The editors and publisher would like to thank Drs. Vinod Malhotra and Patricia Fogarty-Mack for contributing to this chapter in the previous edition of this work. It has served as the foundation for the current chapter.
Clinical anesthesia practice is often labeled as a model for quality and safety in medicine. In 1999, the Institute of Medicine (now the Health and Medicine Division of the National Academies) report, “To Err Is Human: Building a Safer Health System,” specifically identified anesthesia as “an area in which very impressive improvements in safety have been made.” Such attention to a specialty comprising approximately 5% of U.S. physicians highlights the many contributions to overall perioperative quality and safety generated by the specialty of anesthesia. Although actual reductions in anesthesia-specific mortality rates are controversial, ailing patients are anesthetized for more invasive operations than a few decades ago. The principles by which anesthesiologists transformed the inherently dangerous task of reversibly blunting human responses to pain and physical damage and controlling vital life-support functions into a safe and almost routine occurrence should be familiar to all practicing anesthesia providers.
This chapter reviews the history of anesthesia quality and safety, identifies key approaches and strategies that have contributed not only to anesthesia but to other medical specialties, and examines current and future challenges in anesthesia-related quality and safety.
Definitions: Quality Versus Safety
Quality and safety are related terms but are not identical. Safety refers to a lack of harm and focuses on avoiding adverse events. If patient injury is avoided, then the process is safe. In contrast, quality refers to the optimal performance of a task, which may refer to outcome, efficiency, cost, satisfaction, or some other metric of performance in addition to avoiding injury.
It is easy to see how quality and safety do not always overlap. As an example, a process can in principle always be made somewhat safer by installing an additional check or adding extra equipment. Taken to its extreme, it can be argued that an anesthesia provider is not fully safe unless a fiberoptic scope is in the operating room for induction of anesthesia. Another example is concluding that safety could be improved by having a second (or third) anesthesia provider in the room as well! Clearly, such an approach could incrementally create more safety but would not necessarily produce more quality. In contrast, quality includes an “optimization” element, so if a process is changed to produce better patient satisfaction, for example, or a shorter length of stay, it represents higher quality but not necessarily better safety.
In the anesthesia realm, the use of ultrasound to place central lines is an example of a strategy that improves both quality and safety. By reducing the incidence of carotid puncture, ultrasound clearly improves safety. By reducing the time to successful insertion (and the number of misses), ultrasound improves quality as well. In contrast, pin indexing backup oxygen tanks adds safety, but does not really change quality.
Historically, advances in anesthesia performance have addressed both quality and safety as described in this chapter.
Specific Approaches to Anesthesia Safety
Learning From Experience
Because the mechanisms by which most anesthetics exert their effects are not fully understood, and because many intraoperative states (one-lung ventilation, muscle relaxation, cardiopulmonary bypass) are not found in normal human activity, a large component of anesthesia safety is derived from a history of empiric observation and experience. Driven by the goal of minimizing anesthesia-specific fatality and the shockingly high mortality rate during the early years of anesthesia practice, anesthesiologists have over time systematically accumulated an experience base of observations about safety. Emery A. Rovenstine’s case series of nine cardiac arrests, published in 1951, is an example of this empiric approach. Although he offered no definitive solution, his practical observations (e.g., cardiac massage through the diaphragm is ineffective, the differential diagnosis of shock versus cardiac standstill can be difficult) allowed anesthesia providers to incrementally and empirically improve anesthesia safety.
Beecher and Todd’s exhaustive 1954 study of anesthesia-associated deaths in 10 centers over 4 years stands as a prime example of the empiric approach to anesthesia safety. Involving 21 physicians and 11 secretaries over 5 years, Beecher tracked the outcomes of 599,548 anesthetics, identified 7977 deaths (more than 1 in 100) and cataloged the causes as from patient disease, surgical error, or anesthesia. Their observation that patients who had received neuromuscular blocking drugs had a significantly higher perioperative morbidity rate is still a subject for anesthesia trainees today.
