Clinical trial researchers must balance the at times contradictory needs for transparency, with protecting participants’ privacy and looking after companies’ intellectual property. Transparency and information sharing can be voluntary and motivated by the desire to progress science through collaboration. However, increasingly sharing of trial documents and data is driven by regulatory requirements.
The different global regulations ensure that the information released is governed by best practices which reduce the risk of re-identifying individuals represented in various data sets. In what may seem to be a conflicting end result, sponsors and regulators must find the delicate balance between protecting participant privacy, through redaction or anonymisation strategies, and providing data utility. This can be defined as the degree to which a reader can analyse and make meaningful interpretations from the information.
So how can pharmaceutical or academic researchers perform this delicate balancing act? And when operating in multiple geographic regions how can they meet the expectations of different regulatory bodies? In this article we outline the requirements of different regulators. We suggest ways that researchers can assess risk, plan and take steps that maximise data utility while meeting the expectations of regulators and patients.
Redaction vs Transformation Redaction is a method of masking information by applying a box over direct or indirect identifiers. It can be done manually or semiautomatically using most common software tools. Because of this it can be perceived as attractive and cost-effective. For short documents with little personal information (PI) or Protected Personal Data (PPD) it may be the logical choice. However, it has little to no data utility since all of the information is fully hidden. In addition, deciding what to redact can be subjective.
Transformation is the process of pseudonymising, offsetting or generalising direct or indirect information relating to participants. Direct identifiers can be the full name, subject numbers, phone number, email address or a government ID number. Indirect identifiers alone might not lead to reidentification, but combined with other information could be used to identify an individual. They include city, state, demographics and sensitive medical information. In transformation, instead of including a participant’s age, participants’ ages would instead be banded into groups. Validated software tools can assess the risk of reidentification, establish transformation strategies and implement anonymisation techniques. This can save time by automatically applying anonymisation to multiple clinical datasets and documents. This automated quantitative approach, when combined with robust quality control steps, can ensure confidence in the outputs.
