What are cancer patients’ preferences about treatment at the end of life, and who should start talking about it? A comparison with healthy people and medical staff
- First Online:
- Cite this article as:
- Sahm, S., Will, R. & Hommel, G. Support Care Cancer (2005) 13: 206. doi:10.1007/s00520-004-0725-z
- 218 Views
Goals of the work
In order to strengthen cancer patients’ autonomy and to improve quality of palliative care, it is necessary to know what are the patients’ preferences for treatment at the end of life, whether they accept the idea of advance directives, and who should initiate the process of fulfilling such a document.
Patients and methods
We compared cancer patients’ preferences with respect to particular treatment options at the end of life, acceptance of the idea of advance directives, and preferences for whom should initiate writing such a document with that of healthy controls, nursing staff, and physicians (n=100 each group) using a structured questionnaire.
Cancer patients wanted treatment with antibiotics and infringing treatments such as chemotherapy and dialysis significantly more often than healthy controls, nursing staff, and physicians (p<0.01 and p<0.001, respectively). Determinants associated with the wish to opt for these treatments were reduced health condition and older age. The groups did not differ with respect to their acceptance of advance directives; 58–75% of all those surveyed wanted their physicians to initiate a discussion about writing such a document if they thought it appropriate.
Cancer patients’ preferences for treatment at the end of life significantly differ compared to other groups. Oncologists should initiate a discussion about an advance directive when/if the course of the illness seems to make this appropriate, which corresponds to the wish of the majority of cancer patients, healthy controls, and medical staff.
KeywordsCancer patientsPalliative careAutonomyAdvance directive
Decisions about treatment at the end of life often present a moral challenge when the patient can no longer decide for himself/herself. To preserve patients’ autonomy in case of their inability to decide, the advance directive for medical care is recommended as a means to guide medical decisions by doctors and health care proxies/guardians. Yet, despite intensive public debate in many countries, only a minority of patients has written a directive [14, 15, 22]. This is also the case for cancer patients, despite their intimate experience of potential life-threatening disease , Obviously, there are impediments that hinder people to fulfill a directive, although the majority of surveys maintain that people have the intention to lay down wishes concerning medical treatment at the end of life [8, S. Sahm, unpublished data].
Critics argue that healthy people are not able to make clear decisions in advance with respect to life-threatening clinical conditions. Patients’ wishes might change, and doctors should not follow a “dictatorial” reading of what has been laid down in times when people are better off (so-called change in perspective) [16, 23]. In order to strengthen cancer patients’ autonomy and to improve quality of palliative care, it is necessary to know what patients’ preferences are for treatment at the end of life. Is there a change of perspective, as is often stated if people are confronted with life-threatening disease?
We investigated cancer patients’ wishes concerning treatment at the end of life and compared them with those of healthy controls, nursing staff, and physicians. In addition, the following questions were to be answered: What is the overall acceptability of advance directives in these groups? Which people should take the initiative to write a patient’s advance directive of medical care? Are there emotional impediments that hinder people from fulfilling such a document?
Cancer patients treated at the Oncological Clinic and the Breast Cancer Centre of the German Clinic for Diagnosis were interviewed by means of a questionnaire developed by ourselves. Patients suffering from a newly diagnosed or advanced stage cancer were included. They were asked by their attending physicians whether they would be prepared to take part in the survey, and permission was granted. Patients were informed that data collection was strictly anonymous. The study was approved by the ethics committee of the Medical Council of the State of Hesse, Germany.
Healthy controls, nursing staff, and physicians
As a control group, healthy people who had visited the clinic for a medical checkup were asked to fill in the questionnaire. In addition, nursing staff and physicians of the German Clinic for Diagnosis and participants at a conference for nursing staff as well as participants of a doctors meeting during the annual congress of the German Society for Digestive and Metabolic Diseases in Nuremberg (2003) were interviewed.
Patients and test people (healthy controls, nursing staff, and physicians) were asked to fill in a questionnaire (29 questions) developed by us. It contained questions pertaining to treatments that the person surveyed would like to see included in an advance directive; in particular, people were asked to state their view about the following treatment options at the end of one’s life if one were no longer able to decide for oneself: (1) the giving of fluids through a vein (hereafter called “i.v. fluid”), (2) artificial nutrition, (3) antibiotics for the treatment of pneumonia in the occurrence of serious and fatal disease (“antibiotics”), (4) treatment with pain killers in doses that may potentially dull consciousness (“pain killers”), (5) debilitating treatments such as chemotherapy or dialysis (“chemo/dialysis”) and (6) life-sustaining artificial ventilation (“artificial ventilation”). People could choose out of three answers: “Yes, I would demand in any case,” “never,” “do not know.” We preferred these alternatives as answers instead of offering, e.g., scale of agreement/denial since when writing an advance directive, treatment preferences have to be clear cut for legal reasons.
