HTN and a panel of digital leaders hosted a webinar focusing on legacy systems, and the challenge of technical, architectural, and organisational debt. Panellists included Ben Attwood, CDIO at Oxford University Hospitals; David Reilly, director of digital at Countess of Chester Hospital; Mike Hardman, practice lead for development at Aire Logic; and Stuart Stocks, principal enterprise architect at Aire Logic.
Starting out building a picture of what kinds of legacy systems and technical debt NHS organisations are currently working with, and the journey to date; our panel then moved on to discuss approaches to tackling common challenges, lessons learned, benefits, impact, and future outlook.
Starting off with an introduction, Ben told us about his role as CDIO at OUH and also as the senior responsible owner for the Secure Data Environment for Thames Valley and Surrey. “I’m also still a practicing clinician, working as an intensive care and anaesthetics consultant,” he added, “which helps me understand what it’s like as a user of systems.” OUH has 14,000 colleagues across four different sites, and a number of community areas as well, presenting challenges around hardware and clinical applications.
“I think all NHS organisations will recognise that those technical debts build up servicing those sorts of structures, and we are working hard to ensure everything is up-to-date,” Ben continued. “We have a long term EPR programme, and that’s been an interesting journey, as when we started there were quite a lot of areas that didn’t have a capability or functionality, so lots of clinical and supporting systems have had to come in.”
Mike gave us a bit of background on Aire Logic, and his work at national, regional, and local levels across primary and secondary care. “We work on greenfield new developments, and on running, maintaining, and improving programmes, which means we interact quite a lot with an organisation’s technical and organisational debt,” he explained. “We help navigate that space and figure out what the debt truly is, how it can be managed, and what maybe needs urgent replacement.”
David described working in a number of different roles across Cheshire and Merseyside ICB, spending part of his career specialising in integration and systems integration. “Like anywhere else, we don’t have an endless pot of money to be refreshing things all the time, we work on a risk assessment basis, forward planning and looking at technologies coming out of service or out of support in the next few years,” he said. The Countess is also focusing on maximising its EPR wherever possible. “We do have gaps, and a number of legacy systems we don’t have functionality in the EPR to replace, so they may require smaller specialist systems with specific integrations,” he noted.
From a server point of view, the trust is currently undertaking regular refreshes, but has plans for consolidation and collaboration across the region, David told us. “Over the last 12 months we’ve moved from local PACS to a regional PACS platform, and we’re moving to a regional LIMS solution that will take it out of the EPR and into a dedicated LIMS that serves the entire region.” Now, the trust is in “an OK position”, he went on, “doing what we can with what we have, and with a sight of our risk, so we know what our tech debt is, we regularly monitor that, and when opportunity comes up we see what we can do to improve that position.”
“I’ve been with Aire Logic for six years, working similarly to Mike across NHSE, arms-length bodies like NICE, secondary care, and moving into that summary care record space as well,” Stuart said. “There’s been a lot of interest in centralisation of EPR systems, updating EPR systems, removing barriers to providing care that are features of the tech landscape.” Aire Logic has been running for about 18 years, assisting with application landscape reviews, application portfolio management, technical debt assessments, EPR procurements, and more, he added.
What works when tackling legacy and technical debt?
David elaborated on his risk assessment approach at The Countess of Chester, highlighting the benefits of having that insight on overall position and how that has helped with bringing that position to the board or non-digital colleagues. “We had to do a full storage area network replacement as it had reached end of support, and when we applied for capital funding we had the board’s support, because they knew that was coming,” he explained.
Conversely, what doesn’t work so well about that approach, is that “sometimes things you don’t perceive to be a risk can become one very quickly”, David acknowledged. “You focus on what you foresee as the highest risk areas, but we all know the world we’re living in with various vulnerabilities and threats to our networks and organisations; it might be something lower down that list that ends up becoming a problem. The risk-based approach very much reflects a single point in time.”
Ben agreed that the risk-based approach “resonated”, with OUH including organisational debt in its risk register up to the board, but stated that that should then flow through to outcomes and impact. “On a more granular level, we’ve got lots of departmental systems which have been historically point solutions in specialties – local LIMS, PACS modules, local audit and registry tools, and initially a host of apps based on access or other older applications.” For infrastructure, upgrading to Windows 11 has been helpful in getting people to focus-in on technical debt and cyber risk, he considered.
“Understanding our HL7 interfaces and batch messaging, how we manage our integration engine – I think every NHS organisation will recognise those as potential areas of debt, and then workarounds which have been put in necessarily to manage workflows, that are well meant, but our job is to remove the need for those heroics,” Ben continued. The trust has focused on consolidating and retiring systems where possible, and absorbing things into its EPR, adhering to modern architecture, APIs, and a FHIR-led approach. “What we’ve found is that when you’re lifting and shifting something without a redesign, you’re just moving cost, you’re not moving value,” he said.
