Primary care clinicians actively promoting the signing up of patients to established and moderated OHCs and prompting them to interact with other members of that community can be considered a primary care digital social intervention. Patients’ level of engagement with an OHC could be important for gaining benefit, at least for some people, as there is some evidence suggesting that greater OHC engagement translates to improved outcomes [30-35]. Engagement with OHCs, though, is varied, with engagement predictors still being unclear [36]. To maximize engagement with a digital social intervention in primary care, current evidence suggests undertaking developmental work, such as co-designing the intervention with key stakeholders, performing pilot studies, and obtaining feedback on the ideal characteristics of moderators, on use of notifications, and on sharing of discussion topics and resources [17]. Co-designing approaches in health care are important as they set out to “design experiences of patients and staff,” thereby improving “day-to-day experiences of giving and receiving health care” [37]. This is especially true for primary care, a setting with unique features, as pointed out earlier.

Two ethical issues emerge from promoting engagement with OHCs in primary care: firstly, the safety of patients and their interactions with existing users of online communities; and secondly, the support by clinicians of those patients who engage themselves with OHCs.

The first point, the safety of OHC interactions, is discussed more in depth later (see “Potential for Harm” section). About the second point, that is, primary care clinicians looking after patients who engage with OHCs, evidence is sparse and concentrated around cyberbullying of patients engaging with social media in general (rather than with OHCs specifically) as well as around identifying misinformation in social media and either correcting it or warning patients accordingly [38-42]. The possibility for social media engagement to negatively affect psychological well-being due to encountering posts with negative feelings (eg, anxiety and worry) is described in the literature [43,44]. Our previous research has indeed highlighted some difficulties faced by highly active OHC users who take on a guiding role (superusers), with their role being stressful at times [45]. Superusers are both patients and caregivers of patients with asthma, have a wide age range, tend to take part in more than one OHC, and spend considerable time in a role sometimes similar to that of moderators. However, most HCPs are unaware of patient-superusers’ engagement with OHCs and are therefore unable to provide support [45]. This is in contrast with the general agreement among superusers that patient engagement with trusted and thriving OHCs should be promoted within health care [45]. Although some people might naturally be drawn to the role of superuser and enjoy the responsibility, this might not always be the case. The lack of support by primary care HCPs might put some superusers off providing advice to peers, causing OHCs to lose the members who significantly contribute to the cohesion of OHCs and the spread of self-management information. Losing OHC superusers and cohesion may mean that primary care patients joining the OHC may not experience the full benefits of online peer support. Our patient and public involvement (PPI) coauthor (BD) has indeed witnessed instances of superusers disappearing from OHCs due to disagreements with other members and high levels of stress.

The potential for social media to breach boundaries between personal and professional lives and confound roles in health care has been highlighted in the literature [21,46,47]. For example, our previous research revealed a case of 2 retired HCPs, one due to ill health and the other due to age, who assumed the role of OHC superusers [45]. They raised the issue of the need to develop a code of conduct within their registering bodies to engage with users in OHCs. Similarly, signing up to OHCs as users might potentially enable clinicians to “digitally track the personal behavior of patients” [21], including accessing data about patients’ lifestyle, complaints about the care received, or other information not otherwise disclosed to clinicians. However, “spying” on patients is widely recognized as detrimental to the doctor-patient relationship, against the principles of professionalism, and discouraged by existing guidance on social media use by clinicians [48].

With digital interventions in general (not specifically digital social interventions), there are fears that they might “dehumanize” the relationship between HCPs and patients [20,49]. Patients, for example, might increasingly use digital interventions as partial surrogates for information and support that they would have otherwise obtained from HCPs, thereby resulting in fewer interactions and weaker mutual bonds with HCPs. Conversely, participation in an OHC or social media platforms or engagement with digital interventions (in general) might trigger patient queries and requests for subsequent contact with HCPs; hence, promotion of online peer support may end up constituting an extra, time-consuming task for HCPs [20,50,51]. This is important, bearing in mind the increasing workload of clinicians in the UK primary care setting. A survey of OHC members found that patients tend to discuss information they come across in OHCs with their HCPs and that their relationship with HCPs may benefit from OHC engagement due to being more informed about their condition and thus better prepared for consultations [52]. Along the same lines, analysis of an online forum with stroke survivors showed that feedback from online peers facilitated a “distributed deliberation process” (ie, inherent need of patients to discuss with others) and ultimately “shared decision-making with clinicians” (“distributed deliberation” being a prerequisite for “shared decision-making”) [53]. Our PPI coauthor (BD) has indeed confirmed the importance of contacting HCPs before information obtained on the internet from peers is actioned (anecdotal knowledge about different treatment options, for example, when feeling that a condition is being suboptimally managed) and the reassurance that this generates.

