Risk of Anesthesia

Types of Studies

To interpret the literature related to anesthetic and perioperative risk, the strengths and limitations of various study designs must be understood. Prospective cohort studies involve the identification of a group of subjects who are monitored over time for the occurrence of an outcome of interest. The goal is to identify patients in whom the outcome develops. For studies of perioperative mortality, individual cases can be reviewed to determine the cause of mortality. Alternatively, data on all patients in the cohort study can be obtained, and discrete factors associated with the development of morbidity or mortality can be determined, often using multivariate regression techniques. An example of a prospective cohort study to identify factors associated with perioperative cardiac morbidity and mortality is that of Goldman and colleagues, which led to development of the Cardiac Risk Index.

Although prospective cohort studies have important value in identifying risk factors for perioperative outcomes, they also have significant limitations. The range of patients enrolled in the cohort study, both in terms of baseline characteristics and the care they receive, may impact the generalizability of the study findings. Additional biases may be introduced by loss of patients to follow-up. Failure to anticipate the potential impact of some variables and collect data on them may limit the insights gained from a cohort study. Similarly, the inability to collect data on all potential confounders of the relationship between a putative risk factor and a given outcome limits the extent to which cohort studies can support causal inferences.

Randomized clinical trials offer stronger evidence of causality than do observational cohort studies. In a randomized trial, subjects are assigned by random allocation to one of two or more treatments (potentially including a placebo) and are observed for the development of a particular outcome. In the context of perioperative risk, randomized trials may be used to determine the efficacy of an intervention or anesthetic regimen intended to improve postoperative outcomes. For example, hypothermia in the perioperative period has been associated with an increased incidence of perioperative ischemia, a surrogate marker for morbidity. In a randomized clinical trial, the use of forced-air warming to maintain normothermia was associated with a significantly less frequent incidence of perioperative morbid cardiac events. Randomized clinical trials often build on hypotheses generated in cohort studies regarding the determinants of outcomes by testing interventions directed at a specific risk factor associated with adverse outcomes.

Randomized clinical trials derive their strength from their high degree of internal validity; the randomization scheme and the use of placebo (or accepted alternative treatments) provide strong evidence that the results are related to the intervention. Importantly, these trials may have a lower degree of external validity because the intervention tested in a particular trial may not work as well or in the same manner as when it is diffused into a more heterogeneous population. Further, as a result of sample size limitations, clinical trials may often be unable to detect subtle differences in outcomes among study groups or differences in rare events.

Retrospective studies involve the identification of patients who have sustained an outcome and definition of risk factors associated with the outcome. An example of a retrospective design is a case-control study. Case-control studies identify patients with the outcome of interest. Frequently, these patients are included as part of a prospective cohort study. The prevalence of a risk factor in patients with the outcome (i.e., cases) is compared with the prevalence of the risk factor in matched control participants to maximize the efficiency and power of the results. The ratio of cases to control participants can be varied to yield greater power with an increasing number of controls. An alternative retrospective design involves the systematic review of identifiable adverse events for patterns of error. For example, Cheney and colleagues developed the American Society of Anesthesiologists’ Closed Claims Project (ASA-CCP) to assess the risks associated with anesthesia care. By obtaining the records of major events that led to legal litigation, they were able to identify factors that contributed to bad outcomes. With this methodology, selected morbidities that led to litigation can be identified. The limitation of this methodology is that the actual rates of complications in the overall population are not known; only the number of closed legal claims is identified. Cases that do not result in litigation are not included in the database.

Problems Inherent in Studying Anesthesia-Related Risk

Studying anesthesia-related risk involves a range of methodologic challenges. On the most basic level, multiple definitions exist for key outcomes, such as perioperative mortality. In particular, the timeframe in which a death can be attributed to the surgery or the delivery of anesthesia or both varies. Notably, many events related to surgery may occur after discharge when monitoring of outcomes becomes more challenging. For this reason, the National Surgical Quality Improvement Program (NSQIP), a large, prospectively collected U.S. registry of surgical care and outcomes, requires 30-day follow-up on all patients to allow for consistent assessments of outcomes for all patients.

