The Normalization of Deviance

Abstract

Anesthesia professionals recognize that perioperative care of both routine and seriously ill patients has the potential for devastating patient injury. The Institute of Medicine1 and other reports suggest that many patients experience complications during the course of high-intensity care in the nation's >5000 acute care hospitals. The last decade has witnessed multiple campaigns, efforts, and rhetoric aimed to improve the understanding of error and focus individuals and systems on procedures to improve safety outcomes. The last decade has also seen increased emphasis on economic productivity, driven in part by concern about endlessly increasing health care costs. Is the approach of “doing more with less” in the operating room (OR) any different from the “faster, cheaper, better” mantra of the National Aeronautics and Space Administration (NASA)? It is appropriate to ask whether economic pressure has affected patient safety.2 More bluntly, are anesthesia professionals at risk of cascading down the slippery slope of faster rather than better? LESSONS FROM SPACE Theories of organizational safety, such as Normal Accidents Theory, describe the complexity of a system and the tight coupling of interactions between its subsystems. Both the Challenger and Columbia space shuttle accident investigation boards applied Normal Accidents Theory analysis to their investigations and asked similar questions. Why did NASA continue to fly the Challenger shuttle while O-ring erosion problems were documented numerous times before the cold January 1986 launch? Why did NASA continue to fly the Columbia shuttle despite knowing foam insulation was regularly striking vulnerable areas of the vehicle years before Columbia's fatal accident? One explanation is that these mishaps had been “normalized” over many occurrences and many years until managers and engineers began to believe that these flaws were expected and therefore acceptable. Diane Vaughan,3 in her exhaustive book, The Challenger Launch Decision, coined this behavior the “normalization of deviance.” This incremental process is a gradual erosion of normal procedures that would never be tolerated if proposed in 1 single, abrupt leap. Instead, small incremental deviations are observed and tolerated. Lacking an accident, they become “normalized.”4 When the shuttle was originally designed, no allowance was made for the possibility that the Challenger would be launched in subfreezing temperatures, leaving rocket booster O-rings to contract, weaken, and leak. Nor was it ever anticipated that insulating foam debris could fall off the main tank and strike the vehicle's wing. When these events were first experienced, the obvious safety implications were recognized. However, faulty analyses concluded that the vehicle could tolerate these abnormal events. Managers and engineers decided to either implement a temporary fix or simply accept the risk. This approach established a precedent for accepting safety violations as technical deviations that can be tolerated and managed.4 As the problems recurred and the shuttle kept flying, the fallacy that the errors were acceptable was reinforced. Thus, foam strikes were no longer even defined as safety violations. They had become “normalized”: considered a normal part of a shuttle liftoff.4 Normalization of deviance breaks the safety culture, substituting a slippery slope of tolerating more and more errors and accepting more and more risk, always in the interest of efficiency and on-time schedules. This toxic thinking often ends with a mindset that demands evidence that these errors would destroy the vehicle (or harm a patient), instead of demanding proof that the shuttle (or patient) is safe and not being harmed. The boundaries are soon pushed to extremes without understanding where and why the original limits were established.4 LESSONS FROM THE OR: IS THERE A COLUMBIA/CHALLENGER DISASTER IN OUR FUTURE? Anesthesiology shares a core value with space flight and aviation: safety. We in health care can pretend that what happened at NASA cannot possibly happen to us. As of 2003, space shuttles had flown 112 missions. Two of those missions ended catastrophically, a failure rate of 1.7%. Analysis of those 2 spectacular failures demonstrates the same normalization of deviance that we believe is incrementally undermining perioperative safety. Surely, the stakes for anesthesia and perioperative care are just as high as for NASA. Although our failures are not as spectacular as shuttle explosions, they are much more frequent and equally devastating to patients and their families. Therefore, we must avoid the temptation to normalize deviations from safety standards, such as the frequently observed behaviors noted in Table 1, while we acknowledge that correction or avoidance of routine equipment failures or human errors may be insufficient to avert complex mishaps and systemic failures.5 Nevertheless, we advocate an attitude change that fundamentally opposes normalization of deviance. From 4 different institutions, academic and private practice, and from diverse regions around the country, we observe common practices that represent erosions of patient safety in the interest of efficiency. Three clinical vignettes illustrate our concerns.Table 1: Anesthesia Practices that Should Not Be “Normalized”Example 1 Production pressure has promoted the practice