This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Major depressive disorders are common among adolescents and can impact all aspects of their daily life. Traditional therapies, cognitive behavioral therapy (CBT), and interpersonal psychotherapy (IPT) have been delivered face-to-face. However, Internet-based (online) delivery of these therapies is emerging as an option for adolescents. Internet-based CBT and IPT involve therapeutic content, interaction between the user and the system, and different technological features embedded into the online program (eg, multimedia). Studies of Internet-based CBT and IPT for adolescent depression differ on all three aspects, and variable, positive therapy effects have been reported. A better understanding of the treatment conditions that influence therapy outcomes is important to designing and evaluating these novel therapies.
Our aim was to examine the technological and program delivery features of Internet-based CBT and IPT for adolescent depression and to document their potential relation to treatment outcomes and program use.
We performed a realist synthesis. We started with an extensive search of published and gray literature. We included intervention studies that evaluated Internet-based CBT or IPT for adolescent depression. We included mixed-methods and qualitative studies, theoretical papers, and policy/implementation documents if they included a focus on how Internet-based psychological therapy is proposed to work for adolescents with depression/depressive symptoms. We used the Mixed-Methods Appraisal Tool to assess the methodological quality of studies. We used the Persuasive System Design (PSD) model as a framework for data extraction and analysis to examine how Internet-based CBT and IPT, as technology-based systems, influence the attitudes and behaviors of system users. PSD components described for the therapies were linked to reported outcomes using a cross-case comparison method and thematic synthesis.
We identified 19 Internet-based CBT programs in 59 documents. Of those, 71% (42/59) were of moderate to high quality. The PSD features surface credibility (competent “look and feel”), dialogue support (online program + in-person support), liking and similarity (esthetics and content appeal to adolescent users), the reduction and tunneling of therapeutic content (reducing online content into simple tasks, guiding users), and use of self-monitoring were present in therapies that resulted in improved therapy engagement, satisfaction, and adherence, as well as symptom and functional impairments.
When incorporated into Internet-based CBT for adolescent depression, PSD features may improve adolescent adherence, satisfaction, and depression-related outcomes. Testing of these features using hypothesis-driven dismantling approaches is recommended to advance our understanding of how these features contribute to therapy effectiveness.
In their lifetimes, as many as 1 in every 9 adolescents will meet criteria for major depressive disorder (MDD) [
Cognitive behavioral therapy (CBT) and interpersonal psychotherapy (IPT) are recommended psychological therapies for adolescents with MDD [
While CBT and IPT have been traditionally delivered as face-to-face therapies, Internet-based (online) delivery of these therapies is proposed as a solution to access, availability, and uptake barriers. As technology-based treatments, Internet-based psychological therapies consist of (1) therapeutic content, (2) interaction between the user, their computer, and treatment material on the webpage, and (3) technological features embedded into the program (eg, multimedia, interactive treatment components). Recent systematic reviews of studies of Internet-based CBT and IPT for MDD have shown that Internet-based CBT and IPT therapies for clinical (adolescents diagnosed with MDD) and general (adolescents with subthreshold symptoms, adolescents considered at high risk for MDD) populations result in improvements in depressive symptoms and moderate to high satisfaction with the therapies [
We report on a realist review that we conducted to examine the technological and program delivery features of Internet-based CBT and IPT for adolescent depression and to document their potential relation to treatment outcomes and program use. The realist approach provided a lens to explore two main questions: (1) Under what conditions are Internet-based CBT and IPT for adolescent depression being delivered? and (2) Within these conditions, what are the technological features of Internet-based CBT and IPT programs that may explain outcomes reported in studies?
Our review used realist synthesis philosophy and principles as recommended by Pawson and Tilley [
Many realist reviews begin by identifying a theory or theories to develop a preliminary list of C-M-O configurations. The evidence identified in the review is then used to determine which C-M-O configurations are upheld when reviewing the evidence. In contrast, we identified potential theories for our review using an iterative process during project development—brainstorming within the review team and reviewing literature on human-technology interaction and studies of Internet-based psychological therapies for adolescent depression. Persuasive System Design (PSD) emerged from this process as a key framework for our C-M-O configurations. Being derived from both behavior change models and information technology systems models, PSD provided a comprehensive framework for exploring how systems influence the attitudes and behaviors of system users [
Using the PSD model, we then theorized what persuasive system features were likely to be linked to each outcome (ie, which features were mechanisms) reported for Internet-based CBT and IPT, and what delivery contexts were the most relevant in allowing that to happen. The end result of this discussion was a list of C-M-O configurations that would guide data analysis.
