Two recent manuscripts published in the journal BMC Research Notes and JAMIA describe the framework and implementation of a new method of connecting patient data across disparate healthcare ecosystems in PCORnet while maintaining the highest levels of privacy and security. Privacy-preserving record linkage, or PPRL, connects electronic health records (EHRs) and administrative claims databases, unlocking the ability to fill important gaps in data, while still making sure the private information of the individual stays protected.
“There’s a growing appreciation for the fact that traditional health systems are not the only places where important information about our overall health resides,” said Tom Carton, who co-led the PPRL Implementation Workgroup for PCORnet. “PPRL is a method that allows us to fill these gaps in data while maintaining the highest privacy standards so we can better understand what is driving good or bad health outcomes.”
How PPRL works
PPRL has been used as an encryption technique since the 1990s in the computer science community, but only recently has it made the jump to healthcare. The method involves a technique called hashing to create an irreversible and unique code (such as a number) given a set of inputs, like patient demographics or first and last names. These codes allow researchers to link patient records across data sources in a way that is compliant with HIPAA, the Health Insurance Portability and Accountability Act. As a result, users of PCORnet can now assign a universal patient identifier code across multiple datasets from PCORnet Partner Networks without compromising any personal information.
“PPRL is going to bring tremendous value by solving two key problems that currently plague large-scale research,” said Keith Marsolo, co-lead of the PPRL Implementation Workgroup for PCORnet. “First, it will allow users to flag and de-duplicate records across multiple datasets, meaning that people who get care from multiple health systems in the same geography will be more accurately counted. And second, it will allow us to much more easily link EHRs and claims data sets across PCORnet, providing a more complete picture of patients’ health experiences.”
The current state of PPRL in PCORnet
The PCORnet PPRL Implementation Workgroup successfully demonstrated the ability to use PPRL to identify overlap across approximately 170 million patient records accessible via PCORnet; they also created a de-duplicated summary of demographic and clinical characteristics for patients from 61 Partner Networks. This project, which was one of the largest linkage efforts of its kind in the U.S., enables current and future studies—including several PCORnet rare disease studies—to benefit from connected insights across EHR claims, and patient reported data.
“This is an exciting advance for researchers and patients alike,” said Marsolo. “With PPRL in place, meaningful research can proceed uninhibited by gaps in understanding with zero sacrifice to patient privacy and security. It’s a win-win.”