One of the concerns with introducing digital interventions in primary care and in health care in general is their potential to widen health inequalities due to unequal access to digital technologies, associated with differences in patient characteristics [19,20,54-56]. For example, there have been fears that older age groups, ethnic minorities, people with reduced digital and health literacy, those from rural and deprived areas, and those at lower income levels might be excluded from the benefits of digital interventions in primary care (eg, video or email consulting, SMS text messages, mobile health apps, and access to detailed health records). In addition, significantly fewer adults with disabilities engage with the internet compared to people without disabilities [57].

As such, digital social interventions, although removing geographic and physical access barriers as well as some temporal barriers (ie, support is available 24 hours per day rather than restricted to certain times), also have the potential to increase health inequities. However, an overview of systematic reviews of trials looking at the effects of digital social interventions on health outcomes, behavior change, and health equity concluded that OHCs may be effective in promoting health equity [6,7]. Social media interventions have proven not to exacerbate health inequities in certain populations at risk for disadvantage, such as youth, older adults, those with low socioeconomic status, and residents in rural areas [6,7]. Indeed, 70% of homeless people are using social media, and the estimated penetration of broadband connection ownership and the tendency to be influenced by web-based content are wider in ethnic minorities [58]. Analysis of data from a stroke-related OHC found evidence that survivors with a wide range of disabilities successfully engaged with the OHC, and severe disability did not preclude participation [59]. Nevertheless, the concept of “digital inclusion” needs to be at the forefront when adopting digital interventions in primary care, as the risk of the “digital divide” cannot be completely eliminated [60,61], and primary care is usually the major commissioner of health care services and has historically been seen as well positioned to tackle health inequalities [62].

With the AD HOC intervention, our aim is principally to work toward providing evidence that the use of OHCs is indeed beneficial for primary care patients. Proving the benefits of online peer support will encourage primary care clinicians to promote it and patients to engage with OHCs. Although “inclusivity” is not among our primary goals at this early stage, we will seek “digital inclusion” by delivering the AD HOC intervention in areas of London with multiple ethnicities and high levels of social deprivation.

While messages are generally posted publicly (eg, Facebook), some platforms offer the option of posting privately at a one-to-one level (eg, HealthUnlocked [63]). Patients’ public posts in OHCs represent a rich data source to understand the needs, concerns, and preferences of members and to transform posted health information into knowledge that can be disseminated [64]. However, a number of ethical questions pertaining to privacy and informed consent need to be addressed before data from OHCs can be used for subsequent research analyses [65]. Professional bodies such as the British Psychological Society have released ethics-related guidance on internet-mediated research (ie, research involving the remote acquisition of human data through the internet and relevant digital technologies) [66].

Paying attention to patient privacy and confidentiality when social media are used in health care is vital due to the possibility of “patient information to be sold” [22]. There have been fears that promoting engagement with OHCs translates to patients being led to “put their views out there” without necessarily being informed (in advance) how their posts will be used [21,22]. The fact that patients seldom read through the terms and conditions of OHCs [22], along with the frequent difficulties patients have in managing their individual privacy settings [67] and regular modifications in OHC privacy policies without notifications [17,22], further restrict comprehension of privacy among OHC members.

