Dr B was interviewed by a Perspectives editor.
DR B: When the patient had her crisis, we were in a passionate attempt to try more aggressive treatment to palliate her symptoms. It was on a scheduled appointment that I noticed [her] distended neck veins, irritated eyes, and difficulty breathing. She said it all happened in the 24 hours preceding the appointment. We immediately decided that we needed a plan that everybody was comfortable with. First, we had to decide how aggressive to be. I had to make sure that going to the hospital and embarking on emergent treatment was something she thought was appropriate. It was obvious to all of us that the overall journey was nearing its end. This may have been something that would have accelerated her end, and I had to know from her if she wanted to go forward.
Physicians, together with patients and their families, share the burden of decision making throughout the course of a terminal illness. As death nears, such decisions become increasingly difficult as the emotional context becomes magnified. This chapter reviews available evidence and discusses an approach to help physicians provide appropriate care in times of crises near the end of life.
Decision Points in a Terminal Illness
DR B: She had been in excellent physical shape with no muscle mass loss, some decrease in her energy, but otherwise, remarkably asymptomatic. Then, over a very short period of time, [she] began to develop increasing pain. This prompted our switch to a different kind of chemotherapy treatment. Literally, within a week of embarking on our new path, she developed clinically obvious superior vena cava syndrome.
The course of a terminal illness is typically marked by sudden worsening in the clinical condition between periods of relative stability.1 Such events can be asymptomatic (eg, new metastases detected by radiographic studies) or symptomatic (eg, aspiration pneumonia). In each case, the clinician, patient, family, or surrogate decision makers are confronted with deciding whether to pursue continued treatment aimed at prolonging life or instead to adopt an approach focused on specific end-of-life goals and comfort, with the recognition that death is approaching.
As medical technology has advanced, the dividing line between life-sustaining and palliative treatments has blurred. Current evidence suggests that physicians overuse technologically aggressive, life-prolonging treatments and underuse communication skills that can assist patients in choosing from a range of treatment options.2-4 In a study of 164 patients with advanced dementia and metastatic cancer admitted to a tertiary care teaching hospital, Ahronheim et al5 found that 47% of patients received aggressive nonpalliative treatments and that CPR was attempted in 24% of both groups of patients. Teno et al6 reported results from a family survey of 1578 patients who had died in a range of care sites. Compared with hospice care, patients dying in hospitals and other institutions had significant unmet needs concerning emotional support, physician communication, and symptom control.
Beyond the complexity of decision making based on medical facts is the enormous emotional burden these decisions carry. Physicians may view death as a personal failure and struggle with the issue of providing hope while trying to provide truthful information.7,8 Patients, families, or surrogates often experience feelings of guilt, anger, and loss at times when major decisions are required; thus, decision making is a challenge under the best of circumstances. Emotional reactions and the subsequent medical decisions are further influenced by the cultural and spiritual backgrounds of both the physician and the patient.9 (See also Chapter 4, Dealing With Conflict in Caring for the Seriously Ill, Chapter 29, Spiritual Issues in the Care of Dying Patients, and Chapter 32, Negotiating Cross-cultural Issues at the End of Life.)
