From the Literature: Has Anesthesia Mortality Been Decreased?

Zeitlin GL

Zeitlin GL: Possible decrease in mortality associated with anesthesia, A comparison of two time periods in Massachusetts, USA. Anaesthesia 1989; 44:432A33.

Epidemiology is increasingly recognized as a useful research tool in anesthesiology. Ultimately, the specialty’s safety record and its improvement over time can be documented only with epidemiologic data involving mail numerators and large denominators, both of which are counted accurately. By attempting to pin down the anesthetic mortality rate at two points in time (circa 1960 and circa 1980), Zeitlin has demonstrated some of the pitfalls and frustrations in trying to collect useful epidemiologic data.

For a baseline, the author cites a previous study of anesthetic mortality’ in the years 1955-64 which demonstrated a mortality rate of 2.16 per 10,000 anesthetics. This irate was based on a numerator of 15 primary anesthetic deaths observed in a busy private practice in Massachusetts, and a denominator of 69,291 anesthetics from which the 15 deaths were drawn. It is noteworthy that other roughly contemporaneous studies found similar rates. For instance, Harrison2 reported an anesthetic mortality rate in South Africa of 2.2 per 10,000 for the period of 1967-76. So far, so good; we have a nice piece of epidemiologic work, with results comparable to those of other workers.

Regarding the more recent past: Has the anesthetic mortality rate improved since 1960? Using data from Massachusetts two decades later, Zeitlin attempts to answer that question, citing a dramatically lower figure of 0. 1 6 deaths per I 0,000 anesthetics. For the numerator (number of deaths), he drew data from a Massachusetts malpractice insurance company, counting 31 claims involving deaths due to anesthesia between the years 1977-84. But because insurance companies don’t count denominators (they record failures, not successes), Zeitlin” could only estimate the number of cases from which the numerator came(that is, his denominator)to be 800 anesthetics per insured anesthesiologist per year, or 1,920,000 anesthetics.

This is not good epidemiology, as Zeitlin, to his credit, admits. Estimating the denominator cannot substitute for counting (A lession for anesthesiologists who want to do epidemiology-. Counting denominators is just as important as counting numerators, and a lot harder). Worse yet, the numerator hem is not the number of deaths, but the number of malpractice claims resulting from those deaths. The assumption that all deaths result in claims is hardly credible, even in Massachusetts. (the lesson here: data from malpractice claims studies, while useful in themselves, cant be compared with clinically derived numerators.)

The admittedly shaky assumptions of the second Part of Zeitlin’s study might be justified as an attempt to test an hypothesis for which frustratingly little direct evidence currently exists: that anesthesiology is Setting safer. Moreover, a result of 0. 16 deaths per I 0,000 would be believable in the context of other studies with similar results. Unfortunately, other studies of anesthetic mortality in the late 70s and early 80s (from Finland3, Canada4 and France5, for example) have continued to report mortality rates of approximately I to 2 per I 0, 000. Zeitlin’s recent mortality figure–, an order of magnitude smaller than those of other contemporaneous studies, may be therefore literally “too good to be true.”

Zeitlin’s report to the contrary, there is evidence pointing to an anesthetic mortality rate of I or 2 per I 0,000 even possibly up to the mid-1 980s. Anestheliogists have very good reason to believe that the rate has fallen within the past five or more years, but no concrete epidemiologic proof. We will not have that proof until good epidemiologic studies of contemporary practice make their way into print.


1 Memery HN:Anesthesia mortality in private practice. A ten year study. JAMA 1965; 194:1185-8.

2. Harrison GG:Death attributable to anesthesia. BiI Anaesth 1979; 50:1041-1046.

3. Hovi-Viander M: Death associated with anesthesia in Finland. Br I Anaesth 1980; 52:483-9.

4. Duncan, PG: Postoperativecomplications: factors of significance to anaesthetic practice Can I Anaesth 1987; 34:2-8.

5. Tiret L, Desmonts JM, Hatton F, Vourc’h G: Complications associated with anaesthesia a prospective survey in France. Can Anaesffi Soc 1 1986; 33.-336-44.

Review by Richard L. Keenan, M.D., Professor and chairman of Anesthesiology at the Medical College of Virginia, Richmond, and epidemiologic investigator of cardiac arrests during anesthesia care.

Editor’s Note: Dr. Zeitlin replies.”Dr. Keenan’s review is fair comment, but please note further that two recent studies also suggest a rate of mortality attributable solely to anesthesia care dramatically lower than 1-2 per 10,000. The British CEPOD)

study1 and also the review of the Harvard Medical School data before and after the adoption of the monitoring standards’ both showed an anesthesia caused mortality of 0.05 per 10,000 patients.”


1 Lunn YN, Devlin HB: lessons from the confidential enquiry into P-ostoperative deaths in three NHS regions. L-ancet 1987; 2:1384.

2. Eichhom JH: Prevention of intraoperative anesthesia accidents and related severe injury through safety monitoring. Anesthesiology 1989; 70:572-577.