The other thing, Ben mentioned, is “pilots without a runway” – when pilots or proofs of concept are created but then can’t be scaled. “That creates technical debt, because you’re then trying to manage something in one part of the organisation that isn’t scaled across.” That can be common due to lack of funding, which means there is difficulty with managing data quality and configuration to support it, he concluded.
“For me, there’s two kinds of debt in this space – intentional debt that we accrue so we can learn and de-risk, knowing there will be a clean-up cost later; and accidental debt related to rushed changes, unclear design,” Mike noted. “Those things accrue over time, and if we don’t get on top of them they can quickly snowball.” That means monitoring debt as cost and risk is integral, including a time sensitivity element which recognises that risk may increase with the introduction of new things, becoming a bigger debt.
Organisations should also consider what guardrails they have in place to prevent debt from “silently accruing” in the background, which might mean it becomes an issue for clinical or operational teams, Mike suggested. Things like regular organisational checks, hygiene checks, and architectural boundaries that can be put in place. “It’s a constant assessment looking at areas with planned changes that might need a focus, and areas with no changes planned which might be fine to leave for now,” he said. “Quite often things crop up due to people not paying the right attention to this as a concept and just assuming teams are going to manage it themselves.”
“Creating a level playing field is my headline for this discussion,” Stuart said. “Do all those involved in decision making around where the money gets spent have an equal understanding of what the risks are and what the debt is?” When working with organisations, often there are pockets of highly motivated people pointing to the need to make fixes, but at board level they are met with silence, he went on. “They don’t see it, they don’t realise what’s going on. Risk registers are a brilliant way of exploring that. It needs an enterprise architecture approach to understanding the different layers – if people are complaining about Wi-Fi, is that the problem, or is it a debt problem?”
The strategy of the organisation influences how architectural and technical debt is described, Stuart continued. “One person’s debt might be another person’s legacy, so they may be happy to live with that legacy for a while, depending on the strategy of the organisation, whether it is to centralise onto a single EPR, or go best of breed. It’s all down to context, and the context of your debt assessment, how you can understand implications, and have a set pathway to make improvements.”
Workforce skills
Part of the issue with an acute organisation is the scope, size, and variety of things in provision, covering a whole suite of clinical and operational scenarios, Ben told us. That makes workforce skills very important, bringing people up to speed on the latest developments. “With EPR that’s harder to do, because the workarounds build up, those are cumulative, and every time you add a click you are accumulating a debt. We have to look at reducing those, which has both productivity and wellbeing improvements, so we’re not asking our colleagues to do more work and spend more time clicking through things.”
Business-level things like email, documents being moved to SharePoint, and surfacing data through Power BI, make upskilling all the more necessary, according to Ben. “Copilot is worth talking about because it’s the first AI offering across all organisations. We happen to be on a local tenant, so that meant we were able to deploy earlier – we ran a trial with 1,000 licences, and we have seen benefits through that, but it’s also enabling people to be more structured and efficient with those operational processes and day-to-day tooling.” Generally upskilling is crucial, and is reflected in the 10-Year Plan and Medium Term Framework, which state people will need to be AI literate in a short timeframe, he went on, “so we need to get people much more on board with skills and the tools that are becoming available”.
When looking at debt, it’s integral to consider its impact on the ability to change, Mike said. “Some of this infrastructure has existed since the very earliest implementations of a hospital, and because it was done well, it has changed and adapted as the trust and the technology around it has grown. A small thing that is implemented, but which is very dependent on the existing structures and systems, suddenly becomes a risk to changing anything it touches, because it is inflexible and can’t be changed very easily itself.”
Many organisations struggle with siloed thinking, Stuart highlighted, driven by different departments and their leadership implementing differently or taking alternative approaches. Organisations have to understand the consequences of not having a standardised way of doing something, like having multiple systems staff need to know how to use or have multiple logins for. “Ideally, you would have a more tailored solution at the front end, with a data structure behind it allowing things to be done in a more collaborative way behind the scenes. When we look at the New Hospital Programme, they’ve taken a design-led structure with an enterprise-level blueprint taking a view of how the building itself can support the flow of clinical abilities. That, by design, is breaking down siloes and barriers – part of being able to see where debt is and manage it, so we can see what is going to have the biggest bang for our buck.”