Clinical decisions that occur near the end of life fall into 2 broad categories: decisions to use potentially life-prolonging treatments for emergency conditions such as respiratory failure and decisions for situations that are nonemergent and typically involve the use of treatment modalities that emphasize quality of life.10 Box 3-1 provides an outline of patient and disease characteristics that clinicians should consider in helping patients make informed decisions. Obvious factors include a treatment risk-benefit analysis and a clear, mutual understanding of the prognosis of the underlying illness.11,12 Less obvious factors include the treatment burden to families (eg, time off from work), the patients' life stage, and the temporal pattern of the illness.13,14
Box 3-1Patient and Disease Characteristics That Affect End-of-Life Decision Making |Favorite Table|Download (.pdf) Box 3-1 Patient and Disease Characteristics That Affect End-of-Life Decision Making
|Cancer vs noncancer diagnosis |
|End-of-life patient goals |
|Treatment risk vs benefit |
|Comorbid illnesses |
|Symptom burden |
|Treatment burden to caregiver |
|Past response to treatment interventions |
|Age and life stage |
|Temporal pattern of illness |
|Treatment financial cost |
|Personality and psychological variables: hope, denial, anticipatory grief, depression |
|Personal and family culture |
The nature of the terminal disease, often divided into cancer vs noncancer, will affect decision making. The pattern of rapid functional decline that occurs in the last 3 months of life for most cancer patients is generally recognized by both patients and clinicians as the beginning of the dying process.2,15 Thus, discussions regarding the appropriateness of various treatment options during this time are held with the recognition that death is approaching. In contrast, there is no clear dividing line for most noncancer diseases, for which the functional decline is often protracted across years.15 Patients, their families, and physicians are more likely to have difficulty recognizing when such a patient is dying and, by implication, when further disease-modifying therapies should be abandoned. This paradox is reflected in the frequent ICU admissions for patients with end-stage chronic diseases.5,16
The patient's end-of-life goals at the time of decision making are also critical factors. In fact, a review of multiple published studies concluded that the ability to make decisions about one's own dying and death is commonly considered a necessary component of a "good death."17 Patients who are at peace with the notion that death is approaching or those who have no specific goals left to accomplish (eg, attend a family wedding) will commonly decline further interventions aimed at life prolongation.18 In contrast, patients with a strong sense of hope for future improvement or those with a strong denial of impending death are likely to seek out additional life-prolonging treatment options. For example, in 2008, Mack and colleagues19 developed a questionnaire to measure the extent to which patients with advanced cancer accepted their terminal illness peacefully or struggled with their illness. The authors administered the questionnaire to 160 cancer patients and assessed other measures, including cognitive acknowledgment that their illness was terminal. End-of-life outcomes were also examined in the subset of 56 patients who died during the study. The 12-item questionnaire had 2 subscales: a 5-item "peaceful acceptance" subscale (Cronbach α = 0.78) and a 7-item "struggle with illness" subscale (Cronbach α = 0.81). Both subscales were significantly associated with the cancer patients' self-reported peacefulness (correlation coefficient [r] = +0.66 for "acceptance" [P < .0001] and r = 0.37 for "struggle" [P < .0001]). "Struggle with illness" scale scores were associated with cognitive acknowledgment of the approach of end of life (mean scores, 14.9 for patients who were aware that their illness was terminal vs 12.4 for patients not aware that their illness was terminal; P = .001). In addition, among the patients who died, the "acceptance" score was inversely associated with use of a feeding tube at the end of life (P = .015).19
The patient's psychological state can affect decision making. An important physician skill is to distinguish clinical depression from the normal grief and sadness associated with the dying process.20 Validated depression screening tools can be helpful.21 Chochinov et al22 demonstrated that using a single question, "Are you depressed?" correctly identified patients with depression in a sample of 24 cancer patients; this single question performed better than the Beck Depression Inventory or depression visual analog scale. Indications for psychiatric referral and treatment strategies have been reviewed elsewhere.20,23 (See also Chapter 24, Psychological Considerations, Growth, and Transcendence at the End of Life.)
Finally, physicians must be aware of how decisions are made and by whom, within the patient's family system, an issue that includes understanding the patient's culture and spiritual belief system.24 The ever-increasing ethnic and cultural diversity of critically ill patients presents a challenge to clinicians who want to provide culturally competent care. For example, minority patients and families who face difficult decisions about end-of-life care in critical care units do not always believe in Western medicine's core values of patient autonomy and self-determination.25 In a study in the United States, Allen et al26 examined the effect of providing enhanced information regarding the risks, benefits, and alternatives for life-sustaining treatments on hypothetical medical decisions and decisional conflict in a study among 78 older, community-dwelling white and African American adults. Measures included the Life Support Preferences/Predictions Questionnaire–modified (LSPQ-m); the Decisional Conflict Scale; and whether enhanced information was provided, such as detailed descriptions of life-sustaining treatment options for each LSPQ-m illness scenario, and the risks, benefits, and alternatives regarding each treatment. Results showed that enhanced information reduced decisional conflict (P = .049) for the hypothetical life-sustaining treatment decisions and produced different patterns of desire for life-sustaining treatments in African Americans than in whites. The authors concluded that clinicians can reduce decisional conflict in patients by providing enhanced information regarding treatment risks, benefits, and alternatives and that, by so doing, they might reduce the desire for life-sustaining treatments in African Americans.
