This report is the fifth in a series summarizing meetings of the International Committee for Prevention of Anesthesia Mortality and Morbidity*.(1,2,3,4) This group meets periodically to review studies of anesthesia mortality and morbidity and to discuss issues related to anesthesia patient safety in countries throughout the world. The objective of the group is to provide a forum for discussing patient safety related issues and particularly for sharing experiences in conducting studies of adverse anesthesia outcomes. This year 48 invited participants attended, representing 16 countries. Some of the presentations focused on the methodology of studies previously published; others described studies that have not yet been reported. The afternoon session focused on crisis management.
MORNING SESSION: PATIENT SAFETY STUDIES
As in previous meetings, when previously published studies are presented, the speaker is asked to discuss issues of methodology, particularly those elements of the method that were especially successful or not so successful.
Randomized Trial of Pulse Oximetry in Copenhagen (Dr. Jakob Trier-Moller)
Dr. Trier-Moller briefly summarized the multicenter trial of pulse oximetry, which involves 20,802 patients divided into two groups: oximetry or no oximetry intraoperatively and through the stay in the PACU.(5,6) No differences were found in most intra-operative or post-operative events, but there was significantly more hypoxemia and more endobronchial intubations and myocardial ischemia detected in the oximetry group intra-operatively and more hypoxemia and hyperventilation detected in the oximetry group in the PACU. There were, however, no differences in any events between the two groups post-operatively. Dr. Moller addressed three methodological questions: was bias, the learning effect or the study design responsible for the absence of any differences?
Why No Improvement With Pulse Oximetry?
There were a few statistically significant differences in the characteristics of the patients in the two groups. These could have been due to violations of the randomization process. But, because the chiefs of the five participating anesthesia departments carefully monitored the study, it is believed that randomization violations were minimal and that differences in characteristics were due to chance. Did the staff learn from using pulse oximetry and apply new knowledge to the control group? That question was addressed by comparing the difference in the rate of events and differences in the way the patients were treated for each group. The rate of events in each group was stable throughout the study. Although oxygen therapy was ordered more often for the oximetry group, the rate of ordering for both groups was stable throughout the course of the study. Was the study design efficient? The estimate of Type 11 error was 10% for cardiovascular complications, 6% for respiratory complications and 0.2% for all complications with the premise of a reduction in rate of 25% and 2 a of 5%. Still, for the most serious post-operative complications, myocardial infarction and anesthesia-contributory death, which had incidences of 0.15% and 0.034% respectively, the sample size required for a statistical power of 90% with the same assumptions, would have been 500,000 and 1,000,900 respectively. Thus, to assess the effect of pulse oximetry on the most serious outcomes would require sample populations many times greater than used in this large study. Dr. Moller still believes the randomized control trial is the gold standard. But, to be effective, studies should be conducted on an international scale and should be restricted to study populations for which changes are more likely to be seen, e.g., patients with cardiovascular diseases, ASA physical status of III or greater, age greater than 60, upper abdominal procedures.
What has been learned about Reviewer Bias from the ASA Closed-claim study? (Dr. Frederick Cheney)
Dr. Cheney, one of the leaders of the ASA closed-claim studies, briefly summarized the history of this project, which started in 1985. Because the analysis of claims for identification of causes, assessment of the contribution of substandard care, and development of strategies for reduction of risk have all depended on independent reviewers, the investigators have examined the possible bias introduced by that process.(7,8) The report by Caplan, et al. used 21 cases from the entire set and altered the outcome of each from permanent to temporary or temporary to permanent.(9) One hundred and twelve reviewers judged, among other things, appropriateness of care for a set of cases that included only one version of each case. It was shown that the reviewers were much more likely to judge that care was inappropriate when the outcome was permanent rather than temporary regardless of which way the outcome had been changed from the original.
In another study,(10) the overall agreement of reviewers about appropriateness of care was examined, with pairs of reviewers independently examining the same claims’ files. Although the agreement for all claims was fair to good, there was only slight agreement for claims in which there was only a temporary injury. The agreement on outcomes with permanent injuries was also fair to good, but this could be attributed to the greater tendency of reviewers to judge care inappropriate for such outcomes, as described above.
