Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

More than one in three proxies do not know their loved one’s current code status: An observational study in a Maryland ICU

  • Alison E. Turnbull ,

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Visualization, Writing – original draft

    turnbull@jhmi.edu

    Affiliations Outcomes After Critical Illness and Surgery (OACIS) Group, Johns Hopkins University, Baltimore, Maryland, United States of America, Division of Pulmonary and Critical Care Medicine, School of Medicine, Johns Hopkins University, Baltimore, Maryland, United States of America, Department of Epidemiology, Bloomberg School of Public Health, Johns Hopkins University, Baltimore, Maryland, United States of America

  • Caroline M. Chessare,

    Roles Data curation, Writing – review & editing

    Affiliations Outcomes After Critical Illness and Surgery (OACIS) Group, Johns Hopkins University, Baltimore, Maryland, United States of America, Division of Pulmonary and Critical Care Medicine, School of Medicine, Johns Hopkins University, Baltimore, Maryland, United States of America

  • Rachel K. Coffin,

    Roles Data curation, Writing – review & editing

    Affiliation Medical Intensive Care Unit, Johns Hopkins Hospital, Baltimore, Maryland, United States of America

  • Dale M. Needham

    Roles Conceptualization, Funding acquisition, Methodology, Resources, Supervision, Writing – review & editing

    Affiliations Outcomes After Critical Illness and Surgery (OACIS) Group, Johns Hopkins University, Baltimore, Maryland, United States of America, Division of Pulmonary and Critical Care Medicine, School of Medicine, Johns Hopkins University, Baltimore, Maryland, United States of America, Department of Physical Medicine and Rehabilitation, School of Medicine, Johns Hopkins University, Baltimore, Maryland, United States of America

More than one in three proxies do not know their loved one’s current code status: An observational study in a Maryland ICU

  • Alison E. Turnbull, 
  • Caroline M. Chessare, 
  • Rachel K. Coffin, 
  • Dale M. Needham
PLOS
x

Abstract

Rationale

The majority of ICU patients lack decision-making capacity at some point during their ICU stay. However the extent to which proxy decision-makers are engaged in decisions about their patient’s care is challenging to quantify.

Objectives

To assess 1)whether proxies know their patient’s actual code status as recorded in the electronic medical record (EMR), and 2)whether code status orders reflect ICU patient preferences as reported by proxy decision-makers.

Methods

We enrolled proxy decision-makers for 96 days starting January 4, 2016. Proxies were asked about the patient’s goals of care, preferred code status, and actual code status. Responses were compared to code status orders in the EMR at the time of interview. Characteristics of patients and proxies who correctly vs incorrectly identified actual code status were compared, as were characteristics of proxies who reported a preferred code status that did vs did not match actual code status.

Measurements and main results

Among 111 proxies, 42 (38%) were incorrect or unsure about the patient’s actual code status and those who were correct vs. incorrect or unsure were similar in age, race, and years of education (P>0.20 for all comparisons). Twenty-nine percent reported a preferred code status that did not match the patient’s code status in the EMR. Matching preferred and actual code status was not associated with a patient’s age, gender, income, admission diagnosis, or subsequent in-hospital mortality or with proxy age, gender, race, education level, or relation to the patient (P>0.20 for all comparisons).

Conclusions

More than 1 in 3 proxies is incorrect or unsure about their patient’s actual code status and more than 1 in 4 proxies reported that a preferred code status that did not match orders in the EMR. Proxy age, race, gender and education level were not associated with correctly identifying code status or code status concordance.

Introduction

In the United States, the number of intensive care unit (ICU) beds and annual cost of critical care services are steadily increasing.[1] Simultaneously, the use of mechanical ventilation during the last month of life[2] and the proportion of hospitalized patients with advanced dementia who are mechanically ventilated[3] have risen markedly. Ethicists, health-services researchers, and clinicians have worried these simultaneous trends put patients at risk of over-treatment[4] as a result of both cognitive biases (i.e. the cascade effect [5]) and system-level phenomena (e.g. clinical momentum [6]).

In response to these concerns, critical care professional societies endorse shared decision making with the goal of choosing treatments that are “medically appropriate and consistent with the patient’s values, goals, and preferences”[7]. Most patients lack decision-making capacity at some point during their ICU stay, [8,9] and only a third of American adults creates a legal document describing their preferences[10]. Therefore ICU clinicians rely heavily on proxy decision-makers to report the goals of their critically ill loved ones and to participate in deliberation about preference-sensitive treatments. However, it is unclear how well these ICU proxies understand fundamental decisions about their loved one’s care, including how the medical team should respond in the event of acute respiratory failure or cardiac arrest.

To help address this issue, we conducted in-person, structured interviews with all available healthcare proxies present in our adult medical intensive care unit (MICU). Our objectives were two-fold: 1) to assess whether proxies know their patient’s actual code status as recorded in the electronic medical record (EMR), and 2) to determine whether code status orders reflect ICU patient preferences as reported by proxy decision-makers.

