Clinical decision support (CDS) for electronic health records (EHR) and prescribing systems has been promoted to improve patient outcomes. One type of CDS are drug-drug interaction (DDI) alerts. The Office of the National Coordinator for Health IT meaningful use criteria includes the implementation of DDI detection and warnings to physicians and other healthcare professionals. Nearly all healthcare organizations rely on DDI alerts generated from commercial drug knowledge databases. Warnings are currently generated using simple drug combination rules, ignoring drug attributes and the wealth of information available in the EHR that could make the warnings specific to the patient. As a result, providers are bombarded with useless warnings and often miss important ones. Our approach is to change the framework for DDI alerting from basic look-up tables to a more complex, but meaningful, clinical algorithms. Our plan is innovative because it will: 1) eliminate alerts for DDIs that are not clinically important given the patient and drug context; 2) develop implementable and tested algorithms using existing and new evidence; and 3) support the dissemination, implementation, and evaluation of these algorithms across the spectrum of healthcare facilities and organizations. The central hypothesis of this project is that individualizing DDI alerts to specific patient circumstances will result in a much greater proportion of alerts that physicians, pharmacists, and other healthcare providers will be more likely to heed. We will accomplish our objectives and test our hypothesis by pursuing the following aims:
Specific Aim 1 : Design sharable evidence-based individualized DDI algorithms that capitalize on the wealth of patient data located within electronic health records;
Specific Aim 2 : Validate the function of newly designed DDI algorithms using electronic health record data;
and Specific Aim 3 : Conduct a prospective evaluation of DDI algorithms in a variety of healthcare environments including ambulatory and institutional settings. This project will greatly improve CDS for DDIs by incorporating contextual factors into evidence-based and validated alert algorithms, which will reduce alert fatigue and result in more meaningful CDS. Our approach, involving partners across multiple organizations and environments and experts in drug interaction and biomedical informatics, will result in safer healthcare with respect to the use of medications.
The promise of computer systems to improve patient safety has remained largely unfulfilled with respect to drug interaction warnings. Current alerts are simplistic and fail to include specific patient factors, leading to excessive warnings about interactions that are not relevant and healthcare facilities need assistance to implement evidenced-based clinical decision support. This project will develop computer rules that will provide relevant warnings about drug interactions that have been tailored to the specific patient and seek to implement the rules in over 50 healthcare facilities across the United States.
