The role of link workers in weight management for people with severe mental illness: a qualitative study | BMC Primary Care

Design and reporting

This was a qualitative study involving focus groups and dyad interviews with people with SMI and link workers. We followed the Consolidated Criteria for Reporting Qualitative Studies (COREQ) criteria to report it (see Supplementary Data S1) [17, 18].

Patient and public involvement and engagement

The study was co-designed in collaboration with a patient and public involvement and engagement (PPIE) champion who had a lived experience of SMI. A research psychologist, with an academic interest in weight management for people with SMI, co-designed a semi-structured topic guide with the PPIE champion, which was piloted with an external member of the research team. The topic guide included 10 open-ended questions, with probes to scaffold participants’ responses or to explore comments relevant to the research question (see Supplementary Data S2). Focus groups and dyad interviews were co-facilitated by the researcher and PPIE champion, who received training to lead these discussions.

Sampling procedure and recruitment

People with SMI (i.e., community, inpatient, or both) were recruited through convenience sampling via the McPin Foundation– a charity that puts the lived experience of people with mental illness at the heart of research (www.mcpin.org). An online advertisement was posted on the McPin website and anyone interested could directly contact the study team email. Interested participants received an information sheet, followed by a telephone screening call to assess their eligibility, obtain informed consent, and collect demographic data. Two participants were already known to the research team through previous involvement in unrelated studies, but had no ongoing therapeutic or professional relationships with us at the time of recruitment.

The eligibility criteria included: aged 18 or older; given a primary diagnosis of SMI (i.e., ICD-10 codes F20-39 including schizophrenia, schizophreniform disorder, schizoaffective disorder, delusional disorder, brief reactive psychosis, or psychosis not otherwise specified, and bipolar disorder); living with overweight (body mass index [BMI] 25–29.9 kg/m2) or obesity (BMI ≥ 30 kg/m²); willing and able to provide informed consent; and access to Microsoft Teams.

Link workers were also recruited through convenience sampling via the Social Prescribing Research Network– a network that includes members of the public, clinical providers, the voluntary community sector, and policy makers (www.socialprescribing.phc.ox.ac.uk). An online advertisement was emailed to people who had subscribed to the network’s newsletter. As with the SMI group, interested participants were given an information sheet, and invited to a telephone call to assess their eligibility, obtain informed consent, and collect their demographic data.

The eligibility criteria included: aged 18 or older; working in a community-based role in England; willing and able to provide informed consent; and access to Microsoft Teams. There was no exclusion on seniority or length of experience in their role.

While participants were not explicitly asked to recruit through word of mouth or their own networks, any referrals made by them were welcomed by the research team. Recruitment continued until we reached theoretical data sufficiency or the end of the 8-month study period, whichever came first.

Data collection

All focus groups and dyad interviews were conducted remotely as a co-design feature, advocated by the PPIE champion, to promote accessibility and reduce participation burden, especially for people with SMI who have reported mobility issues and concerns about judgement or harm from others. Our decision to run separate focus groups was also informed by the PPIE advice during our study design phase, who indicated that people with SMI might prefer to engage in small group discussions with others who share their lived experience. Additionally, conducting small groups discussions of this size allowed greater flexibility to explore consensus (or lack of) agreement among the participants.

To our knowledge, only the researcher(s) and interviewee(s) were present during the discussions. All discussions were recorded on a separate audio device and transcribed verbatim. No field notes were taken, but the researcher and PPIE champion debriefed after each focus group they co-facilitated. Focus groups and dyad interviews were scheduled for up to one hour with one five-minute break and took place between 25 July 2023 and 31 March 2024. All participants were renumerated with a £25 per hour voucher for their participation. There were no repeated focus groups and transcripts were not returned to the participants.

Data analysis

We followed an inductive-deductive codebook thematic analysis since we aimed to learn about and describe phenomena which little is known ahead of designing an intervention [19, 20]. We did not intend to generate new theories or concepts, but rather explore how all participants accepted the link worker role.

This analysis followed the following steps. First, we deductively developed a priori codes based on our PPIE input and previous research to create an initial coding framework [15]. Second, two researchers read and re-read the transcripts to identify and discuss potential codes, revising the initial coding framework where necessary. Third, the revised coding framework was applied to the interview transcripts. In this step, we attached labels to meaningful units of text in each transcript and inductively-deductively coded them under the coding framework. Fourth, the study team met to discuss the data coding and interpretation, make further iterations to the framework, and reach a consensus agreement. Fifth, once all transcripts had been (re)read and (re)coded, they were electronically managed, which allowed us to easily visualise the data and produce a codebook. The codebook was read and re-read to identify possible patterns between and within interviews, and to identify deviant cases. Sixth, codes were organised and re-organised into sub-themes and then descriptive themes to produce top-level analytical themes.

We acted on Miles and Huberman’s recommendations to enhance trustworthiness in our findings [21, 22]. This meant we used data triangulation, which involved analysing data from multiple sources, namely people with SMI, link workers, and health promotion workers, to account for any convergence and divergence in their perspectives. We used investigator triangulation by engaging in regular discussions and scrutinising the researcher’s reflexivity log to maintain rigour. We also applied theoretical triangulation and sought the PPIE champion’s feedback on the findings to ensure our interpretations remained close to their own account. Our ontological position was relativism, and our epistemological stance was rooted in subjectivism. All data were managed in NVivo 1.7.2 software [23].