A second major challenge in any study of postoperative outcomes is the low observed rate of many key outcomes in the population of interest. Although some recent writers have called into question the safety of contemporary anesthesia care, anesthesia-related death remains relatively uncommon in absolute terms. For example, the rate of anesthesia-related mortality described in the Confidential Enquiry into Perioperative Deaths (CEPOD) of 1987 was 1 in 185,000 patients as opposed to the 1 in 2680 cases reported by Beecher and Todd approximately 30 years earlier. As a result, efforts to identify the range of factors that now contribute to anesthetic mortality are likely to require large patient cohort studies available either from administrative sources or collected over several years from multiple institutions. Several attempts have been made to establish large epidemiologic databases to address this challenge. One example of such an approach has been the work of Dennis Mangano and the Multicenter Study of Perioperative Ischemia Research Group with regard to cardiac surgery. This group used its database to evaluate issues such as the rate and importance of atrial fibrillation after cardiac surgery and the association of perioperative use of aspirin with cardiac surgical outcomes. Other approaches include the development of cardiac surgery databases by the Society of Thoracic Surgeons, the U.S. Veterans Administration NSQIP, and the Northern New England Cardiovascular Disease Study Group. These databases are used to define risk factors for poor outcome, to compare local with national complication rates, and as educational tools. In the United States, the Multicenter Perioperative Outcomes Group has undertaken such an enterprise by pooling electronically collected intraoperative and postoperative data. Although these databases may provide extremely important information to improve care, the ability to generalize results to centers that do not have sufficient infrastructure to participate in such projects (e.g., smaller hospitals) is unknown.

Variations in care and outcomes across institutions may further complicate efforts to develop meaningful estimates of perioperative risk for use in clinical decision making by individual patients. Beyond the impact of patient illness, type of surgery, or anesthetic approach, hospital-level differences in postoperative care may have a profound impact on outcome. For example, the incidence of pulmonary embolism may be related to nursing care and the frequency of patient ambulation after surgery ; similarly, the presence of an intensivist who makes daily rounds and higher nurse staffing ratios may also affect outcome.

Finally, issues of risk adjustment complicate efforts to determine changes in anesthesia risk over time. Common endpoints, such as mortality, are influenced by patient factors as well as by anesthesia and surgical care; as such, temporal trends in patient acuity may influence the apparent adverse outcomes associated with anesthesia and surgery in a given period. With appropriate risk adjustment, changes in mortality rates over short periods may provide some indication of changes in the quality of anesthesia or surgical care. When viewed over longer periods, however, it may be more difficult to reach firm conclusions regarding temporal changes in the safety of anesthesia or surgery based on differences in mortality rates over time. For example, if improvements in anesthetic technology have allowed for older and sicker patients to undergo surgery, then the safety of anesthesia may have improved without any apparent change in mortality rates because a sicker patient population is now offered surgery that, in the past, would have been avoided. Similarly, the rapid adoption of new but relatively high-risk procedures complicates simple comparisons of anesthesia-related complications over time.

Studies of Anesthesia-Related Mortality

Efforts to understand the specific risks imposed by anesthesia care, above and beyond the surgical procedure itself, have represented an important dimension of research in anesthesia since the early 20th century. Although more recent trends in anesthesia research have emphasized a broad view of perioperative outcomes not strictly limited to events primarily caused by anesthesia care, the history of efforts to determine the safety of anesthesia management represents an important chapter in the development of modern perioperative medicine. This history also serves as important background for understanding current research and practice.

Research performed before 1980 demonstrated wide variation in reported rates of anesthesia-related mortality ( Table 30.3 ). Beecher and Todd’s 1954 report of anesthesia-related deaths at 10 institutions represents the earliest published major analysis of anesthesia outcomes. Their study included 599,548 anesthesia procedures and found a rate of all-cause mortality of 1 per 75 cases (1.3%). In 1 out of every 2680 procedures, anesthesia represented the primary cause of mortality, and it was a primary or contributory cause of mortality in 1 of 1560 procedures. The work of Dornette and Orth investigating perioperative deaths over a 12-year period at their institution corroborated these findings: they reported a mortality rate attributable to anesthesia in 1 in 2427 cases, and totally or partially attributable to anesthesia in 1 in 1343 cases. In contrast, Dripps and colleagues found the anesthesia-attributable mortality rate to be 1 in 852 in a similar single-institution longitudinal study. These differences may be partially explained by Dripps’ observation of 30-day, rather than intraoperative or 48-hour mortality, or differences in patient severity across studies.

