Data analysis Analysis of the first interviews began while the interviews were going on. The data were analysed using the constant comparative method as described in grounded theory (10) with the goal of concept development and beginning theory development. Interviews and field notes were coded. Codes were grouped, compared and categorized into concepts. Both descriptive coding and theoretical coding were used: descriptive coding to explore the experiences and needs, and theoretical coding to develop a theory in order to clarify relationships between concepts and to facilitate the development of a theoretical framework. After analysis of eight interviews, a preliminary theory has been developed. The remaining five interviews were used to validate this theory. Data triangulation and investigator triangulation (14) were used to broaden the analysis, and to try to obtain content validity as much as possible (11, 15). Respondents with varying socio-demographic variables participated in the study, and results of the family interviews were checked with 10 patient interviews, two interviews with bereaved carers, and four provider interviews (data triangulation). There were four interviewers who conducted interviews (investigator triangulation). In addition, other (senior) researchers were involved in the analysis and coding process (peer debriefing).