Methods

Setting

This study was conducted in the MICU at Johns Hopkins Hospital (JHH) in Baltimore, Maryland. JHH is an university-affiliated, teaching hospital with 1,177 beds in an urban environment. The MICU has 24 beds, admitted 1,550 patients in 2017, and specializes in the care of high acuity patients with complex medical diagnoses. Participants were recruited as part of a Phase I study[11] of a brief activation intervention to prepare family members of ICU patients to be effective proxy decision-makers. All interviews with proxies were conducted in-person, and outside of the patient’s room. Participants provided oral consent and received a $10 gift card at the end of the interview. Johns Hopkins Medicine Institutional Review Board Committee X approved this study number: IRB00080137. The IRB approved oral consent because participation did not involve procedures normally requiring written consent outside of the research context.

Recruitment

Between January, 2016 and May, 2016, we screened the MICU census 7 days per week to identify eligible families for this study. Families [12] were eligible once their patient had been in the MICU for >24 hours and the family remained eligible for 7 consecutive days thereafter or until patient death or ICU discharge. Family members were excluded if they were non-English speaking, <18 years old, or never physically visited the MICU. We preferentially enrolled a patient’s healthcare agent(s) or surrogate(s), as defined by Maryland State Law, and refer to both agents and surrogates using the term “proxy” as recommended [13]. In Maryland, healthcare agents are individuals appointed in advance directives and their authority is defined by the individual creating the advance directive. If a patient has not named an agent or the agent is not available and the patient lacks decisional capacity, other individuals are approached to act as surrogates in a pre-specified hierarchical order dictated by the state.[14] If a legal proxy was not able to visit the MICU, an adult family member who was physically present in the MICU and interacting with the clinical team as a patient representative was approached for this study. If the initial family member enrolled was not the patient’s legal healthcare proxy, we continued to attempt to recruit the patient’s legal proxy, and included both interviewed family members in analyses.[15]

Study procedures

Participating proxies provided their demographic information and then, as part of the parent study, were read aloud a graphically illustrated booklet explaining the role of a healthcare proxy in the ICU. The booklet was written at a Flesch-Kincard reading grade level of 3.8, required 3.5 minutes to read aloud, and can be accessed via the study website at www.piperscience.org/proxy-activation. The booklet did not mention code status orders or explain CPR. After reading aloud, study investigators gave each participant a copy of the booklet and asked 1 open-ended question and 1 multiple-choice questions about their patient’s goal of care. Response options for the multiple-choice question were derived from previous research categorizing the goals of hospitalized patients [1618]. Next, we asked 2 multiple-choice questions about their patient’s current use of life support and preferred code status avoiding medical terminology. (S1 Fig) After enrolling 11 proxies, an additional question was added to the interview to determine if proxies understood their patient’s actual code status at the time of the interview. Additional data were collected from the EMR regarding patient age, sex, race, zip code, location prior to hospitalization, ICU admission diagnosis, use of life support, and actual code status at the time of the proxy interview, as well as subsequent hospital vital status. Patients without any code status order in the EMR at the time of the interview were recorded and analyzed as being full code because these patients will be resuscitated if indicated.

Analyses

Tables were used to display 1) proxy’s report of the patient’s current code status versus actual code status in the EMR at time of interview, 2) proxy’s report of the patient’s preferred code status vs. actual code status, with a “concordant code status” defined as agreement between preferred and actual status, and 3) proxy report of the patient’s preferred code status versus reported patient goal.

Characteristic of proxies who incorrectly identified their patient’s code status or reported that they did not know their patient’s code status were compared to proxies who correctly identified code status using the Wilcoxon-Mann-Whitney two-sample test, the Chi-square test, and the Fisher’s exact test (if cell counts <10), as appropriate. Statistical significance was defined as a two-sided P-value <0.05. Proxies and patients with concordant versus discordant code status (as defined above) were compared using the same statistical tests. In addition to reporting the P-value for these comparisons, we report the absolute effect size (Cohen’s d) as a measure of the magnitude of the difference to help with appropriate interpretation of the comparisons.[19,20] We conducted the following sensitivity analyses: 1) comparing concordant versus discordant proxies excluding proxies who were unsure of the patient’s code status, 2) comparing concordant versus discordant proxies excluding interviewed family members who were not legal healthcare proxies, and 3) comparing proxies who were incorrect or unsure of their patient’s actual code status versus proxies who correctly identified code status excluding interviewed family members who were not legal healthcare proxies. Statistical analyses were performed using the R programming language version 3.3.2 (Vienna, Austria).

Results

There were 122 unique family members interviewed for 111 consecutive eligible patients (Fig 1). Among these participants 79 (65%) were the patient’s legal healthcare proxy, median age was 51 years old (range: 18–79), 39 (30%) were male, 55 (45%) identified as Black or African-American, and median years of education was 14 (range: 10–24) (Table 1). Five of the 111 patients (4%) had a legal document naming a healthcare agent in their medical chart.[15] Analyses regarding the external validity of this study sample have been previously reported.[15]

thumbnail
Fig 1. Patient and proxy screening, study eligibility and enrollment.

