The Invisible Frontier: Practitioner Perspectives on the Privacy Implications of Utilising Social Media in Mental Health Social Work Practice

In the digital age, traditional approaches to accessing information are disrupted by the ‘public’ nature of social media. The notion of accessing service users’ social media in social work practice is fraught with ethical issues around privacy and boundaries, yet lacks clear guidance from existing law and policy. This qualitative study sought to identify how mental health social workers were navigating these issues and how they thought practice could be developed. Ten mental health social workers, from one NHS Trust in England, were interviewed about their views and experiences around accessing service users’ social media without express permission. Semi-structured interviews were analysed using thematic analysis. Practitioners shared a variety of justifications for utilising social media, based upon statutory responsibility and professional values, but felt direct guidance/legislation and training would be helpful. A process of ‘digital reflexivity’ was outlined, which balanced reasons to access social media against considerations of privacy and the potential effects on the therapeutic relationship. Implications include the need for definitive guidance and training around if, when and how to access service users’ social media; inclusion of digital reflexivity in supervision and multidisciplinary meetings; educating service users about privacy controls; and a potential reconfiguration of theoretical boundaries to include the ‘public’ domain.


Introduction
Social work is a profession that relies on information about those it seeks to protect and empower. What practitioners garner from service users, their families, other professionals and clinical records drives effective risk management and the implementation of appropriate personcentred interventions. But what happens to this traditional flow of information-based practice when a much larger, more public and more connected data source comes along? This is the daunting question frontline practitioners are facing in this age of digital technologies, which 'play a vital role in the mediation of personal and professional human interaction and the access and distribution of information' (Young et al., 2018, p. 13). Social media platforms like Facebook, Twitter and Instagram are a part of daily life for many people (Foley and Woollard, 2019) and to varying degrees document people's personal lives and, critically, their vulnerabilities.
Social workers viewing service users' social media profiles is a highly contentious issue. It raises questions around privacy, morality, ethics, safeguarding and trust. A scoping review into the application of social media in social work identified the concept of accessing service user social media as an 'ethical panic' amongst the workforce (Chan, 2016,p . 272). Concerns around surveillance and regard for human rights on the one hand clash with safeguarding and support on the other (Foley and Woollard, 2019). If a practitioner commits to searching for clients on Teaser text The advent of social media has brought a new dimension to social work practice, with practitioners unsure whether service users' social media could or should be accessed as part of their information-gathering practices. Previous research has shown that there are positives and negatives in respect of safeguarding, privacy and overall well-being, but official guidance on accessing service user social media is limited. In this study, ten mental health social workers in England were interviewed about their experiences of navigating the issues around service user social media and shared their thoughts of how to ethically approach this issue. Many had engaged in the practice but were guided by professional values, knowledge and responsibilities, rather than specific policy or law, which was felt to be lacking. Practitioners attempted to balance the potential benefits of accessing social media against considerations of privacy and the potential effects on professional relationships. Implications include: The need for definitive guidance and training around if and when to access service users' social media.
Enhanced reflection and supervision that would address social media and its related issues.
Further consideration of service users' own understanding of privacy controls and how to empower them in this area.
social media and using any information discovered to inform assessments or intervention, there emerges two major issues. First, the extent to which this breaks a professional/private boundary, and eschews trust and principles of consent and confidentiality, thus endangering relationships with the observed individuals (Kolmes and Taube, 2014). Second, how exactly the gathered information should be interpreted-social media is hardly an objective source, and has multiple forms of expression, selfconstruction and behavioural implications that workers may not be equipped to analyse (Frey, 2018).
The option of disregarding social media as a squarely private part of a service user's life neglects the profession's statutory safeguarding responsibility to make use of all available information sources when conducting assessments pertaining to the protection of vulnerable children and adults (Sage and Sage, 2016), and there is a lack of clarity around whether social media profiles without sufficient privacy settings, that are in the public domain, are or to some extent should be, included in the definition of 'all available information sources' (Boddy and Dominelli, 2017;Cooner et al., 2020). This is complicated further when practitioners do not themselves look at service users' social media but have information from social media sources brought to their attention by the carers/ families of service users or by other practitioners or managers (Cooner et al., 2020). Are practitioners to ignore or to act upon pieces of information that are provided to them?