Multiple additional studies on anesthetic mortality appeared between 1960 and 1980. In the United States, these included the Baltimore Anesthesia Study Committee, which reviewed 1024 deaths occurring on the day of or the day after a surgical procedure, and several single-institution studies. Overall, the rate of anesthesia-related mortality in these studies varied widely, ranging from 1 in 1232 cases in a study by Schapira et al to 1 in 7692 cases in the Baltimore Anesthesia Study Committee report. Results from the international community during that period were similarly heterogeneous in methodology and findings.

Studies of anesthetic risk published before 1980 varied widely in the definitions used for anesthesia-related mortality and in the mortality rates they reported; however, they suggested that death related solely to anesthesia was a relatively uncommon event. Moreover, an overall trend toward lower rates of anesthesia-related mortality across studies over time suggested potential improvements in anesthesia safety.

Studies since 1980 have generally been performed on a regional or national basis with a particular emphasis on documenting changes over time in anesthesia-related mortality. For example, Holland reported deaths occurring within 24 hours after anesthesia in New South Wales, Australia. The incidence of anesthesia-attributable deaths decreased from 1 in 5500 procedures performed in 1960 to 1 in 26,000 in 1984. Based on these estimates, the investigators asserted that for all patients receiving surgery, it was more than five times safer to undergo anesthesia in 1984 than it was in 1960.

Under the direction of the French Ministry of Health, Tiret and colleagues carried out a prospective survey of complications associated with anesthesia in France between 1978 and 1982 from a representative sample of 198,103 anesthesia procedures from hospitals throughout the country. Death was solely related to anesthesia in 1 in 13,207 procedures and partially related in 1 in 3810 ( Table 30.4 ). The French survey confirmed previous findings that major complications occur more frequently in older patients, those undergoing emergency surgical procedures, and those with more extensive comorbid conditions as measured by ASA physical status classification. More notably, the investigators found that postanesthesia respiratory depression was the leading principal cause among cases of death and coma that were solely attributable to anesthesia. Moreover, almost all the patients who had had respiratory depression leading to a major complication had received narcotics, as well as neuromuscular blocking drugs, but they had not received anticholinesterase medications for reversal of the agents.

Despite these observations, the low rates of anesthesia-attributable mortality documented in the French study offered compelling evidence of improvements in anesthesia safety. Such findings were reinforced by other, concurrent work in Finland and in the United Kingdom, resulting in the development of the United Kingdom CEPOD, which assessed almost 1 million anesthetics during a 1-year period in 1987 in three large regions of the United Kingdom.

Beyond confirming earlier work, CEPOD’s findings suggested that anesthesia care was far safer than had been found in prior studies. Examining deaths within 30 days of surgery, CEPOD investigators observed 4034 deaths in an estimated 485,850 surgeries for a crude mortality rate of 0.7% to 0.8%. Anesthesia was considered the sole cause of death in only three individuals, for a rate of 1 in 185,000 cases, and anesthesia was contributory in 410 deaths, for a rate of 7 in 10,000 cases ( Table 30.5 ). The five most common causes of death in the CEPOD cohort study are shown in Table 30.6 . Notably, of the 410 perioperative deaths, gastric aspiration was identified in 9 cases and cardiac arrest in 18 cases. Ultimately, CEPOD researchers concluded that avoidable factors were present in approximately 20% of the perioperative deaths. Contributing factors for anesthesiologists and surgeons tended to be failure to act appropriately with existing knowledge (rather than a lack of knowledge), equipment malfunction, fatigue, and inadequate supervision of trainees, particularly in off-hours shifts ( Table 30.7 ).

