Individual in-depth interviews were conducted to explore dementia care experts’ views on EDC Internet support. Experts from different institutions and regions within the Netherlands were recruited via email. Inclusion criteria were (1) professional caregiver in the Dutch dementia care field, (2) daily interaction with people with dementia (PwD) and their caregivers, and (3) ample experience in supporting EDC. The number of participants (N=11) was determined by data saturation. The redundancy of themes emerged from interviews [34]. Professional backgrounds of the experts were psychiatrists (N=1), clinical neuropsychologists (N=3), registered health psychologists (N=4), occupational therapists (N=1), social psychiatric nurses (N=1), and nurse practitioners (N=1), with an average of 13.64 (SD 7.43) years of professional experience. A semistructured interview guide was developed by authors LMMB and MEdV, which was validated by author FRJV. Topics included EDC needs, relevance and feasibility of EDC support, and themes for an EDC intervention. Brief summaries of the key points were made throughout the interview to obtain participant verification [35]. All interviews were audiotaped. The content of the verbatim transcriptions of the interviews was analyzed by summarizing common themes based on deductive content analysis.

The experts emphasized that caregiver support needs to be tailored to the dementia stage. Concerns were raised about providing early support in the absence of later-stage problems, when caregivers are not in need of help yet and could possibly reject early stage support. Experts considered education on the disease and its course as most important. Other important themes involved accepting the disease, coping with relationship changes, stress, role management, and rumination. The importance of interaction between PwD, caregivers, and environment was also stressed. Correcting or accepting care recipients’ mistakes and notifying social network can be primary stressors in daily interaction. Too much negative information not fitting the early stages, for example, behavioral problems, care homes, and end-of-life decisions, could lead to adverse reactions and should be avoided in EDC support.

Combining the themes chosen by the experts and the caregivers resulted in 9 separate themes (Table 3). The theme “practical tips” was incorporated in the module structure to provide tips thematically.

Intervention content was proposed by authors LMMB and MEdV based on a literature review [10], EDC needs [24], identified themes in Step 1, and conceptual frameworks on self-management. The Stress and Coping paradigm [25] served as the theoretical basis for the content of the modules. According to this model, stress is experienced when a person perceives that the demands (caring for a person with dementia) exceed their personal and social resources. Caregivers’ responses to their stress situation might be mediated by their understanding of the situation and their beliefs about their ability to cope. The latter fits Bandura’s [26] concept of self-efficacy (belief in one’s capabilities). Consistent with this theory, models of dementia management emphasize the need to maintain self-worth and control [28]. An intervention aimed at increasing self-efficacy should not only educate the caregiver, but should also foster self-management by combining education with problem-solving skills, and work toward a change in behavior [36].

The proposed self-management intervention program “Partner in Balance” (PiB) encourages caregivers to actively manage their lives and identify solutions for their specific needs [37]. Increasing knowledge, identifying and setting goals, and learning skills to achieve these previously set goals served as the basis for the intervention program. Module content was focused on role management (eg, balancing activities in daily life) and emotional management (eg, dealing with fear and insecurity about the future) [38]. Formulating, planning, and executing personal goals can be learned using a proactive 5-step change plan (Textbox 1) often used in self-management [38], which was integrated into each module. By formulating and planning a personal change plan, caregivers learn to anticipate on stressful situations and gain confidence in their ability to take care of the situation and themselves [38]. Because caregivers greatly varied in their needs, personal goals, and interest, a flexible choice of modules was used. Successful elements that were identified in the literature review [10], including tailored caregiving strategies and contact with a coach and/or other caregivers, were included in the program content likewise.

Dementia care experts (N=4) not involved in the initial interviews were asked to read the material and provide comments with respect to language, tone, amount of information, and significance and propriety of the content. They worked as clinical neuropsychologist (N=1), registered health psychologist (N=1), occupational therapist (N=1), and social psychiatric nurse (N=1), with an average of 9.75 (SD 2.22) years of professional experience.

The experts provided comments concerning the content of each module. Stigmatizing, complex, or unclear language was reported and alternatives were provided. The textual content of the program was adapted accordingly.

A Web-based fully operational program was developed based on the senior-friendly website checklist [39]. Initial usability flaws were tested with the “think-aloud” method [40]. Potential users were asked to think aloud while using the system, allowing the researchers to understand the reasons behind their usage behavior. Participants (N=4) were randomly selected from the focus group interviews (N=2) and the expert interviews (N=2). All aspects of the senior-friendly website guidelines were checked (eg, font size, contrast, menus and navigation, button style and size, phrasing, illustrations and videos, and Web assistance) and additional comments on user-friendliness were explored in-depth. The interviews were audiotaped and the verbatim transcripts were combined with field notes made by author LMMB on experienced difficulties during the walk-through. A coding scheme based on the interview protocol was used [40].

