- Email: [email protected]
Documentation of code status and discussion of goals of care in gravely ill hospitalized patients
Journal of Critical Care (2009) 24, 288–292
Documentation of code status and discussion of goals of care in gravely ill hospitalized patients Abigail Holley MD a,⁎, Steven J. Kravet MD, MBA b,1 , Grace Cordts MD, MPH, MS c,1 a
Section of Geriatrics, Department of Medicine, University of Chicago, Medical Center, Chicago, IL 60637, USA Division of General Internal Medicine, Department of Medicine, Johns Hopkins Bayview, Medical Center, Baltimore, MD 21224, USA c Division of Geriatrics, Department of Medicine, Johns Hopkins Bayview, Medical Center, Baltimore, MD 21224, USA b
Keywords: Code status; End of life; Goals of care; Medical education
Abstract Background: Timely discussions about goals of care in critically ill patients have been shown to be important. Methods: We conducted a retrospective chart review over 2 years (2003-2004) of patients admitted to our medical service who were classified as “expected to die.” Charts were evaluated for do-notresuscitate (DNR) documentation and discussions of goals of care. Detailed chart reviews for demographic information, cause of death, site of death, length of stay, and duration of resuscitation attempt were performed. Results: Of 497 charts identified, 434 (87.3%) had a DNR on file at the time of death. After exclusion of patients who died in less than 24 hours, 18 no-DNR charts remained. Seven noted a decision to continue aggressive care and 11 had no code status discussion documented. Younger patients and patients with cardiovascular disease were less likely to have a DNR. Resuscitation times were longer in the nodiscussion group. All patients who died without a DNR died in the intensive care unit. Seventy-six percent of discussions were done by medicine housestaff. Conclusions: Although the overall rate of DNR documentation was high, several trends emerged. Medicine housestaff in the intensive care unit would be a logical group to target for an educational intervention to address these discrepancies. © 2009 Elsevier Inc. All rights reserved.
1. Introduction Timely discussions about goals of care in critically ill patients have been shown to be important for a variety of reasons. Patients and their families list one of the most important elements of quality end-of-life care as “not to be ⁎ Corresponding author. Tel.: +1 773 834 5887; fax: +1 773 702 3538. E-mail address: [email protected] (A. Holley). 1 Work completed at Johns Hopkins Bayview Medical Center. 0883-9441/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.jcrc.2008.03.035
kept alive on life support when there is little hope for a meaningful recovery.” [1] Recognition of goals of care early on in a hospitalization may allow for mobilization of resources, such as a Palliative Care Consulting Team, which is likely to improve symptom management, perceived quality of life, and perception of in-hospital treatment [2]. In one study, Ryan et al [3] noted that explicit consultation of a Palliative Care Team regarding goals of care resulted in improvement not only in discussions of goals of care, but also in management of symptoms such as pain and nausea.
Code status in gravely ill patients From a medical systems perspective, intensive communication has been shown to reduce hospital and intensive care unit (ICU) length of stay as well as duration of use of ultimately nonbeneficial advanced life support measures in critically ill patients [4]. Early research in this area found that without direct dialogue between physicians and patients, the physician's opinion about the patient's attitude toward cardiopulmonary resuscitation was often incorrect [5]. This emphasizes the need for open communication between physicians and critically ill patients. In previous studies, families of ICU patients have rated physician communication skills to be equally important or more important than clinical skills [6]. Unfortunately, studies have revealed that these discussions often do not take place or take place only a few days before death [7]. There are many barriers to adequate discussion of goals of care [8-10]. However, there is also evidence that educational interventions may improve communication and quality of discussions between medical housestaff and patients and their families [11]. The authors evaluated the documentation of do-notresuscitate (DNR) orders and discussions of goals of care in the charts of patients who had died while being attended on the medical service who were classified as “expected to die” during their hospital admissions. Our hope was that patterns identified might help to shape quality improvement initiatives, especially in cases of patients who were expected to die and no discussion took place.
