Mariken Stegmann

hypotheses exist and an intervention is known to be feasible, an RCT is appropriate to deliver more robust data about the effect of the intervention in comparison with a control group. However, the above ‐ mentioned suggestions about recruitment and outcome measures still apply. Researchers tend to reach for the highest level of evidence, the RCT, whenever possible, even where this does not suit the research question. Where research focuses on exploring opinions, developing hypotheses, and/or assessing the feasibility of interventions, qualitative research has an important role to play, either alone or in a mixed ‐ methods design. In retrospect, for the OPTion RCT (Chapters 2 and 4), we were too eager to perform an RCT and should have performed a pilot study before progressing. Generalisation A last point to consider is the generalisability of the findings concerning medical communication. Because the organisation of health care differs substantially between countries it can be difficult to extrapolate findings. The role of primary care, for example, is very different among European countries. To help health care providers interpret findings, the research setting should be always be described, preferably with a concise description of the health care system. Furthermore, we must acknowledge that all patients and medical doctors are unique, and that a one ‐ size ‐ fits ‐ all solution to the existing problems with medical communication probably does not exist. Implications of our findings Further research Answering one research question rarely fails to uncover new questions. Based on the current research, the following topics would be worth pursuing in the future: ‐ Can a conversation with the GP about treatment goals reduce anxiety, as shown in Chapter 4, when measured as the primary outcome? Is our finding reproducible, and does it add clinical value for patients? What is the most appropriate outcome measure for SDM interventions? ‐ Can we reproduce the findings of Chapter 6 regarding goals changing or remaining stable among patients with non ‐ curative cancer and explore this topic further? Specifically, it would be interesting to analyse the audio files of consultations and/or to use patient interviews. ‐ Is it feasible to implement changes to the format of correspondence between primary and secondary care, as suggested in Chapters 6 and 7? (i.e., a short and well ‐ organised referral letter and a specialist letter with information about all treatment options.) What effects would these changes have in practice? What is the most appropriate outcome measure for these effects? What are the potential barriers to, and facilitators of, such implementation? ‐ Based on the findings in Chapters 4 to 7, can improved communication between primary and secondary care about a patient’s treatment goals lead to patients making better treatment decisions and altering those decisions if their goals change? Methodological advices ‐ When performing an RCT, it is prudent to perform a pilot study first to explore potential problems with recruitment and to check the performance of the outcome measure. 8 115 Summary and General discussion