Ethical dilemmas and reflexivity in qualitative research

Procedural ethics, the formal approvals required for a study to commence, are dependent on the justification for the study and a stated commitment to adherence to ethical principles. Even before embarking on the formal application, the good intentions of the research need to be carefully weighed up against the vested interests of the researcher involved. Tracy [2] advocates that researchers practise (self-) reflexivity ‘even before stepping into the field … [to assess] … their own biases and motivations’ (P. 842). An illustration of this is provided by JB, an experienced academic and researcher in postgraduate medical education. He reflects on his role conflict as gatekeeper in pursuit of integrity and altruism in research practice. Gatekeepers are those who give access to a research field, or, ‘are able to control or limit researchers’ access to the participants’ [12, P. 452]. Another aspect of the gatekeeper’s dilemma is described by SJ in relation to her role in managing the research of others. This follows JB’s story.

JB: The Universities UK ‘Concordat to Support Research Integrity’ [13] calls for integrity and research excellence throughout the duration of a project. However, one, often hidden complexity, concerns the personal motivation of the researcher, a troubling question for me in leading a research program funded by a postgraduate medical department. Despite beneficent intentions and careful adherence to research governance processes, I retain lingering doubts as to the extent to which hidden coercive influences impact on participant recruitment.

Medical education research is relatively poorly funded and time pressured as many researchers are practitioners with substantial academic and/or clinical roles and responsibilities. As such, these researchers need to reflect upon, and be transparent about, their competing interests [14, 15]. Although this issue has been highlighted for some time in the field, change has been slow, as evidenced by the relatively low number of competing interests disclosed in a survey of medical education related journals [16]. Walsh [17] suggests that medical education researchers may not deliberately fail to disclose, but rather fail to recognize their own competing interests due to the ‘bias blind spot’.

Bearing this in mind, I am conscious of relying heavily on the goodwill of participants with whom I may have relationships before, during and after the study. I believe that the participant should be exercising judgement based on the validity of the study in which they are consenting to participate, rather than on that relationship. Despite the best intentions to contribute to new knowledge, there must be an open and honest realization that research outputs are key benchmarks to measure how effectively researchers are performing. This dilemma was brought home to me while teaching postgraduate students, one of whom posed the question, ‘Can research ever be truly altruistic?’

Initially I was unsure of how to respond, but on reflection, admitted that in my experience, despite endeavouring to be honest and truthful to our participants, there may be a conflict between our personal motivations as researchers and reliance on the goodwill of participants. These reflections are unlikely to lead to any dramatic changes in practice; however, recognizing this potential conflict can at least remind us never to take for granted the commitment of willing participants.

SJ: Balancing the aims of beneficence, the value of research to the medical education community, with the principle of non-maleficence, avoiding harm, is an ethical conundrum which I have frequently encountered in managing an undergraduate medical education program. As a member of the medical education research community I am keen to support research, but am conflicted when faced with frequent requests to include our students as study participants. This particularly arises with survey research which is often the subject of such requests. The issue of survey fatigue is well-recognized [18], so much so, that my institution instigated a policy to restrict this [19]. The policy aims to prevent survey fatigue and ensure that students remain well-disposed to completing important internal and external surveys including the National Student Survey [20], the results of which are important key performance indicators for the institution.

On becoming director of a masters’ program in health professions education, tensions arising from my role as gatekeeper were exacerbated. The needs of the postgraduate students to recruit study participants potentially conflicted with the interests of the undergraduates. The availability of undergraduates as study material for postgraduate students and career academics is not new, and two decades ago undergraduate psychology students were described as ‘a captive population with little power’ [21 P. 74]. In a similar vein, Keune et al. [22] noted ethical issues arising in a scenario whereby a surgical resident introduced a mandatory team/trauma simulation training session. There was no indication that this session was also the basis of a research study until the trainees were presented with consent forms on arrival, clearly suggesting possible coercion. Another study explored the motivation of Indian medical students to participate in research conducted in their university and hospital learning environments [23]. Of the 300 participants, 61% admitted to participating against their genuine wishes, of which 26% agreed because a faculty member had asked them and 4% because they had not appreciated the right to refuse.