2. Methods We conducted a retrospective chart review over 2 years (2003-2004) of patients who died during an admission on our medical service and whose deaths were classified as “expected” during admission. Initial identification of expected deaths was performed by the hospital's trained quality management team, and the identification of expected deaths subsequently confirmed by 2 of the authors (AH, SJK). Deaths were identified as expected if (a) the patient was removed from life support and/or placed on comfort care measures and/ or (b) there was a note from a physician in the chart that the patient's death was imminent, the prognosis was grim, or the patient was not expected to recover. The charts were grouped by presence or absence of DNR documentation. A more detailed chart review was done on charts where no DNR order was on file. Information collected included age, sex, cause of death, length of stay, and duration of resuscitation attempt (if applicable). Cases in which the patient was admitted to the hospital for less than 24 hours before death were excluded, under the presumption that there was not time for an adequate discussion of goals of care to take place. For comparison, we surveyed all “expected death” cases from 2004 (aggregated 2003 data were unavailable) and
289 collected information on demographics, length of stay, cause of death, and site of death. Again, only cases where the patient had been in the hospital for greater than 24 hours were evaluated. A total of 18 “no DNR” charts were compared to 187 “expected death” charts from 2004.
2.1. Statistical analysis We constructed graphical displays and frequency distributions for patients with and without code status documented. Medians of resuscitation time and hospital length of stay were calculated for comparison between patients grouped as “no DNR order, no discussion” and “no DNR order, aggressive care.” P values based on Wilcoxon rank sum test were also computed for the significance of difference between groups. All analyses were carried out using Microsoft Excel and S-plus. After log transformation, the outcomes appear to be normally distributed. We used ordinary linear regression with these 2 groups as predictors, controlling for age and sex effects.
3. Results For years 2003 to 2004, 497 charts for expected deaths were identified. Of these 497 expected deaths, 434 (87.3%) had a DNR order on file at the time of death. Of the remaining 63 charts, 45 died within the first 24 hours of hospital admission and were excluded from chart review. Eighteen patients were in the hospital for more than 24 hours before an expected in-hospital death and did not have a DNR on file. In 11 cases (61%), there was no discussion documented with the patient or family about the grim prognosis or code status. In 7 cases (39%), a discussion took place with the patient and/or their family member or health care agent, and a decision was made to continue aggressive care (Fig. 1). The average age of patients with a DNR on file was 71.7 years, whereas the average age of those without a DNR was
Fig. 1
Flowchart of retrospective chart review.
290
A. Holley et al.
Table 1 Do-not-resuscitate documentation by cause of expected death (2004) Cause of death
No. with DNR on file (%)
MI/CV disease Infection Malignancy Pulmonary disease All others
29/35 48/55 18/20 49/50 25/27
(82.9%) (87.3%) (90%) (98%) (92.6%)
MI, myocardial infarction; CV, cardiovascular.
64.1 years. There was a slightly higher percentage of males in the no-DNR-order group (61.1% vs 45.9%). The difference between groups for age and sex was not statistically significant. Of the patients who died of myocardial infarction or other cardiovascular disease, 82.9% had a DNR on file at the time of death compared to 90% of patients with malignancy and 98% of patients with pulmonary disease (Table 1). The median length of stay was 9 days in the DNR-order group and the no-discussion group, and 4 days in the aggressive-care group (P = .14). All 18 patients without a DNR on file ultimately underwent cardiopulmonary resuscitation before death. Median duration of resuscitation attempt was 33 minutes in the nodiscussion group and 13 minutes in the aggressive-care group (P = .035) (Table 2). Of all expected deaths for 2004, 67.5% occurred in the ICU, but 100% who did not have a DNR on file died in the ICU (P = .004). Seventy-six percent of code status discussions were done by medicine housestaff.