Patients admitted to the medical ICU were eligible if they remained in the ICU for ≥24 hours. If the first proxy enrolled was not the patient's legal surrogate decision-maker under Maryland state law, then we attempted to also enroll the legal surrogate decisions-maker. a 42 patients were not visited by any friends or family members during their ICU stay b 11 patients had 2 proxies enrolled.

https://doi.org/10.1371/journal.pone.0211531.g001

At the time of the 122 interviews with ICU proxies, 103 (84%) patients were full code. Among the 111 proxies asked to report their patient’s actual code status, 69 proxies (62%, 95% CI 52% -71%) were correct, 32 (29%, 95% CI 21% - 38%) were unsure, and 10 (9%, 95% CI 5% - 16%) were incorrect (Table 2). Compared to the 69 proxies who correctly understood their patient’s code status, the 42 (38%) proxies who were incorrect or unsure had similar characteristics (P>0.20, effect sizes (d) ≤0.35) except for in admission diagnosis (P = 0.05, d = 0.98) (S1 Table). Excluding interviewed family members who were not legal healthcare proxies did not qualitatively changed these findings (S3 Table).

There were 54 of 122 (44%, 95% CI 35% - 54%) proxies who reported a preferred code status that was concordant with actual code status, while 31 proxies (25%, 95% CI 18% - 34%) didn’t know what their patient would prefer (Table 3). Among the 35 (27%, 95% CI 21% - 38%) proxies reporting a preferred code status discordant with the patient’s actual code status, 32 (91%, 95% CI 76% - 98%) reported the patient would prefer less use of life support. Patients reporting a discordance in preferred versus actual code status were similar in age, race, years of education, and ICU day at the time of interview (all P>0.20; all Cohen’s d effect sizes <0.40) (S2 Table). Neither limiting analyses to proxies who provided a response other than “I don’t know”, nor excluding family members who were not legal healthcare proxies substantially changed these findings (S4 and S5 Tables).

thumbnail
Table 3. Concordancea between preferred code status and actual code status.

https://doi.org/10.1371/journal.pone.0211531.t003

There were 111 proxies who answered survey questions about both their loved ones preferred code status and current code status. Within the 69 proxies who correctly identified their patient’s current code status, 15 (22%) reported a discordant preferred code status. In contrast, among the 42 proxies who could not identify their patient’s code status, 18 (43%) reported a discordant preferred code status.

When asked to identify their patient’s primary goal of care from a list of goals derived from previous research categorizing goals of hospitalized patients, 106 proxies (88%) chose one of the following options: be cured, live longer, improve health, maintain health, accomplish a personal life goal (Table 1). There were 13 proxies (11%) who selected “to be comfortable”, 2 proxies (2%) were unsure of their patient’s goal, and 1 (1%) refused the question. Among the 70 proxies (57%) who reported that their patient was not on life support at the time of interview, 13 (19%) were mechanically ventilated, 3 (4%) were receiving continuous renal replacement therapy, and 4 (6%) were receiving both. Among the 106 proxies who did not endorse being comfortable as the patient’s primary goal, 24 (23%) believed the patient would not want any life support machines or CPR, 8 (8%) did not want CPR, and 27 (25%) were unsure of the patient’s wishes (Table 4).

thumbnail
Table 4. Proxy-reported preferred code status in patients for whom comfort was not the prioritized goal of carea.

https://doi.org/10.1371/journal.pone.0211531.t004

Discussion

In this study of 122 proxy decision-makers in a single MICU, 38% of proxies could not identify their patient’s code status when described using lay terminology, and less than half believed that the patient would want the actual code status recorded in the EMR. Among proxies reporting a discordant code status, 91% reported that the patient would prefer to forego life support permitted by the code status recorded in the EMR. Among ICU proxies, 98% reported that they knew their patient’s prioritized goal of care, but only 75% provided a response other than “I don’t know” when asked about the patient’s preferred code status. Proxies who were incorrect or unsure (vs were correct) about the patient’s actual code status were not substantially different based on readily identifiable characteristics of the proxy or patient. Patients and proxies characteristics also were not associated with concordance between preferred and actual code status.

