Journal Logo

Empirical Investigations

Anesthesiologist Management of Perioperative Do-Not-Resuscitate Orders: A Simulation-Based Experiment

Waisel, David B. MD; Simon, Robert EdD; Truog, Robert D. MD; Baboolal, Hemanth MD; Raemer, Daniel B. PhD

Author Information
Simulation in Healthcare: The Journal of the Society for Simulation in Healthcare: July 2009 - Volume 4 - Issue 2 - p 70-76
doi: 10.1097/SIH.0b013e31819e137b
  • Free


In 1993, the American Society of Anesthesiologists (ASA) published guidelines recommending that Do-Not-Resuscitate (DNR) orders for patients coming to surgery be reevaluated for the perioperative period.1 This was followed by similar statements by other groups with a stake in reevaluating DNR orders for the perioperative period, including the American College of Surgeons and the Association of periOperative Registered Nurses.2,3 Current guidelines emphasize perioperative reevaluation of the DNR order, discussion of resuscitation options and the active agreement of the patient or surrogate decision maker.4 All the Harvard hospitals represented in this study have adopted the ASA Guidelines with minor variations as hospital policy with minor variations.

Three options are suggested for determining perioperative resuscitation. A patient may temporarily suspend DNR status and choose full resuscitation. In the procedure-directed approach, patient and physicians determine which specific interventions will be performed, much like the more familiar ward DNR orders. The procedure-directed approach often imposed rules too rigid for the vagaries of the operating room. Physicians may feel stuck having to choose between honoring the specifics of the agreement with the patient and honoring what they thought to be the patient's true intent. To permit physicians to have flexibility in determining whether resuscitation would be consistent with the patient's goals, and to take advantage of the time-limited, one-on-one relationship between operating room providers and patients, a goal-directed approach was developed. Goal-directed approaches permit patients to communicate to providers their preferences in common language (eg, “I want to be able to function like I am functioning now”).5 The most common preference is for a patient to desire resuscitative efforts during surgery and in the post anesthesia care unit only if the adverse clinical events are believed to be both temporary and reversible in the clinical judgment of the attending anesthesiologist and surgeon. This option requires the patient to trust the judgments of the providers to use resuscitative interventions judiciously based on their understanding of the patient's values and goals of treatment.5 For example, consider a situation in which tracheal intubation would likely provide necessary short-term support that would lead to extubation in several hours. Although a procedure-directed approach may forbid tracheal intubation, the most common goal-directed perioperative approach would permit tracheal intubation.

Despite numerous education efforts, reevaluation of DNR orders for the perioperative period does not seem to be commonly performed.6 We undertook this study 1. to assess the extent and nature of reevaluation of DNR orders in the perioperative period by practicing anesthesiologists and 2. to observe any relationship between the preoperative interview regarding reevaluation of the DNR order and the performance of resuscitation in the operating room when the patient's condition deteriorated.


With approval of the Institutional Review Board, the experiment was conducted as part of an educational course on Anesthesia Crisis Resource Management (ACRM) held at the Center for Medical Simulation.7 All attending anesthesiologists in the Harvard system are required to participate in an ACRM course at least once in every 3 years with three to six individuals participating in the 6-hour program on a given day. All participants in this experiment had one or more prior exposures to an ACRM course of this kind. Participants were not informed that the study was evaluating aspects of perioperative DNR orders.

At the beginning of each course, a single participant was arbitrarily selected as the primary subject. The subject was brought into a room where a patient (an actor) was waiting with a nurse (also an actor). The nurse told the subject that the patient, Mr. Samuels, was added to the operating room list for a central venous catheter line (Porta-cath, Smiths Medical, St. Paul, MN) placement later in the morning. It was explained that Mr. Samuels had been seen by an anesthesia resident in the preoperative clinic a week before. His surgery had been moved up to today and no one had seen him this morning. The nurse handed the subject a chart and asked the subject to interview Mr. Samuels. The nurse then informed the pair that he would return in 10 minutes or so to place Mr. Samuels on a stretcher and to insert an intravenous catheter. Mr. Samuels, dressed in a hospital gown, appeared to have difficulty finding a comfortable position due to back pain but was conversant and animated. The chart indicated that Mr. Samuels had prostate cancer with extensive metastases to the lumbar spine and a grim prognosis. The chart contained a complete history and physical, a preanesthetic assessment, signed anesthesia and surgical consents, a letter from the patient's oncologist to the primary care physician, radiology report, ECG and report, and various laboratory reports (complete blood count, coagulation studies, electrolyte panel, thyroid panel, and urinalysis). The chart also contained a properly documented end-of-life form specifying that the patient wished not to be resuscitated in the event of a cardiac arrhythmia, not to be intubated or mechanically ventilated, and not to have a feeding tube inserted (a “DNR” order). The history and physical, preanesthetic assessment, and letter all noted that the patient was “DNR.” Mr. Samuels was reported as healthy except for his cancer.

