there is a significant difference between CCBT + MCI and CCBT (p = 0.013).

FIGURE 6.

Box plot of change in severity of child’s pre-treatment ADIS-C/P primary anxiety diagnosis at 6 months.

TABLE 47 - Change in severity of child’s pre-treatment ADIS-C/P primary anxiety diagnosis at 6 months, frequency (%)

Treatment allocation	–7	–6	–5	–4	–3	–2	–1	0	1	Total
CCBT	0 (0.0)	12 (24.5)	10 (20.4)	3 (6.1)	3 (6.1)	8 (16.3)	4 (8.2)	6 (12.2)	3 (6.1)	49
CCBT + MCBT	2 (3.8)	16 (30.2)	10 (18.9)	1 (1.9)	2 (3.8)	6 (11.3)	11 (20.8)	5 (9.4)	0 (0.0)	53
CCBT + MCI	2 (3.9)	18 (35.3)	9 (17.7)	5 (9.8)	2 (3.9)	11 (21.6)	4 (7.8)	0 (0.0)	0 (0.0)	51
Total	4	46	29	9	7	25	19	11	3	153

TABLE 48 - Proportion of children with at least a 2-point reduction in severity of their pre-treatment ADIS-C/P primary anxiety diagnosis at 6 months

Chi-squared test, comparison with control group.

Treatment	n	n (%) with a 2 or more point reduction	p-valuea
CCBT	49	36 (73.5)
CCBT + MCBT	53	37 (69.8)	0.682
CCBT + MCI	51	47 (92.2)	0.013

Analysis of questionnaire data at 6 month follow-up assessments

Analyses of questionnaire results were conducted on the change scores from baseline to the relevant time point. The change scores were more normally distributed. The change scores were modelled using linear regression, adjusted for baseline scores and minimisation factors.

In this section, only patients who are included in the analysis are included in the summary tables.

Adjusted analysis of change in child-reported questionnaire scores at 6 months (intention-to-treat analysis)

Table 49 shows the results for the analysis of the child-reported questionnaire scores, and the change in scores from baseline to 6 months. For all four of the questionnaires there were no significant differences between CCBT + MCBT and CCBT or between CCBT + MCI and CCBT. The adjusted mean scores for each treatment group and every questionnaire are negative, which shows that all groups improved. A summary of the descriptive statistics is given in Appendix 5, Table 136.

TABLE 49 - Adjusted analysis of change in child-reported questionnaire scores at 6 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder, baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder and baseline questionnaire score.

Variable name	Treatment	n	Adjusteda mean change (95% CI)	Adjusteda mean difference (95% CI)	p-value
CAIS-c total score	CCBT	38	–12.45 (–15.06 to –9.84)	Ref.	–
	CCBT + MCBT	43	–11.02 (–13.50 to –8.54)	1.43 (–2.21 to 5.07)	0.4374
		41	–10.75 (–13.29 to –8.21)	1.70 (–1.97 to 5.37)	0.3614
SCAS-c total score	CCBT	41	–17.85 (–22.73 to –12.98)	Ref.	–
	CCBT + MCBT	43	–16.59 (–21.35 to –11.83)	1.26 (–5.56 to 8.08)	0.7146
	CCBT + MCI	41	–17.60 (–22.52 to –12.67)	0.26 (–6.77 to 7.28)	0.9427
SDQ-c conduct subscale	CCBT	42	–0.88 (–1.35 to –0.42)	Ref.	–
	CCBT + MCBT	44	–0.91 (–1.37 to –0.46)	–0.03 (–0.68 to 0.62)	0.9245
	CCBT + MCI	42	–0.81 (–1.28 to –0.34)	0.07 (–0.60 to 0.74)	0.8317
SMFQ-c total score	CCBT	40	–3.81 (–5.33 to –2.29)	Ref.	–
	CCBT + MCBT	44	–3.52 (–4.97 to –2.06)	0.30 (–1.82 to 2.42)	0.7828
	CCBT + MCI	39	–3.97 (–5.52 to –2.41)	–0.16 (–2.36 to 2.04)	0.8872

Adjusted analysis of change in mother-reported questionnaire scores at 6 months (intention-to-treat analysis)

Table 50 shows results of linear regression looking at mother-reported questionnaires about child symptoms/impact. For the SDQ-p conduct subscale a significant difference was seen between the CCBT + MCI and the CCBT arms (p = 0.022), with the CCBT + MCI arm seeing a bigger reduction in score on average. A summary of the descriptive statistics is given in Appendix 5, Table 137.

TABLE 50 - Adjusted analysis of change in mother-reported child symptoms questionnaire scores at 6 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder, baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder and baseline questionnaire score.

Variable name	Treatment	n	Adjusteda mean change (95% CI)	Adjusteda mean difference (95% CI)	p-value
CAIS-p total score	CCBT	35	–9.25 (–12.12 to –6.37)	Ref.	–
	CCBT + MCBT	37	–12.13 (–14.95 to –9.31)	–2.88 (–6.92 to 1.16)	0.1599
	CCBT + MCI	34	–9.69 (–12.68 to –6.70)	–0.44 (–4.61 to 3.73)	0.8348
SCAS-p total score	CCBT	36	–17.44 (–21.03 to –13.84)	Ref.	–
	CCBT + MCBT	41	–16.62 (–20.00 to –13.25)	0.82 (–4.11 to 5.74)	0.7432
	CCBT + MCI	38	–19.17 (–22.72 to –15.61)	–1.73 (–6.86 to 3.40)	0.5050
SDQ-p conduct subscale	CCBT	39	–0.47 (–0.86 to –0.09)	Ref.	–
	CCBT + MCBT	42	–0.99 (–1.36 to –0.62)	–0.51 (–1.05 to 0.02)	0.0600
	CCBT + MCI	41	–1.11 (–1.49 to –0.73)	–0.64 (–1.19 to -0.09)	0.0224
SMFQ-p total score	CCBT	36	–4.25 (–5.82 to –2.67)	Ref.	–
	CCBT + MCBT	38	–5.86 (–7.40 to –4.32)	–1.61 (–3.83 to 0.61)	0.1542
	CCBT + MCI	36	–4.90 (–6.49 to –3.30)	–0.65 (–2.91 to 1.61)	0.5701

Adjusted analysis of change in teacher-reported questionnaire scores at 6 months (intention-to-treat analysis)

The number of respondents was low. As shown in Table 51, no significant differences were seen. A summary of descriptive statistics are given in Appendix 5, Table 138.

TABLE 51 - Adjusted analysis of change in teacher-reported questionnaire scores at 6 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder, baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder and baseline questionnaire score.

