Linking patients' records across organizations while maintaining anonymity.

Area of Science:

• Health Informatics
• Data Security
• Biostatistics

Background:

• Health Information Exchange (HIE) facilitates seamless data sharing across healthcare organizations.
• Maintaining patient anonymity is a critical challenge in HIE to protect sensitive personal health information.
• Effective record linkage requires robust methods for de-identifying patient data.

Purpose of the Study:

• To evaluate the quality of anonymous identifiers for linking patient health records in an HIE context.
• To assess the efficacy of different combinations of personal identifiers in creating anonymous IDs.
• To determine the specificity and sensitivity of these anonymous IDs for accurate record linkage.

Main Methods:

• Generated anonymous identifiers using one-way encryption functions from combinations of personal data.
• Utilized two synthesized databases for evaluating linkage quality.
• Assessed linkage performance based on specificity and sensitivity metrics.

Main Results:

• Anonymous identifiers derived from date of birth, zip code, and last name demonstrated high linkage quality.
• Combinations including first name also showed promising results for record linkage.
• The evaluated methods achieved high specificity and sensitivity in linking records.

Conclusions:

• Anonymous identifiers based on demographic data (DOB, zip, name) are effective for patient record linkage in HIE.
• This approach balances the need for data interoperability with robust patient privacy protection.
• Further validation in real-world HIE settings is warranted to confirm these findings.