Methods: Subjects were hospitalized between January 2010 and June 2015 with monomicrobial ESBL bacteremia and received empiric PTZ or carbapenem. To remain in the study, all patients had to receive carbapenem therapy after their bloodstream isolate was identified as ESBL-producing. We analyzed patients based on their level of care and clinical outcome was determined as day of death or discharge.
Results: 79 ESBL blood isolates were included in our study; 35 received empiric PTZ and 44 received empiric carbapenem. The 14-day mortality was 15 (43%) in the empiric PTZ group vs 5 (11%) in the empiric carbapenem group. In a bivariate Cox regression analysis, patients who received carbapenem had 0.32 the hazard rate of those receiving PTZ (95% CI, 0.13-0.79; p=0.013). Patients on the regular medical floor had a lower hazards than those in the ICU (HR 0.046; 95% CI, 0.12-0.96; p=0.003). Twenty-seven patients were known to be colonized with ESBL-producing organisms; 21 of them received empiric carbapenem therapy and 3 received empiric PTZ; these latter 3 died in the first 14 days.
Conclusion: Our results suggest that, in our instituition, carbapenems should be used as preferred empiric therapy for patients at high risk of ESBL bloodstream infections. Poorer clinical outcome in these patients may be attributable to a delay in initiating appropriate antimicrobial therapy. However, the liberal use of carbapenems is not without consequence and can result in the emergence of drug resistance. The decision to use empiric carbapenem therapy should be carefully considered after factoring in relevant data.