Secondary Logo

Share this article on:

Long-Term Functional Outcome Data Should Not in General Be Used to Guide End-of-Life Decision-Making in the ICU

Wilson, Michael E., MD1–3; Hopkins, Ramona O., PhD4–6; Brown, Samuel M., MD, MS4,5,7,8

doi: 10.1097/CCM.0000000000003443
Viewpoints

1Division of Pulmonary and Critical Care Medicine, Department of Medicine, Mayo Clinic, Rochester, MN.

2Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery, Mayo Clinic, Rochester, MN.

3Biomedical Ethics Program, Mayo Clinic, Rochester, MN.

4Center for Humanizing Critical Care, Intermountain Healthcare, Murray, UT.

5Pulmonary and Critical Care Medicine, Department of Medicine, Intermountain Medical Center, Murray, UT.

6Department of Psychology and Neuroscience, Brigham Young University, Provo, UT.

7Pulmonary and Critical Care Medicine, Department of Medicine, University of Utah School of Medicine, Salt Lake City, UT.

8Division of Medical Ethics and Humanities, Department of Medicine, University of Utah School of Medicine, Salt Lake City, UT.

Dr. Wilson and Brown wrote the first draft. Dr. Hopkins revised for critical content. All authors approved the final article.

Dr. Hopkins’ institution received funding from a grant from Intermountain Research and Medical Foundation to study ICU outcomes. The remaining authors have disclosed that they do not have any potential conflicts of interest.

For information regarding this article, E-mail: samuel.brown@imail.org

Postintensive care syndrome (PICS) represents a new or worsened health status experienced by survivors of critical illness (1). PICS may include impairments in physical, cognitive, and mental health (2–4). Although knowledge of long-term outcomes is useful in developing strategies to diagnose, treat, and prevent PICS, clinicians may also use PICS as grounds for withdrawal or withholding of life support therapies in the ICU. Our anecdotal observations of such practices are consistent with a scenario-based, randomized trial: intensivists primed to consider long-term functional outcomes had a higher probability of proposing life support limitations (5).

Using PICS to justify decisions to limit life support may be ethically inappropriate in many circumstances. Challenges such as the inability to accurately predict functional outcomes, errors of affective forecasting, the disability paradox (including failure to consider adaptation to future disability), clinicians’ pseudo-empathy, and errors in understanding patients’ values and priorities may jeopardize the usefulness of PICS data in decision-making. Where it has been studied, the large majority of patients who survive critical illness, even with PICS, adapt to new disability and do not wish they had been allowed to die (6). Given these challenges and realities, we argue that long-term functional outcome data should, in general, not be used to advocate for limiting life support in the ICU. The risks of making irreversible, potentially inauthentic decisions to limit life support in patients who would have otherwise survived, adapted, and even thrived are simply too high.

We acknowledge that many patients and surrogates express interest in information about future disability (7) and several consensus frameworks recommend its consideration (8). Although discussions with patients and families about PICS are important in planning recovery from critical illness, we draw attention to the substantial risks to the integrity of the decision-making process associated with integration of PICS into deliberations to limit life support. When clinicians focus on PICS, the risk of a therapeutic nihilism that is not authentic to a given patient is substantial. The risks fall into several categories (Table 1), which we explore in turn. Alongside the theoretical considerations, we also bear in mind empirical observations, such as the fact that in one important study of a decision support tool for long-term ventilation information regarding expected short-term and long-term outcomes within the decision support tool did not increase the number of decisions to limit life support (9). In our experience, patients and families tend to limit life support when patients are already near the end of their lives and they perceive the proposed treatments to be overly burdensome.

TABLE 1

TABLE 1

Back to Top | Article Outline

PROGNOSTIC UNCERTAINTY

The application of outcomes data to treatment decisions depends exquisitely on the accuracy of individualized prognostication. For mortality, ICU clinicians’ predictions are often inaccurate and risk becoming self-fulfilling prophecies (10). In one study, about half of all patients predicted to die in the hospital by the healthcare team survived to discharge (11). Length of patient/clinician relationship, level of consciousness, and enthusiasm for novel treatments may all affect prognostic accuracy (12).

Physician prognostication of functional outcomes appears to be even worse than prognostication of mortality. Among approximately 300 patients in Pennsylvania, clinician predictions poorly discriminated functional outcomes and discriminated worse than predictions of mortality (13). In a single-center study, over half of ICU attending physicians recalled being surprised to discover that patients in whom they had recommended limiting life support on the basis of anticipated poor outcomes had in fact survived, sometimes with no functional impairment (14).

