A recent study shows that using anonymised medical notes—such as doctors’ reports and hospital letters—for research poses little risk when strong privacy measures are in place. Researchers found that carefully removing personal details and storing data securely can protect patient identities while providing valuable insights to improve healthcare.
The research team, based at Brighton and Sussex Medical School, reviewed current techniques for removing identifiable information from clinical text and evaluated how securely the data is handled. They focused on the UK’s “Five Safes” model, which includes checks for safe projects, people, data, settings, and outputs. They concluded that, when used properly, these methods make it highly unlikely for anyone to be re-identified from the anonymised text.
Funded by the National Institute for Health and Care Research (NIHR) Applied Research Collaboration Kent, Surrey and Sussex (ARC KSS), the study team previously explored public attitudes toward the use of de-identified clinical data. Most people were supportive, especially when the purpose is to improve health outcomes and privacy is protected. The researchers emphasised the importance of continued public engagement to build trust and ensure transparency around how health data is used.
Key Recommendations:
- Anonymised clinical text should be routinely stored and made available in secure data environments for research.
- Policymakers and data holders can be confident that the risk of re-identification is very low.
- Greater use of clinical free text could lead to significant benefits in healthcare research and delivery.
- Ongoing dialogue with the public is essential to maintain support and trust.
Dr Liz Ford, ARC KSS Data Science Lead, says: “Our findings show that anonymised clinical notes should be routinely stored and used in secure environments to support healthcare research”.
“This approach can help improve healthcare while protecting patient privacy, and it’s vital that we keep the public informed to build trust in how their data is used.”
Read the full paper here.