IDIA - Individualized Drug Interaction Alerts Drug-drug interactions (DDIs) are a preventable medical error and numerous studies show that up to 5% or more of patients receive combinations that could be harmful. Clinical decision support in electronic prescribing systems has been promoted to prevent DDIs and improve patient safety. However, promises of improved efficiency and patient outcomes remain unfulfilled due to the excessive volume of DDI alerts and the perception that these warnings are irrelevant or unhelpful. These alerts are generated using proprietary drug knowledgebases that were not initially envisioned to support DDI checking, but rather pricing and inventory systems. Alerting systems employ simple drug combination rules, ignoring drug attributes and the wealth of information available in the electronic health record (EHR) to make the warnings more specific to the patient. Existing software fails to incorporate factors that influence the risk of an advers drug reaction such as the dose, route of administration, duration of therapy, and concomitant therapies. The simplistic logic of these systems also ignores patient-specific characteristics that influence an individual's susceptibility to adverse drug reactions, such as genetics, age, and renal function. The lack of specificity has resulted in clinicians being inundated with interaction alerts that are irrelevant - leading to widespread desensitization to DDI warnings. Our research has identified numerous gaps in the patient safety net for DDIs. To close these gaps, we propose to change the underlying framework for DDI alerting from basic look-up tables to a more contextual knowledgebase and accompanying rules architecture.
In specific Aim 1 we plan to construct a knowledgebase for the most commonly overridden DDI alerts. The knowledgebase will incorporate evidence-based modifying factors that increase or decrease the risk of patient harm.
In Aim 2, we will construct clinical algorithms that extract and use data fro an existing commercially available EHR system and integrate this information with the DDI knowledgebase. Once these components have been developed, in Aim 3 we will evaluate and validate the algorithms using simulated and actual data from inpatient admissions. The development of a DDI-specific knowledge database combined with clinically validated algorithms will increase the specificity of warnings concerning dangerous drug combinations. These developments have the strong potential to drastically reduce the occurrence of irrelevant alerts while simultaneously improving patient safety.
Despite the expanding use of computer systems in healthcare delivery, the promise of these systems to improve patient safety has remained largely unfulfilled with respect to drug interaction warnings. Current alerts are too simplistic and fail to include specific patient factors that increase or decrease the risk of harm. This project seeks to develop a knowledgebase and computer rules that will provide relevant warnings about drug interactions that have been tailored to the specific patient.