Mr. Samuels's responses to questions in the interview were generally scripted. For example, if asked about his past surgery, he indicated that he had undergone a uvulopalatopharyngoplasty for sleep apnea 20 years ago (not a tonsillectomy as indicated in the chart) and that he suffered “nausea and vomiting and bleeding” postoperatively. If follow-up questions were asked, he would respond that the nausea and vomiting were controlled with medication. If asked, he would respond that the sleep apnea had resolved with the surgery and weight loss. Some of Mr. Samuels's scripted responses hinted at possible problems, but if the subject followed up, it would become clear that no further data were required beyond that which was already contained in the written chart. For instance, Mr. Samuels's chart and his verbal report indicate an allergy to erythromycin manifested as upset stomach.

Mr. Samuels's responses to discussion about his DNR status were scripted (Table 1). Mr. Samuels's general position was that he would agree to what would have to be done in the operating room to receive his central venous catheter but under no circumstances did he want a lengthy intensive care unit (ICU) stay. Mr. Samuels's response was designed to echo a common goal-directed perioperative DNR order that he would accept resuscitative efforts only “if the adverse clinical events were believed to be both temporary and reversible” and that he would likely be returned to his current state of function without significant burden. Regardless of the subject's position, Mr. Samuels never agreed to suspend DNR for the perioperative period. Mr. Samuels always asked the subject whether he thought his wife could pick him up around 4 pm and whether he would be able to eat the meatloaf dinner she was preparing. When the interview seemed to be nearing completion, the nurse returned to bring the patient to the preoperative area.

Table 1
Table 1:
Sample DNR-Related Discussion Points

Approximately 4 hours later, the subject was called to the simulated operating room. The subject found a sedated and conversant Mr. Samuels prepared and draped for the central venous catheter placement under monitored anesthesia care. The patient was now a mannequin simulator (MedSim Eagle), although the patient actor (Mr. Samuels) spoke to the subject through a speaker located in the mannequin head. The anesthesiologist who had just begun the case transferred care of the patient to the subject and made it clear that this was the same patient that the subject had interviewed earlier in the day. Mr. Samuels unfailingly recognized and engaged verbally (albeit in a somewhat somnolent state) the physician from the morning conversation. If at any time the subject requested help, after a short delay other course participants were allowed to enter the operating room to assist. The surgeon (an actor) injected local anesthetic and proceeded to insert the central venous catheter. Although inserting the catheter, the surgeon made a scripted statement that air was withdrawn and there may have been a lung puncture (event 1). Mr. Samuels then complained of difficulty in breathing, and the pulse oximetry reading began to decrease indicative of a pneumothorax. If a chest tube was requested by the subject, the surgeon's scripted response was, “Are you sure we should put in a chest tube? Isn't this patient DNR?” Regardless of the subject's response, the surgeon inserted the chest tube and the pneumothorax was relieved. After an interval of stability, the patient became tachycardic, hypotensive, and new onset S-T elevations began to appear in lead II of the electrocardiogram. As the hypotension worsened and the S-T elevations increased to 4 mm, the patient eventually became unresponsive (event 2). If the subject attempted to intubate the patient's airway, the surgeon's scripted inquiry was, “Are you sure you want to intubate? I think the patient is DNI as well as DNR” (DNI stands for Do Not Intubate. Anesthesiologists from participating institutions use this terminology to distinguish intubation from other possible resuscitation methods). Regardless of treatment, the patient became unstable and the cardiac rhythm degenerated to ventricular fibrillation (event 3). If the subject directed someone to begin chest compressions, the surgeon's scripted response was, “This is certainly resuscitating him. Isn't this patient DNR? Are you sure you want to do this?” If the subject repeated the order or ordered cessation of the resuscitation efforts, the team complied. If the subject ordered electrical defibrillation (event 4), the surgeon's scripted response was, “Does this patient want to be shocked? I thought shocking a patient with a DNR order was going against their wishes. Are you sure you want to do this?” If the subject or designee delivered electrical defibrillation correctly, sinus rhythm was restored, but the patient remained hypotensive, demonstrated ectopy on the ECG, and remained unresponsive. The surgeon's scripted comment was then, “Well if we are resuscitating him fully, why don't we call cardiology and see if we should get him a balloon pump?” (event 5) Subsequent to a clear answer from the primary subject and the subject's verbal orders to the circulating nurse to call the cardiac catheterization laboratory, the case scenario was ended. After the case, a facilitated debriefing using video was conducted with all subjects.

