Concordance between Experiences of Bereaved Relatives, Physicians, and Nurses with Hospital End-of-Life Care: Everyone Has Their ‘‘Own Truth’’

When patients die relatives and healthcare professionals may appreciate the quality of the dying phase differently, but comparisons are rare. In a cross-sectional study (June 2009–July 2012) the experiences of bereaved relatives, physicians, andnurses concerning the quality of dying in a largeDutch university hospital were compared, and the relation to communicationwas explored.Measurements were concordance on the quality of dying (QOD) (0–10 scale), awareness of impending death, and end-of-life communication. Results. Data on all three perspectives were available for 200 patients. Concordance in general was poor. Relatives’ scores for QOD (median 7; IQR 5–8) were lower than physicians and nurses’ (both median 7; IQR 6–8) (P = 0.002). 48% of the relatives, 77% of the physicians, and 73%of the nurses had been aware of impending death. Physiciansmore often reported to have informed patients and relatives of end-of-life issues than relatives reported. When both physicians and relatives reported about such discussion, relatives’ awareness of impending death and presence at the patient’s deathbed weremore likely.Conclusion. Relatives, physicians, and nurses seem to have their “own truth” about the dying phase. Professionals should put more emphasis on the collaboration with relatives and on verification of relative’s understanding.


Introduction
As death comes to us all and most people die after a period of declining health and a significant symptom burden, knowledge on how end-of-life care can contribute to a good deathbed is highly needed. Research on patient reported outcomes and the effectiveness of interventions at the end of life is complex, for example, because of methodological and ethical considerations [1][2][3]. Furthermore, perspectives vary on what good quality of dying is and how this should be achieved, as was recently shown in the debate on the Liverpool Care Pathway for the Dying Patient [4][5][6][7][8][9]. A major limiting factor is that the main person involved, that is, the dying patient, cannot participate in the debate on outcome measures and in the evaluation of experiences in the dying phase [2]. Proxy assessments are needed, including those of relatives and healthcare professionals (HCPs) who may have various perspectives and values [3,[10][11][12][13].
Studies have shown that relatives and patients share important priorities, such as honest and clear communication, involvement in medical decisions, relief of symptom burden, and having the opportunity of completing life and saying goodbye [10,[14][15][16][17][18]; therefore relatives might be able to represent some of the patient's interests. Most studies on proxy evaluation of end-of-life care compare symptom scores of patients and relatives or HCPs. These studies show a tendency of relatives to overestimate the severity of symptoms in comparison to patients' self-reports, whereas nurses and physicians tend to underestimate them [10,12,16,[19][20][21].
Only a few small studies have compared the experiences of the dying phase of relatives and HCPs, showing low to moderate concordance between these groups [13,22]. In the national audit of end-of-life care in Irish hospitals physicians evaluated care more positively than relatives, whereas nurses reported intermediate opinions [23]. More insight is needed into the appreciation of relatives of the care and communication in 2 International Journal of Palliative Care the last days of life and how this is related to the perspectives of HCPs. The purpose of this study was to better understand evaluation of the dying experience from different perspectives and how end-of-life communication may affect the quality of dying. Therefore we compared bereaved relatives and physicians and nurses' assessments of dying patients' quality of life and quality of dying, their awareness of impending death, and their communication; furthermore, we analyzed how communication was related to relatives' appreciation of aspects of the quality of dying and the quality of care, such as preparation for death and satisfaction with care.

Materials and Methods
The study was embedded in a larger study on the quality of palliative and terminal care in the hospital (the PalTeC-H study). We included cases of patients who died in 18 nonintensive care wards of a 1300-bed university hospital in Netherlands (i.e., 951 consecutive deaths), between June 2009 and July 2012, after a hospitalization of at least 6 hours. The protocol of this study is described elsewhere [24]. Physicians and nurses who had closely been involved in daily care for a dying patient were asked by the team coordinator of the ward to complete a questionnaire within at most two weeks after the patient's death. Ten to thirteen weeks after the death of a patient a relative was invited by post by the primary investigator (F. E. Witkamp) to complete a questionnaire. When needed, a written reminder was sent after four weeks.

Ethical Considerations.
Approval for this study was given by the Medical Ethical Research Committee of the Erasmus MC. According to Dutch legislation, written informed consent of the patients or respondents was not required because data were gathered after patients' death and the study involved minimal risk or burden to the respondents.

