A randomized controlled trial of online versus clinic-based CBT for adolescent anxiety.
Summary (5 min read)
A randomized controlled trial of online versus clinic-based CBT for adolescent anxiety
- It has been well established that anxiety disorders in children can be treated efficaciously with cognitive-behavior therapy (CBT), with average remission rates of between 56% and 67% at post-treatment (Cartwright-Hatton, Roberts, Chitsabesan, Fothergill, & Harrington, 2004; James, Soler, & Weatherall, 2008; Silverman, Pina, & Viswesvaran, 2008 ).
- Clearly, there is a need for alternative modes of treatment delivery that increase the chance that young people and their parents will seek help and participate in treatment.
- Finally, communication applications on the internet, such as email, instant messaging, blogs and chat rooms have become a very important and familiar component of the lives of many adolescents (Subrahmanyam & Lin, 2007) .
- Spence, Holmes, March et al. (2006) compared clinic-based group CBT, the same treatment partially delivered via the internet, and a waitlist control with anxious children.
- Given that both parents and the young person are typically involved in the decision to seek and participate in treatment, it is important that both parties regard online CBT as a credible form of therapy.
Participants
- The majority of adolescents were born in Australia (91%), with the remainder born in Europe, New Zealand, Asia, South Africa and the United States of America.
- In general, parents were relatively well educated with 46% of fathers and 51% of mothers having completed a university education.
- A proportion of adolescents also met criteria for a co-morbid disorder other than anxiety (overall 18%), including depression (2.6%), dysthymic disorder (6%), attention deficit hyperactivity disorder (9.6%), and oppositional defiant disorder (1.7%).
- Of the 288 families who approached the study, 30 did not meet broad inclusion criteria at the telephone screen stage (e.g. outside age range, non-anxiety related difficulties, location, lack of internet access) and 258 were invited to provide informed consent and participate in the more detailed assessment to determine eligibility.
- The remaining 115 families met all inclusion criteria, provided informed consent, and were randomly allocated to an experimental condition.
Anxiety Disorders Interview Schedule for DSM-IV: Child and Parent versions.
- Diagnostic status was assessed using the Anxiety Disorders Interview Schedule -Child and Parent versions (ADIS-C/P: Silverman & Albano, 1996) , administered over the telephone.
- Assessors independent of the study and blind to treatment condition conducted the assessments.
- A combined report was formed by the interviewer to reflect the responses provided by both the adolescent and the parent.
- There was high inter-assessor agreement, with a kappa of 0.94 for the primary diagnosis and a correlation of .92 for the severity ratings between the initial interviewer and second, independent assessor.
- Inter-assessor reliability for the CGAS in the present study was .91 between the initial interviewer and the independent assessor.
Secondary outcome measures (Child-and parent-reported anxiety questionnaires).
- The Spence Children's Anxiety Scale (SCAS: Spence, 1998 Spence, , 1999) ) child (SCAS-C) and parent (SCAS-P) versions were employed to measure level of adolescent anxiety.
- The psychometric properties of both questionnaires have been well established (Achenbach & Rescorla, 2001) .
- Items were summed to form a total score, with higher scores representing greater treatment outcome expectancy and perceived credibility of the program.
Procedure
- Data collection occurred across three sites from 2005-2008: The University of Queensland and Griffith University in Brisbane, Australia, and Macquarie University in Sydney, Australia.
- Adolescents who met broad inclusion criteria, together with at least one of their parents, were then provided with information about the study and asked to provide online, informed consent, to complete a battery of online questionnaires, and to participate in diagnostic interviews.
- Families who met all inclusion criteria following diagnostic assessment and who provided informed consent were randomly assigned to one of three treatment conditions; clinic-based treatment (CLIN), internet-based treatment (NET) or a waitlist control condition (WLC).
- Order of random assignment was established ahead of time through a computer program and implemented by a research team member not involved in the recruitment process.
- This study was conducted in accordance with the University of Queensland, Macquarie University and Griffith University Human Ethics Committees.
Content of the Interventions
- Participants in the internet-based treatment condition completed BRAVE for Teenagers -ONLINE (Spence, Holmes, & Donovan, 2006) .
- A detailed description of the BRAVE interventions has been provided elsewhere (March et al., 2009; Spence et al., 2008) , and therefore the following contains only a brief overview.
- The intervention is based on theoretical and empirical research relating to the psychosocial determinants of child anxiety (Dadds & Roth, 2001; Rapee & Spence, 2004; Silverman & Treffers, 2001) and evidence-based, cognitive-behavioral interventions (Barrett, 1998; Rapee, Wignall, Hudson, & Schniering, 2000; Spence, Donovan, & Brechman-Toussaint, 2000) .