whereby some, or all, standard anesthesia monitors are discontinued before the end of general endotracheal anesthesia. A recent Anesthesia Patient Safety Foundation Internet poll (http://www.APSF.org; Fig. 1) suggests that 1 anesthesia professional in 8 will remove blood pressure and/or electrocardiogram (ECG) cables before the patient's emergence from anesthesia and tracheal extubation. A significant number, approximately 1 in 13 respondents, will also remove the pulse oximetry oxygen saturation (SpO2) probe. Some practitioners believe this practice speeds OR turnover or contributes to “efficiency” in some way. We disagree. Nothing in the American Society of Anesthesiologists' (ASA) or the American Association of Nurse Anesthetists' guidelines supports such practice. Moreover, this maneuver removes vital monitors from patients during a period of intense physiological stress—the transition from surgical planes of anesthesia to return of consciousness with spontaneous ventilation, stable hemodynamics, and patient cooperation. As a representative anecdote, we recently observed a patient's cardiac rhythm suddenly convert from normal sinus rhythm to a perfusing monomorphic ventricular tachycardia during this period of emergence. This life-threatening arrhythmia was unexpected and unheralded. But for direct observation of the ECG during this vulnerable transition period, a key diagnosis would have been missed and vital treatment delayed, exposing the patient to further rhythm deterioration and risk of significant morbidity or death. A pulse oximeter alone would have failed to capture this diagnosis, because the perfusing ventricular tachycardia produced no immediate interruption in the oximetry signal or display.Figure 1: Columns represent percent responses to the Anesthesia Patient Safety Foundation Web poll that queried nurse anesthetists and anesthesiologists about removal of routine monitors at the end of surgery but before anesthesia emergence and tracheal extubation. One in 8 respondents removed 1 or 2 monitors, whereas 1 in 13 removed the triad of electrocardiogram (ECG), blood pressure (BP), and pulse oximetry standard monitors. SpO2 = pulse oximetry monitor. (Used with permission of Anesthesia Patient Safety Foundation Newsletter editor, and available at: www.apsf.org, 2009.)Whenever clinical decisions are made that contradict or modify current practice, it is imperative that we ask not, “What is the evidence that this will hurt the patient?” but that we ask instead, “What is the evidence that this decision will not lower safety?” In the absence of such evidence, we should not assume the practice to be safe or acceptable. Example 2 The literature is replete with studies, surveys, commentaries, letters, and editorials that document the continued occurrence of perioperative residual neuromuscular blockade.6,7 Failure to recognize residual muscle weakness can be attributed to many interacting factors but ultimately can be traced to either a failure to monitor neuromuscular blockade or a lack of understanding in neuromuscular pharmacology.8 Current practice involves the use of modern, intermediate-acting neuromuscular blocking drugs with minimal potential for accumulation. In many instances, induction of general anesthesia proceeds without prior documentation of baseline neuromuscular responses, or, frequently, stimulating electrodes are never placed on the patient at all. A recent case illustrates the multiple failures along the clinical continuum. On surgical closure after upper abdominal surgery, an elderly patient received full-dose pharmacologic reversal and then his trachea was extubated after documentation of “train-of-four (TOF) recovery without fade.” Although chart documentation may at first seem appropriate, the patient was actually received in the postanesthesia care unit with minimal respiratory effort, low saturation (SpO2 <88%), and hyperdynamic circulation indicative of hypercarbia. Reassessment of TOF at the ulnar nerve by objective monitoring (acceleromyography) documented a TOF of 0.36. On further investigation, it was noted that the intraoperative assessment consisted of visual assessment of TOF of the face muscles (jaw) and confirmation of “sustained tetanus” (delivered for approximately 2 seconds) through electrodes placed on the patient's temple and forehead. Although technically the monitoring and assessment criteria fulfilled the medicolegal requirements, it is clear that multiple failures occurred: from reliance on subjective (visual) assessment, to lack of baseline response documentation, to inappropriate application of tetanic stimulation (for <5 seconds), to improper location of stimulating electrodes that induced direct muscle stimulation and provided a false index of recovery. Unfortunately, this clinical scenario is not uncommon.7 It is likely that the transition from the “older generation” long-acting neuromuscular blocking drugs (e.g., curare, pancuronium, doxacurium, and pipecuronium) to the newer, intermediate-acting drugs that have much lower potential for residual paralysis has made us think that postoperative residual paralysis was no longer possible and has contributed to the current “normalization of deviance” from the basics of neuromuscular monitoring. We must be proactive and realize that the impetus to learn the nuances of neuromuscular monitoring will continue to decrease, unless we stop