4. Discussion Establishing goals of care is an important component of patient autonomy. It was encouraging to see that the overall rate of DNR discussions was high in our population. However, in a patient-centered system, all patients, especially those that are critically ill, have a right to a discussion about goals of care. The number of patients with DNR orders on file before in-hospital deaths was similar to data from other studies [7]. A direct comparison cannot be made to the SUPPORT (The Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment) trial, as that study did not explicitly differentiate between expected and unexpected deaths. It is disappointing to note, however, that little improvement was made in the 15-year period between the initiation of their study and ours, despite the rapid emergence of academic palliative care programs during the 1990s [12]. There was a trend toward older patients being more likely to have a DNR on file than younger patients, a finding not particularly surprising. The differences in age and sex between the 2 groups, however, were not statistically significant. Other studies have previously described discrepancies in rates of code status discussions based on cause
of death [11]. We had similar findings in that patients who died of malignancy-related illnesses and pulmonary disease were more likely to have a DNR on file at the time of death than those who died of cardiovascular disease. A targeted intervention would raise awareness of the need to initiate code status conversations with younger patients as well as those with cardiovascular disease. It is likely that hospitals and patients would benefit in other ways from early and consistent code status discussions as well. The no-discussion group had longer periods of ultimately futile resuscitation attempts, even when compared to those patients who desired full code status. This may reflect the lack of establishment of a clear plan of care for these very ill patients and/or lack of confidence on the part of the medical team as to how best to treat these patients in the context of their coexisting illnesses. The length of stay was equivalent in the group with DNR on file and the no-discussion group (median, 9 days for both), and shortest in the aggressive care group (median, 4 days). One possible interpretation of this is that although lack of discussion may not necessarily increase length of stay, overly aggressive intervention in these critically ill patients may have actually hastened their deaths. Another possibility is that although average length of stay may not have been changed by the presence of a DNR order, the number of very long (N30 days) ICU stays may be decreased in patients with code status documentation as shown in previous work [13]. It is important to note that all of the patients without a DNR on file died in the ICU, making the ICU a target for further emphasis on goals of care discussions. A high-quality discussion of code status should involve educating the patient and their loved ones about what full code or no code means relative to their overall medical condition [14]. Many barriers to initiation of these discussions have been outlined in the past including physician discomfort, fear of harming the patient-provider relationship, challenging family dynamics, and the time-consuming nature of these conversations when done appropriately [15]. Uncontrolled symptoms in gravely ill patients can also interfere with initiating these discussions [3]. A number of these barriers have been addressed in recent years with the addition of palliative care consultation teams to many hospitals, including active palliation of pain and other symptoms to allow patients to Table 2 groups
Length of stay and resuscitation time in no-DNR No discussion Aggressive 95% CI care
Median LOS 9 (d) 33 Median resuscitation time (min) LOS indicates length of stay.
4 13
−0.4118 to 1.0954 0.0924 to 0.8008
P .14 .035
Code status in gravely ill patients participate in these discussions. Proactive identification of critically ill patients and subsequent palliative care consultation can lead to earlier identification of prognosis and establishment of end-of-life treatment goals, reduced ICU length of stay, and reduced cost of care [16,17]. This first requires identification, however, on the part of the ICU team, of patients who may benefit from palliative services. The ability to communicate with families and establish goals of care is certainly not limited to physicians with expertise in palliative care. Studies of ICU care that have involved explicit communication interventions between ICU caregivers, patients, and their families have also shown the ability to decrease length of stay, increase patient and family satisfaction, and decrease anxiety and depression [4,18,19]. Although this may seem a daunting task at first, if done in a thoughtful and consistent manner, we would hope that providers would increase their level of comfort in talking about code status and improve upon their relationships with patients and their families. There were a number of limitations to our study. A large number of patients did have DNR documentation on file at the time of their death, leaving a relatively small number of charts without code status for review. In addition, the classification of expected deaths was based on local criteria, which may not readily translate to other institutions. We evaluated only charts from the medical service of our hospital and thus cannot comment on how well other services, such as surgery or neurology, document goals of care. We also chose to exclude charts of patients without a DNR on file who were admitted for less than 24 hours. Although we did not analyze that data specifically, it is likely that most, if not all, of the 45 patients who died within 24 hours without a DNR order had attempts at resuscitation, as the default in our society for gravely ill patients in the absence of specific code status directives is toward aggressive care. It is important to acknowledge that all patients who present critically ill, and are expected to die, should ideally have discussions on goals of care. We felt that in these cases, the patients may not have been able to express their wishes verbally or through an advanced directive, and there may not have been family immediately available. We feel it is important to attempt discussions in all patients but chose to present data that allowed for the most ideal opportunity for discussions. Thus, our conclusions may be understated. Finally, our setting was a large, academic hospital with “teaching” attending physicians overseeing patient care. Documentation of code status and goals of care may be different in the case of private attendings overseeing their own inpatients. Despite these limitations, a number of trends came to light that may help to shape future practice, behaviors, and training of housestaff in academic centers. Medicine housestaff, because of their high level of face-to-face contact with patients and their families, are a logical group to target for an educational intervention. Given that all of the patients in our study who died without a DNR on file were admitted to the
291 ICU, explicit teaching regarding addressing goals of care during rotations in the ICU would seem to be warranted. In fact, we have recently incorporated such a core component into the orientation to our medical ICU rotation. In addition, the data from this study were collected just before the establishment of a palliative care consultation team at our institution. Giving housestaff the opportunity to rotate with the palliative care team may also provide them with an opportunity to develop these important skills. The Accreditation Council for Graduate Medical Education has identified 6 core competencies: communication, professionalism, systems-based practice, practice-based learning and improvement, knowledge, and patient care [20]. Departments of medicine have been noted to successfully teach about morbidity and mortality in the context of the core competencies [21]. Improving the practice of endof-life discussions meets elements of communication and professionalism. Patient care can be improved through less unnecessary morbidity associated with prolonged resuscitation attempts. With the recent initiation of inpatient palliative care services at many hospitals, the possibility exists to raise awareness of a systems approach. Lastly, reflecting on (as if in a mirror) [22] and incorporating this knowledge could result in improved practice for housestaff and all members of the team. In summary, we hope that by providing evidence of the opportunity to continue to improve end of life discussions, we can better meet the needs of gravely ill patients and their families.