Sessions include explanations of CBT techniques, real-life examples to demonstrate
- How each skill can be implemented and an opportunity for the participant to apply the technique to their own anxiety-provoking situations.
- The content of the intervention is designed to meet the developmental and cognitive level of teenagers, with age-appropriate scenarios, examples, and activities (example situations include school exams, job interviews, dating, and oral presentations).
- The structured nature of CBT makes it relatively straightforward to adapt session and homework content from a clinic modality to internet delivery.
- In addition, personalised, automated computer-generated emails are sent on behalf of the online therapist to congratulate participants for completion of sessions and to provide feedback about responses to quiz tasks.
- Finally, personalized, automated reminder emails are sent if the session is not accessed within two days of becoming available.
Clinic-based treatment (CLIN).
- In the clinic condition, participants received the equivalent program to the online therapy in terms of content, length and number of session activities, but all sessions were conducted face-to-face with a therapist within a clinic setting.
- BRAVE-CLINIC is a manualized individual CBT program for adolescent anxiety (Donovan, Holmes, & Spence, 2006) adapted from the child program evaluated by Spence, Holmes, March et al. (2006) .
- Adolescents and parents received a workbook to use during the sessions and a copy of the relaxation CD.
- The WLC did not have any form of planned contact with the project team during the 12-week period.
- Following this period, WLC participants were offered NET treatment and subsequently ceased to be part of the study.
Power
- Power calculations were based on chi-square tests using diagnostic status as the primary outcome measure.
- The study was powered on the comparison between the two active treatment conditions as this comparison required the most power.
Statistical Analysis
- Primary outcomes for the investigation were clinical diagnoses (based on the ADIS-C/P), clinician severity ratings (CSR), and clinician-rated global functioning (CGAS).
- Analyses for the continuous variables were conducted using an intention-to-treat (ITT) approach, and included all participants who were randomly assigned to conditions.
- Mixed model, repeated measures analyses using SPSS Mixed were used separately to evaluate the continuously scaled outcome measures over time as a function of treatment condition.
- Maximum likelihood estimation was used to determine parameter estimates.
- The analyses first examined differences between the three experimental conditions (WLC, NET and CLIN) from baseline to the 12-week assessment point.
Baseline Comparisons
- Initial differences between the three experimental conditions were examined for baseline demographic and treatment outcome measures.
- No significant differences were evident for demographic variables including age, gender, family income, or parental education.
Impact of Missing Data
- There were no significant differences in baseline levels of primary and secondary outcome variables or demographic characteristics of the child and family between participants who were retained in the study compared to those who were missing at one or more assessment occasions.
- Dummy variables were computed for the primary outcome measures in which the number of occasions in which there was a missing value on that variable was computed.
- There were no significant interactions between missingness and the condition by time interaction, nor with time or condition for either CSR or CGAS ratings.
Completion of Therapy Sessions
- The average number of sessions completed by NET participants at the 12-week assessment point was 7.5 out of 10 for adolescents and 4.48 out of 5 for parents, although only 39% of adolescents and 66% of parents had completed all of their treatment sessions at this point.
- This was not significantly different to that of CLIN families, who completed on average, 8.25 out of 10 sessions for adolescents and 4.43 out of 5 sessions for parents, with 57% of adolescents and 70% of parents completing all treatment sessions by the 12-week assessment.
- Families in both conditions completed further sessions over the next few weeks.
- No significant differences were evident between NET and CLIN adolescents or parents in terms of the mean number of sessions completed by 12-month follow-up.
Outcome: Baseline to 12-week Assessment Effects
- Diagnostic status was determined on the basis of combined parent and adolescent reports on the ADIS-P and -C.
- Diagnostic status was considered both in terms of the percentage of children who no longer met DSM-IV criteria for their primary anxiety disorder and for any anxiety disorder.
- For the completer sample, analyses were conducted with participants for whom diagnostic data was actually available at the 12-week assessment.
- The criterion for diagnosis free required the CSR for the primary diagnosis to be at a subclinical level (3 or lower).
Primary outcome measures -Continuous measures.
- A linear mixed model analysis was conducted to compare the three conditions over time from baseline to the 12-week assessment for primary and secondary continuous outcome measures.
- Results pertaining to fixed effects for intercept and slopes and effect sizes are shown in Table 3 , and indicate that both NET and CLIN conditions decreased in CSR scores significantly more than the WLC from baseline to 12-week assessment.
- When adolescent gender was included in the model, there were no significant interaction effects between gender, condition and time.
