Validating a novel deterministic privacy-preserving record linkage between administrative & clinical data:...

Area of Science:

• Health Informatics
• Data Security
• Biostatistics

Background:

• Combining research and administrative data offers unique insights but is hindered by data encryption.
• Traditional record linkage methods fail with encrypted personal health data due to privacy and institutional restrictions.
• Privacy-preserving record linkage is essential for secure data integration.

Purpose of the Study:

• To assess the feasibility and validity of a deterministic privacy-preserving data linkage protocol.
• To evaluate a protocol using homomorphically encrypted data for secure record linkage.

Main Methods:

• A deterministic privacy-preserving data linkage protocol was implemented using homomorphically encrypted direct identifiers (health card number).
• Feasibility was determined by direct identifier match rates, and validity by indirect identifier match rates (sex, date of birth).
• Encrypted data was linked on a third-party server without decrypting individual records, followed by verification using indirect identifiers.

Main Results:

• The linkage successfully matched 99.9% of records using direct identifiers and 99.8% using indirect identifiers.
• The protocol demonstrated high feasibility and validity, exceeding the 95% threshold.
• Linking 8,128 individuals to over 3.2 million records took 36 days.

Conclusions:

• The privacy-preserving data linkage protocol is both feasible and valid for integrating encrypted health datasets.
• This approach ensures patient data privacy and security while improving data quality.
• While resource-intensive initially, increased automation makes the process sustainable.