Acknowledgments The authors are indebted to QiLu Yu, PhD, for her statistical assistance. Dr Kravet is a Miller-Coulson Family Scholar through the Johns Hopkins Center for Innovative Medicine.
References [1] Heyland DK, Dodek P, Rocker G, et al. What matters most in end-oflife care: perceptions of seriously ill patients and their family members. Can Med Assoc J 2006;174(5):627-33. [2] Schrader SL, Horner A, Eidsness L, et al. A team approach in palliative care: enhancing outcomes. S D J Med 2002;55(7):269-78. [3] Ryan A, Carter J, Lucas J, et al. You need not make the journey alone: overcoming impediments to providing palliative care in a public urban teaching hospital. Am J Hosp Palliat Care 2002;19(3):171-80. [4] Lilly CM, De Meo DL, Sonna LA, et al. An intensive communication intervention for the critically ill. Am J Med 2000;109(6):469-75. [5] Bedell SE, Delbanco TL. Choices about cardiopulmonary resuscitation in the hospital. When do physicians talk with patients? N Engl J Med 1984;310(17):1089-93. [6] Curtis JR. Communicating about end-of-life care with patients and families in the intensive care unit. Crit Care Clin 2004;20(3):363-80, viii.
292 [7] The SUPPORT Principal Investigators. A controlled trial to improve care for seriously ill hospitalized patients. The study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT). The SUPPORT Principal Investigators. JAMA 1995;274 (20):1591-8. [8] Gorman TE, Ahern SP, Wiseman J, et al. Residents' end-of-life decision making with adult hospitalized patients: a review of the literature. Acad Med 2005;80(7):622-33. [9] Tulsky JA, Chesney MA, Lo B. See one, do one, teach one? House staff experience discussing do-not-resuscitate orders. Arch Intern Med 1996;156(12):1285-9. [10] Sulmasy DP, Song KY, Marx ES, et al. Strategies to promote the use of advance directives in a residency outpatient practice. J Gen Intern Med 1996;11(11):657-63. [11] Sulmasy DP, Geller G, Faden R, et al. The quality of mercy. Caring for patients with 'do not resuscitate' orders. JAMA 1992;267(5):682-6. [12] Hayley DC, Sachs GA. A brief history and lessons learned from twin efforts to transform medicine. Clin Geriatr Med 2005;21(1):3-15, vii. [13] Rapoport J, Teres D, Lemeshow S. Resource use implications of do not resuscitate orders for intensive care unit patients. Am J Respir Crit Care Med 1996;153(1):185-90. [14] VonGunten CF. The art of oncology: when the tumor is not the target. Discussing Do-Not-Resuscitate Status. J Clin Oncol 2001;19(5):1576-81.
A. Holley et al. [15] Calam B, Andrew R. CPR or DNR? End-of-life decision making on a family practice teaching ward. Can Fam Physician 2000;46: 340-6. [16] Campbell ML, Guzman JA. Impact of a proactive approach to improve end-of-life care in a medical ICU. Chest 2003;123(1):266-71. [17] Campbell ML, Guzman JA. A proactive approach to improve end-oflife care in a medical intensive care unit for patients with terminal dementia. Crit Care Med 2004;32(9):1839-43. [18] Lautrette A, Darmon M, Megarbane B, et al. A communication strategy and brochure for relatives of patients dying in the ICU. N Engl J Med 2007;356(5):469-78. [19] Curtis JR, Patrick DL, Shannon SE, et al. The family conference as a focus to improve communication about end-of-life care in the intensive care unit: opportunities for improvement. Crit Care Med 2001;29(2 Suppl):N26-33. [20] Swing S. ACGME launches outcome assessment project. JAMA 1998;279(18):1492. [21] Kravet SJ, Howell E, Wright SM. Morbidity and mortality conference, grand rounds, and the ACGME's core competencies. J Gen Intern Med 2006;21(11):1192-4. [22] Ziegelstein RC, Fiebach NH. ‘The mirror’ and ‘the village’: a new method for teaching practice-based learning and improvement and systems-based practice. Acad Med 2004;79(1):83-8.