Variable name	Treatment	n	Adjusteda mean change (95% CI)	Adjusteda mean difference (95% CI)	p-value
CAS-t total score	CCBT	11	–1.43 (–3.85 to 0.99)	Ref.	–
	CCBT + MCBT	17	–2.92 (–4.91 to –0.94)	–1.50 (–4.65 to 1.66)	0.3432
	CCBT + MCI	23	–1.40 (–3.02 to 0.23)	0.03 (–2.92 to 2.98)	0.9845
SCAS-t total score	CCBT	4	–1.88 (–15.75 to 12.00)	Ref.	–
	CCBT + MCBT	9	–15.07 (–22.85 to –7.29)	–13.19 (–30.33 to 3.94)	0.1227
	CCBT + MCI	15	–10.26 (–15.89 to –4.64)	–8.39 (–24.79 to 8.01)	0.2955
SDQ-t conduct subscale	CCBT	12	0.57 (–0.36 to 1.50)	Ref.	–
	CCBT + MCBT	18	–0.21 (–0.98 to 0.57)	–0.78 (–1.98 to 0.42)	0.1961
	CCBT + MCI	22	0.31 (–0.37 to 1.00)	–0.26 (–1.42 to 0.90)	0.6540

Child outcomes at 12-month follow-up

Presence of child’s Anxiety Disorder Interview Schedule primary diagnosis at 12 months

Table 52 shows that the proportion of missing data for the assessment of the child’s primary diagnosis at 12 months is very similar in the CCBT (39%) and CCBT + MCI (35%) arms, but lower in the CCBT + MCBT arm (28%).

TABLE 52 - Presence of pre-treatment ADIS-C/P primary diagnosis at 12 months (including missing data)

Treatment allocation	Missing, n (%)	No diagnosis, n (%)	Diagnosis, n (%)	Total, n (%)
CCBT	28 (39.44)	31 (43.66)	12 (16.90)	71
CCBT + MCBT	19 (27.54)	30 (43.48)	20 (28.99)	69
CCBT + MCI	25 (35.21)	34 (47.89)	12 (16.90)	71
Total	72	95	44	211

Table 53 shows that recovery from primary diagnosis is very similar in the CCBT (72%) and CCBT + MCI arms (74%), whereas in the CCBT + MCBT group it is lower (60%).

TABLE 53 - Presence of pre-treatment ADIS-C/P primary diagnosis at 12 months

Treatment allocation	No diagnosis, n (%)	Diagnosis, n (%)	Total, n (%)
CCBT	31 (72.09)	12 (27.91)	43
CCBT + MCBT	30 (60.00)	20 (40.00)	50
CCBT + MCI	34 (73.91)	12 (26.09)	46
Total	95	44	139

The results of the linear regression are shown in Table 54. The results show no statistically significant difference between the CCBT + MCBT or CCBT + MCI treatment arms in comparison with CCBT.

TABLE 54 - Analysis of pre-treatment ADIS-C/P primary anxiety diagnosis at 12 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder and baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder.

Analysis conducted using the modified Poisson regression framework with robust error variance.

Parameter		Adjusted RRa	95% CI	p-valueb
Treatment	CCBT	Ref.
	CCBT + MCBT	0.85	0.65 to 1.12	0.257
	CCBT + MCI	1.04	0.82 to 1.30	0.766

Child’s Clinical Global Impression – Improvement at 12 months

The proportion of 12-month CGI-I assessments that were missing is shown in Table 55. The CCBT arm had 39% missing, CCBT + MCI had 35% missing and CCBT + MCBT had 28% missing.

TABLE 55 - Clinical Global Impression: child – Improvement at 12 months (including missing data)

Treatment allocation	Missing, n (%)	Much/very much improved, n (%)	Not much/very much improved, n (%)	Total, n (%)
CCBT	28 (39.44)	33 (46.48)	10 (14.08)	71
CCBT + MCBT	19 (27.54)	39 (56.52)	11 (15.94)	69
CCBT + MCI	25 (35.21)	37 (52.11)	9 (12.68)	71
Total	72	109	30	211

As shown in Table 56, the proportion of children who were ‘much/very much improved’ was similar between treatment groups; between 77% and 80%.

TABLE 56 - Clinical Global Impression: child – Improvement at 12 months

Treatment allocation	Much/very much improved, n (%)	Not much/very much improved, n (%)	Total, n (%)
CCBT	33 (76.74)	10 (23.26)	43
CCBT + MCBT	39 (78.00)	11 (22.00)	50
CCBT + MCI	37 (80.43)	9 (19.57)	46
Total	109	30	139

The results of the linear regression of CGI-I at 12 months are shown in Table 57; there are no statistically significant differences between treatment groups.

TABLE 57 - Analysis of CGI-I at 12 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder and baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder.

Analysis conducted using the modified Poisson regression framework with robust error variance.

Parameter		Adjusted RRa	95% CI	p-valueb
Treatment	CCBT	Ref.
	CCBT + MCBT	1.02	0.82 to 1.27	0.834
	CCBT + MCI	1.05	0.85 to 1.30	0.628

Presence of any Anxiety Disorder Interview Schedule anxiety diagnosis in children at 12 months

As shown in Table 58, the proportion of missing data for all anxiety diagnoses at 12 months was lowest in the CCBT + MCBT arm (28%). In the CCBT + MCI arm it is 35% and in the CCBT arm 39%.

TABLE 58 - Presence of any ADIS-C/P anxiety diagnosis at 12 months (including missing data)

Treatment allocation	Missing, n (%)	No diagnosis, n (%)	Diagnosis, n (%)	Total, n (%)
CCBT	28 (39.44)	23 (32.39)	20 (28.17)	71
CCBT + MCBT	19 (27.54)	23 (33.33)	27 (39.13)	69
CCBT + MCI	25 (35.21)	24 (33.80)	22 (30.99)	71
Total	72	70	69	211

The proportion of recovered children was similar in the CCBT arm (53%) and the CCBT + MCI arm (52%). In the CCBT + MCBT arm the proportion was slightly lower at 46% (Table 59).

TABLE 59 - Presence of any ADIS-C/P anxiety diagnosis at 12 months

Treatment allocation	No diagnosis, n (%)	Diagnosis, n (%)	Total, n (%)
CCBT	23 (53.49)	20 (46.51)	43
CCBT + MCBT	23 (46.00)	27 (54.00)	50
CCBT + MCI	24 (52.17)	22 (47.83)	46
Total	70	69	139

The results of the linear regression of recovery from all ADIS-C/P anxiety diagnoses by 12 months are shown in Table 60. There are no statistically significant differences between treatment groups.

TABLE 60 - Analysis of recovery from any ADIS-C/P anxiety diagnosis at 12 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder and baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder.

Analysis conducted using the modified Poisson regression framework with robust error variance.

Parameter		Adjusted RRa	95% CI	p-valueb
Treatment	CCBT	Ref.
	CCBT + MCBT	0.89	0.61 to 1.31	0.569
	CCBT + MCI	1.01	0.70 to 1.45	0.972

Severity of child’s pre-treatment Anxiety Disorder Interview Schedule primary diagnosis at 12 months

By 12 months 77% of the control group, 74% of the CCBT + MCBT group and 87% of the CCBT + MCI group children had seen an improvement of at least 2 points (Figure 7 and Tables 61 and 62).

