Using causal rule mining to identify opportunities for value improvement in regional CABG care: A proof-of-concept...

Area of Science:

• Cardiac Surgery
• Medical Informatics
• Causal Inference

Background:

• Cardiac surgery outcome monitoring is crucial for quality improvement.
• Identifying causal patterns behind adverse events and high care consumption remains challenging.

Purpose of the Study:

• To evaluate Causal Rule Mining (CRM) as a method to complement outcome monitoring in cardiac surgery.
• To identify potential causal patterns explaining undesired clinical outcomes and elevated care consumption.

Main Methods:

• CRM was applied to data from 1,068 patients undergoing elective isolated coronary artery bypass grafting.
• Outcomes included mortality, stroke, deep sternal wound infection, re-explorations, reinterventions, and length of stay.
• Relevant causal rules were defined by odds ratio and confidence interval, and reviewed by clinical experts.

Main Results:

• CRM identified 114 significant rules, with 7 deemed particularly interesting.
• Three rules warrant further investigation: absence of fractional flow reserve (FFR) testing linked to fewer reinterventions, and red blood cell (RBC) transfusion linked to both reduced and increased re-explorations.
• Clinical experts validated the potential relevance of these findings.

Conclusions:

• Causal Rule Mining offers a structured approach to uncover potential causal explanations for clinical outcomes.
• CRM provides valuable input for hypothesis-driven quality improvement initiatives in cardiac surgery.
• This method supports efforts to enhance patient value and optimize healthcare delivery.