Despite acknowledging some of these ethical issues, the existing guidance tends toward avoidant or non-specific methods to actually tackle them. For example, previous research has advised that practitioners 'avoid practices that may blur [the professional/personal line]' (Groshong and Phillips, 2015, p. 148) or simply that practitioners 'be reflexive' (Sitter and Curnew, 2016, p. 273), whilst the British Association of Social Workers' (BASW) social media policy stipulates that 'decisions should be made on case-by-case basis' and does not directly reference accessing service users' social media (BASW, 2018, p. 10). More recently, BASW's Digital Capabilities: Ethical Considerations (BASW, 2020) and the NHS's review of digital implications on mental health care (Foley and Woollard, 2019) have highlighted the links between ethics, power and digital technologies/platforms, but how these are being acknowledged or utilised in the field remains unclear. As Cooner et al. state, 'although technology is being used, there is a significant level of confusion amongst the organisations, managers and social work practitioners in its usage, all of which is underpinned by an element of risk ' (2020, p. 151). Despite calls for the profession to carefully develop guidelines for professional development and practice (Boddy and Dominelli, 2017;Taylor, 2017;Cooner et al., 2020), there lacks a picture of how or if this is actually being implemented (Reamer, 2015).
A recent national survey suggested that this lack of clarity is a country-wide issue: just 6 of the 134 councils surveyed had any specific guidance on social workers' surveillance of social media accounts (Community Care, 2018a). Meanwhile, the need for such guidance continues to mount. For example, a serious case review into 'Child G' recommended that 'social-media checks would enhance and triangulate information' relating to vulnerable service users and help prevent harm (Wolverhampton Safeguarding Children Board, 2018, p. 3). Furthermore, research has shown that GPs and children's social workers have been conducting these 'checks' on their own initiatives (Clinton et al., 2010;Dolinsky and Helbig, 2015) and also that children's social workers were sometimes 'drawn in' to the practice by a workplace culture that normalised and justified such surveillance for purposes of child protection (Cooner et al., 2020).
This echoes Chan's scoping review, which found that workers were guided by 'their own sense of what is appropriate or not' when using information from client social media (Chan, 2016, p. 273). This 'sense' was guided by the profession's overarching ethical principles, such as the Health Care Professional Council's standards to 'act in an honest and trustworthy way' and 'respect confidentiality and maintain appropriate boundaries' (Health and Care Professions Council, 2016, p. 1). These suggest a workable framework but perhaps could be more comprehensive to provide clear guidance to practitioners on practicing within the evolving digital environment, and in particular if, whether, when and how to access service users' social media. The ramifications of any such access need to be scrutinised.
The 2000 Regulation of Investigatory Powers Act (RIPA) (Directed Surveillance and Covert Human Intelligence Sources Amendment, Order 2010) identifies the viewing of social media by agents of the state, including social workers, for investigatory purposes not to be surveillance and not to require RIPA authorisation if it only occurs once. However, repeat accessing of profiles by professionals without the consent of the service user is defined as unlawful surveillance (unless in certain circumstances) and requires, under RIPA, the prior consent of the individual (Judge, 2016;Cooner et al., 2020). In the absence of such consent, local authorities could be at risk if they do not have appropriate policies in place to protect against or mitigate this risk (Community Care, 2018a).
In the background sits the Human Rights Act's Eighth Article-the right to respect for private and family life. Crucially, the right is not absolute, and therefore can undergo interference from a public authority if deemed necessary for public safety or protection of others' rights and freedoms (European Court of Human Rights, 2018). But counter to this is whether the accessing of genuinely 'public' posts from service users would violate privacy rights, as some practitioners have reasoned (Kolmes and Taube, 2014). Guidance on Article 8 stipulates that services' continual collection/storage of data 'on particular individuals constituted an interference with these persons' private lives, even if such data were collected in a public place or concerned exclusively the person's professional or public activities' (European Court of Human Rights, 2018, p. 37). This then counters the 'public domain' argument.
One way of understanding the impact of such data-collection and surveillance is through a Foucauldian perspective: That the detriment of privacy contributes to a detriment in trusting relationships, as an assumption of non-confidentiality prevents comfort and confiding, especially with people representing authority (Fuchs, 2012;Lyon, 2015). This resonates with Foucault's (1975) fears of a society where citizens are treated as 'the object of information, never a subject in communication ' (p. 200).
This current study proposed to contribute to the evidence base by gathering the views and experiences of mental health social workers who may highlight nuances to the arguments around looking at service users' social media. Although it is arguable that surveilling social media profiles without permission for signs of 'relapse signatures' or exploitation may be justifiable as a preventative measure (Foley and Woollard, 2019), this must be weighed against the potential harm that service user discovery of such surveillance could provoke, for example, 'evidencing' paranoid delusionary beliefs about government control and oppression, and thus diminishing trust in the social worker or service and potentially affecting treatment adherence and service user recovery.
The study thus aimed to answer the research question: 'What are mental health social workers' views and experiences of utilising service users' social media in practice while still upholding individuals' rights and maintaining professional relationships?'

