The conduct of sensitive and timely conversations about the possibility of death or other lasting harm as a consequence of critical illness is a central aspect of PICU practice. Not all families with children in PICUs require conversations about the potential of death, to disclose significantly increased risk of death, nor need reassurance about routine risk. A PICU mortality of less than 5% in most modern PICUs (1) and increasing numbers of children living with chronic conditions and technology dependence (2, 3) have supported the expectation of survival as routine. Consequently, disclosure of an increased risk of death may be unintentionally avoided unless explicitly considered. Decisions should be consciously made for each child in the PICU.
This article is about the circumstances when PICU physicians consider discussion about risk of death and the practical questions that require consideration to inform these important and potentially difficult conversations. When is the best time to have the conversation? Which children are at increased risk? What information is useful to have in advance of the conversation and how to prepare for the conversation itself? The perspectives presented are based on experiences of the authors and supported by available literature. Please refer to Table 1 for a summary of the core principles that inform the identification through to the disclosure of an increased risk of death.
WHAT ARE THE POTENTIAL BENEFITS AND HARMS OF TIMELY CONVERSATIONS ABOUT RISK OF DEATH?
The principal benefit and therapeutic value of timely discussions about the probability of death during ICU admission is simple: time. Early active consideration of the possibility of death enables time for information gathering about the child’s background—disease prognosis, previous critical illness, possible and feasible treatment options for underlying and current critical illness—to better understand the social situation and subsequently for involvement of bereavement services, engagement of other supports, and the creation of legacy items. Timely consideration also allows the PICU team to incorporate community or subspecialist perspectives on disease-specific treatment options, prognosis, and to gather patient- and family-specific knowledge. The passage of time usually increases clarity about illness trajectory and underscores to the importance of considering when a conversation should occur—or more specifically when should the first formal conversation with the family occur.
Timely conversations help align the families perceived risk with the PICU-perceived risk. Common understanding about risk of death can underscore the relevance of subsequent decision-making. Conversations should distinguish between the ideal or desired outcome, probable versus “possible” outcomes and identify areas of potential conflict. Separation of these items into those hoped for and those prepared for can be a constructive prelude to discussions about extent of therapy.
Timely conversations with families about risk of death can create bookmarks to follow-up during future conversations, enable iterative adjustment to static or evolving realities, and permit engagement of wider support networks. A family who did not previously appreciate an imminent risk of death may use the opportunity to engage in various end-of-life processes—including bringing relevant family members from local and remote places.
Conversations about death may have adverse consequences. There are several reasons this may occur. First, a conversation may be discordant with the child’s actual condition. In our experience, this is uncommon, and more often parents or families are not prepared to hear the information or do not realize the severity of the situation. In either situation, initiating an explicit conversation about death may introduce doubt or uneasiness that does not benefit the family or physician-patient relationship. This may undermine perceptions of expertise in the PICU physician, amplify interpretations of the magnitude of uncertainty, and complicate future conversations (4). The importance of having the potentially unwanted conversation should be balanced with efforts to ensure that the conversation occurs and the message is delivered well. Delivering the message well includes understanding how families and providers wish to be involved in this process (5) and understanding, respecting, and incorporating the apparent discordance between hope and reality (6, 7).
Many children for whom risk of death conversations occur will survive. Recollections of statements that “death will occur” or similar may be interpreted as a PICU team who at best “worked hard and did well for the child” or at worst were unrealistically “pessimistic” or were simply “ill-informed” about the actual risk. Perceptions that PICU expertise is limited may reduce uptake of future “bad news” and lead to experience-based optimism. Families of children with chronic disease may not appreciate their child as having increased baseline risk of subsequent critical illness and death following successful PICU discharge. Prospective documentation of the possible outcomes described—for example, death, survival with new disability, survival—can provide an important reference and enable more accurate integration of past conversations into present understanding and communication between the PICU and the family.
For families with an unrealistically heightened sense of the risk of death, a brief, reassuring conversation may acknowledge the risk and share the noncomplacent expectation of the PICU team that the child will survive. This acknowledgment can mitigate these unspoken fears, promote a culture of transparency, and strengthen relationships between PICU and family. Opportunities to neutrally include information about anticipated PICU outcome as a natural part of routine interaction should be used. Positively framed statements may include discussion of PICU discharge, time in hospital, return to usual activities rather than a “special” message about risk of death.
WHEN AND HOW DO WE IDENTIFY THAT A CHILD IS AT INCREASED RISK OF DEATH?