Code status orders dictate how clinicians respond when a patient develops acute respiratory failure or experiences a cardiac arrest. By default, hospitalized patients are “full code”, meaning that they will be intubated and resuscitated, unless there is a code status order specifically instructing clinicians not to perform these interventions. Guidelines issued by ethical and professional associations recommend that code status orders be discussed when a provider anticipates that a patient may experience cardiac arrest.[21,22] Previous research suggests significant heterogeneity among providers in whether they choose to discuss code status at admission, [2326] many proxies don’t understand what CPR entails, and nearly one third do not know how the medical team will respond to a patient with a DNR order.[27]

We are not aware of previous publications evaluating whether proxies of adult ICU patients know their code status. Proxy reports of patient’s prioritized goals of care in this study were similar to a previous study conducted in an academic MICU in Iowa, USA despite substantial differences in the demographics of these two patient populations.[18] In a recent study conducted in 12 acute care hospitals in Canada comparing preferred vs. actual code status, 30% of elderly patients and 31% of their families reported a code status that matched the actual code status in the EMR.[28] Our results were very similar this Canadian study (29% vs 31%) despite patients in our study being much younger, >50% non-white, and having much higher in-hospital mortality. In contrast, the frequency of discordant code status in our study was higher than the prior study in Iowa (29% vs 16%).[29]

Racial and socio-economic differences in end-of-life care in the ICU are well documented,[3032] but neither race nor indicators of socio-economic status were significantly associated with code status discordance in this study. We also found that proxies with post-graduate education were no more likely to understand their patient’s code status than proxies with less formal education (62% vs 62%, P = 1.0). Hence, discussions of current patient status, prognosis, and treatment options using simple, direct language are appropriate regardless of proxy demographics. [33,34]

Proxies of critically ill patients are frequently, and quite appropriately, sad or anxious. It is sometimes assumed that these strong emotions impede their judgement or ability to understand basic information, and make them inappropriately optimistic. This is supported by a study which found that when asked about a loved one’s prognosis, 43% of ICU proxies were more optimistic than their intensivist while only 10% were more pessimistic than their intensivist.[35] But emotion seems an unlikely explanation for why among the families of full-code patients, 26% reported their loved one would prefer less use of life support, 27% said they were unsure what kind of care their loved one would choose, and 35% were unsure or assumed CPR would not be performed. These responses do not fit the prevailing image of the emotional proxy who insists their family member would want all possible interventions to extend their life.

Origins of discordance between EMR orders and the proxy-reported preferred code status

The origins of discordance between preferred vs actual code status in this cohort are unknown. However there are several possible explanations. First, clinicians may have been unaware that some patients’ preferred code status had changed since MICU admission. Second, it is possible some patients had never been able to communicate, and their proxy had not been asked about preferred code status. This is consistent with a recent analysis of recorded meetings between intensivists and the families of critically ill patients that found patients’ prior treatment preferences were discussed in only 19% of meetings.[36] Third, presenting code status as a multiple choice question, in our study, may have normalized treatment limitations and contributed to the frequency of discordant responses. The framing of treatment options and the environmental conditions under which choices are made can have dramatic effects on patient and families choices.[37,38] Proxies may have changed their minds about what the patient would prefer after hearing all the code status options in the context of a multiple-choice question without a default response.[39]

Fourth, in some cases proxies may have been incorrect about the patient’s preference, and the actual code status in the EMR reflected a prior discussion between the patient and clinician. We did not interview patients because this study was embedded within a larger study specifically interested in medical proxies.[40] However proxies for ICU patients are very frequently asked to make decisions [9] since most ICU patients lack decisional capacity at some point during their ICU stay, and the vast majority of end-of-life decisions are made by proxies.[8,41,42] Therefore, the patient’s current code status and preferences previously expressed to ICU clinicians should be disclosed to proxies unless the patient has specifically requested otherwise.

Finally, the single proxy who reported that their patient would prefer to be full code despite a DNR/DNI order in the EMR was demonstrating strict substituted judgment for their patient. Specifically, this proxy explained that their unconscious spouse would ask for “[illicit] drugs and cigarettes,” want to be full code, and try to prematurely leave the hospital against medical advice. However the spouse’s death was imminent, and the proxy felt certain that prioritizing comfort, and foregoing intubation and resuscitation from cardiac arrest, was appropriate.

Failure to elicit patient goals and treatment preferences is potentially harmful even when patients do not require life support interventions. In a randomized survey of U.S. internal medicine residents, respondents were significantly less likely to pursue dialysis, bronchoscopy, and surgical consultation for a patient with a DNR order because of assumptions about these patients’ goals of care.[43] Conversely, clinicians may assume that patients without a DNR order are willing to accept invasive treatments in pursuit of longevity and fail to recognize that each significant preference-sensitive decision requires shared decision-making–a challenge referred to as “decision recognition”.[44] Importantly, a modified Delphi consensus process identified common preference-sensitive decisions that should trigger clinicians to clarify patient goals and consider initiating shared decision-making in the context of critical illness. [45]

Eliciting patient goals vs code status

The U.S. National Academy of Medicine recommends that clinicians work “with patients and families to ensure that care provided matches closely with each individual’s goals.”[46] Interestingly, only 2% of proxies in this study reported they were unsure of their patient’s goal of care, but 26% were unsure whether their loved one would accept the use of life support machines and CPR to help achieve that goal. This difference illustrates proxy confidence about their patient’s values and how such confidence may not translate into decisions about specific treatments without clinician guidance.

