Ethical approval was awarded by University College London Research Ethics Committee (Project ID: 23073.001) (Additional file 1: Supplementary material 1).

Participants were recruited purposively in two ways: through clinical contacts of the research team and through a private hospital group that specialises in eating disorder care. Further participants were recruited by inviting existing participants to share the study with potentially suitable colleagues. Email and social media were used to inform clinical contacts about the study. The research team aimed to recruit a range of professionals, with a range of length of practice, to increase the diversity of the results. All participants had to have experience of working with severe and complex eating disorder presentations in inpatient settings.

Interested volunteers were invited to email the research team and were then sent the participant information sheet and consent form via email. No participants dropped out, and all gave fully informed written consent. Participants were given the option of face-to-face interview or remote interview.

The interviews followed a semi-structured format. The interview topic guide (see Additional file 1: supplementary material) was collaboratively developed in advance by the research team and a small number of multidisciplinary staff from the private hospital group, who have lived experience of providing specialist inpatient eating disorder care. Interviews were recorded and transcribed verbatim by the interviewer, using NVivo followed by manual correction. Any potentially identifying information about the participant, their place of work, or service-users was removed from interview transcripts. Pseudonyms have been applied to the participants.

Transcripts were analysed using reflexive thematic analysis (TA), guided by the six principles of TA developed by Braun & Clark, (2019). Familiarization with and immersion in the data included reading and re-reading transcripts. Inductive organisation of participant quotes and semantic codes was completed to gain an overview of the prominent content across transcripts. Latent codes were then searched for and organised. Connections between codes were searched for, which enabled higher order inductive themes to be generated. Subthemes were explored and generated, before themes and subthemes were reviewed, re-organised, defined and named with the specific research question in mind. Reflexive notes were taken throughout this process, with the position and experience of the interviewer and research team in mind. We aimed to increase credibility of our study by presenting some of the participants with our findings and reflecting on whether it felt as though we encompassed their experiences. We aimed to increase validity by presenting a draft of our research paper to a small focus group from the private hospital group we recruited from. The focus group did not partake in interviews and as such their resonation with the findings demonstrated further face validity of the findings.

The research team was mindful about the sensitive nature of some of the experiences that clinicians may choose to talk about. Signposting to support organisations was included in the participant information sheet and a clear safeguarding protocol was developed before interviews commenced. We offered flexible interview times to accommodate shift patterns and work demands. We also aimed to recruit widely across services and professional roles, with the intention of not placing pressure on specific teams or services. Participants were provided with a £20 e-voucher to thank them for participating and as compensation for their time.

The research team themselves were supported with training and supervision in response to the potential impact of hearing about mental health professionals’ distress, especially as some of the experiences raised by clinicians were similar to those of the research team.

Concepts that demonstrate rigour and quality in quantitative research, such as reliability and generalisability, have been widely critiqued as inappropriate for qualitative research [31]. Therefore, the research team aspired to fulfil quality criteria relevant to qualitative research, such as trustworthiness [28]. To do this, we used existing frameworks to guide us at each phase of our research process and analysis, such as Nowell et al. [28] and Braun & Clarke [4].

In addition to credibility and validity checks, we have made efforts to be open and specific about our decision making, the steps we have taken to generate our themes, and the potential impact of the background of the researchers. This contributes towards increasing transferability, should other researchers wish to replicate the study. We have established dependability as a traceable process is documented. Confirmability was sought by demonstrating that data is clearly derived from the data through our inductive codes, as well as use of verbatim quotes from the transcripts that demonstrate our arguments.

The research team consisted of three white women from working- and middle-class backgrounds.

S.B is a MSc student on a masters level clinical mental health sciences course at the time of the study. She has worked in a private specialist eating disorder inpatient service for over four years, in three different roles, including Healthcare Assistant and Assistant Psychologist. The latter was her employment role at the time of the research.

J.B is a consultant clinical psychologist and professor with over 23 years of experience of working in the NHS with specialist expertise in trauma, mental health and well-being in high-risk occupational groups.

H.N is a doctoral student with experience in qualitative research in healthcare settings.

The research team’s experience of specialist clinical work and related settings was advantageous, particularly for S.B who conducted the interviews, as this experience contributed towards a shared understanding between interviewer and interviewee. S.B.’s work experience positions her as an ‘insider’ of specialist inpatient care, as a previous frontline worker and current multidisciplinary team member. She can also be viewed as an ‘outsider’ by frontline staff as a current multidiciplinary team member or by public sector staff as privately employed. This may have influenced, positively or negatively, the participants’ level of comfort or openness in sharing stories about their work.

There are potential disadvantages in our closeness to the research area. We were mindful of a ‘pull’ towards statements we personally resonated with. This was addressed throughout all stages of data collection and analysis by taking a curious approach to our data and using the whole research team to search for potentially overlooked themes or perspectives. We also conducted credibility and validity checks with participants and other professionals with lived experience of working in similar settings.