The ReThink study: a 3-arm parallel randomized trial of cognitive bias modification, with and without adherence promotion, for adolescent anxiety disorder: trial design and protocol

This study is approved by the Kaiser Permanente Northwest (KPNW) Internal Review Board (IRB) in Portland, Oregon. This study is a randomized controlled trial (RCT), with participants blinded to the type of CBM they receive (low vs. high ratio) but not blinded regarding receipt of AP contacts. All evaluators are blinded to participant study condition.

Participants will be randomized via computer-programming to one of three arms in a 1:1:1 ratio, stratified by age at recruitment (< age 15 vs. > = age 15) and minority status (white non-Hispanic vs. all others). Assignment will be conducted with computer-generated random block sizes of 6 or 9, masked from study staff.

Arm 1 is a Low-Ratio version of the CBM intervention, nearly indistinguishable from the other CBM version but with a lower proportion of trials that shift attention to the neutral stimuli (details below). Arm 2 is a High-Ratio CBM program, with a higher proportion of trials shifting attention to the neutral stimuli. Arm 3 is the same High-Ratio CBM program but with a component consisting of adherence promoting (AP) telephone calls from a study phone coach, from 1 to 4 total as needed. All CBM programs are self-administered on participants’ home computers.

The primary clinical outcome will be diagnostic remission at 6-month follow-up from all anxiety diagnoses present at baseline, as confirmed by the Anxiety Disorders Interview Schedule (ADIS) (see below). The secondary clinical outcomes will be the PARS, SCARED, attention bias, depression (PHQ-9), psychopathology severity (CGI-S), and psychosocial functioning (CGAS). We will also determine whether there are differences in treatment acceptability (PAQ) and satisfaction (CSQ-8). Another secondary outcome will be cost-effectiveness of the CBM intervention, as measured in incremental cost per anxiety-free-day (AFD) and cost per quality adjusted life year (QALY). Additional analyses will test whether baseline CGI-S, PHQ-9 (depression), severity of cognitive bias (ADP), gender, socio-economic status, and race/ethnicity moderate the effect of the intervention on the primary outcome. Additionally, we will test whether attention bias mediates the effect of the intervention on the 6-month remission rate and anxiety outcomes (PARS and SCARED). Lastly, we will examine whether CBM process variables (i.e., number of trials [dose], acceptable error rates, reaction time) is associated with CBM outcomes.

Youth between the ages 12 to 17 years old will be enrolled from the KPNW healthcare organization, a large, non-profit health maintenance organization in the Pacific Northwest region of the USA. We aim to enroll a total of 498 participants in the study.

Potentially eligible youth will be identified by reviewing the healthcare organization electronic health record (EHR) for anxiety diagnoses (general anxiety disorder [GAD], anxiety disorder unspecified, or social phobia [SP], and/or separation anxiety). These are not qualifying criteria themselves, but a preliminary case-finding stage that identifies youth who are likely study eligible whom we will approach for recruitment. This initial EHR stage is meant to improve case-finding efficiency. Teen and parent participants will be recruited by mail, with a follow-up telephone call. Interested families will complete a brief, initial phone screening interview to determine if they meet preliminary inclusion and exclusion criteria. If preliminary eligibility criteria are met, parents and youth will give consent to participate in the full baseline assessment.

Inclusion criteria for this study include: age between 12 and 17 years old at enrollment; a research anxiety diagnosis confirmed by the ADIS at the baseline interview (see below); youth and parent able to complete assessments in English; youth access to a Windows home computer with Internet access. Exclusion criteria include: youth diagnosis of psychotic disorder; significantly impairing learning disorder or processing problem, such that completion of study procedures would be difficult; ADHD (except if symptoms are stable and controlled by medication for at least one month prior to enrollment); or youth primary complaint of condition other than anxiety. Comorbid diagnoses leading to study exclusion will be confirmed by reviewing the EHR.

We considered whether to require a minimum level of cognitive bias at baseline as an inclusion criterion. Based on the literature at the time of our proposal, higher pre-treatment levels of negative bias were associated with improved outcomes [33], although this question has been little tested in anxious youth. Ultimately, we elected to not require a minimum bias level, for several reasons. First, the literature on this issue is suggestive but not definitive, especially for youth. Second, the effectiveness nature of this trial argues for the broadest inclusion possible with the most representative sample. Third, including all youth regardless of baseline bias provided an opportunity to test this relationship in a very large sample.

After the preliminary screening call, interested and potentially eligible participants will complete a baseline interview where final eligibility is determined. All measures at baseline will be administered to parent and youth participants over the phone, prior to randomization. See Table 1 for the assessment schedule.

