Most patients who die in the intensive care unit (ICU) do so after having life-sustaining interventions withheld or withdrawn. Evidence suggests that more than 70% of elderly patients in the United States require a surrogate decision maker at the end of life, and involving surrogates in medical decision making at the end of life can be exceedingly challenging. For families of patients with life-threatening illnesses in the ICU, many factors can affect decisions to either continue or withdraw life-sustaining interventions. Clinical status and prognosis can affect how both surrogate decision makers and physicians approach such end-of-life decisions. Patient and family factors, including race, ethnicity, culture, language, religion, and spirituality, and socioeconomic status can also shape how these decisions are approached. Fortunately, advance care planning can help both surrogate decision makers and clinicians better understand the wishes of patients who lack decisional capacity. Physician factors, including their own race, religion, and geographic location, can shape attitudes toward end-of-life care decisions and specifically toward opinions about withholding or withdrawing life-sustaining interventions. Communication strategies used by clinicians and institutional and system factors can also have an effect on surrogates’ experiences with decision making.
Medical/Health Status and Prognosis
Only a small minority of patient surrogates cite physicians’ explicit statements of prognosis as their sole source of information on possible outcomes—reporting reliance on their own perceptions of patient factors (e.g., the patient’s character and will to live, the patient’s history of illness and resilience, and the patient’s physical appearance) and surrogate factors such as their own personal outlook, faith, and intuition. In addition, most surrogate decision makers express doubt in physicians’ ability to prognosticate accurately, which may be understandable given that doctors are often poor at prognostication for individual patients and have been shown to overestimate prognosis in terminally ill patients by a factor of 5.3. Advanced patient age, functional limitations, and comorbidities appear to have an important effect on the intensity of care at the end of life—influencing both patients and clinicians’ approaches to these types of care decisions.
Patient Characteristics, Values, and Preferences
Race, Ethnicity, and Culture
Broad differences in preferences for end-of-life care have been noted across racial and ethnic groups, although there is great heterogeneity within groups. As such, clinicians are advised to address treatment preferences specifically with patients and surrogates rather than relying on broad generalizations about group preferences and values. In general, African Americans tend to prefer more aggressive use of life support at the end of life when compared with other racial and ethnic groups. Nonwhite patients use more life-sustaining interventions at the end of life than white patients. Asians and Latinos are more likely to favor a family-centered decision-making process, and nonwhite racial and ethnic groups are less likely to have knowledge of and support with advance care planning. One study comparing white and African American caregivers of patients with advanced lung cancer suggested that African American caregivers had more optimistic expectations for treatment outcomes, which may be part of what shapes these observed differences in expressed preferences by patients and their surrogates. Another study of families of critically ill patients compared family perceptions among patients of similar severity of illness and found that African Americans tended to perceive the illness as less enduring and serious, reported more confidence in treatment efficacy, and reported lower illness comprehension compared with whites. A review found that some of the differences between African American patients and non-Hispanic white patients were related to historic mistrust of the health-care system, differences in knowledge and access to services, and differences in spiritual beliefs between these two groups. In addition, clinicians appear to be influenced by patient race, with physicians being more likely to recommend withdrawal of life support with nonwhite patients, although nonwhite patients are more likely to die with full life support in place.
In a study of U.S. caregivers, those who were less “Americanized” or acculturated to the United States (as determined by language preference and questions related to cultural identity) expressed different preferences for end-of-life care, including a more positive preference for feeding tubes/artificial nutrition, a feeling that they received too much information from physicians, and a desire to receive additional services, including complementary therapies and mental health and nutritional counseling.
Families with limited English proficiency receive less information and fewer explicit statements of support in family conferences, although it is not clear how this affects their decisions about issues such as withdrawal of life support. Use of professional interpreters rather than ad hoc interpreters (e.g., family members) can foster accurate communication and minimize errors in translation, particularly when discussing emotionally charged topics such as end-of-life care. Interpreted family conferences also contain fewer elements of shared decision making and a greater ratio of physician/family speech, which has been associated with decreased family satisfaction.
Religious and Spiritual Beliefs
Patients who identify as religious or spiritual or who identify as using spiritual coping strategies are more likely to choose life-prolonging treatments at the end of life and are more likely to oppose do-not-attempt-resuscitation (DNAR) orders. Specifically, individuals who believe strongly that the length of one’s life is controlled by a higher power are less likely to engage in advance care planning, and those who express deference to God’s will tend to choose more life-prolonging treatments (e.g., continuation of life support). There is great variability in the approaches to medical decision making dictated by theology and by religious authorities, and increasingly, religious authorities have been called on to weigh in on matters related to use of medical technology, medical ethics, and end-of-life care. Furthermore, there is considerable heterogeneity in the application and interpretation of religious tenets within individual religious traditions, and many of the approaches common to specific religious groups are more appropriately attributed to cultural beliefs, not being grounded in specific religious doctrine or teachings. In addition, when patients’ or families’ spiritual care needs go unmet, patients rate their care more poorly. Indeed, evidence suggests that unmet spiritual care needs are associated with increased medical costs at the end of life.
