This research was primarily conducted online, with participants accessing the survey through a link on the forum’s home page. There were 8 follow-up face-to-face or telephone interviews. Ethical approval was obtained from the University College London Research Ethics Committee.

The forum studied was Talking Point, the UK Alzheimer’s Society’s online forum for carers of people with dementia [11]. Its home page describes it as “an online support and discussion forum for anyone affected by dementia. It’s a place to ask for advice, share information, join in discussions, and feel supported.” The forum is well used; on August 20, 2013 there were 873 active members, and at the time of visiting (18:00), there were 444 users online (87 members and 357 guests). It contained a number of different subforums, including “I care for a person with dementia” forum, which was the most active. Figure 1 shows a screenshot of discussion threads on Talking Point.

A recent systematic review [6] found that networked technology interventions for carers of people with dementia had “moderate effects on improving carer stress and depression.” Assuming a medium-small effect size (0.35), a sample size of 67 was required for this analysis [12]. The actual sample size achieved was 61, which gave a power of 0.80 at an alpha of .05 to detect an actual effect size of Cohen’s d=0.36.

New users of the forum were invited to take part in the research through an advertisement on the home page, from which they could click on a link to the consent form and survey. An email was also sent to users to alert them to the research, which was also promoted on the Alzheimer’s Society’s Facebook and Twitter pages. At baseline, participants completed 3 standardized measures: the 9-item Patient Health Questionnaire (PHQ-9), the 7-item Generalized Anxiety Disorder scale (GAD-7), and the Scale for the Quality of the Current Relationship in Caregiving (SQCRC). They also answered demographic questions and questions about their role as a carer.

At 12 weeks after completing the first survey, participants were emailed a link to complete the 3 standardized measures again, as well as questions about their use of the forum over the 12-week period. Those who did not complete the second survey within 2 weeks were sent an email reminder.

At baseline, participants indicated if they were interested in an optional interview about their experiences of using the forum. Interview participants were selected according to inclusion criteria and in order to sample a range of different users, in terms of their sex, age, person they were caring for, and length of time they had been caring.

In accordance with the study risk protocol, participants scoring in the severe range on the PHQ-9 or the GAD-7 were advised via email to contact their general practitioner. A total of 49 participants were sent this email.

The Patient Health Questionnaire (PHQ-9 [13] is a widely used 9-item measure of depression. Scores of 20 or more suggest severe depression. It has high sensitivity and specificity for diagnosing depression [13], good internal consistency, convergent and discriminant validity, robustness of factor structure, and responsiveness to change [14].

The Generalized Anxiety Disorder scale (GAD-7 [15] is a widely used 7-item measure of anxiety. Scores of 15 or greater indicate severe anxiety [15]. It has good sensitivity and specificity for GAD [15] and is a valid and reliable measure for detecting GAD in the general population [16], as well as social anxiety, panic disorder, and posttraumatic stress disorder [17].

The Scale for the Quality of the Current Relationship in Caregiving 14-item version of the SQCRC (SQCRC-14) [18] asks carers about their relationship with the person that they are caring for, giving equal weight to positive and negative aspects. A higher score implies the presence of warmth and affection and the absence of conflict and criticism in the relationship. The measure has high internal consistency and good face validity [19], but has had little further psychometric investigation.

The qualitative interview asked participants about their experiences using the forum. It covered what they found more and less useful, how they liked to make use of the forum (eg, whether they preferred to write posts or read other people’s), and whether they felt that it had made a difference to them and their role as a carer. The rationale was to capture the variety of users’ experiences related to both our main outcome variables and also user-defined outcomes. The interview schedule was developed with reference to the literature on peer support, as well as through discussion with the forum’s manager and the charity’s head of evaluation. The manager and volunteer moderators provided feedback during the development of the interview schedule. Seven interviews were conducted over the telephone and one was face-to-face. The interviews lasted approximately 40 minutes.

It is recommended that the researchers’ perspective be disclosed to enhance the credibility of qualitative research [20,21]. The first author, who conducted the interviews and led the analysis, was a white, middle-class, female clinical psychology doctoral student in her twenties. She had no direct experience of caring for a friend or family member with dementia, but had seen the impact of the caring role on other people and was aware of the challenges and stress that this role can bring. The other 2 authors were white, male clinical psychologists, who were generally favorably disposed toward mutual and peer support. All authors attempted to bracket their preconceptions during the analysis.

Paired sample t tests were used to analyze changes in anxiety from baseline to 12 weeks, and depression and quality of relationship for all participants who completed the survey at 12 weeks. Nonparametric correlations were conducted to examine the relationship between forum usage level and changes in outcome.

