Algorithmic Detection of Boolean Logic Errors in Clinical Decision Support Statements.

Area of Science:

• Health Informatics
• Clinical Decision Support Systems
• Electronic Health Records

Background:

• Clinical decision support (CDS) systems are crucial for healthcare quality and safety.
• Errors within CDS systems, particularly those using Boolean logic, can lead to unintended patient safety issues.
• Accurate specification of Boolean logic in CDS is challenging for analysts.

Purpose of the Study:

• To determine the prevalence of specific Boolean logic errors in CDS statements.
• To evaluate the effectiveness of an automated algorithm in detecting these errors.

Main Methods:

• Nine healthcare organizations extracted Boolean logic statements from their Epic electronic health records (EHR).
• An open-source software tool utilizing the Espresso logic minimization algorithm was developed to identify logic errors.
• Three classes of Boolean logic errors were targeted for detection.

Main Results:

• A total of 260,698 logic statements were submitted, with 44,890 minimized by Espresso.
• Errors were identified in 209 logic statements across all participating organizations.
• Every organization reported at least two errors and expressed willingness to address them.

Conclusions:

• Automated algorithms can effectively detect specific categories of Boolean CDS logic errors.
• These errors, though a minority, necessitate correction to prevent patient safety issues.
• The identified errors, while few per site, indicate a widespread problem affecting all participating organizations.