Documentation of code status and discussion of goals of care in gravely ill hospitalized patients Abigail Holley MD a,⁎, Steven J. Kravet MD, MBA b,1 , Grace Cordts MD, MPH, MS c,1 a
Section of Geriatrics, Department of Medicine, University of Chicago, Medical Center, Chicago, IL 60637, USA Division of General Internal Medicine, Department of Medicine, Johns Hopkins Bayview, Medical Center, Baltimore, MD 21224, USA c Division of Geriatrics, Department of Medicine, Johns Hopkins Bayview, Medical Center, Baltimore, MD 21224, USA b
Keywords: Code status; End of life; Goals of care; Medical education
Abstract Background: Timely discussions about goals of care in critically ill patients have been shown to be important. Methods: We conducted a retrospective chart review over 2 years (2003-2004) of patients admitted to our medical service who were classified as “expected to die.” Charts were evaluated for do-notresuscitate (DNR) documentation and discussions of goals of care. Detailed chart reviews for demographic information, cause of death, site of death, length of stay, and duration of resuscitation attempt were performed. Results: Of 497 charts identified, 434 (87.3%) had a DNR on file at the time of death. After exclusion of patients who died in less than 24 hours, 18 no-DNR charts remained. Seven noted a decision to continue aggressive care and 11 had no code status discussion documented. Younger patients and patients with cardiovascular disease were less likely to have a DNR. Resuscitation times were longer in the nodiscussion group. All patients who died without a DNR died in the intensive care unit. Seventy-six percent of discussions were done by medicine housestaff. Conclusions: Although the overall rate of DNR documentation was high, several trends emerged. Medicine housestaff in the intensive care unit would be a logical group to target for an educational intervention to address these discrepancies. © 2009 Elsevier Inc. All rights reserved.
1. Introduction Timely discussions about goals of care in critically ill patients have been shown to be important for a variety of reasons. Patients and their families list one of the most important elements of quality end-of-life care as “not to be ⁎ Corresponding author. Tel.: +1 773 834 5887; fax: +1 773 702 3538. E-mail address: [email protected] (A. Holley). 1 Work completed at Johns Hopkins Bayview Medical Center. 0883-9441/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.jcrc.2008.03.035
kept alive on life support when there is little hope for a meaningful recovery.” [1] Recognition of goals of care early on in a hospitalization may allow for mobilization of resources, such as a Palliative Care Consulting Team, which is likely to improve symptom management, perceived quality of life, and perception of in-hospital treatment [2]. In one study, Ryan et al [3] noted that explicit consultation of a Palliative Care Team regarding goals of care resulted in improvement not only in discussions of goals of care, but also in management of symptoms such as pain and nausea.
Code status in gravely ill patients From a medical systems perspective, intensive communication has been shown to reduce hospital and intensive care unit (ICU) length of stay as well as duration of use of ultimately nonbeneficial advanced life support measures in critically ill patients [4]. Early research in this area found that without direct dialogue between physicians and patients, the physician's opinion about the patient's attitude toward cardiopulmonary resuscitation was often incorrect [5]. This emphasizes the need for open communication between physicians and critically ill patients. In previous studies, families of ICU patients have rated physician communication skills to be equally important or more important than clinical skills [6]. Unfortunately, studies have revealed that these discussions often do not take place or take place only a few days before death [7]. There are many barriers to adequate discussion of goals of care [8-10]. However, there is also evidence that educational interventions may improve communication and quality of discussions between medical housestaff and patients and their families [11]. The authors evaluated the documentation of do-notresuscitate (DNR) orders and discussions of goals of care in the charts of patients who had died while being attended on the medical service who were classified as “expected to die” during their hospital admissions. Our hope was that patterns identified might help to shape quality improvement initiatives, especially in cases of patients who were expected to die and no discussion took place.
2. Methods We conducted a retrospective chart review over 2 years (2003-2004) of patients who died during an admission on our medical service and whose deaths were classified as “expected” during admission. Initial identification of expected deaths was performed by the hospital's trained quality management team, and the identification of expected deaths subsequently confirmed by 2 of the authors (AH, SJK). Deaths were identified as expected if (a) the patient was removed from life support and/or placed on comfort care measures and/ or (b) there was a note from a physician in the chart that the patient's death was imminent, the prognosis was grim, or the patient was not expected to recover. The charts were grouped by presence or absence of DNR documentation. A more detailed chart review was done on charts where no DNR order was on file. Information collected included age, sex, cause of death, length of stay, and duration of resuscitation attempt (if applicable). Cases in which the patient was admitted to the hospital for less than 24 hours before death were excluded, under the presumption that there was not time for an adequate discussion of goals of care to take place. For comparison, we surveyed all “expected death” cases from 2004 (aggregated 2003 data were unavailable) and
289 collected information on demographics, length of stay, cause of death, and site of death. Again, only cases where the patient had been in the hospital for greater than 24 hours were evaluated. A total of 18 “no DNR” charts were compared to 187 “expected death” charts from 2004.
