Virtual Pooled Registry Cancer Linkage System for Cohort Matching and Duplicate Identification

Authors: Deapen D, Penberthy L, Kohler B, Clerkin C

Category: Electronic Health Records (EHRs), Lifestyles Behavior, Energy Balance & Chemoprevention
Conference Year: 2018

Abstract Body:
Purpose: The North American Association of Central Cancer Registries (NAACCR) and the National Cancer Institute (NCI) are developing a Virtual Pooled Registry-Cancer Linkage System (VPR-CLS) that allows researchers to submit cohort data for matching with multiple cancer registries through a single, web-based system. Background: Matching cohorts with multiple cancer registries is useful for confirming cancer diagnoses and obtaining additional information including outcomes. However, this process is difficult, labor-intensive and costly if separate permission to match is needed for each registry. The use and interpretation of different matching protocols in each setting may affect study validity. registries and two large cohorts of 1.) Environmentally exposed military personnel; and 2.) Radiation technologists with occupational exposures dating to the 1950s. Confidentiality of the data was maintained by conducting the match behind each Registry's firewall. We have developed sophisticated matching software for these linkages thereby ensuring that standardized criteria are used across sites. The VPR-CLS has also facilitated inter-registry linkages to identify duplicate cases and multiple primaries shared across geographic jurisdictions. Results: Pilot linkages between two large research cohorts and over 40 registries were conducted. Nearly all registries performed the linkage and provided match counts for the researcher within 2-3 weeks of receiving the standardized, edited cohort file. Both pilot linkages identified CEC matches in all participating registries. In one cohort, historical linkage with four registries identified 26% under-ascertainment using self-report alone. Linkage with registries also provided these researchers with additional, curated data items of interest. The estimated cost was $58,000 as compared with $1.28M for the previous data collection method. Discussion: The VPR-CLS was developed as a collaboration among the US registry programs (SEER and National Program of Cancer Registries), NCI, NAACCR and their statistical partner IMS, Inc. The VPR-CLS benefits cancer registries and researchers by providing a single, efficient system to facilitate multi-registry linkages, providing much faster results at significantly lower cost.

Keywords: cohort studies, cancer registries, linkage studies