We also manually searched the table of contents in medical informatics journals (ie, Journal of Medical Internet Research, Internet Interventions, Journal of Cybertherapy and Rehabilitation, Journal of Telemedicine and Telecare). Snowball searching was also conducted (ie, reviewing eligible article reference lists) to identify relevant documents that may have been missed in the search process.
The Persuasive Systems Design model.
Category | Persuasive feature | Definition |
Primary task support | Reduction | Reduces complex behavior into simple tasks |
Tunneling | Guides a user through a process or experience | |
Tailoring | Tailors the experience to the potential needs, interests, personality, or use context | |
Personalization | Personalizes content (eg, allows you to customize the interface or populates your name) | |
Self-monitoring | Keeps track of the user’s performance or status towards goal achievement | |
Simulation | Provides simulations to enable the user to observe link between cause and effect | |
Rehearsal | Provides a way for user to rehearse a skill or task | |
Dialogue support | Praise | Offers praise as a form of feedback |
Rewards | Rewards target behaviors | |
Reminders | Reminds the user of their target behavior | |
Suggestion | Offers fitting suggestions | |
Similarity | Reminds the user of themselves in some meaningful way | |
Liking | Is visually attractive for the user | |
Social role | Adopts a social role | |
System credibility support | Trustworthiness | Provide information that is truthful, fair, and unbiased |
Expertise | Provides information showing knowledge, experience, and competence | |
Surface credibility | Has a competent look and feel | |
Real-world feel | Provides information of the actual people behind its content and services | |
Authority | Refers to people in the role of authority | |
Third-party endorsement | Provides endorsements from other sources | |
Verifiability | Provides means to verify the accuracy of program via outside sources | |
Social support | Social learning | Can use the system to observe others performing tasks or behaviors |
Social comparison | Can use the system to compare their performance with the performance of others | |
Normative influence | Leverages normative influence or peer pressure | |
Social facilitation | User is able to discern via the system that others are performing the behavior along with them | |
Cooperation | Leverages drive to cooperate to complete tasks or behaviors | |
Competition | Leverages drive to compete against others in completing a task or action | |
Recognition | Offers public recognition for an individual or group |
Search results from academic databases were downloaded into Endnote (Thomson Reuters, Version 7.2) and then screened for eligibility by 2 trained raters (authors LW, AR). Inclusion criteria were as follows: (1) intervention studies (eg, clinical trials) were eligible for inclusion if they evaluated Internet-based CBT or IPT with adolescents with depression, (2) theoretical papers, mixed-methods and qualitative studies, and policy/implementation documents were eligible if they included a focus on how Internet-based CBT/IPT is proposed to work for adolescents with depression/depressive symptoms, and (3) documents that met Criteria 1 and 2 were eligible if they were published in English language. Any reviews (systematic, meta-analysis, etc) identified during our screening process were appraised to identify intervention studies and other potentially relevant documents.
During Stage 1 screening, the eligibility of a random subset (10 citations) was assessed independently by 2 team members (authors LW, AR), and interrater agreement was assessed within the “substantial” Kappa range (Cohen kappa=.74).
The evidence for each Internet-based therapy was assessed for relevance and rigor by consensus of 2 reviewers (authors LW, AR). These two reviewers also conducted co-coding and debriefing activities periodically during analysis.