There were no significant differences between the CCBT + MCBT and CCBT arms or between the CCBT + MCI and CCBT arms (p = 0.760 and p = 0.210, respectively).

FIGURE 7.

Box plot of change in severity of child’s primary ADIS-C/P anxiety diagnosis at 12 months.

TABLE 61 - Change in severity of child’s pre-treatment ADIS-C/P primary anxiety diagnosis at 12 months, frequency (%)

Treatment allocation	–7	–6	–5	–4	–3	–2	–1	0	1	Total
CCBT	3 (6.98)	10 (23.26)	9 (20.93)	5 (11.63)	2 (4.65)	4 (9.30)	7 (16.28)	1 (2.33)	2 (4.65)	43
CCBT + MCBT	3 (6.00)	10 (20.00)	12 (24.00)	1 (2.00)	3 (6.00)	8 (16.00)	6 (12.00)	7 (14.00)	0 (0.00)	50
CCBT + MCI	1 (2.17)	18 (39.1)	8 (17.39)	4 (8.70)	3 (6.52)	6 (13.04)	6 (13.04)	0 (0.00)	0 (0.00)	46
Total	7	38	29	10	8	18	19	8	2	139

TABLE 62 - Proportion of children with at least a 2-point reduction in severity of their pre-treatment ADIS-C/P primary anxiety diagnosis at 12 months

Chi-squared test, comparison with control group.

Treatment	n	n (%) with 2 or more point reduction	p-valuea
CCBT	43	33 (76.7)
CCBT + MCBT	50	37 (74.0)	0.760
CCBT + MCI	46	40 (87.0)	0.210

Analysis of questionnaire data at 12-month follow-up assessments

Analyses of questionnaire results were conducted on the change scores from baseline to the relevant time point owing to the skewed distribution of the raw scores at each time point. The change scores were more normally distributed. The change scores were modelled using linear regression, adjusted for baseline scores and minimisation factors.

In this section, only patients who are included in the analysis are included in the summary tables.

Adjusted analysis of change in child-reported questionnaire scores at 12 months (intention-to-treat analysis)

As shown in Table 63, no significant differences were seen. A summary of the descriptive statistics is given in Appendix 5, Table 139.

TABLE 63 - Adjusted analysis of change in child-reported questionnaire scores at 12 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder, baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder and baseline questionnaire score.

Variable name	Treatment	n	Adjusteda mean change (95% CI)	Adjusteda mean difference (95% CI)	p-value
CAIS-c total score	CCBT	29	–12.83 (–17.14 to –8.51)	Ref.	–
	CCBT + MCBT	33	–9.25 (–13.41 to –5.09)	3.57 (–2.59 to 9.74)	0.2525
	CCBT + MCI	37	–11.71 (–15.48 to –7.93)	1.12 (–4.59 to 6.83)	0.6978
SCAS-c total score	CCBT	31	–18.11 (–23.18 to –13.03)	Ref.	–
	CCBT + MCBT	34	–18.92 (–23.82 to –14.01)	–0.81 (–7.96 to 6.34)	0.8225
	CCBT + MCI	37	–17.98 (–22.59 to –13.38)	0.12 (–6.73 to 6.97)	0.9718
SDQ-c conduct subscale	CCBT	31	–1.21 (–1.87 to –0.55)	Ref.	–
	CCBT + MCBT	34	–0.94 (–1.58 to –0.30)	0.27 (–0.65 to 1.18)	0.5622
	CCBT + MCI	38	–1.00 (–1.59 to –0.40)	0.21 (–0.68 to 1.10)	0.6373
SMFQ-c total score	CCBT	28	–4.08 (–5.95 to –2.21)	Ref.	–
	CCBT + MCBT	35	–4.20 (–5.91 to –2.49)	–0.12 (–2.71 to 2.48)	0.9284
	CCBT + MCI	37	–2.09 (–3.72 to –0.47)	1.99 (–0.48 to 4.46)	0.1132

Adjusted analysis of change in mother-reported questionnaire scores at 12 months (intention-to-treat analysis)

As shown in Table 64, no significant differences were seen. A summary of descriptive statistics is given in Appendix 5, Table 140.

TABLE 64 - Adjusted analysis of change in mother-reported questionnaire scores at 12 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder, baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder and baseline questionnaire score.

Variable name	Treatment	n	Adjusteda mean change (95% CI)	Adjusteda mean difference (95% CI)	p-value
CAIS-p total score	CCBT	28	–9.50 (–12.44 to –6.56)	Ref.	–
	CCBT + MCBT	27	–12.11 (–15.19 to –9.04)	–2.61 (–6.88 to 1.66)	0.2270
	CCBT + MCI	31	–12.15 (–14.99 to –9.32)	–2.65 (–6.72 to 1.42)	0.1985
SCAS-p total score	CCBT	30	–22.37 (–26.62 to –18.12)	Ref.	–
	CCBT + MCBT	31	–16.36 (–20.58 to –12.15)	6.01 (–0.01 to 12.02)	0.0502
	CCBT + MCI	33	–20.74 (–24.78 to –16.71)	1.62 (–4.27 to 7.51)	0.5850
SDQ-p conduct subscale	CCBT	32	–1.04 (–1.56 to –0.53)	Ref.	–
	CCBT + MCBT	32	–0.89 (–1.40 to –0.38)	0.16 (–0.57 to 0.89)	0.6685
	CCBT + MCI	38	–0.85 (–1.31 to –0.38)	0.20 (–0.50 to 0.89)	0.5755
SMFQ-p total score	CCBT	30	–5.14 (–6.81 to –3.48)	Ref.	–
	CCBT + MCBT	29	–4.54 (–6.25 to –2.82)	0.61 (–1.82 to 3.04)	0.6194
	CCBT + MCI	33	–5.97 (–7.55 to –4.39)	–0.83 (–3.11 to 1.46)	0.4739

Adjusted analysis of change in teacher-reported questionnaire scores at 12 months (intention-to-treat analysis)

As shown in Table 65, no significant differences were seen. A summary of descriptive statistics is given in Appendix 5, Table 141.

Note the small number of observations when looking at these results.

TABLE 65 - Adjusted analysis of change in teacher-reported questionnaire scores at 12 months

Ref., reference category.

Adjusted for child age, child gender, type of child anxiety disorder (GAD, social phobia, SAD, other), baseline severity (ADIS-C/P CSR) of the child’s primary anxiety disorder, baseline severity (ADIS-IV mother self-report) of the mother’s primary anxiety disorder and baseline questionnaire score.