All participants (N=4) commented on the layout, font size, contrast, tone, and navigation. The website layout was considered professional and attractive, although a uniform composition of all pages was proposed to foster cohesion. Alterable font sizes and increased contrast with the background color were suggested. Vignettes of caregivers were considered useful, but addressing people with their given name instead of their family name was considered more appealing. The tips from other caregivers were perceived as crucial. It was suggested to conclude every module with these tips to increase layout uniformity.

An uncontrolled pre-post-intervention pilot study with EDC was conducted to establish feasibility, as this is recommended before moving on to a larger scale effect study [31,41]. The Medical Ethics Committee of the Maastricht University Medical Centre approved this study (No NL44475.068.13, Dutch Trial registration number NTR4217). Caregivers were included in the study if they (1) were spousal caregivers of people with mild cognitive impairment [42] or mild dementia of all subtypes [43], and (2) had access to the Internet. Exclusion criteria were (1) insufficient cognitive abilities to engage in the online self-management program, (2) overburdening, (3) severe health problems (determined by the study staff), or (4) caring for PwD caused by human immunodeficiency virus, acquired brain impairment, Down syndrome, Huntington’s chorea, or alcohol abuse. Participants were recruited at memory clinics and ambulatory mental health clinics. Based on comparable feasibility studies, we aimed to include 10 participants [44,45]. Of those contacted, 17 of 43 caregivers (40%) were willing to participate and signed the informed consent form.

Feasibility was evaluated face-to-face at the caregiver’s home by a semistructured interview developed for this study—the Program Participation Questionnaire (PPQ). The PPQ was based on measurement scales for perceived usefulness and ease of use and overall acceptance of information technology [46,47] and included 30 items on usability, clarity, comfort with, and acceptability of the format on a 7-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). For the individual PPQ items, see Multimedia Appendix 1. Mean scores were calculated with descriptive statistics. Because there were no external criteria to properly define feasibility [48], we followed the conventional strategy of using the median score of the questionnaire as a cutoff. This approach was previously adopted in a Delphi study as evidence of agreement of intervention feasibility [49]. Based on the PPQ scale (range 54-234, median 144), scores of 145 or higher were considered “acceptable feasibility.” Mean item scores (range 1-7) will be used to make decisions on positively and negatively appraised aspects of the program. Mean item scores of 5 (slightly agree) or higher will be considered positive, mean item score below 4 (slightly disagree or lower) will need further revisions. Participants were asked to elaborate their scores. Comments were audio recorded and transcribed verbatim. Meaningful data units based on the PPQ items were identified and derived independently from the qualitative data by authors LMMB and HEJW with deductive content analysis. Furthermore, the actual accessed data use of the program (number of log-ins and features used) was compared with self-reported data.

The study population consisted of 17 participants, of whom 10 completed the postintervention assessment. Participants who did not complete the postintervention assessment were replaced to meet the sample size suggested by previous studies [44,45]. The main reasons for not completing the program and the postintervention assessment were difficulties with the online aspect of the program (N=4), private circumstances (N=2), and disagreements with the care recipient (N=1). Completer and noncompleter characteristics are listed in Table 4.

The PPQ showed a good internal consistency (alpha=.89) and had a mean sum score of 209 (SD 22.14). Given the threshold of 145 or higher, this score indicated an acceptable feasibility. Mean item scores above 5 (slightly agree or higher) were found for convenience of completing the program at home (5.9, SD 1.8), clarity of the website (6.0, SD 1.2), module structure (6.1, SD 0.6) and content (6.6, SD 0.5), privacy (6.6, SD 1.3), tailored assignments (6.1, SD 1.3), guidance by the coach (6.6, SD 0.5), and general contentment (6.4, SD 0.9). A mean item score below 4 (slightly disagree or lower) was found for usefulness of the discussion forum (2.8, SD 2.7). Table 5 shows the positive, negative, and neutral themes derived from the additional comments. Self-report usage data were comparable to tracked usage data: 106.41 (SD 96.15) minutes spent per module, including scoping the website (4.5 minutes, SD 4.13), completing the assignments and change plan (79.14 minutes, SD 77.16), contacting the personal coach (15.31 minutes, SD 16.96), and visiting the discussion forum (7.46 minutes, SD 7.09), spread out over 2.38 (SD 1.38) weeks.