If the risk of mortality is both apparent and low, not discussing death is probably the best option. One rule of thumb we use is that if we are thinking that a patient may die, then they are at increased risk. At worse, this rule is diffuse, circular and prone to confirmation bias. Conversely, clinical intuition is an important part of quality care that incorporates published data, local experience, and generic observations for patterns of disease and treatment in specific populations (8).
Conversations may occur for children who are at “increased” risk of death if a treatment is required to prevent death and where one or more stakeholder groups (parents, healthcare professionals) think it is reasonable to withhold the treatment. Examples include invasive mechanical ventilation in a child with untreatable malignancy or Extracorporeal Life Support (ECLS) in a child with severe hypoxic-ischemic encephalopathy. Cues from Severity of Illness scores such as Pediatric Risk of Mortality, Pediatric Index of Mortality, and PELOD provide supportive data but are not intended for use in individual patients (1, 9). Increasingly, brain injury and CNS failure have been cited as the most proximate cause of death in PICU populations (10) and may represent the relative importance of neurologic outcome and functional quality of life in discussions about death. This in turn likely contributes to withdrawal of life-sustaining therapy accounting for 40–70% of deaths in the PICU (3, 11, 12).
Some patients admitted to the PICU have already been given a life-limiting diagnosis with known population survival curves and risk of death. For other patients, the diagnosis is unknown or uncertain, and prognostication is based on measures of multiorgan dysfunction, the extent of technology dependence, response to treatment, and degree of CNS impairment. Often both diagnosis-dependent and diagnosis-independent risks of death overlap and attempts at disentangling progression of chronic disease from an intercurrent illness can be challenging.
More than 70% of children who die in a PICU have an underlying chronic illness (13). “Chronic” may be also defined by time in ICU—perhaps 10, 14, or 28 days in one, or over several, admissions (14). ICU mortality is not 100%; however, 6-month mortality is significant and often associated with prolonged hospitalization, recurrent ICU admission, and interval technology dependence (15). Relative stability in ICU may obscure accumulating mortality risk, and day-to-day management of acute illness may overshadow a persistent background disease. It is conceivable that conversations for these children may be inadvertently avoided, during the first, second, or subsequent ICU admissions.
HOW CAN WE BE CERTAIN THAT THE RISK OF DEATH IS ACCURATE?
Intensive care physicians must routinely contend with the uncertainty associated with predictions of survival for patients where the literature to assist prognostication is limited (16). This is particularly true for the growing population of patients with rare congenital disorders now beginning to transition to adult programs and the changing landscape of conditions such as spinal muscular atrophy (17–20) and bone marrow transplantation (21). Existing medical evidence, current diagnostic information, personal and local clinical experience, and local or remote subspecialty consultation are all important inputs into assigning a thoughtful and accurate risk of death.
Consideration of alternate therapies—to either exclude or suggest as reasonable “options”—can help refine risk of death assessment. Exploratory questions such as the role of ECLS, surgical options, and limitation of resuscitative efforts can provide a tangible construct around which to weigh the risks and benefits of reasonable alternates. Discussions should focus on feasible therapies with clear “potential” to benefit and the risks of those therapies. Balancing of risks should be based on a clear understanding of the underlying disease process and worst- and best-case impact on the patient’s quality of life and the nature and quality of death.
The clinical opinions and judgments of physicians may be susceptible to various cognitive biases that can unintentionally lend greater weight to certain elements of prognostication and risk of death assessment (22, 23). Clinical decision-making uses cognitive pathways that have been described as intuitive and analytical (24, 25). Intuitive (type 1) pathways are fast, take shortcuts to reduce the cognitive burden that requires increased cellular metabolism and in many circumstances, are unconscious. Analytical (type 2) pathways are slow, more deliberate and reliable, and ultimately require increased energy expenditure (24, 25). Type 1 processes are more prone to biases and more likely to occur at times of fatigue, sleep deprivation, hunger, interruptions, and emotional distress (24)—things that occur in both clinicians and parents in ICU (26–28). Allocating uninterrupted time for meetings at times of lower clinical workload, having others “cover” during meetings, otherwise limiting interruptions, and ensuring well-rested parents may be helpful strategies to improve the quality of meetings particularly when decisions or recommendations are the intended outcome.
