Examining the value of using naturally occurring data to facilitate qualitative health research with ‘seldom heard’ ‘vulnerable’ groups: A research note on inpatient care

This research note explores two inter-connected ideas. First, that the category of ‘vulnerable’ groups has expanded and has sometimes been used to exclude seldom heard groups from research. The underpinning protectionist polemic, whilst well-intended, has arguably led to disempowerment of the very groups they intended to safeguard. Learning lessons from advances in research with children can be utilised to develop a more inclusive and empowering approach with adults while simultaneously addressing safeguarding. Second, the note values the potential of using naturally occurring data as a mechanism for including vulnerable groups in research. An example is offered attending to frequently excluded inpatient voices in healthcare studies. The practical and ethical benefits of using naturally occurring data are explored for countering arguments against the inclusion of vulnerable groups using an example of autistic inpatient research from the UNITE study. Researchers are invited to make practical adjustments to maximise involvement opportunities, and a reminder that capacity to take part in research activity may be facilitated by these modifications. Naturally occurring data offer the possibility of understanding complex hospital practices without additional burdens of researcher-generated methods.


Context
In this research note, we consider the critical and pragmatic challenges of engaging certain groups in qualitative health research. Specifically, we untangle some of the taken-forgranted concepts in ethical discourses, such as 'hard-to-reach,' 'seldom heard', and 'vulnerability' while also offering alternative ways of collecting qualitative data that promotes the empowerment and inclusion of frequently excluded populations. For practical solutions to some existing barriers to be offered, it is important to unpack the underpinning assumptions made in research and ethics. This note therefore offers reflections on concepts, processes and ethics of undertaking research with 'hard-to-reach,' 'seldom heard,' 'vulnerable' groups, using examples from inpatient mental healthcare.
Historically, some groups were exploited and harmed in the process of advancing scientific knowledge. It was scandals in health and social science research that led to wider ethical governance to provide a level of protection for populations rhetorically positioned as marginalised, oppressed and at greater risk from mistreatment (Hunter, 2008). Marginalised groups may potentially be further stigmatised by a research spotlight which, while having ostensibly laudable aims to hold people in power to account, continues to perpetuate the construction of powerless groups as requiring special treatment (Tuck, 2009). Any perpetuating of poor treatment of disadvantaged groups is unacceptable and the governance structures implemented were crucial, yet the homogenising of vulnerability and the overt protectionist discourse has arguably resulted in exclusion and disempowerment.
Contemporary arguments, therefore, have sought to redress the consequential imbalance of exclusion afforded by the protectionism ideology in ways to maintain necessary protection from harm, but simultaneously empower those populations to have a voice in research, be involved in decision-making and promote person-centred thinking. This is most evident in research with children. Previously, protectionist discourses were utilised to exclude children from research, but this ignored the value of their voices and failed to respect their rights to participation (Campbell, 2008). Children are now viewed as active agents, included in decision-making, and their contribution to research is respected (e.g. UNCRC, 1989). Consequently, we have seen a paradigmatic shift as researchers move to do research with children, rather than on children (Clark, 2010). Arguably, we should take a similar approach with other groups to ensure voices are represented, science is advanced and those people are at the heart of our understanding of issues that impact their lives. This reflects the methods literature on diversity, inclusivity and equality in participation and means not excluding voices as a form of good practice.
Introducing 'hard-to-reach', 'seldom heard', 'vulnerability' and naturally occurring data A commonly deployed concept in qualitative research is that of 'hard-to-reach' and this has different meanings. Hard-to-reach can simply mean that certain groups are more challenging to access for research and refers to possible barriers like gatekeepers, ethical processes and communication. Slightly more contentious readings refer to these participants as hidden populations, inconvenient samples or those with low social status, disadvantaged and stigmatised groups (Duvnjak, 2013). Thus, the very notion of hard-toreach can be pejorative, and caution must be exercised in employing the concept (Williams, 2009) as it tends to position responsibility with those populations. In modern work, therefore, a newer conceptualisation of 'seldom heard' tends to be used instead, to represent the challenge of engagement rather than the invisibility of populations. Notwithstanding such critical and tautological discourses, pragmatically, there are certain populations within society that are more difficult to recruit to qualitative research, and barriers operate at a methodological, ethical and local level.
In this note, we reference a study exploring the communication experiences of autistic inpatients 1 during hospital ward rounds (the UNITE study). Whilst autistic voices were traditionally excluded from research like other disability categories, recently, there has been a spotlight on autistic voices due to the growing neurodiversity literature and the increasing impact of the autistic community. However, while the autistic voice is taking ground, there is still an absence of voice from hospital inpatients (as they are labelled clinically). Inpatients are still frequently excluded from research based on their perceived 'vulnerability', the complexity of accessing them and concerns about their capacity to provide informed consent (issues we critique shortly). While it is essential to safeguard this group from exploitation, it is equally important to ensure that they are not denied opportunities simply because of their autistic or hospital status. It is therefore incumbent on qualitative researchers to find ways to represent their narratives and include them, without harm or coercion.
We suggest one solution which is to utilise naturally occurring data to capture experiences without additional impacts. Naturally occurring data are data arising from events that would happen in natural environments regardless of researcher involvement (Potter 2002). Thus, videoing institutional settings, like routine hospital ward rounds, is an example. This has potential to capture the nuanced and complex non-verbal and verbal interaction and illustrates real-world practices as they occur. Naturally occurring data are typically contrasted with researcher-generated data (more typical in qualitative work) and are data generated specifically for research purposes, for example, interviews or focus groups (Potter, 2002). Such naturalistic research thus provides opportunities for notions of vulnerability to be critiqued and for participant voices to be foregrounded with less researcher contamination.