In the third related study, several questions were asked about how likely various types of monitors would prevent complications. Although the focus of the study was the question of the usefulness of capnography, it was again found that agreement on individual cases was poor. Dr. Cheney noted that in cases he has reviewed in which pulse oximetry was used, the cause is rarely attributed to ‘inadequate ventilation’ even though this label was the most frequently ascribed for serious outcomes in the closed-claim study. The implication is that pulse oximetry identifies the cause of an event as hypoxemia where otherwise that would not have been known.
In response to questions, Dr. Cheney noted that the study has emphasized individual reviewers because the U.S. legal system is so dependent on expert witnesses. He believes the study has demonstrated the underlying deficiency of that approach since there is so little agreement between expert reviewers about the standard of care. He suggests that practice guidelines are needed but recognized how expensive it is to create them.
Update on the Australian Incident Monitoring System (Dr. William Runciman)
As at previous meetings, Dr. Runciman reported on the progress in the ANS project which was started in July 1988 and now involves 80 hospitals in Australia. The plan is to report on the first 2,000 cases; presently, preliminary analysis has been performed on the first 1,500 cases. A meeting is scheduled this October for a thorough discussion of the study. Reportable incidents are any that did or could have resulted in the demise of the patient. A very detailed report form is used with many structured data items. Dr. Runciman noted the importance of the FOS (Person-C)n-the-Spot), who coordinates activities in each hospital, e.g., assures availability of forms, coaches clinicians in correct use of the forms, assists in gathering follow-up data, designs preventive strategies.
The pattern of incidents in many ways follows that of earlier critical incident studies regarding the type of incident, associated factors and preventive strategies. Dr. Runciman noted that 40% of events have no (undesirable) outcome, but that awareness, death or major morbidity were involved in 12% of events. The errors are categorized in various ways. As an example of items of interest, it has been found that the wrong drug was involved in 6% of events and 50% of those were potentially harmful. Surprisingly, in the majority of the cases, the wrong drug was injected from a syringe that was correctly labeled.
The project has been concentrating on the effectiveness of monitoring in identifying incidents and the mechanisms by which this happens. Some form of monitoring was responsible for detecting the event in 33%, 41% and 45% of the first, second and third set of 500 events, respectively, suggesting an increasing influence of monitoring in this regard. Of the events first identified by a monitor, the pulse oximeter, capnograph and ECG, in that order, are the three monitors most likely to have detected the event. In 70% of the events detected by a monitor, the pulse oximeter gave the first indication. Dr. Runciman commented that there are numerous incidents reported that involve a preventable death that otherwise will not be reported as an anesthesia related occurrence. He argued in favor of using incident reporting as the basis for understanding accidents since so much information is withheld when an actual accident occurs.
Update on the National Confidential Enquiry into Perioperative Deaths (Dr. John Lunn)
The NCEPOD** has been active since 1985, collecting reports of all deaths occurring within 30 days of operations in the United Kingdom, not including Scotland. Information about each event is reported by both the surgeon and anesthetist. Each year, a somewhat different sample is analyzed to focus on a different issue.(11,12) Dr. Lunn noted that they are no longer interested in searching for .causes’ of anesthetic deaths because the variance in judgments is recognized to be too great, confirming the observations reported by Dr. Cheney earlier. Further, since it has been learned from earlier years of NCEPOD that anesthesia is rarely the sole cause of a perioperative death, the focus of efforts is now on quality assurance.
This year, the objective was to collect control cases for comparison to deaths of patients with similar characteristics and procedures. The method chosen was to select the first case performed by each surgeon every day. Unfortunately, it was recognized only later that this was not a random selection and introduced bias because many surgeons schedule their less complex cases for early in the day. Dr. Lunn noted another difficulty encountered in the last year was in gathering official data on the number of deaths as reported from the Department of Health to confirm the reports received by NCEPOD. Unfortunately, data separating patient deaths from hospital discharges are no longer available. Thus, there are now no accurate data on rates of deaths; changes in death rates cannot be measured. Further describing difficulties in conducting such a large scale study, Dr. Lunn noted the delays experienced in having the detailed questionnaires returned. One reason is that the system depends upon the surgeon providing the questionnaire to the anesthetist after a death has occurred. This communication is not always effective.