An international study suggested that the physician's own spiritual beliefs matter. Between January 1, 1999, and June 30, 2000, a prospective, observational study of consecutive patients admitted to 17 European ICUs examined the influence of their physicians' religious affiliation and culture on limitations of life-sustaining therapies at the end of life. Among 31 417 ICU patients, 3086 had limitations. Withholding life-sustaining therapies occurred more often than withdrawing them if the physician was Jewish (81%), Greek Orthodox (78%), or Moslem (63%). Withdrawing therapy occurred more often among physicians who were Catholic (53%), Protestant (49%), or of no religious affiliation (47%). The median time from ICU admission to first limitation of therapy was 3.2 days but varied by religious affiliation, from 1.6 days for Protestant to 7.6 days for Greek Orthodox physicians. In addition, decisions around limitations of life-sustaining treatments were more often discussed with families if the physician was Protestant (80%), Catholic (70%), Jewish (63%), or of no religious affiliation (66%).27 (See also Chapter 29, Spiritual Issues in the Care of Dying Patients.)
Organizing an Approach to End-of-Life Decision Making
DR B: She has metastatic breast cancer. My treatment's working very well. She's continuing to work, she's raising a daughter. She's doing everything that life should be about, yet she has an incurable cancer. And we all know that's going to ultimately end her life, yet we have not gone through scenarios at the end of life early enough to be prepared for these kinds of emergencies. I was fortunate that she could talk, that she had all her mental faculties, and she and her husband had discussed this previously with themselves. I was also fortunate in that I had had the opportunity to work with her very closely for almost 2 years and had a close enough relationship with her that I could predict what would be some of her consistent decisions. But that still never makes up for having those discussions earlier in the palliative care treatment of cancer patients.
Physicians can organize their approach to helping patients to make critical end-of-life decisions by assessing the patient's current physical symptoms and psychological and spiritual needs, assessing family and social support systems, estimating and communicating prognosis, and asking the patient to define his or her end-of-life goals.28,29 The optimal timing for this discussion is during a routine outpatient visit for a patient with any chronic life-limiting disease. The patient should have an opportunity to learn from the physician the future expected disease course, potential treatment options, and, together with the physician, define specific goals of care before an acute medical crisis. This type of discussion may or may not result in a change from continued treatments designed to prolong life, and patients with noncancer diseases may be less willing to accept the finality of their illness. Furthermore, as Ms J's case illustrates, not every crisis or possible intervention can be anticipated and discussed before the event. However, beginning the conversation can develop the physician's understanding of the patient's preferences, reassure the patient that the physician is open to discussing end-of-life care, and begin what may be a slow process toward acceptance of a terminal diagnosis.
For optimal communication at the end of life, there must be information gathering, information giving, and relationship building and interventions to enhance communication among patients, families, and clinicians. The existing literature suggests that patients have highly individualized desires for information and clinicians cannot predict patient preferences. In addition, physicians and other health care professionals often do not sufficiently discuss treatment options and quality of life or do not respond to emotional cues from patients. However, physicians and other health care professionals can be taught to communicate better through intensive communication courses, and such communications interventions can improve some patient outcomes.30
Unfortunately, a National Institutes of Health consensus panel concluded that, although enhanced communication among patients, families, and clinicians is crucial to high-quality end-of-life care, end-of-life care is often fragmented among clinicians and health care settings, leading to a lack of continuity of care and impeding the ability to provide high-quality, interdisciplinary care.31
Physician Factors That Influence End-of-Life Decisions
DR B: We tend to postpone our decisions about the end of life…it is too often an abrupt shock to both the patient and the doctor when an emergency like this comes to the forefront. Suddenly, oh my God, we haven't talked about intubation. We've never talked about CPR.
Physician factors can influence end-of-life decisions, including the physician's culture, spiritual beliefs, and personal values.32-34 These issues can transcend medical knowledge or currently accepted bioethical and legal principles. For example, some physicians feel uncomfortable with the concept of withdrawing or withholding parenteral or tube-feeding nutrition near the end of life, fearing that patients will die a painful death, a belief not supported by research findings. In a study in which artificial nutrition and hydration were not initiated for 32 dying cancer patients, McCann et al35 found that symptoms of hunger, thirst, and dry mouth were uncommon and could be easily palliated with small amounts of oral food and use of ice chips.
