Phase 3

Phase 3: Reviewing algorithm performance and refining as needed

The last phase is the periodic assessment of the matching algorithm to determine if refinements would improve its performance and of the matching process to ensure that it continues to perform effectively and efficiently over time. These reviews will likely need to be more frequent early in the implementation of person marching, such as annually, and become less frequent as the process matures. This review process repeats steps described in Phase 1: Characterize the Data and Determine Matching Algorithm. Including the following areas:

Personnel and communication

• Ensure the list of responsible personnel is up to date.

• Ensure the communication channel used for messages is up to date.

Review identifiers and data source limitations (Phase 1, Step 1A and 1B)

• Determine appropriate identifier combinations

• Address data limitations

Linkage validation (Phase 1, Step 1B).

• Perform a quick linkage review.

• Perform a formal linkage review.

Tuning of matching algorithm (Phase 1, Step 1C).

• Assess the balance between precision and recall and adjust as needed.

• Review the run-time of the linkage process.

A final step: celebrate!

As this module has described, designing and implementing a person matching process can seem complicated, involving many steps and many people. However, once the process has been established and potential duplicates adjudicated, take a moment to celebrate reaching this milestone and acknowledge all those who have made it possible. The process described here will help to ensure that each person is uniquely represented within or across systems and datasets and forms the basis for ongoing improvement of the system, future linkages, and potential expansion to additional data sources — all to facilitate improved patient care and accurate population health assessments. Congratulations!