One of the major caveats against using advance directives for decision making at the end of life is that people may not be able to understand what is meant by any treatment option without being appropriately informed. Yet, being informed by a physician when writing a directive is recommended in many countries, but it is not a precondition for the directive’s legal significance. In the questionnaire, we used terms that are considered to be intelligible to everybody as much as possible. We did not use medical language. For example, instead of “artificial ventilation”, the notion “machines to support respiration” was used.
Furthermore, there were questions about knowledge of advance directives, emotional impediments to dealing with this subject, and who should take the initiative to talk about or to write an advance directive. Physicians were asked whether they have ever mentioned to patients the possibility of writing an advance directive, and all people were asked whether they had been approached about this possibility by their physicians.
In addition, demographic data were collected (age, gender, marital status, having/not having children, living alone/with children/partner). People were asked to give a self-assessment of their state of health; to report educational qualifications (for physicians and nursing staff, their specialities/areas of present employment); to give a self-judgment of their religiousness; and to report about experiences in dealing with serious illness, either in oneself or in a family member.
From a group if items on the questionnaire that asked about knowledge of, intention to write, and emotional reactions when fulfilling an imaginary advance directive, an acceptance score was calculated (maximum acceptance score 18).
Comparisons of the frequency of categorical variables in the groups and the association of categorical variables were calculated with the chi-square test. The comparison of the frequency of ordinal variables in the groups, their correlation with categorical variables, and the comparison of the score between the groups were carried out using the Wilcoxon or Kruskal-Wallis test for nonparametric distributions. The relationship of the score to quantitative variables (age) were calculated with the Spearman correlation and the association of the score with categorical variables with the Kruskal-Wallis test.
Patients and test people
One hundred cancer patients, 100 healthy controls, 100 nursing staff, and 100 physicians fulfilled the questionnaire during August and October 2003. As the evaluation of the questionnaire was carried out anonymously, the percentage of tumor patients who returned the questionnaire of the total number of patients being treated in the clinic at that time can only be estimated. It was estimated at 40%.
Demographical characteristics of thegroup. Data are given as mean (SD) for age and in percent for other items; n=100 in each group.SD standarddeviation, NS not significant
Mean age (SD)
Self-assessment of health state
Very good or good versus tolerable or poor
Healthy controls (n=100)
Nursing staff (n=100)
Preferences concerning treatment options at the end of life
Attitudes towards treatment preferencesat the end of life as would be stated in advance directives. Figures are givenin percent for each group for three alternative answers that interviewees couldchoose from: “Yes, I would demand in any case,” “never,”“do not know.” NS not significant
Significantly more patients wished for treatment with antibiotics in the case of pneumonia in addition to their other illness as well as chemotherapy and dialysis. Far fewer healthy people and medical staff wished for these treatments. The majority in all groups wished for pain therapy, even if this dulled consciousness. However, a low but significant proportion of those surveyed rejected this (at least 8% of patients).
It is important to note the high proportion of those surveyed in all groups who were undecided with regard to therapy options (11–61%, depending upon the therapy option and the group). It was lowest in the question of the administration of painkillers, although even here 19% and 20% of patients and healthy people, respectively, were undecided.
Determinants of treatment preferences
Patients’ preferences for treatment at the end of life may be influenced by the stage of the disease. As data collection was strictly anonymous, we investigated correlations between the wishes expressed by the surveyed people with regard to treatment options (as they would state it in advance) and demographic data, self-assessment of health condition, formal educational qualifications in patients and healthy controls, specialities/area of employment for nurses and doctors, religiousness, and experience with previous serious illness in the surveyed person’s own life or in that of relatives, and the existence of an advance directive.
Surveyed people who did not live with a partner or who were not married tended to reject artificial nutrition (40.9% of those without a partner versus 26.8% of those with a partner; p<0.05). Those surveyed who had written a medical directive tended to reject artificial nutrition (50.8% versus 25.7%; p<0.005) and were less inclined to be undecided with regard to these treatment options (29.8% versus 49.6%).
There was a correlation between the wish for antibiotic treatment and the self-estimation of the state of health. Those who were in a very good and good state of health wished for this treatment significantly less often than those in tolerable or poor health (24.2%, 21.4%, 35.1%, and 40.0%, respectively; p<0.05).