Relevance was defined as the level of contribution to the review, and rigor was defined by the methodological quality of a study conducted on the Internet-based therapy (intervention, mixed-methods, and qualitative studies). Relevance was assessed by reviewing the details provided for an Internet-based therapy’s (1) context (eg, user, program features/design components), (2) mechanism(s): hypotheses as to how specific elements of the therapy worked, was proposed to work, or did not work, and (3) outcomes: reasons for therapy effect or lack of effect on specific adolescent outcomes. These details were obtained by reviewing documentation of usability evaluation, therapy/study protocols, and publications related to evaluations (eg, clinical intervention studies evaluating efficacy or effectiveness). Relevance was rated as low/none (no or little information), medium (some information), and high (well-described information).
The methodological quality of evidence (rigor) around each therapy was assessed, where possible, using the Mixed Methods Appraisal Tool (MMAT), an effective and practical quality assessment tool [
NVivo software (QSR International; Version 11) was used to extract data. We extracted data for several aspects of the context of the Internet-based therapy: (1) user context (eg, urban/rural, age group, sociocultural composition, clinical severity), (2) usage context (eg, therapy objectives, adjunct versus stand-alone therapy), and (3) technology context (eg, synchronicity, use of multimedia, software and bandwidth requirements, mobile phone versus desktop). We also extracted information on therapy design (ie, conditions under which an adolescent completed the therapy: sequence, structure, timing; CBT/IPT features). In instances where the description of the therapeutic content was not available in the article, we included information from the article’s citations that described a therapy. We also extracted available information on PSD system features (primary task support, dialogue support, system credibility support, social support), therapy usage (eg, attrition, engagement), as well as clinical (eg, symptom reduction) and therapeutic (eg, therapeutic alliance) outcomes. Where available, information on full or partial C-M-O configurations was also extracted for individual therapies (ie, we sought data on particular therapies that could explain what Mechanism led to an Outcome, under which Contexts).
To promote consistency during data extraction, a coding guide with operational definitions for each code was used. PSD principles were used to guide the coding of mechanisms [
We used a multistep approach to identify and organize information about what contexts of Internet-based CBT/IPT and persuasive system design attributes may contribute to adolescent outcomes. Our initial list of C-M-O configurations was revised based on consensus between team members. Drawing from qualitative synthesis methods, we selected an “index” therapy (CATCH-IT [
As a final step we compiled a framework matrix for each C-M-O configuration in NVivo in order to map “demi-regularities”, semi-predictable patterns of therapy outcomes [
Figure 1 presents a flow diagram outlining the document search and appraisal process. A total of 15,760 unique and potentially eligible documents were reviewed for inclusion in this review. Of these, 59 documents were deemed eligible for inclusion: published studies (n=45), gray literature documents (n=8), and clinical trial protocols (n=6). These documents were published between 2006 and 2016, and they detailed 19 unique Internet-based psychological therapies.
An overview of the structure and delivery features of the 19 Internet-based therapies is provided in
Most therapies (14/19, 74%) were designed to support contact with a health care or teaching professional. However, the nature of this contact varied significantly across therapies ranging from a 10-minute, initial face-to-face motivational interview (CATCH-IT) to a fully synchronous, online chatroom moderated by a trained coach over multiple weeks (Master Your Mood). Three therapies included the option of synchronous computer-mediated communication (ie, chat, text messaging); most therapies relied on email-based communication. ChilledPlus, CATCH-IT, and CURB involved the adolescent’s primary health care provider and adjunct education for parents as part of the therapy. For brief content descriptions of each Internet-based therapy, see
Reported structure and delivery characteristics of Internet-based psychological therapies for adolescent MDD.