The scenario was designed so that according to American Society of Anesthesiologists guidelines, subjects would address perioperative resuscitation with the patient, document an agreed upon goal-directed or procedure-directed order, treat the pneumothorax with a chest tube, provide endotracheal intubation to support the patient's breathing, and possibly perform cardiopulmonary resuscitation and defibrillation. It is not totally clear, however, that in the circumstance of ventricular fibrillation inserting an endotracheal tube and performing cardiopulmonary resuscitation that the physicians were acting according to the patient's stated goal of not facing an extended stay in the ICU. At the beginning of the resuscitation, anesthesiologists could reasonably view the inciting events and effects as reversible and temporary. Later in the resuscitation, it would not be reasonable for anesthesiologists to consider the events as reversible and temporary. Because of the subsequent commitment to a long ICU stay and the decreased likelihood of returning Mr. Samuels to his baseline, an intraortic balloon pump should not be inserted and resuscitation should cease. Decisions regarding acceptable interventions were made, in a reflective manner, by consulting experts in perioperative ethics and perioperative DNR orders, official statements, and peer-reviewed literature.


Primary subjects, those who assisted the primary subject in patient care, and those who were strictly observers were asked to complete a questionnaire immediately after the case before any discussion took place. The questionnaire is included in Appendix.

Data Analysis

Scores for each question of the questionnaire were analyzed by computing the frequency of each response. Video review of all patient interviews was conducted independently by two of the investigators (DBW, HB). Reviewers noted the duration of the interview and duration of any DNR discussion. In addition, reviewers coded whether any discussion of the following topics occurred: (1) confirmation of the DNR order, (2) differentiation of anesthesia and resuscitation, (3) possible complications that could result in the need for resuscitation, (4) involvement of any other care team members (eg, surgeon, circulating nurse), and (5) documentation of the perioperative DNR understanding. If DNR was discussed, the investigators further coded whether the anesthesiologist (subject) guided the DNR discussion toward suspending DNR during the perioperative period, led the discussion toward selection of specifically allowed and denied procedures, or led the discussion toward discussing the goals of the patient with respect to DNR status. A review by the third investigator (DBR) resolved differences. The response of the subjects during the case scenario to each of the five intraoperative events was noted by agreement of two of the investigators (DBR, RS) during the case in real time. Whether the team decided to discontinue resuscitation at any point in the case scenario was noted similarly.

Differences in proportions were tested using χ2 analysis and two-tailed two-sample t test between proportions. Significance was assumed if P ≤ 0.05 unless noted otherwise.


Simulation Results

In 30 of 31 classes, all attendees gave informed consent and participated in the study. Table 2 reviews the attributes of the preoperative interviews. For all 30 subjects, 17 addressed the DNR order and 13 did not. Of the 30 subjects performing the preoperative interview, 22 stated postsimulation that they recognized that the patient had a DNR order; of these 22 subjects, 17 addressed resuscitation during the preoperative interview and five did not. The mean length of all 30 preoperative interviews was 13:35 ± 6:19 (mean ± SD, expressed in minutes:seconds). The mean length of the 17 interviews in which perioperative resuscitation was addressed was 14:53 ± 7:42. The mean length of the perioperative resuscitation discussion was 3:52 ± 2:48. The mean length of the 13 interviews in which perioperative resuscitation was not addressed (the five subjects who recognized the DNR order but did not discuss the order and the eight subjects who did not recognize the DNR order) was 11:52 ± 3:26. Exclusive of the perioperative resuscitation discussion, comparisons of the mean times between those who did or did not address DNR status were not significantly different.