Those living alone more often rejected painkillers that might dull the consciousness (11.7% of those living alone versus 3.7% of those living with a partner; p<0.05). Also, 79% of those living with a partner but only 64% of those living alone (p<0.05) demanded painkillers constantly. Among those who considered themselves religious, only 3% rejected painkillers; of the nonreligious, 8.4% rejected them (p<0.063, tendency only).
With regard to the wish for more aggressive treatment options such as chemotherapy and dialysis, there were clear differences between the healthy and the sick, and there was also an age dependency. Of those who judged their state of health to be good or very good, 19.1% and 27.3%, respectively, wanted this treatment constantly; the proportion of those undecided was 40.8% and 45%, respectively. On the other hand, those who judged their state of health to be tolerable or poor, 47.3% and 50.0%, respectively, (p<0.001) wanted this treatment constantly; the proportion of undecided was smaller: 36.4% and 35%, respectively. Those who rejected this option were younger than those who wanted this treatment (p<0.005) or who were still undecided.
Those who lived alone tended to reject artificial ventilation (76% versus 57%; p<0.05) and were not so often undecided (19.6% versus 34%). Those who had written a medical directive rejected artificial ventilation more often at the end of their lives (p<0.01)
Who should take the initiative for writing an advanced directive for medical care?
Preferences about who should initiatediscussion about the writing of an advance directive. Figures are given in percentof each group for the answers yes/no. Interviewees were asked to react positivelyor negatively to the suggestions on a prepared list
Only the affected patients
Physicians, if they think it appropriate
Physicians as routine matter
Clergy or psychologists
The option that physicians should broach the subject when and if they consider it appropriate was the most popular in all groups. However, only 39% of the physicians had ever mentioned to patients the possibility of writing an advance directive for medical care. Only a small minority in all groups had ever been approached by their doctor on this subject (10%, 3%, 6%, and 6%, respectively, for patients, healthy controls, nursing staff, and physicians).
The expectation that physicians should initiate a discussion about advance directives as a matter of routine was rejected by the majority in all groups, most rigorously by physicians themselves. They were prepared to do this only if they considered it to be appropriate as a result of the individual situation. They would much rather delegate this initiative to other people, usually to the patient’s relatives.
Emotional impediments to writing an advance directive
Emotional barriers to writing an advanceddirective. Interviewees were asked to choose out of these alternative statementswhich positively or negatively express a feeling when writing an imaginary directive.Wording of the question: “Imagine you were now writing an advance directiveor discussing it with your doctor. How would you feel? Please indicate your feelingsout of these statements: a: It makes me feel good/I find it unpleasant; b: It isgood to be able to talk about it/It makes me feel very uncomfortable; c: I am gladthat my relatives now know how they should decide/I am afraid that my relativesmight use the directive in a way which I did not intend; d: It is good to knowwhat will happen to me when I can no longer decide for myself/Now the doctors willnot think and only follow the directive in a dictatory manner.” Figures aregiven in percent for each group of alternative answers.NS not significant
Acceptance of advance directives
Acceptance scores calculated from 18items that asked about knowledge, trust in advanced directives, and positive emotionalreactions when faced with the subject (maximum acceptance score 18); n=100 in each group
If we looked at the groups independently, we found an association between the willingness to accept the idea of advance directives and formal educational qualifications within the group of patients. Willingness to accept was lower among those who left school early with low qualifications than those who left with the equivalent of IGCSE “O” level and those with a college or university education (p<0.005). Acceptance was higher among those who left with the equivalent of IGCSE “O” level than among patients who completed a technical training or apprenticeship (p<0.05).
Wishes about treatment at the end of one’s life as stated in an advance directive
There was a significant difference between patients and the other groups regarding the administration of antibiotics and aggressive treatments such as chemotherapy and dialysis. The patients’ group wanted these types of treatment significantly more often. Those who wanted chemotherapy/dialysis were older than those who rejected them. In our opinion, this points to the existence of a change in perspective, which is often referred to in bioethical debate. This change means a patient’s about-face from rejection of to the wish for a particular type of treatment if he/she is confronted with serious disease.
One could as well argue that this change is not a change of attitudes; rather, it is a change of understanding because the patient has been informed properly. It seems to be impossible to foresee in advance any effect a treatment option may have. For example, chemotherapy may alleviate symptoms even in advanced stages of a disease. One has to keep in mind that we asked interviewees about their preferences as they would state them in advance, not when they were at the end stage of life.
