A Web-Based Psychoeducational Program for Informal Caregivers of Patients With Alzheimer’s Disease: A Pilot Randomized Controlled Trial

Introduction

Due to the worldwide aging population, the number of persons with dementia (35.6 million currently) is expected to double by 2030. The socioeconomic consequences of this rapid rise and the absence of an effective pharmacotherapy have positioned dementia as a major public health concern in recent years [1]. The Alzheimer’s Association reported that in 2012 more than 15 million caregivers provided an estimated 17.5 billion hours of unpaid care, representing US $216 billion [2]. Today, the majority of persons with Alzheimer’s disease (PWAD) living at home are cared for by their spouses, children, or friends [3]. Nevertheless, the amount of time dedicated to their relative, the physical efforts, and the strong emotional involvement associated with caregiving may induce chronic stress in caregivers and weaken their physical and mental health [4-7]. Such repercussions can also negatively affect other areas of their lives (eg, professional or social) [8].

Various nonpharmacological intervention programs for caregivers are available on-site (ie, [9,10]). Nevertheless, some caregivers are not willing or available to attend a face-to-face program due to a lack of respite, the distance, or owing to care-recipients’ behavioral or physical problems. For them, technology-based programs may represent an interesting complementary strategy to regular care management [11,12].

Based on a face-to-face psychoeducational program [13], we adapted and developed the Diapason program, based on a user-centered design, including a proof of concept and 2 usability tests [14]. Although other recent Internet-based programs have been tested [15,16], to our knowledge, the use of mixed research methods still remains rare in randomized controlled trials (RCTs) [17]. Yet, including qualitative analysis in the evaluation of these programs may improve results interpretation, help “trialists” become more sensitive to individual differences, and save money “by steering researchers toward interventions more likely to be effective in future trials” [18].

The main aim of this pilot RCT was to evaluate the impact of the Diapason program on caregivers’ perceived stress. We hypothesized that this program offering information, skills training, and a forum for caregivers would significantly reduce their perceived stress and burden, and enhance caregivers’ self-efficacy, self-perceived health, and self-perceived knowledge about the disease. Qualitative analyses would facilitate the identification of subgroups benefiting from the program and would guide us to improve content and methods to evaluate this type of intervention.

Methods

Study Design

We carried out an unblinded monocentric pilot RCT (NCT01430286) between 2011 and 2014 in a day care center geriatric unit (Paris, France). Informed consent was obtained before participation. French ethical CPP approved this protocol in July 2011. The in-depth description of the protocol study has been reported elsewhere [19].

Recruitment and Participants

The recruitment strategy included flyers and posters placed in the hospital. During the consultations, geriatricians proposed this protocol to caregivers of PWAD. The caregivers interested in the study filled out a contact form. Then a psychologist provided them with the information form, confirmed inclusion criteria, and collected the signed informed consent.

Eligible participants were required to be French-speaking caregivers of community-dwelling PWAD who met the criteria of the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition [20]. Caregivers had to spend at least 4 hours per week with their relative, be aged 18 years or older, scored 12 or more on the Perceived Stress Scale (PSS-14), and to have access to a computer with Internet connection. Professional caregivers were ineligible.

Based on the literature, a 6-point difference on PSS-14 was expected between the experimental and control groups at 3 months [21]. With an assumed SD of 9, 40 participants per group needed to be included to detect this difference with an 80% power (Cronbach alpha=.05; 2-tailed).

Intervention

The Diapason program [22] was delivered in a free, password-protected, fully automated website to be used in an individual fashion, at home, by the caregivers. The program’s content was based on cognitive theories of stress, a literature review [23], and the results of a study conducted by our team [13]. In the latter, caregivers who improved their understanding of cognitive and behavioral symptoms reported feeling less stressed. Furthermore, caregivers with a perceived personal time restriction or poor social support suffer more stress, burden, and depression [24,25]. Consequently, our intervention targeted (1) caregivers’ beliefs about the illness and the caregiving role, (2) caregivers’ skills to manage daily life difficulties, and (3) caregivers’ social support and help-seeking behavior to obtain respite or financial support, and to meet and discuss with peers through a forum. Twelve thematic sessions were sequentially and weekly unblocked once the previous one was entirely viewed (see layout in Figure 1). Owing to the variability of behavioral and psychological symptoms depending on the type of dementia and the important impact of some of them on caregivers’ stress (eg, hallucinations, delusions), only Alzheimer’s disease was targeted by this program.

