Anonymized, But Vulnerable: Developing A Better Checklist For Traumatic Stress Data

In traumatic stress research, as in other health and social science research, ethical, legal, and regulatory frameworks require careful attention to data privacy. Internationally, researchers in traumatic stress are working to increase data sharing and other FAIR data practices in order to accelerate advances in understanding and ameliorating the impact of trauma. Most traumatic stress researchers are aware of basic methods for de-identifying/anonymizing datasets, but systematic methods for de-identification are not widely known or used.

The authors investigated two very different anonymized datasets that had been shared in an open repository for traumatic stress data. Reviewing these data let to the discovery of several vulnerabilities that the original anonymization efforts had missed. One factor that may arise particularly often in traumatic stress data is the potential to infer the study participant sampling frame based on the nature of the trauma being studied, giving attackers access to a “reidentification frame”. We used these findings to inform development of an improved checklist for deidentification of traumatic stress data.