Logistic Regression Modela to Identify Variables Independently Associated with Failure to Receive Appropriate Empiric Antimicrobial Therapy.

aModel included a generalized effect estimate to account for clustering among hospitals and included the following confounders: admission from a facility (p = 0.17), presence of a central line at the time of BSI (p = 0.53), in the ICU prior to BSI onset (p = 0.59), and presence of a Foley catheter at the time of admission (p = 0.82).bTwo interaction terms were included in the model: Interaction between Charlson score and community-onset, healthcare associated location of acquisition (0.04) and interaction between Charlson score and hospital-onset, healthcare associated location of acquisition. No specific effect measure available for these variables due to interaction.cMulti-level variable, therefore no effect measure available.