Program |
Participants | Program details | ||||||
Target age |
Testing |
Parent |
Time commitment | Contact | Adapted from | |||
Before |
During |
|||||||
Blues Blaster (USA) [ |
11-15 | P | No | Total: 60-90 minutes over 1 wk |
None | None | Face-to-face Coping With Depression | |
CATCH-IT (USA) [ |
14-21 | P | Yes | Total: 660-840 minutes over 7-8 wks |
In-personb | Phone | ||
Chilled Plus (AUS) [ |
12-17 | T | Yes | Total: 600 minutes over 8 weeks |
In-person | Face-to-face Chilled | ||
Cope2thrive (USA) [ |
13-18 | P, T | No | Total: 350 minutes over 7 weeks |
None | Email/ Phone | Face-to-face COPE group | |
CURB (USA) [ |
13-17 | P | Yes | Total: 660-840 minutes over 7-8 weeks |
In-person | Phone | CATCH-IT | |
DEAL (AUS) [ |
18-25 | T | No | Total: 240 minutes over 4 wks |
None | Computerized SHADE | ||
DWD (CAN) [ |
13-18 | P, T | No | Total: unspecified |
None | None | Manualized DWD | |
Feeling Better (CAN) [ |
16-30 | T | No | Total: 120-200 minutes over 6-10 wks | None | Email/ Phone | Telehealth Family Help | |
iRFCBT (UK) [ |
15-22 | P | No | Total: 360 minutes over 6-12 wks |
None | Internet MindReSolve | ||
iTreAD (AUS) [ |
18-30 | T | No | Total: minimum 240 minutes + social networking over 12 months |
None | Email/ Online chat | Includes DEAL as component | |
Master Your Mood (NZ) [ |
16-25 | T | No | Total: 720 minutes over 8 wks |
None | Online chat | Face-to face group Grip op je dip | |
MAYA (Chile) [ |
12-18 | T | No | Total: <20 minutes over <1 wk |
None | None | ||
Mood Mechanic Course (AUS) [ |
18-25 | T | No | Total: 1924 minutes over 8 wks |
None | Email/ phone/ text | Internet UniWellbeing | |
MoodGym (AUS) [ |
12-17 | P, T | No | Total: 150-300 minutes over 2-3 wks |
Nonec | None | ||
MoodHelper (USA) [ |
18-24 | T | No | Total: unspecified |
None | None | Internet for adults ODIN | |
OIPE (USA) [ |
12-17 | T | No | Total: unspecified number of minutes over 12 wks |
In-person | Text | ||
Rebound (AUS) [ |
15-24 | RP | No | Total: unspecified number of minutes over 12 wks |
None | Social network moderation | Internet Horysons for youth psychosis | |
SPARX (AUS) [ |
13-18 | P, T | No | Total: 210 minutes |
Noned | Phone | CD-ROM SPARX | |
Thiswayup (AUS) [ |
12-16 | P | No | Total: 228-263 minutes over 7 wks |
In-person | In-person | Face-to-face CLIMATE Schools |
aTesting context refers to the type of population who received the therapy: P=prevention (ie, recruited participants with subthreshold depression), T=treatment (ie, inclusion criteria stipulated that participant meet threshold for depressive symptomology, risk, or diagnosis), RP=relapse prevention (ie, required participant have had a previous depressive episode)
bContact with primary care provider was either motivational interview or brief advice.
cMoodGym was tested in different implementation contexts; some included no in-person contact and some with in-person contact.
dSPARX was tested in different implementation contexts; some included no in-person contact and some with in-person contact.
Reported therapeutic and persuasive system design features in evaluated Internet-based psychological therapies for adolescent MDD.
All 19 therapies were based on principles of CBT and used essential “ingredients” described in clinical practice guidelines (eg, [
The PSD features that were present in the therapies are also presented in
Details of the level of contribution and quality assessments are provided in
The availability and contribution of evidence related to Internet-based psychological therapies.
Therapy | Level of contributiona | Documentation available for review | Associated MMAT scoresa | ||||
Context | Mechanism | Outcome | Usability | Protocol | Efficacy/Effectiveness | ||
CATCH-IT | High | High | High | [ |
[ |
[ |
4 [ |
MoodGym | High | High | High | None | None | [ |
3 [ |
SPARX | High | High | High | [ |
[ |
[ |
4 [ |
Blues Blaster | High | High | Medium | None | None | [ |
3 [ |
DEAL | High | Medium | High | [ |
[ |
[ |
3 [ |
Master Your Mood | Medium | High | High | None | [ |
[ |
2 [ |
MoodHelper | Low | High | Medium | None | None | [ |
3 [ |
Feeling Better | Medium | Medium | Low | [ |
None | None | 2 [ |
Thiswayup | Medium | Low | Medium | None | None | [ |
2 [ |
Maya | Low | Medium | Medium | [ |
None | [ |
2 [ |
OIPE | Low | Low | Medium | None | None | [ |
2 [ |
Mood Mechanic Course | Low | Low | Medium | None | None | [ |
3 [ |
Rebound | Medium | Medium | Low | [ |
None | [ |
3 [ |
Cope2thrive | Medium | Low | Medium | None | None | [ |
3 [ |
CURB | Medium | Low | None | [ |
None | None | N/A |
iRFCBT | Low | Low | None | None | [ |
Ongoing trial | N/A |
iTreAD | Low | Low | None | None | [ |
Ongoing trial | N/A |
Chilled Plus | Low | Low | None | None | None | Ongoing trial | N/A |
DWD | Low | Low | None | None | None | Open Access | N/A |
aFollowing published guidelines for MMAT scoring, in instances where multiple documents reported on the same data set, a single MMAT score was calculated.