Design
This qualitative study consisted of semi-structured interviews with mental health social workers within a single NHS Trust in England. The study of privacy and social media lent itself to explorative, often philosophical, reflections that were best provoked by the open-ended and indepth nature of qualitative inquiry (Denzin and Lincoln, 2008). Given the relatively underexplored connections between social work, social media and privacy, this design was used with a view to procuring what Goldstein (1991) calls 'practice wisdom': a wide range of descriptive ideas and experiences from participants that reflect the holism and complexity of practice and ethics (p. 117). Data were collected in summer 2019, prior to COVID restrictions; interviews were conducted face-to-face as this enabled the researcher to develop a rapport with the participants and more easily discuss sensitive topics (Irvine, 2011).

Ethical considerations
The study conforms to internationally accepted ethical guidelines and relevant professional ethical guidelines. It received approval from an institutional ethics review committee, the NHS Health Research Authority (HRA) and the Research and Development Office of the NHS Trust in which the study took place. All participants gave written consent.
The research was undertaken as part of a Master's degree in Social Work Practice at the University of York, during which time the researcher was also employed as a practising social worker in the Trust in which the study took place. Such 'insider research' could have meant that known colleagues were recruited; however, to avoid additional challenges around responder bias and confidentiality (Chammas, 2020), recruitment was targeted at practitioners outside of the researcher's team. Anonymity and confidentiality were emphasised during recruitment and all contact was via the researcher's University email address. No one else within the Trust was informed of who had taken part or what anyone had said. All data were anonymised and pseudonymised and kept strictly confidential.
In the study information sheet and the applications to the University of York research ethics committee (SPSW/MTA/2018/48) and the NHS HRA, we explained that the purpose of the research was to understand how mental health social workers were responding (or not) to social media and its privacy implications, and did not intend, therefore, to critique or challenge the approach being used by participants. Equally, we did not encourage, or participate in, any use of service users' social media. The information sheet assured potential participants that their conduct was not under scrutiny and that all information shared about practitioners' uses of social media would be treated in strict confidence and would not be reported to the Trust. Reasons for this decision were twofold. First, we were aiming to gather and report on if, and potentially how, practitioners were using service users' social media in order to help inform future policy and practice, and this information may not have been forthcoming without such assurances; second, similar to the approach taken by Cooner et al. (2020, pp. 145-146), it is possible that any use of service users' social media could have been helping to keep a service user and/or their family safe and as researchers we did not seek to interfere with practitioners' judgement calls on this. Thus, any disclosure of practitioners looking at service users' social media accounts were not reported to authorities. Potential participants were informed that confidentiality would only be broken if the researcher believed there was risk of harm to the participant or another person.

