Abstract
Background
Patients labeled with penicillin allergies often receive alternative antibiotics, leading to increased cost, higher risk of adverse events, and decreased efficacy of procedural prophylaxis. However, most of those patients can tolerate a cephalosporin. University of Washington Medical Center – Montlake (UWMC-ML) Interventional Radiology (IR) frequently administer a pre-procedure prophylactic cephalosporin. We worked with the clinicians in IR to develop tools to allow them to better assess penicillin allergies, make the most appropriate antibiotic choice, and update the patient’s allergy documentation.
Methods
We identified all patients who underwent procedures in IR between 2017–2019. Chart review was done to determine the procedures performed, patient demographic information, allergies, allergy documentation, and prophylactic antibiotics received.
In May 2020 we implemented new Clinical Decision Support tools, including an online assessment app (https://tinyurl.com/IRPCNAllAssess) and handouts to guide antibiotic decision making to clinicians in IR.
Results
From 2017 to 2019, 381 patients underwent 958 procedures in IR. Of those, 379 patients underwent 496 procedures for which the recommended first line choice for antibiotic prophylaxis is a cephalosporin. Of patients who received pre-procedure prophylactic antibiotics for those procedures, 15.9% [n=11] of patients with penicillin allergies received the first line antibiotic, compared to 89.9% [n=319] of patients without a reported penicillin allergy. Since implementation, the online app has been used to evaluate 9 patients, of whom 8 had penicillin allergies. All 8 patients safely received the first line antibiotic (3 were delabeled, 4 reported a history of mild reactions, and 1 reported a history of an immediate IgE mediated response to penicillin but safely received cefazolin).
Conclusion
IR evaluates hundreds of patients who may receive prophylactic antibiotics each year. By providing tools to assess penicillin allergies, we were able to improve both their prescribing and de-label patients which will provide a much broader impact on their care than on just their current procedure. Our free tool can be accessed at the website above, and we will demonstrate in person.
Disclosures
All Authors: No reported disclosures
EFFECTIVE COMMUNICATION OF MACHINE LEARNING MODELS IN CLINICAL DECISION SUPPORT TOOLS FOR PAH RISK PREDICTION