Of the candidate C-M-O configurations initially put forward using the PSD model, five configurations were substantively supported by available evidence (
In this review, we found that
Real-time guidance involved adolescents completing the Internet-based therapy with an individual (doctor, teacher, therapist) in a setting available to support them (or supervise them) while they completed the activities. Results from studies suggested that completion rates increased if the therapy was delivered with real-time guidance in contexts such as schools or connected with primary/secondary care versus having the adolescent complete the therapy on their own (self-guided) [
Studies of MoodGym have compared in-person guidance to self-guidance in a setting of their choice (eg, home) with synchronous support from the virtual guide (eg, avatar). One study found that there was a 10-fold difference between the approaches (favoring the in-person guidance) in terms of the number of online exercises adolescents completed [
Summary of the C-M-O configurations substantively supported by evidence.
C-M-O configuration | C, M, O | Supporting programs | |
1. Computer-mediated dialogue required real-time support and monitoring to optimize therapy adherence. | C: real-time support |
CATCH-IT, MAYA, MoodGym, SPARX, ThisWayUp | |
2. Therapies with surface credibility led to engagement and satisfaction with the therapy. | C: user interface |
Blues Blaster, CATCH-IT, Feeling Better, Master Your Mood, MAYA, SPARX | |
3. Therapies that included liking and similarity features led to engagement and satisfaction with the therapy. | C: user interface |
Blues Blaster, CATCH-IT, CURB, DEAL, Feeling Better, MoodGym, MAYA, SPARX | |
4. Reduction and tunneling of therapy content were necessary for adolescents to complete more of the therapy. | C: user interface |
Blues Blaster, CATCH-IT, CURB, DEAL, Feeling Better, Master Your Mood, MAYA, MoodGym, SPARX | |
5. Self-monitoring was a key PSD component for facilitating symptom improvements among adolescent users with a MDD diagnosis or functional impairments. | C: users with a MDD diagnosis and/or functional impairments |
Blues Blaster, CATCH-IT, DEAL, Feeling Better, Master Your Mood, Mood Helper |
In this review, we found that adolescents assess the
Potential therapy users quickly assess credibility [
Therapies that incorporated the PSD elements of
Across programs, liking related to the program’s appearance. The issue of color palette was consistently identified during usability testing [
Feedback on the experience of the online video game MAYA, which had low satisfaction and engagement, suggested that, “it would be desirable that [the game] portrayed a social context more similar to [the participant’s] reality” [
The use of
Reducing the amount of text, improving navigational instructions, and reducing the length of therapy modules to improve ease of use were associated with higher adherence (eg, [
A defining feature of the therapies that were successful in treating adolescents with severe symptomology at baseline was the use of
Increasing self-awareness of emotions is an important clinical program feature for treating depression as it prepares individuals for changing their cognitions, beliefs, and schemas [
Mood Helper was one of the only pure stand-alone Internet-based therapies in our review [
The second program, Blues Blaster, was designed primarily to teach adolescents how to monitor their mood and to do so using engaging methods [
Advances in technology have allowed for health care programs to connect users to treatments in dynamic ways. The dramatic growth of technologies designed to persuade and motivate represents a significant shift in focus toward end-user computing in health behavior change therapies [
A key argument in favor of developing and offering Internet-based psychological therapies to young people is increased access [
The findings from our review, in terms of surface credibility and liking, suggest that adolescents’ visual experiences lead to esthetic and credibility judgments [
Fogg has argued that persuasive systems will work only if the user has sufficient motivation and the user’s ability is being adequately triggered to perform the new behavior [
Future studies should operationalize each PSD feature, hypothesize its intended effect, and measure its use and effect. As technology and methods of human-technology interaction evolve, this documentation and evidence will provide a valuable roadmap for the depression treatment field. Studies that use a factorial design or fractional factorial design would move the field forward by providing an opportunity to compare intervention groups that include multiple, and different combinations of, persuasive design functionality. For example, there could be conditions within both study arms in which some groups receive tailored feedback and others do not, some include a social networking forum and others do not, and some provide reminders and others do not. In this way, the impact of PSD features can be isolated. It might also be useful to develop hybrid designs that include both standard “randomization” as well as “preference” arms, in order to determine which groups of adolescents might be more attracted to certain PSD features.