Variable name	Treatment	n	Adjusteda mean change (95% CI)	Adjusteda mean difference (95% CI)	p-value
CAS-t total score	CCBT	9	–2.56 (–4.76 to –0.37)	Ref.	–
	CCBT + MCBT	10	–0.04 (–2.23 to 2.14)	2.52 (–0.62 to 5.66)	0.1099
	CCBT + MCI	12	–0.40 (–2.29 to 1.48)	2.16 (–0.77 to 5.09)	0.1391
SCAS-t total score	CCBT	4	–7.90 (–17.17 to 1.37)	Ref.	–
	CCBT + MCBT	4	–0.04 (–18.30 to 18.22)	7.86 (–12.83 to 28.55)	0.2437
	CCBT + MCI	5	–11.90 (–23.14 to –0.66)	–4.00 (–18.25 to 10.25)	0.3510
SDQ-t conduct subscale	CCBT	9	–0.09 (–1.21 to 1.03)	Ref.	–
	CCBT + MCBT	11	–0.27 (–1.30 to 0.75)	–0.18 (–1.71 to 1.34)	0.8051
	CCBT + MCI	12	0.41 (–0.54 to 1.36)	0.50 (–0.99 to 2.00)	0.4903

Secondary research questions

The extent to which improvement in child anxiety was associated with changes in maternal anxiety and parenting responses is shown in the following tables of correlations (see Tables 66–75). In each case Pearson’s correlation coefficient (r), the p-value and the number of observations are shown. For each of the three child anxiety variables that are used to correlate with other variables, a higher score implies a worse outcome for the child (as we are looking at the change from baseline).

Adverse events

Adverse events were defined as follows:

Adverse or unexpected events resulting in physical or psychological injury that arise from the administration of research procedures or the provision of treatment within the trial. This will include events such as breach of confidentiality, adverse therapeutic interventions, diagnostic error, improper staff behaviour, falls and injury.

There were no adverse events.

Introduction

The aim of the economic evaluation was to assess the cost-effectiveness of the CCBT + MCBT and CCBT + MCI treatment arms compared with the CCBT treatment arm. This design mirrored the study research questions, and hence was equivalent to conducting two separate economic evaluations, that is CCBT + MCBT versus CCBT, and CCBT + MCI versus CCBT. An incremental comparison between CCBT + MCBT and CCBT + MCI was not a research question and the statistical and economic analyses were not powered for this comparison. The primary economic analyses followed an ITT approach and adopted a health service provider perspective. The economic analyses aligned with the primary aim of the trial, namely to evaluate whether or not the CCBT treatment arm could be improved by the addition of (i) treatment of maternal anxiety disorders (MCBT), or (ii) treatment focused on maternal responses (MCI). Secondary economic analyses complemented the results of the primary analyses by measuring additional mother and child resource use of health and personal social services beyond the main costs of the treatment. Broader impacts on other sectors, including education and employment, were also measured. Data were collected on time off school for children and time off work and usual activities for mothers/carers.

Methods

The primary economic analyses for both comparisons consisted of a cost–utility analysis (CUA) conducted from a health service provider perspective. Secondary economic analyses supplemented the primary results by identifying, measuring and valuing resource use impacts from a wider social and personal social service perspective in addition to measuring the impact on the education and employment sectors. Recent methods guidance on the conduct, reporting and presentation of economic evaluations were adhered. 75–77 Costs and outcomes were combined within a CUA framework and presented using incremental cost-effectiveness ratios (ICERs) with uncertainty represented using the cost-effectiveness plane. The results were also reported using net monetary benefit (NMB). Prices were reported in 2011/12 as the base year, adjusted for inflation using Retail Price Index (RPI) 201278 or Hospital and Community Health Service (HCHS) index 2011/1279 as appropriate. All statistical analyses were performed using Stata version 12.1. Statistical significance was set at p-values < 0.05.

Results

Cost–utility analysis

Table 87 and Figures 8–10 report the results of the CUA from a health-care perspective for CCBT + MCBT compared with CCBT. The bootstrapped ICERs for this comparison are shown in the cost-effectiveness plane in Figure 8. With higher mean costs (albeit statistically insignificant) and lower mean utility (albeit statistically insignificant) differences between the groups, Figure 8 reveals that, given current thresholds for commonly accepted levels of cost-effectiveness (£20,000–30,000), CCBT + MCBT is not likely to be a cost-effective alternative to CCBT. The CEAC shown in Figure 9 reveals that the probability that CCBT + MCBT will be cost-effective in comparison with CCBT is < 10%. The NMB curve (see Figure 10) confirms that CCBT + MCBT confers no monetary benefit over CCBT for a broad range of societal willingness-to-pay thresholds and would not be deemed cost-effective given commonly accepted threshold values representing value for money.

TABLE 87 - Cost–utility analysis (health service perspective): ITT approach – CCBT + MCBT vs. CCBT

SE, standard error; WTP, willingness to pay.

CUA results	CCBT + MCBT (n = 69), mean (SE)	CCBT (n = 71), mean (SE)
Cost of intervention	£2124.85 (84.98)	£1891.30 (87.14)
QALY gain	0.794 (0.022)	0.827 (0.024)
Incremental cost (95% CI)	£233.55 (–£6.81 to £473.92)
Incremental QALY gain (95% CI)	–0.033 (–0.101 to 0.035)
ICER, incremental cost per QALY gain	–£7077
(95% CI) bootstrap method	Lower limit, £12,373; upper limit, –£91
(95% CI) Fieller’s method	Lower limit, £10,000; upper limit, £187
NMB for WTP = £20,000	–0.033 × £20,000 – £233.55 = –£893.55
NMB for WTP = £30,000	–0.033 × £30,000 – £233.55 = –£1223.55

FIGURE 8.

Cost-effectiveness plane showing bootstrapped replicates of the ICER: CCBT + MCBT vs. CCBT. LL, lower limit; PE, point estimate; UL, upper limit.

FIGURE 9.

Cost-effectiveness acceptability curve showing the probability that the intervention is cost-effective at different willingness-to-pay thresholds: CCBT + MCBT vs. CCBT.

FIGURE 10.

Net monetary benefit curve and limit curves: CCBT + MCBT vs. CCBT. LL, lower limit; UL, upper limit.

Table 88 and Figures 11–13 present the results of the CUA from a health-care perspective for CCBT + MCI compared with CCBT. The bootstrapped ICERS for this comparison are shown in the cost-effectiveness plane in Figure 11. With higher mean costs (albeit statistically insignificant) and higher mean utility (albeit statistically insignificant) differences between the groups, Figure 11 reveals that, given the distribution of the ICERS, CCBT + MCI is highly likely to be a cost-effective alternative to CCBT. The CEAC shown in Figure 12 reveals that, given current NICE thresholds for accepted levels of willingness to pay for an extra QALY (£20,000–30,000), the probability that CCBT + MCI will be cost-effective in comparison with CCBT is around 75%. The NMB curve (see Figure 13) confirms that CCBT + MCI confers additional monetary benefit over CCBT alone and would be deemed a cost-effective alternative given commonly accepted threshold values representing value for money.