Other examples of empirically derived anesthesia safety observations include the surprising difficulty in detecting esophageal intubation (or arterial desaturation), the tendency of some anesthetics (e.g., desflurane) to trigger hypertensive tachycardic responses, the dangers of circuit disconnection, and the potential for delivery of a hypoxic gas mixture. In all, the anesthesia approach has been to identify and describe such events, determine how they might occur in clinical practice, develop and test countermeasures, and disseminate the results through technical improvements or education. Although most of these anesthesia-related adverse events are by now rare in occurrence, they highlight a key approach: recognize a potentially preventable event, evaluate its likelihood, and systematically develop countermeasures to reduce the incidence. Taken together, observations such as these have led to reductions in anesthesia-related mortality rates, with current estimates ranging from 1:250,000 for healthy patients to 1:1500 for those with complex medical problems.
In addition to empiric observations about patient-related safety, anesthesiologists have addressed safety issues related to provider performance. An everyday example is in the interface between the anesthesia provider and anesthesia delivery system (also see Chapter 15 ). As in aviation, the human–anesthesia machine interface has been designed specifically to reduce inadvertent errors. In the same way that levers in an airplane for landing gear and flap control have a knob shaped like a wheel and a flap, for example, so is the knob on an anesthesia machine for oxygen gas flow shaped differently from knobs controlling air and nitrous oxide, and it is always located on the right. Similarly, the potentially dangerous delivery of hypoxic gas mixtures is prevented by “linking” the oxygen flow to the nitrous oxide flow so that oxygen is always present in fresh gas flow. Nonuniversal connectors to ensure that oxygen is being delivered through the oxygen flowmeter, and an oxygen analyzer to serve as a final check on the delivered gas mixture are other examples of safety mechanisms designed to avoid the inadvertent delivery of a hypoxic gas mixture.
Even though adverse events due to failure of mechanical ventilation or hypoxic gas delivery have almost been eradicated in anesthesia, this process of empiric observation continues today. Recent awareness of the dangers of anemia during spine surgery (also see Chapter 32 ), hypotension in the sitting position (also see Chapter 19 ), or the role of fibrinogen in coagulopathy during maternal hemorrhage (also see Chapter 33 ) are current examples of issues identified through empiric observation.
Adoption of Specialty-Wide Standards
Because anesthesia is normally administered in conjunction with therapeutic or diagnostic procedures, identifying adverse outcomes attributable specifically to the anesthesia practice is challenging. In fact, one of Beecher and Todd’s explicit goals in their landmark study was to define “the extent of the responsibility which must be borne by anesthesia for failure in the care of the surgical patient.” Because adverse events clearly attributed to anesthesia are rare, promulgating appropriate countermeasures across the specialty is difficult. Nevertheless, anesthesia was the first medical specialty to embrace universally applicable standards, developing and promulgating a set of monitoring recommendations with the goal of reducing anesthesia-related adverse events. Driven in part by high malpractice awards, these standards included continuous anesthesiologist presence and vital sign monitoring including blood pressure, heart rate, electrocardiogram, breathing system oxygen concentration, and temperature and were initially published as a research article from a single health care consortium and developed from a database of adverse events.
Although not evidence-based, these standards were incorporated as intraoperative monitoring standards by the American Society of Anesthesiologists (ASA) 2 months later and have remained as one of only three practice standards endorsed by the ASA (the other two being standards for pre- and postoperative care). Since their adoption, conclusive evidence for the efficacy of these standards has remained elusive, but retrospective observations have suggested benefit. In a follow-up study, the authors of the monitoring standards published a case series of 11 major intraoperative accidents attributable to anesthesia from 1976 to 1988, but found that only one occurred after universal adoption of the monitoring standards. Observations from the ASA Closed Claims Project database also suggest a reduction in the number of claims for death or permanent brain damage during that period.
Whether monitoring standards or (possibly) new technologies were responsible for a perceived reduction in adverse events, the willingness of anesthesia providers as a group to adopt practice standards remains an approach almost unique to anesthesiologists and a marker for the priority anesthesiologists put on safety.
Patient Safety–Focused Programs
A third element characteristic of the anesthesia approach to patient safety is the formation of patient safety–focused specialty entities. Existing only for the promulgation of safety, these societies represent an important aspect of the anesthesia approach to patient safety.
Foremost among these groups is the Anesthesia Patient Safety Foundation (APSF), an independent nonprofit corporation begun in 1985 with the vision “that no patient shall be harmed by anesthesia.” Supported by the ASA and corporate sponsors, APSF members include anesthesiologists, nurse anesthetists, manufacturers of equipment and drugs, engineers, and insurers.
