illustrated eye with data inside

Case study: INSIGHT Health Data Research Hub

Tue Jan 12, 2021
$download_content = get_field('download_content');

We are working with INSIGHT to better involve the public, patients and other stakeholders in deciding how health data is used and shared

At the Open Data Institute, we are working with the INSIGHT – a Health Data Research Hub for eye health – to involve the public, patients and other stakeholders in deciding how health data is used and shared

INSIGHT – which launched in September 2019 – is one of seven Health Data Research Hubs funded through Health Data Research UK.

INSIGHT brings together and makes available anonymised data from eye scans and images, and advanced analytics, to NHS, academic and industry researchers. It aims to unlock new insights in disease detection, diagnosis, treatment and personalised healthcare.

The hub is focused on eye diseases such as age-related macular degeneration and glaucoma, and – as many other systemic diseases manifest in the eye – the application of eye health to wider health. This focus is a timely one, given that the number of people with sight loss in the UK is expected to double to over four million by 2050.

Involving the public, patients and other stakeholders in deciding how the data made available by INSIGHT is shared and used is central to the hub’s plans.

We are supporting INSIGHT to do this, working closely with the charity Action Against Age-related Macular Degeneration (AAAMD) and University Hospitals Birmingham and Moorfields Eye Hospital’s existing public and patient involvement and engagement teams. This builds on our experience and interest in empowering people to play a more active role in data stewardship – collecting, maintaining and sharing data, and in particular, determining who has access to it, for what purpose and to whose benefit.

Key facts and figures

  • Age-related macular degeneration (AMD) is the leading cause of sight loss in the UK.
  • Optical coherence tomography (OCT) scans are used to detect the earliest signs of disease such as AMD.
  • Scans of the eye can also detect a number of non-vision-related medical conditions, such as dementia, diabetes and heart disease.
  • INSIGHT is one of seven data hubs across the UK that are funded through Health Data Research UK.
  • INSIGHT brings together anonymised data from eye scans and images, and makes them available to NHS, academic and industry researchers.

Why is early-stage data on age-related macular degeneration important?

One of the most powerful tools for the diagnosis and monitoring of sight-threatening diseases is an eye scan. These scans are routinely carried out on most patients with diseases affecting the retina, to help ophthalmologists to diagnose AMD (and other eye conditions). However, these scans can also uncover markers of other health conditions that are not necessarily related to the eye. For example, a study in 2019 found that people with Alzheimer’s disease had fewer blood vessels and less blood flow in the retina, suggesting that eye scans could help to detect the condition.

Data on these scans is a rich source of information that could help uncover early markers of disease, demonstrate the mechanisms behind diseases and lead to the discovery of new treatments.

For AMD, the only treatment currently available is for late-stage disease. Having access to data on early-stage disease could help researchers and scientists understand how the disease develops, how many people have early-stage disease and whether there is anything that could be done to stop the disease before it gets to the sight-threatening stage.

Enabling the access to, and use of, this data holds huge potential for creating health and societal good through groundbreaking scientific discoveries. However, ophthalmic data in the UK is held by a large number of different organisations and there are limited ways for researchers to access large collections of scans.

INSIGHT aims to bring these scans and related data together in a systematic way.

Rather than a one-off task, this will form a framework for most, if not all, future ophthalmic data, which can then be made available through one single-request access point.

INSIGHT is currently focused on secondary care data (that is, hospital rather than general practice data), and so is closely linked with NHS Foundation Trusts. ‘Foundation Trust’ is a status awarded to hospitals who have shown they demonstrate the highest clinical standards, quality leadership and a great record of patient responsiveness and safety.

Involving patients and the public

Through INSIGHT, academic, industry and NHS researchers and organisations will be able to apply to access anonymised collections of OCT scans and related data from across Moorfields Eye Hospital and University Hospitals Birmingham.

Involving the public, patients and other stakeholders in deciding how the data made available by INSIGHT is shared is central to the hub’s plans. The plans include a mix of different deliberative methods, such as making INSIGHT’s work visible in the hospitals involved, and running workshops and events for patients, the public and other stakeholders to share their views.

As part of this work, we have supported INSIGHT to convene a new group – the INSIGHT Data Trust Advisory Board, or ‘INSIGHT DataTab’ – to provide meaningful oversight and scrutiny of the hub’s use and sharing of data. The INSIGHT DataTab is responsible for:

  • assessing and providing recommendations on individual applications to access the data made available by INSIGHT
  • developing and iterating the criteria for assessing the applications
  • providing feedback on the overall procedure for granting or denying requests to access INSIGHT data.

The members are representing not a finite set of institutions – which could reduce the diversity of the group and arguably block truly innovative thinkers – but are there as citizens, who bring a wide range of opinions and experiences.

We worked closely with AAAMD and the hub’s existing Patient and Public Involvement and Engagement teams to recruit a set of 10 initial members with a broad range of opinions, experiences and perspectives. This includes patients with sight-threatening conditions and members with experience in health data research and regulation. You can see the initial members of the INSIGHT DataTab here.

We have helped to design a process whereby the hub receives an application from a researcher to access data, and then processes the request to check that it is technically and legally feasible. If it is, the application will be passed to the INSIGHT DataTab to review. The group will review the application using a set of assessment criteria that it is currently developing itself.

Following its deliberations, the INSIGHT DataTab will provide one of four possible recommendations back to the hub to say that the application should be: granted; granted with further conditions; deferred pending receipt of additional information or clarification; or denied.