Predictive models for functional outcomes are also inadequate. Acute severity of illness is not associated with functional outcomes among survivors (15). Baseline impairments dominate subsequent outcomes but are as yet inadequately characterized in clinical practice and research (16). In addition, methodological problems plague prognostic models. First, most predictive models are validated in terms of discrimination (how often, on average, nonsurvivors have higher predicted mortality than survivors), when patients and surrogates will be more interested in calibration (what proportion of patients with a predicted outcome experience that outcome). Even if discrimination were actually of interest, even highly discriminative models are notoriously uninformative at the individual level (17). Second, prognostic models perform worst at extremes (e.g., the highest and lowest deciles of risk), but the decisional thresholds for patients and families likely cluster at those extremes. Importantly, clinicians may not understand or acknowledge the crucial differences between their own and patients’ decisional thresholds.

Back to Top | Article Outline

THE PROBLEMS OF AFFECTIVE FORECASTING

Functional impairments cannot be interpreted without context. The meaning of a future functional impairment requires predicting future emotional response to a given state (“affective forecasting”) (18). In the ICU, patients and surrogates are asked to predict their emotional response to future survival with new disability. Unfortunately, affective forecasting is highly inaccurate and made worse when surrogates are forecasting on behalf of a patient (18 , 19). Errors of affective forecasting include the disability paradox, underestimating adaptation, immune neglect, and the focusing illusion. The disability paradox is the well-established observation that healthy people rate the quality of life of people with disabilities lower than do the people with those disabilities (20). New disability often produces negative emotions, but the negative emotions are briefer and less intense than predicted (21).

Furthermore, people generally adapt to new disabilities (22). Physical adaptation includes new ways of accomplishing the same task such as using a walker to achieve mobility. Psychologic adaptation may involve developing new sources of happiness and meaning (22). People routinely fail to predict or even consider adaptation to future disability during affective forecasting (23). Patients, surrogates, and clinicians are all prone to disregard adaptation. In the few ICU studies where adaptation has been measured, most patients who survive do not regret surviving—even with new disability. In one study of 67 sepsis survivors (39% overall response rate), every participant reported they would be willing to be treated in the ICU all over again (6). ICU survivors will likely adapt better in a supportive environment, providing some rationale for increasing support to survivors (24 , 25).

Similarly, patients may be unaware (“immune neglect”) that they will likely employ coping mechanisms to manage discomfort—employing what psychologists call the psychologic immune system (26). These mechanisms include adaptive—for example, meditation, humor, and compartmentalization—and maladaptive—for example, avoidance and substance abuse—coping mechanisms. Adaptive coping dramatically improves emotional well-being in the face of new disability (27). Notably for the ethical questions here, individuals aware of anticipated coping strategies are better at affective forecasting (28).

The “focusing illusion” occurs when patients focus exclusively on the negative aspects of a potential disability. For example, a person anticipating colostomy may focus on the appliance and its effect on wearing a bathing suit, but not realize that other life domains will be minimally affected (22). These biases are ubiquitous threats to the integrity of decisions to limit life support based on PICS.

Back to Top | Article Outline

STRESS MAY AMPLIFY BIASES IN DECISION-MAKING

ICU patients, surrogates, and clinicians experience a myriad of physical, cognitive, psychologic, financial, social, and moral stressors, which may greatly amplify decision-making biases (29). People experiencing stress have decreased attention to the decision-making task, increased reliance on the judgment of others, decreased executive functioning, are more likely to favor higher risk short-term rewards and are more likely to reach premature diagnostic and prognostic closure (30 , 31). People in an emotionally aroused or “hot” state often inaccurately predict their preferences when they return to an emotionally unaroused or “cold” state, and vice versa (32).

Back to Top | Article Outline

PITFALLS OF ASSESSING PATIENT VALUES AND PRIORITIES

Anchoring decisions to limit life support to future health states may also interfere with identification of patients’ values and priorities. ICU clinicians are at risk for pseudo-empathy, an inaccurate belief that patients see the world the way the clinicians do. In a distressing situation, clinicians experience their own strong emotions about the potential inappropriateness of treatment, which may induce pseudo-empathy (33).

Clinicians may also fail to recognize that patients’ values and priorities change. Older adults’ perceptions of states worse than death (the relevant health state for most decisions to limit life support) have at best moderate stability when measured months apart (34). Among 100,000 nursing home residents, approximately 40% had code status order changes over a period of 5 years; reversals of do-not-resuscitate (DNR) were common after hospital readmission (35). Anecdotally, the confrontation with critical illness may authentically cause a patient to scrutinize and ultimately reject a prior DNR/do-not-intubate (DNI) order based on hypothetical preferences. Over half of DNR/DNI orders in hospitalized patients do not even reflect accurate patient preferences (36).

