What does “do not resuscitate” (DNR) mean? Since DNR status first appeared in the New England Journal of Medicine 40 years ago,1 this acronym has been associated with much controversy; more recent articles offer a historical perspective2 but little clarity on the challenges of operationalizing the concept of “DNR.” At first, the meaning seems clear: to forego aggressive resuscitative measures in someone for whom death is imminent. Health care providers offer DNR as the pinnacle of patient autonomy, explicitly conferring the ability to refuse care that may prolong life and allowing them to die in a manner consistent with their wishes and beliefs. However, for some patients and/or family members, DNR may be an indicator of less-than-aggressive care or no care at all,3 causing fear and reluctance to invoke this tool to aid in a dignified death. Concerningly, previous work has shown that doctors and nurses may give less attention to patients with active DNR orders4 or withhold treatments not related to resuscitation.2
Despite the heterogeneity of emotions and beliefs DNR orders provoke—and the corollary that there is no universal definition of acceptable intraoperative practices in DNR patients—patients who choose not to undergo aggressive end-of-life resuscitation measures nonetheless do present for surgical care that is appropriate and even desirable. For example, emergent treatment of hip fracture is the most common non-eye surgery performed in those over 80 years of age5; a single-center study of hip fracture patients in the Netherlands revealed a 45.6% prevalence of DNR orders,6 although the prevalence in the United States is no doubt lower. But this begs the question: are we, as anesthesiologists, complicit in these chilling scenarios—a patient ignored, or appropriate preventive care or treatment for perioperative complications withheld, because of the presence of a DNR order?
In this context—a lack of clarity of the implications of DNR orders in surgical patients, a limited understanding of the outcomes of palliative surgical procedures, and uncertainty over the potential unintended consequences of DNR orders—the article by Walsh et al7 in this issue of Anesthesia & Analgesia is timely and compelling. The authors used the well-validated NSQIP database to study procedures performed in 5629 patients with an active DNR order. They matched patients with active DNR orders to 443,555 patients undergoing the same principal operative Current Procedural Terminology code who did not have a DNR order in place to explore patterns of care, including resuscitative efforts, and evaluate for a “failure-to-rescue” phenomenon in this cohort.
Unsurprisingly, patients with an active DNR order were typically older, were more likely to be functionally dependent, and had high rates of serious comorbid disease. Fully 40% were American Society of Anesthesiologists physical status IV or V, indicating life-threatening systemic disease or a patient not expected to survive 24 hours whether or not surgical care is provided, respectively. Compared with non-DNR patients, the most common procedures in DNR surgical patients were emergent and were performed for immediate relief of intolerable symptoms: eg, hip fracture repair, or surgical palliation, or treatment of a bowel obstruction. Rates of 30-day mortality were high (24.4%), but adjusted rates of perioperative morbidity were not different—arguing against failure-to-rescue, as defined by Walsh et al7—and, furthermore, these patients appropriately underwent fewer aggressive postoperative interventions, presumably consistent with their wishes.
Anesthesiologists should be aware of 3 important implications from this work. First, as far as this secondary data analysis can suggest, we are doing a good job respecting end-of-life preferences in patients with a DNR order. Although the study has no information on intraoperative resuscitation, odds of intubation and cardiopulmonary resuscitation were much lower than would have been expected based on severity of illness alone. The authors found no evidence for failure-to-rescue, with rates of morbidity no different than adjusted rates in the DNR population. Transfusion practices were also similar between the 2 groups, further arguing that DNR was not equivalent to “do-not-treat” and assuaging the very real concern that nonresuscitative interventions may be withheld as an unintended consequence of DNR status.
Second, while most of us would readily accept that patients with an active DNR order are at higher risk of perioperative mortality, the authors’ figure of a 25% 30-day mortality is impressive. NSQIP does not offer reliable documentation of timing (intraoperatively, immediately postoperatively, or perhaps several weeks postoperatively, of causes unrelated to the surgery itself) or mechanism of death. The latter is particularly interesting: what proportion died after withdrawal of life-sustaining therapies, where interventions had to be deliberately terminated as causing harm (or care not within the patient’s stated goals), instead of not offered in the first place (where overall care might have been better aligned with patient goals)? A recent single-center study of vascular patients offers a window into this question: only 14% of vascular surgery patients who died in hospital had an advance directive in place, 73% died with transition to comfort-focused care, and the median time from palliative care consultation to death was only 10 hours.8 We might hypothesize that patients with an active DNR order would be less likely to find themselves among this high-intervention, poor-prognosis group, but this study does provide a pessimistic window into the realities of end-of-life care for some surgical patients.
Finally, although the authors enriched their sample by restricting to certain Current Procedural Terminology codes, 1.3% of their cohort had an active DNR order in place at the time of surgery. In the absence of available data on the overall prevalence of DNR orders in surgical patients, this suggests that up to 250 patients with active DNR orders may present for surgery at a large hospital with a volume of 20,000 anesthetic cases annually. Anesthesiologists must be comfortable with this population, which presents unique challenges to our field: namely, that a key component of our care (endotracheal intubation) may be against the patient’s wishes in another context, and any of the extensive palette of life-saving interventions that we might routinely offer could be construed as not within the patient’s goals of care.
