Despite the devastating impact of psychiatric and psychological problems, mental health remains under-recognised as a priority for funders and policy makers alike (MQ Transforming Mental Health, 2019). Yet the past 18 months have underlined the crucial need for assessing population mental health and wellbeing, and adapting an already stretched health service to meet those needs (Holmes et al, 2020). At the same time, the pandemic showed that mental health research can provide timely responses to pressing issues (Demkowicz et al., 2021). Funders launched special calls for research on the mental health impact of the pandemic and the lockdowns, ethics committees rapidly turned around decisions, participant recruitment was sped up using online technology, and open access repositories facilitated dissemination of non-peer reviewed findings. The importance of having a coordinated and efficient infrastructure to support mental health research has never been so clear – and the foundation of that research is of course data.
Challenges in data sharing
The National Health Service (NHS) generates and manages routinely collected medical data in the UK. These data document patients’ experiences with mental health services and can also supplement epidemiological studies with clinical information. These patient-level data are crucial for researchers to produce evidence that can inform and guide policy makers, practitioners and also the general public. However, an authoritative commentary by Tamsin Ford and her colleagues published inThe Lancet Digital Health last year sets out the complex issues currently encountered in accessing administrative mental health data under the guardianship of the (NHS) and highlights the critical consequences for research (Ford et al., 2021). The issues are multi-layered and multifaceted, reflecting an infrastructure that is cumbersome and not fit for purpose.
As Tamsin Ford outlines, researchers are reported to have faced strenuous bureaucratic processes and long delays when attempting to access these valuable data. Some of the issues relate to working with potentially identifiable data and navigating the legal framework on data protection and privacy set out in the General Data Protection Regulation (GDPR). The new legislation blurred the boundaries between personal and anonymised data and generated uncertainty around how to deal with this information. Furthermore, researchers requesting data via the Data Access Request Service have been asked to demonstrate the value of their research for public interest. While the ultimate purpose of mental health research is to improve detection, prevention and treatment of illnesses, some valuable research also aims to “better understand” and may not have a direct and immediate impact for patients and the public. In addition, communication between researchers, data custodians and the public has been marked by a lack of trust because of staff shortages and rapid turnaround of people in posts.
While GDPR aims to facilitate research and protect patients’ information from data breaches, the new legislation has instead hampered the progress of mental health research. And while protecting patients’ sensitive and identifiable data is paramount, the consequences of preventing researchers from using these valuable and rich data are not trivial. For instance, the implementation of innovative projects focusing on addressing mental health problems can be deterred because of the complexity of accessing data. Research findings can be out of date and that is detrimental, especially when we need rapid evidence to guide our response in times of crisis. Data scientists can be discouraged from creatively exploring new forms of data. For the NHS, patients and the public, these delays mean services that are not anchored in the most up to date sources of clinical data.
The complex issues with sharing mental health data appear to go against the public’s view about the use of their routinely collected data. A poll set up by Mental Health Data Science Scotland indicated that people view the use of their data for research purposes favourably. This project focuses on Best Practice for Mental Health Data Science and it generated a useful checklist, co-created by community experts and the research team, that provides researchers “…information about what is acceptable to, and important for, people with lived experience of mental illness” (Kirkham et al., 2020). This checklist, structured around key themes including data security, anonymity, transparency, and community, suggests to researchers and the wider scientific community best practice with data usage for today and for the future. While data protection is essential, it should be part of best practice for mental health research instead of being an obstacle for accessing data. Educating researchers about issues around mental health data including ethics, stigma, and privacy can reduce malpractice. Breaking down the bureaucracy that hinders mental health data sharing can also be useful for data custodians. This roadmap is also extremely valuable in highlighting the importance of involving patients and study participants in mental health data science, alongside data users and data custodians.
One of the great values of Tamsin Ford’s commentary is that in addition to highlighting an important problem, it also proposes a series of remedies. These involve a variety of stakeholders including the government, data custodians, data users, and academic institutions. Actions needed include establishing a system of data sensitivity levels (low, medium or high) with different steps for accessing data; capitalising on training and supervision to educate researchers; and making the access process transparent, consistent and clear. These are all important steps in the right direction.