Ideally, physicians should be able to provide accurate information concerning the range of treatment options in such a manner that their personal biases do not influence decision making while still allowing the patient hope. However, physicians often offer a better prognosis than they believe to be the case,36 and patient characteristics may affect communication. Lamont and Christakis37 performed a prospective cohort study examining physicians' prognostic abilities among cancer patients in community hospice programs. Physicians reported that even if patients asked for prognostic information, they would only provide frank estimates 37% of the time. Cooper-Patrick et al32 completed a telephone survey of 1816 adults from primary care practices and found that blacks rated their visits with physicians as less participatory than did whites, suggesting racial differences within the physician-patient relationship. In a meta-analysis of 29 studies, Roter et al33 found that women physicians spent more time with patients, on average 2 minutes per visit; engaged in more active partnership behaviors; and provided psychosocial counseling. Other factors that anecdotally affect decision making include physicians' feelings that arise in the face of their patient's progressive terminal illness: inadequacy ("I don't know what else to do."); guilt ("I should have been able to do more."); and family, patient, or peer pressure ("What will my colleagues think of me if I don't keep fighting?").
First Steps in Decision Making
DR B: The first decision point was in my office, when all I knew was that she had SVC [superior vena cava] syndrome. I told the patient, "This is what you have, and this is what it means—you can die from this really quickly if we don't do the following treatments." The next thing was explaining to her the various treatment options, including to not do anything at all. She indicated that all measures, including the most aggressive measure, would be appropriate because she wanted to continue to fight.
Several features of Dr B's approach warrant discussion. First, the physician appropriately recognized the urgency of the situation. Second, Dr B allowed sufficient time to explore treatment options and to understand the patient's values and goals. Although it would have been easier to say, "You need to go to the hospital for radiation or you may die," Dr B detailed various options, including no antineoplastic treatment, and allowed the patient an opportunity to express her goals.
The phrasing with which Dr B presented the third treatment option of not doing anything at all can lead to misunderstanding. Dr B was undoubtedly referring to the option of not using antineoplastic treatment. However, the patient may have interpreted this as meaning that nothing would or could be done to relieve shortness of breath and, by extension, that death would occur by suffocation. Patients need reassurance that no matter what treatment option is chosen, the goal of comfort will always be paramount: "A final option is not to use any further treatments to slow down or shrink the cancer but to refocus our efforts to ensure that you are comfortable for whatever time remains." Without a clear description of pharmacologic and nonpharmacologic options for the control of dyspnea, as an alternative to more invasive options, the overwhelmed patient cannot truly make an informed decision.38 Techniques, including sample phrases for end-of-life goal-setting discussions, have been published.8,18,39-43 (See also Chapter 1, Initiating End-of-Life Discussions With Seriously Ill Patients.)
In the midst of making urgent decisions, assessing and conveying prognosis become especially important. Physicians are often reluctant to provide specific information about prognosis, largely out of fear of destroying hope, so information provided is often overly optimistic.7,44-47 Yet, prognostic information is often the most important piece of information that patients need to make informed choices. Physicians who actually know but obfuscate by saying, "Only God can tell" or "There is no way to know for sure," may not be meeting the needs of patients. In a study of 1032 cancer patients undergoing palliative treatments, 84.9% expressed a desire for as much information as possible, good or bad, 7.3% wanted only good news, and 7.8% did not want any details.8 Degner et al,48 reporting results from a cross-sectional survey of 1012 women with breast cancer, found that 1 of the 2 most highly ranked types of information desired from physicians was information about the spread of disease. Finally, Gerle et al49 reported more than 40 years ago that of 101 patients in Sweden with inoperable cancer, the highest rates of anxiety and depression were found among patients who were not told the truth of their condition.