Although some changes in anesthesia practice have been instituted in the UK, Dr. Lunn felt that there had been too little response from clinicians or managers to the findings of NCEPOD. This year, the project will focus on 15 procedures, requesting detailed data on the most recently performed by each surgeon to collect a more random sample of control cases. The procedures selected for this sample are those in which it is known that there are differences in management between clinicians or hospitals.
Recent Advances in Anesthesia Safety in Poland (Dr. Jacek Jestrzebski)
Poland is confronted with a shortage of physicians and particularly anesthesiologists, of which there are only I per 20,000 population versus about I per 6,000-8,000 population in most European countries. Approximately 15% of anesthetics are administered by non-anesthesiologists and the distribution of anesthesiologists throughout the country is very unbalanced. This creates an extreme workload, with some anesthesiologists working more than 85 hours per week in smaller hospitals. There is also a shortage of centralized operating rooms, so anesthesiologists often must move between distant locations. Over 50% of anesthesia machines are greater than 20 years old and there are shortages of drugs and other equipment. Thus, the focus of safety efforts has been on education. Five years of practical and specialty training is required of physicians before the specialist exam can be taken. In Dr. Jastrzebski opinion, more important for patient safety has been the emphasis on regional anesthetic techniques. The use of regional anesthesia has increased from 8% from 1971 through 1980 to 35% during 1981 through 1990. Using data collected from 17 hospitals in 3 districts, representing a total of 5,700 beds, Dr. Jastrzebski reported that the complication rate has decreased from 0.6 to 0.47 per 10,000 patients during these two time periods. Similarly, the anesthesia-related rate of mortality has decreased from 1.29 to 0.35 per 10,000 patients. Dr. Jastrzebski attributes this finding not only to the emphasis on all regional techniques, but particularly on training and techniques other than spinal and epidural anesthesia. In response to questions, Dr. Jastrzebski explained that the data is reported by the chiefs of the various hospitals, who use the anesthesia records as their primary source of information.
Anesthesia Deaths and Complications in St. Petersburg (Dr. Alexander Dizhe)
With Uvarov and others, Dr. Dizhe previously reported on the causes of deaths in the operating room in a number of medical institutions (presumably in the same region) during the period 1973 to 1985. This report is based on a similar study conducted during the period 1986 through 1991 in St. Petersburg. The investigations are based on retrospective analysis of anesthesia records. But, the absence of important information from the records and the lack of complete monitoring make it difficult to analyze these cases.
This report is based on 121 cases of death in the operating room (not including the perioperative period). About 42% of procedures were urgent; the perioperative condition of 118 patients was labeled as ‘serious’. Fifteen percent of the cases were said to be associated only with anesthesia, 25% indirectly associated with anesthesia and 60% were primarily a result of complications of the surgical intervention, including cases of trauma with uncompensated blood loss. The 48 cases in the first two groups are described further. The pattern of deaths was similar to that in the previous study. In 13 cases, the primary reason for death was related to respiratory insufficiency. Among these, there was obvious evidence of the need for artificial ventilation in three although the patient was allowed to breathe spontaneously. Three cases involved aspiration, two migration of the tracheal tube and two others pneumothorax.
Other cases of anesthesia related death were distributed among other causes that are typically reported including hypovolemia in association with epidural or spinal anesthesia. Anaphylaxis was identified in only one of the cases. Among the second group of cases, where anesthesia was only partially responsible for the death, 20 appeared related to cardiovascular collapse at the onset of surgery or shortly thereafter. Overall, Dr. Dizhe reports that the frequency of lethal outcomes associated with anesthesia during surgery appears to fluctuate between I per 10,000 and 4 per 10,000 among the institutions studied. He cites as the major factors for these variations in rate the differences in degree of specialist training, organization of assistance and maintenance of equipment.
