Development of an Alert System to Detect Drug Interactions with Herbal Supplements using Medical Record Data

While potential medication-to-medication interaction alerting engines exist in many clinical applications, few systems exist to automatically alert on potential medication to herbal supplement interactions. We have developed a preliminary knowledge base and rules alerting engine that detects 259 potential interactions between 9 supplements, 62 cardiac medications, and 19 drug classes. The rules engine takes into consideration 12 patient risk factors and 30 interaction warning signs to help determine which of three different alert levels to categorize each potential interaction. A formative evaluation was conducted with two clinicians to set initial thresholds for each alert level. Additional work is planned add more supplement interactions, risk factors, and warning signs as well as to continue to set and adjust the inputs and thresholds for each potential interaction.