Enterprise Master Patient Index (EMPI)

The core function of an EMPI is to link a single patient's identifiers from multiple independent systems—such as a hospital EHR, a primary care practice management system, and a regional lab—into one unified view. Unlike a standard MPI confined to a single organization, an EMPI operates across the entire health system or Health Information Exchange (HIE). It ingests demographic feeds from each source and maintains a persistent enterprise identifier that cross-references all local medical record numbers (MRNs).

EMPI systems rely on sophisticated probabilistic matching algorithms rather than simple deterministic rules. Because real-world patient data is messy—containing typos, nicknames, transposed dates, and missing fields—the engine calculates a statistical likelihood that two records belong to the same person.

• Weighted field comparison: A match on Social Security Number is weighted more heavily than a match on gender.
• Fuzzy logic: Handles phonetic variations (e.g., 'Jon' vs 'John') and common misspellings.
• Threshold tuning: System administrators configure match thresholds to balance duplicate creation against false merges.

Once records are linked, the EMPI constructs a golden record—a single, best-source composite of truth for that patient. This is not a copy of the full clinical record but a curated set of core demographics.

• Survivorship rules: Logic determines which source system's data is most trusted (e.g., the state immunization registry's birth date over a patient-entered portal value).
• Persistence: The enterprise ID and its cross-reference map remain constant even if source system data changes.
• Downstream synchronization: Updates to the golden record can be broadcast back to subscribing systems to correct local demographic errors.

A critical operational workflow is the continuous identification and merging of duplicate patient records. EMPIs run both real-time matching during patient registration and batch processes to catch historical duplicates.

• Potential duplicate queues: High-probability matches that fall below the auto-merge threshold are flagged for manual review by a data steward.
• Unmerge capabilities: A robust EMPI allows for the reversal of an incorrect merge, restoring the original distinct records without data loss.
• Overlay prevention: The system actively prevents the creation of a new patient record if an existing match is found during an ADT (Admit, Discharge, Transfer) feed.

The EMPI issues and governs a unique, non-semantic enterprise identifier (EUID) for each patient. This identifier has no embedded meaning (unlike a coded MRN) and serves solely as a persistent pointer.

• ID integrity: The EUID is never reassigned or recycled, even if the patient dies.
• Cross-reference table: The EMPI maintains a dynamic lookup table mapping every local MRN from every source system to the single EUID.
• Query by identifier: External systems can query the EMPI using a local MRN to retrieve the EUID and the full set of linked identifiers for that patient.

Modern EMPIs expose their identity services through standard interoperability protocols to integrate seamlessly into the clinical workflow.

• IHE PIX (Patient Identifier Cross-Referencing): An integration profile that defines how systems query an EMPI for a list of correlated patient identifiers using HL7 v2 or v3 messages.
• IHE PDQ (Patient Demographics Query): Allows a client system to search for a patient by demographics and retrieve the full set of identifiers.
• FHIR $match operation: A RESTful API endpoint that accepts demographic parameters and returns a confidence-scored match result, aligning the EMPI with modern SMART on FHIR architectures.