Team morale may impact communal thinking. If there is low morale and a high degree of moral distress related to the care of a patient, it may increase the risk of “groupthink” and can threaten optimal decision-making process and decisions (22, 29). For individual clinicians, confirmation bias may occur when information supporting preconceived beliefs about the outcomes of patients with certain conditions or acuity of illness are favored, and contradictory information ignored (22, 23). Similarly, recency bias can occur when the outcomes of most recent similar “cases” dominate the present, and clinicians overlook or downplay other relevant data (22, 23). These biases may be mitigated by consultation with ICU colleagues to review mortality risk and may provide peer-to-peer support for the primary clinician.
Individuals may reach a different conclusion from the same set of information depending on how the information is presented or perceived (23). This framing may be highlighted best in how differences in quality of life for children with disabilities are viewed by families compared with others (30, 31). Healthcare professionals in ICU see these children when they are critically ill—a very different frame of reference that may be held by their parents or other clinicians who conduct regular outpatient appointments when the children are well (32). Involvement of family doctors, general pediatricians, subspecialists, and other multidisciplinary care providers in collaborative team meetings can describe the patient’s premorbid functioning to mitigate these potential framing effects.
PREPARING AND CONDUCTING THE CONVERSATION
Conversations about death should be matched with the patient’s circumstances (33). The child with fulminant critical illness may never reach stable state to permit a formal antemortem conversation between the ICU physician and parents. For these patients, an imperfectly informed first conversation with ongoing, brief updates may ensure that the above benefits of a timely conversation are not lost while waiting for opportunity to have a more comprehensive and precise conversation. Parents may be in the room, by their preference, supported by institutional policy, which may enhance their appreciation of the disease severity (34). To healthcare professionals, risk of death may be apparent, conversations may seem an unnecessary distraction from the act of providing care, and the act of stepping away may be perceived as “standing down.” A short, compassionate conversation from an available member of the ICU team or a conversation with a historically involved non-ICU healthcare professional can help the family with acute understanding, communications to extended family, and framing of short-term expectations before a more formal conversation.
We prefer formal conversations occur in private spaces, separate from the bedside, with seating, light, and consideration to safety for the healthcare providers if tensions run high. Situation-specific consideration should complement the inclusion of parents, and/or relevant family members. The patient’s nurse can provide current context, insights of bedside conversations, and questions from the family (35). Furthermore, first-hand exposure to the conversation better informs future bedside conversations frontline staff may have about the range of expected outcomes, the rationale for plans made, and the resulting actions. Other healthcare team members complementing the ICU team may bring disease-specific expertise, historical patient knowledge, and provide temporal continuity for the family (36).
Meetings can have too many people. Involvement of multiple services with trainees, nurse practitioners, social workers, and others must strike a balance between “inclusive and informed” versus “voyeuristic and overwhelming.” This distracting environment may undermine the therapeutic intent of the conversation. In our experience, a brief premeeting or huddle with key stakeholders usually ensures a shared mental model of the intended goals of the meeting, the social situation, and consistency in understanding of, and communications about, disease, treatments, and prognosis (36). After this premeeting, nonessential persons are offered the opportunity to leave, and the seating arranged such that parents are well-placed to interact with the remaining team members. The conversation should open with introductions that include the name, title and role of each participant, and addressing the child by their accepted name, after which the broad purpose of the meeting should be described, additions sought, considered and incorporated, or constructively deferred. Clarifying questions to probe the parental understanding of their child’s background and present illness can establish a baseline on which the conversation is grounded (35).
The purpose of the meeting should be apparent to its participants. The goals may be to communicate a new or increased risk of death and set the stage for a future meeting to discuss treatment options, to present a single clear recommendation, to explore several treatment options that may all be “reasonable” in the professional judgment of the ICU team, or to discuss mode of death: how a child may die. We see the discussions about risk of death, recommendations and end-of-life decision-making as separate and ideally sequential. If families are struggling to understand or accept the objective information presented about risk of death, then they will likely perceive any recommendations in a similar fashion. This highlights the importance of separating disclosure of risk of death and ensuring the understanding of the family before articulating subsequent recommendations. It is common for families to require more than one meeting to reach consensus about limitation or withdrawal of life-sustaining therapy (11).
Good recommendations and decisions are based on well-understood data and the patient’s journey to the current point in time. Understanding should be clear and objective and include some appreciation of the degree of uncertainty. The data may not be perfect, but it should be the best currently available and framed multiple ways for families and others. The weight given to the value judgments, preferences, and recommendations of ICU physicians and other stakeholder groups about these risks has changed over time (see personal reflection articles), is a sociopolitical phenomenon, can be a cause of moral distress in healthcare professionals (29, 37) and is not the focus of this article.