Eligibility and recruitment
To be eligible, potential participants had to be mental health social workers within a particular NHS Trust in England. There were no restrictions based on the amount of experience working in mental health, though student social workers were excluded.
Purposive sampling was used to recruit mental health social workers (Padgett, 2008). A recruitment email was circulated by administrative staff to all mental health social workers within the Trust (excluding those in the researcher's own team). This included an invitation to participate and an attached information sheet detailing the study rationale, interview process and guarantees around confidentiality, anonymity and data security. A reminder email was circulated three weeks later. In addition, the researcher attended various team meetings within the Trust to introduce the study. Ten mental health social workers responded to the invitation and were contacted to discuss the study and arrange interview times and dates. As the target of ten practitioners was achieved, there was no need to employ a sampling frame.

Procedure
All interviews were conducted in-person, in offices without door-windows to protect participants' anonymity. At the start of each interview, participants were taken through the information sheet and consent form and invited to ask any questions.
Interviews lasted approximately forty-five minutes and centred around three main headings: Understanding of social media (what participants understood as social media, own usage). Social media in social work practice (use or consideration of service users' social media to inform practice, prospective positive and negative usages, potential ethical issues). Perspectives on privacy and policy within the Trust around utilising social media in practice. (Note, the social media policy in the Trust pertained only to employees' own conduct on socialnetworking platforms and made no reference to service users' social media.) The semi-structured topic guide allowed for flexible, participant-led discussion of the issues surrounding social media and privacy, which is essential when exploring emergent phenomena (Denzin and Lincoln, 2011).
Interviews were audio-recorded, with consent, on an encrypted digital recorder, then transcribed verbatim by the first author, anonymised and pseudonymised. Audio files were transferred to the secure University fileserver and immediately deleted from the digital recorder.

Data analysis
Interviews were analysed using the Framework method of thematic analysis, as this allowed the explorative nature of the discussions to be grouped into recurrent themes for more thorough deduction (Ritchie and Lewis, 2003). Each transcript, identified by a participant identification number, was coded in Microsoft Word by highlighting relevant text and using the comment function to add codenames and notes. The coding frame was initially devised from the topic-guide, and then refined iteratively following the re-reading of transcripts and each round of coding. Each transcript was coded twice; a sample was also coded by the second author to check for researcher bias and ensure the codes' robustness. Themes were identified and grouped through mind maps to determine the overarching themes then charted in Excel with rows corresponding to participants and columns relating to themes.

Sample
Ten social workers participated in the study from across the crisis team, psychosis recovery team, affective-disorders team, dementia service and Early Intervention in Psychosis. Participants ranged from newly qualified to senior positions; four were also trained as Approved Mental Health Professionals (AMHPs). Participant age spanned from early twenties to late fifties, with an average age of thirty-six years.

Findings
Five main themes were identified in the data: practitioners' justifications for (not) looking at service users' social media; understanding privacy; transparency, consent and relationships; the absence of policy, legislation and training; and future directions.

