Anonymising documents for disclosure

Publishing a trial's clinical documents puts the people who took part at risk of being identified from them, and this branch covers the work that protects them. It divides along the parts of that work: writing and anonymising the documents, the software that measures re-identification risk and transforms the text until it falls below a regulator's threshold, and the anonymisation report that explains how a package was de-identified, followed by one published package worked through decision by decision. It leaves the programmes that publish clinical documents to regulators' publication of clinical documents, who carries out the anonymisation within a package to staffing and scheduling a disclosure package, commercially confidential information to deciding what is commercially confidential in a disclosed document, the law on personal data to confidentiality and data protection, and surveys of writing software that list anonymisation tools among others to software and tools.

• Protecting participants in documents written for disclosure — how participants are re-identified, anonymisation against pseudonymisation, what a disclosed report contains, redacting a finished report against anonymising it while drafting, measuring the risk that remains, and signatures and metadata
• Anonymisation software for public disclosure — why anonymisation is automated, the tools available and how they are bought, detecting personal data, measuring re-identification risk on participant profiles, transforming a document until it falls below the threshold, who operates the software, and how accurately it detects
• The anonymisation report — the document that explains how a published clinical data package was de-identified: the templates it follows and the reports published so far, the methodology, risk assessment and data utility behind it, the techniques the packages used, how software generates it, and how the agency has commented on it
• Anonymisation decisions in one published package — what the Voydeya package retained, recoded and redacted, and what writing its pivotal report with disclosure in mind saved