Critical care guidelines endorse a shared decision-making model in which the patient, if able, and/or the family play an active role in decision-making [34, 35]. Shared decision making is based on a relationship of mutual respect between the clinical team, who are experts on prognosis and treatment options and the patient/family, who are experts on the patient’s values. The goal is consensus regarding value-laden treatment decisions by discussing the nature of the decision to be made, exchanging relevant information and discussing preferred roles in decision-making [36]. The clinical team shares the burden of decision-making by being available to help clarify the patient’s values and the impact of each option on the patient’s goals.

Clinicians need guidance to operationalize shared decision making into practice [37]. A qualitative study of 51 family conferences in the ICU setting regarding EOL treatment decisions showed substantial variation in the frequency with which individual elements of shared-decision making occurred. Clinicians more frequently provided medical information and made efforts to understand the patient’s treatment preferences, but were less likely to ensure understanding of the medical facts or ask about role preferences in decision-making [37]. Multiple studies have now confirmed a wide variation of preferred roles in decision making among patients and families. Some prefer to retain total control, while others prefer to defer to clinicians [38–40]. (See Table 5.3 for specific suggestions on facilitating shared decision-making.)

Few patients prepare advanced directives specifying desired care in the event of incapacity (a living will) or authorize a specific surrogate decision maker (a health care proxy or durable power of attorney for health care) [44]. Even when present, these advance directives often do not anticipate the specific set of circumstances the patient faces in the ICU [45, 46] and may not diminish the substantial distress surrogates feel in having to make these decisions [45–47]. In a systematic review of 40 studies examining the effect on surrogates of making treatment decisions for incapacitated adults, 27 of which focused on EOL treatment decisions, Wendler et al. reported that surrogates frequently experience significant stress, anxiety, guilt and moral distress [47]. In one prospective cohort study of surrogate decision-makers of incapacitated adults in ICUs, researchers found that 27 % of the sample had relatively low confidence in their ability to act as a surrogate [48]. Lower surrogates’ confidence in their role was associated with poorer quality of communication between the clinician and the family, suggesting that better communication may improve surrogates’ confidence. Surrogates may misconstrue their role. They may underestimate the importance of their input regarding the patient’s values and beliefs or believe the decision should be based on their own values and desires rather than those of the patient [49]. Clinicians often miss opportunities to explain surrogate decision making [50]. Clinicians should clarify sensitively that the patient’s values should drive EOL treatment decisions and acknowledge the difficulty of being a surrogate decision-maker, especially when patient preferences conflict with the family’s emotional needs.

In order to ensure the patient and family understand the patients’ diagnoses and prognosis, clinicians must convey information clearly and avoid medical jargon. Information may need to be repeated over time as integrating complex information under stressful circumstances is difficult. Delivery of information should be structured in order to maximize opportunities for family members to speak and ask questions [51]. An increased proportion of family speech during family meetings is significantly associated with increased satisfaction with clinician communication [51]. One recommended strategy for information disclosure is “Ask-Tell-Ask.” The clinician begins the discussion by asking patients or surrogates to describe their understanding, including what they have been told by other clinicians, and asking permission to provide them information. The clinician then continues with a concise update in layperson’s terms. The family is then asked to summarize what was discussed and are given the opportunity to ask questions or address their concerns [52].

Clinicians may be reluctant to provide information about prognosis or expected outcomes out of concern they would be judged negatively for the inherent uncertainty of the information or that they may extinguish hope or worsen emotional distress [53]. However, patients’ families and surrogates generally desire frank information about prognosis [54], recognize the uncertainty of prognostic estimates and are interested in using prognostic estimates to emotionally and logistically prepare for potential worst outcomes [55, 56]. The process of surrogates preparing and accepting the possibility of death as an outcome is a slow and incremental one [54–56], rife with intrapersonal conflicts and tensions [57, 58] and often limited by psychological biases that make interpreting grim prognosis more difficult for patients and surrogates [59]. In a mixed qualitative and quantitative study in which 80 surrogates of critically ill patients were asked to use a probability scale to interpret sixteen prognostic estimates, researchers found that surrogates were able to accurately interpret statements expressing a high probability of a good outcome but were statistically more optimistic in their interpretation of statements that conveyed a high risk of death [59]. Techniques to enhance surrogate understanding of grim prognostic estimates include: checking that surrogates are ready to hear the information, giving adequate forewarning that the news is not good, acknowledging uncertainty and devoting time to exploring surrogates’ perceptions [60]. In one qualitative study of 179 surrogates’ perception of their critically ill family member’s chance of survival, most appeared to weigh the clinician’s judgment against their own knowledge of the patient’s strength of character, their own observations of the patient while in the ICU, and their own optimism, intuition and religious faith [55]. By sensitively listening to surrogates explain their world view, clinicians can build trust and may decrease the potential for conflict.

Research in decision psychology suggests that strong emotions and affect impact reasoning and information processing [61]. Specifically, emotions can change people’s estimates of both value and risk. For example, when emotions are heightened, people are more likely to substitute how they feel about a particular outcome for a more logical reasoned assessment [61]. In addition, people appear to have difficulty forecasting what they will want and how they will behave in affective states different from their current state (so called, “hot-cold empathy gaps”) [62]. This has relevance to EOL care in the ICU in several ways: surrogates are often making decisions under stress and while grieving over bad news; patients previously expressed wishes were often made when they were in relatively cool affective states making it hard for surrogates to be sure of what they would have wanted during the stress of illness. Despite the extreme importance of emotions and affect in decision-making, clinicians often miss opportunities to acknowledge and address emotions [50, 63]. Strategies for attending explicitly to emotions are listed in Table 5.2.