Questionnaires.
Previous research provided us with a multidimensional framework of quality of life, quality of dying, and quality of care, which are overlapping but distinguishable constructs [25,26]. Quality of life (at the end of life) involves physical, psychological, social, and spiritual experiences, and quality of dying additionally includes life closure and death preparation, the character of health care, and the circumstances of death [25]. Quality of care at the end of life refers to the extent to which these domains are adequately affected by healthcare [27].
Based on this framework three complementary selfadministered questionnaires for relatives, physicians, and nurses were developed by a group of experts in palliative care and research, because previously described instruments met neither our requirements of complementarity of the three perspectives nor those of completeness regarding the last days of life and the dying phase [28][29][30][31]. The questionnaire for relatives was critically evaluated by a representative of the hospital patients' council. All questionnaires were tested on relevance and face validity among members of all targeted groups, including persons who had recently lost a relative, hospital physicians, and nurses. Subsequently, all questionnaires were piloted in the first 30 cases [24]. Some topics were assessed from all perspectives (Table 1), including two summarizing items, in which the respondents were asked to rate the patient's quality of life during the final three days (further referred to as QOL3) and his/her quality of dying (QOD), both on a 0-10 scale with zero being "very poor" and ten "almost perfect. " Further, all were asked whether HCPs had discussed patient's impending death with relatives and patients (no or yes) and whether they had been aware of the patient's impending death (no, more or less, yes). "Being aware" was defined as being aware that death was likely to occur within hours or days, and "being more or less aware" as knowing that the patient would die, but not expecting that death would occur so soon. We also asked when the respondents had become aware of the patient's impending death (<24 hrs, between 24 and 48, or 48 and 72 hrs and >72 hrs in advance). Finally, we assessed aspects of the quality of dying and the quality of care according to relatives, including opportunities to prepare for death, satisfaction with care, saying goodbye (no, more or less, yes), and presence at the moment of death (yes or no). The validity and reliability of the questionnaire for relatives were previously analyzed with simple and multivariate regression analyses and Principal Component Analyses [32]. In addition, in all groups we analyzed the association between the QOD score and answers to verbal terms, reflecting the way of dying as experienced by the respondents, such as quiet, peaceful, traumatic, sad, shocking, and moving. The QOD score was positive related to "positive" terms, such as quiet, beautiful, with dignity, and peaceful and negative related to "negative" terms, such as shocking, traumatic, and troubled ( values <0.001).

Data Analysis.
We analyzed data on all patients for whom respondents from all three groups (relatives, physicians, and nurses) had completed a questionnaire. Linear regression analyses were performed to study the effect of patient characteristics (i.e., age, gender, diagnosis cancer or no cancer, and duration of the final hospitalization) on QOL3 or QOD scores. Differences of QOL3 and QOD scores between the groups were estimated using multilevel regression analysis. The multilevel model included random effects for the intercepts of the regression model to allow for the multiple measurements per patient and fixed effects for the three perspectives. Correlations between participants' awareness of the patient's impending death and the moment of their awareness were tested using cross tables with Chi 2 tests and Spearman correlation coefficients, respectively. Agreement between physicians and relatives on whether physicians had informed relatives of the patient's impending death was tested using Cohen's Kappa ( ): values were interpreted as <0.20 = poor; 0.21-0.40 = fair; 0.41-0.60 = moderate; 0.61-0.80 = good; and 0.81-1.0 = very good concordance [33]. Furthermore we analyzed the relation of the communication of HCPs about a patient's impending death with patients and relatives' awareness and death preparation, relatives' presence at the moment of death, and their satisfaction with care, using Chi 2 tests. Level of significance was set at < 0.05 for all analyses. All data were analyzed with statistical packages SPSS 21 and R version 2.13.

General Characteristics.
During the study period 951 patients died. All invited participants responded independently; the response rates per group were 52% for relatives, 45% for physicians, and 54% for nurses. For 200 patients all three questionnaires were completed. At the time of death, the mean age of these patients was 69 years and the mean duration of the final hospitalization was 14 days; 54% died of cancer; at the time of admission to the hospital the treatment goal was symptom control for 61% and terminal care for 5% (Table 2). Relatives were mostly the spouse (50%) or a child (34%) of the patient, and 88% were involved in patient care during final days; their mean age was 57 years. Participating physicians were mostly interns, and 37% had cared for more than five dying patients during the previous 12 months. Of the participating nurses 28% had cared for more than five dying patients during the previous 12 months. In 61% of the cases, nurses had cared for the patient during at least 2 shifts. Characteristics and outcomes were similar to those of the total study populations [32,34].

