Analysis

To manage the material, relevant sections of the transcripts, i.e. sections that referred to attendance or non-attendance, were extracted from the original transcript and exported into NVivo [24] and analysed using this software. During extraction of data from original transcripts we took a cautious approach and included any section referring to attendance or motivations. The framework method [25] is a form of qualitative analysis that uses a matrix output to organise large datasets to identify commonalities and differences within the data to draw descriptive and/or explanatory conclusions clustered around themes [26]. This analytic procedure was particularly useful for this study to help manage and organise the large amount of data. It also provided the opportunity to compare and contrast across different typologies, e.g. modality of groups.

To create the coding framework the first author (MD) open coded a random selection of transcripts (n = 10), while the second author (PM) coded five transcripts also selected at random. Initial categories were discussed by the two analysts, and early codes and categories that could be applied to all subsequent transcripts were decided upon. This initial ‘analytic framework’ went through many iterations as novel codes and categories were produced through the analysis process, and as discussions continued about the meaning of the findings between the analysts. A framework matrix was created once the analytic framework was finalised and data was charted into it [28].

Summarising the data in the matrix allowed the findings to be presented and discussed with all authors, the wider research group of the authors, as well as a patient and carer group (Service User and Carer Advisory Group on Research, run by City University of London and East London NHS Foundation Trust).