᭜ EDITORIAL VIEWS

Anesthesiology 2002; 96:1039–41 © 2002 American Society of Anesthesiologists, Inc. Lippincott Williams & Wilkins, Inc. Perioperative Risk How Can We Study the Influence of Provider Characteristics? IN the 20th century, the medical community was in- prove outcome.7–10 The relatively high probability of tensely interested in understanding the factors associ- death associated with patient disease has allowed the ated with perioperative risk. In the current issue of majority of these studies to be performed in a single 1 ANESTHESIOLOGY, Silber et al. have added to this literature center or a small group of centers. and studied the influence of anesthesia provider charac- Although most of the research has focused on identi- teristics on outcome. fying and minimizing the impact of patient disease, in-

One of the earliest systematic analyses of the factors creasing attention has been given to the two other com- Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/96/5/1039/404427/7i0502001039.pdf by guest on 27 September 2021 associated with perioperative risk occurred in 1935 ponents of perioperative risk: surgery and anesthesia. when Ruth helped to establish a commission to analyze Because of the lower probability of death associated perioperative deaths.2 The commission relied on volun- with these two factors, a different methodologic ap- tary submission of cases and determined the cause of proach is often needed. Single-site studies are frequently death by majority vote. This methodology was deemed not feasible, and assessments of cause of death by indi- inadequate in subsequent studies. vidual clinicians may not be appropriate. Because death A major advance in the analysis of perioperative risk is less common, outcomes other than death may be was the report by Beecher and Todd3 of the factors studied to obtain a statistically reliable number of adverse associated with anesthetic death in 10 institutions, pub- events. This approach has particularly been used with re- lished in 1954. Their study reviewed the administration spect to the development of clinical predictors of cardiac of 599,548 anesthetics to patients with a wide range of events, including nonfatal myocardial infarction.7,11 diseases. The cause of mortality was determined at the Although investigators have continued to rely on mul- local institution by consensus of a surgeon and the chief ticenter studies, the problem has increasingly been stud- anesthetist of the institution. Although the methodology ied through the use of administrative data sets.12 Exam- for assigning the cause of death was criticized as being ples of administrative databases include Medicare claims arbitrary, the major contributions were the recognition files, private insurance company claims, and hospital that there are three major causes of death in the periop- electronic records. These databases have a limited erative period and the quantification of these risks. They amount of data for an extremely large number of sub- found that a patient had an overall chance of mortality of jects. For example, the Medicare database includes both 1 in 75 cases in the perioperative period. Anesthesia was financial data and ICD-9 (disease) and CPT (procedure) reported as the primary cause of mortality in 1 in 2,680 codes for each patient. They also include information cases, and was either the primary or contributory cause regarding location of care and provider. of mortality in 1 in 1,560 cases. Surgical error in diagno- Administrative data sets have been used to derive risk- sis, judgment, or technique was the primary cause of adjusted provider specific mortality. For example, New death in 1 in 420 cases, and patient disease was the York and Pennsylvania have used this type of data to primary cause in 1 in 95 cases. develop a system of individual physician report cards for Subsequent researchers have adopted this framework cardiac surgery, which allow consumers to determine of the tripartite components of perioperative risk (sur- their surgeon’s risk-adjusted mortality rate.13,14 4–6 gery, anesthesia, patient disease). Because it is the In studying the influence of anesthesia-related factors, major cause of mortality, most of the subsequent analy- previous studies have suggested that its contribution is ses have focused appropriately on patient disease. Spe- several orders of magnitude lower than the influence of cifically, investigators have focused on identifying med- patient disease or surgical