Implications for clinical research and health services researchers

Research evaluating whether patients are receiving care consistent with their goals and preferences is challenging.[47,48] Patient goals are often personal, difficult to categorize, dynamic, [16,17,49] and often not documented in the electronic medical record (EMR) in a form that’s easily extracted or analyzed.[5052] Consequently, clinical researches and health services researchers frequently use code status, which is almost universally available in EMRs as a categorical variable, as an indicator of patient treatment preferences.[53,54] While code status is often the only information on patient treatment preferences systematically recorded and easily extractable from the EMR, our results suggest it is a weak and imperfect indicator of treatment preference. Therefore, prospective clinical studies that are able to interview patients and proxies should ask about goals and treatment limitations (i.e. preferred code status) directly using lay terminology.

Implications for clinicians

The majority of proxies who reported that the patients’ most important goal was not comfort, also reported the patient would not choose to be full code or were unsure of the patient’s preferred code status. Hence, when a proxy says their patient would want to “go home”, “get better”, or “live longer” clinicians should not assume the patient would choose to be full code or would consent to invasive procedures. Rather, avoid premature closure, ask additional questions to determine whether the patient would accept the use of life support and CPR in an attempt to achieve their goals, and then enquire whether the proxy would like to know your professional recommendation. This is consistent with current recommendations to avoid evoking unrealistic aspirations with the question: ‘What would your loved one want?’ and instead focus conversations with proxies on how the patient would feel about likely clinical options and outcomes within the bounds of clinical reality.[55] Patient goals of care and treatment limitations are distinct attributes requiring separate querying and deliberation. Finally, clinicians should not assume that proxies will tell clinicians when they believe the current code status is not aligned with their patient’s preferences. We identified 15 proxies who correctly identified their patient’s current code status, reported it was not what they believed the patient would want, but did not notify the ICU clinical team. This is consistent with a recent study of current and former ICU proxies in which less than 50% reported being comfortable speaking up when they believed their love one would prefer less aggressive, or more aggressive medical care than they were currently receiving.[56]

Implications for patients and family caregivers

Critically ill patients and their proxies can help ensure that life support interventions are used in ways that are consistent with patient preferences by actively engaging ICU clinicians. This means asking direct questions about whether patient goals are likely to be achievable, expected long-term patient outcomes, and whether physicians expect life support interventions to help achieve valued goals and outcomes. Patients and their proxies may need to initiate conversations about their goals and preferences multiple times, and should not assume that these conversations are conveyed across treatment teams or during physician hand-offs[57].

Limitations, strengths and next steps

This study was conducted in a single MICU. Both within and between ICUs, there is great variability in how clinicians discuss,[24,58] and interpret[5961] code status orders. As a next step, we encourage replicating this study using a multi-center design. We also could not determine the incidence or content of discussions that occurred between patients, proxies, and providers prior to enrollment because the majority of these discussions are not consistently documented, as common in many ICUs [5052]. The study did not enroll ICU patients and thus could not determine how frequently proxies correctly identified patient goals and preferences about the use of life-support. We also did not analyze responses to open-ended questions about the patient’s goals of care or interrogate how proxies interpreted the word “life support” in survey questions. Finally, we did not ask proxies how confident they were in their answers, although the response option “I don’t know” was used to discourage guessing. Study strengths include enrolling a socio-economically diverse study population and addressing questions that are central to both clinical practice and research.

Conclusions

In conclusion, code status orders were poor indicators of ICU patient goals of care and treatment preferences as reported by their proxy decision-makers. More than 1 in 3 healthcare proxies who were physically present in the ICU were either incorrect or unsure about their patient’s current code status. Readily identifiable characteristics, such as age, gender, race, and education level, were not associated with discordance in preferred vs actual code status, or with proxy misunderstanding or uncertainty about actual code status.

Supporting information

S1 Fig. Survey questions about patient goals and preferences.

https://doi.org/10.1371/journal.pone.0211531.s001

(PDF)

S1 Table. Proxy and patient characteristics by understanding of code status (n = 111).

https://doi.org/10.1371/journal.pone.0211531.s002

(PDF)

S2 Table. Proxy and patient characteristics by code status concordance (n = 122).

https://doi.org/10.1371/journal.pone.0211531.s003

(PDF)

S3 Table. Proxy and patient characteristics by code status concordance excluding proxies who were unsure of the patient’s preferred code status (n = 89).

https://doi.org/10.1371/journal.pone.0211531.s004

(PDF)

S4 Table. Proxy and patient characteristics by code status concordance for the subset of interviews with legal healthcare proxies (n = 79).

https://doi.org/10.1371/journal.pone.0211531.s005

(PDF)

S5 Table. Proxy and patient characteristics by understanding of code status for the subset of interviews with legal healthcare proxies (n = 72).

https://doi.org/10.1371/journal.pone.0211531.s006

(PDF)

