Master Person Index: Considerations When Transitioning to a New Solution

The backbone of an All-Payer Claims Database (APCD) is a strong Master Person Index (MPI) solution that allows data users to reliably track individuals over time and across payers and data systems. This ability is fundamental for many use cases (especially research and evaluation) and is often a central justification for creating and operating data resources.

Over time, as data systems grow, new data management structures are required. As part of this process, when transitioning to a new MPI solution, data administrators should take special steps to ensure consistency in this essential process.

These data systems can encounter two common transition scenarios:

1. Processes and algorithms unique to the vendors: Each vendor may rely on and/or prioritize different data elements for assigning a unique member ID, so that matches may not be consistent between solutions. A person with an “old” MPI could seem to vanish or end care on the transition date and appear as a totally “new” person entering the system after the transition date.
2. Moving from hashed or encrypted member identifiers to “live” or “direct” identifiers: Some systems remove selected direct identifiers as the data files are initially submitted. This hashed/encrypted data only supports exact matches. In contrast, direct identifiers enable stronger MPI results with probabilistic matching (also known as “fuzzy matching”). The challenge here is to maximize the match rate between the “old” and the “new” MPIs.

When transitioning from one MPI solution to another, data managers should consider the following actions during the implementation phase.

• Data quality strategy: Before the transition, identify the key performance indicators that will be used to evaluate the new MPI solution compared to the current MPI solution to assess the sensitivity (percentage of true positives) and specificity (percentage of true negatives) of both solutions.
• Consistency of Unique Member IDs over time: Changing the MPI introduces risks in consistently tracking members across multiple time periods. The transition plan should include strategies to move from the old to the new system such that longitudinal records are accurately maintained. Data managers should consider policies and processes such as:

• • Mapping previous unique IDs to newly assigned IDs. This is especially helpful when IDs have previously been assigned using hashed or encrypted data. To implement, the MPI vendor receiving direct identifiers would hash or encrypt the data using the same methodology applied under the previous MPI solution and create a crosswalk between the historical hashed/encrypted data and the new MPI created with direct identifiers. The resulting crosswalk is then used to assign unique IDs across the entire data set. This also allows for improving the MPIs previously assigned using only hashed/encrypted data.
  • Rerunning historical member data through the new MPI process. This would provide one unique and consistent identifier using the same methodology over the entire data set.
  • Consider whether any adjustments need to be made to the sequencing methodology used to assign the MPIs across solutions. The more the method and logic change, the more it makes sense to reprocess historical data through the new MPI solution.
  • Creating rules for handling duplicate records and how to classify them. Consider whether historical identifiers should be maintained as is, or if it is acceptable to merge or split these records based on the new MPI solution.

• Solution to link to other data sources: In addition to the impact of MPI transition on an APCD, other changes may be required to support linkages to other data sources. Questions to consider include:
  • Does the format of the finder/cohort files need to be revised?
  • Does the new MPI solution rely on new required data elements?
  • How will changes to existing approved linkages across solutions be addressed?

• Communication to stakeholders: Data managers should provide clear communication and advance notice to data users (internal as well as external researchers who receive ongoing data refreshes) about the transition and start discussing how a new MPI might impact their research and analytic projects.

To learn more about how FHC can help your program efficiently and effectively transition to a new data management vendor, contact us at [email protected] to connect with one of our experts.