Not So Private

by Stacey A. Tovino

Click here for a PDF file of this article

Abstract

Federal and state laws have long attempted to strike a balance between protecting patient privacy and health information confidentiality on the one hand and supporting important uses and disclosures of health information on the other. To this end, many health laws restrict the use and disclosure of identifiable health data but support the use and disclosure of de-identified data. The goal of health data de-identification is to prevent or minimize informational injuries to identifiable data subjects while allowing the production of aggregate statistics that can be used for biomedical and behavioral research, public health initiatives, informed health care decision making, and other important activities. Many federal and state laws assume that data are de-identified when direct and indirect demographic identifiers such as names, user names, email addresses, street addresses, and telephone numbers have been removed. An emerging reidentification literature shows, however, that purportedly de-identified data can—and increasingly will—be reidentified. This Article responds to this concern by presenting an original synthesis of illustrative federal and state identification and de-identification laws that expressly or potentially apply to health data; identifying significant weaknesses in these laws in light of the developing reidentification literature; proposing theoretical alternatives to outdated identification and de-identification standards, including alternatives based on the theories of evolving law, nonreidentification, non-collection, non-use, non-disclosure, and nondiscrimination; and offering specific, textual amendments to federal and state data protection laws that incorporate these theoretical alternatives.

Not So Private

by Stacey A. Tovino

Click here for a PDF file of this article

Abstract

Federal and state laws have long attempted to strike a balance between protecting patient privacy and health information confidentiality on the one hand and supporting important uses and disclosures of health information on the other. To this end, many health laws restrict the use and disclosure of identifiable health data but support the use and disclosure of de-identified data. The goal of health data de-identification is to prevent or minimize informational injuries to identifiable data subjects while allowing the production of aggregate statistics that can be used for biomedical and behavioral research, public health initiatives, informed health care decision making, and other important activities. Many federal and state laws assume that data are de-identified when direct and indirect demographic identifiers such as names, user names, email addresses, street addresses, and telephone numbers have been removed. An emerging reidentification literature shows, however, that purportedly de-identified data can—and increasingly will—be reidentified. This Article responds to this concern by presenting an original synthesis of illustrative federal and state identification and de-identification laws that expressly or potentially apply to health data; identifying significant weaknesses in these laws in light of the developing reidentification literature; proposing theoretical alternatives to outdated identification and de-identification standards, including alternatives based on the theories of evolving law, nonreidentification, non-collection, non-use, non-disclosure, and nondiscrimination; and offering specific, textual amendments to federal and state data protection laws that incorporate these theoretical alternatives.