Impact of Anesthesia Management Characteristics on Severe Morbidity and Mortality: Are We Convinced?
Arbous, M Sesmu M.D., Ph.D.; Meursing, Anneke E. E. M.D., Ph.D.; van Kleef, Jack W. M.D.; Grobbee, Diederick E. M.D., Ph.D.*
We would like to express our thanks for the critical assessments of our article “Impact of Anesthesia Management Characteristics on Severe Morbidity and Mortality.”1
All five letters address the same issue: Can causal associations be addressed in case–control studies, and should controls be fully matched to cases?
Observational studies have a record of successful contributions to medicine in establishing causal relations and increasing knowledge on pathogenesis, etiology, prognosis, and diagnosis (smoking and lung cancer, asbestos and mesotheliomas, long-term propofol infusion and cardiac failure).2,3
For a rare outcome such as death after anesthesia, a case–control study is a particularly efficient research design.4–6
Our study, with the extensive collection of information on each anesthetic procedure of cases and controls, was not a fishing expedition, nor did the statistical analysis dredge for associations. It was based on previous knowledge of potential anesthesia management–related determinants of postoperative mortality that had previously been qualitatively addressed but rarely quantified, and on common sense. The risk factors that, in our study, were significantly and independently associated with postoperative mortality and severe morbidity confirm what many anesthesiologists suspect and support our previous knowledge. Like any study, whether observational or interventional, ours only increases the causal probability of previous views and decreases the size of the confidence interval. Consequently, intervention studies will need to confirm what we have found.
The second issue to be addressed is the marked lack of understanding of the principles of the case–control design as expressed in the letters—in particular, the erroneous view that confounding bias is reduced (as completely as possible) by matching of controls to the cases. Three questions should be addressed: (1) Should we have fully matched the controls; (2) can the risk factors in anesthetic management as observed in our study be explained by the fact that the cases were sicker, more frequent emergency cases outside working hours, and so on; and (3) is there evidence for important residual confounding? The answer to all of these questions is no.7–12
Controls in a case–control study should represent those at risk of becoming a case; they provide estimates of the background frequency of an exposure (such as anesthesia management–related factors) in individuals who are free of the outcome. Fortunately, cases and controls are different. There are several factors that cause someone admitted to surgery to become a case or stay a control. What is the consequence of attempts to match as completely as possible? A major part of the thus highly selected group of controls would comprise patients who should belong in a museum for being alive. Importantly, it would not have been possible to correct in the multivariate analysis for the matched factors, and it would be unclear which bias we had introduced of unknown factors by matching on a few known factors.
Two authors (Drs. Robertson and Schmidt) specifically raise the question of whether the difference we found in some anesthetic management factors between cases and controls could be explained by the fact that the cases were sicker, more frequent emergency cases outside working hours, and so on. Indeed, in this study, there should be concern that the condition of the patient (American Society of Anesthesiologists physical status classification) and the characteristics of the procedure (emergency, outside office hours) are potential confounders, because patient and procedure factors do influence anesthesia management and also affect outcome. However, rather than extensive matching, the proper way to address this issue in a case–control study is to measure the potential confounders and correct for them in the analysis. This is what we did. Indeed, introduction of important characteristics of patients and procedures did change the risk estimates related to anesthesia management factors, showing that they are confounders of the relation. There was, however, a limit to the effect and the number of relevant confounders. In our view, therefore, residual confounding was not an important issue in our study.
Apart from anesthesia management, the American Society of Anesthesiologists physical status classification was obviously and not surprisingly the most important determinant of outcome. The number of patients with American Society of Anesthesiologists physical status III–V among the controls was large enough for proper adjustments in the analysis (23%).