Tackling legacy and promoting standardisation
David talked about Countess of Chester’s EPR, highlighting its set of model contents, or functionality that has been developed that can then be incorporated into the core platform. “Sepsis is one we’ve adopted, and as a principle we’ll be looking to adopt those wherever possible. There will always be elements of localisation, where you need to break away from the prescribed solution, but we try to stick to HL7, FHIR, and open APIs to help drive development and growth.” That comes with limitations, however, like with the example of electronic prescribing, which has been around in primary care for a long time, but is still in its infancy in secondary care. “That has led to disagreements on what standards should be followed and who is dictating them, and sometimes you need to find a supplier who can jump in and fill that gap, which might not meet national standards, but will solve the problem for that time period.”
There’s always a need for customisations, but the trust doesn’t do a lot of in-house development, because it doesn’t have the skills, David said of the Countess of Chester. “They’re difficult skills to bring in and keep within the NHS, we don’t have the luxury around a lot of our own bespoke developments in terms of apps and so on, but we do try and use the tools where we can to stay within standards.”
“We’re always in a cycle of needing to move faster or take on functionality that isn’t fully developed,” Ben told us. “The NHS App is a good example – we’ve got a clear mandate that organisations need to be able to connect to it, but to do that and align various services means differential upgrades or differential reconfiguration. The other thing is to look at the lifecycle of EPR, patient portal, and so on, and then list all the things we need to do which will form the spec of this next big reprocurement or procurement exercise and then build those into that requirement.” The way NHS infrastructure and architecture works, it’s not possible to implement things without some impact or compromise, he added.
The best approach to mitigating some of those challenges is being more agile, Mike suggested, “embracing change and not necessarily making everything such a massive piece of work or approaching it as a huge multi-year procurement”. Making change more continuously and doing that in smaller pieces as part of the day-to-day has benefits of preventing the build up to “Big Bang” implementations and steadily reducing debt. “That’s a much more holistic approach to managing debt over time, because when you’re changing continuously you can tackle large risks almost straight away,” he said, “rather than leaving it to the next procurement cycle which might be years down the road.”
Funding is always a barrier when it comes. to taking on new things and sustaining services, David noted, stating that Countess of Chester’s approach is often to look at what’s available in existing funding and being upfront at the point of a business case that there isn’t a line in the revenue budget to support it, but outlining the benefits it would offer for the organisation. “A good example is our patient portal, which we received funding for a few years ago. That funding is due to run out in March, but in our initial business case we showed the benefits of the portal that would help fund it in things like reduced postage and DNA rates.”
Things have changed very rapidly over the last couple of years, and the potential for AI moving forward is “mind boggling”, David said. “Digital investments have to be made with that in mind. This isn’t something you’re just putting in for three years, this is something you hope will transform the way things are done within an organisation, and that should be supported from a revenue point of view. We have looked for tangible, quantifiable benefits we can evidence and been upfront with them at the outset, because otherwise you just create a problem for yourself when the money runs out.”
Future plans
Our panellists discussed their plans for the next 12 months or so, with Ben pointing to the need to prepare for AI. “AI needs a big infrastructure layer behind it to make it really work, and I’m keen to see how the platforms on offer might consolidate over the next year or two, because I think the major tech organisations are starting to want to recoup cost of research and development.” Those costs might be reflected in things like RAM costs and devices, he assumed.
OUH is currently having conversations with the local system and key provider collaboratives around what can be achieved via shared platforms and shared work, Ben went on. “We’re talking about how we can align some of our cycles, because that helps align some of the debt other organisations might have – in one way it adds complexity, and in another it helps consolidate and think about where we want to get to on data storage, servers, contracts, etc. We do have a full EPMA, and the challenge has been to re-regulate our EPR to adapt and adopt model content, and we’re also focusing on clinician experience and getting back to patient outcomes.”
Mike echoed plans to consolidate and make change more of a continual process, assessing debt position regularly, and working with organisations to try and identify whether something is debt or legacy. “Often things get labelled as technical debt, when it’s something we could probably just make a small organisational change and resolve,” he said.
David spoke about a regional Ambient Voice pilot preparing to go live in the next month or two, that should help make the shift from the Countess of Chester’s current e-communications solution to an EPR-developed solution. “We’ve got some work to do on our core network switches, which is a bit frightening, but we’ve got capital approval to do that, and that’s been part of making people aware of the risk. To replace those is a big task, and that will happen later this year. We’re also going to replace our legacy proxy solution, so we might secure funding to procure a new solution, and then we’ll continue to develop and enhance our EPR to bring new services or digitise processes where possible.”
“I’m going to champion the sharing of information and knowledge,” Stuart shared. “Focusing on getting everyone on board and building a shared understanding of what AI means in the organisation. Look at whether any work is needed to clean data up, and so on. My ambition for the next 12 months is to minimise the creation of new debt environments, and make sure environments people operate in are more debt resistant.”
We’d like to thank our panellists for taking the time to share these insights with us.
