Search

The National Clinical Cohort Collaborative (N3C) uses privacy-preserving record linkage (PPRL) to formulate longitudinal patient histories and for de-duplication across disparate data including electronic health records, administrative claims, and mortality data. The N3C employs a linkage honest broker implementation model which separates the entity holding de-identified PPRL tokens, from the entity holding the substantive data payloads. As a result of this data governance structure, linkage quality assessments are necessarily distributed into 2 distinct processes: 1) Linkage Honest Broker receipt and processing of linkages, 2) Data Aggregation entity assembling data based on received linkage crosswalks. Data is released on a weekly basis. This presentation will focus on Step 2 for the Mortality data linkages where the downstream data linkage pipelines must assess linkage quality across several dimensions of quality, focused on plausibility. There are 4 types of mortality data spanning the electronic health record, government Social Security Administration death master files, third-party obituary data, and administrative claims. Each of these sources has different levels of authoritativeness and data latency, compounding the challenges of reconciling fact of death including accurate death reporting dates. To address these challenges, we implemented an automated linkage quality pipeline which generates reports that assess linkage plausibility across the various data types, classifying concordance and discordance between data sources including age, dates of death, surfacing plausible and non-plausible activities post-death. Sharing our real-world implementation will aid conference attendees in understanding how to plan for such activities when considering specific implementation models for privacy-preserving linkages.

Data centralization is increasingly non-viable in the changing landscape of data privacy and data governance. New models of data sharing are needed, in particular when privacy-preserving record linkages (PPRL) are being used to connect data within restricted data enclaves and public open enclaves. We share a privacy-preserving data-sharing infrastructure that employs a linkage honest broker implementation of PPRL connecting structured electronic health record data between the National Center for Advancing Translational Science (NCATS) National Clinical Cohort Collaborative (N3C), a restricted data enclave, with the National Cancer Institute (NCI) Cancer Imaging Archive (TCIA), a public open enclave. In particular we will address: • Data governance models needed when linking data from restricted vs. public enclaves • Specific privacy-preserving consideration when linking to imaging data which contain specific structured data and imaging considerations • Privacy-preserving data aggregation when combining multi-modal data from restricted data enclaves with publicly discoverable imaging data

Objectives Patient matching rates between organizations can be as low as fifty percent. Challenges to matching include the variation in quality and availability of patient attributes. Here we describe the changing nature of patient attributes available over the past 11-years across a diversity of care settings in the United States. Approach Our expert panel identified 64 patient attributes that are currently used or could potentially be candidates for patient matching. We identified a national sample of 14 health care sites who sent us aggregated information on the 64 patient attributes from 2010 to 2020 (inclusive). The information included overall counts and percent availability, overall counts and percent availability by race, and counts and availability by year. Only patients having at least one visit to the site since 2010 and who were between 18 and 89 years of age at time of extraction were included. Results The aggregated results revealed that first name, last name, gender, postal codes, and date of birth are highly available (>90%) across healthcare organizations and time. Patient reported social security number, work phone number, and emergency contact declined markedly, potentially reflecting privacy concerns. Email addresses (from 18.0% to 63.7%) and phone numbers (from 14.7% to 69.4%) increased greatly over the past 11 years. Novel patient matching attributes such as blood type, facial image, thumb print, or eye color are rarely collected across sites for all years. We observed emerging attributes including sexuality, occupation, and nickname with a small number of sites collecting these over 70%, reflecting the feasibility of wider adoption in the future. Conclusion In this study, we examined the availability of 64 patient attributes across 14 sites from 2010 and 2020. Our findings could inform policy makers and readers about patient attributes that are used for current patient matching and emerging data attributes that could be considered for incorporation into future matching algorithms.