Massachusetts’ Center for Health Information & Analysis Selects Onpoint to Build De-Identification Model Using Expert Determination Approach

Ma chia

September 2017 -- Onpoint is pleased to announce that it has been selected by the State of Massachusetts’ Center for Health Information and Analysis (CHIA) as the organization’s new consultant and data management contractor for the creation of a statistically de-identified data set, employing an Expert Determination methodology consistent with section 164.514(a) of the HIPAA Privacy Rule. This de-identified data set will support CHIA’s mission of supplying providers, payers, policymakers, researchers, and citizens alike with the most accurate, useful, and robust set of data on the Massachusetts healthcare system while safeguarding patient privacy.

CHIA is the hub of information and analysis for the Massachusetts healthcare system, charged with providing reliable data and meaningful analyses to those seeking to improve the quality, affordability, access, and outcomes of services across the State’s communities. The organization’s decision to utilize a statistics-based Expert Determination model in approaching the de-identification of the MA all-payer claims database data set for public use comes with several benefits compared to alternative models such as the commonly employed Safe Harbor method.

Whereas the Safe Harbor method defined in 45 CFR 164.514b(2) relies on the removal of a broad set of specific patient identifiers to eliminate the possibility that a patient’s identity could ever be traced back to sensitive health information in a data set, the Expert Determination method requires knowledge and experience with statistical principles and risk management exercises to render information not individually identifiable.

To illustrate, the Safe Harbor method generalizes all ZIP codes contained in a data set to their initial three digits, which may in turn prevent researchers from conducting a town-by-town health outcomes analysis; the method also requires suppression of all patient ages over 89 years, which could be essential to support a gerontological study. The Expert Determination method, on the other hand, would only mask this type of information if, and only if, it is discovered that releasing such information, while retaining other identifiers, on a per patient basis introduces a threshold of risk indicating that the information could potentially be linked to an individual by the anticipated recipient(s). 

The Expert Determination approach effectively evaluates functional needs, mitigating controls, motives, capacity, and data sensitivity in an effort to design a de-identification model that delivers a data set with the greatest utility while ensuring the protection of patient privacy. In its consulting role, Onpoint will work collaboratively with CHIA to deliver such a model and de-identified data set for general consumption.