Chaired by Lord James O’Shaughnessy and Professor Mike Bewick, the first inquiry session of the Commission took was held on the 28th June. This session focused on fostering the health data ecosystem, bringing together the recommendations of leaders in healthcare, life sciences and local government/regional authorities.
The inaugural inquiry session of Curia’s NHS Innovation and Life Sciences Commission contained a rich discussion of health data, aided by the distinguished panellists of thought leaders in this space.
The first half of this inquiry session focused on the role of National Health Service (NHS) staff in the health data infrastructure and establishing best practices to achieve greater health outcomes and reduce inequalities. The panel for this session included:
• Dr Ben Goldacre MBE, Director, Bennett Institute
• Breid O’Brien, Director of Innovation & Digital Health, NHS England and Improvement (NHSEI)
• Dr Claire Bloomfield, Deputy Director, Data for R&D at NHS Transformation Directorate
Training programmes and competency frameworks
The first topic of discussion centered on how NHS staff and current leadership teams can enhance their digital and data skills and what programmes or frameworks are needed to achieve this at scale. Opening the session, Ben Goldacre outlined his research when writing the Goldacre Review1. He noted the three main groups of NHS staff that require focus in achieving the skills and competency needed across the NHS:
Analysts find opportunities to optimise care delivery logistics, which involves improving the quality of the safety and cost-effectiveness of NHS delivery. At present, these positions have been neglected, yet the Government Economic Service (GES), Government Statistical Service (GSS) and Government Operational Research Service (GORS) provide a clear roadmap of how new technical professions could be created. The core recommendations of the Goldacre review highlight how policymakers should mirror the Government Analyst Function, career pathways and training pathways in the NHS, which has been taken up the least energetically in the data strategy.
Senior leaders with technical skills
Senior strategic leadership roles are needed for developers, data scientists, data architects etc. There is also a missing emphasis on leadership training for existing technical roles and how to make these roles more attractive to people from technical backgrounds.
Crossover technical skills
There must be an emphasis on crossover skills – combining software development skills with domain knowledge of NHS data, the basics of epidemiology and administrative structures.
To achieve this, the NHS may need to enforce a ‘golden handcuffs’ approach – meaning that the health service ensures staff who are paid to train in new data processes are contracted to stay within the system for a length of time. However, the only issue implementing this approach is the NHS are ‘recruiting’ instead of ‘selecting.’
Further to this, the issue of who sets these standards and how they are measured was discussed in relation to achieving success across NHS analysts, senior leaders and cross over technical skills. Goldacre explained that the NHS should follow the GES and GSS model, further noting the external Due Diligence Assessment (DDA) framework has real value in relation to this issue.
Goldacre noted the problem that people with practical technical skills are not certificated, disregarding ‘formal’ training as an indication of true value. The NHS also struggles with new forms of technical competency. Despite being able to tier job descriptions, salaries etc. to existing roles, the health service cannot provide the same for new software developers and data architect roles. On available solutions, he explained the need to focus on getting people with deep technical skills into senior leadership roles, not to write the code itself but to manage and recruit such people. Goldacre explained this will take a “couple of years” but it is achievable, allowing the NHS to plant the seed for progress which will snowball moving forward.
On establishing a chief officer role over these training and competency frameworks, Goldacre explained the need to combine multiple heads of professions from different areas. The case of NHSE showed that the existing organisational structures for new frameworks and training programmes were simply not there; with NHS analysts continuing to operate in very isolated groups. He detailed that expecting analyst groups to self-subsidise and self-organise was not going to work. Instead, we need to pay for the structures to be built.
Supporting existing staff and data analytical roles
Led by Richard Stubbs, the next section of this topic focused on supporting the existing staff within the NHS in fostering the data infrastructure. Breid O’Brien emphasised the need to capture data and for data to be engaging and accessible for staff, patients, carers and the voluntary sector. It is important to ensure frontline systems are user friendly and accessible. For NHS staff specifically, a significant focus on data is needed within their training. However, the biggest challenge is changing the public perception on the use of data. Implementing a narrative around data to ‘tell a story’ should help break down barriers that exist in this respect.
In addition, Dr Claire Bloomfield stated the need for a technical workforce but also a multiplicity of expertise and bringing together different disciplines to harness the UK’s health data potential. Currently, the NHS leaders are leaving data to technological people, but a more wide-ranging approach is needed. Bloomfield emphasised the need to focus on the quality of the data alongside how we are using it for insights. To achieve this level of quality, NHS leaders must question why this matters and “what the goals and the impact of the high-quality data will be.” Finally, she addressed the need for partnerships across the ecosystem; as the NHS cannot compete with the expertise from industry and academia, a level of fluidity between organisations and collaborative ways of working is needed to cross-fertilise different areas.
On incorporating data education into healthcare professional training, Bloomfield emphasised integrating higher education into the NHS England ‘family’ but that the previous attempts to integrate genomics showed that specific programs needed to be created for the clinical workforce. Unfortunately, there is not a ‘silver bullet’ training course to achieve success, but incorporating academic institutions is needed to kickstart this process.
