Both the technical platform and the contents of the treatment were developed de novo (Multimedia Appendix 1). The treatment content was developed according to an evidence-based manual [30]. Decisions not informed by previous research were made in collaboration with an expert group consisting of 2 psychologists and 1 psychiatrist from the Research Clinic for Functional Disorders and Psychosomatics at Aarhus University Hospital in Denmark. This group was supplemented with 2 external psychologists with expertise in developing and conducting research on internet-delivered programs. Two focus group interviews with former patients were conducted to establish relevant themes to address in new treatment content (eg, patient videos). The technical platform was developed in collaboration with a Web-developer. New procedures—patient self-referral and video-based assessment—were applied.

A total of 18 patients were self-referred of which 3 (3/18, 17%) withdrew prior to assessment (Figure 1). Of the patients, 15 (15/18, 83%), including 12 females (12/18, 67%) were assessed for eligibility, all of whom met the inclusion criteria and were included in the study. Socio-demographic and clinical details are shown in Table 2. Of the patients, 80% (12/15) completed treatment (ie, performed a minimum of 3 modules). The majority of the completers were considered active during all 12 weeks of treatment, independently of the number of modules finalized. One patient (1/15, 7%) dropped out prior to module 1 due to technical obstacles. Thus, post-treatment data on 14 (14/15, 93%) patients and 3-month follow-up data on 12 (12/15, 80%) patients were included in the final statistical analyses (Figure 1).

Of the patients, 12 (12/15, 80%) agreed to participate in the semi-structured interviews at post-treatment. Due to cancellations, only 10 interviews were conducted. The patient who dropped out and 1 of the non-completers participated in the interview ensuring a varied evaluation.

Online patient recruitment through the clinic’s webpage attracted the patient population of interest. All self-referred patients were judged to have health anxiety as their primary problem based on the first screening and they were invited for assessment. Patients reported that it was easy to find the trial and go through the steps of self-referral.

At the end of the video-based interview, patients commented on their user experience. Overall, patients found it convenient, and the majority expressed feeling more secure at home than having to go to the hospital. A few patients felt anxious prior to the assessment due to the new video-format, and technical issues did interrupt a few interviews. In spite of this, most patients expressed that the video-format did not interfere negatively with their sense of contact with the clinician during the assessment. All clinicians had no prior experience with the video-format, and they reported positively about the amount of clinical information the assessment allowed for.

The post-treatment patient interviews revealed overall satisfaction with relevance and progression of treatment themes and a positive response to the use of patient videos. There was no consistent pattern in content feedback (eg, some patients emphasized the importance of mindfulness exercises, and some patients did not use them at all). Patients sent an average of 12.9 (SD 7.76) messages (range 4-29) to their clinician, and clinicians sent an average of 18.5 (SD 3.46) messages (range 12-25) to their patient. Some patients expressed that this form of communication allowed for more honesty compared with earlier experiences with face-to-face therapy, whereas for others it was a hindrance. Therefore, some patients wrote many extensive messages where others consulted the clinician only to a limited extent. The clinicians reported that it was possible to have a good clinical sense of the patient. However, for those patients who did not use the clinician, the therapeutic reinforcement was missing to a larger extent. Furthermore, the patients were generally surprised by the amount of time they needed to complete a module.

Mean, standard deviation, and percent changes at baseline, post-treatment, and 3-month follow-up are shown in the Table 3. Mean difference, CI, P values, and effect sizes are shown in Table 4. Patients reported statistically significant reductions in health anxiety symptoms from baseline to post-treatment with a difference in WI-7 score of 37.2 points (95% CI 24.1-50.4, t₁₃= 6.11, P<.001), and at 3-month follow-up with a difference of 33.9 points (95% CI 13.6-54.3, t₁₁= 3.66, P=.004) with large within-group effect sizes (SRM 1.06-1.63).

There were no significant changes in scores from post-treatment to 3-month follow-up which suggests stability of clinical outcome. Furthermore, patients reported significant reductions in symptoms of anxiety and depression, and significant increases in life satisfaction and psychological flexibility (Table 4). Effect sizes were moderate to large on all secondary outcomes (SRM 0.73-1.80), with the exception of the SF-12 PCS measuring physical functioning. Apart from life satisfaction, which further increased significantly from post-treatment to 3-month follow-up (SRM 0.96), there were no significant changes in scores from post-treatment to 3-month follow-up.

The patient interviews gave rise to revision and further extension of some treatment themes and exercises. Furthermore, we learned that patients need to not only be motivated for internet-delivered treatment, but also to be able to work independently, prioritize their time with regard to treatment engagement, and have a minimum of technical know-how. To further improve adherence prior to the RCT, a patient recruitment video was developed to provide a clear picture of the treatment format, emphasizing the time and effort needed to participate in this type of treatment. This was also further addressed in the assessment interview. Patients who were ambivalent had up to 2 weeks to decide whether to enter the randomization for the following trial. Based on the experience with the drop-out of 1 patient due to technical obstacles, we included a question about computer skills in the baseline questionnaire. We also enquired more about this in the assessment interview. Finally, due to discussions raised about changes in medication during the pilot study, an extra inclusion criterion was added for the RCT to ensure that patients had been stable on anxiety medication for at least 2 months prior to inclusion.

