Adjustment disorder with anxiety (ADA) is the most frequent and best characterized stress-related psychiatric disorder [2]. According to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), a diagnosis of ADA is appropriate with the finding of anxiety occurring within 3 months of a psychosocial stressor or life event (eg, divorce, job loss, serious physical illness) [3] with symptoms generally abating by 6 months after the event. Considered a common disorder ranging from mild to moderate intensity, ADA is a true diagnostic entity. A general medical study showed that the level of anxiety is comparable to that of other anxiety disorders such as generalized anxiety disorder [4].

Patients are generally not treated with a tailored approach but with medication, mainly benzodiazepine monotherapy. Benzodiazepine misuse is a public health problem. Although stress management is considered the most appropriate psychological treatment for ADA, the evidence base for this approach is limited [5] . However, several studies compared cognitive behavioral therapy (CBT) with other treatments for generalized anxiety disorders and found greater improvements in stress and anxiety symptoms from CBT. In a study comparing CBT with relaxation, a posttreatment progression for State-Trait Anxiety Inventory scale (STAI-T trait subscale) scores from 53.04 to 46 and for Penn State Worry Questionnaire (PSWQ) scores from 61 to 51.13 were shown for the group that received CBT [6]. A study comparing psychodynamic psychotherapy with CBT found a greater improvement in the symptoms of stress and anxiety for the CBT intervention (a posttreatment progression of STAI-T trait subscale scores from 58.83 to 43.41 and of PSWQ scores from 63.48 to 49.86 in this group) [7]. In addition, eMental health options are considered uniquely suited for offering early intervention after patients experience stressful life events that can potentially trigger adjustment disorders [8].

It seems legitimate, given the prevalence and the human, economic, and social costs of this pathology linked to stress, to develop preventive and curative measures for ADA. It is not always possible to intervene upstream to reduce the exposure to stress and prevent the occurrence of ADA (primary prevention). Secondary and tertiary prevention measures are therefore useful to limit their consequences.

It is a multicenter, comparative, prospective, unblinded, randomized, controlled study in 3 parallel groups. As it is not possible to mask the different treatment groups, patients were not blinded from their intervention group.

The project was approved by the local ethics committee, Comité de Protection des Personnes Nord Ouest IV (approval number CPP 15/12), as is required for medical intervention research in France. Data processing will be carried out in accordance with the requirements of the Commission Nationale de l’Informatique et des Libertés reference methodology, MR 06001 (Multimedia Appendix 1).

The study is being conducted at 3 university hospitals (Lille, Amiens, Caen) in the northwest area of France as part of the university communities (InterRégional). Patients are referred by their general practitioner to a psychiatric consultation service for psychological treatment of anxiety symptoms in a context of recent stress. To improve recruitment, general practitioners in the 3 areas were informed of the study by a local investigator during continuing medical education, during scientific meetings, and through all types of collaborative contacts between primary care and psychiatry services. We also directly informed the general practitioners who were involved in previous research on ADA and CBT and invited them to refer patients.

During a medical interview, participants will receive oral and written information detailing the progress of the trial and be allowed a period of reflection. Informed consent will be collected from each subject before they enter the study (Multimedia Appendix 2).

Participants are told they are free to leave the study at any time. Participants will be released from the study by the investigator in cases of adverse events such as diagnosis of an illness requiring hospitalization or surgery, beginning a new medication, or changing medication doses. The current research does not involve any risk, with the exception of the possible negative psychological impact of completing the psychological questionnaires.

Each newly referred patient is asked to answer the optional adjustment disorder section of the Mini International Neuropsychiatric Interview (MINI) [46], French version [47], to confirm he or she meets to the ADA criteria according the DSM-5. The MINI is administered during a face-to-face interview by clinical investigators who are trained in psychiatry. See Textbox 1 for selection criteria for study subjects.

Immediately after inclusion and assessment, patients are randomly allocated in a 1:1:1 ratio into 3 groups using a Web-based central randomization: patients using a digital stress management module (group 1), patients following the stress management module guided by a therapist in attendance (group 2), and patients on a waiting list and benefiting from usual care through their attending physician (group 3).