The exploration of target population characteristics is also needed to determine how motivational (eg, readiness for change, self-regulatory skills), developmental/age-related, sociocultural, technical competency and modality features (synchronous, ambient, etc), and depression severity differentially interact with program mechanisms and impact adolescent outcomes. For example, research has shown gender differences in the perceived persuasiveness of numerous health intervention components, with females being more receptive to most persuasive behavior change strategies [
This review is the first to use a realist framework for studying Internet-based psychological therapies for adolescent depression. This framework allowed us to consider studies and theories together to understand how the therapies worked. We included numerous therapies identified in the gray literature, which allowed for a comprehensive appraisal of the current evidence base and reduced the risk of publication and selection bias. Previous reviews of Internet-based therapies for this population have focused only on empirical literature, and therefore, have provided limited insight into the complex causal pathways that may underpin therapy effects. Including multiple research designs, while challenging from a data integration standpoint, enabled the analysis to benefit from the strengths of each approach and corroborate findings across divergent contexts and theoretical orientations. From a realist perspective, this diversity has huge explanatory value and can help uncover contexts and conditions not typically captured in meta-analytic or traditional systematic reviews. In addition to offering a more thorough assessment of Internet-based therapies, this review supersedes existing reviews by including substantially more therapies and documenting the body of work around each one. A further strength of the review is the use of gold standard review methods (notably, duplication of screening, quality assessment, and consensus-building with research team members).
The greatest challenge in applying the PSD model is that no explicit heuristics have been defined for it yet, and so nuances between different PSD features are still being mapped [
Another challenge in our review was the lack of information on the nature of adolescent’s interaction with PSD features. For example, current descriptions of dialogue support provided in Internet-based therapies are lacking. While authors described using email, reminders, and options for peer engagement, there was little detail about actual engagement with these features (eg, How often were reminder emails triggered?, How many adolescents elected to publish journal entries to their peers?, How many adolescents spontaneously emailed their assigned coach and how often?). In terms of peer-based dialogue support, some have argued that there is not yet very strong evidence for what type of peer-based social support therapies ought to provide [
Results from our review suggest there is room for improvement in both designing and implementing Internet-based therapies for adolescent depression and in elucidating how persuasive mechanisms are designed and ultimately function. We offer that many of the assumptions that implicitly shape Internet-based therapy development and delivery—adolescents are highly competent technology users, adolescents want to complete programs on their own, the more persuasive design components the better, or that compliance will result in improved outcomes—are vastly under-acknowledged and are based on pervasive assumptions about adolescents, what they prefer, and what they need. Improved engagement of adolescents with MDD in the design and development of future therapies is crucial if we hope to provide effective Internet-based therapies for this population.
Review search strategies.
Brief description of Internet-based psychological therapies for adolescent depression.
cognitive behavioral therapy
context-mechanism-outcome
Interpersonal Psychotherapy
major depressive disorder
mixed method appraisal tool
persuasive system design
This work was supported by a knowledge synthesis grant from the Canadian Institutes of Health Research (CIHR: KRS2014). AN and LH are supported by CIHR New Investigator Salary Awards.
None declared.