TABLE 88 - Cost–utility analysis (health service perspective): ITT approach – CCBT + MCI vs. CCBT

SE, standard error; WTP, willingness to pay.

CUA results	CCBT + MCI (n = 71), mean (SE)	CCBT (n = 71), mean (SE)
Cost of intervention	£2124.46 (83.06)	£1891.30 (87.14)
QALYs	0.855 (0.018)	0.827 (0.024)
Incremental cost (95% CI)	£233.16 (–£6.81 to £473.92)
Incremental benefit, QALY gain (95% CI)	0.028 (–0.030 to 0.086)
ICER, incremental cost per QALY gain	£8327
(95% CI) bootstrap method	Lower limit, –£173; upper limit, –£11,021
(95% CI) Fieller’s method	Lower limit, –£95; upper limit, –£8881
NMB for WTP = £20,000	0.028 × £20,000 – £233.16 = £326.84
NMB for WTP = £30,000	0.028 × £30,000 – £233.16 = £606.84

FIGURE 11.

Cost-effectiveness plane showing bootstrapped replicates of the ICER: CCBT + MCI vs. CCBT. LL, lower limit; PE, point estimate; UL, upper limit.

FIGURE 12.

Cost-effectiveness acceptability curve showing the probability that the intervention is cost-effective at different willingness-to-pay thresholds: CCBT + MCI vs. CCBT.

FIGURE 13.

Net monetary benefit curve and limit curves: CCBT + MCI vs. CCBT. LL, lower limit; UL, upper limit.

Discussion

The aim of the economic evaluation was to assess the cost-effectiveness of the CCBT + MCBT and CCBT + MCI treatment arms in relation to the CCBT treatment arm from a health service perspective. The economic analyses aligned with the primary aim of the trial, namely to evaluate whether or not CCBT could be improved by the addition of (i) treatment of maternal anxiety disorders (MCBT), or (ii) treatment focused on maternal responses (MCI). Combining the total treatment costs with maternal and child QALYs revealed that, within commonly accepted levels of value for money (i.e. £20,000–30,000 per extra QALY gained), CCBT + MCBT was not likely to be a cost-effective alternative to CCBT. However, combining the total treatment costs with child QALYs revealed that in the comparison of the CCBT + MCI with CCBT treatment arms, the CCBT + MCI treatment arm was highly likely to be a cost-effective alternative to the CCBT arm. A limitation of these analyses, however, was that the resource use and costs of additional health and personal social services, beyond the current treatment costs, were not included in the primary analyses owing to very small sample sizes for these data components. Insufficient statistical power meant that conclusions could not be drawn about the impact of these additional costs on the overall cost-effectiveness. These additional data, however, do provide insights about the type and range of services this group of children and mothers use. Children undergoing treatment for anxiety disorders can be seen to be accessing a broad range of services beyond treatment from GPs, including social workers, psychologists, psychiatrists, education welfare officers, family liaison officers, teachers, paediatricians, audiologists, ophthalmology, accident and emergency (A&E), dieticians, physiotherapists, mental health workers and ‘other services’. Notwithstanding the limited evidence these small sample sizes convey, comparisons of CCBT + MCBT and CCBT + MCI with the CCBT treatment arm reveal increasing costs prevalent in the CCBT arm in the majority of the assessment periods. Furthermore, counting the number of services accessed by each group it can be seen that overall participants in the CCBT arm were accessing approximately one-third more services than those in the CCBT + MCBT and CCBT + MCI arms. In addition, the total costs are always higher in the CCBT arm for both comparisons. If these differences translated into actual differences then this would only increase the likelihood of the CCBT + MCBT and CCBT + MCI arms being more cost-effective than the CCBT arm. It is only by replicating these data collection exercises with larger samples that these results can be confirmed or refuted. On a cautionary note, however, it is important to outline that increased use of services may not represent an inferior quality-of-life outcome owing to the benefits of increased awareness about the health and well-being advantages of accessing additional services.

Broader impacts on other sectors, including impacts on education, employment and impacts on leisure time including time off school for children and time off work and usual activities for their mothers, were also presented but not included in the primary analyses. Further analyses will explore the impact of these effects on a broader societal perspective.

Conclusions

These CUAs have shown that when adopting a health service perspective, only the addition of MCI to standard CCBT is highly likely to represent a cost-effective use of resources for these mother/child pairs within commonly accepted thresholds of cost-effectiveness. Further, analyses reveal that when adopting a health service perspective only the addition of MCBT to standard CCBT is unlikely to be a cost-effective use of resources for these mother/child pairs. However, the latter result should be interpreted with caution because of the high percentage of missing data in some variables which, despite being dealt with using appropriate imputation techniques, may still be viewed as a shortcoming. However, further analysis incorporating the additional health and social care costs has indicated that, depending on the representativeness of these data, there are possible improvements in the cost-effectiveness of both CCBT + MCBT and CCBT + MCI depending on the assumptions made about these costs. Further analysis of the data exploring inclusion of the additional health and personal social service costs and employment and educational impacts using multiple imputations within sensitivity analyses may provide further insight to the cost-effectiveness of these interventions. This economic evaluation provides insight to the broad range of services accessed by this client group, hence it is recommended that future economic evaluations in this area incorporate data collection on this full range of services to fully capture the impact of new interventions.

Summary of findings

Children with anxiety disorders whose mothers are also highly anxious have been shown to have a poorer response to treatment than those whose mothers are not anxious,9,100 yet the impact on children of treating the maternal anxiety has been unclear. Further, the clinical impact on children with anxiety disorders of targeting potentially anxiogenic maternal behaviours has not previously been evaluated systematically. The current study set out to address both these issues in a large RCT.

There was some evidence that the MCBT and MCI treatments were successful in altering the maternal factors which they were designed to address. Specifically, MCBT conferred short-term benefits in maternal recovery from primary diagnoses. However, by the post-treatment assessment, when children in all treatment arms had received CCBT, mothers in all arms had experienced a good level of recovery from their primary disorder and differences between treatment arms were no longer apparent. It is important to note that there were no significant differences between treatment arms on maternal-reported symptoms of anxiety at any time point. For the CCBT + MCI treatment arm there was evidence of change in maternal overprotection and expectations relating to a lack of child coping with challenge. There were no differences between treatment arms in change in other potentially anxiogenic parenting responses (such as expressed anxiety and positive behaviours).

Despite the success in changing some aspects of maternal anxiety and parenting responses, adding neither treatment of maternal anxiety (MCBT) nor treatment of maternal responses (MCI) conferred a significant benefit on children on the basis of the primary child treatment outcomes. Although both adjunct treatments achieved higher child recovery and global improvement rates post treatment than the group in which neither maternal anxiety nor potentially anxiogenic parenting received specific therapeutic attention, the advantages were neither statistically significant nor consistent across treatment arms and outcome measures.