Other forms of mental health data need to be more accessible too
Some of the issues raised by Ford and her colleagues are not exclusive to NHS administrative data. Indeed, we need greater efforts in making mental health data from a wide range of sources more discoverable and more usable. These are the aims of DATAMIND, the recently MRC-funded mental health data hub housed at Health Data Research UK (HDRUK). The goal of this new data hub is to improve people’s mental health by changing the way the NHS, charities, industry, and researchers use and share information that is already harvested. The hub will maximise the value of existing data by bringing together diverse sources including health records, school and administrative data, charity data, and data from research trials, genomics, longitudinal and cohort studies.
The hub is a virtual space where researchers and other stakeholders can find and use mental health data, to benefit patients and the public and improve care. DATAMIND works with patients and people with personal experience of mental illness to understand whom they trust to use their data, and to develop ways people can work together on mental health.
As part of DATAMIND, the UK Longitudinal Linkage Collaboration (UK LLC) is developing a new approach for linking well-established longitudinal studies to routine records. It provides a safe and secure virtual space where information collected from these studies can be used in COVID-19 research across the UK. The next aim for this project is to expand its scope to mental health data.
Also part of DATAMIND, the Catalogue of Mental Health Measures aims to provide easy access to information about the mental health measures included in British cohort and longitudinal studies to maximise the uptake of existing data and facilitate mental health research. While the Catalogue does not itself provide access to data, it does provide details on the mental health measures collected through the years and information about how to access data.
Mental health data are invaluable resources that should be securely shared to support research and ultimately benefit the public, patients and the NHS. We need a coordinated infrastructure that is supportive and flexible in promoting the use of mental health data. Commentaries like the one from Ford and her colleagues (Ford et al., 2021) are important in raising concerns over existing arrangements. Work from Mental Health Data Science Scotland and DATAMIND are important in developing an infrastructure for the safe and efficient sharing of mental health data. This will happen when the views of people with personal experience of mental illness are properly considered, and when mental health becomes a priority and is supported by adequate funding.
Statement of interests
None to declare.
Ford T, Mansfield KL, Markham S, McManus S, John A, O’Reilly D, Newlove-Delgado T, Iveson MH, Fazel M, Das Munshi J, Dutta R, Leavy G, Downs J, Foley T, Russell A, Maguire A, Moon G, Kirkham EJ, Finning K, Russell G, Moore A, Jones PB, & Shenow S (2021). The challenges and opportunities of mental health data sharing in the UK. The Lancet Digital Health, 3, e333-336.
Demkowicz O, Panayiotou M, Parsons S, Feltham A, Arseneault L, Ingram B, Patalay P, Edge D, Pierce M, Creswell C, Victor C, O’Connor RC & Qualter P (2021). Looking back to move forward: Reflections on the strengths and challenges of the COVID-19 UK mental health research response. Frontiers in Psychiatry, 12, 622562. (doi: 10.3389/fpsyt.2021.622562).
Kirkham, E.J., Iveson M., Beange, I., Crompton, C.J. Mcintosh, A. & Fletcher-Watson, S. (2020). A stakeholder-derived, best practice checklist for mental health data science in the UK. University of Edinburgh. (mhdss.ac.uk/best-practice-mental-health-data-science).
Holmes E, O’Connor RC, Perry H, Tracey I, Wessley S, Arseneault L, Ballard C, Christensen H, Cohen Silver R, Everall I, Ford T, John A, Kabir T, King K, Madan I, Michie S, Przybylski AK, Shafran R, Sweeney A, Worthman C, Yardley L, Cope CL, Hotopf M, & Bullmore E. (2020). A call for action: Mental health science and multidisciplinary research priorities for the COVID-19 pandemic. Lancet Psychiatry, 7, 547-560. (doi.org/10.1016/S2215-0366(20)30168-1).
MQ Transforming Mental Health. UK Mental Health Research Funding 2014–2017 (2021). AMRC Open Res 2021, 3:9. (doi.org/10.21955/amrcopenres.1114943.1).