error. During the previous ical conditions associated with poor outcomes and have two decades, there have been attempts to quantify the attempted to identify specific interventions that will im- risk associated with anesthesia using a variety of sources. Slogoff and Keats15 reported on rates of perioperative This Editorial View accompanies the following article: Silber JH, myocardial ischemia and infarction after cardiac surgery ᭜ Kennedy SK, Even-Shoshan O, Chen W, Mosher RE, Showan AM, and found that one anesthesia provider (Anesthesiologist Longnecker DE: Anesthesiologist board certification and patient 7) had a significantly higher rate than the others. The outcomes. ANESTHESIOLOGY 2002; 96:1044–52. Confidential Enquiry into Perioperative Deaths assessed nearly 1 million cases of anesthesia during a 1-yr period in 1987 in three large regions of the United Kingdom.5 Accepted for publication January 29, 2002. The authors are not supported by, nor maintain any financial interest in, any commercial activity that may be There were 4,034 deaths within 30 days in an estimated associated with the topic of this article. 485,850 operations, resulting in a crude mortality rate of

Anesthesiology, V 96, No 5, May 2002 1039 1040 EDITORIAL VIEWS

0.7–0.8%. Surgery had contributed totally or partially in First, it is not clear from the data whether the actions 30% of all patients. Progression of the presenting disease of the anesthesiologist contributed to the patient’s had contributed to death in 67.5% of all patients, with death. As noted in the article, the estimated number of progress of an intercurrent disease being relevant in excess deaths derived in this article—3.8 excess deaths 44.3% of patients. Anesthesia was considered to be the per 1,000 cases—is several magnitudes greater than the sole cause of death in only 3 individuals, for a rate of 1 in often-quoted risk of death from anesthesia of 0.005 185,000 cases, and anesthesia was contributory in 410 deaths per 1,000 cases. This is in the range of mortality deaths, for a rate of 7 in 10,000. Therefore, the ability to associated with the other components of perioperative determine the influence of the anesthesia provider on risk: surgery and patient disease. The data do not permit perioperative mortality would require either an ex- the authors to link specific actions of the anesthesiolo- tremely large sample size or alternative outcome mea- gist to the death rate, and therefore, there is no direct sures. Silber et al. have used large data sets and have evidence that all of the excess deaths can be attributed developed additive outcome measures in the study of to whether the anesthesiologist had a board certification. Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/96/5/1039/404427/7i0502001039.pdf by guest on 27 September 2021 the factors associated with anesthesia.12,16 However, extrapolating from the foundations of the fail- 1 In this issue of ANESTHESIOLOGY, Silber et al. have contin- ure-to-rescue concept, more skilled providers of care ued their inquiries into hospital and provider characteris- (e.g., board certification as a surrogate for “skill”) may tics for perioperative mortality by evaluating death and reduce the severity and fatality of a complication, which failure to risk in a large cohort of Medicare patients. Silber may not be directly related to anesthesia care. This may et al. developed the concept of failure to rescue as a means be related to such interventions as more skilled resusci- of studying quality of care for the manner in which hospi- tation or better pharmacologic management of coexist- tals handle 41 known “emergencies” or complications that ing disease. Information not available in the claims data are coded in the Medicare files.17,18 In brief, they evaluated would be required to demonstrate this link. the 30-day death rate in patients who died without a re- Second, the article attributes the difference in death corded complication or in whom a complication devel- rates to board certification based on a set of odds ratios. oped. The theoretical foundation for such an approach is In statistics, omitted variable bias occurs when variables that the presence of more “skilled” care would prevent not in the equation are correlated with a variable that is complications from becoming deaths. By evaluating both already in the equation. Board certification is included in mortality and failure to rescue, the number of relevant the equation; however, many variables that are