Patient matching is the process of identifying and linking records that belong to the same person across systems, organizations, or data sources. It affects interoperability, care coordination, eligibility, analytics, and AI workflows that depend on complete or accurate patient data.
Healthcare AI buyers should ask how matching errors, duplicates, incomplete records, and uncertain matches are detected, reviewed, and audited.
Application scenario: In workflow review, this term helps teams map a vendor claim to the care setting, data flow, integration point, user handoff, and oversight step where it applies. Procurement impact: Buyers should evaluate evidence, interoperability effort, security and privacy controls, pricing assumptions, support, and compliance responsibilities before shortlisting or contracting for a tool that depends on this capability.
Sources and review notes
These links support definition-level research and do not establish the regulatory status, safety, or suitability of any product.
ONC defines patient matching as identifying and linking one person's data within and across health systems and notes that implementations commonly compare multiple demographic fields such as name, birth date, phone number, and address. ONC's Interoperability Standards Advisory lists production and emerging standards for demographic query, identifier cross-reference, and cross-organizational FHIR matching, with different maturity, adoption, federal-requirement, and test-tool statuses that buyers should not treat as equivalent. ONC's registrar playbook emphasizes accurate, consistently verified demographic capture, and the Project US@ specification standardizes United States domestic and military address representation to support matching while explicitly leaving other addresses and a complete matching algorithm out of scope. ONC's 2014 final report documents the threshold tradeoff between false-positive and false-negative matches and discusses duplicate reports and correction workflows; it remains background evidence rather than a current product benchmark. The 2025 SAFER Patient Identification guide provides a current safety-review resource for organizations using electronic health records. These sources do not validate a vendor, prescribe one algorithm or universal threshold, make standardized addresses sufficient by themselves, or prove that a linked record belongs to the same person. Evaluation should use adjudicated reference pairs representative of local sources and separately report true matches, true nonmatches, false matches, missed matches, uncertain or manually reviewed pairs, duplicate creation, erroneous merges, and unresolved records, with precision, recall, and threshold behavior tied to each workflow's harm. Teams should test transposed and missing fields, name and address changes, aliases, multiple births, shared contact information, language and naming conventions, source-system formatting, and cross-organization data loss; preserve source identifiers, match scores, rule or model versions, reviewer actions, merge and unmerge history, and downstream corrections; restrict demographic access; and provide quarantine, escalation, rollback, and patient-record correction paths before automated linking affects care, billing, exchange, or analytics.
