Development of a drug allergy alert tiering algorithm for penicillins and cephalosporins

Introduction

Limited research is available regarding recommendations about which drug allergy alerts (DAAs) in clinical decision support (CDS) systems should interrupt provider workflow. The objective was to evaluate the frequency of penicillin and cephalosporin DAA overrides at two institutions. A secondary objective was to redesign DAAs using a new tiered alerting system based on patient factors.

Methods

A retrospective, observational study evaluated CDS DAA overrides for penicillins and cephalosporins at two large academic medical centers. Included patients were at least 18 years of age and had a penicillin or cephalosporin DAA fired at the time of medication ordering. We developed a rule-based algorithm to classify DAAs into three groups: no alerts presented to user, non-interruptive (informational) alerts, and interruptive alerts requiring a coded response. The rule-based algorithm includes drug class or cross-sensitivity matches and reaction types with designated severities (high, medium, or low).

Results

DAAs for penicillin and cephalosporins were overridden 55% of the time at each institution. Of the DAAs overrides, 85% were cross sensitivity matches and 15% were drug class matches. Reactions were classified as 22% high severity, 29% medium, and 48% low. Most low severity reactions were rash (25%), unspecified reactions with no comments (13%), nausea/vomiting (4%), and GI upset (3%). High severity reactions were mostly other reactions with comments (19%) and anaphylaxis (4%). Approximately 30% of the penicillin and cephalosporin alert overrides could have been non-interruptive alerts based on the penicillin or cephalosporin allergic reaction documented in the EHR at each institution.

Conclusion

The majority of penicillin and cephalosporin DAAs were overridden, largely for cross sensitivity in lower severity reactions. The data can be used to inform DAA redesign, reduce override rates, and improve patient safety.