Method
Twenty one GPs, PNs, DNEs and specialist physicians were purposively sampled from 179 respondents to a survey in which relational coordination between health professionals involved in insulin initiation was measured. In this survey 58% of DNEs, 37% of GPs, 34% of PNs and 37% of specialist physicians consented to being contacted regarding participating in an interview. Maximum variation sampling was utilised to capture a wide range of perspectives and experiences. Two to three health professionals of each type above and below the median relational coordination score were purposively selected to participate in an interview to explore their current professional roles, working relationships with other health professionals and practice or organisational factors impacting initiation of insulin in the general practice setting. Where possible health professionals were also selected on the basis of gender, the type of clinic they worked in (e.g. hospital or general practice) and at least one health professional per group was from a regional or rural setting. This sample size is in keeping with those typically seen with purposive sampling, which usually consists of 30 cases or less. The aim of these interviews was not to obtain saturation, generalisability or to link responses to individual scores, but rather to explore possible factors underlying the relational coordination domains in this purposively selected group.
Interviews were conducted either face to face or via telephone if this was requested by the participant. The interviews were semi-structured according to a pre-written interview schedule which was flexible, updated as the study progressed and allowed exploration of the factors underlying the interviewee's responses. All interviews were conducted by the first author (a female GP). Interviews took between 30 and 45 minutes, were digitally recorded and then transcribed by a professional transcription service. The transcript was reviewed and then uploaded into NVivo 9 (QSR International) for framework analysis, a matrix based method for ordering and synthesising data. Framework analysis was developed by Ritchie and Spencer at the National Centre for Social Research, United Kingdom, in the 1980s. Analysis consists of six stages: familiarisation, identification of descriptive categories, indexing, charting, investigation and interpretation and report findings.
Each transcript was entered as a case node and then characterised according to gender, health professional type (GP, DNE, PN, specialist physician), the health care setting in which the participant worked (general practice, private, hospital, community health centre), duration of practice, geographical (RA) classification and relational coordination survey scores.
The key themes were initially based on relational coordination, collaboration and factors impacting on these as determined from a literature review and new themes were identified as the interviews were conducted and analysed. These were used to build the index for the study which included barriers and facilitators to insulin initiation, education and training, intra and interprofessional relationships, health care setting, professional roles, trust and relational coordination domains between health professional dyads. Key points identified within the transcripts were summarised (charted) into the framework with mapping to the original text in the transcripts (indexing). Two interviews were reviewed with the second author, a GP qualitative researcher, and results compared. The purpose of this was not to reach concordance but rather to introduce different viewpoints regarding coding and interpretation of the data and to refine the coding and analysis. Comparisons were made within and between different health professional groups and common and contrasting themes were developed.
This study received ethical approval from the Human Research Ethics Committee (HREC) at The University of Melbourne (HREC ID: 1238199). All participants provided written consent prior to participating in the interview.