References

  1. 1. Halpern NA, Pastores SM. Critical care medicine in the United States 2000–2005: an analysis of bed numbers, occupancy rates, payer mix, and costs. Crit Care Med. 2010;38: 65–71. pmid:19730257
  2. 2. Teno JM GP. Change in end-of-life care for medicare beneficiaries: Site of death, place of care, and health care transitions in 2000, 2005, and 2009. JAMA. 2013;309: 470–477. pmid:23385273
  3. 3. Teno JM, Gozalo P, Khandelwal N, Curtis JR, Meltzer D, Engelberg R, et al. Association of Increasing Use of Mechanical Ventilation Among Nursing Home Residents With Advanced Dementia and Intensive Care Unit Beds. JAMA Intern Med. 2016;176: 1809–1816. pmid:27723891
  4. 4. Winzelberg GS, Hanson LC. Intensive Care Unit Bed Availability and Use of Mechanical Ventilation in Nursing Home Residents With Advanced Dementia: When We Build It, Why Do They Come? JAMA Intern Med. 2016;176: 1816–1817. pmid:27723856
  5. 5. Bruce CR, Fetter JE, Blumenthal-Barby JS. Cascade Effects in Critical Care Medicine: A Call for Practice Changes. Am J Respir Crit Care Med. 2013;188: 1384–1385. pmid:24328766
  6. 6. Kruser JM, Cox CE, Schwarze ML. Clinical Momentum in the Intensive Care Unit: A Latent Contributor to Unwanted Care. Ann Am Thorac Soc N Y. 2017;14: 426–431. https://doi.org/10.1513/AnnalsATS.201611-931OI
  7. 7. Kon AA, Davidson JE, Morrison W, Danis M, White DB. Shared Decision Making in ICUs: An American College of Critical Care Medicine and American Thoracic Society Policy Statement. Crit Care Med. 2016;44: 188–201. pmid:26509317
  8. 8. Prendergast TJ, Claessens MT, Luce JM. A national survey of end-of-life care for critically ill patients. Am J Respir Crit Care Med. 1998;158: 1163–1167. pmid:9769276
  9. 9. Torke AM, Sachs GA, Helft PR, Montz K, Hui SL, Slaven JE, et al. Scope and outcomes of surrogate decision making among hospitalized older adults. JAMA Intern Med. 2014;174: 370–377. pmid:24445375
  10. 10. Yadav KN, Gabler NB, Cooney E, Kent S, Kim J, Herbst N, et al. Approximately One In Three US Adults Completes Any Type Of Advance Directive For End-Of-Life Care. Health Aff (Millwood). 2017;36: 1244–1251. pmid:28679811
  11. 11. FAQClinicalTrials.gov—Clinical Trial Phases [Internet]. [cited 26 Jan 2017]. Available: https://www.nlm.nih.gov/services/ctphases.html
  12. 12. Davidson JE, Aslakson RA, Long AC, Puntillo KA, Kross EK, Hart J, et al. Guidelines for Family-Centered Care in the Neonatal, Pediatric, and Adult ICU. Crit Care Med. 2017;45: 103–128. pmid:27984278
  13. 13. Schwartz AAGJS. Who Can Make Health Care Decisions For Another? Defining Health Care Proxies Under Maryland Law [Internet]. 2006. Available: http://www.marylandattorneygeneral.gov/Health%20Policy%20Documents/proxies_definition.pdf
  14. 14. DeMartino ES, Dudzinski DM, Doyle CK, Sperry BP, Gregory SE, Siegler M, et al. Who Decides When a Patient Can’t? Statutes on Alternate Decision Makers. N Engl J Med. 2017;376: 1478–1482. pmid:28402767
  15. 15. Turnbull AE, Hashem MD, Rabiee A, To A, Chessare CM, Needham DM. Evaluation of a strategy for enrolling the families of critically ill patients in research using limited human resources. PLOS ONE. 2017;12: e0177741. pmid:28542632
  16. 16. Kaldjian LC, Curtis AE, Shinkunas LA, Cannon KT. Goals of care toward the end of life: a structured literature review. Am J Hosp Palliat Care. 2008;25: 501–511. pmid:19106284
  17. 17. Haberle TH, Shinkunas LA, Erekson ZD, Kaldjian LC. Goals of care among hospitalized patients: a validation study. Am J Hosp Palliat Care. 2011;28: 335–341. pmid:21097876
  18. 18. Brandt DS, Shinkunas LA, Gehlbach TG, Kaldjian LC. Understanding Goals of Care Statements and Preferences among Patients and Their Surrogates in the Medical ICU. J Hosp Palliat Nurs. 2012;14: 126–132. pmid:22423214
  19. 19. Sullivan GM, Feinn R. Using Effect Size—or Why the P Value Is Not Enough. J Grad Med Educ. 2012;4: 279–282. pmid:23997866
  20. 20. Chavalarias D, Wallach J, Li A, Ioannidis JA. Evolution of reporting p values in the biomedical literature, 1990–2015. JAMA. 2016;315: 1141–1148. pmid:26978209
  21. 21. Guidelines for the Appropriate Use of Do-Not-Resuscitate Orders. JAMA. 1991;265: 1868–1871. pmid:2005737
