Background: The Cancer Registry of North-Rhine-Westphalia stores exclusively encrypted personal identifiers of registered cancer patients. Therefore, comparisons with secondary data sets can only be performed by record linkage procedures that are based on encrypted personal identifiers. We report on a pilot study which linked encrypted personal data from the disease management program for patients with diabetes mellitus type 2 (DMP-DM2) with the database of the EKR NRW in order to test the feasibility and efficiency of these record linkage procedures.
Methods: Personal identifying variables of the DMP records were encrypted in a 2-stage process before being sent electronically to the EKR NRW where they were subsequently submitted to a probabilistic record linkage with the registry data. The study included 27 450 participants who were insured at the AOK NordWest, residents of the district Münster and who were aged 40-79 years at the time of first enrolment to the DMP-DM2 between June 2003-July 2008.
Results: The electronic processing time of the semi-automatic record linkage procedure took about 24 h. Approximately 2% of the records had to be reviewed manually. After exclusion of prevalent cancer cases, multiple primaries and inadequate data, 26 742 participants (47.3% men; 52.7% women) remained in the data set. About 1 364 cohort members (759 men, 605 women) were diagnosed with cancer after submission to the disease management program.
Discussion: The DMP-DM2 records were encrypted and linked to cancer registry data with a moderate personnel and financial input and high efficiency. Linked records were instantly usable for epidemiological analyses. Experiences of the pilot study suggest that future linkage studies can further advance the level of data protection, without losses in efficiency, by moderately complex software modifications and amendments of the data flow.
© Georg Thieme Verlag KG Stuttgart · New York.