Anonymising Clinical Data for Secondary Use

TL;DR

This paper discusses the challenges and methods of anonymising clinical data for secondary use, balancing data utility with subject privacy, and exploring quantitative metrics and new technologies for risk management.

Contribution

It provides an overview of anonymisation approaches tailored for different data sharing scenarios in clinical research, emphasizing the importance of quantitative metrics and emerging technologies.

Findings

Different scenarios require tailored anonymisation strategies.

Quantitative metrics can guide de-identification levels.

Emerging technologies offer new solutions for privacy protection.

Abstract

Secondary use of data already collected in clinical studies has become more and more popular in recent years, with the commitment of the pharmaceutical industry and many academic institutions in Europe and the US to provide access to their clinical trial data. Whilst this clearly provides societal benefit in helping to progress medical research, this has to be balanced against protection of subjects' privacy. There are two main scenarios for sharing subject data: within Clinical Study Reports and Individual Patient Level Data, and these scenarios have different associated risks and generally require different approaches. In any data sharing scenario, there is a trade-off between data utility and the risk of subject re-identification, and achieving this balance is key. Quantitative metrics can guide the amount of de-identification required and new technologies may also start to provide alternative ways to achieve the risk-utility balance.