Semistructured interviews were transcribed verbatim, with all identifiable data removed to preserve anonymity. They were analyzed using thematic analysis [22], taking an inductive, data-driven approach. Interview transcripts were repeatedly reviewed to become familiar with the data and to ensure that information was represented accurately. During this process, an initial list of ideas was generated. These were grouped into codes and then brought together into meaningful themes, which were then checked against initial codes and the overall dataset, and amended in some instances. Finally, themes were organized into overarching domains. Credibility checks [21,23] involved a third party examining sections of analyzed interview transcripts and providing feedback on codes, themes, and domains.

Respondent validation [20] was used as a further credibility check. Each interviewed participant was emailed a summary of the themes generated from their interview and asked to complete a feedback form. Seven of the 8 interview participants responded to this, either by completing the form or simply by replying to the email, and said that they felt that the list of themes was a good summary of their interview.

The quantitative and qualitative data were analyzed separately, using a concurrent analysis approach [24]. This allows each analysis to stand on its own as an independent perspective on the data, rather than have one influenced by the other.

From the 61 participants who completed the 12-week survey, 40 expressed an interest in being interviewed and 13 were invited (with the aim of having a variety of backgrounds and forum usage). Of these, 8 were able to be interviewed. Table 5 gives their characteristics.

Thematic analysis of the 8 semistructured interviews resulted in 18 themes across 3 domains (Table 6). Each domain is discussed subsequently, and illustrative quotes can be found in Table 6.

There was an improvement in the quality of the relationship with the person with dementia. There was no change in users’ depression or anxiety over the 12-week study period in contrast to other studies of the outcome of online support [27]. Interview participants reported a range of positive experiences and benefits from using the forum. Limited negative experiences were also reported.

The hypotheses that after 12 weeks of forum usage, users’ anxiety and depression would decrease were not supported, although depression and anxiety levels did not increase. This may indicate no effect of forum usage. However, interpretation is dependent upon understanding of the normal expected trajectory of carer mood, and this is complicated because models predict anything from deteriorating to increasing psychological health over time [28]. Additionally, the progression of psychological distress in carers is affected by a range of factors, including intrinsic variables, such as carer gender [29], and more contextual factors, such as the relationship between carer and care recipient [30], which varied across the sample. It may also be that participants’ low usage of the forum accounts for these findings because there is evidence of a dose-response relationship in online forums for mental health conditions [31].

Qualitative results provide some insight into how the forum may have improved the carer’s relationship with the person with dementia; for example, through carers learning more about how to interact with someone with dementia and feeling as though the information that they had gathered helped them to become a better carer. Many participants’ forum usage levels were low, and it is possible that being aware of the forum as a resource that was available should they wish to use it was enough for some participants to improve the quality of their relationship with the person that they were caring for. In addition, it is possible that those who signed up to the forum were individuals who had made the decision to learn and develop as much as possible in their role as a carer, and these individuals were motivated to have a good quality relationship with the person with dementia. It is also possible, however, that the improvement in scores on this measure simply represents regression to the mean.

For all 3 measures, most participants’ scores neither reliably improved nor reliably deteriorated. More participants reliably improved than reliably deteriorated. Although the present study only covered a 12-week period, this finding is encouraging given that relationship quality is likely to deteriorate over time [9,10]. Research into the trajectories of anxiety and depression is more mixed [28], but this is also an encouraging finding in relation to these measures.

The 8 carers interviewed were generally very positive. Their reasons for using the forum echo previous research, which found that the primary functions of such groups are to exchange information, connect to others, and to obtain emotional support [32].

For interviewees, the fact that other forum users were in a similar situation to them had a range of benefits. Carers of people with dementia frequently report feelings of isolation and inadequate social support [33], but interview participants reported that the perceived similarity of other users reduced isolation and loneliness, allowed them to “let off steam,” and feel more normal and understood. These experiences, which correspond to Yalom and Leszcz’s [34] therapeutic factors in group therapy (universality, altruism, guidance, imparting information, and catharsis) are similar to those reported in a study of mental health service users’ information needs [35].

Several interviewees noted that seeing that others were worse off helped them to feel better about their own situation. No interviewee mentioned comparing themselves to individuals better off than them. Benefits of downward comparison have been seen found in online support groups for heath conditions [36,37]. However, for several participants, reading the stories of other users who were in worse situations was distressing.

Interviewees reported deriving benefit from helping others, in-line with the helper therapy principle [38], which is unique to peer support. Providing help can increase feelings of competence, equality, social usefulness, independence, and social value, and allows individuals to view themselves as having strengths as well as needs [39]. The reported benefits of lurking (using the forum without actually posting) are echoed in another study [40] that found lurkers on an online support group reported a range of benefits indicating that reading messages may be as beneficial as interacting with the group.