- A linear mixed model analysis was then used to examine differences in primary outcome measures across the four time points for the two treatment conditions.
- There were no significant effects for condition or between condition and time for the primary outcome measures.
Outcome: Baseline to 12-week Assessment and 6-and 12-month Follow-ups
- Follow-up data were examined only for the NET and CLIN treatment conditions, as the WLC participants were not available beyond the 12-week assessment point.
- For the purposes of the ITT sample, missing data at 6and 12-month follow-up were replaced with the value from the previous time point at which they provided assessment data.
Clinically significant change at 6-month follow-up. The percentage of adolescents in
- The ITT sample who no longer met DSM-IV criteria for their primary anxiety disorder continued to increase from 12-week assessment to 6-month follow-up, for both NET and CLIN conditions (see Table 1 ), suggesting that diagnostic status was further improved over time.
- Similar improvements were evident for both treatment conditions in the completer sample.
- Participants also continued to improve at 6-month follow-up with respect to the presence of any anxiety disorder (see Table 1 ).
Secondary outcome measures -Continuous measures.
- There were no significant condition, or condition by time interaction effects, indicating that adolescents in both conditions showed significant and equivalent decreases in anxiety symptoms over time.
- Table 5 shows the parameter estimates and effect sizes for the mixed model analysis and demonstrates that there were no significant differences in changes over time between the CLIN and NET conditions for the secondary outcome measures.
Satisfaction with the Program
- Satisfaction data was collected for 40 adolescents and 41 parents in the NET condition and for 36 adolescents and 37 parents in the CLIN condition.
- Overall, the results indicate that adolescents and parents reported moderate to high satisfaction with the treatment received, although parents in the CLIN condition reported slightly higher program satisfaction.
Discussion
- The results of the study demonstrate the significant benefits of online delivery of CBT for the treatment of anxiety disorders amongst adolescents, with outcomes at 12 month followup being similar to those found for clinic-based therapy.
- Furthermore, the online program was regarded as a highly credible approach by both parents and adolescents, at a level equivalent to clinic-based CBT, and produced moderate to high ratings of satisfaction with treatment.
- Overall, the outcomes for both the CLIN and NET groups at 12-week post-baseline tended to be somewhat lower than those reported for clinic-based therapy (Cartwright-Hatton, et al., 2004; James, et al., 2008; Silverman, et al., 2008) .
- It is feasible that both clinic and internet CBT of the length and type of content used in this study may not be sufficient to bring about significant improvements across the broad spectrum of anxiety symptoms in those young people with more complex, comorbid presentations.
- There are, however, some limitations in the design of the study.
Did you find this useful? Give us your feedback
Citations
1,025 citations
Cites background from "A randomized controlled trial of on..."
...Of the 15 studies that mention adapting the design of the intervention to increase adherence, 8 studies do so without any theoretical basis or reference [64-71], 4 studies make the adaptation the focus of their study [72-75], and 2 studies have adapted the design based on a prior study on the same intervention [76, 77]....
[...]
865 citations
Cites result from "A randomized controlled trial of on..."
...However, a study by Spence et al (50) on adolescents is clearly in line with our findings, suggesting equivalence....
[...]
836 citations
787 citations
Cites background from "A randomized controlled trial of on..."
...Most studies have been on adults, but there are also studies on children/adolescents (60,61) and older adults (62)....
[...]
775 citations
Cites background from "A randomized controlled trial of on..."
...In addition, studies needed to target an adult population, excluding a study on clinically anxious adolescents aged 12–18 years and their parent (Spence et al., 2011) and on treating youth depression and anxiety (Sethi, 2013)....
[...]
...In addition, studies needed to target an adult population, excluding a study on clinically anxious adolescents aged 12–18 years and their parent (Spence et al., 2011) and on treating youth depression and anxiety (Sethi, 2013)....
[...]
References
11,458 citations
6,882 citations
3,281 citations
"A randomized controlled trial of on..." refers background or methods in this paper
...The CGAS has demonstrated good inter-rater reliability (r = .84) and test-retest reliability (r = .85) (Dyrborg et al., 2000; Rey, Starling, Wever, Dossetor, & Plapp, 1995; Schaffer et al., 1983)....
[...]
...The blind assessor also completed the Children's Global Assessment Scale (CGAS: Schaffer et al., 1983)....
[...]
...Scores ranging between 81 and 100 on the CGAS represent normal levels of functioning, scores of 61-80 indicate slight disability, moderate disability is represented by scores between 41 and 60, and scores of 1-40 indicate serious disability (Schaffer et al., 1983)....
[...]