Each session included theoretical and practical information, videos of health professionals, and a practice guide for applying the session’s content in real life. The length of the intervention was 3 months, with each weekly session lasting 15 to 30 minutes on average, but there was no time limit and the participants could access different website sections (eg, relaxation training, forum) for as long as they wished at any time. The program content is summarized in Textbox 1.

‎View this figure

Procedure

Participants were recruited and randomized offline in 2 parallel groups based on a computer-generated randomization list using blocking and stratification by sex and relationship (spouses vs nonspouses).

The experimental group participants received at baseline a 10-minute training session on how to use the website, a log-in and password, a printed version of the user’s manual, and a notebook to write personal ideas about their application of the program’s content. Each week, participants had to read through an entire thematic session and fill out a printed satisfaction questionnaire. Other website sections (eg, relaxation training, forum) were available but not mandatory to complete the program. No modification regarding methodology, program content (except for forum discussions), or the website was done during the course of the study.

The control and experimental group participants received usual care, in which they were provided with information about the illness during their semiannual follow-up with their geriatrician. The control group participants were given access to the Diapason program at the end of their participation. All participants were advised to look for additional help if necessary and were asked to inform the researcher about it.

An individual face-to-face assessment was conducted at the Broca hospital by research psychologists (VCL or JW) at baseline, at the end of the program (at month 3), and after 3 months of follow-up (at month 6). Each 90-minute assessment visit consisted of a structured interview, standardized questionnaires, and visual analog scales (VAS). Additionally, experimental group volunteers participated in an optional one-to-one semistructured interview at the 3-month follow-up.

Measures

Based on a cognitive approach inspired by Lazarus and Folkman’s [26] stress and coping theory and Bandura’s [27] self-efficacy model, we hypothesized the program would have a direct impact on perceived stress levels, self-efficacy, and burden, and may influence depression and self-perceived health status.

To evaluate the perceived stress of caregivers (primary outcome), we used the 14-item Perceived Stress Scale (PSS-14) [28]. The total score ranges from 0 to 56, higher scores representing higher stress levels. In order to target the caregiving stress, we adapted the instruction by proceeding with heteroevaluation and adding the following underlined text: “This scale asks you about your feelings and thoughts about your experience with your relative during the last 4 weeks.”

The secondary outcomes were:

1. Self-efficacy measured by the Revised Scale for Caregiving Self-Efficacy (RSCS) [29], which distinguishes 3 self-efficacy domains: obtaining respite, responding to disruptive behavior, and controlling upsetting thoughts. Scores in each domain range from 0 to 100, higher scores indicating a higher degree of confidence for each situation.
2. Perception and reaction to cognitive or behavioral symptoms of PWAD were evaluated with the Revised Memory and Behavior Problems Checklist (RMBPC) [30]. This instrument rates 24 problems on 2 scales. They evaluate (5-point scale) frequency and caregiver’s bother or strain for each problem. A global score ranging from 0 to 4 was calculated for both scales. Higher scores indicate higher frequency or higher emotional effects.
3. Subjective burden was evaluated with the French version of the Zarit Burden Interview [31]. The total score ranges from 0 to 88, a higher score meaning a higher burden level.
4. Depressive symptoms were measured with the second version of the Beck Depression Inventory (BDI-II) [32] including 21 items with a total score range from 0 to 63. Higher scores indicate higher levels of depressive symptoms.
5. Self-perceived health was measured with the French version of the Nottingham Health Profile (NHP) [33]. We analyzed social isolation, emotional reactions, and sleep quality subscores and rated each from 0 to 100, which provided a percentage of the perceived illness impact.