These examples demonstrate a tension between protecting potential participants and respecting their autonomy to choose whether or not to participate in research. One solution as gatekeeper may be to give agreement for participants to be recruited via a general email to a whole cohort, but prohibit purposive sampling of a specific group who may feel more pressurized. Managing access to students as participants highlights for me the ‘grey area’ between protecting the vulnerable and behaving paternalistically in upholding ethical principles.

AC: My research followed principles of naturalistic inquiry [24], capturing the phenomenon of teaching and learning in the operating theatre, as it happened, through ethnographic observation, video and audio-recordings [25]. One challenge of capturing naturalistic data is gaining prior consent of participants, as it can be difficult to predict who may be co-present within an operating theatre. In my study there was potential for inadvertent participants to be captured during data collection, defiling the research principle of autonomy—the right to determine participation or non-participation [5].

One approach to uphold autonomy was in the choice of microphone for audio capture: the ‘XTag RevoMic’ which has a short capture range. This meant that talk at the operating table of consenting participants was recorded but that others entering the operating theatre were not inadvertently recorded. The recording was taken from a camera placed within the operating lamp for open surgery, and from a laparoscopic (internal body) camera so that the site of the operation was captured but not the faces of those present. Assurance of this was key to maintaining trust between myself as researcher and the clinical teams involved.

Potential patient participation in the study also raised issues of autonomy. The study explored interactions between surgeon and trainee, but the patient’s body cavity formed the backdrop for video recordings of hand movements and the interactions involved. This meant that the patient would be co-present during the data collection episode although under anaesthetic. Murphy and Dingwall [26] state that ‘one must distinguish between those for whom the research is likely to be consequential and those who are tangential to it’. The ethics committee deemed that patient participants were peripheral, and that for this non-interventional research, patient consent was not required. General Medical Council (GMC) guidance specifies that doctors may use recordings such as laparoscopic video streaming or images of internal organs for secondary purposes (such as research), without seeking consent from patients, provided that the recordings are captured as part of patient care and are anonymized [27]. In this study recordings were captured by the research team and therefore were not a routine part of patient care.

The clinician participants, in particular the nursing staff, expressed very strong feelings that the patient should also be recognized as having the right to consent in making an informed judgement in whether or not to participate. Despite this presenting an extra hurdle for me as researcher, in my surgeon role I was used to gaining patient consent for operations and agreed that patients should be given the right for their procedure to be part of the study or otherwise. I discussed the study with patients, providing a bespoke ‘plain language’ Participant Information Sheet and consent form. Inevitably this raised separate ethical difficulties as patients were anxious that by consenting, they were agreeing to be operated on by a learner surgeon (given that the study was investigating clinical teaching and learning). I had to deal with patient vulnerability sensitively, explaining that participation in the study or otherwise would have no bearing upon their clinical treatment, or the person performing the operation, emphasizing that this was a non-interventional study. In this way the autonomy of patients to choose to participate was respected. No patient declined to be part of the study on this basis.

AMR: Power asymmetry is a feature of research with the balance generally considered to be in favour of the researcher who directs the process while the participant responds [28, 29]. Ben-Ari and Enosh [30] argue that this power is actually co-constructed through the process, as participants exert power in shaping knowledge through choosing what to reveal. An illustration of such power dynamics occurred during my study on partnership working in the development and design of a new healthcare degree, commissioned by the Health Authority [31]. As program leader, I led the development and delivery of the curriculum and managed the team involved. This brought me the benefit of access to the university team and senior healthcare managers as participants, but also role conflict in managing relationships as the study progressed.

I was acutely aware of my potential biases and power dynamics as both researcher and manager; when interviewing members of my own team I was sensitive to the ‘ethics of care’ [1]. A specific dilemma arose during an interview with a lecturer when I asked a question from the schedule. Rather than respond to this, the interviewee took the opportunity to air personal grievances. The interview may be viewed as a dialogic process, and my sense of control of the dialogue was challenged by the participant taking ‘counter-control’ [11] in deviating from the schedule. I was unsure of how to react, feeling it inappropriate as researcher to challenge or make a direct response. In order to avoid distractions and regain control, I steered the conversation back to the interview schedule.

