Identifying and reducing inappropriate use of medications using Electronic Health Records

Inappropriate use of medications (IUM) is a global problem that can lead to unnecessary harm to the patients and unnecessary costs across the health care system. Identifying and reducing IUM has been a long-lasting challenge and currently, no systematic and automated solution exists to address it. IUM can be manually identified by experts using medication appropriateness criteria (MAC). In this research I first conducted a review of approaches used to identify IUM and reduce IUM. Next, I developed a conceptual model for representing the MAC, and then developed a tool and a workflow for translating the MAC into structured form. Because indications are an important component of the MAC, I conducted a critical appraisal of existing knowledge sources that can be used to that end, namely the medication-indication knowledge-bases. Finally, I demonstrated how these structured MAC can be used to identify patients who are potentially subject to IUM and evaluated the accuracy of this approach. This research identifies the knowledge gaps and technological challenges in identifying and reducing IUM and addresses some of these gaps through the creation of a representation for MAC, a repository of structured MAC, and a set of tools that can assist in evaluating the impact of interventions aimed to reduce IUM or assess its downstream effects. This research also discusses the limitations of existing methods for executing computable decision support rules and proposes solutions needed to enhance these methods so they can support implementation of the MAC.