February 2025
New paper published - What is the patient re-identification risk from using de-identified clinical free text data for health research?
Patient medical records, notes, letters, reports, all hold important clinical information about an individual and are vital in delivering essential and continued care. This written, often detailed information, is referred to as clinical free text data. This type of data has the potential to improve the lives and health outcomes of not only the individual patient, but also the wider population and future generations if used for health research. Currently, this detailed information is trapped and under-used for health research. This is primarily due to concern from data controllers about patient privacy risks.
This paper explores and evaluates the risk of patient re-identification from clinical free text data. It presents a model to conceptualise and evaluate risk of re-identification and provides a framework by which decision-making bodies might assess the risks posed. The paper describes the additional protections in place regarding free text data, and recommends that access to such data (by approved researchers) is done safely within an NHS Secure Data Environment (SDE). The paper goes on to highlight real case studies of successful governance of free text data for research in the UK and argues that with strong safeguards in place, it is now time for the UK and NHS to transition to regularly, routinely and safely using clinical free text to conduct vital health research and improve healthcare for all.
Read the full paper here >
New paper published - “I don’t see a reason why we should be hidden from view”: Views of a convenience sample of people living with HIV on sharing HIV status data in routinely collected health and care databases in England.
People Living with HIV (PLWH) now have near normal life expectancy, yet little is understood about how HIV affects conditions of ageing. It can be difficult to recruit enough PLWH to study conditions of ageing across a large sample. One route around this is to use anonymised patient health record data, a common source of data for epidemiological research in the UK. Currently, HIV diagnostic codes are stripped out when data is processed for research, which means no research about HIV can be done using these datasets. This represents a barrier to understanding common health problems and the health trends of PLWH, which leads to health inequity for PLWH.
This study explores the views of PLWH on sharing their HIV status data in routinely collected health and social care databases in England. 37 PLWH participated in deliberative online focus groups, where they were invited to share their views and suggest safeguards they felt would allow them to feel more comfortable with this type of data sharing.
Participants were cautious about sharing their HIV status; their concerns largely stemmed from prior discriminatory experiences and the risk of re-identification. However, with assurances put in place around data security, participants felt more confident in supporting this type of data sharing. Safeguards participants identified as important in building trust and confidence were: high-tech data security; limited access and clear accountability for data users; meaningful involvement of PLWH in data and research systems; communication about data flows and uses; education about HIV to reduce stigma. Moving forward the authors recognise and encourage a commitment to long term relationship building with the HIV community, fostering trust and encouraging transparency and openness.
Having engaged deliberately with PLWH from a range of backgrounds, the authors suggest from their findings that sharing HIV status in NHS research datasets could be achieved and if well planned, transparent and collaborative with the HIV community, would likely be supported by PLWH. Sharing HIV status for research will: build a deeper understanding of the ageing HIV population, improve services, and encourage better health, care and wellbeing for PLWH.
Read the full paper here >