ANESTHESIA is an intrinsically hazardous undertaking and anesthesiologists have struggled for years to determine the incidence of catastrophic adverse outcomes. There is a wide belief that anesthesia is safer now than it was 30 or more years ago, and anesthesiology has been acknowledged by some for its nearly perfect safety record. 1In this issue of Anesthesiology, that belief is challenged. 2Dr. Robert Lagasse, reviewing literature from the past 5 decades and presenting new data, claims provocatively that “the emperor is not wearing any clothes” and that “we must dispel the myth that anesthesia-related mortality has improved by an order of magnitude.” Is this commonly held belief of markedly improved safety imaginary or does Lagasse's analysis miss the mark? The truth is somewhere in between. Assessing the contribution of anesthesia care to perioperative mortality and morbidity is notoriously difficult. Making fair comparisons across epochs in time is even more problematic. Yet, absence of unequivocal data is not evidence of absence of the effect. Even so, the available data may reflect several points: anesthesia for healthy patients is “safer” than it once was (but further progress may be possible); the rate of anesthesia-related mortality for all surgical patients is still higher than desired; and, safety levels can “plateau” or even diminish over time without constant effort at improvement.

Perhaps the last thorough peer-reviewed summary of anesthesia mortality studies was in 1987. 3Fifteen years later, the issue is still plagued by confounding variation in definitions, relatively small sample sizes from selected institutions, and the lack of large population studies, especially in the United States. The new review in this issue by Lagasse of 23 studies, dating back to the mid-1950s, provides an interesting, but not totally clear perspective. The reported rates of anesthesia-related mortality vary widely by year and by country. Only four reports are from the United States and only one (The Confidential Enquiry into Perioperative Deaths) 4is a systematic analysis of a large population. Lagasse's control chart suggests randomness in anesthesia-related mortality during the past 30 yr. As Lagasse asks, is this real or an artifact of varying techniques and definitions? There is no way to be certain. The definitions for deaths in which anesthesia was “associated,”“related,”“contributory,” or “preventable” varied widely as did the time windows for the perioperative period (24 h to 30 days). The locales studied were disparate. Yet, even if we accept the definitions to be roughly comparable over the years, much else has changed that is not accounted for in the comparison of these studies.

Are the nature of surgical patients and the operations performed unchanged over time? It seems likely that, compared with the period before 1970, more complex procedures are now performed more readily on sicker patients. This phenomenon might be studied using large databases if they provide equivalent data far enough back in time. Moreover, while it might be possible to track changes over time in the distribution of patients’ ASA physical status scores, we cannot determine whether there have been shifts in the way that scores would be assigned to the same patients in different epochs. It seems entirely possible that patients previously judged as ASA 3 or 4 may now be scored as ASA 2 or 3 as medical management for many diseases has improved and anesthetizing such patients has become routine.

Despite the difficulties, there are data to suggest on the order of a ten-fold improvement in anesthesia safety if the focus is on studies in the United States, the country for which claims of dramatic improvements have been made. Add to Lagasse's review the pioneering study of Beecher-Todd, which covered the period 1948–1952. 5It reported a rate of 1:1560, for deaths in which anesthesia was at least “a very important contributing factor.” Despite the caveats of interpretation, it's fair in definition and methods to compare this order of magnitude with that provided by Lagasse's new data, where the mortality rate was approximately 1:13,000. That's close to a ten-fold improvement. Considering differences in patient risk and complexity of surgery from 5 decades ago, it's easy to see why many claim a dramatic increase in safety from years past despite the absence of hard data. The reports of Marx and Memery, also from the United States, support this. 6,7 

The changing relative risk factor for anesthesiologists in malpractice insurance premiums provides different supportive evidence. The risk factor has dropped dramatically. Though many factors affect malpractice risk, changes of this magnitude would be unlikely without a substantial reduction in losses. The inflation-adjusted premium for Harvard Medical School insured anesthesiologists is approximately one-quarter of its rate in the mid-1980s (Personal communication between author J. Cooper and Robert Hanscom, Risk Management Foundation of the Harvard Medical Institutions, Cambridge, MA, August, 2002). That's not an order of magnitude, but it's a large change. Large claims are typically for major injury in relatively healthy patients; thus these data suggest that anesthesia care has become safer for that cohort.

