Secondary Logo

Journal Logo

Is Anesthetic-Related Mortality a Statistical Illness?

Levy, Warren J., MD

doi: 10.1213/01.ANE.0000173753.68064.F1
Letters to the Editor: Letters & Announcements
Free

Department of Anesthesia; Hospital of the University of Pennsylvania; Philadelphia, PA; levyw@uphs.upenn.edu

Back to Top | Article Outline

To the Editor:

The recent article by Monk et al. (1) and the accompanying editorial by Cohen (2) discuss a number of potential limitations of the work. Omitted from this discussion is one critical statistical limitation that has major implications for the interpretation of the results.

The primary, and statistically overwhelming, predictor of postoperative mortality is preoperative condition, as assessed by the Charlson Comorbidity Index. In statistical analysis, the optimal scale of measurement is continuous and linear. Successively less desirable scales include ordinal (rank-ordered), nominal (grouped), and dichotomous (a subclass of nominal scales), although sometimes measurement or analysis can only be performed using nominal or ordinal scales. When continuous data are grouped, measurable reductions in the statistical power of a data set are observed.

This issue becomes important in assessing the importance of the findings of deep hypnotic time or intraoperative systolic hypotension as predictors of postoperative mortality. The Charlson Comorbidity Index, an ordinal scale, is not a perfect measure; and the analysis was performed using it in dichotomous fashion, thus reducing its effectiveness in eliminating all of the statistical effects of comorbidity. The residual effects are likely to be covariate with hypotension or Bispectral Index <45; thus the observed statistical relationships may represent the inadequacies of the measure of comorbidity rather than a fundamental effect of deep anesthesia.

This does not invalidate the results of Monk et al., but it suggests that considerable caution be exercised in the clinical application of the results. Not only is further research into anesthetic effects indicated but additional research into the measurement of comorbidity is needed, as the accuracy and robustness of the measure of comorbidity is critical to the validity of conclusions relating anesthetic depth and postoperative mortality.

Warren J. Levy, MD

Department of Anesthesia

Hospital of the University of Pennsylvania

Philadelphia, PA

levyw@uphs.upenn.edu

Back to Top | Article Outline

References

1. Monk TG, Saini V, Weldon BC, and Sigl JC. Anesthetic management and one-year mortality after noncardiac surgery. Anesth Analg 2005;100:4–10.
2. Cohen NH. Anesthetic depth is not (yet) a predictor of mortality. Anesth Analg 2005;100:1–3.
© 2005 International Anesthesia Research Society