2.1. Statistical analysis We constructed graphical displays and frequency distributions for patients with and without code status documented. Medians of resuscitation time and hospital length of stay were calculated for comparison between patients grouped as “no DNR order, no discussion” and “no DNR order, aggressive care.” P values based on Wilcoxon rank sum test were also computed for the significance of difference between groups. All analyses were carried out using Microsoft Excel and S-plus. After log transformation, the outcomes appear to be normally distributed. We used ordinary linear regression with these 2 groups as predictors, controlling for age and sex effects.
3. Results For years 2003 to 2004, 497 charts for expected deaths were identified. Of these 497 expected deaths, 434 (87.3%) had a DNR order on file at the time of death. Of the remaining 63 charts, 45 died within the first 24 hours of hospital admission and were excluded from chart review. Eighteen patients were in the hospital for more than 24 hours before an expected in-hospital death and did not have a DNR on file. In 11 cases (61%), there was no discussion documented with the patient or family about the grim prognosis or code status. In 7 cases (39%), a discussion took place with the patient and/or their family member or health care agent, and a decision was made to continue aggressive care (Fig. 1). The average age of patients with a DNR on file was 71.7 years, whereas the average age of those without a DNR was
Fig. 1
Flowchart of retrospective chart review.
290
A. Holley et al.
Table 1 Do-not-resuscitate documentation by cause of expected death (2004) Cause of death
No. with DNR on file (%)
MI/CV disease Infection Malignancy Pulmonary disease All others
29/35 48/55 18/20 49/50 25/27
(82.9%) (87.3%) (90%) (98%) (92.6%)
MI, myocardial infarction; CV, cardiovascular.
64.1 years. There was a slightly higher percentage of males in the no-DNR-order group (61.1% vs 45.9%). The difference between groups for age and sex was not statistically significant. Of the patients who died of myocardial infarction or other cardiovascular disease, 82.9% had a DNR on file at the time of death compared to 90% of patients with malignancy and 98% of patients with pulmonary disease (Table 1). The median length of stay was 9 days in the DNR-order group and the no-discussion group, and 4 days in the aggressive-care group (P = .14). All 18 patients without a DNR on file ultimately underwent cardiopulmonary resuscitation before death. Median duration of resuscitation attempt was 33 minutes in the nodiscussion group and 13 minutes in the aggressive-care group (P = .035) (Table 2). Of all expected deaths for 2004, 67.5% occurred in the ICU, but 100% who did not have a DNR on file died in the ICU (P = .004). Seventy-six percent of code status discussions were done by medicine housestaff.
4. Discussion Establishing goals of care is an important component of patient autonomy. It was encouraging to see that the overall rate of DNR discussions was high in our population. However, in a patient-centered system, all patients, especially those that are critically ill, have a right to a discussion about goals of care. The number of patients with DNR orders on file before in-hospital deaths was similar to data from other studies [7]. A direct comparison cannot be made to the SUPPORT (The Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatment) trial, as that study did not explicitly differentiate between expected and unexpected deaths. It is disappointing to note, however, that little improvement was made in the 15-year period between the initiation of their study and ours, despite the rapid emergence of academic palliative care programs during the 1990s [12]. There was a trend toward older patients being more likely to have a DNR on file than younger patients, a finding not particularly surprising. The differences in age and sex between the 2 groups, however, were not statistically significant. Other studies have previously described discrepancies in rates of code status discussions based on cause
of death [11]. We had similar findings in that patients who died of malignancy-related illnesses and pulmonary disease were more likely to have a DNR on file at the time of death than those who died of cardiovascular disease. A targeted intervention would raise awareness of the need to initiate code status conversations with younger patients as well as those with cardiovascular disease. It is likely that hospitals and patients would benefit in other ways from early and consistent code status discussions as well. The no-discussion group had longer periods of ultimately futile resuscitation attempts, even when compared to those patients who desired full code status. This may reflect the lack of establishment of a clear plan of care for these very ill patients and/or lack of confidence on the part of the medical team as to how best to treat these patients in the context of their coexisting illnesses. The length of stay was equivalent in the group with DNR on file and the no-discussion group (median, 9 days for both), and shortest in the aggressive care group (median, 4 days). One possible interpretation of this is that although lack of discussion may not necessarily increase length of stay, overly aggressive intervention in these critically ill patients may have actually hastened their deaths. Another possibility is that although average length of stay may not have been changed by the presence of a DNR order, the number of very long (N30 days) ICU stays may be decreased in patients with code status documentation as shown in previous work [13]. It is important to note that all of the patients without a DNR on file died in the ICU, making the ICU a target for further emphasis on goals of care discussions. A high-quality discussion of code status should involve educating the patient and their loved ones about what full code or no code means relative to their overall medical condition [14]. Many barriers to initiation of these discussions have been outlined in the past including physician discomfort, fear of harming the patient-provider relationship, challenging family dynamics, and the time-consuming nature of these conversations when done appropriately [15]. Uncontrolled symptoms in gravely ill patients can also interfere with initiating these discussions [3]. A number of these barriers have been addressed in recent years with the addition of palliative care consultation teams to many hospitals, including active palliation of pain and other symptoms to allow patients to Table 2 groups
