Consecutively implanted ICD patients from six hospitals in the Netherlands (ie, Erasmus Medical Centre, Rotterdam; Amphia Hospital, Breda; Catharina Hospital, Eindhoven; Onze Lieve Vrouwe Gasthuis, Amsterdam; Canisius Wilhelmina Hospital, Nijmegen; Vlietland Hospital, Schiedam) were approached for study participation between April 2010 and February 2013. Patients were eligible for participation if they fulfilled the following inclusion criteria: first-time ICD implant, age 18-75 years, proficient in the Dutch language, and with Internet access and a sufficient level of Internet skills. Exclusion criteria were the following: life expectancy less than 1 year, history of psychiatric illness other than affective/anxiety disorders, or on the waiting list for heart transplantation.

Patients were approached by the ICD nurse or ICD technician prior to or briefly after ICD implantation. They were informed both verbally and in writing about the study. If the patient met the inclusion criteria and was willing to participate, informed consent was signed. Patients who could not decide at that time were approached again after ICD implant while still hospitalized. Prior to discharge from the hospital, consented patients were provided with the first set of questionnaires (baseline) and their medical records were accessed for information on their demographic and clinical variables. After completing the questionnaires, patients returned them in a self-addressed and pre-stamped envelope to Tilburg University, Netherlands, which served as the core lab for WEBCARE. If the questionnaires were not returned within two weeks, patients received up to 3 reminder phone calls. Patients who did not want to participate but who were willing to give access to information from their medical records also signed an informed consent form. The study was approved by the Medical Ethics Committee of all participating centers and was conducted in accordance with the Declaration of Helsinki. All patients provided written informed consent.

After receiving the baseline questionnaires and signed informed consent, participants were randomly assigned to either of two conditions: (1) the WEBCARE (WC) group, receiving questionnaires at baseline, 3 months, 6 months, and 12 months by mail, and getting access to the Web-based intervention for a time period of 12 weeks to complete 6 modules online [24] in addition to usual care, or (2) the Usual Care (UC) group, receiving only questionnaires by mail at all time points and usual care.

Patients were randomized using block randomization by computer, randomizing 20 patients per hospital, at each time point. Randomization lists were generated by an independent, blinded statistician and sealed by a research assistant. For the current analyses, we will only focus on patients who were randomized to the WC group.

Patients who signed the informed consent form but who decided to quit the intervention and/or the study prematurely were contacted by telephone 12 weeks after randomization and asked why they had decided to quit. This time interval was chosen in order to not interfere with possible intervention effects (patients were allowed to work at their own pace, some chose to finish the intervention within the first two weeks, while others decided to do the 6 lessons within the last two weeks. For that reason, it was clear at 12 weeks who had quit or finished the intervention). Hence, patients were contacted at the time that they should have received their 3-month follow-up and finished the 6-module online course.

The intervention was based on the previously developed Web-based treatment “Alles Onder Controle” (Everything Under Control) [25] and was for the purpose of the WEBCARE trial adapted for ICD patients. The Alles Onder Controle treatment was developed for the healthy depressed population and has proven to be effective in reducing distress [25,26]. The Web-based course for ICD patients is a 12-week intervention of 6 online lessons addressing distress based on the cognitive behavioral model (problem-solving treatment). The first lesson focused on psycho-education with respect to living with an ICD (eg, what are “normal” adaptation problems post ICD implantation). In the second lesson, patients received homework assignments and were provided with therapist feedback (feedback was provided by master’s-level psychologists and was intended as minimal guidance to help patients get through the lessons—encouraging patients to continue with the lessons and giving guidance on how to address their problems according to problem-solving theory). In addition, patients received a relaxation training CD, which they were allowed to use throughout the intervention.

Patients were allowed to work at their own time and pace; however, if a lesson was not finished within two weeks, a reminder email was sent, with up to 3 reminders per lesson. Patients could proceed to the next lesson only when the previous one was finished and the homework assignment was sent to the therapist. If patients did not log in within the first two weeks, a reminder email was sent. Twelve weeks after receiving the log-in information, patients’ accounts were automatically closed.

Continuous variables were compared using the Student’s t test, while discrete variables were compared using the chi-square test. Data are represented as percentages for nominal variables and mean (SDs) for continuous variables. To compare groups on psychological variables, ANOVAs (analysis of variance) were performed. If group differences were observed, the Tukey-Kramer post-hoc test for unequal group sizes was used to identify which groups differed significantly. Descriptive data were coded and analyzed using “frequencies”. A P<.05 indicated statistical significance. All tests were two-tailed. Data were analyzed using SPSS Statistics 19.0 for Windows.

A detailed description of the patient recruitment for WEBCARE is displayed in Figure 1. A total of 1024 patients were approached for participation, 735 (71.78%) were excluded due to not meeting the inclusion criteria (n=492), refusing to participate (n=192), or not returning baseline measures (n=51). Eventually 289 patients were randomized to either the WC group (n=146) or the UC group (n=143).