The clinical impact of the APSF has been immense. The APSF newsletter, published four times a year, has become one of the most widely circulated anesthesia publications in the world and is dedicated solely to safety. Identifying aspects of anesthesia practice with significant potential for adverse consequences, the APSF newsletter has highlighted diverse issues such as the anesthesia machine checkout, opioid-induced respiratory depression, residual neuromuscular blockade, postoperative visual loss, and emergency manual use. Instructional videos, research grants, and other special conferences are also part of the APSF effort to promote safety.
A second entity with a unique approach to safety is the ASA Closed Claims Project. Operating in cooperation with malpractice lawyers, the Closed Claims Project group reviews data from settled anesthesia lawsuits to identify anesthesia safety concerns that may be amenable to targeted efforts. In a series of academic publications since 1988 and continuing into the present, the Closed Claims Project has investigated a wide range of topics ( Table 48.1 ) focusing on rare events difficult to study systematically. Although such analyses cannot estimate incidences or risk factors, they provide a wealth of descriptive information that has helped anesthesiologists address patient safety issues. Among these are the recognition that listening to the chest may not be a reliable method of detecting esophageal intubation and that a common factor in adverse outcomes due to massive hemorrhage is late recognition.
|Year||Title||No. of Claims||Notable Finding(s)|
|1988||Cardiac arrest during spinal anesthesia||14||Bradycardia was the most common presenting symptom with hypotension as the second. |
Epinephrine was not given until 8 minutes (mean) after onset of asystole.
|1990||Adverse respiratory events in anesthesia||522||Death/brain damage occurred in 85% of cases. |
In 48% of esophageal intubations, auscultation of breath sounds was performed and documented.
|1999||Nerve injury associated with anesthesia||670||Ulnar nerve injuries were most frequent, were associated with general anesthesia, and occurred predominantly in men.|
|2006||Injury associated with monitored anesthesia care||121||Monitored anesthesia care claims involved older and sicker patients than general anesthesia claims. |
Respiratory depression due to sedative/opioid administration was the most common mechanism of damage (21%).
The combination of electrocautery and oxygen was a recognized mechanism in 17%.
|2014||Massive hemorrhage||3211||30% of claims involved obstetrics, and thoracic/lumbar spine procedures were also overrepresented. |
Recognition and initiation of transfusion therapy were commonly delayed.
|2015||Postoperative opioid-induced respiratory depression||357||88% of events occurred within 24 hours of surgery, and somnolence was noted in 62% before the event.|
The Anesthesia Quality Institute (AQI) is the newest and potentially largest patient safety project sponsored by organized anesthesia. Begun in 2008, the goal of the AQI was to “to be the primary source of information for quality improvement in the clinical practice of anesthesiology.” Sponsored by the ASA, the AQI administers and supports an Anesthesia Incident Reporting System (AIRS) and the National Anesthesia Clinical Outcomes Registry (NACOR), which currently captures information on approximately 25% of all the anesthetics administered in the United States. The goal is to capture enough anesthetic data that accurate benchmarking of clinical outcomes related to anesthesia can be performed and informed efforts to improve quality can occur.
From Safety to Quality: Making Anesthesia Both Safer and Better
Although most observers believe anesthesia care to be safer today than 50 years ago, whether the quality of anesthesia care has improved is less clear. Incorporating not only safety but efficiency, cost, and patient comfort and satisfaction, anesthesia quality has many more dimensions than the avoidance of adverse outcomes.
Several barriers exist to measuring and improving anesthesia quality. Because the relative contribution of anesthesia to the outcome of surgical procedures is difficult to define, identifying how anesthetic care might have made a difference is likewise challenging. It is easy to see that if a patient goes home a day sooner after a colectomy, for example, determining whether that improvement is due to anesthesia, surgery, or hospital care is extremely difficult. More than likely, this type of improvement is a result of all variables.
