A Case Report in Using a Laboratory-Based Decision Support Alert for Research Enrollment and Randomization.

Area of Science:

• Clinical Informatics
• Rheumatology
• Health Services Research

Background:

• Positive antinuclear antibody (ANA) tests are common but do not always indicate autoimmune disease.
• Predictive models can identify individuals at higher risk for developing autoimmune conditions.
• Pragmatic studies require efficient methods for participant randomization within electronic health records.

Purpose of the Study:

• To identify barriers in implementing a custom clinical decision support (CDS) alert for participant randomization in a pragmatic study.
• To evaluate the feasibility of using a CDS alert triggered by positive ANA tests for autoimmune disease risk prediction and early referral.

Main Methods:

• Integrated a validated logistic regression model into the electronic health record (EHR) to predict autoimmune disease risk in ANA-positive individuals (titer ≥ 1:80).
• Developed a silent, custom CDS alert to randomize eligible individuals into a pragmatic study.
• Randomized participants into an intervention arm (risk score review, provider notification, expedited referrals) and a control arm (standard care).

Main Results:

• Successfully assessed 3,961 individuals and randomized 2,105 since June 2023.
• Encountered technical challenges including laboratory vendor changes, equipment failure, and altered test names, which impacted CDS alert functionality.
• Demonstrated the feasibility of a laboratory-based CDS alert for randomization across a large healthcare system.

Conclusions:

• Laboratory-based custom CDS alerts can facilitate participant randomization in pragmatic studies within large healthcare systems.
• Close collaboration with laboratory teams and a deep understanding of laboratory testing workflows are essential for successful CDS alert implementation.
• Addressing technical challenges proactively is key to ensuring the continuous operation of CDS alerts in research settings.