Decisions may need to be made in the meeting, or subsequently. Scheduling a next meeting in a reasonable timeframe can enable those with longstanding relationships and professional friendships to provide emotional support for children and their families; can provide time for the family to process the information of the meeting; and similarly, provides time for healthcare professionals to recalibrate their longer term hopes and expectations with the current reality. In this interval, exploring preferences and values of parents and other stakeholders may help inform a decision and can identify potential areas of disagreement (35).
In some countries, the responsible physician will be expected to decide, in others decisional authority resides with parents. Communicating concern or worry about the child’s chances even with on-going active treatment may give permission for families to reveal their perspectives about an alternative therapy or withdrawal of life-sustaining therapy and lead to a clear, consensus-based plan. Irrespective of the individual, group or groups with decisional authority, these important decisions deserve the objective and compartmentalized presentation of best available information about the probabilities of possible outcomes. Consensus is desirable but may not be necessary. Lack of consensus does not exclude understanding of the rationale and thoughtful intent of family preferences and/or physician recommendations underpinning subsequent decisions.
The death of a critically ill child is infrequent and can have long-lasting impacts on families and healthcare providers. Healthcare professionals in ICU should actively consider the need to have conversations about possible death in all children in ICU and the benefit of time for families, pre-ICU healthcare professionals, and the ICU team to prepare should death occur. Conversations should be carefully considered, information gleaned from broad sources, potential physician biases considered and mitigated, and sociocultural perspectives for methods of sharing information understood and incorporated. Separation of facts from recommendations and decision-making can increase the quality of communication, increase alignment of family and the healthcare team, and ensure the child remains the focus of care.
1. Wolfler A, Osello R, Gualino J, et al; Pediatric Intensive Therapy Network (TIPNet) Study Group: The importance of mortality risk assessment: Validation of the pediatric index of mortality 3 score. Pediatr Crit Care Med 2016; 17:251–256
2. Spratling R. Understanding the health care utilization of children
who require medical technology: A descriptive study of children
who require tracheostomies. Appl Nurs Res 2017; 34:62–65
3. Burns JP, Sellers DE, Meyer EC, et al. Epidemiology of death
in the PICU at five U.S. teaching hospitals*. Crit Care Med 2014; 42:2101–2108
4. Roscigno CI, Grant G, Savage TA, et al. Parent perceptions of early prognostic encounters following children
’s severe traumatic brain injury: ‘Locked up in this cage of absolute horror.’ Brain Inj 2013; 27:1536–1548
5. White DB, Malvar G, Karr J, et al. Expanding the paradigm of the physician’s role in surrogate decision-making: An empirically derived framework. Crit Care Med 2010; 38:743–750
6. Roscigno CI, Savage TA, Kavanaugh K, et al. Divergent views of hope influencing communications between parents and hospital providers. Qual Health Res 2012; 22:1232–1246
7. Sisk BA, Malone JR. Hope, optimism, and compassionate communication. JAMA Pediatr 2018; 172:311–312
8. Brush JE Jr, Sherbino J, Norman GR. How expert clinicians intuitively recognize a medical diagnosis. Am J Med 2017; 130:629–634
9. Pollack MM, Holubkov R, Funai T, et al; Eunice Kennedy Shriver National Institute of Child Health and Human Development Collaborative Pediatric Critical Care Research Network: The Pediatric Risk of Mortality score: Update 2015. Pediatr Crit Care Med 2016; 17:2–9
10. Au AK, Carcillo JA, Clark RS, et al. Brain injuries and neurological system failure are the most common proximate causes of death
admitted to a pediatric intensive care unit. Pediatr Crit Care Med 2011; 12:566–571
11. Garros D, Rosychuk RJ, Cox PN. Circumstances surrounding end of life in a pediatric intensive care unit. Pediatrics
