Developing advanced analytic capabilities in the NHS has been saved
Developing advanced analytic capabilities in the NHS
6 min read
In the second series of blogs on data leadership in UK public healthcare, we look at the challenges of enabling advanced analytics. We’re delighted to have perspectives from Adam Steventon, Director of Analytics at the Health Foundation, Ronan O’Leary, Course Director for Cambridge University’s Masters in Healthcare Data, and Specialty Lead Neurosciences and Trauma Critical Care at Cambridge University Hospitals NHS Foundation Trust and Monica Jones, Chief Data Officer for HDRUK North and DATA-CAN.
Where are the gaps in advanced analytics and data science capabilities in the NHS and what can we do to tackle them?
Adam: The NHS should be ambitious in its use of data, we have national hospital administrative data sets, the NHS number, and very good universities with plenty of people emerging with excellent data science skills. So, there's an opportunity to be at the forefront of using data to drive treatments, outcomes, and efficiencies across the system. However, we do fall short of that and the challenges can be grouped into three areas. Firstly, there's the demand for analysis, how non-specialists understand how data can help them in their jobs.
Secondly, there are supply issues around analytical skills, access to data, and right development opportunities and career pathways for analytics.
The third area is about reconciling demand and supply, the need for good analytical leaders who can engage with leaders in their organisation to understand where analysis can add value.
Ronan: Broadly speaking, there are three components to the gap in capabilities. Firstly, the NHS, as a national organisation, has perhaps been a little less effective at identifying resources for informatics. For example, the Epic electronic patient record at Cambridge University Hospitals has been an internationally recognised, award winning deployment of a comprehensive EPR across a complex organisation.
The second reason is the professionalisation of data sciences within the workforce - during training healthcare professionals are not taught to think about recording and reporting data, classifying diseases, and accurately describing treatments. As healthcare systems, at an international level, we need to go through a professionalisation of how people work with informatics systems and data.
The third reason will under pin the future success of data within the health environment. We need to train people to use the electronic patient records as health informaticians, not just to show them how to order and prescribe.
Monica: Analysts working in the frontline of the NHS lack of opportunity to do any advanced analytics as they predominantly work on data wrangling and cleansing, with a focus on the operational running of the Trusts. The analysts have the skill sets, but they don’t get the time or space to develop these further. Providing the space and upping the skills was an objective of the Yorkshire and Humber Care Record via the Population Health Management (PHM) Academy. Through efficiencies of automating the ordinary work, frees up more time to do the extraordinary, and upskilling the analysts using latest technologies like Google Cloud Platforms and advanced analytics tools (Python, R) gives them that leg up.
What key things need to be done at a national level to close these gaps?
Adam: There needs to be a strategy for developing analytical capability across the NHS, addressing the supply of analysts - how we get the right number of analysts with the right skills. National programs, such as the long-term plan or integrated care systems, should use their influence to help describe what good analysis could contribute to those programs, and to set very clear expectations of how this would be delivered.
Ronan: The NHS is good at using a national model to impose improvements. For example, we’re well regarded internationally for strokes from trauma because of the mandated care systems we have ensured are delivered in all acute trusts. This model has been effective in a clinical sphere and may be effective in the informatics environment. A nationally delivered way of ensuring that Trusts develop a supportive training and utilisation culture around health informatics could deliver a significant change and improvement.
Monica: We need to have closer ties with academic institutions. In Leeds we have the Universities and Leeds Teaching Hospital Trust. Having a common way of working and the expectation that working at the university in health research will provide you with an honorary contract at the hospital and vice versa can get people working in these multidisciplinary teams.
Each side bring in their own advantages; individuals from academia can energise and manage expectations, whereas those coming from the NHS route have in-depth knowledge of the systems and how things work, putting them together makes them greater then when working apart.
What would have the most impact at the local level?
Ronan: Firstly, improving accessibility is key - connecting the data to clinicians who are delivering a service.
Secondly, not using electronic systems like paper records with lots of free text, but instead standardising ways of putting data in, so that we don’t compromise the richness of the questions that we can ask.
Finally, using data to be more responsive about changing services. With real-time electronic patient records, we should be able to make real-time changes, we should be able visualise problems, model them, do small scale tests and then scale up to effective system wide change.
Monica: Joining up the different systems (primary care, secondary care, community care, mental health, commissioning function and the local authority), as you then really understand the value of data and unlock the power. NHS still measures things episodically, so often this is not in context, but by joining up those organisations up the town/city gets an identity.
What examples have you seen recently that model these changes?
Adam: Data sharing of coronavirus test figures has allowed local organisations to rapidly adjust their critical care capacity, and data has been critical in informing the public understanding of the disease and the progression.
The national data strategy for health and social care has been published in draft form, and includes a focus on building analytical capability, as well as a commitment to open and reproduceable analytics. This ambition is welcome, though needs to be supported by a clear and credible implementation plan.
The NHS R community is built around open-source programming languages in health and social care and is another example with tremendous potential to benefit from learning across organisations. The community is a really good way to engage analysts - we all like talking about the software that we use, learning how to code better, making better use of the data and the technology. Finally, the Association of Professional Health Care Analysts (APHA) are doing good work to progress the professionalisation of the health analytical workforce with professional registration.
Monica: On a local level it would be the Rotherham Health Record is an electronic system for sharing health information from GP, hospital and social care providers with the health and care professional who provide direct care to the individual. Then in terms of the symbiotic relationship between academia and NHS at Leeds there are three universities and the Teaching Hospital Trust, with the YHCR bringing that all together. Finally, at a national level the NHS Digital Cancer Trusted Research Environment (TRE) and HDRUK.