Quality of Life and Quality of Dying.
Both the QOL3 and the QOD were scored with a range from 0 to 10, in all three groups of participants. The median QOL3 score of the relatives was 3 (Interquartile Range (IQR) 1-6), whereas physicians and nurses scored QOL3 significantly higher (both median 5; IQR 3-6) ( = 12.76; < 0.001) ( Table 3). The median QOD score was 7 in all three groups, with a IQR of 5-8 in relatives and 6-8 in HCPs, the latter being significantly higher than relatives' scores ( = 6.47; = 0.002). QOL3 and QOD scores were not related to patients' diagnosis, age, or gender. Relatives and physicians' QOL3 scores were slightly lower if the duration of the hospitalization had been longer ( = 5.63, = −0.03; = 0.02; = 4.56, = −0.02; = 0.03, resp.).

Awareness of Impending Death.
Relatives reported to have been aware of the patient's impending death in 48% of all cases. Physicians and nurses had been aware in 77% and 73%, respectively (Table 4). There was no correlation in the awareness between relatives and physicians (Rho = 0.10; = 0.17) and poor correlation between relatives and nurses (Rho = 0.26; = 0.000) and nurses and physicians (Rho = 0.20; = 0.004). Of all cases, 29% of relatives had not been aware of the patient's impending death and in another 29% they became aware during the final 24 hours. In 38% of cases physicians and nurses became aware of the patient's impending death in the final 24 hours and correlation between physicians and nurses moments of awareness was moderate (Rho = 0.41; = 0.000). Correlation between relatives and HCPs' moments of awareness was poor and in the opposite direction (Rho = −0.25; = 0.003 and Rho = −0.26; = 0.001, resp.).

Communication.
Physicians more often reported to have informed patients and relatives of end-of-life issues than relatives and nurses reported (Table 5). In 57% physicians reported that they (32%) or a colleague (25%) had informed patients about their impending death. In 27%, this had been impossible, due to unconsciousness of the patient. Physicians reported to have informed 90% of relatives of the patient's impending death, whereas 64% of the relatives reported to have been informed ( = 0.18, = 0.001).
Compared to cases in which only the physician reported that the relative was informed about impending death, relatives who had confirmed this communication were more likely to have been aware of the patient's impending death (63% versus 27%), to have said goodbye (64% versus 36%), to be satisfied about symptom management (82% versus 69%) and about emotional and social support (76% versus 53%), and to have been present at the moment of death   (92% versus 75%) ( Table 6). Discussion between relatives and physicians about impending death did not affect the patient's preparedness. When the physician reported to have discussed impending death with the patient relatives more often reported that the patient had been aware of impending death (41% versus 14%) and had said goodbye (58% versus 27%) and that relatives themselves had said goodbye (59% versus 40%). When nurses had discussed the end of life with the patient this was also associated with increased percentages of patients' awareness, patients and relatives' goodbyes, and relatives' presence at the moment of death.