Patients’ and families’ religious and spiritual values may impact how they make sense of illness and approach EOL decisions. Tolerance and respect for religious and spiritual beliefs should be conveyed [64]. Surrogates with strong spiritual beliefs express more doubt about the accuracy of physician prognostication [65]. When surrogates doubt the clinician’s prediction of poor prognosis, they are more likely to request continued life support [66]. Families may cite spiritual or religious explanations for why they are not willing or ready to discuss transitioning care from restorative to comfort-oriented. In general, such explanations should be seen as an opening for the sensitive exploration of the patient’s or family’s spiritual needs and beliefs. Tools are available for assessing spirituality in the patient encounter [67–69] and reviews have been written for approaching a patient or family with expectations of a miracle [70], or responding to a request for prayer or other spiritual ceremonies [71].

Clinician communication can be supplemented with printed, online or video informational material which may be helpful in improving family members’ understanding of complex medical information [31, 72, 73]. In a multicenter randomized clinical trial to improve the effectiveness of information provided to family surrogates in ICU, standard information (at least one meeting with the physician during the first week of ICU admission) was compared with standard information plus an information leaflet [72]. Family members in the standard communication arm expressed poor comprehension of the diagnosis, prognosis or treatment (as measured by recall questionnaires very soon after the family meeting) about 40 % of the time. This was decreased to 11 % with the addition of the information leaflet [72]. Brochures have been developed to help families prepare for face to face meetings with the clinical team [74] and better understand the syndrome of chronic critical illness, the hallmark of which is the prolonged dependence on mechanical ventilation and other intensive care therapies [75, 76]. Video decision support tools are being studied as an adjunct to advance care planning discussions. A small randomized clinical trial showed that patients with advanced cancer, who viewed the video in addition to communicating verbally with the clinician, reported improved knowledge about CPR and an increased preference for comfort-oriented care near EOL [77]. In a feasibility pre-post study, a CPR video support tool for surrogates in the ICU setting was associated with improved CPR knowledge without any change in their CPR preferences [78]. More research is needed on how to integrate technologies like video and internet with direct clinician communication in order to improve information understanding, relationship building and ultimately improve patient and surrogate outcomes.

Patients with decision making capacity may choose to forgo (withdraw or withhold) any treatment. There is no ethical distinction made among various LSTs, such as mechanical ventilation, extracorporeal membrane oxygenation, left-ventricular assist devices, vasopressors, dialysis, antibiotics, blood products, intravenous fluids or artificial nutrition. The decision is made based on the benefits and burdens of the treatment in the context of the patient’s prognosis and values. The patient’s values may dictate that LST would be undesirable if it merely served to prolong life in a state s/he considers a fate worse than death or the benefit does not justify the burden of treatment. Clinicians can help patients and families navigate decisions about LST by clarifying prognosis and the expected benefit of treatment and exploring patient’s values. Decisions by surrogates to forgo LST on behalf of incapacitated patients are ethically justifiable when made according to patient’s wishes, if known, or in patient’s best interests. The legal authority and responsibility of surrogate decision makers regarding forgoing LST vary from state to state in the United States. Clinicians must be familiar with applicable local laws and regulations.

Although secular ethical principles hold no distinction between withholding and withdrawing LST [79, 80], in practice, stopping a treatment already in place may feel different emotionally from not starting a treatment. Withdrawing a treatment may be emotionally more difficult, especially when death follows soon after, even though the intent is to remove an undesired treatment for which the burden outweighs the benefit or which is merely serving to prolong death from the underlying illness. Particularly when there is significant uncertainty about prognosis, not starting a treatment may be more difficult and leave unresolved the question of whether the patient could have benefited. Reluctance to start a particular treatment in order to avoid deciding to discontinue it later deprives families and clinicians of a time limited trial of therapy [15]. The time limited trial is especially useful when prognosis is uncertain and allows families and clinicians to directly assess the benefits of a particular treatment with the recognition that the treatment may be stopped if it does not achieve the patient’s goals [15].

Withdrawing LST requires the same careful preparation and quality expectations as all other treatments and procedures performed in the ICU. The communication process and the specific plan for the withdrawal of LST should be documented in the chart. Novices should be supervised by expert clinicians, given direct feedback and opportunities for debriefing. Clinicians in training should not be allowed to withdraw LST independently until they have demonstrated competence in a supervised setting.

The goal of withdrawing LST is to remove treatments that are no longer wanted or do not provide any comfort to the patient. In practice, clinicians tend to withdraw treatments in a non-random sequence (for example, withdrawal of mechanical ventilation tends to occur after forgoing dialysis or vasopressors) [81, 82]. This sequential approach (also called stuttering withdrawal) may help clinicians and families create psychological distance between their decisions and the inevitable death and may make it easier for both clinicians and families to place the responsibility for the death on the underlying illness or failing body [83]. In general, guidelines suggest that once a decision has been made to withhold one LST (such as renal hemodialysis or vasopressors), clinicians should carefully consider the utility of other forms of LST.

Interdisciplinary teams should aim for consensus about the decision to forgo LST. All team members with direct patient care roles should feel included in decision making. Clinicians’ decisions to limit LSTs are based on patient characteristics and influenced by multiple non-clinical, idiosyncratic factors [84–86]. A survey of clinicians across multiple university hospitals in Canada showed, that clinicians were more likely to withdraw LST from older patients, patients who they felt were unlikely to survive the current illness episode, and those who had poor baseline cognitive and physical function [85]. Provider level factors like the clinician’s age, number of years since completing training, sub-specialty training, gender, race, religious beliefs may underlie variability in clinician’s attitudes about withdrawal of life sustaining therapies [84, 87, 88]. These variations in approach to forgoing LST may lead to conflict among team members about decisions for individual patients. An approach which emphasizes consensus building among the interdisciplinary team is important for conflict prevention and management [89].