Length of stay and resuscitation time in no-DNR No discussion Aggressive 95% CI care
Median LOS 9 (d) 33 Median resuscitation time (min) LOS indicates length of stay.
4 13
−0.4118 to 1.0954 0.0924 to 0.8008
P .14 .035
Code status in gravely ill patients participate in these discussions. Proactive identification of critically ill patients and subsequent palliative care consultation can lead to earlier identification of prognosis and establishment of end-of-life treatment goals, reduced ICU length of stay, and reduced cost of care [16,17]. This first requires identification, however, on the part of the ICU team, of patients who may benefit from palliative services. The ability to communicate with families and establish goals of care is certainly not limited to physicians with expertise in palliative care. Studies of ICU care that have involved explicit communication interventions between ICU caregivers, patients, and their families have also shown the ability to decrease length of stay, increase patient and family satisfaction, and decrease anxiety and depression [4,18,19]. Although this may seem a daunting task at first, if done in a thoughtful and consistent manner, we would hope that providers would increase their level of comfort in talking about code status and improve upon their relationships with patients and their families. There were a number of limitations to our study. A large number of patients did have DNR documentation on file at the time of their death, leaving a relatively small number of charts without code status for review. In addition, the classification of expected deaths was based on local criteria, which may not readily translate to other institutions. We evaluated only charts from the medical service of our hospital and thus cannot comment on how well other services, such as surgery or neurology, document goals of care. We also chose to exclude charts of patients without a DNR on file who were admitted for less than 24 hours. Although we did not analyze that data specifically, it is likely that most, if not all, of the 45 patients who died within 24 hours without a DNR order had attempts at resuscitation, as the default in our society for gravely ill patients in the absence of specific code status directives is toward aggressive care. It is important to acknowledge that all patients who present critically ill, and are expected to die, should ideally have discussions on goals of care. We felt that in these cases, the patients may not have been able to express their wishes verbally or through an advanced directive, and there may not have been family immediately available. We feel it is important to attempt discussions in all patients but chose to present data that allowed for the most ideal opportunity for discussions. Thus, our conclusions may be understated. Finally, our setting was a large, academic hospital with “teaching” attending physicians overseeing patient care. Documentation of code status and goals of care may be different in the case of private attendings overseeing their own inpatients. Despite these limitations, a number of trends came to light that may help to shape future practice, behaviors, and training of housestaff in academic centers. Medicine housestaff, because of their high level of face-to-face contact with patients and their families, are a logical group to target for an educational intervention. Given that all of the patients in our study who died without a DNR on file were admitted to the
291 ICU, explicit teaching regarding addressing goals of care during rotations in the ICU would seem to be warranted. In fact, we have recently incorporated such a core component into the orientation to our medical ICU rotation. In addition, the data from this study were collected just before the establishment of a palliative care consultation team at our institution. Giving housestaff the opportunity to rotate with the palliative care team may also provide them with an opportunity to develop these important skills. The Accreditation Council for Graduate Medical Education has identified 6 core competencies: communication, professionalism, systems-based practice, practice-based learning and improvement, knowledge, and patient care [20]. Departments of medicine have been noted to successfully teach about morbidity and mortality in the context of the core competencies [21]. Improving the practice of endof-life discussions meets elements of communication and professionalism. Patient care can be improved through less unnecessary morbidity associated with prolonged resuscitation attempts. With the recent initiation of inpatient palliative care services at many hospitals, the possibility exists to raise awareness of a systems approach. Lastly, reflecting on (as if in a mirror) [22] and incorporating this knowledge could result in improved practice for housestaff and all members of the team. In summary, we hope that by providing evidence of the opportunity to continue to improve end of life discussions, we can better meet the needs of gravely ill patients and their families.