Back to Top | Article Outline

IMPLICATIONS AND CONCLUSIONS

PICS data has a clear role in helping patients, surrogates, clinicians, and researchers prevent, diagnose, and treat PICS. But for patients and surrogates who are amid critical illness and making decisions about whether or not to limit life support, communication of data about long-term outcomes is fraught with challenges and potentially adverse consequences. Given known failures of affective forecasting, poor prognostic accuracy for functional outcomes, and the distorting role of disability stigma, we do not feel that in general data on long-term functional outcomes should be used to guide decisions to limit life support.

We believe that the primary goal of ICU decision-making is to make decisions that are true to the patient as a person, what we term an “authentic” decision (37). An authentic decision requires knowledge of a patient’s values and priorities; accurate and relevant information about expected burdens and outcomes of therapy; and an accurate mapping of values and priorities onto specific medical treatments (38). This is a tall order; it’s little wonder that it’s so often fraught for all participants.

When making decisions to determine whether life support should be considered (or avoided), we recommend emphasizing current realities rather than anticipated futures. For example, rather than identifying which long-term future health states seem unacceptable (fraught with significant bias which we will subsequently discuss), decisions to limit life support would be based on identifying patients who are currently in premorbid conditions from which they would not want a burdensome medical rescue or patients who are currently in the final phases of their life.

We do not advocate a return to strong parentalism or hiding truth. We believe that discussions about PICS are important to the recovery trajectory and should be held with most patients. However, if patients or surrogates wish to use knowledge of PICS to make decisions to limit life support, ethically such data should be presented in context of discussions about adaptation to new disability, the disability paradox, failures of affective forecasting, and the profound prognostic uncertainty. In our clinical experience, such safeguards are employed rarely if ever. We also advocate research to better address these limitations. In addition, clinicians must understand their risks of pseudo-empathy and discrepant decisional thresholds. Methods such as guided metacognition (i.e., helping clinicians reflect on their thought processes) may limit such clinician errors (39).

The observations from research into long-term functional outcomes that may be most relevant to life support decision making are the observation that baseline quality of life is highly predictive of longer-term outcomes. Although various cognitive biases may also affect decisions about further treatment based on baseline quality of life, we feel that such decisions are likely to be associated with greater familiarity and authenticity. There—with careful attention to the biases we have described—it may be quite appropriate to consider with patients and surrogates the role of their current phase of life and phase of illness as they pursue an authentic path through the thickets of critical illness.

