# Carbapenem Versus Non-carbapenem Therapy in Hematology Patients: Extended-Spectrum Beta-Lactamase Positive Enterobacteriaceae Colonization Impact

## TL;DR

This study examines whether using carbapenem or non-carbapenem antibiotics affects outcomes in hematology patients colonized with drug-resistant bacteria.

## Contribution

The study evaluates the clinical impact of empirical antibiotic choices in patients colonized with ESBL-E, challenging assumptions about colonization leading to infection.

## Key findings

- Rectal colonization with ESBL-E was common, but prospective infections were rare.
- Empirical antibiotic choices based on colonization status did not significantly affect mortality.
- Switching antibiotics was associated with higher mortality, even when bacteria were susceptible.

## Abstract

Background: Extended-spectrum beta-lactamase-producing Enterobacteriacea (ESBL-E) infections are a major source of mortality and morbidity in patients with hematologic cancers. One of the most significant risk factors for bacterial illness is prior colonization with resistant germs. Empiric usage of carbapenems is recommended for the treatment of infections in patients with a positive colonization history.

Objectives: We aimed to determine the outcome of empirical carbapenem (de-escalation) versus non-carbapenem (escalation) therapy in adult hematology patients who have rectal extended-spectrum beta-lactamase positive ESBL-E colonization.

Methods: Two hundred three rectal swab cultures were collected from 130 patients, admission or during hospitalization between June 2014 and May 2015. Patients were followed till January 2016 for future infections due to ESBL-E. Empirical antibiotic treatment was started according to the patient’s medical condition without consideration of previous colonization status. Stable patients received empirical escalation therapy. All-cause and early mortality of infections are analyzed.

Results: Seventy-three (36%) swabs were positive for ESBL-E. Patients with rectal ESBL-E colonization were defined as cases; patients without colonization were defined as controls. Prospective infection due to ESBL-E in the case and control group was 6.8% and 2.3%, respectively. No statistically significant relation was found between colonization and prospective infection due to ESBL-E (p=0.110). There was no all-cause or early mortality in prospective infections with ESBL-E. Among case patients, one patient each died from all-cause mortality in the escalation (n=55) and de-escalation (n=3) group. The all-cause mortality in the antibiotic switch group (n=30) was eight, including five patients in the early mortality group although the bacteriologic agents were susceptible to the given antibiotics.

Conclusion: In our institution, rectal colonization with ESBL-E was high, but contracting an infection due to ESBL-E was surprisingly low. Colonization with ESBL-E may not necessarily end with an infection in some situations. In stable patients, disregarding colonization features before empirical therapy did not seem to be inappropriate.