  22. 22. Deciding to Forego Life-Sustaining Treatment [Internet]. Mar 1983 [cited 27 Jul 2017]. Available: https://repository.library.georgetown.edu/handle/10822/559344
  23. 23. Smith AK, Ries AP, Zhang B, Tulsky JA, Prigerson HG, Block SD. Resident Approaches to Advance Care Planning on the Day of Hospital Admission. Arch Intern Med. 2006;166: 1597–1602. pmid:16908792
  24. 24. Auerbach AD, Katz R, Pantilat SZ, Bernacki R, Schnipper J, Kaboli P, et al. Factors associated with discussion of care plans and code status at the time of hospital admission: results from the Multicenter Hospitalist Study. J Hosp Med Off Publ Soc Hosp Med. 2008;3: 437–445. pmid:19084893
  25. 25. Loertscher L, Reed DA, Bannon MP, Mueller PS. Cardiopulmonary resuscitation and do-not-resuscitate orders: a guide for clinicians. Am J Med. 2010;123: 4–9. pmid:20102982
  26. 26. Anderson WG, Chase R, Pantilat SZ, Tulsky JA, Auerbach AD. Code status discussions between attending hospitalist physicians and medical patients at hospital admission. J Gen Intern Med. 2011;26: 359–366. pmid:21104036
  27. 27. Wilson ME, Krupa A, Hinds RF, Litell JM, Swetz KM, Akhoundi A, et al. A Video to Improve Patient and Surrogate Understanding of Cardiopulmonary Resuscitation Choices in the ICU: A Randomized Controlled Trial*. Crit Care Med. 2015;43: 621–629. pmid:25479118
  28. 28. Heyland DK, Barwich D, Pichora D, Dodek P, Lamontagne F, You JJ, et al. Failure to Engage Hospitalized Elderly Patients and Their Families in Advance Care Planning. JAMA Intern Med. 2013; 1–10. pmid:23545563
  29. 29. Gehlbach TG, Shinkunas LA, Forman-Hoffman VL, Thomas KW, Schmidt GA, Kaldjian LC. Code status orders and goals of care in the medical ICU. Chest. 2011;139: 802–809. pmid:21292755
  30. 30. Johnson RW, Newby LK, Granger CB, Cook WA, Peterson ED, Echols M, et al. Differences in level of care at the end of life according to race. Am J Crit Care Off Publ Am Assoc Crit-Care Nurses. 2010;19: 335–343; quiz 344. pmid:20595215
  31. 31. Mack JW, Paulk ME, Viswanath K, Prigerson HG. Racial disparities in the outcomes of communication on medical care received near death. Arch Intern Med. 2010;170: 1533–1540. pmid:20876403
  32. 32. Muni S, Engelberg RA, Treece PD, Dotolo D, Curtis JR. The influence of race/ethnicity and socioeconomic status on end-of-life care in the ICU. Chest. 2011;139: 1025–1033. pmid:21292758
  33. 33. Bernacki RE, Block SD, for the 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; pmid:25330167
  34. 34. Warrillow S, Farley KJ, Jones D. Ten practical strategies for effective communication with relatives of ICU patients. Intensive Care Med. 2015;41: 2173–2176. pmid:25904186
  35. 35. White DB, Ernecoff N, Buddadhumaruk P, Hong S, Weissfeld L, Curtis JR, et al. Prevalence of and factors related to discordance about prognosis between physicians and surrogate decision makers of critically ill patients. JAMA. 2016;315: 2086–2094. pmid:27187301
  36. 36. Chiarchiaro J, Ernecoff NC, Scheunemann LP, Hough CL, Carson SS, Peterson MW, et al. Physicians Rarely Elicit Critically Ill Patients’ Previously Expressed Treatment Preferences in Intensive Care Units. Am J Respir Crit Care Med. 2017;196: 242–245. pmid:28707977
  37. 37. Barnato AE, Arnold RM. The Effect of Emotion and Physician Communication Behaviors on Surrogates’ Life-Sustaining Treatment Decisions: A Randomized Simulation Experiment. Crit Care Med. 2013; pmid:23660727
  38. 38. Patel D, Cohen ED, Barnato AE. The effect of framing on surrogate optimism bias: A simulation study. J Crit Care. 2016;32: 85–88. pmid:26796950
  39. 39. Anesi GL, Halpern SD. Choice architecture in code status discussions with terminally ill patients and their families. Intensive Care Med. 2016; 1–3. pmid:26951425
  40. 40. Turnbull AE, Chessare CM, Coffin RK, Needham DM. A brief intervention for preparing ICU families to be proxies: A phase I study. PLOS ONE. 2017;12: e0185483. pmid:28968409
  41. 41. Ely EW, Margolin R, Francis J, May L, Truman B, Dittus R, et al. Evaluation of delirium in critically ill patients: validation of the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU). Crit Care Med. 2001;29: 1370–1379. pmid:11445689
