Nvivo 11 qualitative analysis software

All interviews were recorded and transcribed for analysis. Research notes were taken during and after the interview process whenever it was informal or inappropriate to record. Data analysis was a flexible and iterative process where emerging findings and hypothesis were continuously tested in the field until saturation (i.e. no new findings emerging). The analytical induction process involved the iterative testing of theoretical ideas, which were used to refine and categorise themes grounded in the data whilst emerging themes were evaluated in dialogue with existing social science theory on adherence. This resulted in an adherence framework that was systematically applied in subsequent data collection and analysis. All interviews were systematised and analysed with NVivo 11 Qualitative analysis software (QSR International Pty Ltd. Cardigan UK).

Data analysis was a flexible and iterative process. Preliminary data from in-depth interviews and participant observations were jointly analysed by FJ and JM to further adapt the interview guide and process. Additional data collection was used to confirm or refute temporary findings until the point of saturation (i.e. no new findings emerging). All interviews were systematized and analysed with NVivo 11 Qualitative analysis software (QSR International Pty Ltd, Cardigan UK).

Verbatim transcriptions in Chinese were analyzed using a thematic analysis [27 ]. Researchers first read through all transcripts to be familiar with the data. Themes were then developed based on the interview guide and discussion among researchers to capture the important features of the data. Codes were generated for each transcript line by line and grouped into categories. Each category was then reviewed and examined again, and groups of main categories were refined before finalizing the major themes. Coding process and key themes were discussed within the research team. Coding was performed by WG and reviewed by EG, followed by a discussion to ensure the trustworthiness. Quotations used in this paper were translated from Chinese to English and then back-translated into Chinese to increase the transparency of the interpretation. NVivo 11 qualitative analysis software (QSR International) was used to facilitate the coding approach.
The survey data in this study were analyzed by using STATA statistical software (Version 15, StataCorp LLC). A descriptive analysis was performed, and frequencies and proportions were reported for categorical variables.