Practitioners' justifications for (not) looking at service users' social media
Seven of the participants disclosed that they had accessed service users' social media, without consent, as part of their practice. Six participants reported looking at Facebook accounts and single participants reported looking at Instagram and Twitter profiles and running a Google search on service users. Participants' justifications were based around four key reasons: 1. Urgency and risk, in the case of informing a Mental Health Act assessment and checking a person's identity. 2. Statutory responsibility, for example, investigating potential exploitation or fraud. 3. Seeking assurance, for example, checking that advice had been adhered to, checking if an intervention had been successful, or checking for known risks on a disengaged service user. 4. Understanding personal narratives, for example, seeking closure on a professional loss.
Some participants stated that they intentionally sought out service users' social media in order to better safeguard their clients. Many participants spoke about digital platforms as places where vulnerable adults could be especially exposed to exploitation or abuse, and this gave them specific safeguarding reasons for accessing their social media: We would have a duty-of-care to see if this person's being bullied over social-media or they're being threatened or being asked to provide indecent photos -loads of things that could happen. (P6) Another set of arguments for accessing social media was to gain insight into an individual in order to inform practice where that was not otherwise available (e.g. from clinical records or reportage from service users/families/other professionals): I feel like there were quite significant gaps in the professional information and this helped fill some of them. It made [this high-risk service-user] seem slightly more human, because I was getting this impression of this quite monstrous figure, but there were people on [this page] who clearly liked him and loved him and were worried about his safety. (P4) This spoke to the wealth of 'non-clinical' information that could be gleaned from looking, and its power to significantly shape perceptions, particularly if it was thought 'they might reveal themselves a bit more' through social media (P1).
Two participants spoke about 'resorting to' social media as an information source in urgent cases, on the basis of risk. For example, one explained that they had accessed their service user's Twitter profile to help triage an individual for a Mental Health Act Assessment: Because it was a higher risk case, I suppose it was about how quickly do we need to do this? And how can we actually get a hold of this person and get credible evidence? (P10) Other participants expressed discomfort, however, due to the subjectivity of social media: Other practitioners discussed how they did not proactively look at service users' social media, but if information was presented to them by a third party, then they could not ignore that information and may need to look at the account themselves. For example, one practitioner noted: A family member was telling us, this is on their Twitter feed, and that's where it initially came from, otherwise we wouldn't have gone there.
Practitioners who were less inclined to consider social media drew on criticisms that it was essentially 'false', due to the pressure to portray a positive image of oneself (P6), or that you couldn't verify that the actual individual was behind the profile (P7).
But the more consistent argument against accessing service users' social media was concern that the informational power it afforded could simply be unnecessary and-not so simply-constitute a breach of a person's private life, with far-reaching ethical consequences.

Understanding privacy
Participants were largely consistent in defining privacy as 'a way of controlling who you share information with' (P5), particularly personal details that did not relate to their involvement with services. Many argued that the Human Rights Act 'should extend to what you do on social media', therefore determining that accessing it could be an infringement because it involved reaching for information outside of the individual's sphere of control (P8). However, a counter-argument made by some participants was that information was only accessible when service user profiles were 'open' and in the 'public domain', and thus could be legitimately viewed by anybody: [If] we're talking about a high level of risk and we're highly concernedif it's open profiles I have no real issue with it. (P9) But on further reflection, the 'public domain' justification faded because privacy was potentially 'distorted' by digital platforms and individuals' lack of digital literacy and/or their vulnerabilities/conditions, which could impact their understanding of online privacy or how to set controls around it: Social-media has massively blurred those boundaries 'cause I guess it's encouraged people to share without the consequences being thought of... people who don't necessarily understand how public that information is when you share it. (P3) It was common for participants to think about privacy and ethics via analogy to physical practice, where going into service users' homes, reading sensitive case histories and asking about their 'intimate thoughts' proved that social workers already 'work with people's private lives all the time' (P3). These privacy-related elements were seen by participants to be essential in understanding service users' environments and assessing their needs. Reflections led to whether there was a difference between utilising private information in the 'real world' and the 'virtual world' of social media. One participant likened it to the approach toward medical records, where professionals only look for the information relevant at that time. But others disputed that comparison, noting that clinical records were intended for professional perusal and not authored by the service user, unlike Facebook statuses and Tweets. Furthermore, social media access lacked the same transparency with which current information-gathering practices were enacted, such as 'explain[ing] why I'm asking and how that might be used to help them' (P2).

