deidentification

Deidentification is the process of removing or masking information that identifies individuals within a data set to prevent disclosure of personal information. The goal is to reduce privacy risk while preserving useful data for analysis. There are two terminologies: deidentification and anonymization. Anonymization implies that reidentification is not reasonably possible, whereas deidentification often allows the possibility of reidentification if additional information becomes available.

Techniques include direct identifier removal (names, addresses, social security numbers), pseudonymization (replacing identifiers with codes), data masking and generalization (reducing precision), data suppression, data perturbation, aggregation, and more formal approaches such as k-anonymity, l-diversity, t-closeness, and differential privacy. The choice depends on data type, use case, and acceptable risk.

Regulatory context: In the United States, HIPAA Safe Harbor requires removal of 18 identifiers, while expert determination may be used to deem data deidentified. The European Union’s GDPR treats pseudonymized data as still personal data, requiring safeguards. Other jurisdictions set standards for data sharing and research.

Risks and governance: Reidentification can occur when deidentified data are linked with other datasets or when external information is obtained. Ongoing risk assessment, governance, data stewardship, and privacy-preserving techniques help manage risk. Deidentification is not a one-time act; it is an ongoing process that may require revisiting as data ecosystems change.

Applications: Used in healthcare research, public health statistics, and open data releases to balance data utility with privacy.