Abstract
BACKGROUND: The growing volume of health data provides new opportunities for medical research. By using existing registries, large populations can be studied over a long period of time. Patient-level linkage of registries leads to even more detailed and extended information per patient, but brings challenges regarding responsibilities, privacy and security, and quality of data linkage. In this paper we describe how we dealt with these challenges when creating the Primary Secondary Cancer Care Registry (PSCCR)- Breast Cancer. METHODS: The PSCCR - Breast Cancer was created by linking two existing registries containing data on 1) diagnosis, tumour and treatment characteristics of all Dutch breast cancer patients (NCR), and 2) consultations and diagnoses from primary care electronic health records of about 10% of Dutch GP practices (Nivel-PCD). The existing registry governance structures and privacy regulations were incorporated in those of the new registry. Privacy and security risks were reassessed. Data were restricted to females and linked using postal code and date of birth. The breast cancer diagnosis was verified in both registries and for a subsample of 44 patients with the GP as well. RESULTS: A collaboration agreement was signed in which the organisations retained data responsibility and accountability for 'their' registry. A Trusted Third Party performed the record linkage. Ten percent of the patients with breast cancer could be linked to the primary care registry, as was expected based on the coverage of Nivel-PCD, and finally 7 % could be included. The breast cancer diagnosis was verified by the GP in 42 of the 44 patients. CONCLUSIONS: We developed and validated a procedure for patient-level linkage of health data registries without a unique identifier, while preserving the integrity and privacy of the original registries. The method described may help researchers wishing to link existing health data registries.