Transparency, consent and relationships
Lack of transparency and consent when looking at service user profiles appeared to be the primary contributor to the sense of 'invading privacy' and feeling 'uncomfortable' (P8) or 'guilty' (P5) afterwards. Participants explored requesting consent from service users to look at their social media if it ever became relevant, but this was met with scepticism: 'I can't think of anyone who'd be like "yeah go for it".' (P4) Participants typically identified looking as an issue of power, further complicated in regard to mental health where service users may already experience paranoia or concerns around surveillance: Professional relationships were seen to be at stake if social media was accessed without serious consideration. Younger participants felt more at risk of this happening, as their digital familiarity gave rise to unconsciously utilising it: Many participants also expressed concern about using their own personal accounts to access service users' profiles, which might muddy professional boundaries and jeopardise their own privacy; however, creating 'professional accounts' was felt to be too much like 'state interference' (P8). Thus, interviewees found themselves in a persistent wrestle with the myriad potential implications on practice. This unresolved tension was met with a choral call for official guidance to make sense of it all.

The absence of policy, legislation and training
All participants recognised their knowledge gap of specific policy or law relating to the accessing of service users' social media. One participant explained how their AMHP training covered general information governance which contained 'legal and moral principles' that could 'transfer to social-media' (P10). In place of specific knowledge or guidance, the instinct to rely on the values and spirit of other legislation was universally reported by participants, with the Human Rights Act consistently quoted as something that reinforced privacy consciousness. In contrast, some participants acknowledged that other pieces of legislation directly necessitate breaches of privacy, for example, how the s. 135 warrant under the Mental Health Act would permit entry into a private residence. Relying solely on existing legislation and frameworks could therefore cut both ways and lacked the desired clarity.
Recently qualified participants spoke of receiving training about ensuring their social media profiles were private/closed, but gave no guidance around approaching service users' social media. Participants who had been in practice longer reported no training (despite one participant repeatedly requesting it).
The sense amongst most participants as to why guidance did not yet exist was that there was a top-down lack of awareness about the importance and complexity of social media issues: Because most of the people at upper management levels are not of the digital generation, there's kind of a lapse between what's going on on the ground and what needs to happen at policy-level to make the two meet in a professional way. (P4) Younger participants tended to see themselves as more comfortable and confident around new technologies compared to older colleagues. They argued that a work-culture shaped largely by older practitioners, whom they believed saw social media as 'ambiguous and vague and dangerous' (P3), posed a barrier to more in-depth consideration of social media in the guidance.
This perceived generation gulf was also noted in relation to service user populations; some participants stated that their teams did not see social media as relevant, due to working with mostly older people. However, those same participants saw social media issues cropping up with younger families/carers, and foresaw a future where even those under dementia services were engaged with, and vulnerable on, social media.
Two participants were aware of workers losing their jobs for unofficially engaging with service user social media, which led to a resigned belief that accessing service users' social media was the remit of police or child-protection, not adult services: It's not even acknowledged, and there's no mandate for us to even think about it... that's just not very acceptable in this day and age. (P3) The quote above illustrates the frustration of most participants when it came to discussions of current guidance, but this did spur on plenty of ideas for 'futureproofing' (P8).

Future directions
Participants wanted a clear framework to help determine the circumstances in which it would be acceptable and as-ethical-as-possible to look at a service user's social media. One participant envisioned this as a checklist: What do you need the information for? Why? What are you expecting it to tell you? Have you got the person's consent? If not, why not? Have they got capacity to give consent? Can you get the information another way, in a consensual way? (P10) It was suggested that with a clear policy on the circumstances in which practitioners could look at service user profiles then transparency and the process of seeking consent would be clarified: You need to be clear from the start... if they disclose something to do with social-media, and it's a safeguarding issue... then the worker might The Invisible Frontier Page 13 of 20 Downloaded from https://academic.oup.com/bjsw/advance-article/doi/10.1093/bjsw/bcab184/6362461 by guest on 08 September 2021 have to ask if they can see their social-media, maybe they'd have to consent to that. (P6) It was frequently acknowledged that social media issues could be raised with service users in initial assessments and incorporated into recovery/relapse-plans if it was significant to the individual. A participant who had appeared on a service user's Facebook Live stream during their Mental Health Act assessment, posited: Something like an advance-statement where you would say, in this instance if I became unwell, I would want my social-media usage to be controlled by my Mum or something, to prevent this going on Live. (P2) Finally, there was a suggestion that Information Governance departments could help to develop training and share 'lessons learned' from serious case reviews.