Although in the moment I felt that I was acting ethically in prioritizing the quality of the research [32], the issue troubled me later. It seemed inappropriate to raise the grievance with the interviewee in my management role after the event as I had assured participants of the confidentiality of the proceedings. With hindsight, given the power asymmetry in the manager-employee relationship, I now believe that seeking a later opportunity to discuss the grievances would have respected the autonomy of the staff member and provided duty of care. On reflection, a different interviewer may have avoided such circumstances occurring, but in this instance I was expected to conduct my own interview as part of the doctoral training.

JS: My doctoral study involved conducting a multi-site, observational, longitudinal research study, focusing on clinical reasoning development in final year medical students as they transitioned to junior doctors¹. This required managing a number of boundary issues through my overlapping roles of clinician, researcher, teacher and mentor. As I observed participants in both simulated and clinical workplaces, the ethical challenges of these roles developed. I had to ensure that participants could opt in or out of various clinical scenarios as well as at different data collection points. I also had to be mindful of my own potential biases and the need to treat them equitably, neither advantaging nor disadvantaging them as participants in the study. Furthermore, I had to consider the secondary participants who were the patients and ward staff in each workplace. Given the opportunistic and serendipitous nature of the workplace, audio-recorded consent was initially sought in the moment from secondary participants, followed by full written consent.

My ontological position, in particular my predominantly insider role, required a high degree of self-reflexivity [2]. Despite extensive prior approvals and permissions, I anticipated and faced ethical dilemmas arising from my relationships as both an ‘insider’ (a middle grade clinician), and an ‘outsider’ (clinical settings beyond my own workplace and field of expertise). The potential for role conflict also arose between my clinical and researcher role. It was agreed that should an emergency occur within the workplace, I would intervene if necessary in my clinical role in accordance with the GMC’s Good Medical Practice [33]. Furthermore, I would identify a senior member of staff in advance for each shift to whom I could report any instances of unsafe medical practice. Participants were made aware of my role and ethical position as this may have influenced their decision to be involved.

Ethical dilemmas arising during the study included critical incidents ranging from observing unsafe medical practice to observing primary participants being expected to deal with situations beyond their capabilities. One example occurred in observing a junior doctor who was covering obstetrics, an area in which they had little expertise. The junior doctor had been ‘emergency-paged’ to review an obstetric patient following a significant bleed, where no obstetrician was available. Walking to the emergency, as I unpicked the participant’s clinical reasoning, it became clear that this individual lacked the experiential knowledge and practical skills required. I felt very concerned by this and decided in the moment, that in the interests of non-maleficence (avoiding harm), I would switch role and take an active part in the patient management. Fortunately, shortly after our arrival at the emergency, an appropriate senior doctor appeared and no intervention on my part was required. Had there been unsafe practice, I would have reported the incident following the protocols in place to protect patient safety. Although in this case the situation was resolved, recognising the vulnerability of the junior doctor, I alerted the clinical supervisor to the incident to provide support and ensure that the clinical team debriefed after the event.

For presentation at medical education meetings there was a lower chance of participant identification. The audience was particularly interested in analysis of exact timing of ‘teaching talk’, the guidance given and subsequent responses of the trainee surgeon including pauses, hesitations and hand movement. To enable this, with support I digitally altered audio data so that the pitch of speakers’ voices was modified to conceal identity. On occasion, specific clips with names removed but not completely anonymized format were presented using speakers’ own voices. Express written permission was received from participants for use of that particular clip, recognizing my ability as practitioner-researcher to gauge the likely sensitivity of the data and the audience. For publication purposes, Jeffersonian [34] transcription notation was used. This technique records pauses, intonation, pace and stresses in the delivery of speech so that the transcript indicates not only what is said, but also how it is said. These different methods of presenting the data were selected in the spirit of non-maleficence, avoiding potential harm to participants, acknowledging that complete anonymity could not be entirely guaranteed. This was made clear to participants at the outset.