There was some evidence of an advantage for the CCBT + MCI treatment arm on the primary outcomes at the 6-and 12-month follow-up assessments, but this did not reach statistical significance. There was a significant advantage of CCBT + MCI over CCBT on change in child anxiety severity at the 6-month follow-up, and a similar pattern existed at the 12-month follow-up (although it was no longer statistically significant). There was a general lack of significant differences between treatment arms on child-, mother- and teacher-reported anxiety symptom questionnaires. Where statistically significant differences did exist, these were contrary to expectations, with children in the CCBT group reporting a greater reduction in symptoms of anxiety and low mood than children in the CCBT + MCI arm at the post-treatment assessment. In contrast, mothers in the CCBT + MCI arm reported a greater reduction in child conduct problems at the 12-month follow-up assessment than mothers in the CCBT arm.

The secondary research questions considered whether or not improvement in child anxiety was significantly associated with change in (i) maternal anxiety, and (ii) maternal parenting responses. No significant associations were found between change in maternal anxiety and change in child anxiety symptoms, severity or improvement at the mid-treatment, post-treatment and 6-month follow-up assessments. Contrary to expectations, greater change in maternal anxiety symptoms was associated with less change in the severity of the child’s primary anxiety diagnosis. It is important to note that a large number of correlations were conducted to examine this research question and the lack of a consistent pattern of results highlights the fact that no clear conclusions can be drawn. In relation to maternal parenting responses, significant associations were found between change in maternal behaviours and change in child anxiety, most commonly for maternal overprotection. Specifically, and contrary to expectations, a greater increase in overprotection was associated with a greater reduction in the severity of the child’s primary anxiety diagnosis post treatment and global improvement at both the post-treatment and the 6-month follow-up assessments. These indices are both assessor rater; when child symptom ratings were used the opposite pattern was found, with a greater reduction in maternal overprotection being associated with greater reduction in child anxiety symptoms. The difference in the pattern of findings according to who is reporting on child anxiety is interesting and warrants further investigation, but for now clear conclusions about mechanisms of change cannot be drawn.

Economic evaluation

The CUAs demonstrated that, when adopting a health service perspective, the addition of MCI to standard CCBT is highly likely to represent a cost-effective use of resources for these mother/child pairs within commonly accepted thresholds of cost-effectiveness. Further, analyses reveal that when adopting a health service perspective the addition of MCBT to standard CCBT is unlikely to be a cost-effective use of resources for these mother/child pairs. Those cost-effectiveness results should be interpreted with caution owing to the high percentage of missing data on some variables which, despite being dealt with using appropriate imputation techniques, may still be viewed as a shortcoming. Furthermore, analysis incorporating the additional health and social care costs has indicated that, depending on the representativeness of these data, there are possible improvements in the cost-effectiveness of both CCBT + MCBT and CCBT + MCI depending on the assumptions made about these costs. Further analysis of the data exploring inclusion of the additional health and personal social service costs and employment and educational impacts using multiple imputations within sensitivity analyses may provide further insight to the cost-effectiveness of these interventions. The economic evaluation provides insight into the broad range of services accessed by this client group; hence, it is recommended that future economic evaluations in this area incorporate data collection on this full range of services when evaluating new interventions.

Strengths and limitations

The study had several notable strengths, including the referred clinical sample, the use of reliable, blind raters to make assessments of child and maternal anxiety and maternal behaviours before and after treatment, and a design which allowed for isolating the effects of specifically targeting maternal anxiety and parenting responses. A further strength of the study was the inclusion of non-specific interventions designed to balance therapist contact. However, the data collected for health economic analyses indicated that therapist contact did not end up entirely balanced within each phase of treatment. Most notably, more therapist time was spent delivering the eight sessions of MCBT treatment than the eight sessions of NDC that were delivered in the CCBT arm. Similarly, the MCI treatment took more time to deliver than the FH-oriented control. In both cases this may have resulted from the more directive treatment manuals in the MCBT and MCI treatments requiring longer sessions, or from therapist difficulties in maintaining engagement in the NDC and FH treatments so moving through the material more quickly. This suggestion is consistent with the fact that the highest rate of dropout was found during the eight-session NDC phase of treatment. Despite these differences, the overall time and cost of interventions across the entire treatment period was not significantly different across arms, supporting a good balance overall in therapist contact across treatment arms.

The strengths of the study need to be considered in the light of various other limitations. Although we allowed for 20% loss to follow-up, by the 1-year post-treatment assessment retention was down to 61% in the CCBT condition. This limits the conclusions that can be drawn about differences between treatment conditions in the longer term. Although there were no clear baseline differences between completers and those who dropped out, it is of concern that the greatest amount of dropout occurred during the eight-session maternal counselling phase. Therefore, this form of intervention appears not to have been an acceptable treatment approach for some families. This finding presents a challenge for future research; the inclusion of non-specific interventions presents a conservative test of the specific effects of particular interventions but requires the delivery of credible, alternative interventions. Whether longer-term dropouts over-represented those with good or bad treatment outcomes cannot be determined, although the sensitivity analyses that were conducted suggest that this was not the case. Our impression was that, in some cases, those that dropped out from follow-up assessments did so because they felt that their child had made a good recovery and had ‘moved on’ (and so did not want them to have to take part in a long diagnostic assessment). Future studies might benefit from an abridged follow-up assessment which places a minimal burden on participants. The degree to which long-term outcome was also influenced by involvement with help seeking elsewhere is also unclear. Although families agreed not to initiate any other treatment during the course of the intervention, they may have sought help elsewhere during the follow-up period. However, as indicated in the economic analyses (see Appendix 3, Tables 91–111), use of other resources was low across all treatment arms.

The lack of additional health and social care resource use beyond the treatment costs is a limitation for the economic evaluation presented here. Unfortunately there was a large amount of missing data on this measure which precluded its inclusion in these analyses. However, preliminary analysis incorporating the limited data available on the additional health and social care costs has indicated that, depending on the representativeness of these data, there are possible improvements in the cost-effectiveness of both CCBT + MCBT and CCBT + MCI depending on the assumptions made about these costs. Further analysis of the data exploring inclusion of the additional health and personal social service costs and employment and educational impacts using multiple imputations within sensitivity analyses may provide further insight to the cost-effectiveness of these interventions.

Other limitations include the relatively restricted demographic characteristics of the participating group, who were predominantly of non-minority ethnicity and relatively high socioeconomic status (SES), although this was lower than in a comparable trial on the basis of parent education. 19 Our study population also had somewhat higher rates of child- and parent-rated anxiety levels than in similar studies. 19 In terms of marital status, the population was representative of the wider population. 101 We elected to focus on middle childhood (ages 7–12 years) and intervening with mothers as a methodological expediency as it is likely that the nature of parental influences on child anxiety varies with child age102,103 and parent gender;34 however, as a result, the findings cannot be generalised to young children or adolescents or to interventions with fathers or other caregivers. The extent to which characteristics of other parental figures (e.g. paternal anxiety and parenting responses) moderate treatment outcomes warrants further examination. In our trial each phase of treatment was delivered by a different therapist, this meant that there was a very large number of different combinations of therapists (116 combinations) which precluded us from examining therapist effects.

