Morgan DJ, Johannsson B, Schweizer ML, Braykov NB, Weisenberg SA, Uslan DZ, Kelesidis T, Young H, Cantey JB, Septimus EJ, Srinivasan A, Perencevich EN, Laxminarayn R. The Frequency Of Antibiotic De-Escalation Over Six US Hospitals: Results from a Multicenter Cross-Sectional Study (Poster #770). IDWeek. October 2012. San Diego, CA.

ABSTRACT

Background:

Use of patient-specific culture data to optimize empiric therapy is a cornerstone of rational hospital antibiotic use. The frequency with which cultures are obtained and therapy tailored to results is unknown.

Methods:

We performed a cross-sectional study using retrospective chart review of 1,200 adult inpatients, hospitalized >24hrs, with >=1 active antibiotic order. Patients were enrolled for 4 index dates at quarterly intervals during a 1-year study period (9/2009-10/2010). Infectious disease (ID) specialists recorded demographics, comorbidities, antibiotic therapy, imaging studies and culture results in a 17d window, and categorized changes to therapy. No change was defined as the continuation of the course as initially ordered; de-escalation was a change resulting in narrower coverage; escalation was a switch to/addition of an antibiotic resulting in broader coverage. A Cox proportional hazard model stratified by infection site was used to model time to de-escalation. Patients receiving <=1 antibiotic prescription and/or exclusively prophylactic courses were excluded from the analysis.

Results:

Of 1,200 charts that were reviewed, 631 patients(52.6%) were included in the analysis. Of these, 288 (45.7%) were not changed, 192 (30.4%) were de-escalated and 151 (24%) were escalated. De-escalated prescriptions included 18 fully discontinued courses  (2.85%), 61 de-escalations without culture results (9.7%), and 113 de-escalations based on cultures (17.9%). The no-change category included 250 continued as initially ordered (39.6%) and 38 switches to equivalent antibiotics (6%). De-escalation was most common for urinary infections (46%). Patients that received fewer prescriptions, were started on broad-spectrum antibiotics, had elevated WBC at start of course, shorter pre-therapy LOS had higher probability of de-escalation. However, positive culture and imaging study had no significant effect.

Conclusion:

Although patients with suspected infections were frequently cultured, clinicians changed antibiotics in less than half of patients receiving multiple therapies. Availability of positive culture and/or imaging study suggestive of infection did not have a significant impact on de-escalation probability.