The INSIGHT DataTab will convene quarterly to review the batch of applications that have been received in the preceding three months. The minutes and recommendations from these meetings will be published openly in a timely fashion, allowing others to monitor progress and decisions of the board. Over time, the group might choose to consider only applications that are novel or contentious, especially if the volume of applications grows considerably or if many prove to be similar in nature.

Frequency of meetings and ensuring there is flexibility to restructure DataTAB is also important, as this will ensure nimble decision making to cater for a wider range of possible use access requests.

We expect the group to begin reviewing applications in early 2021.

Who else was involved?

INSIGHT is a partnership led by University Hospitals Birmingham, alongside Moorfields Eye Hospital, the University of Birmingham, Roche, Google Health and AAAMD.

What was challenging?

  • Recruiting a diverse group of members.

Any group designed to oversee data sharing and use is only as strong as the diversity of its membership. So this is something we started to think about early on in this work.

To try to ensure that the INSIGHT DataTAB would reflect a wide range of views, we thought about its diversity in terms of its members’ experiences and their characteristics. For example, we were keen to ensure that the opinions, experiences and perspectives of patients with sight-threatening conditions were represented, as well as those with experience in health data research and regulation. We also wanted to make sure the group was diverse in terms of characteristics protected by the Equality Act 2010, such as age, race and sex.

We then worked closely with AAAMD, different patient and public representative groups and the hub’s existing Patient and Public Involvement and Engagement teams to recruit members with these different experiences and characteristics.

As with bringing together any group, reflecting the diversity of the population you’re sampling from is hard. We were surprised that there isn’t more structured advice, guidance or tooling out there for organisations like us trying to design data governance processes to be inclusive and representative.

  • Establish something practical while also testing theory.

The INSIGHT DataTab is very real. It has a set of members onboard, an agreed terms of reference and the hub is committed to receiving its recommendations. With a more theoretical lens, we also set out to use this work to test our thinking around ‘data trusts’.

At the ODI, we describe that data trusts provide independent, fiduciary stewardship of data. They involve one party authorising another to make decisions about data on their behalf, for the benefit of a wider group of stakeholders.

With data trusts, the independent person, group or entity stewarding the data takes on a fiduciary duty. In law, a fiduciary duty is considered the highest level of obligation that one party can owe to another – a fiduciary duty in this context involves stewarding data with impartiality, prudence, transparency and undivided loyalty.

While the INSIGHT DataTab is responsible for providing oversight of the hub’s stewardship of eye scans and other ophthalmic health data, University Hospital Birmingham and Moorfields Eye Hospital remain, as data controllers, legally accountable for the data and liable for any risks associated with sharing it. We didn’t think it was appropriate for the INSIGHT DataTab members to take this accountability away from the hospital staff, nor try to formalise a fiduciary obligation around the members’ duties. With guidance from BPE Solicitors, we therefore decided that the INSIGHT DataTab members should be bound by a memorandum of understanding, which outlines their role, responsibilities and commitments.

In short, the INSIGHT DataTAB does not currently provide independent, fiduciary stewardship of data, so we do not consider it to be a data trust. This project has contributed to our thinking on data trusts, although balancing the need to deliver something tangible and effective for the hub with our interest in exploring this theory has not always been easy.

It is important to note that we are dealing with health data which is already subject to an established fiduciary structure and accountability. The existing structures, while not perfect, are trusted and well utilised. Understandably, people are reluctant to move to a new model. The work with INSIGHT serves as a pilot model, generating evidence to guide future discussions of implementing data trusts to govern and steward health data.

What went well

  • Learn from existing efforts.

There’s a lot going on in the health data research space, and a number of initiatives and groups already involve patients and the public in decisions about data.

At the beginning of our work, we invested time to dig into some of these. Some had been established to oversee access to data collected across the NHS, such as NHS Digital’s Data Access Request Service and the Health Research Authority Confidentiality Advisory Group. Others are focused on data collected for specific purposes – for example, the Personal Information and Access Sub-Committee helps UK Biobank to steward data and samples it has collected on around half a million people.

In other instances, some groups and processes that involve the public and other stakeholders lacked transparency. For example, we couldn’t find the membership lists for some groups, or the criteria used for assessing data access requests and the reasons for their advice being disregarded. As a result we enshrined provisions in the Terms of Reference that would seek to enable fuller transparency. For example, DataTab will publish its access criteria, make available its full membership list, produce detailed minutes of its meetings and receive an explanation from INSIGHT if certain advice is not taken on board.

As well as helping us identify areas for us to improve on with the INSIGHT DataTab, we also found aspects to pick up and adapt. For example, in order to avoid meeting to consider each and every request for data, many groups only convene to decide on those that are novel or contentious – over time, this is something we expect the INSIGHT DataTab to do too.

  • Involving stakeholders is a process rather than an act.

From the beginning, we saw the involvement of public, patient and other stakeholders not as a singular conversation to have or piece of research to do, but as an ongoing partnership. We began by attending events where patients were sharing their attitudes and expectations around the use of data, including a great workshop where members of patient advisory groups shared their experiences. We also spoke with organisations like Understanding Patient Data to test some of our thinking and highlight our blind spots.

Most importantly, we presented a very early outline of our plans for the INSIGHT DataTab to a panel of patient and public representatives at a workshop organised by Moorfields Eye Hospital. The workshop was vital in challenging some of our assumptions and in particular, in determining whose views should be represented on the group we were about to convene.

While we’re delighted with the group that we’ve brought together, we will actively seek to understand whether its membership should evolve over time – for example, based on feedback from its members about voices they feel are missing from the process. Helping to bring patients closer to decision making around data is not a single action, it’s a process.