The randomization sequence is provided by an independent statistician (who does not take part in assessing the patients at any point in the study) using computer-generated random numbers with block sizes of 6 and center stratification consistent with the Consolidated Standards of Reporting Trials (CONSORT) [48]. The randomization sequence is implemented in the electronic case report form (eCRF) system to ensure a centralized, real-time randomization procedure.

A document describing the randomization procedure is kept confidential in the Clinical Investigation Centre of Lille University Hospital.

Assessments are conducted for the 3 groups at baseline, 2 months, and 6 months. At baseline, participants complete the self-administered questionnaires for anxiety and depressive symptoms, and care contacts and medications are recorded. At 2 and 6 months, an evaluation of overall satisfaction as well as adverse events is additionally carried out. Self-assessments are collected in the face-to-face sessions with the investigator.

It is the same psychiatrist investigator for each site who generated the random allocation sequence, enrolled the participants, assigned the participants to the interventions, and assessed the participants, so he is not blinded to the intervention group.

The flowchart in Figure 1 summarizes the experimental design.

The program includes 5 weekly sessions lasting 1 hour that patients follow from a program accessible on a computer in our unit.

Participants have minimal contact with a clinician to encourage adherence and engagement with the program. This minimal contact of 5 minutes is performed before and after every session by a trained nurse, who investigates the adverse events and drug dose changes since the last session, answers any questions, discusses the progress of the session, and possibly guides participants in the navigation of the computer program.

To avoid connection problems, a flash memory drive is supplied to the participant at the first session that contains videos, audio files, self-help books, a portfolio in the form of an eGuide, and a log book with the program of exercises to be completed between 2 successive sessions of the program. The patient is encouraged to practice 1 or several daily exercises for 20 minutes each, 5 or 6 days per week (Multimedia Appendix 3).

The program includes 5 weekly sessions lasting 45 minutes or 1 hour with trained clinical psychologists (graduate of a master’s program in cognitive and emotional therapy with a minimum of 1 year of practice in CBT and cognitive behavioral stress management). Information, exercises, and homework assignments are delivered by the therapist without self-help support. At the beginning of each session, the therapist asks the participant about adverse events and changes in drug doses since the last session.

Control group patients received usual care consisting of contact with their general practitioner and nonspecific psychoeducation about stress and anxiety. Medication is prescribed as a stable dose during the period prior to assessment at 2 months.

The Mini International Neuropsychiatric Interview Version 5.0.0 (MINI) is a fully structured diagnostic interview that assesses a major axis for the diagnosis of disorders [46]. Each newly referred patient was asked to answer the optional adjustment disorder section of the MINI from the French version [47]. The MINI was administered by research assistants who were trained to established reliability criteria. Any participant who met diagnostic criteria for a DSM-5 Axis I diagnosis was excluded from the study. There was no major change between DSM-IV and DSM-5 for anxiety and depressive disorders.

The visual analog scale of satisfaction (VAS-satisfaction) measures the overall satisfaction with the program.

The change in the STAI scale between baseline and the 2-month visit was the main criterion. We chose this criterion because it is a self-report measure, and self-report is commonly used in CBT and seems particularly indicated for Internet-based interventions [60]. In addition, the trait of anxiety is the core of a diagnosis of ADA, and the STAI-T measures the trait of anxiety, which is distinguished from a state of anxiety. The HADS scale and the Hamilton Anxiety Rating scale do not make this distinction.

In a study on the clinical and psychometric characteristics of ADA, the mean (standard deviation) of the STAI baseline score was 52.1 (SD 14.6) [4]. Two clinical studies with patients with generalized anxiety disorder, 1 comparing CBT with relaxation [61] and the other comparing CBT with brief psychodynamic psychotherapy [62], showed changes in the STAI-T scores of 7 and 16, respectively. In view of these results, we hypothesize that an average difference between each experimental arm and the control arm of 11.5 (the mean of 7 and 16). To demonstrate this difference in variation with a standard deviation of the STAI score change of 14.6 (assuming a conservative correlation coefficient of 0.5 between the baseline and 2-month STAI measures), a 2.5% type I error (using a Bonferroni correction for the 2 prespecified comparisons), and a power of 80%, it is necessary to include 32 patients per group. Although analysis of the primary outcome will be adjusted for baseline values, the sample size calculation does not take into account this adjustment to maximize the power. Finally, considering 20% ​​missing data, 120 patients (40 per group) are planned in this clinical trial.