Discussion
The frequency of, and reasons for, accessing service user social media reported in this study reflect that previously reported (Chan, 2016;Community Care, 2018b). Echoing findings by Cooner et al. (2020), some practitioners reported proactively looking at service users' social media profiles, whilst others were 'drawn in' if they received information from third parties that they felt they needed to act upon. The issue, therefore, is not solely if, when and whether social workers could look at service users' social media profiles, but what they should do when presented with information by third parties about things that have been said or done on social media that might indicate exploitation, abuse, or signs of relapse. Notably, none of the participants mentioned the 2000 RIPA, which suggests an ignorance of the Act as also found by Cooner et al. (2020).
In line with Sitter and Curnew (2016), practitioners' decisions to look at service user social media were steered by 'reflexivity'. What the current study most strongly demonstrated is what that process of 'digital reflexivity' actually looks like: 1. Accepting that, in the appropriate circumstances, service user social media is an arena that mental health social workers may have a responsibility to enter into. 2. Acknowledging that there are major ethical issues that need to be understood first, and that these were reasons to approach social media in the 'right' way, not reasons to avoid it altogether. 3. Drawing from various pockets of personal and professional knowledge to ensure that they were entering into service user social media as proportionately and defensibly as possible (in the absence of clearer, specific guidance).
Reflexivity was driven by a professional sensitivity towards privacy that previous research had not identified, with strong perceptions that it is a right for people to have autonomy regarding their information, as historic definitions have suggested (Westin, 1967). Practitioners were not under the illusion that social media accessing could be benign, as many did in an earlier study (Kolmes and Taube, 2014), and instead engaged with privacy implications at a high level. The ethical risks were composed of two interlocking factors: the potential for systemic disempowerment (e.g. surveillance-type fears) and jeopardising trusting relationships, which could consequently harm mental health and service engagement, as others have suggested (Fuchs 2012;Lyon, 2015). The focus on these areas in the current study supports findings that they are of increased concern when working with individuals with mental health issues (Khoury and del Barrio, 2015). Cooner et al. (2020) explored arguments for social workers looking at social media in relation to child-protection work and some of the same issues arose in the current study in relation to mental health service users, in particular concerns around exploitation and abuse, as well as considerations that the subjects could lack understanding of the consequences of their words and actions in digital environments.
Some of the respondents' proposed practice solutions were largely informed by their grounding in mental health work, as ideas like incorporating social media usage into recovery/relapse plans drew clear inspiration from current mental health policy (Department of Health, 2008). It seemed that transparency was a higher priority in mental health, for the sake of maintaining trust with anxious, paranoid or disengaged service users. For some, there was a crucial reliance on family/ friends/carers to inform services if an individual's social media activity was causing concern for their mental health-this ensured professionals themselves did not have to surveil social media in the uncomfortable state-surveillance way that Romele et al. (2017) envisioned and yet could acknowledge the importance of 'insider knowledge' in the interpretation of individuals' actions (Doody et al., 2017).
Intergenerational differences within the workforce marred a cohesive team-approach to social media similar to that identified by Diercksen et al. (2013), but this was not a major barrier to practitioners' individual reflexivity towards privacy. What older participants felt they lacked in digital competency, they made up for in inter-legislative knowledge that generated privacy consideration. This can also be linked to a general societal shift in the familiarity with social-networking technologies and their intrinsic issues with privacy and power: high-profile stories of digital data breaches, manipulation or regulation (e.g. Edward Snowden's National Security Agency (NSA) leaks, Facebook's Cambridge-Analytica scandal, introduction of the UK's General Data Protection Regulation legislation (GDPR)), on top of increased dependence on data-driven technologies throughout the COVID-19 pandemic, have increased awareness of digital platforms as spaces of privacy vulnerability (Manokha, 2018;Goldkind et al., 2020).
Of particular difficulty for practitioners was hypothesising 'how' one should access service user social media if required. The anxiety of using their own personal accounts mirrored other studies that spoke to the reluctance to potentially blur professional/private boundaries in this way (Reamer, 2015). Yet, the discussion around social media distorting boundaries of privacy highlighted a need for more flexible theory to approach relationships in the digital age: the traditional '3 Ps of Social Pedagogy', for example, does not account for when the personal/private/ professional start to intersect like this (Eichsteller and Holthoff, 2011). Perhaps, the addition of a fourth 'P'-'Public'-is required?