At each visit, we collected information on caregiving variables (structured questionnaire). On the 4 VAS, caregivers evaluated their current levels (from 0=low to 100=high) of (1) knowledge about Alzheimer’s disease, (2) overall stress, (3) self-efficacy for coping with the illness, and (4) the caregivers-PWAD relationship quality.

Web metrics (session length and rate of visits) were collected for each experimental group participant automatically and anonymously. Participants completed a weekly satisfaction questionnaire focused on utility, clarity, and comprehensiveness (5-point Likert scale). They rated from 0 to 100 the applicability and positive emotional impact of each session and reported their opinion of the program (open-ended question). At the end of their participation, we proposed a semistructured interview exploring their opinion of the program.

Concerning the PWAD, we collected at baseline the Mini-Mental State Examination (MMSE) [34] from the medical record and Instrumental Activities of Daily Living (IADL [35]) and the date of symptom onset (reported by the caregiver).

Data Analysis

All available data at baseline were analyzed by intention-to-treat analysis. Descriptive statistics (means and percentages) were calculated for caregivers’ and PWAD’s characteristics. Moreover, t tests (or Mann-Whitney tests) and Spearman or polyserial correlations were used to assess associations between variables. The missing data within each scale were treated according to the recommendations of the literature when available. Otherwise, simple mean imputation was used. The last observation carried forward method was used for participants who dropped out. After checking normality and homoscedasticity of primary outcome (PSS-14), we conducted an analysis of covariance (ANCOVA), controlling for regression to mean phenomenon and effects of potential confounders at baseline on primary outcome. All analyses were conducted using R Software for Windows (version 3.0.0).

Interviews and open-ended questions were concurrently analyzed by two trained psychologists (JW and VCL) following the thematic analysis method, using a semantic approach, driven by analytic interests and an essentialist/realist approach [36].

Results

Participants

As summarized in the flowchart (Figure 2), of the caregivers met by the physicians, 129 were prescreened between December 2011 and August 2013. Among them, 40 did not meet inclusion criteria (ie, did not use the Internet, did not accept/know the diagnosis, were not available for 3 assessments at the hospital), 23 were unreachable, and 17 declined. After an 8-month recruitment extension, the main investigators (ASR and VCL) stopped recruitment (in total 20 months) because the rate of inclusions did not exceed 2 persons per month on average.

We randomized 49 participants. Of the 25 participants allocated to the experimental group, 17 (71%) finished the protocol and validated at least 10 of 12 online sessions. Four participants ended their participation in the study without withdrawing consent.

Demographics and other characteristics of participants are summarized in Table 1. At baseline, the groups were imbalanced regarding the number of weekly hours of professional help and IADL and BDI-II scores. The PSS-14 scores were correlated with weekly professional help received (ρ=.33) and BDI-II scores (ρ=.49), whereas the correlation with IADL scores was weak (ρ=–.11).

‎View this figure

Primary Outcome: Self-Perceived Stress

Mann-Whitney tests did not show significant differences between the experimental and control groups over time (Table 2). We conducted ANCOVA with the PSS-14 at month 3 as dependent variable and the PSS-14 at baseline, group, stratification factors (sex and relationship), and potential confounders at baseline (BDI-II and professional help received) as independent variables. Only the PSS-14 at baseline (P<.001) and weekly help received (P=.01) were significantly associated with PSS-14 at month 3. Thus, no significant relationship was found with the intervention (P=.34). ANCOVA showed similar results when stratification factors were not included.

Secondary Outcomes

Only the VAS evaluating knowledge of the disease showed significant change at month 3 scored a high effect size (Cohen’s d=.79, P=.008). Indeed, the experimental group scores increased by 13.8 points (SD 15.1), whereas the control group scores decreased by 0.04 points (SD 17.4) (Table 2). However, no significant differences were found between the groups from baseline to month 6.

