1789. Inpatient Penicillin Skin Testing: Outcomes From a Propensity-Matched Case–Control Study

BACKGROUND: Nearly 10% of patients report an allergy to penicillin, yet fewer than 10% are confirmed to have a true allergy. Reported allergy frequently leads to the use of costly, broad-spectrum or less-effective antibiotics. We launched a penicillin skin testing (PST) service offering real-time skin testing for inpatients. Here we present clinical outcomes for the first 80 consecutively tested cases compared with propensity-matched controls. METHODS: PST was performed on 80 adults with a reported penicillin allergy admitted to Duke University Hospital between November 2016 and March 2018. A logistic regression model predicting PST receipt was developed using a cohort of penicillin-allergic, untested adults. Covariates included age, gender, diagnosis, and Charlson co-morbidity index. Using this model, the PST cases were propensity-matched 1:1 with untested, penicillin-allergic controls admitted in the preceding year (October 2015–October 2016). Rates of first-line antibiotic receipt were compared between PST cases and their propensity-matched controls. RESULTS: PST cases and controls had similar demographics, reported allergies, diagnoses, and co-morbidities. Cases were more likely to receive a first-line antibiotic (83% vs. 57%, P = 0.003, Table 1). Rates of clinical cure were similar between groups. Ninety-day recurrence and C. difficile infection were numerically higher in the untested group but did not reach statistical significance. A single allergic reaction (rash upon receipt of a cephalosporin) occurred in the PST group. CONCLUSION: Penicillin skin-testing significantly increased the proportion of patients receiving first-line antibiotics. While rates of recurrence and C. difficile infection were lower for skin-tested patients, these differences did not reach statistical significance. As this study was not expressly powered to detect such differences, we plan to reassess these outcomes once we have accrued a sufficiently large cohort of tested patients. DISCLOSURES: All authors: No reported disclosures.