probably outcomes dramatically increases, allowing investigators to correlated with board certification are not included in study the influence of care providers on outcome on a the equation. These include training at better medical smaller number of cases. schools and residency programs, being educated at US 16 In a previous issue of ANESTHESIOLOGY, Silber et al. medical schools, and other factors. These variables, not compared anesthesia care personally performed or med- board certification, could be responsible for the higher ically directed by an anesthesiologist with the outcomes death rate. Statistically, the greater the correlation be- of patients whose anesthesia was not personally per- tween the omitted variables and the variable included in formed or medically directed by an anesthesiologist and the equation (board certification), the greater the omit- reported that both 30-day mortality rate and failure to ted variable bias. That is, it is possible that some other rescue were lower when anesthesiologists directed an- factor highly correlated with board certification (i.e.,an esthesia care. Work by this group and others has shown omitted variable) is actually responsible for the higher that hospitals with higher percentages of board-certified death rate. anesthesiologists on staff had lower adjusted mortality This hypothesis of some other factor being responsible and failure-to-rescue rates among surgical patients.17,19 is given some legitimacy in figure 1. The odds ratio The current study further evaluates the importance of seems to be meaningfully different from 1.0 only in the board certification by specifically comparing outcomes 16- to 20-yr postgraduation cohort. In the other years, in anesthesiologists who were board certified with those the ratio is not significantly different from 1.0. There- who were not. fore, it does not seem to be the fact that the physician Although this article represents a significant contribu- passed the Boards that is important, but that they passed tion to the literature, any clinical or policy recommen- the Boards 16–20 yr ago. Further analysis of what those dations to be derived from this article must recognize individuals 16–20 yr after graduation had in common is the limitations imposed by the data set and this type of warranted to determine what denotes a more “skilled” analysis. Clinicians need to know that this type of anal- clinician and whether it is the certification process itself ysis is designed to generate hypotheses, not to test hy- or the factors leading to board certification. potheses. Because all the relevant data are not available Third, there were only 75 anesthesiologists who were from Medicare claims data, it is not possible to reach not board certified 11–25 yr after medical school gradu- definitive conclusions about the causes of death or the ation. This is a very small number of anesthesiologist to other outcomes using this analytic approach. draw definitive clinical or policy conclusions. Because

Anesthesiology, V 96, No 5, May 2002 EDITORIAL VIEWS 1041 these 75 anesthesiologists can then be subdivided into References US-trained and foreign-trained medical graduates, are 1. Silber JH, Kennedy SK, Even-Shoshan O, Chen W, Mosher RE, Showan concentrated in relatively few hospitals, and probably are AM, Longnecker DE: Anesthesiologist board certification and patient out- located in relatively few geographic areas, the ability to comes. ANESTHESIOLOGY 2002; 96:1044–52 2. Ruth HS: Anesthesia study commissions. JAMA 1945; 127:514–7 generalize the results becomes even more problematic. 3. Beecher HK, Todd DP: A study of deaths associated with anesthesia and Despite these limitations, analysis of Medicare claims surgery. Ann Surg 1954; 140:2–34 4. Lunn JN, Hunter AR, Scott DB: Anaesthesia-related surgical mortality. An- offers a unique opportunity to generate hypotheses that aesthesia 1983; 38:1090–6 warrant further study. The previous study by Silber et 5. Lunn JN, Devlin HB: Lessons from the confidential enquiry into periopera- al.16 regarding medical direction appropriately has led to tive deaths in three NHS regions. Lancet 1987; 2:1384–6 6. Tiret L, Desmonts JM, Hatton F, Vourc’h G: Complications associated with calls for further research about the relative risks associ- anaesthesia: A prospective survey in France. Can Anaesth Soc 1986; 33:336–44 ated with anesthesiologists and nurse anesthetists. Simi- 7. Goldman L, Caldera DL, Nussbaum SR, Southwick FS, Krogstad D, Murray