Table 2
Table 2:
Attributes of the Preoperative Discussions of DNR Status

Table 3 describes the resuscitation patterns performed by the 30 groups. There was no discernable relationship between the preoperative discussion of resuscitation and the intraoperative pattern of resuscitation. Of the 27 groups that chose to continue the resuscitation through the end of the scenario, 16 had addressed perioperative resuscitation during the preoperative interview and 11 had not addressed perioperative resuscitation. More specifically, of the 19 groups that provided full resuscitation (resuscitation pattern 1), eight of the preoperative interviews resulted in a procedure-directed or goal-directed limitation of resuscitation. Of the three groups that decided to discontinue resuscitation (resuscitation patterns 3, 4, and 5), one group had suspended the DNR order, and two groups had not addressed the DNR order. No group performed the scenario conforming to our interpretation of the American Society of Anesthesiologists guidelines. One group (resuscitation pattern 3) that arbitrarily considered the DNR order suspended performed the intraoperative resuscitation (preoperative interview notwithstanding) in a way that we would consider consistent with the guidelines. All primary subjects were called for help at some point between event 1 and event 2. Between 1 and 4 “helper” subjects entered the operating room at various intervals.

Table 3
Table 3:
Patterns of Resuscitation

Questionnaire Results

Thirty primary subjects and 102 additional class attendees completed questionnaires. Of those 102, 72 became involved in the case at some point and 30 remained observers throughout. The mean experience of all the subjects as attending anesthesiologists was 13.1 ± 10.4 years with a median of 12 years, range of 0 to 40 years, lower quartile of 3.5 years, upper quartile of 19.5 years, and an interquartile range of 16 years.

Thirty-eight (50/132) percent of subjects indicated that they had attended an academic presentation on the topic of perioperative DNR orders within the last year. There was no association between those who were exposed to an academic presentation of perioperative DNR orders and familiarity with the ASA guidelines on perioperative DNR orders. Five percent (6/132) of the respondents reported that they were quite familiar with the ASA guidelines, 45% (60/132) were fairly familiar, 39% (52/132) were a little familiar, and 11% (14/132) were not at all familiar.

Responses of the subjects to questions of their current practice are shown in Figure 1. There was no difference between those who were exposed to an academic presentation of perioperative DNR orders in the past year and those who were not.

Figure 1.
Figure 1.:
Survey responses of all subjects (n = 132) when asked about their current clinical practices regarding perioperative DNR. Data indicate that a significant number of anesthesiologists do not follow the American Society of Anesthesiologists guidelines.

Participants were surveyed as to how they would have responded to the events of the scenario. Of the surveyed 132 participants, 100% (132/132) responded that they would place a chest tube, 94% (124/132) would intubate the trachea, 94% (124/132) would perform chest compressions, 93% (123/132) would perform defibrillation, and 27% (36/132) would take the patient to the cardiac catheterization laboratory to place an intraortic balloon pump. Figure 2 shows the reasons for performing any given intervention. The most common reasons for intervening, regardless of type of intervention, was that the situation was considered reversible, that the patient was likely to return to baseline, that the situation was iatrogenic, and that intervention would be consistent with the patient's goals. Reasons of concerns about legal ramifications, consistency with the anesthesiologist's values, and concern that the family may have a different opinion about the DNR order were significantly different than the other reasons (P > 0.05).

Figure 2.
Figure 2.:
Reasons subjects chose to perform the five interventions of insertion of chest tube, tracheal intubation, chest compressions, defibrillation, and insertion of intraaortic balloon pump. The consistency in the reasoning suggests that participants view their obligation to Mr. Samuels in an all-or-nothing manner.