Autism and hospital care
Epidemiological studies suggest that autism affects approximately 1% of the population (Brugha et al., 2009) and hospital inpatient wards often admit autistic individuals for various reasons. Autism is regarded clinically to be a lifelong neurodevelopmental condition with qualitative impairments in social interaction, social communication and inflexibility of thinking (American Psychiatric Association, 2013). Given that differences in communication are a core characteristic, this may partly explain why autistic voices were not well-represented in research historically and why researchers relied on proxy involvement from carers and families.
Notably, autism research has been reinvigorated by genetics, as funding bodies have levied significant monies at biomedical work. However, the dominance of genetics has been critiqued, as 56% of funded research has been based on brain and biology, and 15% on aetiology in the United Kingdom, with similar statistics in the United States (Pellicano et al., 2013). Critics argued that genetics research threatens the existence of a condition which is beneficial to society (Waltz, 2008). Indeed, autistic communities and parents favour more interventionist research, experiential studies, work that facilitates educational need and qualitative work (O'Reilly et al., 2015;Pellicano et al., 2013).
Fortunately, autism research is growing and diversifying, partly because of an emerging autistic voice. Scholars have begun to recognise the importance of undertaking social research on the communication of autistic individuals as a way of demonstrating competence (see recent special issue; O'Reilly et al., 2016b) and work is recognising that it is not appropriate to consider autism a disease or deficit (Brownlow, 2010) or perhaps even a disability. Underpinning the growing body of critical thinking is a compelling autistic voice promoting greater involvement of autistic people in research as participants, co-producers and leaders of projects; yet, the same is not true of autistic inpatients in mental health hospitals, or indeed any inpatient group.

Exclusion of inpatients from qualitative research
There are clear differences between the position of autistic individuals in community settings compared with those in hospitals in relation to doing research. In the United Kingdom, persons can be detained in hospital under the Mental Health Act (MHA, 1983) for assessment and treatment against their wishes based on their best interests by professionals. This means they can be prevented from leaving hospital, have their possessions rationed or confiscated and be given treatment under duress. However, whether formally admitted or voluntarily, all inpatients are deemed as being in such poor mental health that hospitalisation is warranted because of the assessment of risk to self or others. It is because of this status, and the fact that under the MHA, they have an a priori 'disorder of mind', they are deemed to be vulnerable, or for some, incapacitated, and therefore require special protections in research or exclusion from it.

A critique of vulnerability
The notion of vulnerability has created tensions, and the assumption that inpatients are 'vulnerable' is insufficient to deny them opportunities, rights or voices. Indeed, increasing numbers of groups are being listed as vulnerable based on assumptions of risk, which has led to special protections (Wendler, 1998). This growing trend towards overprotectionism has led to disempowerment and exclusion of vulnerable adults based on the deontological principle of autonomy. Indeed, most discussion of vulnerability in research intrinsically ties to ethics and falls under the remit of research governance (Kenny et al., 2016). As Nickel (2006, p. 247) argued, 'we expect people to safeguard their own rational interests' and thus where their ability to do so is diminished, additional protection is necessary.
Notably, there is a contrast to the categorical definition of vulnerability whereby group membership is synonymous with vulnerability. For example, Wendler (1998) argues in favour of a comparative approach which considers context; and there has been recognition that the notion of vulnerability is indexical, not static (Nordentoft and Kappel, 2011). Wendler argues that not everyone in those broad categories should be automatically excluded from participating in research, and likewise, not everyone who is not listed as a member of a vulnerable group should be included. Furthermore, it is not merely the vulnerability characteristics of a person that matters, but how that vulnerability is important in the context of health research (Nickel, 2006). Evidently, the traditional focus on risk and harm fails to account for the very voices they are seeking to protect and fails to allow for the participant-centred ideology of the qualitative paradigm.
For autistic inpatients, these arguments are especially pertinent. Autistic individuals are viewed as vulnerable due to their mental health category, receptive and expressive communication differences, and their potential compliance may put them at risk due to possibilities of coercion. However, the decision to participate in research is different from other decisions. Taking the approach that vulnerability is context-dependent, researchers can adjust how they do research to ensure that even though participants may be 'vulnerable' in some situations (e.g. making medication decisions), they may be less vulnerable in the context of research.