Implications for policy and practice
Mental health social workers in this study, as in Cooner et al.'s (2020) study on child-protection social workers, were unanimous in their calls for definitive guidance on if, when and how to access service users' social media. The first implication, therefore, is the identified need for policy, with accompanying training, to clarify the boundaries within which social workers (and other frontline practitioners) must operate in relation to clients' social media profiles, in particular when working with vulnerable groups. Such explicit guidance would need to be developed with and ultimately adopted by professional bodies such as BASW (Boddy and Dominelli, 2017).
A second, related, implication is that digital reflexivity should be encouraged in settings like professional supervision and multidisciplinary meetings to help eradicate 'taboo' sentiments and share defensible practices when resorting to accessing social media. This could also aid other frontline professions and draw learning from their unique perspectives whilst acknowledging, as argued by Sage and Sage (2016), that 'policy cannot replace critical thinking, clinical considerations, or address all ambiguous practice situations' (p. 83).
A third implication focuses upon the exploration of preventative measures suggested by practitioners that could reduce service users' vulnerability to social media, for example, assessing social media risks in early assessments, educating service users on, and advocating for, the use of privacy controls, and using advance statements to safeguard against future non-capacitious social media usage (Boddy and Dominelli, 2017). These stem from existing laws, the Mental Capacity Act and The Care Act, so could be included in training around those legislations, as well as in dedicated digital training for social workers as noted elsewhere-for both students (Young et al., 2018) and experienced practitioners (Taylor, 2017).
A final implication was around reconfiguring theoretical boundaries, as it was acknowledged that social media has distorted conventional pedagogical demarcations of personal, private and professional. Greater acknowledgement of and research into the emergent 'public' sphere of life is required to better understand social workers' remit to engage with it.

Limitations
Though they represented a broad range of experience, roles and generations, the ten participants all worked within the same Trust, so their reflections on organisational perspective and policy limitations was not necessarily representative of social workers nationally. It is also likely that participants had a pre-existing interest in or engagement with the issue, meaning potential populations of the workforce who lack social media awareness or interest were not as well represented. It was also clear from the results that attitudes to social media varied in relation to different service user populations, which could not be meaningfully explored within this study's focus on mental health social work.

Conclusion
The study aimed to provide insight into how mental health social workers were responding to the challenges of service user social media and privacy in practice. From interviews with ten practitioners in one NHS Trust in England, results showed that service users' social media was occasionally accessed, always in thoughtful and cautious ways that prioritised privacy, yet without awareness of specific legalities. Practitioners highlighted a number of interwoven ethical issues-privacy, transparency, consent, trust, boundaries, disempowerment-and brought these together in a way that aided overall understanding. Through a reflexive approach to privacy and power, workers were able to identify several potential ways to work 'with' the apparent ethical risks, which upheld person-centred and least-restrictive practices. Whilst current guidance is still underdeveloped, practitioners' reflections indicate positive future directions: the need to develop a robust policy framework supported by staff training, encouragement of digital reflexivity across teams, and working in preventative ways that support service user privacy. Ultimately, in recognising the necessity of both privacy and protection, and in developing relationships and supporting well-being, mental health social workers might be amongst those best-placed to lead the profession through the frontier of social media in practice.