Twenty Years of the Health Insurance Portability and Accountability Act Safe Harbor Provision: Unsolved Challenges and Ways Forward

Abstract

The Health Insurance Portability and Accountability Act (HIPAA) was an important milestone in protecting the privacy of patient data; however, the HIPAA provisions specific to geographic data remain vague and hinder the ways in which epidemiologists and geographers use and share spatial health data. The literature on spatial health and select legal and official guidance documents present scholars with ambiguous guidelines that have led to the use and propagation of multiple interpretations of a single HIPAA safe harbor provision specific to geographic data. Misinterpretation of this standard has resulted in many entities sharing data at overly conservative levels, whereas others offer definitions of safe harbors that potentially put patient data at risk. To promote understanding of, and adherence to, the safe harbor rule, this paper reviews the HIPAA law from its creation to the present day, elucidating common misconceptions and presenting straightforward guidance to scholars. We focus on the 20,000-person population threshold and the 3-digit zip code stipulation of safe harbors, which are central to the confusion surrounding how patient location data can be shared. A comprehensive examination of these 2 stipulations, which integrates various expert perspectives and relevant studies, reveals how alternative methods for safe harbors can offer researchers better data and better data protection. Much has changed in the 20 years since the introduction of the safe harbor provision; however, it continues to be the primary source of guidance (and frustration) for researchers trying to share maps, leaving many waiting for these rules to be revised in accordance with the times.

Best Practices in Evolving Privacy Frameworks for Patient Age Data: Census Data Study

“…In practice, the only instance where current census data are referenced in HIPAA’s Safe Harbor criteria is the list of regions determined by the first 3 digits of a zip code. Any such region with a population under 20,000 is deemed to pose an unacceptable reidentification risk and is not permitted to be part of a Safe Harbor data set [ 15 ]. Any change to the Safe Harbor criteria, such as a dynamic age-constraining threshold, would need to come from the US Department of Health and Human Services.…”

LionVu: A Data-Driven Geographical Web-GIS Tool for Community Health and Decision-Making in a Catchment Area

“…Some of the potential benefits of big data to public health are determining risk factors that lead to diseases, such as cancer, and that generate new knowledge, improve clinical care, and streamline health surveillance [ 30 ]. When using GIS to present public health data, there are issue that restrict data availability, such as HIPAA [ 7 ], but one also has to watch out for data heterogeneity, data security, transparency, and fair use of data [ 30 ]. In the case of LionVu, we are providing hypothesis-driven research that uses smarter GIS functionality in ways that can utilize big data sources (see Table 2 ).…”

The Costs of Anonymization: Case Study Using Clinical Data

“…Overall, we determined that 6 of the 70 variables needed to be protected against reidentification: age, gender, height, weight, BMI, and history of renal biopsy. The underlying reasons and results of the 2-step procedure are provided in Table S1 in Multimedia Appendix 1 [ 40 , 44 ].…”