Adjustments to lessen vulnerability
The direction of ethics is influenced by person-centred approaches in health, leading to new debates about how to best involve participants. Researchers can offer protections and be mindful of deontological principles because research involvement is different from other kinds of activities. Researchers can adjust practical arrangements to facilitate greater participant autonomy and overcome some of the perceived and actual vulnerabilities that the person may have. For example, simpler language can be used with shorter sentences and a slower pace, as well as putting in additional safeguards. Researchers can consult with gatekeepers, carers and experts to ensure strategies are implemented to maintain protection. Visual supports may also be used to help the participant follow conversations, stay focused and remind them of what has been discussed.
For good practice, qualitative researchers recognise the importance of iterative consent/assent, and this ensures that the participant is free to change their mind about their involvement. Critical here are the observation skills of researchers to see possible nonverbal signs that the person is distressed or uncomfortable and wants to withdraw even though verbally they may be saying otherwise; the problem of acquiescence or compliance. In relation to autistic individuals specifically, and anxious individuals more generally, researchers can give reminders about the duration left for participation. This information will help lessen anxiety and concerns about 'unknowns' and reduce any potential harms.
We argue that whilst protections have been in place for inpatients for excellent reasons, researchers can take appropriate steps to involve them in research with appropriate safeguards in ways that meet new ideologies of care and research. Furthermore, we argue that using naturally occurring data collection approaches employing video-based research offers a mechanism for hearing hitherto unheard voices and understanding complex healthcare environments within an ethical framework.

Naturally occurring data
Arguments propose that there is a need to 'advance methods and methodological discussions in the human/social sciences being beyond "about" disability' to be more centred on empowering voices (Lester and Nusbaum, 2018), and we argue that video-recorded naturally occurring data combined with researcher-generated methods is one such solution. Video-based research is well-suited to understand the complex nature of hospital activity to see what is actually going on rather than what people perceive is happening (Carroll et al., 2008). This benefit is succinctly described by the idea of video-data capturing work as done (WAD) rather than work as imagined (WAI) (Hollnagel, 2015). For example, research by O' Reilly et al. (2016a) demonstrated that in child mental health settings, professionals believed risk assessments were regularly conducted when, in contrast, the video data showed this was not always the case in practice. It is evident that video-based studies are increasingly being funded and granted ethical approval for sensitive research, including videoing decision-making in busy A&E departments (Stevenson et al., 2018), diagnostic sessions in autism (Lenne, 2018) and palliative care interactions (Pino et al., 2017).
The collection of naturally occurring data from routine events involving autistic inpatients has considerable practical and ethical benefits. Video and audio recordings provide a detailed picture of the evolving character of staff-patient interactions and the sequences of turn-taking underpinning the communication in real-world situated activities. Thus, the data demonstrate more authentic autistic voices as they unfold in everyday interactions that comprise and shape the natural hospital environment. The benefit is that recordings reveal the two-way character of the interaction so that a social understanding of the communication rather than an atomistic or individualistic view of communication. This shines a spotlight not just on autistic individuals' so-called 'deficits' in communication but on the emergent and co-produced nature of the interactions and their competencies.
Crucially, this means that researchers can closely examine communication as a multifaceted social action combining verbal expression, body language, gesture and pragmatics that could not be revealed through reported accounts alone. In short, it facilitates a perspective on autism that reveals communication competencies in vivo, the resources brought to bear upon the interaction, in a way that a reliance on reports from interviews does not have the power to expose. For the researcher, recording real-life happenings are a rich source of data to properly examine how autistic people actually communicate rather than relying on characterisations of their talk from neurotypical people. Ethically, this means that there is a better representation of autistic people's voices.
The natural character of the data also means that there are few or no additional burdens on the autistic participant which is also pertinent to ethics and the principle of not causing harm or maleficence. This is a significant advantage given that autistic individuals are likely to find neurotypical interaction through social communication, verbal and nonverbal expression, and understanding more challenging, which may impact their willingness to participate in conventional methods like interviews. Recent research illustrates that autistic people are more comfortable interacting with other autistic people than neurotypicals, a finding they attribute to the double empathy theory (Crompton et al., 2020). Researcher-generated methods are highly reliant on spoken and receptive communication skills and often on neurotypical and autistic pairings. In contrast, natural methods examine talk as it occurs normally, and so, there are no additional communication demands made on the autistic participant.