Anesthesia Mortality and Morbidity in Japan (Dr. Yasuhiro Shimada)
This was the first report to ICPAMM of data on anesthesia mortality and morbidity in Japan. Dr. Shimada, after accepting the invitation to participate in the meeting, was surprised to realize there has been no systematic research on anesthesia outcomes or mishaps in his country. Thus, he used the occasion of this talk to conduct his own inquiry. Dr. Shimada first briefly explained the organization of anesthesia care in Japan. Although any physician can anesthetize patients, the title of anesthesia specialist is conferred, only after two years of residency and must be ,followed by a continuing training program of throe years. The title of board certified anesthesiologist is conferred only after successful completion of written and oral examinations. There is a substantial shortage of anesthesiologists in Japan; of 870 hospitals with operating rooms, only 500 have at least one board certified anesthesiologist. Although information on closed claims is collected by the Japan Medical Association, these were not available for examination. The Japan Society of Anesthesiologists was planning to initiate an inquiry of anesthesia accidents beginning in July 1992, but this will include only board-certified anesthesiologists. Thus, Dr. Shimada turned to information from the Medical Malpractice Information Center, established by a group of lawyers specializing in medical malpractice. He sent a questionnaire to 215 lawyers from this group and has, to date, received a response from 59%. A total of III case summaries were reported, however, only 64 had sufficient detail to permit analysis. In these, anesthesia accidents were defined as involving either cardiac arrest, hypotension or hypoxia necessitating special treatment and other critical events. Twenty of the suits involved anesthesiologists; in the other 44, the anesthesia was administered by a non-anesthesiologists. The cases in both groups were examined separately.
Of particular note was a large number of cases of cardiac arrest and hypotension involving spinal anesthesia administered by non-anesthesiologists (21 cases). Within this group of providers, 24 of the cases involved death and severe damage. For anesthesiologists, general anesthesia was administered in 17 of the cases; 6 resulted in death. Among both types of providers, the records suggest that monitoring which is standard in other countries is not always used, e.g., in 24 cases involving non-anesthesiologists there was no ECG monitoring, temperature monitoring was not used in any of the cases, BP monitoring was used in 37 (of the total 44). For anesthesiologists, these standard forms of monitoring were applied more frequently, but even though the date of occurrences are from 1974 to the present, no case involved the use of a pulse oximeter or capnometer. Human error was the most frequently cited cause of the adverse outcome. Dr. Shimada observed that bradycardia and cyanosis were frequently the first indication prior to the onset of cardiac arrest, suggesting that the pulse oximeter could have been useful in preventing the problem. Inadequate air-way management was often related to intubation failure. For cases involving hypoventilation, seven involved a high spinal anesthetic. For cases involving hypovolemia, eight were associated with spinal anesthesia with severe hypotension and/or cardiac arrest.
Dr. Shimada noted his dismay at realizing the lack of information about anesthetic outcomes in Japan and the relatively undeveloped state of monitoring. Not surprisingly, he suspects that these cases represent only the ‘top of the iceberg’. There are now about 300 new cases of medical malpractice reported annually with a total of 1,600 cases still pending resolution. Of the cases completed during the period 1980 through 1986, about 30% to 35% are won by the plaintiff annually. Among the cases examined in this report, the largest payment was about one million dollars.
During the discussion, Dr. Michael Rosen added further insight to the situation in Japan. He and Dr. M. Fugimori have reported that the mortality in Japan is four times higher than that in England and Wales. Data from both countries is available although that from Japan does not identify causality. Yet, if anesthesia in Japan is the cause of death among the same proportion of all cases as in the UK and Wales, about 160 such anesthesia related deaths occur annually. Conversely, the neonatal fatality rate is the same as that in England and Wales. This may result from the extreme shortage of anesthesiologists, who are in much shorter supply per capita in Japan, versus obstetricians and pediatricians, of whom about the same number are available per capita as in England and Wales. Dr. Naosuke Sugai commented that one of the main reasons for the shortage of anesthesiologists in Japan is that they are not permitted to practice privately.
The general discussion focused on two suggestions: a taxonomy for outcome and incident studies and the development of a generic incident report form. The former topic was the focus of the first meeting in 1984. That produced some agreement on definitions for anesthesia-related death and morbidity. But, Dr. Lunn observed that the term ‘anesthesia death” is still being used even though it is inappropriate since patients rarely die solely from the anesthetic. There was strong support for creation of a generic form, perhaps based on the numerous such instruments, e.g., from the AM, ASA closed-claims study, CEPOD and critical-incident studies. Several participants volunteered to join a task force to work on these issues.
