Association between oncologists' death anxiety and their end‐of‐life communication with advanced cancer patients

Early and open communication of palliative care (PC) and end‐of‐life (EoL)‐related issues in advanced cancer care is not only recommended by guidelines, but also preferred by the majority of patients. However, oncologists tend to avoid timely addressing these issues. We investigated the role of oncologists' personal death anxiety in the rare occurrence of PC/EoL conversations.

these issues is a stepwise process within the long-term care of advanced cancer patients. Responsible for initiating this process early on are the treating oncologists, 5,6 which in Germany are both medical oncologists and organ-specific oncologist such as gastroenterologists or gynecologists.
Timely communication results in various benefits for patients, their informal caregivers and the healthcare system, [7][8][9] as it is associated with an increased quality of life, improved symptom control, less aggressive medical care at the very EoL and reduced healthcare costs. [7][8][9] It is possible and necessary to integrate PC/EoLrelated issues and thereby still foster hope to patients. 11 Crucial to this is to reframe hope to realistic goals that are achievable within the context of the incurable illness, such as quality of life, symptom relief, a specific event or even a peaceful death. 11 Contrary to recommendations and advantages, PC/EoL-related issues are often addressed only marginally or too late, sometimes at first when patients are no longer able to decide for themselves or are already in acute crisis. 12 Oncologists tend to avoid PC/EoL conversations and often postpone addressing the issues during cancer care. [13][14][15] Further, they tend to discuss the issues in an ambiguous or indirect way 16,17 or to foster unrealistic hope by mitigating bad news. 18 When giving prognostic information, they often overestimate survival time. 19 Major reasons for this avoidance of timely and clear PC/EoL communication are insecurities and a personal discomfort of caring oncologists. In fact, addressing the EoL is considered the most stressful and difficult part of oncological care. 16 Oncologists report, for example, the fear of causing stress or destroying hope when addressing the issues 1,16,20 or consider the reduced mental capacity of advanced cancer patients as a factor preventing EoL conversations. 21 Moreover, some studies suggest that also the oncologists' personal death anxiety is a determinant for rare or too late PC/EoL conversations. [22][23][24][25] Death anxiety includes all concerns raised by cognitions and awareness about death including distress, fear and the sense of powerlessness. 26 Facing the death of another person can be a cue raising this awareness and reminding physicians of their own finiteness. 26 Hence, death anxiety can lead to distancing from PC/EoL issues in order not to be reminded that the patient's fate could be one's own. 27 Nonetheless, the relationship between oncologists' death anxiety and medical communication has not been sufficiently studied. A systematic review by Draper et al. summarizing the existing studies identified only seven studies on this association. 27 As a major methodological weakness the review criticizes that in all studies, communication skills were only assessed via self-report without including an external perspective. 27 The authors conclude that death anxiety might influence the way physicians experience communication, but claim insufficient evidence if death anxiety does influence actual, observable communication behavior. They demand for more high-quality studies and see the need for further investigation of the relationship between death anxiety and externally rated EoL communication. 27 Thus, in order to gain more insight into the reasons for the discrepancy between guidelines recommending timely conversations and clinical reality, the association between personal death anxiety and PC/EOL communication needs to be further studied, especially with regard to the external perspective. This will allow conclusions on necessary support offers and further communication education for oncologists.
Against this background, this study examined the relationship between oncologists' death anxiety and self-reported as well as

| Death anxiety
Oncologists' death anxiety was measured via the Thanatophobia-Scale, 30  An overview of all self-report questionnaires used within this study including its reference and example items is displayed in Supplementary Material S1 (Table S1; self-developed items are presented in full).

