Purposive sampling was employed (Ritchie et al 2003).Inclusion criteria were COPD diagnosis (GOLD 2005), completion of an 8-week outpatient pulmonary rehabilitation course held either in a hospital gym or in one of four community venues within the last two years, and ability to access the pulmonary

rehabilitation venue independently. Exclusion criteria were no spoken English or requirement for transport provided by the hospital. We set out to include people with a range of experiences in relation to pulmonary rehabilitation to generate rich data and to introduce diversity whilst maintaining overall homogeneity (Finch and Lewis 2003). Using records held by the pulmonary rehabilitation team, eligible participants were placed into two groups, A and B, by the principal this website researcher. Group A had received input from pulmonary rehabilitation staff to assist with ongoing exercise following completion of the pulmonary rehabilitation course, either by choosing to attend a maintenance gym session run by pulmonary rehabilitation staff or by receiving an induction into an existing community class from pulmonary rehabilitation staff. Group B had not received any input

from pulmonary rehabilitation staff regarding ongoing exercise due either to choice or lack of opportunity for pulmonary rehabilitation staff to support their chosen exercise option. Suitable patients were approached via letter. Recruitment Lenvatinib chemical structure continued until nine positive responses had been received from each group, in an attempt to secure six to eight participants per group. Data were analysed manually using a grounded theory approach (Charmaz 2006). Each segment of transcribed GPX6 data from Group A and B was coded openly. Frequently occurring codes were used to re-organise and integrate the data into broader categories and themes, and inter-theme relationships were identified. Mind-maps facilitated this iterative process (Braun and Clarke 2006). An experienced qualitative researcher (HF) reviewed the coding process to enhance analysis credibility. The observer (AG) reviewed

the findings independently and concurred with the themes identified. Respondent validation was carried out by two participants in each focus group, who agreed that the analysis accurately reflected their discussion. To guard against a selective narrative, the researcher purposely chose individuals who, between them, embodied a range of views within the dataset (O’Neill Green et al 2010). The results were reviewed by two expert pulmonary rehabilitation practitioners, who confirmed that the findings were meaningful and credible in relation to personal experience. A critically reflexive account and audit trail were maintained throughout to establish dependability and confirmability (Holloway and Wheeler 2002). Of the 28 people approached by letter, 22 responded initially to express interest and 16 participated in the focus groups.