We thank Dr. Orkin for pointing out the well-described limitations of the American Society of Anesthesiologists Closed Claims Project with reference to our recently published article.1We agree with him that closed claims study results may be misinterpreted by readers who fail to recognize the inherent limitations in analyzing malpractice claims and the problems with attempting to generalize beyond the closed claims population. These concerns are readily acknowledged in the study limitations section and in the discussion of our manuscripts.
We disagree with Dr. Orkin’s statements concerning statistical “indiscretions” resulting from use of logistic regression analysis in our recent papers. Logistic regression analysis is fully justified in the analysis of closed malpractice claims, because it can be used to test an association between two variables. There is no assumption in the logistic regression methodology about the time order of the dependent and independent variables, and there is no requirement that there be a population at risk.2Logistic regression analysis is applied legitimately without an identified population at risk and where the outcome is identified after the potential risk factors. The analysis of case-control studies is such an example in which the entire sample is identified at the time of occurrence of an outcome. The analysis of the closed claims is analogous, where cases are those claims with death and brain damage and controls are those claims with other outcomes. The interpretation of the results in the closed claims studies is limited to the closed claims population and not the general anesthesiology population at large. Hence, the challenge arises with the interpretation of what a positive association means, and inferring from the closed claims population to the anesthesia population at large.
We agree with Dr. Orkin that the American Society of Anesthesiologists Closed Claims Project does offer rich and unique qualitative descriptions of various complications. However, we also believe that more sophisticated statistical analysis can provide useful additional information concerning associations of independent variables and outcomes among the closed claims population.
*University of Washington, Seattle, Washington. firstname.lastname@example.org