We report the results of a simulation of perioperative DNR orders. Our results indicate that even though these guidelines have been in place for over 10 years, only half of anesthesiologists would describe themselves as at least fairly familiar with the guidelines and few follow them correctly. These data do not differ from survey data obtained over the years, which indicate that physicians are often not aware of the issues surrounding perioperative resuscitation orders and that many physicians will not honor a patient's refusal of care.8,9

In the simulation, inadequacies in reevaluating the DNR order for the perioperative periods existed at all stages. Some incorrectly assumed the DNR order to be withdrawn during the perioperative period. Some incorrectly thought that they had reached an agreement to suspend the DNR order. Others either misunderstood or disregarded Jim Samuels's preferences. This, too, is consistent with the literature. Patient preferences to limit resuscitation may not be actualized because physicians seem more capable of understanding patient preferences to receive resuscitation than preferences to refuse resuscitation.10 Clinicians also overestimate the success of resuscitation.11–13

Determining appropriate therapy during escalating interventions highlights the intricacies of using goal-directed DNR orders. The increasing uncertainty of the likelihood of a successful intervention increases the subjectivity of what may be considered a “temporary and reversible” condition. For example, we acknowledge the subjectivity of using CPR for ventricular fibrillation, because a rapid return of a stable rhythm may meet Mr. Samuels's burden-to-benefit requirements. On the other hand, although insertion of a balloon pump followed by angiography and possible stent placement may resolve the ischemia, hypotension and ectopy, we believe that these interventions misunderstand the concept of “temporary and reversible.” Because a functioning stent may not resolve the damages related to the cardiac event, the likelihood of survival to the quality of life Mr. Samuels desires is too unlikely to be worth the burdens from Mr. Samuels's perspective and therefore should be eschewed.

It could be argued that arterial stenting should be given a trial of therapy. But that brings forth greater complexity, because Mr. Samuels would be managed by individuals who did not speak directly with Mr. Samuels and were unaware of his preferred burden-to-benefit ratio. These clinicians may be inclined to continue with therapy that would be appropriate for a patient willing to tolerate a greater burden-to-benefit ratio. This is exactly what Mr. Samuels wanted to avoid.

There was a stunning consistency in the reasoning of the participants, even for different interventions. The similar reasoning for performing interventions with a higher probability of success (eg, chest tube) versus those known to have a lower probability of success (eg, defibrillate) may be due to the aforementioned overestimation of the success of CPR. Although there were differences in whether participants thought that different interventions would be successful, this difference did not translate into any significant changes in action. This suggests that individuals view their obligations to their patients grossly (in a “go versus no go” way) and do not appreciate potential shades of gray. We were also concerned about the number of anesthesiologists who based their decision on the iatrogenicity of the event. Being influenced by iatrogenicity is understandable and not uncommon, but it does not account for the patient's goals.14,15 In other words, if the patient has decided that they do not want to live under certain conditions, say long-term ICU care, then it is irrelevant to the patient how they came to be in that situation. Although it is heartening to see the relatively low percentages of other personal reasons for proceeding, such as fear of legal ramifications or consistency with caregiver values, it is troubling to see these reasons listed at all.

A number of ideas came out of the simulation that would not have been accessible through a standard survey assessment. One is the effect of group interactions on ethical decision making. Although 63% of the groups went forward with an intraaortic balloon pump, fewer than 30% of the respondents said they would have done so if free to make their own decision. One reason brought up in debriefings that may account for this difference is the powerful norm that helping anesthesiologists feel obligated to defer to the primary anesthesiologist regardless of their individual feelings. Another reason may be rooted in group conformity. When in a group, people tend to conform to the group and social norms, in this case the traditional and out of date norm of continuing resuscitation.16 These observations may change how we think about the problem of penetration of perioperative DNR orders or other guidelines.

A second opportunity found by the simulation is the poor quality of the preoperative DNR discussion and the subsequent misunderstanding of Mr. Samuels's position. This finding suggests the need for better education regarding the process of discussing the perioperative resuscitation status. In addition, this finding brings forth hope that inadequate penetration of guidelines and the difficulty honoring patient preferences may be partially due to the lack of knowledge and practice discussing perioperative resuscitation in particular and with difficult conversations in general.