Longitudinal studies about the stability of treatment preferences yielded contradictory results. Among healthy elderly people, expressed wishes proved to be reasonably stable; yet, changes occurred in correlation with changes in the state of health . Among patients with amyotrophic lateral sclerosis, preference with regard to reanimation changed within a few months ; other authors described stable preferences for a time period of 1 year in these patients . This imputed change of perspective is often used as an argument not to take the obligation to honor an advance directive too seriously [12, 18, 23]. Our research supports this assumption.
Many clinicians would prefer an approach that would consider the relevant circumstances of each case at the time the decision is to be made. We think our data support this conviction. Yet, this clinical experience has to be considered against a political background in which the binding nature of advance directives will be strengthened by legal measures in many European countries, e.g., Great Britain and Germany [4, 21]. Against this background, we think our findings are of interest and in line with data from the United States that show that a “change of perspective” at the end of life might change patients’ attitudes toward treatment preferences .
So far, there are no studies about the feasibility of advance directives comparing medical staff, healthy people, and patients in Europe. Our data support that documents written when people are healthy are of little value in many circumstances. In addition, one might conclude that the capacity to imagine in advance what treatment a person will prefer in the future is lacking in most cases. In particular, the high percentage of interviewees who were undecided shows that many were aware of not being able to predict what they will want in a future situation. If patients lack the capacity to decide for themselves, the solution should not be to follow a written document in a dictatorial manner. An alternative would be the designation of health care proxies with whom matters should be discussed.
In our study we could not analyze the group of patients according to their stage of disease. As has been said above, answers might change during the course of an illness. We tried to take health condition into consideration by analyzing determinants of patients’ preferences. Although we cannot dismiss the effect of selection (a higher percentage of cancer patients receive chemotherapy), we cannot establish any factors apart from the expected demographic differences (the greater age of the patients) and their poorer state of health, which would explain the difference between the treatment wishes among patients and those among the other groups. In agreement with this, Slevin et al. found a higher willingness among cancer patients to undergo debilitating treatments even if the chances of success were slight .
Hence, the assumption that there is a change of perspective during the course of an illness seems to be reasonable. Thus, a correlation exists between the wish to receive antibiotics in case of concurrent pneumonia at the end of one’s life and the self-estimation of one’s state of health. Those surveyed who were in a good state of health wished for this treatment significantly less often than those who were in a tolerable or poor state of health. The same goes for the option for the treatment by chemotherapy/dialysis. Half of those who considered themselves to be in a tolerable or poor state of health wanted this option; the proportion of those undecided was lower. The opposite was the case for those who considered themselves to be in a very good or good state of health. They rarely wanted this option and were more often undecided.
We were surprised by the high proportion of those surveyed in all groups who wanted to be given liquids and artificial nutrition at the end of their lives. The question may be raised whether the way questions are asked will have impact on the results gained. With respect to advance directives, this problem turns out to be crucial since the way it is asked will influence the answer. That may lead to an attitude to reject the feasibility of the instrument at all. There will never be a “gold standard” to ask patients in a way in which there will be no misunderstanding.
Our results show that patients probably cannot decide in advance whether receiving i.v. fluids will be futile or not. And, not surprisingly, many interviewees in our study rejected to decide in advance and answered “do not know.” We believe that many people are unnecessarily worried that patients might suffer if they receive no i.v. liquid or calories when they are dying. Rather, studies have shown that hydration may be futile, perhaps even counterproductive and prolong suffering . Hence, any decision to offer these treatments when patients are dying should be based on sound clinical judgment instead of using advance documents that were written in advance when the situation was not foreseeable [2, 10, 18].
It is reasonable when prescribing these measures (i.e., nutrition and hydration), to distinguish between groups of patients: the dying and those for whom the physician’s prognosis is that they are not yet dying . Such a distinction is made in the principles of the German Medical Council for medical care for the dying . For the satisfaction of the subjective needs of the dying, i.e., the stilling of hunger and thirst, these patients benefit from, e.g., moistening their mouth more than from balancing physiological parameters . Thus, the intravenous administration of liquids and artificial nutrition is not necessary in all cases. Yet, it goes without saying that it may be indicated if it reduces suffering depending on circumstances, e.g., to treat loss of consciousness in case of dehydration . Physicians should explain these differences when communicating with patients and their relatives, particularly as the necessity to administer liquids and artificial nutrition to patients at the end of their lives is often the cause of conflict between medical staff and relatives.
