Validation of algorithms to identify elective percutaneous coronary interventions in administrative databases

Area of Science:

• Health Informatics
• Cardiovascular Medicine
• Health Services Research

Background:

• Administrative data often lacks specific codes to distinguish elective from non-elective percutaneous coronary interventions (PCI).
• This ambiguity hinders outcome tracking, medical management assessment, and research on elective PCI patients.
• Accurate identification of elective PCI is crucial for healthcare quality improvement.

Purpose of the Study:

• To evaluate the performance of algorithms in identifying elective PCI procedures using administrative data.
• To determine the most effective algorithm for distinguishing elective PCI from other procedures in healthcare databases.

Main Methods:

• A retrospective study analyzed administrative data from an integrated healthcare system (1/1/2015–6/31/2016).
• PCI encounters were identified, validated via chart review, and classified as elective or non-elective.
• Nine pre-defined algorithms were tested, calculating sensitivity, specificity, and predictive values.

Main Results:

• Of 521 PCI encounters, 497 were valid, with 93 classified as elective.
• An algorithm excluding emergency room visits achieved the highest sensitivity (97.9%).
• An algorithm incorporating prior cardiologist visits and tests showed the highest positive predictive value (62.2%).

Conclusions:

• An algorithm excluding emergency room visits is a practical method for identifying elective PCI in administrative data.
• This approach can improve the ability to analyze and manage patients undergoing elective PCI.
• Further refinement of algorithms may enhance the accuracy of elective PCI identification.