Socioeconomic Status and Education
Socioeconomic status and education may also impact end-of-life care, although very few studies have examined these specific factors in this context. A systematic review found that uninsured patients were more likely to have life support withdrawn in the ICU. In one study, uninsured ICU patients, when compared with those with insurance being cared for at the same hospital, received fewer procedures, including hemodialysis and placement of central venous catheters, and had increased mortality despite being younger in age, having fewer comorbidities, and having a lower probability of death at the time of admission to the ICU. Low literacy has been identified as a risk factor for not having an advance directive in place, which can limit surrogates’ abilities to understand patients’ values and wishes. In addition, low or marginal health literacy has been shown to be associated with a preference for more aggressive care at the end of life.
Role of Advance Directives and Advance Care Planning
Advance care planning generally and advance directives specifically can be helpful for surrogates in clarifying the wishes of their loved one. Indeed, the absence of an advance directive has been identified by ICU directors as a barrier to optimal end-of-life care. Historically, the prevalence of advance directives has been low, ranging from 5% to approximately one third of patients. Several more recent studies demonstrate much higher rates of advance directive use, a change that has been attributed in part to the aging of the U.S. population, as well as increased familiarity with and growing public discourse around the importance of advance care planning. Some evidence suggests that patients who have engaged in advance care planning or who have advance directives are more likely to receive care that mirrors their stated preferences and less likely to receive technologically aggressive interventions. In addition, the presence of a living will has been shown to improve families’ assessments of the quality of death and dying for their loved one in the ICU. Importantly, there is a growing understanding that advance directives are most helpful in the context of a broader process of advance care planning that helps patients and their families prepare to be able to make the best “in the moment” decisions about life-sustaining treatments.
Surrogate Decision Maker/Family Factors
Surrogate Preferences for Control, Role, and Decision Making
Historically, physicians have used a parentalist approach to medical decision making. Calls for increased patient and family autonomy have led to implementation of alternate models of decision control. Decision control can be viewed as a spectrum, with patient and family autonomy/informed consent at one end and clinician parentalism at the other. In between these extremes is shared decision making, a model in which clinicians share medical information; patients and/or surrogate decision makers share information about values, goals, and preferences; and both parties discuss and come to an agreement about an optimal plan of care. Although shared decision making has been endorsed by critical care societies as the preferred default approach, clinicians should recognize that patient/surrogate preferences related to decision control can vary widely and are influenced by factors that include gender, personality, education, socioeconomic status, and culture. To optimize communication, clinicians must assess preferences related to decision control for each patient and family and modify their approach to reflect these preferences. Although it is not clear how these different approaches might affect choices specifically around the withdrawal of life-sustaining interventions, it is important to consider what approach might be preferred and be most effective with surrogate decision makers before beginning family conferences focused on these decisions.
In addition, surrogates use varying approaches to their role as a proxy decision maker. Many medical ethicists and clinicians suggest that surrogates apply the principle of substituted judgment, in other words, asking surrogates to use their knowledge of the patient’s values, goals, and preferences to articulate what the patient would choose were the patient able to participate in medical decision making. Evidence, though, suggests that many surrogate decision makers have difficulty in determining what the wishes of their loved one might be, with about a third of surrogates incorrectly predicting the treatment preferences of patients. This may be in part because some patients’ wishes change and evolve, although most patients show stability in medical preferences over time. In addition, in cases in which surrogates inaccurately predict the wishes of their loved one, their preferences on behalf of their loved one more closely approximate their own personal wishes about end-of-life care, highlighting the challenges of applying a substituted judgment standard. Surrogates use different factors in medical decision making, including factors other than the patient’s perceived wishes; these factors include their own personal values, religious beliefs and preferences, family consensus, and shared experiences with the patient.
Family relationships seem to have an impact on the accuracy of proxies to predict the wishes of their loved one. Spouse proxies have been found to be more accurate than adult children of patients. Patients with highly supportive and well-functioning families are more likely to engage in advance care planning, and lower levels of family conflict have also been associated with higher proxy–patient accuracy in medical decision making.
