Intervention to Reduce Broad-Spectrum Antibiotics and Treatment Durations Prescribed at the Time of Hospital Discharge: A Novel Stewardship Approach

Abstract

OBJECTIVE For most common infections requiring hospitalization, antibiotic treatment is completed after hospital discharge. Postdischarge therapy is often unnecessarily broad spectrum and prolonged. We developed an intervention to improve antibiotic selection and shorten treatment durations. DESIGN Single center, quasi-experimental retrospective cohort study METHODS Patients prescribed oral antibiotics at hospital discharge before (July 2012-June 2013) and after (October 2014-February 2015) an intervention consisting of (1) institutional guidance for oral step-down antibiotic selection and duration of therapy and (2) pharmacy audit of discharge prescriptions with real-time prescribing recommendations to providers. The primary outcomes measured were total prescribed duration of therapy and use of antibiotics with broad gram-negative activity (ie, fluoroquinolones or amoxicillin-clavulanate). RESULTS Overall, 300 cases from the preintervention period and 200 cases from the intervention period were included. Compared with the preintervention period, the use of antibiotics with broad gram-negative activity decreased during the intervention (51% vs 40%; P=.02), particularly fluoroquinolones (38% vs 25%; P=.002). The total duration of therapy decreased from a median of 10 days (interquartile range [IQR], 7-13 days) to 9 days (IQR, 6-13 days) but did not reach statistical significance (P=.13). However, the duration prescribed at discharge declined from 6 days (IQR, 4-10 days) to 5 days (IQR, 3-7 days) (P=.003). During the intervention, there was a nonsignificant increase in the overall appropriateness of discharge prescriptions from 52% to 66% (P=.15). CONCLUSIONS A multifaceted intervention to optimize antibiotic prescribing at hospital discharge was associated with less frequent use of antibiotics with broad gram-negative activity and shorter postdischarge treatment durations. Infect Control Hosp Epidemiol 2017;38:534-541.

Conflict of interest statement

Conflicts of interest: No relevant conflicts of interest to disclose.

Figures

Figure 1

Study diagram. In the pre-intervention study period, 1825 adult cases were identified electronically. A total of 376 charts were manually reviewed until 300 included cases was reached. In the intervention period, 777 adult cases were identified electronically. A total of 288 charts were manually reviewed until 200 included cases was reached. 50 included cases from each period were randomly selected for blinded appropriateness review.