Demographic, clinical, and psychological characteristics of patients who were randomized to the WC group are shown in Table 1. The mean age of the group was 58.23 (SD 9.87) and 120 (82.2%, 120/146) were male patients. In addition, 106 (72.6%, 106/146) had a higher educational level.

Patients who did not return the baseline questionnaires (n=51) and were excluded (not randomized) from current analyses did not differ systematically on demographic variables. However, significant differences on clinical variables were observed with patients who were not randomized, more often having a NYHA Class III/IV (P=.045), peripheral artery disease (P=.022), and using psychotropic medication (P≤.001) (anxiolytics, P=.004 and hypnotics, P=.010). Of the patients who refused to participate (n=192) but fulfilled the inclusion criteria, 60 (31.3%) signed consent and gave permission for medical record screening at the time of implantation. These patients were somewhat older [60.26 (SD 1.80) vs 58.16 (SD 10.30); P=.04], more likely to have a NYHA Class III/IV (34.1%, 14/41 vs 21.3%, 57/267; P=.013), to have experienced a previous myocardial infarction (76.6%, 36/47 vs 50.1%, 170/339; P=.001) or coronary artery bypass grafting (34.0%,16/47 vs 19.2%, 65/339; P=.019), to have peripheral artery disease (13.6%, 8/59 vs 5.3%, 18/339; P=.018), or to have a cardiac resynchronization therapy-defibrillator (CRT-D) (57.9%, 44/76 vs 25.9%, 87/336; P<.001) as compared to patients who signed informed consent for study participation (n=340).

Of the 146 randomized patients to the WC group, 34 (23.3%) completed the treatment and filled in the follow-up assessment (completers), 88 (60.0%) patients dropped out of the treatment but remained in the study and filled in the follow-up assessments (treatment dropouts), and 24 (16.4%) patients dropped out of the treatment and the study (dropouts). Focusing on the treatment, Figure 2 presents an overview of patients’ adherence to the intervention and shows that the number of patients completing the lessons diminishes over time. The first lesson was completed by 83.5% (122/146) of the patients randomized to the WC group (16.5%, 24/146 never logged in), while only 23.3% (34/146) completed the last lesson (and thus the whole treatment schedule).

Completers and (treatment) dropouts did not systematically differ on any baseline demographic, clinical, or psychological measures (Table 2). Although not significant, dropouts tended to be more often employed, have a lower education level, and lower mean score on anxiety and depression as compared to the other two groups. Table 2 shows a detailed overview of the baseline demographic and clinical characteristics and psychological profile of the three groups.

The reasons given by patients for not completing the treatment are displayed in Figure 3. Unfortunately, we were not able to reach all patients to learn about their reasons for dropout. Of the patients that we were able to reach, the majority (20.5%, 23/112) indicated that they faced technical problems (eg, the computer was not working, the website was not responding, problems with Internet connection): “My computer broke and I don’t use it that often, so I would have to buy another one just for the study; I didn’t wanted to do that” and “I got irritated because the website was not responding when I tried to log in, so I decided to quit”.

Time constraints were an issue in 15.2% (17/112) of patients (“I started working, so I don’t have time to do the online course”) and 10.7% (12/112) of patients were feeling fine and did not need additional support: “I had trouble thinking of any problems that I’m experiencing at this moment, as I don’t have any. I’m feeling fine”; and “I don’t have any problems. I’m happy that I have received the ICD. I feel reassured now”. Additional reasons were that the treatment did not apply to the patients’ needs (6.3%, 7/112): “I expected the course to be addressing only ICD specific problems…more technical problems. That’s why I decided to participate. I didn’t want to discuss psychological issues as I don’t have any”, and having already a lot to deal with (5.4%, 6/112): “I have just received an ICD and I’m dealing with depressive symptoms; I have other priorities at this moment”, and “I wanted to do the treatment but not within the 12 weeks. I wanted to start at a later time point because I had a lot to deal with immediately after the ICD implantation”.

Other reasons for dropout included treatment as being too confronting (4.5%, 5/112): “It was too personal, too confronting. I realized that I had more problems than I thought”, and feeling too sick (4.5%, 5/112): “It was too much. I had two surgeries in the past half year and I’m now on the waiting list for heart transplantation. I’m feeling sick all the time”. There were also patients who experienced the treatment as too negative: “I was feeling fine about my ICD, but when I started reading the content of the online course, I started feeling unhappy and therefore I decided to quit”, and “The homework assignments and questionnaires are too negatively worded, while you expect us to start thinking positive. I didn’t want to proceed as I didn’t wanted to start thinking negative about the ICD and how I’m feeling”.

This is the first behavioral randomized controlled trial to address the adherence and dropout issues of a Web-based intervention in the ICD population. Our findings show that 23.3% of patients randomized to the treatment arm completed the full treatment (six lessons), while 16.5% never logged on to the intervention. A gradual decline in adherence was observed with more patients dropping out as the lessons proceeded. The three groups (completers, treatment dropout, and dropout) did not differ systematically on any demographic or clinical baseline characteristics and their psychological profile. The top 3 reasons given for dropping out of the treatment were: technical issues with the computer/website, time constraints, and feeling fine/not needing additional support.

