51 Patient perspectives on a smartphone app to support home-based exercise An “inductive coding” approach was chosen for stage 3, the coding stage, and Microsoft Excel 2016 was used to aid with the analysis. Coding was performed by extracting meaningful quotes from the transcripts to an Excel datasheet, adding a short descriptive code to the quote, grouping related or similar quotes, and repeating the process until the entire transcript was coded. The first 3 interviews were independently coded by 2 researchers (RA and CK) (19). After an interview was coded, the researchers compared results and discussed differences in coding until they reached a consensus, and they labeled the codes with a short descriptive name. If the researchers could not reach a consensus, a third researcher (MP) was consulted. The remaining interviews were coded by 2 researchers (RA and CK) working together. During the coding process, the researchers continuously refined and adjusted the codes to best fit the data. In stage 4, paper prints of the codes and their associated quotes from the first 3 interviews were used to allow a hands-on approach for the creation of categories and an initial analytical framework. Categories were formed by grouping codes that appeared to be related until all codes were assigned to a category. The categories were then grouped under themes based on the topics from the interview guide. To reduce bias introduced by the personal perspectives of a single researcher, the researchers (RA and CK) worked together to construct the framework and discussed each new category and its place within the framework until they reached a consensus. The analytical framework was continuously developed in an iterative process. Categories were merged, split, or relabeled, and codes were assigned to different categories in an attempt to best fit the data until all interviews were analyzed. After each iteration, the members of the research team (RA, CK, MP, TK, RO, and CV) discussed the new framework matrix and used the input from the discussion for the next iteration. The final framework matrix contained all categories with the summarized data from each interview and was used to interpret the data, completing stages 6 and 7 of the analysis. RESULTS Participant Characteristics Once data saturation was reached after 9 interviews, recruitment ended. The characteristics of the patients included in the study can be found in Table 1. Table 1 Participant characteristics. Participant number Gender Age (years) SUSa score (0-100) 1 Male 42 70 2 Female 29 82.5 3 Male 39 90 3
RkJQdWJsaXNoZXIy MTk4NDMw