The study also included children and mothers with a broad range of anxiety disorders. There is emerging evidence that generic treatment approaches, of the sort provided in the study, may be more beneficial for some child anxiety disorders than others104 and that particular potentially anxiogenic parental responses may be disorder specific. 64,105 Both of these sets of findings suggest that further work needs to be done which takes account of the precise form of parental and child anxiety, as well as particular forms of parenting responses. The inclusion of mothers with a broad range of disorders also meant that it was most appropriate for a transdiagnostic anxiety treatment to be delivered. Although there is evidence for the effectiveness of this approach (both here and in previous studies),38 it is unclear if disorder-specific treatments would have conferred greater benefits in terms of maternal anxiety. Furthermore, anxiety disorders are commonly comorbid with depression;32 the extent to which maternal depression moderates treatment outcomes in the context of maternal anxiety disorders warrants further attention. Finally, although we recruited a referred clinical population, their experience of treatment will have been very different from routine child mental health services because of the intensive research assessments and the adjunct interventions that were essential for our research purposes. Further insights in to the experiences of patients and therapists participating in this trial would be of value.

Interpretation of results

The findings suggest that adding specific interventions targeting maternal anxiety or the MCIs to individual CBT for children with anxiety disorders in the context of maternal anxiety disorder do not confer significant benefits in terms of child outcomes when compared with individual CBT (supplemented with non-specific therapist support); although some cost–benefit may obtain from the addition of an intervention to target the MCI.

Our findings, in relation to the treatment of parental anxiety disorder, are consistent with those obtained in a recent study by Hudson and colleagues. 19 Their study also specifically focused on children with anxiety disorders who had a parent with an anxiety disorder, and they also failed to find a significant benefit of the addition of parental anxiety management to CCBT. The findings also run contrary to the idea that changing parental responses is necessary for successful treatment of child anxiety disorder in the context of parental anxiety disorder. 23 Notably, however, the CCBT treatment arm in the current study performed considerably better than expected: immediately post treatment, 48% of children were free of their primary anxiety disorder, and 64% were ‘much’ or ‘very much’ improved; a year after treatment, 72% were free of their primary diagnosis and 77% were ‘much’/‘very much’ improved. These outcomes are particularly notable given the relatively brief nature of the CCBT intervention. 106 They are also higher than the 33% of children (with a parent with an anxiety disorder) who were free of their primary anxiety disorder in the study reported by Hudson and colleagues. 19 Indeed, the success rates reported here are similar to those found from more intensive (14-session) CBT for children with anxiety disorders, regardless of parental anxiety disorder status, where 60% and 72% were ‘much’/’very much’ improved at post treatment55 and 6-month follow-up,107 respectively. The lack of consistent differences between groups on clinical outcomes and the failure to find consistent, significant associations between the hypothesised mechanisms of change and child outcomes, may suggest that the association that has been commonly found between maternal anxiety disorder and child treatment outcomes may be the result of some third factor, for example other stressors experienced by the family, or shared associations with particular child or maternal comorbidities. The fact that we obtained higher child outcomes than expected following CCBT suggests that it is possible that this third factor variable was addressed to some extent by the generic support received by all mothers in this trial.

One consideration in making sense of the lack of main effects of the two active adjunct interventions (MCI/MCBT) is statistical power, given the higher than anticipated success rate on the CCBT arm. However, the extent of differences found between our treatment arms was below our a priori criteria for clinical significance (30% more children free of their anxiety diagnoses). The unexpectedly high rate of recovery within the CCBT group in the current study is unlikely to be a function of particular features of our sample, as our study population was a referred sample with systematically confirmed anxiety diagnoses, comparable to other clinic samples in the literature. One possible explanation may lie in the potential added value of the non-specific interventions. These both provided some level of parental support. Although the design used for the current study was appropriate for determining whether or not the MCBT and MCI interventions conferred specific benefits, controlling for therapist contact time using non-specific interventions is a conservative approach and the effects of the non-specific treatments are unclear. The absence of a significant main effect of either CCBT + MCBT or CCBT + MCI also needs to be considered in the light of the degree to which the adjunctive interventions were successful in altering their respective targets. In the one previous study that assessed the impact of adding CBT for parental anxiety disorders to CBT for child anxiety disorders,12 the failure to find differences in child outcomes was attributed to the fact that the parental CBT did not confer a benefit in terms of reducing parental anxiety compared with when only the children received CBT (i.e. 35.5% vs. 32.7% of mothers were free of their primary anxiety disorder, respectively). In the current study, a more intensive CBT intervention was delivered to the mothers and, as predicted, CCBT + MCBT was associated with a significant reduction in the frequency of maternal anxiety disorder compared with when mothers received NDC (58.5% vs. 36.5% diagnosis free). However, by the end of the CCBT treatment phase there were no group differences in maternal anxiety disorder, with all groups showing high rates of recovery from maternal anxiety disorder (52–66%). The fact that a marked reduction in anxiety disorders across all groups was found following CCBT is consistent with recent findings indicating that reduction in child anxiety promotes reduction in parental anxiety. 25

This study is the first of which we are aware to report on observational assessments of parental responses before-and-after treatment for child anxiety disorders. We were able to rigorously evaluate the extent to which the MCI intervention successfully altered maternal responses to her child when confronted with a challenge. These assessments provided evidence that, in terms of a reduction in overprotective behaviours, the MCI treatment was indeed successful. It is notable that this observation-based finding was confirmed by maternal self-report data. The MCI intervention was also associated with change in maternal cognitions associated with confidence in child coping (i.e. reduced predictions regarding child fear and increased predictions regarding child control). Despite these positive benefits of the MCI intervention, no significant benefit to child outcomes was conferred on primary outcome measures (although MCI added to CCBT seemed to represent good value for money). One possible reason for this may be that the changes were not of sufficient magnitude to be of benefit in terms of the clinical outcomes. Another possibility is that the factors that did change are not central to the maintenance of child anxiety. Indeed, although scores on the self-report maternal overprotection scale used have been found to be associated with the development of child anxiety,108 others have found that they do not discriminate clinically anxious children from their non-anxious peers. 66 It will be important to evaluate whether the association between maternal anxiety disorder and child treatment outcomes is in fact mediated by other shared factors, for example other stressors experienced by the family, which might have been addressed to some extent by the generic support received by all mothers in this trial. It is of interest that no specific benefit was apparent for the MCI intervention on the measures of maternal expressed anxiety, intrusiveness or positive behaviours. Although it is, of course, possible that the intervention was ineffective with respect to these dimensions, it is also possible that these findings reflect a lack of sensitivity of the laboratory-based observational tasks. It is also possible that these parental behaviours changed equally across groups in response to improvements in child anxiety. 13

Although there were no significant differences between treatment arms on the primary outcome measures at the post-treatment assessment, an advantage for CCBT + MCI was found on indices of change in anxiety disorders severity and an advantage which approached significance in terms of being free of the primary anxiety diagnosis at the 6-month follow-up assessment. There was also a trend for an advantage of CCBT + MCI over the CCBT arm in terms of the proportion of children who were free of all their anxiety diagnoses. These findings are consistent with the health economic outcomes which suggest that the CCBT + MCI is a cost-effective use of resources in comparison with the CCBT intervention. As the inclusion of two non-specific interventions within the CCBT arm would be expected to reduce its cost-effectiveness, sensitivity analyses were conducted in which zero costs were attributed to the non-specific interventions. This is a conservative test given that the non-specific interventions might be expected to confer some benefit to children and mothers, yet there was still evidence to support the cost-effectiveness of the CCBT + MCI intervention in this context. As noted above, the mechanisms by which CCBT + MCI conferred a cost–benefit advantage remain unclear.