Ben Goldacre noted the points within his review on training in clinical informatics, data fluency for senior leaders and reciprocity. In his view, the more data is used, the higher its quality will be. He said more needs to be given back to people for using their data as they will pay more attention to improving it. Regarding the workforce, Goldacre explained the tendency to look in the “pools of light” and neglect other areas. He explained NHS data analysts are successful in completing single analysis with a finished
dataset – but what the NHS health data leaders neglect is the vitally important pathway from raw data being collected to its curation and secure storing.
“The ABPI told us during the [Goldacre] review that 80 per cent of the work in an NHS data project, as far as they were concerned was ‘data curation’ and they said that 80 per cent of the spend from government should therefore be on ‘data curation’, and that’s still an area that’s being neglected” – Ben Goldacre
On Lord O’Shaughnessy’s question on ensuring needed public trust in use of patient data and curation, Claire Bloomfield explained there are two strands of curation: curation at source as data is generated and curation for specific research, noting these are not the same thing. The NHS is responsible for driving the quality up, but the wider industry needs to ask questions on how we are curating and reusing curated data from research to optimise efficiency. There is better scope for these conversations between industry, academia and the NHS on how trusted research environments can be curated, who does this and how data is retained and reused.
Infrastructure requirements for patient data storage
On the technical and infrastructure requirements, Ben Goldacre stated that critical change is needed to move away from the pseudonymisation of copied records and instead consolidate all analytic work for self-improvement, academic research and innovation in shared Trusted Research Environments (TREs). He explained the importance of this for several reasons:
- When data is disseminated out of numerous places, it is inherently unsafe and duplicates risk. People are knowledgeable of that now, which is a good thing to avoid privacy catastrophes and noted the three million people currently opting out of health data use.
- None of the data that is over duplicated is portable between different environments – which is a disaster for joining up co-ordinated work across industry, academia and the health service.
He noted the challenge is how does the NHS create TREs avoiding the mistakes of the past, particularly non-delivery. There is a need for a delivery orientated approach, in particular a workforce with a common set of knowledge and high status ‘cadre’ of people who create great platforms for others to use. In the past TREs have either been procured by giving money to closed organisations or given money to people with adjacent skills.
“We need to talk with UK Research and Innovation (UKRI) and the National Institute for Health and Care Research (NIIHR) in particular about how we can support a rich, collaborative and competition ecosystem around data curation, secure analytics and platforms for others to use.” Ben Goldacre
“We also need to think very carefully when we procure large data infrastructure from the NHS about how we can avoid the risks of ‘vendor lock-in,’ which is another form of monopoly and another form of obstruction to a shared ecosystem where people are competing and collaborating.” Ben Goldacre
On delivering this within a changed health policy landscape, Claire Bloomfield explained there will be a shift over the next three years in what we see in terms of the wider system architecture. The policy will push people one way in requiring TREs to be accredited and for them to be the default means by which NHS data is accessed. However, these services need to be effective and deliver – so there is a cultural, operational and policy opportunity to move the system in the right direction.
“We will need an ecosystem wide approach with NHS England, UKRI, NIHR and others all pulling in the same direction as we try and reshape what the landscape looks like over the next three year to five years” – Claire Bloomfield
Remove barriers around data and public trust
On the barriers that currently exist, Breid O’Brien noted the importance of framing the upcoming publication of the target data architecture for different audiences, as language is extremely important. She re-emphasised the significance of multiple use for curated data but noted most people want to know they can access their data when needed.
On the issue of public trust, Claire Bloomfield stated the nature of an ongoing conversation that needs to happen between patients and those using their data. We need to establish a far more comprehensive, overarching conversation at the national level – backed up locally and regionally about how data is used across care planning, population, health and research. Through this, understanding perspectives of different regions around data is achievable and public confidence can be ensured.
Ensuring shared care records for population health management at an ICS level
Dr Claire Bloomfield opened the discussion on care records, noting there is a clear gap that exists with social care records and what the NHS currently holds, highlighting the need to learn lessons from the past and ‘leapfrog’ social care data with learnings from the NHS. There is significant weight placed on Integrated Care Systems (ICSs) in co-ordinating this activity, yet there is huge variability in the components for ICSs and regions to work through. The NHS cannot fix everything overnight, but there should be a focus on what can be a ‘quick win’ and what over time will improve interoperability and standards that will allow other data sources to feed and flow. She listed two priorities which will help in this regard:
- Embedding the research requirement into ICSs
- Through R&D in data, the NHS invests in ICSs and TREs
In addition, Goldacre noted the importance of having practical, shared code and methods to facilitate data curation for others in the ecosystem to use. Particularly with social care, building within this space openly so others can learn more is essential. On the issue of services operating within ‘black boxes’ he said: “There has been no culture or expectation of people sharing the technical work that they did ‘under the bonnet’ to produce a shared dataset, a shared technical environment and so on.”
Goldacre also noted that data curation, secure analytics and running platforms for others is the most difficult challenge at hand, exacerbated by the lack of an open “knowledge commons” and a competitive and collaborative ecosystem.
In closing the first part of the session, Lord O’Shaughnessy discussed the challenge of ‘opt in, opt out’ in patient data, noting the importance of the NHS App in facilitating a constant conversation with the public.