Data are recorded in an eCRF developed using Clinsight (Ennov Inc). The eCRF is used for data entry at each investigator site, and every center is responsible for patient anonymization. The eCRF was created, tested, and validated before the start of data entry. The data necessary for monitoring the primary and secondary endpoints are identified and managed at regular intervals throughout the trial. Data are monitored by the data management team of the data management department of University Hospital of Lille using predefined data management rules, and queries are automatically edited. Finally, overall automated monitoring is performed by the data manager at the end of data entry. In case of discrepancies, queries are edited to resolve the problems encountered. After validation, the database is frozen and exported for analysis.

Statistical analysis will be conducted in a blinded fashion with a blinded code for the intervention. All statistical analyses will be carried out independently in the Department of Biostatistics of the University Hospital of Lille. SAS 9.3 (SAS Institute Inc) software or later will be used. P values will be reported as the actual values, unless P<.001. No interim analyses are planned. A detailed statistical analysis plan (SAP) will be drafted and validated before the database is frozen. Patient characteristics at baseline will be described for each of the 3 arms; the quantitative variables will be described either by the mean and standard deviation in case of a Gaussian distribution or by the median and interquartile range if not. The normality of the distributions will be verified graphically by histograms and by the Shapiro-Wilk test. The qualitative variables will be described by the numbers and percentages of each category. All analyses for the primary and secondary objectives will be performed on all randomized patients in their original group of randomization according to intention-to-treat (ITT) principles. No subgroup analysis will be performed.

Comparison of the primary endpoint (2-month change in STAI scale) between each experimental group and the control group will be performed separately using an analysis of covariance with an adjustment for the STAI baseline value and the center. Since 2 comparisons will be made for the primary analysis, each of them will be performed at the 2.5% significance level (Bonferroni correction). The absolute mean differences and effect sizes (standardized mean difference) will be reported with the 95% confidence interval. In case of deviation from normality of model residuals, nonparametric analysis will be used; absolute changes between baseline and 2-month visits will be calculated and compared between the experimental arm and control arm using nonparametric analysis of covariance adjusted for baseline values [63,64]. The primary analysis will be conducted according to the ITT principle. Missing values will be handled by multiple imputation procedures. Missing data will be imputed under the missing at random assumption using a regression switching approach (chained equation with m=20 imputations) with predictive mean matching method for continuous variables and logistic regression (binary, ordinal, or polynomial) for qualitative variables [65]. The imputation procedure will be performed using the main baseline characteristics, outcome, and variable group of randomization, and multiple imputed data sets will be combined using Rubin’s rules [66,67]. Sensitivity analyses will first be conducted using all available STAI measurements (complete cases analysis) and second in patients without major deviation from protocol (per protocol analysis); major deviations will be specified in the SAP.

For the secondary endpoints of the HADS anxiety subscale, PSWQ, VAS-stress, PSS, BDI, and HADS depression subscale scores, comparisons of the changes between the baseline and the 2-month visit between each experimental group and the control group will be performed separately with the same methodology used for the primary endpoint.

Comparison of the change between the baseline and the 6-month visit for the primary endpoint and secondary endpoints between each experimental group and the control group will be performed separately using a linear mixed model. In this model, we will include the 2 measurements (at 2 and 6 months) for the time effect, the group effect (experimental/control), an adjustment for baseline values and center, and a time × group interaction. A contrast with a 2.5% type I error will be used to compare the 6-month change between the experimental and the control group.

The efficacy of the 2 therapeutic programs (face-to-face versus digital support) at 2 and 6 months will be compared using an analysis of covariance at the 5% significance level to compare the variations in the STAI score between the groups, adjusting for the baseline value.

VAS satisfaction and consumption of tobacco, alcohol, and drug use will be analyzed in each group using descriptive statistics.

The full version of the protocol can be viewed in Multimedia Appendix 4.