Preliminary understanding of the effectiveness of the program was based on the baseline and postintervention assessment 8 weeks later, completed at the participant’s own convenience in an uncontrolled pilot study. At participant’s request, paper questionnaires were used. The Caregiver Self-Efficacy Scale (CSES) was used to measure domain-specific caregiver self-efficacy [50]. The subscales include 4 items on service use and 6 items on care management, with scores ranging from 1 (not at all certain) to 10 (very certain). We found a good internal consistency for both service use (alpha=.73) and care management (alpha=.87). Paired samples t tests were conducted to evaluate pre-post-intervention changes. The goal attainment scaling (GAS) [51] method was used to rate treatment-related change and to compare relative success of previously set personal goals. Baseline scores were set at −2. Postintervention scores can range from −2 (much lower than expected) to +2 (much better than expected), with a score of 0 meaning goal attained. Raw scores were transformed into an individual mean GAS score (T score) to determine goal attainment with a potential weight assigned to the goal(s) [52]. T scores of 50 or more (SD 10) indicate effective goal achievement.

Postintervention, participants (N=10) had significantly higher scores on both the CSES care management subscale (mean 41.1, standard error [SE]=2.5; t ₉=−2.5, P=.03) and service use subscale (mean 32.6, SE=1.7; t ₉=−3.5, P=.01) compared with preintervention scores (mean 36.1 and SE=3.2, and mean 23.2, SE=3.4, respectively), although effect sizes were small (d=0.14 and 0.41, respectively).

In this study, 8 program completers set 13 goals in total; 2 program completers were not able to set goals due to personal difficulties verbalizing the desired change. Two goals (of 2 participants with multiple goals) could not be scored after the intervention, because the goals changed during the course of the study. In total, 8 goals were attained (2 attained, 5 higher than expected, and 1 much higher than expected), and 3 goals were unattained (1 much lower than expected and 1 lower than expected). The mean T score at baseline (set at the −2 level) was 27.8 (SD 3.04). The mean achieved T score after the intervention was 53.7 (SD 12.03). Table 6 shows the number of goals for each domain, with most goals set on communication with the care recipient (N=7), followed by maintaining positive activities together (N=2), obtaining social support (N=2), and planning time alone (N=2).

In the intake session participants are introduced to the website and the self-management concept of the program. The coach and participant set personal goals using a motivational interviewing technique frequently used to identify change objectives and enhance intrinsic motivation [53]. Based on the discussed areas for improvement, participants select 4 of the 9 modules that were previously identified by experts and caregivers. Participants are provided with personal login codes to access their selected modules and edit their personal information. After the online period, participants will discuss their personal goals and their ability to cope with future difficulties in the evaluation session.

During the online period of the intervention, participants follow the chosen modules during an 8-week period. The website consists of (1) a home page with a short description of the goal of the program, personal login option, contact information of the researcher, and the institutional affiliation (Maastricht University); (2) a personal page with a link to the chosen modules and a mailbox for exchanging emails with the personal coach; and (3) an online forum to interact with other caregivers, moderated by the researcher and personal coaches.

Every module has a fixed design of the following 4 components: (1) video clip of fellow family caregivers, (2) education, (3) self-reflection assignment, and (4) the 5-step change plan, guided by the personal coach who will provide individualized online feedback after completion of each module and offers assistance when needed. For every module, 2 weeks are reserved as a starting point. However, participants are allowed to complete the modules at their own pace as informed by the self-management approach [38]. The first week of a module addresses Components 1-4. Participants can send their assignment and the 5-step plan to their coach. The second week of every module is reserved for feedback from the coach, after which participants can adjust their 5-step plan if necessary.

The personal coach is an experienced dementia care professional (psychologist or psychiatric nurse). Coaches will receive a 1-day training in self-management techniques and online help before the start of the intervention. They will receive experienced supervision from an experienced professional in psychology and self-management during the course of the intervention period to ensure quality and alignment of the feedback of the coaches according to self-management principle. Coaches are asked to support participants in choosing modules that fit their personal situation, help participants identify feasible goals, offer techniques to achieve goals, and provide participants with general constructive feedback on their assignments. Using a personal login code, coaches will be matched with the participants assigned to them. The CONSORT EHEALTH checklist is presented as Multimedia Appendix 2.