Rather than the evolution of safety over time, consider the absolute risk of death related to anesthesia in the last 10–20 yr. Several studies indicate a very low mortality rate for ASA 1 and 2 patients. The quoted rate of one death per 200,000 anesthetics likely originated from Eichhorn's study of closed malpractice claims, which counted mostly healthy patients. 8Two periods were studied, before and after the institution of monitoring standards. The rates of anesthesia mortality for ASA 1 and 2 patients was 5 in 757,000 (1/151,000) patients in the first period (1976– June, 1985) and 0/244,000 patients from July, 1985 through June, 1988. A recent study by Arbus et al.  of 869,000 anesthetics in the Netherlands during 1995–1997, reported 7 deaths solely attributable to anesthesia, a rate of 1:124,000. 9Yet another recent study, which appears to be comprehensive in its collection of perioperative adverse events in a Canadian hospital from 1996–2000, reports only 1 anesthesia contributory death in 84,000 patients, including patients in all ASA classifications. 10Finally, in the two hospitals studied by Lagasse, there were no deaths solely attributable to anesthesia among approximately 184,000 patients of all ASA physical status. Anesthesia had a major contribution in only one of 126,000 ASA 1 and 2 patients. 2As hazardous undertakings go, anesthesia's track record for healthy patients is indeed a model for health care. Carefully examining the Beecher-Todd study, we see that for “good risk patients” anesthesia was the primary or major contributing factor to death in approximately 4.1 per 10,000 surgical procedures between the years 1948 and 1952. 5Compared with today's data, that is more than an “order of magnitude” improvement that many have suggested.

But, it's not good enough. The goal of the Anesthesia Patient Safety Foundation, expanding from that articulated by Macintosh 1150 yr ago, is that no patient  shall be harmed by anesthesia. ‡

Lagasse appropriately reminds us that anesthesia is not yet completely safe for ASA 1 and 2 patients (whose main risks may be iatrogenic) and less so for sicker patients. Anesthesia still contributes to serious adverse events and avoidable deaths.

Has anesthesia safety reached a plateau? Lagasse's data from hospitals in the 1990s contain too few patients for a definite conclusion, but such a plateau is possible. Negative outcomes related to anesthesia-care might end up being traded off against chances for successful surgical treatment of serious diseases. The low probability of anesthetic mortality for healthy patients may force safety goals to compete inappropriately against efficiency and cost considerations. Lagasse's paper emphasizes a need to improve anesthesia and system safety for all patients, including a growing cohort of ASA 3 and 4 patients.

The theory of organizational safety teaches that “safety” is a never-ending process whose success may not be measured strictly by epidemiologic methods. 12The profession of anesthesiology itself is a model concerning patient safety processes . 13Anesthesiologists have played important leadership roles in addressing organizational safety in all of health care. Anesthesiology was the first medical profession to treat patient safety as an independent problem. Anesthesiology has implemented widely accepted guidelines on basic monitoring, conducted long-term analyses of closed malpractice claims, addressed fatigue of residents serving in-house call, developed patient simulators as meaningful training tools, and tackled problems of human error. Most importantly the profession has institutionalized safety in its scientific and governing bodies, creating the ASA's Patient Safety and Risk Management Committee and the Anesthesia Patient Safety Foundation. Yet we should not be complacent, believing that we have won the war. Lagasse, in this review and in his prior work, has made valuable contributions to this effort. We all agree that the war must continue. Nonetheless, we believe there is no “myth” as to improvement in anesthesia patient safety. There are semantic disagreements, differences about what are the epochs being compared and little good data to be found. Anesthesiologists should remain aware of the hazards they still face, take pride in having been the leaders in patient safety efforts, and stay motivated to continue the pursuit of “no harm from anesthesia” with the passion it still demands.

Leape LL: Error in medicine. JAMA 1994; 272: 1851–1857
Lagasse RS: Anesthesia safety: Model or myth? A nesthesiology 2002; 97: 1609–17
Derrington MC, Smith G: A review of studies of anaesthetic risk, morbidity and mortality. Br J Anaesth 1987; 59: 815–33
Lunn JN, Devlin HB: Lessons from the confidential enquiry into perioperative deaths in three NHS regions. Lancet 1987; 2: 1384–6
Beecher HK, Todd DP: A Study of the Deaths Associated With Anesthesia and Surgery. Based on a study of 599,518 anesthesias in ten institutions 1948–1952, inclusive. Ann Surg 1954; 140 (1): 2–35
Marx G, Mateo C, Orkin L: Computer analysis of postanesthetic deaths. A nesthesiology 1973; 39: 54–8
Memery HN: Anesthesia mortality in private practice: A ten-year study. JAMA 1965; 194 127–30
Eichhorn JH: Prevention of intraoperative anesthesia accidents and related severe injury through safety monitoring. A nesthesiology 1989; 70: 572–7
Arbous MS, Grobbee DE, van Kleef JW, de Lange JJ, Spoormans HHAJM, Touw P, Meursing AEE: Mortality associated with anaesthesia: Qualitative analysis to identify risk factors. Anaesthesia 2001; 56: 1141–53
Fasting S, Gisvold SE: Serious intraoperative problems: A five-year review of 83,844 anesthetics. Can J Anesth 2002; 49: 545–58
Macintosh R: Deaths under anaesthetics. Br J Anaesth 1948; 21: 107–36
Reason J: Managing the risks of organizational accidents. Aldershot, England, Ashgate Publishing Limited, 1997
Gaba DM: Anaesthesiology as a model for patient safety in health care. BMJ 2000; 320: 785–8