Only one user reported problems watching the videos (Flashplayer was not installed on computer) and another with little experience using the Internet could not use it unaided. The high scores on the weekly satisfaction questionnaire showed that nearly all participants considered Diapason topics to be useful (95%, 19/20), clear (100%, 20/20), and comprehensive (85%, 17/20). Topics describing strategies to maintain relatives’ autonomy and coping skills with the PWAD’s behavioral troubles fostered higher levels of positive emotional impact (mean 61.50, SD 22.83 and mean 61.90, SD 26.68, respectively). The most applicable session was focused on coping skills of the PWAD’s behavioral troubles (mean 72.25, SD 15.22). In contrast, the session describing caregiving stress factors and protectors received the lowest scores for positive emotional impact (mean 49.25, SD 21.75) and applicability (mean 61.00, SD 17.67).

On average, participants used the website 19.72 times (SD 12.88) and for 262.20 minutes (SD 270.74) during the first 3 months. The most frequently visited section was the forum (mean 24.86 times, SD 40.95), whereas only 10 messages and 10 answers were posted. Four spouses (45%) and 4 daughters (33%) visited the website 26 times or more (third quartile). No significant correlation was found between the PSS-14 score (M3–M0) and frequency (ρ=–.15) or duration (ρ=–.05) using the website. After month 3, connection times were near zero.

Qualitative Analysis

Thematic analysis on the participants’ impressions underlined four trends: caregivers without a clear opinion toward the program (5/25, 20%) and those with a clearly positive (3/25, 12%), qualified (11/25, 44%) or negative (6/25, 24%) opinion. These trends were significantly associated to the relationship (Fisher’s exact test, P=.01). Thus, most wives had a negative opinion, whereas daughters primarily expressed a qualified opinion about the program. Only male caregivers expressed a positive opinion (see Table 3). As shown in Figure 3, reasons varied between caregivers of a single category.

‎View this figure

Moreover, we distinguished 4 topics comprising caregivers’ opinions (examples in Table 4):

1. “It was useful for me.” A few participants reported having benefited from the program. They said it improved their understanding on the disease or changed their initial beliefs about the disease or diagnosis.
2. “It would be better for others.” Participants considered the program would be better attuned to the needs of a PWAD in other (earlier or more advanced) stages of the disease than their relatives. Most children thought the overall “message” was more adapted for spouses rather than for them. The contrary was not stated.
3. “I expected something else.” Some participants found the content was not in-depth enough. They expected more specific and individualized advice, and more “human interaction” with professionals or peers.
4. “This is not for me.” Other participants preferred another kind of intervention (eg, individual therapy, respite, financial support) or reported not feeling a need for help. Others considered the program had come too late or did not believe that someone/something could help them. Most of them ended their participation.

Additionally, many experimental and control group participants reported having used other resources to better understand the disease and adapt their behavior (eg, reading books, asking for help, or contacting associations).

Discussion

Principal Findings

Statistical analysis did not show significant differences in self-perceived stress (PSS-14) between the groups over time. This result is most likely due to low statistical power. Perceived stress levels remained stable over time in PSS-14 scores although the disease progressed. This stabilization has been observed in control groups from similar studies, suggesting that caregiving stress rarely increases over a period of 3 months [16]. After 6 months, a few experimental group participants had heightened stress levels. This may be due to a raised awareness of their loved one’s diagnosis. Even if it is a major source of stress, being aware of diagnosis might help caregivers to deploy adapted coping strategies (eg, self-regulation, problem-focused coping, positive emotion) [37], whereas those using avoidance-escape strategies (eg, denial of diagnosis) may suffer from more negative long-term consequences (eg, inability to cope with behavioral problems) [38].

As in other studies [13], the experimental group participants’ self-perceived level of disease knowledge was significantly improved between baseline and month 3, with control group participants reaching similar levels at month 6. During the first 3 months, the program may have accelerated the learning process, but the control group may have improved their perception of disease knowledge at month 6 based on their experience and information from other sources (eg, websites, books, professionals/institutions, friends).

