Clinical Decision Support for Disseminating and Implementing Patient-Centered Outcomes Research Clinical Evidence

There is a growing need to use clinical decision support (CDS) to disseminate patient-centered outcomes research (PCOR)-related guidelines. There is also a need to create interoperable CDS artifacts so that these tools may be able to be used in multiple settings and platforms. This research will demonstrate how interoperable CDS artifacts are developed and disseminated with a use case of CDS applied to the ordering of advanced diagnostic imaging. The Protecting Access to Medicare Act of 2014 requires that, when ordering advanced diagnostic imaging for patients on Medicare, ambulatory providers must consult appropriate use criteria (AUC) through CDS mechanisms. The CDS is to determine when the order complies with AUC, with a goal of helping providers order the most appropriate tests for their patients. 

The researchers created publicly available CDS artifacts via their previous work with the Harvard Medical School Library of Evidence (HMS-LOE). These artifacts are written as a clinical evidence logic statement (CELS) in “if-then” formats and graded using a system developed by the Oxford Centre for Evidence-Based Medicine for strength of evidence. Additionally, 18 percent of the CELS in the HMS-LOE go beyond the ‘if-then’ format and use standardized formats, such as the use of SNOMED CT, Fast HealthCare Interoperability Resources (FHIR), and Clinical Quality Language (CQL).  

In this research, investigators will examine three PCOR guidelines from HMS-LOE: the use of breast MRI to diagnose a second breast cancer in a woman with prior breast cancer, the use of CT scan for pulmonary embolism (PE), and education for those at risk of PE. For each, they will examine the ability of FHIR, CQL, and SNOMED CT to represent them, translate them into CDS artifacts, implement them into a separate healthcare system with a separate EHR, identify lessons learned about scalability, and measure impact on guideline adherence.

The specific aims of the research are as follows:

• Assess the technical capability of FHIR, CQL, and SNOMED CT for representing CDS based on three PCOR guidelines.
• Validate PCOR CELS represented in a structured representation and evaluate scalability in two different workflows and settings.
• Scale previously developed CDS artifacts to a different healthcare system, with a different electronic health record (EHR) and health information technology platform, end users, and workflows.
• Measure the impact of disseminating the CDS by assessing the extent to which CDS recommendations are accepted.

For the three areas, CDS hooks will be used to create pop-up reminders in the EHR to: order an MRI whenever an order for a mammogram without MRI is entered in patients with prior breast cancer, order a D-dimer for CT pulmonary angiogram when appropriate, and discuss PE with a link to print materials. After validating in one healthcare system, the researchers will then implement and evaluate them in a second healthcare system in another State, using a different EHR, with different patients and providers, and utilizing different workflows. The CDS artifacts will be disseminated further by adding them to the CDS Connect repository, as well as their own HMS-LOE.