Data analysis
Data collection and analysis ran iteratively, and in parallel. NVivo 12 (QSR International, Melbourne, Australia) was used to organise data and support analysis. Data were analysed inductively with all data from each case read together to identify and organise codes. Codes were developed and refined reflecting text or expressions that captured significant views in the data. Codes were then grouped to reflect developing themes, with code descriptions and sample quotes assigned. Case summaries were developed and reviewed using constant comparative techniques to assess whether and how to expand or merge themes. Double coding of a proportion of interview transcripts was undertaken by JW. Commonality and variances across cases were discussed between the researchers and used to further shape evolving themes and sampling.