During the first 3 months, the program was highly used, in contrast with other studies [39], most likely due to mandatory reading of weekly sessions. Nevertheless, once the program was finished (after 3 months) almost none of the participants used the website, probably due to the stasis of the program’s content. No significant correlation was found between frequency or duration of website use and stress levels (PSS-14).

Qualitative Results

Our qualitative findings are comparable to previous works. Caregivers considered the program could be useful for people other than themselves [11]. They wished to receive personalized support, extensive information, specific assistance, and more communication with professionals and peers [40]. They preferred the topics offering strategies to maintain the PWAD’s autonomy and teaching skills for coping with behavioral problems [41], but were less interested in self-care [7]. Furthermore, specific subgroups of caregivers benefited from the program [42,43]. Some reported having a better perception of the disease or accepted diagnosis after the program [13]. In contrast with other studies [11], the most interested users were male caregivers. Probably linked to their preference for information and skills-centered interventions (such as Diapason) rather than emotional-focused ones [44]. In our study, daughters expressed more qualified opinions about the program compared to female spouses. In our view, because children caregivers are often active workers, they may recognize distance-based interventions as an interesting alternative for them. Moreover, female spouses facing greater caregiving challenges may be less aware of caregiving consequences for themselves [45] and may need more personalized interventions.

Strengths of the Study

To our knowledge, this is the first pilot RCT based on a mixed methods research design evaluating an online program for caregivers of PWAD. By using a mixed method research design, this program follows current methodological trends [46] using qualitative data to complement and contextualize RCT results [18]. Based on literature recommendations [47], Diapason adopted a multicomponent structure combining information and interaction between caregivers. Furthermore, this study met almost all the “best practice” criteria for a RCT (ie, randomization, intention-to-treat analysis, prior sample size calculation, and restriction of analysis to primary outcomes) [9] and controlled the intervention’s implementation errors [48]. Indeed, we paid particular attention to control implementation error. For instance, we controlled the information viewed by the caregiver according to a specific schedule. Additionally, the website content remained static during the study offering the same content to all participants. Finally, in order to avoid bias associated with the hypothetical imbalance of number of messages exchanged with the professional at the beginning and at the end of the protocol (eg, the first participants would not have benefited from discussions published later in the study), professionals only acted as moderators.

Limits and Lessons Learned

In spite of using different strategies, the recruitment for this study was difficult; only 38.0% (49/129) of prescreened caregivers were actually enrolled. These difficulties occur in Internet-based intervention studies [16], suggesting it may be due to caregivers’ attitudes toward these programs [49]. Nevertheless, their reluctance to face-to-face services was also described [49-51]. Thus, further studies about caregivers’ help-seeking behaviors and readiness facilitators or predictors are warranted [52].

Although face-to-face trials allowed the control of bias, isolated caregivers and those living in remote regions could participate more easily if the trials were conducted online only. In any case, replication with larger samples is necessary to complement our results. In addition, due to the heterogeneity of caregiver populations, we advise limiting the number of inclusion criteria [42] and the number of variables measured to reduce analysis bias. Finally, we pointed out the risk of bias owing to nonblinded assessments in this study [53].

This pilot study evaluated the first online version of the Diapason program. Qualitative results revealed little acceptance of the program and high expectations from caregivers. The Diapason program needs to evolve toward dynamic, flexible, and more customizable content based on a structure that favors interaction with professionals and peers, such as online community support [54].

Conclusions

Although a lack of statistical power prevents any definitive conclusions being reached about the efficacy of this program, the mixed research analysis provided us with valuable information for improving content and methods. Caregivers outlined high expectations about the program’s functionalities and showed little acceptance of our program. Dynamism, flexibility, personalization, and socialization appeared as key characteristics expected by caregivers. Overall, further studies about caregivers’ help-seeking behaviors and readiness facilitators or predictors are warranted.