Despite an Herculean analysis, Silber et al.1
are challenged when exploring the influence of the anesthesiologist’s medical direction on clinical outcome of surgical anesthesia. After adjusting for patient and hospital characteristics, they found “statistically significant” relations between absence of medical direction and both death and failure to rescue, but significance testing and related P
values convey nothing about the magnitude of an intervention’s effect or clinical importance. Indeed, small differences of little clinical importance can be found to be statistically significant with very large study populations because P
values are sensitive to sample size. More indicative of the importance of a factor in such a logistic regression analysis are the odds ratios. However, with odds ratios of 1.08 and 1.10, respectively, the influence of absence of medical direction seems very small, at least for an unselected patient population. By comparison, the odds ratios for relations between customary health risks and specific outcomes (e.g.
, cigarette smoking and lung cancer, asbestos exposure and mesothelioma, chronic alcoholism and hepatic cirrhosis) are in the range of 5 to 20, making inferences about the importance of these risk factors relatively easy.
Trying to extract greater meaning from their analyses, Silber et al.1
note that absence of medical direction corresponds to 2.5 excess deaths per 1,000 anesthetic procedures. Given that there are an infinite number of “excess deaths” that might be avoided generally in health care with different interventions, one might wonder how the benefit of medical direction compares with those of other health care interventions. A handy way to compare interventions involves estimating the number of patients needed to be treated to prevent one negative outcome, 2
which in the Silber example would be 400 (i.e.
, 1,000 anesthetic procedures per 2.5 deaths). Compared with many other interventions (table 1
), medical direction is much less effective but still within the spectrum of interventions that are widely thought to be beneficial.
Fredrick K. Orkin M.D.