| Self-reported communication skills
Confidence. Oncologists' confidence in dealing with specific PC/EoLrelated issues was assessed with five self-developed items (Likert scale 0-4; higher value indicating higher perceived confidence).
Self-efficacy. Oncologists' self-efficacy in PC/EoL communication was assessed via the scale 'communication' of the Self-Efficacy in Palliative Care Scale (SEPC). 31 It consists of eight target behaviors, upon which responders rate their confidence (100 mm visual analog scale; higher values indicating higher self-efficacy). The total score is the mean of all items. The SEPC is valid and reliable. 31 It was translated to German via the TRAPD-method. 32 13 Responders receive a total score for each scale, which is calculated by the sum of all items of HARNISCHFEGER ET AL.
-925 the respective scale (higher scores indicate more intensive use of approach/avoidant communication strategies). The survey has a good reliability. 13 It was translated via the TRAPD-method 32  Five researchers (NH, HR, HB, KH, NR) participated in rating the consultations via a study-specific 13-item rating manual. We conducted a comprehensive training for the raters in order to establish an acceptable inter-rater reliability. About 20% of the videos were rated by two raters, disagreements were discussed and solved by consensus. Videos were assigned for rating in a way ensuring that study participants and rating researchers always belonged to different study sites of the multicenter trial.
The rating manual was developed based on the German versions of the validated COM-ON-checklist 33 and the COM-ON-Coaching rating scales 34 as well as under consideration of the core competencies in PC. 35 We created a study-specific instruction manual with

| Statistical analyses
Descriptive statistics were determined via of means, standard deviations, frequency distributions and percentages. Descriptive communication skills were displayed for total study cohort and separately per gender. Gender differences were analyzed via t-tests.
Inter-rater reliability of the external rating of communication skills was analyzed via the intraclass correlation coefficient. Effect sizes were determined by means of partial eta squared (partial η 2 ), interpreted as small from the value of 0.01, medium from the value of 0.06 and strong from the value of 0.14.

| Descriptive analyses
Mean values and standard deviations of death anxiety as well as selfreported and externally rated communication skills for the total cohort and separately per gender are presented in Table 2. No gender difference was found in any of the variables.
Regarding self-reported confidence in discussing specific PC/ EoL-related issues, oncologists felt most and similarly high confident with discussing fears and concerns, patients' goals, values and wishes as well as symptoms and symptom control. They subjectively felt less confident with discussing feelings and thoughts regarding the EoL and PC services (Table 2, Figure 1).
In external rated communication skills, the best communication performance was observed regarding discussing further cancerspecific treatment as well as pointing out opportunities and giving hope. Against that, oncologists had more difficulties in dealing with emotions and showing empathy as well as addressing the EoL and even considerably more difficulties with explaining the concept of PC (Table 2, Figure 1).
Inter-rater reliability of the external rating of communication skills ranged from 0.619 to 0.936, which can be interpreted as good to excellent. 36

| Association between death anxiety PC/EoL communication skills
Our analyses revealed that death anxiety was significantly associated with more avoidant communication strategies with a high 926 -HARNISCHFEGER ET AL. effect size. No association was found with approach communication strategies.
Death anxiety was also significantly associated with lower perceived self-efficacy (medium effect size), less confidence in discussing feelings and thoughts regarding the EoL (medium effect size) and less confidence in discussing patients' goals and wishes (low effect size). No association was found between death anxiety and confidence in discussing symptoms and symptom control, PC services as well as fears and concerns. None of the sociodemographic characteristics or externally rated communication was significantly related to death anxiety (Table 3).