Each assessment wave will take between 1 and 2 h each for youth and parents to complete. Youth and parent participants will complete four follow-up assessments: “post-treatment” at the end of the active intervention period one month after enrollment; and 3-month follow-up, 6-month follow-up, and 12-month follow-up. Youth will be asked to complete a single CBM session on their home computer at the time of each follow-up assessment, to measure negative bias rather than to reactivate the intervention. Details on the study design and participant flow are displayed in Fig. 1. Some measures (ADIS, PARS, CSSRS, CGAS, CGI-S/CGI-I, and demographics) will be administered by research evaluation staff over the phone, while the remaining self-report measures will be completed online by the parent and/or youth through a study web portal. In some cases when participants are unable or unwilling to complete the phone interview or online assessment, they may be mailed a survey packet to complete at home instead.

To promote adherence and follow-up completion participants receive Amazon gift certificates for completing baseline and each planned follow-up. Assessment staff made multiple calls to ppts to schedule or follow-up on any missed assessments.

Research evaluation staff will be blinded to CBM allocation and whether they receive AP calls. We have procedures to reduce the chance that research evaluation staff are made aware of subjects’ status (i.e., data sent from the participant’s performance on the CBM program will be sent directly to the programmers’ server and kept separate from the rest of the study data).

Youth in all three arms will download one of two versions of the CBM program, depending on their study condition, and install it on their home computers. Neither youth, parent or assessment staff will know what CBM version is installed; all versions appear indistinguishable to users. Youth are asked to ideally complete three intervention sessions per week over a four-week period (a goal of 12 sessions total). Youth in all three study arms may receive non-research Treatment as Usual (TAU) for anxiety and other conditions, as delivered by the HMO or other healthcare providers. TAU may include CBT, other psychotherapies, and/or medications. We will collect data on receipt of TAU from the HMO’s EHR at the study conclusion, augmented with information about any services the youth may have received outside of the HMO during the follow-up assessments.

CBM procedures. The Low-Ratio CBM condition (Arm 1), and the two High-Ratio CBM conditions (Arms 2 and 3) share a common protocol for the delivery of the intervention. After eligibility is determined from the baseline assessment and the participant is randomized, they will be given a link and unique code to install the internet-based CBM program on their home computer. After registering with the program, the youth will complete a practice session over the phone with a study staff member. The same set of stimuli will be used for all three conditions, making the downloaded program appear identical to all users. Each youth will be instructed to complete three 15-min sessions per week for four weeks (12 sessions total). In each session, youth view 160 trials of a computerized probe detection task (simulated screenshots in Fig. 2). In each trial, two faces are briefly displayed one above the other, from a standardized set of 8 individuals. For all three CBM conditions, 80% of trials consist of neutral and negative (disgusted) face pairs, while the remaining 20% of trials consist of two neutral faces. Following a 500 ms presentation, the images disappear and a letter probe (either “E” or “F”) is displayed in the location where either the top or bottom face had just been. Youth are instructed to decide whether the letter was an “E” or an “F” with a left or right click of a mouse button. Youth do not receive feedback indicating the accuracy of their responses. In the 80% of High Ratio CBM trials (Arms 2, 3) where a neutral and disgusted face are present, the probe (E or F) always replaces the neutral face. In this manner, youth are trained to disengage their attention from threat. In the Low-Ratio condition (Arm 1), the probe replaces the neutral and disgusted faces with equal frequency (50–50%) in the trials containing both types of faces, and thus only minimal threat retraining occurs. The same library of stimuli is used for all training sessions. Images of faces will be used, rather than emotional words, because facial stimuli have been used in previous CBM studies with youths [24, 25, 61] and may control for differences in youths’ reading ability. Moreover, faces are the most common stimuli in CBM, across disorder type [62].

Phone Coach Adherence Promotion (AP) (Arm 3 Only). AP procedures are intended to mimic the effects of the nonspecific, yet likely important, support and instruction provided by research staff when CBM has been traditionally delivered in laboratories. This includes technical assistance with use of the program, support/encouragement, motivational enhancement, and brainstorming solutions to barriers to regular sessions. We hypothesize that the addition of AP will lead to greater participant adherence to the program and therefore better clinical outcomes. In the first 4 weeks of each youth’s enrollment, phone coaches will make between one and four brief, 10–30 min AP telephone calls to youth enrolled in the High-Ratio CBM + AP condition (Arm 3). Coaches may also involve parents as appropriate. Call frequency will be determined by need, primarily based on the participant’s level of adherence to the CBM program. Participants who complete at least 80% of required sessions with a accuracy rate of 80% or higher will be considered adherent to the program. Youth who are most adherent to the protocol may receive only one or two AP calls from their coach, whereas youth with inconsistent or suboptimal use of the CBM program will receive more calls, up to a total of four. Phone coaches will receive automated adherence reports by logging into the server that stores the session data (e.g., number of trials, accuracy, reaction time). Phone coaches will use these reports to determine the need for AP calls, and the type of coaching they will provide (e.g., low accuracy rate and short reaction time might suggest that youth should slow down slightly to be more accurate). Calls do not need to be once weekly, but may be concentrated (e.g., twice weekly for the first few weeks) in order to facilitate participant engagement and maximize exposure to the CBM program. Coaches will record the date, duration, and content addressed in each call.