The main finding of this pilot feasibility study was that iACT encompassing online self-referral, video-based assessment, and internet-delivered treatment was a feasible treatment concept for patients suffering from severe health anxiety. We found overall positive outcomes with large improvements in health anxiety symptoms and medium to large improvements in anxiety, depression, HRQoL, and life satisfaction, which improved further during the 3-month follow-up period. The observed reductions in health anxiety symptoms were similar to those reported in a previous, uncontrolled, face-to-face pilot study, and a subsequent RCT on group-delivered ACT for patients with health anxiety [24,32]. Since health anxiety is considered a chronic disorder if untreated [10], spontaneous remission is unlikely to have contributed significantly to this effect. It is often debated whether symptom measures should be used as a primary outcome in ACT trials, since the treatment does not focus on symptom reduction but instead aims to increase behavioral flexibility and valued living. In our study, we also measured psychological flexibility, which did show the largest improvement (SRM 1.80). This supports the model of change suggesting that increased psychological flexibility is associated with general psychological well-being and might even precede symptom reduction [28,41]. Thus, symptom reduction is a welcomed treatment outcome but should not be mistaken as a therapeutic goal in treatment. In the present study, only 1 (1/15, 7%) patient dropped out, and 12 patients (12/15, 80%) completed 12 weeks of treatment. This was comparable to iCBT for health anxiety which has shown a similar completion rate of 85% [15]. Since the majority of patients completed the treatment, iACT seems to be an acceptable and feasible treatment for severe health anxiety.

The thorough standardized clinical assessment and well-defined diagnostic inclusion criteria were strengths of this pilot study. Moreover, both completers and non-completers participated in the post-treatment interview and most data were available for patients at follow-up. Lastly, the study employed validated clinical outcome measures. Even though this study did not employ a randomized design, most outcomes pointed in the same direction: symptoms improved following iACT. The most important limitation of this study was the uncontrolled study design and small sample size. The design was chosen based on the aims of this study, and the sample size did not limit the feasibility evaluation, which was the primary aim. The preliminary observed clinical changes should be replicated in a larger RCT to establish clinical relevance. Another limitation was that the video-based assessment was not evaluated systematically. We did discuss the technical and clinical implications at supervision, and asked the patients about their experience at the interview, but we did not collect our experiences in a structured manner. Lastly, the post-treatment patient interviews were not analyzed systematically and since they were conducted by the dedicated clinician, there was a risk of bias due to socially desirable answers.

Patient self-referral is rarely applied in a publicly funded health care system based on gatekeeping, where patients are referred primarily through the general practitioner. We found that the majority of those who self-referred fulfilled criteria for severe health anxiety, which may indicate that self-referral is a sensible recruitment procedure in patients with severe health anxiety. One explanation may be that patients with health anxiety often seek medical information via the internet, and may have come across the clinic’s webpage while searching for explanations for their symptoms. The patient video on the webpage was likely to have further increased recruitment sensitivity as many patients reported that they recognized their own health anxiety from the patient’s stories. Former studies support that patient self-referral is a feasible recruitment method for internet-delivered psychological treatment [42,43]. However, patient characteristics may differ compared to patients referred from the general practitioner to outpatient treatment. Prior to this study, we anticipated that the treatment setup would appeal to males and younger patients with less severe symptoms of health anxiety. Studies tend to find no age or gender difference in the distribution of health anxiety [10], yet males and younger patients seem to be less frequently referred to treatment [24]. However, in the present study, we recruited 80% (12,15) female patients, with an average age of 38.8 years. Moreover, patients had health anxiety for approximately 15 years and were quite impaired at referral, with an average WI-7 score of 81 (range 0 to 100). Online administration of questionnaires has shown to be a valid method for measuring symptoms of health anxiety, which suggests that the surprising level of symptom severity cannot be explained by this [44]. This might imply that a different patient subgroup was reached, even though the group did not differ in the expected direction. In the DSM-5, patients suffering from health anxiety are categorized as either care-seeking or care-avoiding [2]. It may be that self-referral appeals to more avoidant patients with health anxiety, thus making treatment accessible for a different subgroup. However, these speculations need to be investigated in a larger patient sample.

The overall aim of iACT for health anxiety is to target and improve a patient's psychological flexibility, which previously has shown to be a central process of change [28,45]. This was done through several treatment processes as previously illustrated (Textbox 1). Specifically, we had a distinct focus on acceptance of inner experiences and values in order to broaden the behavioral repertoire from control and avoidance to values-based behaviors. This involved gradually changing behaviors from unworkable short-term strategies to long-term goals in line with the chosen values. The distinct focus on personal values in ACT compared to CBT may explain why iACT tends to have better adherence to protocol compared to iCBT [22]. Future studies should include process measures to further understand the role of values and investigate the mediational effect on symptoms.

Clinician guidance was added to this treatment program because studies generally find that unguided internet-delivered treatment has higher attrition [20,46]. The asynchronous message system enabled a more flexible and patient-tailored contact in terms of individual needs of guidance. This was also reflected in the wide range of messages exchanged between each patient and their allocated clinician. The aim of the written guidance was to motivate patients, to validate patient efforts, to clarify misunderstandings, and to personalize treatment content. This study did not measure the quality of therapeutic alliance, but a former review on internet-delivered treatment across patient populations found client-rated alliances close to those found in face-to-face studies [47]. However, the review also reported that clinicians rated the alliance lower than in face-to-face treatment. This is in line with the feedback from clinicians in the present study regarding patients who did not use their dedicated clinician. This may suggest that clinicians are affected more by the internet-delivered treatment modality than patients.

Severe health anxiety is a disabling and costly disorder with a lack of accessible, specialized treatment options. Here, we showed that online self-referral, video-based assessment, and iACT may be a feasible treatment set-up for patients suffering from severe health anxiety. The promising outcome results also suggest potential efficacy for this iACT program which is currently being tested in a larger RCT. If proven effective, iACT may increase availability and accessibility of specialized treatment for health anxiety. Future studies need to evaluate the sensitivity of patient self-referral, and to compare characteristics in patients self-referred or referred through a health care professional, to determine whether different patient populations are reached.