Anesthesia Advanced Cardiac Life Support: A Guideline Validated?

The goal of a clinical practice guideline is to distill a systematic review of evidence into recommendations intended to optimize patient care.1 Usually generated by experts in the field of interest, an ideal guideline should offer a systematic, unbiased analysis of the available evidence, a transparent description of how that evidence is translated into patient recommendations, and a metric for capturing the unanimity of guideline authors. Some, like those produced by the American Society of Anesthesiologists, go even further by soliciting opinions of society members and experts not on the writing committee.2 Although the creation of practice guidelines is clearly a difficult task, establishing their validity may be even more difficult. A recent search of the “guidelines.gov” clearinghouse for “breast cancer screening,” for example, produced 56 guidelines generated between 2010 and 2017 and originating in the United States, with 11 published in 2017 alone.3 Do all of these guidelines improve either the quality of breast cancer care or mortality? And to what degree? Systematically answering the above questions may not be possible. Compliance, expense, and hospital, regional, and patient level variations in environment all conspire to bury any outcome benefit in within-group variability. In addition, because guidelines unavoidably must comment on some issues with little evidence, recommendations may not be completely evidence based. Fully 30% of the highly cited 2016 Society of Critical Care Medicine Surviving Sepsis guidelines, for example, are graded “1C,” meaning that the expert panel strongly recommends the intervention “1,” even though the supportive evidence is limited to “C.”4 Nevertheless, application of Surviving Sepsis has already generated outcome improvement,5 even in the absence of fully literature-supported recommendations. The example of Surviving Sepsis suggests that successful guidelines based in part on expert opinion can be feasible and effective. In this issue of Anesthesia & Analgesia, Moitra et al6 and McEvoy et al7 offer a 2-part “narrative review with recommendations” focusing on the pathophysiology, diagnosis, and management of cardiac arrest in the intraoperative period. Their combined efforts are an expansion and update of their 2012 publication on the same topic in the Canadian Journal of Anaesthesia8; and, in addition to new literature, they are informed by user feedback from workshops and refresher courses at national meetings. Part 1 describes the epidemiology, physiological mechanisms, and unique features of intraoperative arrest, and outlines a basic response and rescue sequence for when such an arrest occurs. Unlike most cardiac arrests, intraoperative arrest is immediately witnessed by caregivers who are familiar with the patient, and thus, preexisting comorbidities and the antecedents to arrest are usually known. Moitra et al6 persuasively argue that knowledge of the prearrest period and patient comorbidities can permit a focused, tailored response that is not possible with most in-hospital or unwitnessed arrests. Part 2, authored by McEvoy et al7 and the same writing group, focuses on specific resuscitative protocols relevant to unique arrest scenarios, including anaphylaxis, tension pneumothorax, malignant hyperthermia, traumatic arrest, and local anesthesia toxicity. Clinical descriptions of each are presented, as are suggested treatment paths. The result is comprehensive, and provides a roadmap for anesthesiologists diagnosing and managing intraoperative arrest. But, as with nearly all practice guidelines, and particularly with guidelines focused on cardiac arrest for which solid evidentiary report is lacking, readers might wonder how such recommendations were derived and whether they improve outcomes. With respect to the first question, experienced anesthesiologists will note the lack of high-quality randomized evidence but recognize these treatment pathways as a judicious distillation of physiology, evidence, and clinical expertise, and as codification of already known, empirically derived best practice. After all, cardiac arrest during spinal anesthesia was first described nearly 50 years ago,9 and closed claims analyses of such arrests began 30 years ago.10 Such codification facilitates a consistent, effective response … an important goal of Moitra et al.6 Another article in this issue of A&A may provide some clues to the second question: do these guidelines work? In a retrospective analysis of the National Surgery Quality Improvement Program (NSQIP) database from 2012 to 2013, Kim et al11 describe the landscape of postoperative cardiac arrest and identify potential risk factors for patients who arrest after noncardiac surgery. Kim et al11 observe that, in the NSQIP database, although complications were common before cardiac arrest, they did not affect postarrest survival. Thus, prearrest conditions such as acute kidney injury, acute respiratory failure, myocardial infarction, stroke, or deep vein thrombosis/pulmonary embolism do not appear to, after adjustment for preoperative risk, affect postarrest survival. On one level, Kim et al’s11 results are not surprising. They note that mortality after postoperative cardiac arrest is as high as mortality in hospitalized patients who arrest, and it may be that few preoperative factors meaningfully affect that mortality. But Kim et al11 also include a post hoc analysis of arrests on the day of surgery. Although they could not test their hypothesis with postoperative day (POD)#0 data (because of an inability to determine whether the complication occurred before or after the arrest), their day of surgery arrest data are, in themselves, interesting. Kim et al11 observe that POD#0 had the highest absolute number of cardiac arrests, but their supplemental Figure 3b shows better arrest survival (66% vs 49%) and 30-day survival (47% vs 29%). Although one might imagine that POD#0 arrests occur in healthier patients, Kim et al’s11 adjustment for known confounders and primary finding, that preexisting comorbidities do not affect postarrest survival, suggests not that the patients are less sick, but that arrests and resuscitations occurring on POD#0 differ from those occurring after POD#0. Perhaps the most salient difference between POD#0 arrests and those on subsequent days is the mechanism of arrest and subsequent response. Of the many scenarios developed by McEvoy et al,7 several (anaphylaxis, hyperkalemia, pulmonary embolism, malignant hyperthermia, and hypertensive crisis) occur not just intraoperatively, but in the first few postoperative hours. That these scenarios, unique to the perioperative period, can be managed with a better survival than arrests in hospitalized patients or after POD#0 suggest an internal validity to the algorithms described by McEvoy et al,7 which although not fully evidence based, are empirically derived from clinical experience, closed claims, physiological principles, and pharmacology of anesthetic agents. Taken together, the narrative reviews produced by Moitra et al6 and McEvoy et al7 and the description of postoperative arrest created by Kim et al11 suggest that arrest on POD#0 can be treated effectively, that existing treatment pathways rooted more in experience than evidence can be valid, and that the outcomes generated by such experience-derived narrative reviews can equal those generated by arrest management protocols rooted in formal guidelines. Evidence to support the recommendations of Moitra et al6 and McEvoy et al7 will likely accumulate over the next several years as researchers begin to focus on perioperative arrest in greater detail. But the guiding principles used by anesthesiologists for years to keep patients safe in the operating room are likely to remain intact. DISCLOSURES Name: Avery Tung, MD, FCCM. Contribution: This author wrote the manuscript. Conflicts of Interest: A. Tung receives a salary as executive editor for critical care for Anesthesia & Analgesia. This manuscript was handled by: Richard C. Prielipp, MD.