The American Society of Anesthesiologists offers guidance on this point,9 specifically advising that automatic perioperative suspension of DNR orders is inappropriately paternalistic. Instead, 3 potential strategies are suggested, with the ultimate authority over the options belonging to the patient and surrogates: full resuscitation, resuscitation with limitation of certain procedures (eg, chest compressions and defibrillation), or resuscitation that is “defined with regard to the patient’s goals and values,” leaving the specific determination up to the anesthesiologist and procedural team. The last option requires a careful exploration of the patient’s preferences, and a deep rapport must be established before any interventions are undertaken. It is one of the most explicit expressions of the fiduciary relationship between a physician and patient, yet perioperative anesthesia practice today is seldom structured to allow the time or longitudinal relationship that would facilitate development of the necessary rapport. Another challenge in developing this rapport is the fact that most anesthesiologists receive little or no training in initiating sensitive conversations about end-of-life options—and the difficulty may be compounded when language, class, or racial/ethnic differences exist.10–12 Hospital-based palliative care programs are available in 90% of large hospitals and in 67% of all hospitals with 50 or more beds13; intensivists and geriatricians may also be experienced in navigating patient preferences at end of life and may be an important resource. If these teams are already involved in a patient’s care, an effort to involve the intraoperative anesthesiologist when a surgery is planned may help clarify the anesthetic care decisions for the patient and entire care team.
As with all secondary data analyses, there are important limitations to this work, and unanswered questions remain. Because NSQIP does not uniformly report time to event for morbidity and mortality events, the authors were unable to perform competing-risks analysis, which does leave open the question of whether DNR patients died before they could undergo increased morbidity; a different dataset would be required to provide a more definitive answer here. NSQIP also does not collect data on metrics that may be as important as, or even more than, mortality and morbidity in this population: pain control, sleep quality, and the perceived degree to which care benefited the patient (according to patient-defined criteria of benefit). DNR patients were hospitalized twice as long as controls and were between 2.5 and 5 times more likely to be discharged to a care facility; data on an important counterfactual population (ie, DNR patients who chose not to undergo surgery) are not available in NSQIP, but this suggests that DNR patients spend considerably less time at home in the first 30 postoperative days, which may also conflict with end-of-life goals. This analysis cannot be used to determine which surgeries, what hospital care, or what anesthesia care is appropriate in a patient with end-stage disease—an issue which will become increasingly pressing as advanced interventions with palliative and life-prolonging properties, such as transcatheter aortic valve implantation, are refined and become more widespread. For procedures that have a particularly high prevalence in DNR patients, efforts to collect rigorous data about the perioperative experience—on how clinical care decisions are affected, on typical outcomes like mortality, and on patient-centered outcomes focusing on quality of life—should be a priority.
A DNR order declares a choice to limit actions that cause harm in the face of an inevitable, and perhaps rapidly approaching, death. With the institution of a DNR order comes a reminder to explicitly consider whether any major medical intervention would offer only an increased burden of suffering without meaningful likelihood of improving quality of life (a consideration which should be extended to all patients). While mortality rates were high, 3 of 4 patients undergoing palliative surgeries in this dataset survived for at least a month postoperatively and did not suffer elevated rates of morbidity or evidence of failure-to-rescue. With this, Walsh et al7 provide illuminating data on the epidemiology and outcomes of surgical procedures in this vulnerable, yet growing, population.
Name: Elizabeth L. Whitlock, MD.
Contribution: This author helped plan, write, and edit the manuscript.
Name: Rondall K. Lane, MD.
Contribution: This author helped plan, write, and edit the manuscript.
This manuscript was handled by: Thomas R. Vetter, MD, MPH.
1. Rabkin MT, Gillerman G, Rice NR. Orders not to resuscitate. N Engl J Med. 1976;295:364–366.
2. Burns JP, Truog RD. The DNR order after 40 years. N Engl J Med. 2016;375:504–506.
3. Beach MC, Morrison RS. The effect of do-not-resuscitate orders on physician decision-making. J Am Geriatr Soc. 2002;50:2057–2061.
4. Sherman DA, Branum K. Critical care nurses’ perceptions of appropriate care of the patient with orders not to resuscitate. Heart Lung. 1995;24:321–329.
5. Deiner S, Westlake B, Dutton RP. Patterns of surgical care and complications in elderly adults. J Am Geriatr Soc. 2014;62:829–835.
6. Simons AE, Karres J, Nijland LM, Ultee JM, Kerkhoffs GM, Vrouenraets BC. Do-not-resuscitate orders and early mortality in hip fracture patients. Age Ageing. 2017 February 5 [Epub ahead of print].
7. Walsh EC, Brovman EY, Bader AM, Urman RD. Do-not-resuscitate status is associated with increased mortality but not morbidity. Anesth Analg. 2017;125:1484–1493.
8. Wilson DG, Harris SK, Peck H, et al. Patterns of care in hospitalized vascular surgery patients at end of life. JAMA Surg. 2017;152:183–190.
10. Barnato AE, Anthony DL, Skinner J, Gallagher PM, Fisher ES. Racial and ethnic differences in preferences for end-of-life treatment. J Gen Intern Med. 2009;24:695–701.
11. Erickson SE, Vasilevskis EE, Kuzniewicz MW, et al. The effect of race and ethnicity on outcomes among patients in the intensive care unit: a comprehensive study involving socioeconomic status and resuscitation preferences. Crit Care Med. 2011;39:429–435.
12. Harrison KL, Adrion ER, Ritchie CS, Sudore RL, Smith AK. Low completion and disparities in advance care planning activities among older Medicare beneficiaries. JAMA Intern Med. 2016;176:1872–1875.
13. Dumanovsky T, Augustin R, Rogers M, Lettang K, Meier DE, Morrison RS. The growth of palliative care in U.S. hospitals: a status report. J Palliat Med. 2016;19:8–15.