CROs and sponsors under pressure to share anonymized data from clinical trials are developing methods to de-identify information according to new research.
The analysis – available here – is based on a review of published trials that included reporting recommendations on anonymization to enable data sharing.
The EU/UK region provided 53% of the included studies, followed by the US/Canada region with 35%, while the rest (12%) originated from other regions.
And according to the study – while there is no standard approach to data anonymization – most of the trials included in the review used a variation of one of three strategies – either anonymization, de-identification or pseudonymization.
This evolving industry consensus reflects the current lack of guidelines on anonymization according to the authors.
“Currently, there is a strong demand for academic researchers to share their data more readily. In clinical trials, data can be shared more widely if they are anonymized, yet, we do not have standardized recommendations on how to do this.
“As time goes by there seems to be an emerging natural consensus on the definitions of pseudonymization, de-identification and anonymization,” they said.
The authors also point out that many of the trials called for the use of so-called privacy models, with k-anonymity – in which data attributes are suppressed or generalized until each row is identical with at least k-1 other rows – emerging as a front runner.
According to the study twelve of the trials included in the review studies recommended the use of privacy models such as k-anonymity, l-diversity and differential privacy to further guarantee and assess data anonymity to protect datasets from re-identification attacks.
“There are other privacy models, but they are not routinely used in clinical trials, as they could be complex, time-consuming and not practical for clinical trials datasets, which are relatively small when compared against routine health data.”
Image: Stock Photo Secrets