Methods: University of Maryland Medical Center (UMMC) is an 800-bed tertiary care hospital with an average of 600 antimicrobials prescribed daily. The UMMC ASP consists of 2 part-time adult Infectious Diseases (ID) physicians and one clinical ID pharmacist. ID consults on 70% of patients prescribed IV antibiotics. An antibiotic report is generated daily containing patient identifiers, antibiotic, dose, frequency, primary service, prescriber, and ID consulting team. Using Excel, the list is abstracted to generate a database for analyzing antibiotic appropriateness, defined as ASP agreement with original prescription, and for recording interventions. Beginning in October 2014, the ASP implemented a modified 48-hour antibiotic timeout (M48AT) for all antibiotics prescribed without ID consult as well as a subset of restricted antibiotics in all patients. Newly prescribed restricted antimicrobials are reviewed within 24 hours.
Results: There were 1012 antibiotic reviews in adult patients during a 6-month period (10/1/14 – 3/31/15). The majority of reviews were for M48AT without ID consult (75%), followed by M48AT with ID consult (14%) and bug-drug mismatches (2%). Most of the reviewed antibiotics were appropriate requiring no intervention (54%). Of the 469 interventions, 43% were ID consult requests, 28% were de-escalate therapy, 10% were change to effective therapy or escalate therapy, and 4% were change in duration of therapy. In addition, we identified a method of ascertaining specific prescriber patterns with attention to those prescribers that deviate from standard best practice.
Conclusion: Our ASP model is a proof of principle demonstrating a functional ASP using Excel in a setting of resource limitations. Further work is needed in implementing education and feedback to prescribers, as well as assessing the success of our program with outcome metrics.
L. Pineles, None
M. Kleinberg, None