Challenges of doing ward-round research
We are currently conducting a study in a mental health hospital to explore communication practices during routine ward rounds. Ward rounds are a critical site for reviewing a person's progress, planning on-going care and longer-term management. They are a key part of routine hospital processes, established to review an individual's care with them, their healthcare team and family. Ward rounds usually take place in a designated meeting room rather than the bedside so that individuals are invited into the room. Despite their key role in hospital life, ward rounds have been under-researched, especially in mental health, and relatively neglected as the focus of quality improvement initiatives.
Overall, the study methodology is best described as a sequential video-reflexive ethnography comprising videos and interviews in a two-phase design. Following video recordings of ward rounds, the individual involved and two members of their staff team are interviewed about concrete communication events occurring in the ward round as well as wider communication and care issues for autistic inpatients. The dual approach means the researcher can both ask about their views about how well communication happens, perceptions as well as see practice and therefore contrast perceived communication with real-world communication. The value of including autistic voices from the natural environment, juxtaposed with critical reflections from participants has real potential to inform our understanding of inpatient care communication pathways. The process of undertaking this multi-method approach however is fraught with pragmatic and ethical challenges, as highlighted here.
These are challenges that have been acknowledged in other settings, whereby the rhetoric of vulnerability threatens to overpower the value and importance of voice. For example, Parry reported a review of ethics research and policy on videoing in health. She acknowledged three potential risks arising from (i) the potential for data sharing, (ii) the data are patient identifiable data and (iii) 'Hawthorne' effects, that is, that the interaction is detrimentally changed by the existence of a camera. The first two risks can be mitigated by protections such as encryption, digital effects to distort images and voices, and proper safeguards regarding with whom and where raw or disguised data can be shared. For this article, the final risk is probably most salient regarding the potential that video data have for capturing autistic voices. On this point, Parry finds that participants state they quickly forget the presence of video equipment, but if there is any effect, it is to enhance care rather than to impoverish it. Interestingly, the fact that people acclimatise to the presence of the camera has been acknowledged for decades (e.g. Speer and Hutchby, 2003) and yet, despite evidence, continues to remain a concern.
Further criticism levied at naturally occurring data has been that although the data is founded upon real-world conversation and direct individual's voices, they are analysed, interpreted and theorised by the researcher rather than autistic individuals. In response, we argue that the very close and detailed analysis made possible by natural data facilitates a more authentic account. Additionally, the involvement of autistic patients in looking at the data and providing their own accounts as they are given the opportunity to do during the interviews means the researcher can be more confident of the authenticity of their findings. As such, the reflexive video-ethnographic methodology facilitates the involvement of autistic voices more than a video-based research design alone because of its emphasis on collaborative conversations.

Concluding remarks
We began by acknowledging that historically, safeguards advanced the protection of participants and were motivated by good intentions due to past unethical practice. Since then, concerns have been raised that issues of vulnerability and capacity have unintentionally led to over-protectionism resulting in exclusion of certain groups from qualitative research activity. Attitudes are changing, and some groups are now successfully included (e.g. children) and these lessons need to extend to other groups. One of these groups has been the voices of hospital inpatients particularly in mental healthcare services, for example, autistic individuals.
The reasons for this exclusion are that in practice categorical approaches to 'vulnerability' mean that membership of a category such as mental health in-patient or autistic automatically confers a label of 'vulnerable' and strongly suggests incapacity and risk of coercion, regardless of the individual presentation and the specific nature of the decision to be made. We argue that ultimately, this has led to discrimination against such members being able to take part in research and to have their voices heard. Instead, we propose that researchers are more creative in how they do research and how they make reasonable adjustments to involve traditionally excluded groups, including consultation with relevant other parties who have the best interests of the individual in their decisions. We advocate that video-based naturally occurring data could be an important way that hitherto perceived vulnerable groups can be safely included. The article highlights the benefits of this approach as the research design does not place any additional burdens on the participant, yet it provides a mechanism to hear their voices as well as demonstrate communication competencies.
In conclusion, a more active autistic voice could be facilitated if naturally occurring data are combined with researcher-generated interviews as part of a video-reflexive ethnographic approach, whereby there is also critical reflection of the videos by the participant themselves as well as the researcher. It is possible to empower vulnerable adults to be involved in research and simultaneously safeguard them. There is a need to have an inpatient voice in research, policy and practice, and natural data can be a mechanism for respecting autistic voices and increasing our understanding of everyday healthcare practices.