There were several comments about the impact of ICPAMM on members of this group. Anecdotes were told of how information disseminated from the group has been useful in developing methods for studies in other countries and in influencing public health authorities. A question was asked of how this group could be helpful to countries that did not yet have active patient safety efforts. Based on the experiences of those who have participated in ICPAMM, it was felt that ICPAMM could be most effective by making available information on methodology so that other countries could establish their own systems for data collection. The comment was made that publications from other countries are not typically useful to influence any one government. Rather, each country must collect its own data to convince authorities that a problem exists. The lessons learned from studies reported through this series of ICPAMM meetings should be applied to future investigations. Despite the great difficulties, which have deliberately been the emphasis of discussions in these meetings, collection of information about outcomes was generally felt to be important and influential in the countries that have done this work.
The participants agreed to meet again at the European Congress of Anesthesia to be held in Jerusalem in 1994. It was suggested that, given the change in leadership of the WFSA Committee on Safety, the efforts of ICPAMM be coordinated with the WFSA. For information about participation in ICPAMM or reprints of previous reports or information about any of the studies described, contact Dr. Cooper at Massachusetts General Hospital Department of Anesthesia, Boston, MA 02114 USA.
Dr. Cooper is Director of Anesthesia Technology and Director of Biomedical Engineering at Massachusetts General Hospital; Associate Professor of Anaesthesia, Harvard Medical School; Member of the APSF Executive Committee and Co-founder and organizer of ICPAMM.
1. Keats AS and Siker ES: International symposium on anesthesia mortality and morbidity. Anesthesiology 1985; 63:349-350.
2. Cooper JB: Conference Report 1986 Meeting of the International Committee for Prevention of Anesthesia Mortality and Morbidity. Can j Anaesth 1988; 35:287-93.
3. Cooper JB: Report of a meeting of the International Committee for Prevention of Anesthesia Mortality and Morbidity. Anaesthesia 1989; 44:441-443.
4. Cooper JB: ICPAMM Meets in Poland, Preventive Safety Studies. APSF Newsletter 1990; 6:32-35.
5. Moller JT, Pederson T, Johannessen NW, et al: Pulse oximetry does not reduce post-operative complications. A prospective study of 20,802 patients. Anesthesiology 1991; 75:A867.
6. Moller JT, Ravlo 0, Jensen PF, et al: Pulse oximetry uncovers hypoxemia and decreases the incidence of evidence for myocardial ischemia during anesthesia A prospective study of 20,802 patients. Anesthesiology 1991; 75:AI057.
7. Caplan RA, Posner K, Ward RJ and Cheney FW: Peer review of major anesthetic mishaps Evidence for consensus on basic issues of clinical management. Quality Review Bulletin 1988; 14:363-368.
8. Cheney FW, Posner K, Caplan RA and Ward RJ: Standard of care and anesthesia liability. JAMA 1989; 261:1599-1603.
9. Caplan RA, Posner KL and Cheney FW: Effect of outcome on physician judgments of appropriateness of care. JAMA 1991; 265:1957-1960,
10. Cheney KL, Caplan RA and Cheney FW: Physician agreement in judging clinical performance. Anesthesiology 1991; 75:AI057.
11. Buck N, Devlin HB, Lunn JN:The Report of a Confidential Enquiry into Perioperative Deaths. The Nuffield Provincial Hospitals Trust, 3 Prince Albert Road, London, 1987.
12. Campling EA, Devlin HB, Hoile RW and Lunn JN: The report of the National Confidential Enquiry into perioperative Deaths, NCEPOD (London), 1990.
*This meeting was sponsored by the Anesthesia Patient Safety Foundation.
**To purchase the 1990 Report of the National Confidential Enquiry Into Perioperative Deaths, send a check payable to NCEPOD in the amount of 9.00 pounds (in the UK) or 10.00 pounds (overseas) to National Confidential Enquiry Into Perioperative Deaths, 35-43 Lincoln’s Inn Fields, London WC2A 3PN (Tel: 071-831 6430 Fax: 071-430 2958).