| DISCUSSION
To our knowledge, this is the first study investigating the relationship between oncologists' death anxiety and self-reported as well as externally rated, observable PC/EoL communication skills.
Main study finding is that death anxiety is associated with more avoidant self-reported EoL communication, less perceived selfefficacy, less confidence in discussing feelings and thoughts regarding the EoL and less confidence in discussing patients' goals and wishes, but is not associated with more avoidant externally rated communication skills. This is in line with the systematic review of Draper et al., who assumed an association between death anxiety and the way physicians experience communication, but questioned if death anxiety is related to observable communication behavior. 27 This might suggest that oncologists manage to conduct successful consultations without showing their increased discomfort. This conclusion, however, needs further investigation, for example, by interviewing patients.
Dealing with personal anxiety in interaction with patients probably costs oncologists a lot of 'inner work'. This is in line with existing evidence stating that involvement in EoL care can lead to burnout, compassion fatigue, nihilism and moral distress. 37,38 Our results indicate that having to distance oneself from personal fears might also be a contributing factor to these adverse effects of oncological work. Training, self-reflection, group supervision and consultation offers for oncologists might represent effective approaches to reduce this anxiety and personal discomfort, as recognizing and facing own fears can lead to more self-awareness and understanding of those feelings. 26,37 Regarding physicians' general level of death anxiety, with a mean value of M = 20.31, we can report a slightly lower level of death anxiety among our sample of oncologists than a previous study found in a group of non-oncologist primary care physicians who reached a value of M = 22.6. 39 It can be assumed that oncologists with their current exposure to terminal ill patients might be a little less anxious about death and dying than physicians not explicitly working in oncological settings, where possibly habituation effects play a role.
Apart from death anxiety and its associations with communication skills, we could confirm the oncologists' general tendency to avoid PC/EoL-related issues, as it was proposed by several previous studies. [12][13][14][15] In our analysis of oncologists' self-reported communication skills, we found that oncologists subjectively felt more comfortable with discussing patients' goals, values and wishes, fears and concerns as well as symptoms and symptom control than discussing feelings and thoughts regarding the EoL and PC services. In externally rated PC/EoL communication skills, they showed a considerably better performance in discussing further cancer-specific treatment, pointing out opportunities and giving hope than in addressing the EoL and dealing with emotions/showing empathy.
Here, the most difficulties were observable in explaining the concept of PC, suggesting that there might be a lack of knowledge and that more training in the field of PC is needed. Previous literature Comparing external and self-reported ratings, it is noticeable that from both perspectives, the more 'hopeful' issues-such as discussing goals and wishes, further cancer-specific treatment options, symptom control and opportunities-seem to be addressed more easily than the EoL, emotion-related issues and explaining PC/PC services. The only exception seems to be the experienced confidence in discussing patients' fears and concerns. Although fears of advanced cancer patients might objectively relate to death and dying, it is possible that oncologists tend to redirect these fears away from EoL-related issues to hope on their cancer-specific treatment offers, which might make them feel more confident.
When interpreting the results, it should be considered that with

| Study limitations
Limitations include that we are only able to report associations, but no causalities due to the cross-sectional design of our study. Hence, it is only possible to conclude that there is an association between death anxiety and subjective difficulties in PC/EoL communication, self-selection bias of our study cohort is possible. As data were assessed as part of a study evaluating a PC/EoL communication skills training, potentially a higher proportion of oncologists participated that were already more interested in communication and might be endowed with better communication skills than their colleagues. Yet, the insecurities with PC/EoL issues were even found among those possibly 'better qualified' oncologists. Above that, it is possible that the basic PC knowledge that had to be assumed as part of physicians' basic qualification might not have sufficiently existed. As we did not measure PC knowledge directly, we cannot exactly determine the extent of those knowledge gaps.
Further, due to the lack of standardized measurements, external rating of the videos was conducted by a study-specifically adapted rating scale that previously had not been used in this exact form.
The extensive training of the raters still resulted in a good to very good inter-rater reliability. Lastly, externally rated communication skills were measured within one single consultation. Thus, long-term processes could not be assessed.

| Clinical implications
Oncologists might benefit from interventions to reduce their perceived difficulties and personal fears that are elicited in PC/EoL communication with advanced cancer patients. For example, group supervision and consultation offers should be implemented regularly on oncological wards in order to enhance reflection and awareness of oncologists' personal fears. 37 Guidance in dealing with personal discomfort should also receive more attention both in basic communication education and in specific communication trainings for oncologists. Spiritual care training might also be an approach to empower oncologists to communicate about patient's fears and EoL-related issues, because spiritual care sets a great focus on addressing emotions and establishing a compassionate physician-patient-relationship. 45 The detected difficulties in explaining the concept of PC and PC services from the oncologists' own and the external perspective indicate that more education on the concept of PC including its strengths and opportunities needs to be provided to oncologists. The  More group supervision and consultation offers for oncologists are needed in order to increase reflection and awareness of personal fears and reduce personal discomfort in EoL communication with advanced cancer patients. This might facilitate early communication of PC/EoL-related issues, as it is preferred by patients 1-3 and recommended by guidelines, 5,6 but often not executed practice. [12][13][14] Such interventions would possibly also help to reduce adverse psychological effects of working in oncology. 37  Not included in final model due to stepwise exclusion of non-significant variables. *Significant association with death anxiety at an alpha level of 0.05.