Acknowledgments The authors are indebted to QiLu Yu, PhD, for her statistical assistance. Dr Kravet is a Miller-Coulson Family Scholar through the Johns Hopkins Center for Innovative Medicine.
References [1] Heyland DK, Dodek P, Rocker G, et al. What matters most in end-oflife care: perceptions of seriously ill patients and their family members. Can Med Assoc J 2006;174(5):627-33. [2] Schrader SL, Horner A, Eidsness L, et al. A team approach in palliative care: enhancing outcomes. S D J Med 2002;55(7):269-78. [3] Ryan A, Carter J, Lucas J, et al. You need not make the journey alone: overcoming impediments to providing palliative care in a public urban teaching hospital. Am J Hosp Palliat Care 2002;19(3):171-80. [4] Lilly CM, De Meo DL, Sonna LA, et al. An intensive communication intervention for the critically ill. Am J Med 2000;109(6):469-75. [5] Bedell SE, Delbanco TL. Choices about cardiopulmonary resuscitation in the hospital. When do physicians talk with patients? N Engl J Med 1984;310(17):1089-93. [6] Curtis JR. Communicating about end-of-life care with patients and families in the intensive care unit. Crit Care Clin 2004;20(3):363-80, viii.
292 [7] The SUPPORT Principal Investigators. A controlled trial to improve care for seriously ill hospitalized patients. The study to understand prognoses and preferences for outcomes and risks of treatments (SUPPORT). The SUPPORT Principal Investigators. JAMA 1995;274 (20):1591-8. [8] Gorman TE, Ahern SP, Wiseman J, et al. Residents' end-of-life decision making with adult hospitalized patients: a review of the literature. Acad Med 2005;80(7):622-33. [9] Tulsky JA, Chesney MA, Lo B. See one, do one, teach one? House staff experience discussing do-not-resuscitate orders. Arch Intern Med 1996;156(12):1285-9. [10] Sulmasy DP, Song KY, Marx ES, et al. Strategies to promote the use of advance directives in a residency outpatient practice. J Gen Intern Med 1996;11(11):657-63. [11] Sulmasy DP, Geller G, Faden R, et al. The quality of mercy. Caring for patients with 'do not resuscitate' orders. JAMA 1992;267(5):682-6. [12] Hayley DC, Sachs GA. A brief history and lessons learned from twin efforts to transform medicine. Clin Geriatr Med 2005;21(1):3-15, vii. [13] Rapoport J, Teres D, Lemeshow S. Resource use implications of do not resuscitate orders for intensive care unit patients. Am J Respir Crit Care Med 1996;153(1):185-90. [14] VonGunten CF. The art of oncology: when the tumor is not the target. Discussing Do-Not-Resuscitate Status. J Clin Oncol 2001;19(5):1576-81.
A. Holley et al. [15] Calam B, Andrew R. CPR or DNR? End-of-life decision making on a family practice teaching ward. Can Fam Physician 2000;46: 340-6. [16] Campbell ML, Guzman JA. Impact of a proactive approach to improve end-of-life care in a medical ICU. Chest 2003;123(1):266-71. [17] Campbell ML, Guzman JA. A proactive approach to improve end-oflife care in a medical intensive care unit for patients with terminal dementia. Crit Care Med 2004;32(9):1839-43. [18] Lautrette A, Darmon M, Megarbane B, et al. A communication strategy and brochure for relatives of patients dying in the ICU. N Engl J Med 2007;356(5):469-78. [19] Curtis JR, Patrick DL, Shannon SE, et al. The family conference as a focus to improve communication about end-of-life care in the intensive care unit: opportunities for improvement. Crit Care Med 2001;29(2 Suppl):N26-33. [20] Swing S. ACGME launches outcome assessment project. JAMA 1998;279(18):1492. [21] Kravet SJ, Howell E, Wright SM. Morbidity and mortality conference, grand rounds, and the ACGME's core competencies. J Gen Intern Med 2006;21(11):1192-4. [22] Ziegelstein RC, Fiebach NH. ‘The mirror’ and ‘the village’: a new method for teaching practice-based learning and improvement and systems-based practice. Acad Med 2004;79(1):83-8.