Mental health data science: “My NHS patient record is a very precious thing”

A lived experience researcher working with McPin on a ‘big data’ project discusses the far-reaching impact of data in their life

Photo by NASA on Unsplash

My NHS patient record is a very precious thing. It contains lots of very personal information about me from birth onwards: every contact I’ve ever had with a clinician, every medication I’ve ever been prescribed, every illness I’ve ever had, every vaccination, every referral, every test, every address I’ve lived at, my sexual orientation, my ethnicity, details about my lifestyle, my weight, my height, whether I smoke, drink, and eat healthily.

Of even more sensitivity to me, it contains a wealth of information about my mental health; all the above plus details of every conversation I have had with my psychiatrists, my therapists, and my GP. It documents every traumatic life experience, the ups and downs of treatment pathways and experiences, details about my family and friends, every positive clinical outcome, every negative one.

This is information about me collected and recorded during routine healthcare over many years, usually disclosed in confidence. Information that is essential to effective shared decision-making and better clinical care. Therein lies its value but, also, its problem; this is sensitive information.

The value of data

Multiply the above by millions, and you have a huge repository of information with enormous potential – routinely collected ‘big data’.

Mental health researchers are increasingly using this routinely collected data – usually with identifying information removed – from thousands or millions of people to answer pressing and relevant research questions; to evaluate and model service use and design; to audit the quality of services; and to gauge whether there is equity of access and outcomes to mental health services for different segments of the population (spoiler alert: there isn’t).

Routinely collected data vs clinical trials

These datasets are more likely to be representative of the population in contrast to some clinical trials and studies where people actively consent to take part.

Clinical trials may have hidden biases that make the findings less applicable in the real world. For example, trials of psychological therapy usually ensure strict adherence to the therapy manual, offer regular and good quality supervision to the therapists, and sometimes offer a higher-level of support to the participants than in ‘typical’ clinical settings.

The number of participants is usually relatively small and subject to its own recruitment biases, meaning the impact of other demographic factors, such as ethnicity and cultural experiences, are often not captured in the dataset.

Real-world therapy experiences and outcomes may differ widely from the original clinical trial and in different sub-groups of people. Therefore, being able to use big data from these real-world settings has power in numbers and wider generalisability.

Data’s vulnerability

But in the strength of big data also lies vulnerability. Who does the data belong to? Did I give permission for it to be shared with someone outside of my clinical care and for research? Did anyone even ask me? Can I say no? Will someone be able to identify me when they look at the data? Will they keep it safe and secure? Can they sell it on? Can the data be transferred outside the UK? Will the research benefit me and the wider public?

These are all questions that go through many people’s minds when they discover that parts of their health record are being used in health research.

The level of public awareness of health record use in research is currently quite low but slowly improving. The recent controversy around sharing of GP health records has brought this into sharp focus for many people.

However, when effective public engagement work has been conducted, people’s attitudes to sharing their data is usually more positive. This is a gap that needs to be addressed, along with greater transparency about its use and benefit and concerns around profiteering and security.

How clinical experiences affect data sharing

A recent study by Elizabeth Kirkham’s team explored how people’s clinical experiences affected their attitude to data-sharing.

Strikingly, around 9 in 10 participants were willing to share their physical and mental health data for research, with the proportion for mental health data being slightly lower. There was also a strong positive link between higher satisfaction with NHS care experiences and willingness to share data. This is very relatable.

Experiences within the mental health system that have broken my trust – feeling dismissed and disempowered, questions not answered transparently, a data breach – have at times heightened my distrust in the use of my data. What happens in one part of the system can have a knock-on effect elsewhere.

The researchers make an important point in suggesting that addressing NHS satisfaction could go some way towards improving attitudes to data-sharing. More joining up of clinical services and research, and more tailored communication for different groups, also comes to mind.

Creating a best practice checklist

In collaboration with people who have personal experience of both mental illness and data research, the same team also developed a best practice checklist for use in mental health data science. This is a welcome development as the views of people with lived experience of mental illness have been integrated into a practical tool that data science researchers should pay attention to.

The broad areas of security, anonymity, transparency, and community align with findings from other research that has identified people’s key concerns.

I like that it presents immediate actions and considerations along with more aspirational ones that I hope researchers will work to adopt soon. The theme of community also struck a chord. Ensuring that people with lived experience are not just done to, but part of the conversation and more, is crucial to building trust and engagement in all sectors of the population.

Sharing findings in a way that is meaningful to all and giving back to the community is also essential. I hope for a future where more lived experience researchers are really driving the research questions and prioritisation agenda rather than being mere passengers.

Responsibility for the future

Going forwards, I would be interested to understand how data science researchers use the checklist in practice, how easily it is implemented, whether barriers or challenges occur, and how people with lived experience they may be co-producing with regard it.

I would also be interested to know how researchers will be working to incorporate the ‘future’ parts of the checklist and when. Ongoing monitoring and evaluation of its impact is needed.

There is also space for further and in-depth qualitative work exploring the nuances of people with lived experience’s views, experiences, and levels of comfort with mental health data science in different scenarios.

For example, research has shown that people report relatively high levels of trust in sharing data with universities, but lower levels of trust in data-sharing with private sector companies, technology providers and insurance companies.

With an evolving UK healthcare system where the boundaries between public and private sector are increasingly blurry, what does this mean for the ethics of sharing data? How willing will a Chief Data Officer at ‘Tech Company X’ working in healthcare be to embrace the checklist?

Our health data is a powerful shared resource that demands respect and integrity. With power comes responsibility.

To stay up to date with the work McPin does follow us on Twitter @McPinFoundation or sign up to our newsletter

For more on why data science is so important to mental health research, read Elizabeth Kirkham’s blog.

Further reading link: Understanding Patient Data website