Large national studies performed since the 1987 CEPOD report vary in the extent to which their findings agree with those of the CEPOD investigators. In a prospective study of 7306 anesthesia procedures in Denmark, Pedersen and colleagues found complications attributable to anesthesia in 43 patients (1 in 170) and 3 deaths (1 in 2500), an incidence far higher than that documented by the CEPOD investigators. Complications in the 43 patients, in order of incidence, included cardiovascular collapse in 16 (37%), severe postoperative headache after regional anesthesia in 9 (21%), and awareness under anesthesia in 8 (19%).

In the United States, Li and colleagues conducted a population-level study to estimate epidemiologic patterns of anesthesia-related deaths, using International Classification of Diseases (ICD) codes listed in the United States multiple-cause-of-death data files for the years 1999 through 2005. Although the interpretation of Li’s study is complicated by questions surrounding the sensitivity of ICD codes for anesthesia-related mortality, their findings are in accord with those of the CEPOD report in presenting anesthesia-related mortality to be an extremely rare cause of death at the population level. The authors found anesthesia to be the underlying cause of death in 34 patients each year in the United States and a contributing factor in another 281 deaths annually, resulting in a 97% decrease in anesthesia-related death rates since the 1940s.

Recent European studies have taken a broader focus beyond anesthesia-related events to examine perioperative outcomes more generally, particularly among high-risk patients who Lagasse and others previously observed to account for the majority of postoperative deaths. In a 2011 report, NCEPOD investigators prospectively collected data on all patients undergoing inpatient surgery, excluding obstetric, cardiac, transplant, or neurosurgery cases, in United Kingdom National Health Service facilities over a 1-week period. In addition to prospectively collected patient-level data on clinical care and outcomes, the authors conducted a detailed institution-level survey of resources and practices. Although the authors observed an overall 30-day mortality rate of 1.6%, a subset of high-risk patients—approximately 20% of the full cohort—experienced a disproportionate share of adverse outcomes, accounting for 79% of all perioperative deaths. Notably, the authors identified important gaps in the perioperative management of these patients. A minority of the high-risk patients were monitored using an arterial line, a central line, or cardiac output monitoring; still more concerning was their observation that 48% of all high-risk patients who died were never admitted to a critical care unit for postoperative management. Similar findings were obtained in another study of surgical outcomes conducted across 28 European countries between April 4 and April 11, 2011. Such patterns, which the authors describe as a “systematic failure in the process of allocation of critical care resources” in Europe, highlight the potential importance of “rescue”—the prevention of mortality among patients who experience postoperative complications—in determining the outcomes of surgical care. Further, to the extent that critical care use among patients who die after surgery is higher in the United States than in the United Kingdom, such differences may offer insight into potential reasons for earlier observations of lower risk-adjusted postoperative mortality among American versus British surgical patients.

In the United States, Whitlock and colleagues retrospectively analyzed 2,948,842 cases logged in the National Anesthesia Clinical Outcomes Registry between 2010 and 2014. They documented a mortality rate of 33 per 100,000. Increasing ASA physical status, emergency case status, time of day, and age less than 1 year or greater than 65 years were independently associated with perioperative mortality. After adjustment for confounding factors, mortality remained greater for cases started after 6 PM , suggesting that certain factors influencing perioperative mortality might be modifiable. The most common concurrent outcomes in patients who died within 48 hours of anesthesia were hemodynamic instability (35.0%) and respiratory complications (8.1%). Notably, due to data limitations, the authors did not comment on the number of deaths that were anesthesia associated.

In summary, research on anesthesia-related mortality offers a complex and still incomplete picture regarding the risks of anesthesia. Taken from the perspective of the 1987 CEPOD report or the findings of Li and colleagues, modern operative anesthesia could be characterized as an exceedingly safe enterprise with bad outcomes occurring as truly rare events; however, other studies have disputed these findings. More recent work has sought to go beyond efforts to quantify the contribution of anesthesia per se to overall operative risk to explore how anesthesia providers might be able to improve outcomes among high-risk patients—in essence asking not “how safe is anesthesia?” but instead “how can anesthesia providers help make surgery safer?” Ultimately, these studies’ differing messages emphasize not only the dynamic nature of anesthesia risk over time, but also highlight important changes in how anesthetic risk has been defined across different periods and how alternate approaches to evaluating, describing, and mitigating such risk may be more or less relevant at a given moment in time.