The high proportion of those surveyed who were undecided with regard to the various therapy options is of utmost significance. With the exception of the option of painkillers, the proportion wavers between 24% and 61%. The same goes for the professionals, i.e., the nursing staff and the physicians. Considering that it is reasonable to assume that those surveyed were highly motivated to concern themselves with this difficult subject (substantiated by the fact that they completed the questionnaire), these results are remarkable. The extent of uncertainty with regard to wishes for treatment at the end of one’s life may be an obstacle that keeps people from writing an advance directive.
The fact that a significant proportion of patients and healthy controls (8% and 6%) rejected an effective dose of painkillers if these might dull consciousness is worth noting. It is in accordance with studies by other authors . Physicians must understand that patients often fear the loss of control and consciousness more than they fear pain. However, a correctly balanced and administered pain therapy only rarely leads to significant restrictions of consciousness and the ability to make decisions. Physicians ought to communicate this to their patients.
Our results point to a further significant factor: those surveyed who lived alone and who were not married rejected artificial nutrition, the administration of antibiotics, and artificial ventilation more often than those who were married or live with a partner. Not only medical factors but demographic and cultural factors such as race and education determine treatment preferences at the end of life . Our study verifies the influence of lack of family and the loss of intimate social contact on treatment wishes. In our opinion, this should be considered when informed consent is discussed with patients. Doctors should carefully explore the reasons life-sustaining treatment is rejected. Loneliness and loss of a social safety net may be underlying factors. Physicians and institutions should offer appropriate support where possible. These results correlate well with the analysis of determinants that motivate patients in the Federal State of Oregon in the United States to seek help in assisted suicide. Thus, those who ask for help in killing themselves are usually not married and live alone .
Of course, the results with respect to treatment options at the end of life (as imagined in advance) gained in our group of healthy controls must not be extended to the general population. We chose the group of health-concerned people because it seemed to be a group that might be reached by activities to promote the spread of advance directives. The writing of advance directives needs an initiating act. For example, the US ruling (The Patient Self-Determination Act) forces providers to offer people that seek medical advice/treatment the possibility to fulfill a medical directive. So, for purposes of comparison, a group is best suited that seeks contact with the medical system.
Initiative to write an advance directive
Only a minority of those surveyed believed that the initiative to write an advance directive should emanate from the patients themselves. A great majority wanted physicians to initiate the discussion when they consider it appropriate. The physicians themselves agreed with this. This result contrasts with our finding that only one third of physicians has ever had discussions with patients about advance directives and that an even smaller proportion of patients has ever been approached by their physicians on this subject.
Physicians, more often than others surveyed, preferred to delegate this task to relatives and other professional groups. Lay people are usually rejected for this role. Physicians ought to be aware of their patients’ expectations. Patients want doctors to broach the subject and consider doing so as part of their work. Our results can be interpreted as an expression of trust that the physician will recognize the right time at which such a discussion might become necessary. It is evident that physicians rarely meet this expectation. To improve medical care and to uphold patients’ autonomy at the end of life, physicians should start discussion about advance directives with their patients if they think it appropriate in the course of disease. Physicians should be prepared to do so. As others have argued, the aim should be a more comprehensive advanced care planning for patients by improving physicians’ communication skills . The advance directive would then serve as a means to promote communication between patients and physicians. This interpretation is also supported by the results of other studies [5, 13].
Acceptance of advance directives and emotional impediments to write such a document
We found no difference with respect to the overall acceptance of the idea of advance directives between the groups. The great majority of those surveyed associated positive feelings with the writing of an advance directive. However, one quarter of physicians expressed negative feelings. There is nothing known at the moment about factors that cause negative emotions to a higher degree in physicians. This result might explain the factual reserve about initiating a discussion with their patients about advance directives.
In conclusion, our research shows significant differences with respect to particular treatment options at the end of life between cancer patients, healthy controls, and medical staff. Our findings may help palliative care workers to understand their patients’ preferences and may help to improve decision making about life-sustaining treatments together with their patients. To uphold patients’ autonomy at the end of life, oncologists should initiate a discussion about an advance directive when/if the course of the illness seems to make this appropriate, which corresponds to the wishes of the majority of cancer patients, healthy controls, and medical staff. Furthermore, the results of the study show that written documents are of little value only in ascertaining patients’ wishes at the end of life. This should be considered when policies about the use of advance directives are made and strategies to strengthen the binding nature of such documents are discussed. Alternative approaches such as advanced care planning and designation of health care proxies should be taken into consideration.
The authors are indebted to Anne Boediger who helped to prepare this manuscript.