The findings of this study are generally in line with previous findings from the “Alles Onder Controle” Web-based intervention for individuals from the general population with increased levels of anxiety or depression who volunteered to undergo the intervention. Although generally higher percentages of completers were reported (38-55%), the rate of patients who never logged on was between 9% to 16% [25,26,35] and is in accordance with the 16.5%, which was observed in the current study. The higher number of completers in the Alles Onder Controle study might be attributed to these individuals being volunteers and thus highly motivated with respect to investing time and effort in improving their well-being. In addition, the latter patients scored high on anxiety and depressive symptoms, while patients in WEBCARE were included regardless of their anxiety/depression levels, as WEBCARE was set up as a prevention trial. However, there was a comparable gradual decline in adherence in the previous studies as in patients in the WEBCARE treatment arm [25,35].

In general, higher attrition rates are reported in open access Web-based trials without therapist guidance [36], while the number of completers is higher in closed trials where feedback is provided and reminders are sent [23,37]. Looking more specifically into cardiac patients (older adults with depression and comorbid cardiovascular disease), in their qualitative study on online treatment adherence, Donkin et al reported that time constraints, competing priorities, perception of limited worth of the intervention, and anxiety about spending time on the computer were important factors that contributed to an increased attrition [38], which are to some degree comparable to what we found in the current study. Generally, their results showed that adherence might increase if the benefits of participating in the treatment outweigh the costs of participation. In the current trial, the mean depression and anxiety scores were relatively low, patients were not distressed, and likely did not see a need for the treatment as they generally felt fine.

Studies of Web-based interventions have also shown that the duration of treatment [39] and timing [40] of the treatment may affect patients’ engagement. Also, patients’ perceived control over completing the course and coaching via phone vs email are found to be associated with higher completion rates [41]. In addition, printed delivery mode has shown to result in higher attrition as compared to a Web-based delivery mode [42]. These findings can partly explain why a higher attrition rate was observed in the current study. Current results show diminished participation after the third lesson. Perhaps, had the treatment been somewhat shorter, more patients would have completed all lessons. In addition, the treatment was offered immediately after ICD implantation. As indicated by some patients, at that time point they have a lot to deal with already and may for that reason decide not to participate or complete all modules.

A number of limitations of this study must be acknowledged. First, current analyses are based on a relatively small sample and should be replicated in larger studies in the future. Second, results on reasons for dropout are based on descriptive data that were obtained via a telephone call to patients. A structured interview or validated questionnaire would perhaps provide more valid information. Third, unfortunately we were not able to reach all patients at 12 weeks by telephone; hence, current findings are based on patients who answered the phone and were willing to provide us with information regarding their reasons for dropout.

This study also has several strengths. To our knowledge, it is the first behavioral intervention trial in ICD patients to have used a Web-based approach and the first study in ICD patients overall to address issues of adherence and attrition. The information from WEBCARE adds to our knowledge about factors that may influence adherence in trials using a Web-based approach, which can be used when designing Web-based behavioral intervention trials for ICD patients in the future, in order to increase the number of patients enrolled in the study and their treatment adherence.

When offering a Web-based intervention to ICD patients, it seems of great importance to make the intervention as patient-tailored as possible. Not all patients have the same needs at the same time. Thus, giving them time and space to complete the lessons when needed is an important factor as is making it possible to select which lessons to complete (some patients are more interested in technical aspects of the ICD, which would give them more reassurance, while others prefer psychological support in dealing with this new situation). As a proportion of the ICD patients indicated that they were feeling fine and did not need any additional support, it might be more important to focus on patients who have higher distress levels post implant. These patients could be identified using brief and standardized questionnaires that are designed to assess psychological distress. Close monitoring of patients’ psychological needs is warranted as it has been associated with morbidity and mortality [11,12], and may help us to provide the right amount of care that is needed at the right time. Designing behavioral interventions with a collaborative care approach, where a period of “watchful waiting” is employed would perhaps be a way to go. With this approach, we would be able to offer support to patients who have “chronic” levels of distress or who develop distress at a later time point, as we have now learned that the “one size fits all” approach results in high dropout and low adherence. Incorporating a standard intervention in current health care models, which would be offered to all patients, might result in a great loss of resources as only a small proportion of patients would be willing to participate and complete the treatment (as shown in current data).

In conclusion, as Web-based treatments are increasingly being implemented in clinical practice, knowing how to keep patients motivated and compliant with treatment becomes more important every day. Our findings indicate that more attention should be paid to the technical aspects of Web-based treatment and making it more user-friendly. In addition, to overcome the barrier of home computers not working as they should, future studies should examine whether a similar intervention could be delivered using smartphones or tablets in order to decrease dropout. Also, future studies should examine the relationship between adherence and outcomes, as the results to date are inconclusive [22]. Examining the appropriate duration and timing of the intervention is also of great importance, which to date remains unexplored in the ICD population. The provision of patient-tailored interventions at the time when the patient needs it is likely to increase treatment adherence and enhance the effectiveness of such interventions.