Patient Preferences for Surrogate Latitude in Decision Making
In addition to the difficulty in implementing substituted judgment as a surrogate decision maker, patients vary in the latitude they choose to give to their surrogate decision makers. Most patients, in the event of decisional incapacity, would want decisions made on their behalf using both substituted judgment and best interest standards and involving both surrogates and physicians. Many patients show a great deal of trust in the decisions of surrogates as well—with more than three quarters of patients in one study preferring that physicians follow the preferences of their surrogate even when those preferences were at odds with previously stated wishes. The fact, though, that some patients prefer that advance directives be followed even if surrogates disagree highlights the importance of discussing surrogate latitude as part of advance care planning.
Physician Bias and Influence
Many clinician factors can influence decisions around the withholding and withdrawing of life support. For example, clinicians’ overall religiosity and specific religious affiliation can influence the likelihood that life-sustaining interventions are either withheld or withdrawn, with religious physicians more likely to favor more aggressive interventions and less likely to favor withdrawal of life-sustaining interventions. In one European study, withdrawal of life support was more common among physicians who identified as Catholic, Protestant, or nonreligious, whereas a decision to withhold rather than withdraw life support was more common among Jewish, Greek Orthodox, and Muslim physicians.
Most physicians select DNAR status for themselves when presented with a hypothetical end-of-life scenario and express a personal preference to receive less aggressive care in general. It remains unclear why physician and layperson preferences are so different and whether these personal preferences have an impact on treatment approaches toward patients. It seems reasonable, though, that clinical experiences and witnessed suffering might affect physicians’ personal preferences. This hypothesis suggests a possible opportunity for physicians to better communicate with patients and their families about their own clinical experiences and to offer to make recommendations to families who likely have considerably less experience with both critical care and end-of-life care.
Physician factors such as white race, residence in North American or northern Europe, more clinical experience, and experience in ICU care predict provision of less technologically aggressive end-of-life care, although there are conflicting data about the effect that physician age might have on comfort with DNAR orders and, in general, on treatment decisions in patients with advanced illness. Medical residents have been found to be marginally more likely than attending physicians to promote aggressive end-of-life care, with the least experienced residents the most likely to prescribe technologically aggressive care at the end of life.
Communication Strategies and Skill
Communication strategies and skill can have a major influence on surrogates’ medical decision making. There is significant variability in physicians’ roles in navigating complex medical decision making, and few physicians explicitly negotiate their roles with individual families. Patient families are also highly variable in their preferences regarding physicians’ recommendations about care decisions at the end of life. One study found that most physicians think that making recommendations about end-of-life care is appropriate, although there is significant heterogeneity in whether physicians actually make these recommendations to families.
Prognostic information can also be easily misunderstood, and some data suggest that surrogates’ interpretation of a report of a poor prognosis may be overly optimistic. Furthermore, in one study nearly a third of surrogates stated that they would choose to continue life-sustaining interventions even in the face of poor prognosis (<1% chance of survival), and 18% opined that they would choose to continue these measures given a physician’s assessment that there was “no chance of survival.” Estimating prognosis accurately for individual patients is incredibly challenging, and, as a rule, physicians tend to be overly optimistic. More experienced physicians and physicians with shorter physician–patient relationships also tend to be more accurate, and prognostication is more accurate as patients approach the end of life. Although accurate prognostication is difficult, using numerical estimates rather than vague language, framing prognosis from both a positive and a negative perspective, and using consistent denominators when estimating risk can all promote better understanding by patients and their surrogates.
Provision of Emotional Support
Explicit statements of empathy by clinicians for family members have been shown to increase family satisfaction with communication, and although effective communication strategies can promote trust on the part of family members, it is not clear how these statements might affect decisions around life-sustaining interventions. Many family conferences do not contain any explicit empathic statements or statements of support for family members. Specific communication tools, such as the NURSE (Name, Understand, Respect, Support, Explore) mnemonic for acknowledging and validating emotion and the facilitated values history, may help in both supporting family members and enhancing the ability of surrogates to understand their role in decision making and to contribute to decision making that reflects the authentic values and wishes of their family member.
Role of Palliative Care Providers
Not surprisingly, ICU patients for whom palliative care providers are consulted tend to be sicker, with longer lengths of stay in the ICU and higher mortality; that is, the sickest patients with the poorest prognosis seem to be identified as those who would benefit most from palliative care services. A randomized controlled trial evaluating the impact of an inpatient palliative care consultation service found no differences in survival or symptom management, although patients seen by the consult service had fewer ICU admissions on hospital readmission and sustained lower costs over a 6-month time horizon. In addition, patients themselves were more satisfied with their care experience and with provider communication specifically.