Back to Top | Article Outline

REFERENCES

1. Needham DM, Davidson J, Cohen H, et al. Improving long-term outcomes after discharge from intensive care unit: Report from a stakeholders’ conference. Crit Care Med 2012; 40:502–509
2. Hopkins RO, Weaver LK, Collingridge D, et al. Two-year cognitive, emotional, and quality-of-life outcomes in acute respiratory distress syndrome. Am J Respir Crit Care Med 2005; 171:340–347
3. Needham DM, Dinglas VD, Morris PE, et al; NIH NHLBI ARDS Network: Physical and cognitive performance of patients with acute lung injury 1 year after initial trophic versus full enteral feeding. EDEN trial follow-up. Am J Respir Crit Care Med 2013; 188:567–576
4. Pandharipande PP, Girard TD, Ely EW. Long-term cognitive impairment after critical illness. N Engl J Med 2014; 370:185–186
5. Turnbull AE, Krall JR, Ruhl AP, et al. A scenario-based, randomized trial of patient values and functional prognosis on intensivist intent to discuss withdrawing life support. Crit Care Med 2014; 42:1455–1462
6. Cuthbertson BH, Elders A, Hall S, et al; Scottish Critical Care Trials Group; Scottish Intensive Care Society Audit Group: Mortality and quality of life in the five years after severe sepsis. Crit Care 2013; 17:R70
7. Fried TR, Bradley EH, Towle VR, et al. Understanding the treatment preferences of seriously ill patients. N Engl J Med 2002; 346:1061–1066
8. Bernacki RE, Block SD; American College of Physicians High Value Care Task Force: Communication about serious illness care goals: A review and synthesis of best practices. JAMA Intern Med 2014; 174:1994–2003
9. Cox CE, Lewis CL, Hanson LC, et al. Development and pilot testing of a decision aid for surrogates of patients with prolonged mechanical ventilation. Crit Care Med 2012; 40:2327–2334
10. Sinuff T, Adhikari NK, Cook DJ, et al. Mortality predictions in the intensive care unit: Comparing physicians with scoring systems. Crit Care Med 2006; 34:878–885
11. Meadow W, Pohlman A, Frain L, et al. Power and limitations of daily prognostications of death in the medical intensive care unit. Crit Care Med 2011; 39:474–479
12. Christakis NA, Lamont EB. Extent and determinants of error in doctors’ prognoses in terminally ill patients: Prospective cohort study. BMJ (Clinical research ed) 2000; 320:469–472
13. Detsky ME, Harhay MO, Bayard DF, et al. Discriminative accuracy of physician and nurse predictions for survival and functional outcomes 6 months after an ICU admission. JAMA 2017; 317:2187–2195
14. Wilson ME, Rhudy LM, Ballinger BA, et al. Factors that contribute to physician variability in decisions to limit life support in the ICU: A qualitative study. Intensive Care Med 2013; 39:1009–1018
15. Brown SM, Wilson E, Presson AP, et al; With the National Institutes of Health NHLBI ARDS Network: Predictors of 6-month health utility outcomes in survivors of acute respiratory distress syndrome. Thorax 2017; 72:311–317
16. Ferrante LE, Pisani MA, Murphy TE, et al. Functional trajectories among older persons before and after critical illness. JAMA Intern Med 2015; 175:523–529
17. Ware JH. The limitations of risk factors as prognostic tools. N Engl J Med 2006; 355:2615–2617
18. Halpern J, Arnold RM. Affective forecasting: An unrecognized challenge in making serious health decisions. J Gen Intern Med 2008; 23:1708–1712
19. Gilbert DT. Stumbling on Happiness. 2006New York, NY, A.A. Knopf.
20. Albrecht GL, Devlieger PJ. The disability paradox: High quality of life against all odds. Soc Sci Med 1999; 48:977–988
21. Gilbert DT, Pinel EC, Wilson TD, et al. Immune neglect: A source of durability bias in affective forecasting. J Pers Soc Psychol 1998; 75:617–638
22. Ubel PA, Loewenstein G, Schwarz N, et al. Misimagining the unimaginable: The disability paradox and health care decision making. Health Psychol 2005; 24:S57–S62
23. Ubel PA, Loewenstein G, Jepson C. Disability and sunshine: Can hedonic predictions be improved by drawing attention to focusing illusions or emotional adaptation? J Exp Psychol Appl 2005; 11:111–123
24. Modrykamien AM. The ICU follow-up clinic: A new paradigm for intensivists. Respir Care 2012; 57:764–772
25. Mikkelsen ME, Jackson JC, Hopkins RO, et al. Peer support as a novel strategy to mitigate post-intensive care syndrome. AACN Adv Crit Care 2016; 27:221–229
26. Hoerger M. Coping strategies and immune neglect in affective forecasting: Direct evidence and key moderators. Judgm Decis Mak 2012; 7:86–96
27. Halpern J, Arnold RM. Affective forecasting: An unrecognized challenge in making serious health decisions. J Gen Intern Med 2008; 23:1708–1712
28. Dunn EW, Brackett MA, Ashton-James C, et al. On emotionally intelligent time travel: Individual differences in affective forecasting ability. Pers Soc Psychol Bull 2007; 33:85–93
29. Embriaco N, Papazian L, Kentish-Barnes N, et al. Burnout syndrome among critical care healthcare workers. Curr Opin Crit Care 2007; 13:482–488
30. Starcke K, Brand M. Decision making under stress: A selective review. Neurosci Biobehav Rev 2012; 36:1228–1248
31. Croskerry P. A universal model of diagnostic reasoning. Acad Med 2009; 84:1022–1028
32. Loewenstein G. Hot-cold empathy gaps and medical decision making. Health Psychol 2005; 24:S49–S56
33. Halpern J. From Detached Concern to Empathy: Humanizing Medical Practice 2001New York, NY, Oxford University Press.
34. Lockhart LK, Ditto PH, Danks JH, et al. The stability of older adults’ judgments of fates better and worse than death. Death Stud 2001; 25:299–317
35. Mukamel DB, Ladd H, Temkin-Greener H. Stability of cardiopulmonary resuscitation and do-not-resuscitate orders among long-term nursing home residents. Med Care 2013; 51:666–672
36. Jesus JE, Allen MB, Michael GE, et al. Preferences for resuscitation and intubation among patients with do-not-resuscitate/do-not-intubate orders. Mayo Clin Proc 2013; 88:658–665
37. Taylor C. The Ethics of Authenticity. 1991Cambridge, MA, Harvard University Press.
38. Turnbull AE, Hartog CS. Goal-concordant care in the ICU: A conceptual framework for future research. Intensive Care Med 2017; 43:1847–1849
39. Croskerry P. The cognitive imperative: Thinking about how we think. Acad Emerg Med 2000; 7:1223–1231
Keywords:

affective forecasting; decision-making; intensive care unit; postintensive care syndrome; treatment refusal; withholding life support

Copyright © by 2019 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.