12. Sands R, Manning JC, Vyas H, et al. Characteristics of deaths in paediatric intensive care: A 10-year study. Nurs Crit Care 2009; 14:235–240
13. Meert KL, Keele L, Morrison W, et al; Eunice Kennedy Shriver National Institute of Child Health and Human Development Collaborative Pediatric Critical Care Research Network: End-of-life practices among tertiary care PICUs in the United States: A multicenter study. Pediatr Crit Care Med 2015; 16:e231–e238
14. Peterson-Carmichael SL, Cheifetz IM. The chronically critically ill patient: Pediatric considerations. Respir Care 2012; 57:993–1002; discussion 1002–1003
15. Pinto NP, Rhinesmith EW, Kim TY, et al. Long-term function after pediatric critical illness: Results from the survivor outcomes study. Pediatr Crit Care Med 2017; 18:e122–e130
16. Jones PM. Thoughtfulness and grace: End-of-life decision making for children
with severe developmental disabilities. Am J Bioeth 2016; 16:72–73
17. Mendell JR, Al-Zaidy S, Shell R, et al. Single-dose gene-replacement therapy for spinal muscular atrophy. N Engl J Med 2017; 377:1713–1722
18. Bach JR, Baird JS, Plosky D, et al. Spinal muscular atrophy type 1: Management and outcomes. Pediatr Pulmonol 2002; 34:16–22
19. Finkel RS, Chiriboga CA, Vajsar J, et al. Treatment of infantile-onset spinal muscular atrophy with nusinersen: A phase 2, open-label, dose-escalation study. Lancet 2016; 388:3017–3026
20. Wilkinson DJ, Gillam LH. Should long-term ventilation be offered in severe spinal muscular atrophy. J Paediatr Child Health 2013; 49:813–814
21. Zinter MS, Holubkov R, Steurer MA, et al. Pediatric hematopoietic cell transplant patients who survive critical illness frequently have significant but recoverable decline in functional status. Biol Blood Marrow Transplant 2018; 24:330–336
22. Croskerry P. Henriksen K, Battles JB, Marks ES. Diagnostic failure: A cognitive and affective approach. In: Advances in Patient Safety: From Research to Implementation. 2005, pp 241–252Rockville, MD, Agency for Healthcare Research and Quality (US).
23. The Patient Safety Education Program: Module 2: Human Factors Design: Application for Healthcare. 2011Canada.
24. Croskerry P, Singhal G, Mamede S. Cognitive debiasing 1: Origins of bias and theory of debiasing. BMJ Qual Saf 2013; 22(Suppl 2):ii58–ii64
25. Kahneman D. Thinking Fast and Slow. 2013Toronto, ON, Canada, Anchor Canada.
26. Stremler R, Dhukai Z, Pullenayegum E, et al. Sleep, sleepiness, and fatigue outcomes for parents of critically ill children
. Pediatr Crit Care Med 2014; 15:e56–e65
27. Stremler R, Dhukai Z, Wong L, et al. Factors influencing sleep for parents of critically ill hospitalised children
: A qualitative analysis. Intensive Crit Care Nurs 2011; 27:37–45
28. Parshuram CS, Dhanani S, Kirsh JA, et al. Fellowship training, workload, fatigue and physical stress: A prospective observational study. CMAJ 2004; 170:965–970
29. Austin W, Kelecevic J, Goble E, et al. An overview of moral distress and the paediatric intensive care team. Nurs Ethics 2009; 16:57–68
30. Tait AR, Voepel-Lewis T, Zikmund-Fisher BJ, et al. Presenting research risks and benefits to parents: Does format matter? Anesth Analg 2010; 111:718–723
31. Peters E, Hart PS, Fraenkel L. Informing patients: The influence of numeracy, framing, and format of side effect information on risk perceptions. Med Decis Making 2011; 31:432–436
32. de Vos MA, van der Heide A, Maurice-Stam H, et al. The process of end-of-life decision-making in pediatrics
: A national survey in the Netherlands. Pediatrics
33. Janvier A, Barrington K, Farlow B. Communication with parents concerning withholding or withdrawing of life-sustaining interventions in neonatology. Semin Perinatol 2014; 38:38–46
34. Dingeman RS, Mitchell EA, Meyer EC, et al. Parent presence during complex invasive procedures and cardiopulmonary resuscitation: A systematic review of the literature. Pediatrics
35. Hollyday SL, Buonocore D. Breaking bad news and discussing goals of care in the intensive care unit. AACN Adv Crit Care 2015; 26:131–141
36. Hsiao JL, Evan EE, Zeltzer LK. Parent and child perspectives on physician communication in pediatric palliative care. Palliat Support Care 2007; 5:355–365
37. Larson CP, Dryden-Palmer KD, Gibbons C, et al. Moral distress in PICU and neonatal ICU practitioners: A cross-sectional evaluation. Pediatr Crit Care Med 2017; 18:e318–e326