For efficacy
outcomes, we compared the proportions of patients
in the two study groups who could be evaluated
and who had any type of surgical-site infection,
using Fisher’s exact test and calculating the
relative risk of infection and 95% confidence intervals.
The consistency of the effects of the study
intervention on infections across different types
of surgery was examined with the use of an interaction
test. To determine whether the results were
consistent across the six participating hospitals,
a prespecified Breslow–Day test for homogeneity
was performed.