The most significant obstacle to anesthesia quality is knowledge of patient outcomes. Because most of the patient’s pre- and postoperative course lies outside the preoperative clinic, operating room, and postanesthesia care unit, understanding how a patient’s clinical course is affected by alterations in anesthesia care requires considerable effort to follow patients into the postoperative phase. For this reason, early attempts to improve anesthesia quality focused on perioperative processes rather than outcomes. The Surgical Care Improvement Project, or SCIP, was a national test of this approach. By incentivizing the public reporting of hospital performance on evidence-based process measures such as administering antibiotics in a timely fashion and verifying the continuation of preoperative β-adrenergic blockers into the perioperative period, policymakers hoped to improve quality by improving perioperative processes of care. Puzzlingly, however, over the 8-year history of the SCIP project (2006-2014), performance on nearly all process measures included in the project improved, but outcomes (whether surgical site infections or mortality rate ) failed to improve. In fact, because of concern regarding adverse outcomes from potentially harmful process measures, several related process measures were also rescinded. Among these were whether β-adrenergic blockers were given to patients within 24 hours of an admission for myocardial infarction and verifying that antibiotics were given within 4 hours of an emergency room visit for pneumonia.
Why implementation of a suite of process measures, all with literature support, have not clearly improved patient outcomes remains a mystery. Clearly, improving quality by mandating specific processes of care is not straightforward and has led quality experts to be much more reluctant to embrace process measures alone as a method of assessing care quality.
Measuring structural elements can also provide a glimpse into the presence or absence of quality. Structure refers to the presence or absence of specific organizational features that are considered to be integral to the provision of high-quality care. If present, such features then suggest that the clinical care is of high quality.
Examples of structural elements considered to correlate with quality care include the ready availability of diagnostic radiologic testing, having physicians on call for emergencies, an electronic medical record, and mandating a dedicated intensivist for all critical care units. The presence of an active quality improvement mechanism might also be considered a structural feature of high-quality care. Although structural quality is relatively invisible to trainees unfamiliar with diversity in health care environments, hospital rules governing nurse-patient ratios, timely availability of obstetric anesthesia specialists, and a protocol for hand hygiene are examples.
Although structural measures are generally easy to measure, the link between structure and improved outcomes is often difficult to discern. The availability of in-house critical care attending physicians at night, for example, is intuitively reasonable, and would be a relatively easy structural feature to measure. However, more than one study suggests that hospitals that have implemented an in-house night-call system might not see clear improvements in outcomes.
One logical consequence of an inability to identify clinically relevant process measures is to focus instead on outcome. Because there is considerable practice variability in anesthesia, variability in outcomes likely exists. In principle, by identifying “bright spot” institutions that have better outcomes, the corresponding best practices can be identified and disseminated. Although NACOR is not yet mature enough to allow outcome analysis, surgical databases are approaching that goal. The Society of Thoracic Surgeons (STS) Adult Cardiac Surgery Database is perhaps the best example, capturing data from more than 90% of all cardiac procedures in the United States. Other databases include the National Surgical Quality Improvement Program (NSQIP) and National Inpatient Sample (NIS). Because sufficiently complete data for outcome reporting has historically not been available, few hospitals have routinely made outcome data available to their clinical care staff. In addition, outcome reporting for anesthesiologists in particular is challenging because events that occur postoperatively may not be related to anesthesia care per se. Such an approach is changing, however, as hospitals recognize the value of feedback. Monthly central line and catheter-related urinary tract infection rates posted in the intensive care unit or patient satisfaction scores posted in the operating room are examples.
Outcome reporting initially seems straightforward and gives individuals or institutions a benchmark for measuring future performance. But accurately comparing outcomes between individuals or institutions requires some way to adjust for patient conditions unrelated to the anesthesia or surgical (or hospital) care. This “risk adjustment” can be extremely difficult as different adjustment algorithms may produce different results, algorithms may be vulnerable to “gaming” by inducing favorable patient selection, the accuracy of data may be suspect, and the adjustment algorithm itself may not be consistent from year to year.
Current evidence is mixed with regard to whether outcome reporting improves outcomes. Two 2015 studies suggest that knowing one’s outcomes may not by itself drive improvement. In addition, should a “bright spot” institution with unusually good outcomes be recognized, identifying and disseminating lessons from that institution would likely involve developing a set of process measures, which (as the SCIP program demonstrates) may not have the desired effect.
Nevertheless, the use of both process and outcome measures are key to quality improvement. As the management consultant Peter Drucker once noted, “You can’t manage what you can’t measure.” Yet measurement alone is inadequate. Our current experience with both process and outcome measurement is that neither readily leads to improved quality. Further work is needed to better understand how to use outcome and process measurements to drive quality.