Several interviewees reflected that the type of information and support available on the forum was either not available to them elsewhere or that they would not wish to obtain it elsewhere. The forum may either work as a complement to other services, or offer a service to people who do not access other services. A number of interviewees were not the main carer of the person with dementia; therefore, perhaps, more traditional services are not available to them.

The forum’s anonymity provided some interviewees with the freedom to say more than they would be able to otherwise. This online disinhibition effect [41] brings the advantage that users may be more open about their feelings allowing access to emotional support. However, several interview participants also considered some posts to be judgmental or inappropriate. Helpful moderators (as found on Talking Point) are important to ameliorate this more toxic online disinhibition effect [41]. Additionally, some participants said that the forum allowed them to read only the posts that they wanted to, allowing them to avoid distressing posts.

Several participants noted that reading posts helped them to learn more about progression of dementia and to be better prepared for future problems. However, others felt that some posts were too distressing to read. With a disease that involves worsening of symptoms over time, more information and less uncertainty about the future can provide relief, but it can also cause distress and more information is not always positive. This tension has also been found in support groups for people with motor neuron disease and their carers [36].

Participants also noted that the forum taught them more about how to interact with someone with dementia and to become a better carer or companion. Consequently, the forum may have benefits for the person with dementia through improved care.

There is an apparent discrepancy between the positive qualitative experiences of the site and the modest pre-post changes. This may be because of the contrast between psychological outcome variables measured quantitatively and the qualitatively reported benefits, which were largely related to general therapeutic group factors [34] as well as to practical advice and learning. This raises questions about what carers hope to gain from the site and provides useful information about outcome measures for future studies, a contentious issue in this field [42,43].

There were several limitations to the study. There was a low usage rate that affected the ability to reliably assess the impact of the forum and skewed the data. However, this may reflect real-life usage and, as a number of participants noted, a benefit was being able to visit the forum as little or as often as desired. Additionally, inclusion criteria for interviewees excluded those who used the forum less or disengaged from using the forum. Their views may have differed from those who were interviewed, but those interviewed still provided useful information about people who do engage with the forum, of whom there are many. Only 8 participants were interviewed, which is a relatively small number, and it is possible that a larger sample of interviewees would have provided different or additional data. Whether these 8 interviewees’ experiences reflect the experiences of the wider group of forum users is difficult to ascertain.

Participants received various levels of additional formal and informal support in their roles as carers, which may further complicate and limit the scope of conclusions that can be made about the specific role of the forum. In addition, only approximately 3% of possible participants took part in the research. These sampling issues inevitably limit the extent of conclusions that can be made, and mean that findings from the quantitative data, in particular, are tentative. The lack of a control group in the present research means that these outcomes cannot be compared to a group of carers who did not have access to Talking Point; therefore, findings cannot necessarily be attributed to forum usage as opposed to other factors. Future studies would benefit from using a randomized controlled design.

Finally, the caregivers in our sample tended to be late in their caring trajectory (adult child vs spouse caregivers) with the median age suggesting likely sandwich generation carers. These caregiver variables have systematic effects on caregiver mental health outcomes [44]; the number of participants scoring in the severe range for depression was relatively low [45] although this was not the case for anxiety [46]. Given these sample characteristics and evidence of low depression rates, caution is needed when generalizing the results to the wider population of caregivers.

The qualitative data indicated unique benefits from peer support/group therapy, such as not feeling alone and feeling understood through shared experience. Some of these benefits are unique to online support, such as being continually and flexibly accessible as well as enabling honesty within an anonymous online environment. Therefore, clinicians may wish to direct carers to such online peer support forums. In an economic climate in which services are often being reduced, online peer support is likely to become more prevalent and may be the only support that some carers receive. It is important that this area continues to be researched so that carers can derive maximum benefit from online peer support forums.

Future research could investigate further what outcomes are important to carers and specifically consider these outcomes in the evaluation of online carer support interventions. In addition, more research needs to be conducted into whether different types of carers (eg, according to gender and ethnicity) derive different types or levels of benefit from online peer support forums because previous research (eg, [47,48]) suggests that interventions do not affect carers uniformly. In order that carers receive the best support possible, it would also be worthwhile investigating whether there are certain types of support that work well when offered in conjunction with online peer support and vice versa. Interventions for carers of people with dementia are often most effective when they are multicomponent in nature [49,50], although this makes evaluation and attribution of any observed effects more complex [50].