B, Burke DS, O’Malley TA, Goroll AH, Caplan CH, Nolan J, Carabello B, Slater EE: larly, the current study should stimulate further questions Multifactorial index of cardiac risk in noncardiac surgical procedures. N Engl with regard to the importance of board certification with J Med 1977; 297:845–50 Downloaded from http://pubs.asahq.org/anesthesiology/article-pdf/96/5/1039/404427/7i0502001039.pdf by guest on 27 September 2021 8. Mangano DT, Layug EL, Wallace A, Tateo I: Effect of atenolol on mortality respect to patient outcomes. and cardiovascular morbidity after noncardiac surgery. Multicenter Study of The specialty of anesthesiology has been lauded for its Perioperative Ischemia Research Group. N Engl J Med 1996; 335:1713–20 early adoption of systems to improve patient safety and 9. Poldermans D, Boersma E, Bax JJ, Thomson IR, van de Ven LL, Blankensteijn 20 JD, Baars HF, Yo TI, Trocino G, Vigna C, Roelandt JR, van Urk H: The effect of for being at the forefront of the six sigma process. bisoprolol on perioperative mortality and myocardial infarction in high-risk These processes have led to rare fatalities directly related patients undergoing vascular surgery. Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography Study Group. N Engl J Med 1999; to the provision of anesthesia care. The exploration by 341:1789–94 Silber et al. of factors associated with reduced rates of 10. Rose DK, Cohen MM, DeBoer DP: Cardiovascular events in the postanes- thesia care unit: Contribution of risk factors. ANESTHESIOLOGY 1996; 84:772–81 failure to rescue provides one innovative way of further 11. Eagle K, Brundage B, Chaitman B, Ewy G, Fleisher LA, Hertzer N, Leppo L, reducing the elements of perioperative risk. Board cer- Ryan T, Schlant R, Spencer W, Spittell J, Twiss R: Guidelines for perioperative tification of the anesthesiologist, or some unidentified cardiovascular evaluation of the noncardiac surgery. A report of the American Heart Association/American College of Cardiology Task Force on Assessment of aspect of the educational or certification process, war- Diagnostic and Therapeutic Cardiovascular Procedures. Circulation 1996; 93: rants further study as a means of reducing the mortality 1278–317 12. Cohen M, Duncan PG, Tate RB: Does anesthesia contribute to operative associated with perioperative complications related to mortality? JAMA 1988; 260:2859–63 anesthesia, surgery, or patient disease. Because the risk 13. Hannan EL, Kumar D, Racz M, Siu AL, Chassin MR: New York State’s Cardiac Surgery Reporting System: Four years later. Ann Thorac Surg 1994; directly related to anesthesia seems to be almost too low 58:1852–7 to measure easily, the goal should be to identify addi- 14. Schneider EC, Epstein AM: Use of public performance reports: A survey of patients undergoing cardiac surgery. JAMA 1998; 279:1638–42 tional actions or training within the domain of the anes- 15. Slogoff S, Keats AS: Does perioperative myocardial ischemia lead to post- thesiologist that ensures that our patients receive the operative myocardial infarction? ANESTHESIOLOGY 1985; 62:107–14 best quality of care. The specialty should focus on de- 16. Silber JH, Kennedy SK, Even-Shoshan O, Chen W, Koziol LF, Showan AM, Longnecker DE: Anesthesiologist direction and patient outcomes. ANESTHESIOLOGY termining the link between provider characteristics and 2000; 93:152–63 failure to rescue and designing interventions to further 17. Silber JH, Williams SV, Krakauer H, Schwartz JS: Hospital and patient characteristics associated with death after surgery: A study of adverse occurrence reduce all-cause perioperative mortality. and failure to rescue. Med Care 1992; 30:615–29 18. Silber JH, Rosenbaum PR: Measuring quality of hospital care. JAMA 1995; Lee A. Fleisher, M.D.,* Gerard F. Anderson, Ph.D.† *Associate 273:21 Professor of Anesthesiology and Vice Chairman of Clinical Investiga- 19. Hartz AJ, Krakauer H, Kuhn EM, Young M, Jacobsen SJ, Gay G, Muenz L, tion, Joint Appointments in Medicine and Health Policy and Manage- Katzoff M, Bailey RC, Rimm AA: Hospital characteristics and mortality rates. ment, †Professor and Associate Chair of Health Policy & Management, N Engl J Med 1989; 321:1720–5 Joint Appointment in Medicine and International Health, Johns Hop- 20. Kohn LT, Corrigan JM, Donaldson MS: To Err Is Human: Building a Safer kins Medical Institutions, Baltimore, Maryland. lfl[email protected] Health System. Washington DC, Institute of Medicine, 2000

Anesthesiology, V 96, No 5, May 2002