In our study, we planned a qualitative analysis and a quantitative analysis. Both have been published, and they convey fundamentally different information rather than a simple duplication of the same findings as suggested by Drs. Avram and Krejcie.1,13
It is interesting to compare both reports because they show marked agreement but also demonstrate that qualitative and quantitative analyses are complementary. The qualitative analysis looked at overall anesthetic management with all available information collected, as judged by a panel of anesthesiologists. In lengthy discussions, an opinion was formed about the extent of contribution of anesthesia to the fatal outcome. However, although on overall impression no major anesthesia management factors were found to have contributed to the death in 692 cases, they were included in the case–control study because they contain important information about anesthetic care (measured by an extended questionnaire and the anesthetic form) and reveal information that is difficult to extract in a qualitative analysis.
We hope to have addressed the letters to the editor to the satisfaction of the authors. We also hope that this response helps to support the important role of case–control methods in risk research an to a better understanding of its principles. This is not to neglect the point made by Dr. Warner in the editorial accompanying our findings.14
M. Sesmu Arbous, M.D., Ph.D.
Anneke E. E. Meursing, M.D., Ph.D.
Jack W. van Kleef, M.D.
Diederick E. Grobbee, M.D., Ph.D. *
*Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, and Dutch Association for Anesthesiology, Utrecht, The Netherlands. email@example.com
1. Arbous MS, Meursing AEE, van Kleef JW, de Lange JJ, Spoormans HHAJM, Touw P, Werner FM, Grobbee DE: Anesthesiology 2005; 102:257–68
2. Rothman KJ, Greenland S: Causation and causal inference, Modern Epidemiology, 2nd edition. By Rothman, KJ, Greenland S. Philadelphia, Lippincott–Raven, 1998, pp 93–114
3. Cremer OL, Moons KGM, Bouman EAC, Kruijswijk JE, de Smet AGA, Kalkman CJ: Long-term propofol infusion and cardiac failure in adult head-injured patients. Lancet 2001; 357:117–8
4. Miettinen OS: Design of sampling of the base, Theoretical Epidemiology. Edited by Miettinen OSM. New York, John Wiley & Sons, 1985, pp 69–83
5. Miettinen OS: The “case-control” study: Valid selection of subjects. J Chronic Dis 1985; 38:543–8
6. Rothman KJ, Greenland S: Case-control studies, Modern Epidemiology, 2nd edition. By Rothman, KJ, Greenland S. Philadelphia, Lippincott–Raven, 1998, pp 93–114
7. Wacholder S, Silverman DT, McLaughlin JK, Mandel JS: Selection of controls in case-control studies: II. Types of controls. Am J Epidemiol 1992; 135:1029–41
8. Wacholder S, McLaughlin JK, Silverman DT, Mandel JS: Selection of controls in case-control studies: I. Principles. Am J Epidemiol 1992; 135:1019–28
9. Wacholder S, Silverman DT, McLaughlin JK, Mandel JS: Selection of controls in case-control studies: III. Design options. Am J Epidemiol 1992; 135:1042–50
10. Rothman KJ, Greenland S: Matching, Modern Epidemiology, 2nd edition. By Rothman KJ, Greenland S. Philadelphia, Lippincott–Raven, 1998, pp 147–61
11. Vandenbroucke JP: When are observational studies as credible as randomised trials? Lancet 2004; 363:1728–31
12. Grimes DA, Schulz KF: Compared to what? Finding controls for a case-control studies. Lancet 2005; 365:1429–33
13. Arbous MS, Grobbee DE, van Kleef JW, de Lange JJ, Spoormans HHAJM, Touw P, Werner FM, Meursing AEE: Mortality associated with anaesthesia: A qualitative analysis to identify risk factors. Anaesthesia 2001; 56:1141–53
14. Warner MA: Perioperative mortality: Intraoperative anesthetic management matters. Anesthesiology 2005; 102:251–2
This article has been cited 1 time(s).
AtherosclerosisHigh plasma lecithin: cholesterol acyltransferase activity does not predict low incidence of cardiovascular events: Possible attenuation of cardioprotection associated with high HDL cholesterolAtherosclerosis
© 2006 American Society of Anesthesiologists, Inc.
Publication of an advertisement in Anesthesiology Online does not constitute endorsement by the American Society of Anesthesiologists, Inc. or Lippincott Williams & Wilkins, Inc. of the product or service being advertised.