Although destroying hope is feared by physicians, few data support the notion that providing truthful information will destroy hope.8 Loprinzi et al,50 discussing how to respond to patient queries about prognosis, wrote, "Hope should not be static. It should be a dynamic entity that is reframed by circumstance, relationships, and the ongoing development of one's personal world view." In other words, dying patients can still have hope for symptom control, of resolving personal relationships, and for a dignified death. Suggested communication phrases for responding to patient questions about hope have been developed.51 (See also Chapter 1, Initiating End-of-Life Discussions With Seriously Ill Patients.) Ms J was able to make a clear decision based on her personal goals and values. However, many patients and families struggle and ask such questions as "Doctor, what would you do if this were your mother?" The manner in which physicians respond to this query is part of the art of medicine. Some will reflect the question back to the patient: "I really don't know what I would do if this were my mother, but based on what I know of your condition and your goals, I would recommend…" Other physicians feel comfortable personalizing the recommendation: "If this were my mother, I would recommend…" Quill and Brody52 have suggested that it is important for physicians to make clear recommendations using a relationship-centered decision-making approach. They argue that the past 20 years have seen an increase in patient-centered decision making, in which patients are asked to make decisions without physician recommendations. They say that this process represents a limited understanding of the concept of patient autonomy and provides insufficient assistance to patients in navigating the complex medical environment. In Quill and Brody's relationship-centered approach, physicians are urged to consider both the patient's and the physician's values and make recommendations based on both. Unfortunately, there still remains considerable work to ensure a shared decision-making process. White et al53 demonstrated that only 2% of end-of-life communication encounters included all aspects of a shared decision-making approach.
Modifying the Plan: Presenting Critical Information
DR B: Following her first treatment, I went back to the patient with the new information from the CT scan about her airway obstruction. We now had to have an immediate conversation about the various treatment modalities for airway obstruction, up to and including intubation and mechanical ventilation. If radiation produced temporary swelling, if radiation didn't work, and/or she had respiratory failure from an airway obstruction, what should be the plan? During that discussion, she decided not to be intubated, and I supported her 100%. We decided to supplement her with as much oxygen as we could on the ward, but we would not take her to an intensive care unit or intubate her. Her husband had more difficulty with this decision, but after the 3 of us discussed it even more, he understood why.
For Ms J, the new finding of tracheal obstruction necessitated a quick reevaluation of goals and treatment options. Once again, Dr B recognized the gravity of the situation and spent the necessary time to discuss treatment options and end-of-life goals. The new discussion included the options surrounding ICU admission and contingency planning for managing the moment of death, that is, whether or not to perform CPR. Dr B emphasized the treatments that could be used to alleviate distress, such as oxygen and opioids.38
Deciding about ICU admissions, do-not-attempt-resuscitation orders, and use of parenteral hydration or tube-feeding nutrition are among the most challenging discussions in medical practice. The notion of dying may be viewed as an abstract concept early in a chronic illness. However, there is no avoiding the reality of death during discussions of ICU admission or withdrawal of parenteral hydration or tube-feeding nutrition. For many patients and families, deciding to decline further medical interventions represents a relief, a letting go of what has been a long struggle for continued life.54 However, for others, their families, and, all too often, their physicians, the decision to forgo further life-sustaining measures, even in the face of overwhelming evidence of imminent death, is viewed as giving up.55 Guilt, "I should keep fighting"; anger, "Why didn't I go to the doctor sooner?"; shame, "My family will think I am not a fighter"; and fear, "What will happen to me?" are common responses in such situations. Assistance from palliative care specialists, nurses, chaplains, psychologists, and social workers can be especially useful, either individually or together as part of an interdisciplinary patient-family, goals-of-care meeting.16,53,56-63
Negotiating Family Conflicts
The patient-husband dynamic described in this case is familiar to most physicians. The patient was ready to accept defined limitations in future care, whereas the husband, initially, was not. Acceptance of impending death occurs for different people at different times. A truism of palliative care is that dying patients usually know when they are dying and that acceptance of impending death is typically hardest not for the patient but rather for those individuals whose lives intersect with the patient through strong personal connections. Provided the patient consents, the physician can assist the family and other individuals struggling to accept the impending death by including them as active participants in end-of-life discussions with the patient. In this venue, the patient is encouraged to express his/her feelings, concerns, and goals for end-of-life care.64 The physician's role is crucial in helping facilitate these discussions in a manner that empowers patients to openly express their feelings and desires.65 Although nurses, social workers, chaplains, and others can help in this process, the physician's role in providing emotional support, education, and affirming decisions is crucial.66 (See also Chapter 4, Dealing With Conflict in Caring for the Seriously Ill.)