It is notable that there was also a trend for CCBT + MCBT to have an advantage over CCBT in terms of general improvement post treatment. However, differences between arms were not consistent across time points or measures. Thus, although there is a possibility that MCBT helped support the generalisation of benefits in the short term, this speculation received only weak support.

Some unexpected findings should also be noted. In particular, on the basis of child self-reported anxiety symptoms, the CCBT + MCBT group did less well than the CCBT group at the post-treatment assessment. Both the CCBT + MCI and CCBT + MCBT groups also reported less of a reduction in low mood than the CCBT groups. None of these findings were maintained at the later assessment (although CCBT + MCBT had the poorest overall outcomes by the 12-month assessment); however, they are surprising given the content of the child treatment was the same across groups. Whether or not the full course of NDC (eight sessions) received by the mothers in the CCBT group led to some short-term benefit in terms of children’s perceptions of their symptoms remains unclear. There were also some unexpected findings in relation to the secondary research questions, which addressed the extent to which change in maternal anxiety and parenting responses was associated with change in child anxiety. However, as the pattern of findings was not consistent across measures and assessment time points no clear conclusions can be drawn. It is important to note that the sample size for this study was based on providing enough power to assess the primary outcomes. Other outcomes are secondary and, owing to the large number of tests, the results must be interpreted with caution.

Implications for health care

• The novel intervention that focused on modifying maternal parenting responses was associated with some benefit to children and mothers with anxiety disorders, and is likely to be cost-effective (although the latter result needs to be considered with caution because of the high percentage of missing data in the economic analyses). Incorporating effective measures to address maternal cognitions and behaviours when interacting with her child may improve health outcomes for children with anxiety disorders in the context of maternal anxiety disorder.

• We can be confident that supplementing individual CCBT with CBT to target the maternal anxiety disorder is unlikely to confer substantial health benefits and is unlikely to be cost-effective (although the latter result needs to be considered with caution owing to the high percentage of missing data in the economic analyses). Given the intensity of this intervention and its general lack of effectiveness we think it is unlikely that supplementing CCBT with this intervention will improve child outcomes.

Implications for future research

• Given that CCBT alone was sufficient for a good number of patients, it is possible that a benefit of the CCBT + MCI and CCBT + MCBT interventions may be enhanced in particular contexts, for example in the context of particular maternal or child anxiety disorders or high levels of severity. Future research that directly addresses these possibilities is warranted.

• The relatively low level of association between change in parental anxiety and responses and child anxiety may suggest that other factors may account for the modest treatment outcomes typically found among children with anxiety disorders who have mothers with anxiety disorders (such as genetic or broader social/environmental factors). Future research is warranted to address these issues.

• The economic evaluation provides insight as to the broad range of services accessed by this client group, hence it is recommended that future economic evaluations in this area incorporate data collection on this full range of services when evaluating new interventions.

Contribution of authors

Cathy Creswell and Peter J Cooper had overall responsibility for the study.

Cathy Creswell was responsible for the day-to-day operationalisation and of the study and drafting the final report.

Susan Cruddace and Rachel Gitau were responsible for day-to-day management of the research and Lucy Willetts was responsible for clinical management.

Stephen Gerry was responsible for the final statistical analysis of the clinical outcomes, under supervision of Merryn Voysey and Ly-Mee Yu.

Emma McIntosh led on the design of the economic analysis.

The initial statistical analyses were designed by Jill Mollison.

Lynne Murray and Alan Stein were involved in the design, monitoring the trial and interpretation of results.

Rosamund Shafran was responsible for developing and overseeing the MCBT intervention.

Mara Violato led on the analysis and drafting of the economic results in collaboration with Emma McIntosh.

Lucy Willetts and Cathy Creswell were responsible for developing the MCI intervention.

Nicola Williams was responsible for monitoring of randomisation and interim data monitoring.

All authors contributed to drafting the final report.

Ethical approval

The study was approved by the Berkshire Research Ethics Committee (07/H0505/156) and the University of Reading Research Ethics Committee (07/48).

Publications

Data from initial assessments of subgroups of participants included in this trial were included in the following papers:

Creswell C, Apetroaia A, Murray L, Cooper P. Cognitive, affective, and behavioural characteristics of mothers with anxiety disorders in the context of child anxiety disorder. J Abnorm Psychol 2013;122:26.

Clarke K, Cooper P, Creswell C. The Parental Over-protection Scale: associations with child and parental anxiety. J Affect Disord 2013;151:618–24.

Orchard F, Cooper PJ, Creswell C. Interpretation and expectations among mothers of children with anxiety disorders: associations with maternal anxiety disorder. Depress Anxiety 2015;32:99–107.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the MRC, NETSCC, the HTA programme, the EME programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme, the EME programme or the Department of Health.

Public involvement in the conduct of the research

A member of the public was a full member of the TSC from initiation to completion of the trial. This lay member was identified by contacting parents of children that had received treatment for anxiety disorders at the University of Reading/Berkshire Healthcare NHS Foundation Trust/Berkshire Child Anxiety Clinic. The individual appointed was the only parent to express an interest who was available to commit to ongoing participation and so no selection process was needed. The lay member’s contributions to the conduct of the trial included reviewing information sheets for children and parents, providing feedback on the study protocol and providing guidance on strategies for successful recruitment. This proved invaluable, particularly in providing advice on how to best inform potential participants about the trial and recruitment strategies.

Lessons learned

We benefited from the commitment of our one lay representative; however, we were unable to secure a commitment from other potential lay members, and two general practitioners who gave agreement to join the TSC were ultimately unable to attend meetings. We clearly recognise the value of patient and public involvement at all stages of the research process so will include more comprehensive costings to cover the expenses/lost earnings associated with patient and public involvement and will be more explicit in forming patient and public involvement relationships (e.g. through honorary appointments) in future grants.

Health economic logs

(PDF download)

Health economics diary/patient-held resource use diary

(PDF download)