Those choosing to perform similar studies or to build on this work may want to consider using qualitative and quantitative assessments together, in what is known as a mixed methodology approach. Qualitative assessments help researchers improve conceptualization and understanding of the research problems, which may then be used to inform or to complement quantitative methods.17

The generalizability of this study to actual clinical practice in a new situation of perioperative reevaluation of DNR orders is limited by the fact that this study was conducted entirely in simulation. Of note, a self-report completed at the end of the course day by these subjects regarding the realism of the scenarios showed that 83% rated the course “excellent,” 16% rated the course “very good,” 1% rated the course “good,” and no one rated the course lower than good. This is in contrast with the “realism” rating of an earlier crisis resource management course for attending anesthesiologists (N = 303) in which of 54% rated the course “excellent,” 38% rated the course “very good,” 8% rated the course “good,” 1% rated the course “mediocre” and no one rated the course lower than good.

In conclusion, it is clear that perioperative reevaluation of DNR orders is not firmly established in anesthesia practice. Anesthesiologists would benefit from more education on managing the subjective nature of goal-directed orders during escalation of therapy. Simulation of perioperative DNR orders is a useful way to elicit anesthesiologists' actions in the heat of the moment, which may bring us closer to understanding why anesthesiologists act as they do.


1. American Society of Anesthesiologists. Ethical Guidelines for the Anesthesia Care of Patients with Do Not Resuscitate Orders or Other Directives that Limit Treatment. Park Ridge, IL: House of Delegates, American Society of Anesthesiologists; 1993.
2. Statement of the American College of Surgeons on Advance Directives by Patients: “Do not resuscitate” in the operating room. Am Coll Surg Bull 1994;79:29.
3. Association of PeriOperative Registered Nurses. I: AORN Position Statement: Perioperative Care of Patients with Do-Not-Resuscitate (DNR) Orders. 1995. Available at: Accessed January 2, 2007.
4. American Society of Anesthesiologists. Ethical Guidelines for the Anesthesia Care of Patients with Do Not Resuscitate Orders or Other Directives that Limit Treatment. Amended October 17, 2001. Available at: Accessed January 2, 2007.
5. Truog RD, Waisel DB, Burns JP. DNR in the OR: a goal-directed approach. Anesthesiology 1999;90:289–295.
6. Waisel DB, Burns JP, Johnson JA, et al. Guidelines for perioperative do-not-resuscitate policies. J Clin Anesth 2002;14:467–473.
7. Blum R, Cooper JB, Feinstein D, et al. Sustaining development of crisis resource management training for academic faculty: a new approach to continuing education. Anesth Analg 2003;97:S10.
8. Clemency MV, Thompson NJ. “Do not resuscitate” (DNR) orders and the anesthesiologist: a survey. Anesth Analg 1993;76:394–401.
9. Richter J, Eisemann MR. The compliance of doctors and nurses with do-not-resuscitate orders in Germany and Sweden. Resuscitation 1999;42:203–209.
10. Wenger NS, Phillips RS, Teno JM, et al. Physician understanding of patient resuscitation preferences: insights and clinical implications. J Am Geriatr Soc 2000;48:S44–S51.
11. Pembeci K, Yildirim A, Turan E, et al. Assessment of the success of cardiopulmonary resuscitation attempts performed in a Turkish university hospital. Resuscitation 2006;68:221–229.
12. van Walraven C, Forster AJ, Parish DC, et al. Validation of a clinical decision aid to discontinue in-hospital cardiac arrest resuscitations. JAMA 2001;285:1602–1606.
13. Jones K, Garg M, Bali D, et al. The knowledge and perceptions of medical personnel relating to outcome after cardiac arrest. Resuscitation 2006;69:235–239.
14. Casarett DJ, Ross LF. Overriding a patient's refusal of treatment after an iatrogenic complication. N Engl J Med 1997;336:1908–1910.
15. Casarett DJ, Stocking CB, Siegler M. Would physicians override a do-not-resuscitate order when a cardiac arrest is iatrogenic? J Gen Intern Med 1999;35–38.
16. Cialdini RB, Goldstein NJ. Social influence: compliance and conformity. Annu Rev Psychol 2004;55:591–621.
17. O'Cathain A, Murphy E, Nicholl J. Why, and how, mixed methods research is undertaken in health services research in England: a mixed methods study. BMC Health Serv Res 2007;7:85.



Ethics; DNR; Resuscitation

© 2009 Lippincott Williams & Wilkins, Inc.