Improving Quality through Decision Support for Evidence-Based Pharmacotherapy

This project developed a decision support system for medication management. The goal was to promote increased adherence to evidence-based pharmacotherapy (EBP) guidelines both through traditional clinic-based and new models of care, including population health management and cross-disciplinary teams. The decision support system used in this project, known as the System for Evidence-Based Advice through Simultaneous Transaction with an Intelligent Agent across a Network (SEBASTIAN), is the basis for an international Health Level 7 standard for clinical decision support using a service-oriented architecture. Increased availability and use of decision support tools for medication management can be expected to reduce medication errors, improve health care quality at an acceptable cost, and augment disease management for patients and populations.

The project aims were to:

• Expand the functionality of an existing decision support system in use within a regional HIE network for Medicaid beneficiaries to incorporate EBP and to promote medication adherence.
• Implement and evaluate the impact of two complementary interventions for medication management on adherence to EBP among Medicaid beneficiaries in ambulatory care settings through a three-arm randomized control trial.
• Compare resource utilization and assess the economic attractiveness of the interventions to promote medication adherence and EBP.
• Disseminate information regarding the development and impact of the interventions through Web teleconferences, professional meetings, educational lectures, and peer-reviewed journals.

Two interventions were created to detect evidence-based indications for nine classes of medications based on the presence of diabetes, hypertension, asthma, congestive heart failure, ischemic vascular disease, or stroke. The first was a clinic-directed intervention that generated patient-specific reports sent to clinics 1 day prior to a scheduled appointment. These reports displayed a 1-year list of filled prescriptions with summaries of adherence and recommendations for missing EBP. The second intervention was population-oriented, whereby weekly notices were sent to care managers about patients who appeared to be nonadherent and had no record of a clinic visit. To evaluate these interventions, 2,219 Medicaid beneficiaries with at least one of the conditions were randomly assigned to usual care, reports alone, or reports plus care manager notices.

Neither the reports alone nor the reports with the notices groups improved adherence to EBP compared to usual care. No improved adherence was detected for any individual class of medication or for any individual condition. The group randomized to receive notices however, did have significantly increased contact with care managers, demonstrating the potential to address EBP nonadherence at the population level. Site visits, contextual evaluation, and user surveys suggested that the failure to improve adherence to EBP resulted from insufficient time for clinicians to be able to address medication adherence during the clinical encounter. Nonadherence issues might be better addressed by members of the care team outside the clinic, including pharmacists or advanced nurses trained in medication nonadherence.

Despite the lack of improvement, this decision support enhanced population-level approach may prove to be a viable care model that could be utilized to shift care away from the current episodic clinic-based, clinician-centered approach. Expansion of population surveillance with augmentation of data available through health information exchanges could allow more extensive patient-focused care management external to clinics in venues such as the patient’s home. These new models of care could lower costs and increase accessibility because they are not dependent on scarce and expensive clinic and clinician resources. The value of the medication management project intervention is indicated by the fact that the Medicaid care management program paid to operationalize the system to keep it functioning after the grant term ended. In April 2012, the application was successfully integrated into the electronic health record for Duke University Health System; the sustainability of the project is an indication of success.

The study team concluded that resources should be invested to explore how the management of the health of a population enabled through decision support can be extended to improve the coordination, quality, efficiency, and even outcomes of health care delivery in the United States.