Variations of Anonymity, Transparency, and Efficacy in Digital Health Applications

Social Anxiety Disorder (SAD) is one of the most widespread mental illnesses and it has previously been shown that online therapy and online self-help tools such as the one investigated in this study can be successful in reducing not only symptoms of SAD but also symptoms of depression and increase the quality of life of participants. A substantial concern with digital health intervention tools is data security and privacy as many such tools have been found to be unsafe and easy targets for hacker attacks, potentially endangering personal (health) data of the users.

For this research project, an internet-based program targeting social anxiety has been developed that can be used completely anonymously (i.e., without collecting an email address or other information that can or could potentially identify users). It is originally based on the well-established cognitive-behavioral treatment of social anxiety by Clark and Wells from 1995. This new anonymized version will be compared in a partially randomized patient preference trial with an already well-evaluated version in which, as is common in other digital health applications, personal data such as the email address or other information from the users is used, for example, for login. The fully anonymized version has some disadvantages regarding usability and possibly also effectiveness, as no usage history can be stored, users cannot pick up where they left off during a new login, and instead of entering texts in diaries, for example, about anxiety-related thoughts within the program, they need to download a PDF, etc. On the other hand, the anonymized version has advantages regarding the low threshold for use and privacy. People who could benefit from such internet-based programs might be more likely to use them, as there are no data protection risks associated with their use, and a fully anonymized version could immediately be used even without entering an email address. Thus, the reach and impact of evidence-based digital tools for mental problems and disorders such as social anxiety (disorder) could be increased with an anonymous version. However, there are no studies on the effectiveness of digital tools which can be used anonymously and there is no knowledge about the preferences of users (e.g., is it important for users to be able to use an anonymous version, or do they prefer the non-anonymized version with a few practical advantages?).

Study objectives:

1. To investigate the efficacy of a completely anonymous version of the CBT-based program "JOURNeY" and compare it to an already evidence-based non-anonymous version of the program regarding primary social anxiety symptoms, and secondary outcomes such as depressive symptoms, quality of life, usability, adherence to the program, internalized stigma, attitudes on help-seeking, and personality functioning.

2. To investigate patients' preferences and the influence of patients' preferences on outcome.

3. To explore how patient characteristics are associated with participants' preferences.

This study is part of a larger trial conducted with two departments at the University of Bern. One part assesses clinical research questions and the other part is concerned with psychology of digitalization which researches user behavior and concerns regarding privacy in digital health applications.

The study is a single-center partially randomized patient preference trial and for the clinical part, there are two experimental factors (program versions, i.e. anonymous and standard) each evaluated at two levels (randomized vs. preferred).

Recruitment: Participants will be recruited using reports in newspapers, through internet forums, social media (e.g., Facebook), and via a website (https://www.online-therapy.ch/) in German-speaking countries.

Sampling: After checking the inclusion criteria, participants will be randomized with equal probability to one of the two conditions; randomization vs. preference. In the second stage, if a participant is randomized to the choice arm, the participant can choose the preferred treatment; if a participant is randomized to the random arm, the participant is randomized to a program version (anonymous vs. standard) arm. This design is used to estimate the treatment, selection, and preference effect. The random allocation will be unknown to the investigators. In order to keep the sample sizes of the 16 conditions similar, the allocation list will be made using randomly permuted blocks.

A total of 452 participants who exceed pre-defined cut-off scores on social anxiety measures will be recruited for this study.

Assessment procedures: The primary outcome measure is change in symptoms of SAD at post-treatment, i.e., after 8 weeks. Secondary outcomes include sustaining change at follow up (6 months after baseline), changes in depressive symptoms, changes in quality of life, changes in internalized stigma and attitudes towards help-seeking, changes in levels of personality functioning, and negative effects. Furthermore, an exploratory preference analysis will be conducted to gain insights into digital health application users' preferences and potential preference effect.

Treatment: The internet-based self-help program ("JOURNeY") is based on the well-established cognitive-behavioral treatment of social anxiety by Clark and Wells from 1995 and has been proven efficacious in previous studies. The standard version of JOURNeY is accessed through a secure website, with each participant having a password-protected account. It consists of the following four treatment components: (1) psychoeducation (knowledge about SAD and its treatment); (2) cognitive restructuring (challenging dysfunctional social cognitions); (3) attention training (decrease of self-focused attention); and (4) exposure (decrease of avoidance of feared situations and safety behaviors). The anonymous version has the same content but does not require an account and does not collect any personal data of participants.

Data analysis: Primary analyses will be conducted on the Intention-To-Treat (ITT) sample (i.e., all assigned participants) using linear mixed models with repeated measures. In addition, a completers analysis will be carried out, also with linear mixed models with repeated measures.

For the preference analysis, a more exploratory approach will be used. For binary outcome data (e.g. program preference), regression analysis will be conducted. For other data, tests such as t-test, chi-squared tests or other appropriate tests will be used.

Reporting will follow CONSORT E-Health standards.