Mariken Stegmann

we approached refused to participate because they felt too ill. The OPTion study mainly included patients with lung cancer, known to have a poor prognosis, and these were not only included directly after diagnosis but also at the moment of a later treatment decision. These factors may have increased the recruitment problems in our study. However, it should be realised that these issues can be expected, to some extent, in all research among patients with non ‐ curable cancer. Ethics The high dropout rate because of health deterioration raises the ethical question as to whether we should perform research in patients with non ‐ curable cancer. However, some patients in the OPTion study explicitly stated that participation made them feel valuable and that they wanted to help future patients in the same situation. Furthermore, greater attention has been paid to aggressive end ‐ of ‐ life care in recent years, with disease ‐ oriented treatment continued to shortly before death. 38 SDM allows patients to discuss the benefits and negative effects and make informed choices that meet their needs. One may argue that health care should be particularly optimised in this highly vulnerable population. However, achieving this goal will involve further research that will come with important responsibilities to use sound methodology and to recruit adequate samples. Both topics are discussed more extensively in the following sections. There must also be an adequate dissemination strategy that reaches the involved stakeholders. Studies should therefore be published not only in international journals but also in Dutch journals that are read by practicing health care professionals and shared during training sessions. Outcome measures The ultimate outcome measure in medical research is traditionally survival, yet this is not always the most important outcome for patients, particularly those with non ‐ curable cancer. Quality of life has therefore been suggested as a valuable patient reported outcome measure in SDM research, but for patients with non ‐ curable cancer, this is affected by physical symptoms, knowledge of impending death, and other contextual factors. This makes it almost impossible to influence quality of life or other general outcome measures by changes in health care organisation. Consequently, we chose DSE, which is a very specific outcome measure (Chapter 2). However, using this had a ceiling effect, with most patients in both groups scoring so high on DSE that the questionnaire was unable to detect differences. Other studies have also reported high scores for DSE, but without explicitly mentioning the ceiling effect. 39–41 A similar problem has been described when using patient satisfaction as the outcome measure. 42 Researchers should be aware of the problems inherent to using outcome measures that are too general or unresponsive. Conducting a pilot study in the target population can be sufficient to see how the proposed outcome measure performs. Recruitment A pilot study can also yield valuable information about recruitment, identifying potential factors related to the patient, health care professional, and researcher. As mentioned above, health deterioration is a common reason for patients to refrain from participation. To minimise this problem, it is important that the commitments for the questionnaires and other burdens are limited as much as possible. This minimal impact for the participant should then be made clear in any patient information material. In the Netherlands, there are strict regulations on research where patients are subject to new treatments or activities with 8 113 Summary and General discussion