  42. 42. Cohen S, Sprung C, Sjokvist P, Lippert A, Ricou B, Baras M, et al. Communication of end-of-life decisions in European intensive care units. Intensive Care Med. 2005;31: 1215–1221. pmid:16041519
  43. 43. Stevenson EK, Mehter HM, Walkey AJ, Wiener RS. Association between Do Not Resuscitate/Do Not Intubate Status and Resident Physician Decision-making. A National Survey. Ann Am Thorac Soc. 2017;14: 536–542. pmid:28099054
  44. 44. Barnato AE. Challenges In Understanding And Respecting Patients’ Preferences. Health Aff (Millwood). 2017;36: 1252–1257. pmid:28679812
  45. 45. Turnbull AE, Sahetya SK, Needham DM. Aligning critical care interventions with patient goals: A modified Delphi study. Heart Lung J Crit Care. 2016;45: 517–524. pmid:27593494
  46. 46. Dzau VJ, McClellan MB, McGinnis JM, Burke SP, Coye MJ, Diaz A, et al. Vital Directions for Health and Health Care: Priorities From a National Academy of Medicine Initiative. JAMA. 2017;317: 1461–1470. pmid:28324029
  47. 47. Turnbull AE, Hartog CS. Goal-concordant care in the ICU: a conceptual framework for future research. Intensive Care Med. 2017; 1–3. pmid:28656453
  48. 48. Teno JM, Price RA, Makaroun LK. Challenges Of Measuring Quality Of Community-Based Programs For Seriously Ill Individuals And Their Families. Health Aff (Millwood). 2017;36: 1227–1233. pmid:28679809
  49. 49. Berntsen GKR, Gammon D, Steinsbekk A, Salamonsen A, Foss N, Ruland C, et al. How do we deal with multiple goals for care within an individual patient trajectory? A document content analysis of health service research papers on goals for care. BMJ Open. 2015;5: e009403. pmid:26656243
  50. 50. Spronk PE, Kuiper AV, Rommes JH, Korevaar JC, Schultz MJ. The Practice of and Documentation on Withholding and Withdrawing Life Support: A Retrospective Study in Two Dutch Intensive Care Units. Anesth Analg. 2009;109: 841–846. pmid:19690256
  51. 51. Mack JW, Cronin A, Taback N, Huskamp HA, Keating NL, Malin JL, et al. End-of-life care discussions among patients with advanced cancer: a cohort study. Ann Intern Med. 2012;156: 204–210. pmid:22312140
  52. 52. Decato TW, Engelberg RA, Downey L, Nielsen EL, Treece PD, Back AL, et al. Hospital Variation and Temporal Trends in Palliative and End-of-Life Care in the ICU. Crit Care Med. 2013; pmid:23518869
  53. 53. Vranas KC, Jopling JK, Sweeney TE, Ramsey MC, Milstein AS, Slatore CG, et al. Identifying Distinct Subgroups of ICU Patients: A Machine Learning Approach. Crit Care Med. 2017; pmid:28640021
  54. 54. Walkey AJ, Barnato AE, Wiener RS, Nallamothu BK. Accounting for Patient Preferences Regarding Life-Sustaining Treatment in Evaluations of Medical Effectiveness and Quality. Am J Respir Crit Care Med. 2017;196: 958–963. pmid:28379717
  55. 55. Schwarze ML, Campbell TC, Cunningham TV, White DB, Arnold RM. You Can’t Get What You Want: Innovation for End-Of-Life Communication in the ICU. Am J Respir Crit Care Med. 2016;193: 14–16. pmid:26474354
  56. 56. Bell SK, Roche SD, Mueller A, Dente E, O’Reilly K, Lee BS, et al. Speaking up about care concerns in the ICU: patient and family experiences, attitudes and perceived barriers. BMJ Qual Saf. 2018; bmjqs-2017-007525. pmid:30002146
  57. 57. Lane-Fall MB, Collard ML, Turnbull AE, Halpern SD, Shea JA. ICU Attending Handoff Practices: Results From a National Survey of Academic Intensivists. Crit Care Med. 2016;44: 690–698. pmid:26588827
  58. 58. Dzeng E, Colaianni A, Roland M, Chander G, Smith TJ, Kelly MP, et al. Influence of institutional culture and policies on do-not-resuscitate decision making at the end of life. JAMA Intern Med. 2015;175: 812–819. pmid:25845020
  59. 59. Hart JL, Harhay MO, Gabler NB, Ratcliffe SJ, Quill CM, Halpern SD. Variability among us intensive care units in managing the care of patients admitted with preexisting limits on life-sustaining therapies. JAMA Intern Med. 2015;175: 1019–1026. pmid:25822402
  60. 60. Hart JL, Kerlin MP. Interpreting “Do Not Resuscitate”: A Cautionary Tale of Physician Influence. Ann Am Thorac Soc. 2017;14: 491–492. pmid:28362528
  61. 61. Walkey AJ, Weinberg J, Wiener RS, Cooke CR, Lindenauer PK. Hospital Variation in Utilization of Life-Sustaining Treatments among Patients with Do Not Resuscitate Orders. Health Serv Res. 2017; n/a-n/a. pmid:28097649