Discussion
In the acute hospital setting, bereaved relatives and HCPs rated the QOL3 as poor and the QOD as moderate; HCPs scored QOL3 and QOD significantly higher than relatives. The concordance of their evaluation of QOL3 and QOD, their awareness of the impending death of the patient, and discussions about impending death was poor. However, communication between HCPs, patients, and relatives about impending death was associated with better awareness of and preparedness for the end of life and with more satisfaction with care.
A poor concordance of perspectives on end-of-life care was also found in a few other studies [13,22,23]. Different stakeholders seem to experience their "own truth" and probably there is no "one truth. " This finding has important implications regarding the development and use of quality indicators and outcome measures in end-of-life care. The use of such indicators in palliative care is in its infancy and the effects of their implementation have not yet been assessed [35,36]. Our study shows that outcomes are likely to be affected by the type of assessor.
Relatives had been fully aware of the impending death of the patient in 48% of the cases, whereas physicians and nurses had been aware in 77% and 73% of all cases. However, in four out of ten cases HCPs only became aware of the nearness of death during the final 24 hours. The awareness and moment of awareness of relatives and HCPs were not or even negatively correlated. Awareness of impending death is often seen as an important condition for a good death [25,37,38]; however it is known to frequently occur late in the hospital [39][40][41]. Hospital care is typically aimed at improving patients' health and subsequently discharging them to go home. Only in 5% the treatment goal at admission was providing care in the dying phase. This practice affects the expectations of patients, relatives, and HCPs. Most participants in our study might have expected the patient to be discharged to go home after the hospitalization, which instead turned out to end with the patient's death. Recognition of the irreversibility of a worsening health condition and, subsequently, refocusing care to enable the patient and his relatives to have a good death bed, is difficult [39,42]. Our finding that relatives and HCPs evaluated the dying phase differently and recognized the imminence of death at various moments underlines the need for frequent and adequate communication between patients, physicians, nurses, and relatives, during the final days of life and also after death. Every party involved brings his/her own expertise and jointly they are probably best able to assess the patient's needs and prospects: physicians have biomedical knowledge and skills, nurses have expertise in 6 International Journal of Palliative Care  (9)   personal care and daily observations, and relatives probably best know how to interpret patients' appearance and behavior. One-third of the relatively young participating physicians had discussed imminent death with the patient themselves. The provision of clinical care in the learning environment of a teaching hospital is challenging; junior doctors need adequate supervision to learn more about end-of-life care, truth telling, and breaking bad news [43,44].
This study also showed that being informed about the patient's impending death did not obviously result in awareness of the nearness of death. Nevertheless, communication of HCPs with relatives and patients clearly affected the awareness of impending death and the preparation for the end of life, especially when the relative confirmed to have been informed by the physician, which was not always the case. Taking into account physicians' late awareness and the number of cases in which patients could not be informed because they were unconscious, communication about impending death might often have taken place late in the process [45,46]. Communication at the end of life is one of the most difficult tasks of doctors and concerns deliberations on if bad news should be broken, when, and how [39,42,45].
When being confronted with a fatal prognosis, relatives might experience tension between wanting to know what to expect and needing to remain hopeful. This tension might lead to ignoring prognostic information, relying on one's own beliefs and focusing on positive details [47]. To support relatives in "preparing for the worst and hoping for the best, " sensitive and frequent communication is needed, about the patient's prognosis and about the physician's expectations of the effects of treatment [47,48]. When end-of-life communication is regarded as a continuous process, disclosure of an impending death might be less difficult, for the physician and for the patient and relative, which in turn might yield higher levels of satisfaction with care [48,49].
Nurses discussed the impending death with patients in only one-third of all cases, but when they did, this was associated with increased awareness of patients of their impending death, with the extent to which patients and relatives had been able to say goodbye and with relatives' presence at the moment of death. Nurses are known to experience many difficulties in end-of-life communication, such as uncertainty about their role, and their knowledge, and skills. They also need to wait for the physician to disclose bad news and then often hesitate in initiating difficult conversations with patients and relatives [42,50,51]. Although nurses are generally dedicated to providing the best end-of-life care, they may in practice often continue to provide care as usual, until the physician explicitly and openly states that the patient is dying [52,53].

Strengths and Weaknesses.
To our knowledge this is the first study to examine QOL3 and QOD from the perspectives of relatives, physicians, and nurses, in a large number of cases, without restrictions regarding diagnoses or duration of illness. We comprehensively assessed experiences with endof-life communication and their relation to death preparation and circumstances of death. The response rates were similar to response rates in comparable studies and groups of participants [54][55][56], and the characteristics and results in the 200 common cases were similar to those in the separate groups. Still, some selection bias cannot be ruled out, which might result in more positive experiences of physicians and relatives [57]. Our findings provide concrete opportunities to improve care, instead of satisfaction rates, which are often high in evaluation studies in healthcare [58]. Due to the cross-sectional design we cannot draw firm conclusions about causality. Although 84% of participating relatives were close relatives of the patient, and 88% were involved in the care of the patient during his/her final days, physicians might have informed other family members, who again might not have informed the participating relative. The moment after death at which relatives and HCPs filled in questionnaires differed. HCPs were asked to fill in a questionnaire as soon as possible after the patient's death, to restrict recall bias. Relatives were asked later, to diminish disturbance of their bereavement, but their perspectives might have changed in this period, and recall bias cannot be ruled out. However, emotional events, such as the death of a beloved person, are known to be recalled better compared to less moving events. Finally, we used newly developed questionnaires and as a result our findings cannot easily be compared to those of other studies using other instruments.

Conclusion
Relatives, physicians, and nurses had different perspectives on the quality of the final days of patients' lives, on their quality of dying, and on communication about impending death. Everyone seems to experience his/her "own truth. " To achieve improvements in care in the dying phase, more emphasis should be put on the collaboration between HCPs 8 International Journal of Palliative Care and relatives in the therapeutic alliance with patients. Furthermore, physicians should be more willing to timely discuss a poor prognosis and to check whether the patient and relative's understanding align with physicians' .