Drug-Drug Interaction Alerts: Emphasizing the Evidence

Many analysts and users of contemporary clinical decision support ("CDS") systems have expressed grave concerns about the technology’s efficacy and functionality. Alerts generated by CDS systems are often inaccurate, and an excess of alerts leads some physicians to experience "alert fatigue" and to turn off CDS altogether. This article formulates recommendations to improve drug-drug interaction (DDI) alerts. The paper comments upon a proposal by Susan Ridgely and Michael Greenberg, who call for the development of a consensus-based "clinically significant drug-drug interaction list" that could generate limited liability protection for users. We argue that instead of creating a list of always-contraindicated DDIs, experts should develop DDI alerts that offer essential information about DDI risks, supporting evidence, mitigating factors, and appropriate courses of action. Thus, DDI warnings should provide users with concise but comprehensive information. They should not deprive clinicians of discretion, but rather, enable them to make more knowledgeable and effective prescribing decisions. In addition, the article analyzes several other DDI-related issues. It details a process for determining which DDIs should generate alerts in CDS systems. It also examines the extent to which DDI alerts should serve as a basis for liability protection and suggests how data about DDI alert accuracy could be used as evidence in malpractice litigation.