The phone coaches will provide the explicit rationale and instructions for the CBM intervention [29], as well as the “fear priming” instructions which encourage youth to activate anxiety fears prior to each CBM session to potentiate the effectiveness of CBM skills training [63]. The remaining content of phone coach calls will be individualized to the needs/issues of each participant, but will typically address: technical assistance (difficulty downloading CBM program, registering, creating a user account); encouragement and reinforcement for appropriate use; problem solving to overcome barriers and obstacles to use; using brief motivational enhancement techniques to encourage participants to use the program [30‐32]; and/or coaching and advice to ensure high quality CBM sessions (e.g., conducting sessions at times/locations where participants will not be interrupted). If needed, phone coaches will also provide case-management and referrals for emergent clinical issues not addressed by CBM program (e.g., suicidality), although they will explicitly not themselves provide any psychosocial therapy for anxiety or any other issues.

CBM adherence data is also recorded automatically by the CBM program as part of its delivery across all arms. Number of sessions is calculated as the number of sessions completed from baseline to the 1-month follow-up. A session is defined as at least 160 completed trials. Quality of participation includes: missing data due to non-response, inaccurate responses (pressing random buttons), and probe response latencies outside of the task window (< 200 ms, > 3500 ms). As a check on the adequacy of our overall CBM implementation, we will benchmark the quality of participant performance against the CBM literature, for the total sample and by arm. Number of trainings and participation quality are also secondary outcomes, with the hypothesis that the Adherence Promotion arm (Arm 3) will show superior effects on anxiety due to an expected higher number of CBM sessions.

The study uses a rigorous control condition with a Low-Ratio CBM design, which controls for the effects of time and attention. The youth in all three arms will be asked to complete the same number of sessions over the intervention period, and in each session the youth view the same number of trials. These features bring the rigor of a classic efficacy design and permit confident attributions of outcome effects.

This study will have “effectiveness” elements (such as recruitment via electronic medical records and self-administration of CBM), which will allow us to enroll a sample that is 10 times larger than other published CBM trials. This study will be large enough to have the statistical power to pursue a wide variety of substantive, mechanistic, and economic analyses that smaller lab-based studies could not address. However, the sample will not be so large that we are unable to pursue a detailed evaluation of clinical outcomes.

Although self-administered CBM promises to be an easily disseminated treatment for anxiety, past results have been mixed when CBM has been delivered outside of the lab. It is possible that without any in-person staff support, participants will not adhere to the protocol as closely as if they were in the lab. However, the design of the current study has attempted to address this potential limitation by adding the telephone-based adherence promotion arm to address these potential issues, which will be the first to test the importance of this component. This added component will add additional support provided by research staff to give technical assistance with the use of the program, support/encouragement, motivational enhancement, and brainstorming solutions to barriers. The addition of AP in this trial will attempt to recreate some of the support of in-person lab-based interventions, which we hypothesize will lead to greater clinical effectiveness of the self-administered CBM program.

Another potential limitation is that the control condition is itself an active treatment, albeit of a lower “dose” of negative bias retraining opportunities. While this controls for expectation bias and treatment exposure, it also may prove to be a more a more difficult control condition for the High-Ratio CBM to overcome.

The study is a mixed-blind design, where participants are unaware whether they are being administered the High- or Low-Ratio version of the CBM program. However, with the addition of the AP component in Arm 3, it is impossible for the participants and phone coaches in this arm to be blind to their condition. To address this, we have taken measures to ensure that the research evaluation staff remain blinded to participant condition and complete assessments without knowledge of their assignment, including whether a given participant is receiving AP.

Our planned sample will have approximately 141 participants in each arm at 12-month follow-up, accounting for an expected attrition rate of 10%. A sample this size is sufficient to detect expected clinical differences, but will give us a less than optimal level of power to detect differences in the cost effectiveness analyses which ideally compare hundreds or thousands of cases per condition. Although this is a limitation, the economic evaluation will still provide useful information about the likely effect size related to cost effectiveness for the intervention and this information could help refine the intervention in order to increase its potential for implementation in other health-care settings.

This self-administered CBM intervention overcomes barriers to EBTs for anxiety such as cognitive behavioral therapy and SSRI medication, as well as limitations to previous studies that require participants to attend a centrally located clinic and complete sessions in a lab. The at-home format of this intervention makes it uniquely easy to disseminate, and this study will fill a gap in the literature by comparing the costs and outcomes of delivering CBM both with and without an adherence promotion component. This will provide the information needed to consider whether broad dissemination and implementation of this intervention is feasible.