Architectures for person matching and de-duplication

One of the most critical issues within a healthcare systems environment is the strategy for matching and de-duplicating person records between systems. This is especially true for organizations that find themselves with patient data scattered around various systems, either because they were acquired or built over time, or because they represent “best of breed” solutions deployed in their respective subject areas. Even an organization which opts to go for a more streamlined integrated system solution likely still has many interfaces to ancillary and secondary systems, including some outside of the enterprise. Typically, a Master Patient Index, or MPI, is created or acquired to match records or de-duplicate sets of records that appear to have the same patient represented more than once. The presentation offers four models for enterprise-wide matching and de-duplication architectures. Both the strengths and weaknesses of each model, as well as pertinent examples from public health implementations, will be presented. While the models are general, and do not represent product-specific implementations, the functionality that can be achieved under any of these options may differ if different matching/de-duplication products are used. The culture and system strategy of an organization go a long way in determining which approach to de-duplication/record matching services represents a good fit. Organizations whose systems are more distributed among programs or departments have more to gain from thinking about these different strategies. Organizations with a more centralized approach to system deployment likely already have struggled with these issues in the course of their system integration planning. These same models can be applied to health information exchange/regional health information organization planning with many of the same considerations. Depending on how loosely-coupled or tightly-coupled an HIE project wants to be can help determine what strategy they employ for record matching and de-duplication.