The article by Anderson et al1 raises several concerns about study methodology and the appropriateness of their study conclusions. The study contains an acknowledged type II statistical error, ie, despite the sample size of 1070 patients, the study is too small to have statistical power. The ostensible statistical difference in a post hoc analysis of a subset of patients (eg, more "minor wound infections" in the ticarcillin plus clavulanic acid group) is therefore meaningless. Further, extrapolation of estimated costs, rather than measurement of actual resource consumption, led to the untenable conclusion that surgical prophylaxis with either ceftriaxone or cefotaxime is more cost-effective than prophylaxis with ticarcillin plus clavulanic acid.
The definitions of "infectious morbidity" that were used are problematic in several respects. The definitions of "wound infection" omitted any consideration of the appearance of the wound. Nosocomial pneumonia is overdiagnosed by conventional sputum collection, and the criteria used actually