Real-World Matching Performance of Deidentified Record-Linking Tokens

Abstract

Objective: To evaluate tokens commonly used by clinical research consortia to aggregate clinical data across institutions. Materials and Methods: This study compares tokens alone and token-based matching algorithms against manual annotation for 20,002 record pairs extracted from University of Texas Houston (UTH)’s clinical data warehouse in terms of entity resolution. Results: The highest precision achieved was 99.9% with a token derived from the first name, last name, gender, date-of-birth. The highest recall achieved was 95.5% with an algorithm involving tokens that reflected combinations of first name, last name, gender, date-of-birth, and social security number. Discussion: To protect the privacy of patient data, information must be removed from a healthcare dataset to obscure the identity of individuals from which that data were derived. However, once identifying information is removed, records can no longer be linked to the same entity to enable analyses. Tokens are a mechanism to convert patient identifying information into HIPAA-compliant de-identified elements that can be used to link clinical records, while preserving patient privacy. Conclusion: Depending on the availability and accuracy of the underlying data, tokens are able to resolve and link entities at a high level of precision and recall for real-world data derived from a clinical data warehouse.

Real-world analysis of healthcare resource utilization by patients with X-linked myotubular myopathy (XLMTM) in the United States

“…The use of deidentified tokens for record matching across research consortia and between identified research databases and anonymized public databases has been growing [ 18 ], and a recent study reported 99% precision for matching among 20,002 record pairs when first name, last name, gender, and date of birth were tokenized [ 19 ]. However, as this analysis employed a novel use of artificial intelligence capabilities on both commercial and CMS payor databases, namely Medicaid claims, it has inherent limitations.…”

Using Electronic Health Records and Linked Claims Data to Assess New Medication Use and Primary Nonadherence in Rheumatology Patients

“…ENRGY is a national rheumatology practice–based research network that includes more than 700 community rheumatologists throughout the United States. The EHR data were linked to Medicare pharmacy and medical claims using unique person‐specific identifiers (eg, Medicare Beneficiary Identifiers) and to commercial pharmacy and medical claims data (Optum) using third‐party software (Datavant) via deterministic linkage using name, sex, date of birth, and at least one matching physician office visit date 6 . The study period included EHR records and claims data from 2007 through 2020.…”

“…The EHR data were linked to Medicare pharmacy and medical claims using unique person-specific identifiers (eg, Medicare Beneficiary Identifiers) and to commercial pharmacy and medical claims data (Optum) using third-party software (Datavant) via deterministic linkage using name, sex, date of birth, and at least one matching physician office visit date. 6 The study period included EHR records and claims data from 2007 through 2020. Medications were normalized between data sources using RxNorm medication ontology, one of the vocabularies represented in the Unified Medical Language System.…”

Public Health Informatics and the Perioperative Physician: Looking to the Future

“…Tokenization assigns hashes and tokens to a group of identifier fields to encrypt a data file that contains PHI before it is shared with another organization and linked to a larger dataset. 84 The data linkage would then occur by comparing the hashes or tokens, which must match exactly, between the linked files. 85 Through the replacement of individual private health information with random inputs as tokens, the ability to link individual data across datasets (ie, public health data) to gain greater analytical insights and maximize privacy and reverse identification of patients can be achieved.…”