OHC privacy is a complicated concept with little consensus on whether OHCs represent private pages or public documents [21,65,66]. In general, members of OHCs do not expect to act as “research subjects” and privacy is among their key concerns [65,68,69]. It has even been noted that although familiarity with a certain social media platform might motivate engagement with OHCs hosted on the same platform (eg, Facebook), there are patients (especially those affected by stigmatized conditions) who prefer to participate in OHCs hosted by platforms that allow maintaining anonymity [17]. However, levels of perceived privacy may vary depending on whether or not an OHC is closed (ie, requiring some sort of registration) and on the membership size, with larger OHCs more likely to be perceived as public spaces [65,70]. OHC moderators and administrators are a good starting point for advice about the attitudes of members toward privacy [66] and, as such, in the AD HOC intervention, we are having regular discussions with staff managing the OHC. Perceived privacy is crucial in determining the need for informed consent before patients are signed up to an OHC and any data produced by their engagement are analyzed. Members of online communities often perceive consent as required when they don’t expect “observation or reporting” of their activity to happen [65].

Regardless of how privacy is conceived, it is important to ensure the provision of transparent and clear details about data-sharing arrangements to primary care patients invited to participate in OHCs as part of research studies [2]. Within this context of transparency, researchers are advised to monitor the hosting platforms and inform patients who have signed up to OHCs about any substantial changes in terms and conditions or privacy policies [17]. There are increasing suggestions that consent is sought within a “digital/social domain,” especially when data from OHCs are being collected for research purposes (due to possible breaches of anonymity in processing these data) and ideally by offering multiple venues for consenting to ensure patients are aware of what and how their data are being shared [17,66,71]. However, obtaining consent in the sphere of OHCs is not a straightforward procedure, and there have been debates on prospective versus retrospective consenting methods in terms of benefits and drawbacks [65,72,73].

In the AD HOC intervention, we plan to follow a standardized consenting procedure in which primary care clinicians obtain consent at the time of promoting engagement and providing patients with login details to a well-known and moderated OHC after explaining the terms and conditions of the hosting platform to the patient. Face-to-face contact with clinicians will also ensure that competency for consenting is soundly verified, as well as excluding any potential cognitive or other problems that might make them vulnerable to coercion [66]. As part of the consenting procedure in the AD HOC intervention, patients signing up to the OHC are automatically excluded from optional OHC services, such as receiving emails (though they have the option to change these settings themselves by logging into their OHC account).

As with any digital intervention in primary care, the promotion of online peer support might entail potential risks for both patients’ safety (who could be vulnerable due to their long-term conditions) and the thriving of the wider OHC. These risks are important to be addressed as they might be detrimental to the reputation of primary care clinicians actively promoting OHC engagement. Risks for individuals might arise from (1) untrustworthy health information resources and advice in OHCs—these are common concerns of HCPs in works exploring their attitudes toward OHCs; (2) failure to escalate problems due to a “sense of security” as a result of the OHC-enhanced self-management and due to difficulties in appropriately articulating symptoms when writing posts in OHCs; and (3) social isolation and decreased motivation for engaging in in-person interactions as a result of excessive digital interactions [19,20,45,65]. It should be mentioned, however, that incorrect medical information is usually spotted and removed in established and well-moderated OHCs. Nevertheless, open-ended questions of whether available information is personally relevant and applicable to the individual who is using the OHC still persist. Risks for the viability of OHCs, in turn, might emerge from disruption in the community dynamics triggered by the addition of new patients who do not yet share the norms of “gathering toward a common goal.” Another potential risk could result from the intrusion of clinicians or researchers involved in the promotion of online peer support, who might sign up to the community and “interfere” in OHC communications (eg, to attract participation in research studies), although the latter is only a theoretical risk that has never been encountered [17,21,74]. To mitigate these risks, clinical guidelines for the promotion of digital social interventions in primary care should highlight established and well-moderated OHCs that are safe to be promoted, that is, that have been investigated to ensure their moderation and superusers are sound.

Indeed, our previous research reassuringly revealed that superusers in well-moderated OHCs showed awareness of the complexity of coping with asthma and the limits of their advice. They provided emotional and behavioral self-management support and often directed users back to HCPs for medical queries [45,75]. In addition, researchers’ professionalism and their interest in studying peer-to-peer interactions within an OHC reduce the chances for intrusion and interference in the community’s communications, as any interference might alter the content and type of interactions (eg, formation of networks of communication in the OHC) and thus bias the results of research efforts.