Eliciting Further Patient Goals
The outcome of many goal-setting meetings is a decision to forgo further life-prolonging treatments and to focus solely on comfort. At such a time it is important for all physicians and other health care professionals to demonstrate their support: "I want you to know that I fully support your decision and will do my best to honor your wishes." Once the overall goal has been established, an assessment of patient-specific goals can be made by asking, "What do you need or want to do in the time you have left?" Common replies include ensuring physical comfort, being at home with family, or attending an important family event. Physicians can help patients meet these goals by reviewing the current medications, tests, and interventions and for each to ask themselves "Is this contributing to comfort; is this helping the patient achieve their goals?" Any elements not contributing to the patient's goals should be discontinued. Research data from Finucane et al,67 McCann et al,35 and others reviewed elsewhere68 have demonstrated the lack of benefit and potential harm from continuing artificial nutrition or hydration for those with advanced dementia or near the end of life. For other common interventions, such as continuous pulse oximetry, prophylactic antithrombotic treatments, and parenteral antibiotics, there are few specific data to guide physicians in deciding which treatments should or should not be discontinued; when in doubt, combining patient-defined goals with common sense must suffice.
A final set of questions that can be addressed, once the goals of care become clear, concerns the place of end-of-life care and necessary support for the patient and family. Many, but not all, patients prefer to die at home.69 In a prospective study of 138 patient-family caregiver dyads, about half reported a preference for a home death, but half disagreed on preferred location of death.70 Among 246 community-dwelling persons older than 65 years with congestive heart failure, chronic obstructive pulmonary disease, or recent pneumonia, 43% expressed the desire to die at home.71 The Medicare Hospice Benefit is designed to assist families to care for their dying family member at home by providing nursing, social work, chaplain, volunteer, and physician services. Eligibility requirements for the Medicare benefit include a physician-certified prognosis of less than 6 months, assuming that the disease follows the usual course, and that the goals of care are palliative rather than curative. Most private insurers now have a hospice benefit with some or all of the features of the Medicare benefit. Physicians can assist patients and families to accept the concept of hospice care by saying "We have discussed your goals for the time you have left. To best meet these goals, I am recommending that you enroll in a hospice program." (See also Chapter 36, The Role of Hospice and Other Services.)
Management of patients who approach physicians specifically with a request to withdraw life-prolonging treatments, particularly patients whose deaths are not imminent, such as those receiving long-term kidney dialysis, present a different set of issues. Physicians may be concerned that such requests arise from coercion of family members or represent treatable depression or delirium. Jenkins and Bruera72 have suggested assessment and management guidelines based on their experiences as palliative care clinicians; these guidelines conform to modern ethical principles concerning patient autonomy and decision-making capacity, blended with important principles of palliative care, including thorough patient assessment and treatment to ensure maximal symptom control. (See also Chapter 39, Meeting Palliative Care Needs in Post–Acute Care Settings.)
Keeping Focused on the Goal
DR B: There are many things that I found helpful in this woman's case that reminded me of key areas that make decision making in palliative care emergencies so challenging. First, it was important to give her accurate information as quickly as possible. Second, I felt it was critical to explain the full range of potential outcomes, including the worst case scenario. Third, focusing on the patient's goals is central to decision making.
Dr B summarizes the important aspects of decision making near the end of life. Information must be provided that is timely, honest, and focused around the patient's goals. No laboratory test, clinical pathway, or computer algorithm will replace the central role of the physician in helping patients and families navigate an increasingly complex medical care system. Working as partners with patients and families to make difficult decisions requires in-depth medical knowledge, leadership, the ability to self-reflect, and outstanding communication skills. The challenge for today and the future is to ensure that all physicians acquire and demonstrate these attributes.
The following disclosures were reported at the time this original article was first published in JAMA.
Funding/Support: The Perspectives on Care at the Close of Life section is made possible by a grant from the California HealthCare Foundation.