accepting the deviance. As production pressure increases in proportion to the decrease in reimbursement, and as new, drug-specific antagonists of neuromuscular blocking drugs reach the market (e.g., sugammadex and cysteine),9 the perceived need for appropriate, timely, and continual monitoring of neuromuscular function will likely continue to decrease. Are we strong and perceptive enough to resist cascading down the slippery slope of “faster” rather than “better and safer?” Example 3 Improved understanding of the pharmacology of local anesthetics has contributed to a steady decrease in complications caused by local anesthetic toxicity during performance of peripheral nerve blocks.10 In addition, the use of real-time ultrasound to visualize needle location in proximity to neural and vascular structures, collateral anatomy, and the ability to visualize perineural spread of local anesthetic provide additional information to minimize potential complications. However, ECG rate, rhythm, and morphology, arterial blood pressure, and arterial oxygen saturation are sensitive indicators of local anesthetic toxicity.11 Thus, it is concerning to see our profession “normalize” the practice of performing peripheral nerve blocks using only a single monitor, usually a pulse oximeter. Indeed, a survey performed by Corcoran et al.12 revealed that monitoring practices varied at academic anesthesiology departments. There was no relationship between the monitoring technique and the number of peripheral nerve blocks performed per month or the level of training of the provider performing the block. The authors reported that 16% of the studied anesthesia departments used only a pulse oximeter.12 This scenario parallels the perception that monitoring “shortcuts,” eerily similar to our first example, are acceptable. We disagree. By way of example, we recently observed an elderly patient who became confused and lethargic immediately before cardiac asystole ensued during injection of local anesthetic for an interscalene plexus block in the perioperative holding area. Prompt detection of her (potentially) fatal arrhythmia led to immediate intervention, initiation of advanced cardiac life support, and a good outcome. Thus, we believe the safer practice, consistent with Anesthesia Patient Safety Foundation philosophy, is to place all standard ASA-recommended monitors on the patient before the injection of large volumes of local anesthetic for peripheral nerve block. Individual practitioners may claim impressive records in hundreds of patients. However, we know from the rule of 3 that when the numerator is 0, the real incidence may be as great as 3 over N. So, the “perfect practitioner” with 300 blocks without major complication may have a rate of major complications of 3 of 300 or as high as 1%13! Thus, a long track record of deviations from accepted practice without adverse consequence is not adequate evidence to normalize the deviation. In the absence of large studies documenting that monitoring can be relaxed during placement of peripheral nerve blocks, we advocate adherence to ASA monitoring standards to maximize patient safety. SUMMARY There are many elements that contribute to errors within an industry or profession. Several human factors associated with safety breakdowns are outlined in Table 2. Experience and root-cause analyses usually document that 2 or more of these factors coalesce to form a “perfect storm” leading to a mishap. For example, expecting a fatigued provider to care for an emergency patient with concurrent production pressure to maintain the elective schedule, while using new and unfamiliar equipment, is a potent mix of risk factors. As Gaba et al.14 pointed out, production pressure “is a reality for many anesthesiologists and is perceived in some cases to have resulted in unsafe actions.” One solution is to integrate standard protocols and expectations for safe practice and expected behavior throughout the practice. Other potential solutions may involve the design of better and “smarter” monitors that will reduce the noise pollution and attendant distractions in the OR, and variable priority training that helps clinicians focus on “optimal distribution of attention when performing multiple tasks simultaneously with the goal of flexible allocation of attention.”15Table 2: Human Factors Contributing to Anesthesia MishapsWe have also observed the phenomenon of intersecting curves of knowledge versus experience. When we exit our organized training period, our knowledge base is strong. We have studied for specialty examinations, experienced the idealized purity of an academic environment, and have been taught the “right way” to practice by our mentors and role models. As the years pass, our minute, detailed knowledge may decrease, but our practical experience increases greatly, and patient care and safety are assured. However, as we are increasingly challenged to “do more with less,” the temptation will arise to “cut a few corners” where we can to achieve productivity and efficiency benchmarks. To that end, we caution our colleagues to avoid the slippery slope of accepting a decrease in vigilance and safety while striving for “faster, better, cheaper.” We encourage every individual to maintain vigilance, advocate for patient safety, aim for excellence and efficiency, and avoid the temptation of normalizing deviance from accepted safety standards.