A National Survey to Systematically Identify Factors Associated With Oncologists’ Attitudes Toward End‐of‐Life Discussions: What Determines Timing of End‐of‐Life Discussions?

Background. End‐of‐life discussions (EOLds) occur infrequently until cancer patients become terminally ill. Methods. To identify factors associated with the timing of EOLds, we conducted a nationwide survey of 864 medical oncologists. We surveyed the timing of EOLds held with advanced cancer patients regarding prognosis, hospice, site of death, and do‐not‐resuscitate (DNR) status; and we surveyed physicians’ experience of EOLds, perceptions of a good death, and beliefs regarding these issues. Multivariate analyses identified determinants of early discussions. Results. Among 490 physicians (response rate: 57%), 165 (34%), 65 (14%), 47 (9.8%), and 20 (4.2%) would discuss prognosis, hospice, site of death, and DNR status, respectively, “now” (i.e., at diagnosis) with a hypothetical patient with newly diagnosed metastatic cancer. In multivariate analyses, determinants of discussing prognosis “now” included the physician perceiving greater importance of autonomy in experiencing a good death (odds ratio [OR]: 1.34; p = .014), less perceived difficulty estimating the prognosis (OR: 0.77; p = .012), and being a hematologist (OR: 1.68; p = .016). Determinants of discussing hospice “now” included the physician perceiving greater importance of life completion in experiencing a good death (OR: 1.58; p = .018), less discomfort talking about death (OR: 0.67; p = .002), and no responsibility as treating physician at end of life (OR: 1.94; p = .031). Determinants of discussing site of death “now” included the physician perceiving greater importance of life completion in experiencing a good death (OR: 1.83; p = .008) and less discomfort talking about death (OR: 0.74; p = .034). The determinant of discussing DNR status “now” was less discomfort talking about death (OR: 0.49; p = .003). Conclusion. Reflection by oncologists on their own values regarding a good death, knowledge about validated prognostic measures, and learning skills to manage discomfort talking about death is helpful for oncologists to perform appropriate EOLds. Implications for Practice: Oncologists’ own perceptions about what is important for a “good death,” perceived difficulty in estimating the prognosis, and discomfort in talking about death influence their attitudes toward end‐of‐life discussions. Reflection on their own values regarding a good death, knowledge about validated prognostic measures, and learning skills to manage discomfort talking about death are important for improving oncologists’ skills in facilitating end‐of‐life discussions.