The UK’s National Health Service Blood & Transplant has deployed cloud-based BPM and Decision Management technologies from IBM to transform the operation of its organ allocation schemes. The organisation also uses cloud-based collaborative modelling tools to enable varied groups of subject-matter expert stakeholders work together to specify key aspects of the schemes it’s addressing.
Case study key facts
|Organisation||NHS Blood and Transplant – a Special Health Authority within the UK’s National Health Service|
|Current goals||Complete implementation of cloud-based workflow and decision automation systems to support a variety of organ allocation schemes – reducing the costs, risks and time involved in allocating organs to improve outcomes for patients.|
|Current approach||NHS Blood and Transplant is currently part-way through a programme of work to apply BPM and Decision Management technologies to its organ allocation schemes, replacing out-dated and brittle application software. The first scheme to be implemented – for the allocation of hearts – took just 6 months to implement, using a co-located and cross-functional team.|
|Outcome||Through its implementation of BPM and Decision Management technology, NHSBT has vastly simplified its heart allocation scheme and reduced allocation times. New schemes are set to deliver further life-saving and life-improving benefits.
Further, NHSBT has built a platform on which it can continue to streamline and digitise organ allocation schemes – and make these schemes easy to change with confidence as new allocation outcome evidence comes to light.
|Tools and suppliers used||Blueworks Live, Business Process Manager (BPM) on Cloud and Operational Decision Management (ODM) on Cloud from IBM; implementation services from specialist IBM partner T-Impact.|
NHS Blood and Transplant (NHSBT) was formed in October 2005 from the merger of the UK’s National Blood Service (founded in 1946) and UK Transplant (formed from a number of agencies originally formed from 1968 onwards). NHSBT operates as a Special Health Authority: a NHS Trust that operates nationally, rather than providing services to one particular geographic area.
NHSBT is responsible for providing safe blood supply to hospitals in England, and providing tissues and solid organs to hospitals across the UK. Each year, NHSBT works with donors giving around two million donations of blood and 3,500 organs, saving and transforming countless lives. It collects, tests, processes, stores and delivers blood, plasma and tissue to every NHS Trust in England; and it matches, allocates, audits and analyses organ donations across the whole of the UK.
Project background and drivers
Matching and delivering donated organs to recipients involves the co-ordination of a complicated, highly time-critical set of operational processes. Over years, NHSBT has developed and continuously improved allocation schemes for all the types of organ that can be donated: from kidneys (involved in over 80% of organ transplants) to livers, pancreas, hearts and lungs. Each organ type has its own allocation scheme, based on a number of factors – including of course a number of clinical matching factors (age, body size, tissue match, and so on): but also including estimates of the likely benefit of a donation to a given recipient and ‘fairness’ criteria (prioritising recipients based on the length of time they’ve been waiting for an organ).
Crucially, it’s not only the case that the allocation schemes involved in matching donated organs to recipients are complicated and time-critical; they also evolve over time. Advisory groups comprising expert doctors and surgeons meet regularly to review clinical outcomes associated current schemes, and make changes to best-practice guidance.
NHSBT started on its journey to use software to support organ allocation schemes in 2002-03, when it deployed an Oracle PL/SQL-based application supporting a patient-centred kidney allocation scheme. However over time the application became more and more difficult to maintain, and system documentation became more and more out of step with the application’s core logic. It became difficult to understand the potential impact of changes to decision logic across the application, and difficult to design effective functional testing regimes for the application.
By the time Aaron Powell, NHSBT’s current Chief Digital Officer, joined the organisation in 2010 it was estimated that making a significant change to the functionality of the application – in order to update it in line with current guidance – would take two years. With the number of organ donors increasing by 75% in the period from 2008 to 2016 and the resulting need to further streamline allocation processes, it was clear that something had to change.
Powell initiated a significant shift in IT strategy from the end of 2014, with a shift to a cloud-first application and platform posture. There was now another reason to prioritise redesign of the existing organ allocation system.
Implementation characteristics and status
After considering technology from Oracle, Appian and IBM, in late 2015 NHSBT decided to work with IBM to implement new systems to support organ allocation.
The organisation felt that IBM’s combination of workflow and rules automation offered the best fit to its functional requirements, and in addition felt that IBM’s Blueworks Live collaborative mapping and modelling tool would help it drive requirements discovery and definition for its new system. In line with NHSBT’s broader technology strategy, it’s used IBM Business Process Manager (BPM) and Operational Decision Management (ODM) products hosted on IBM’s SoftLayer cloud.
Work started in February 2016, and by October 2016 the first implementation was complete and in live operation.
At the time of writing this case study, in May 2017, NHSBT now has two organ allocation process applications live on its new platform: heart allocation (the first to be implemented) and lung allocation. A scheme for liver allocation is currently being implemented with planned go-live towards the end of 2017.
Right now, the allocation platform is being used to manage around 600 transplants a year across eight transplant centres. There are around 50 users of the platform today, but NHSBT aims to increase this to over 200 as more allocation schemes are implemented.
The implementation team was very clear that in starting to implement new technology to support organ allocation schemes, it had to start with heart allocation. There were two key reasons for this.
First, the opportunity for improvement was well-bounded and clearly understood. Prior to the implementation of the new system, NHSBT’s heart allocation process was partly supported by the organisation’s existing Oracle application; but when prioritising urgent cases, NHSBT had to keep a manual record of individual recipients – having staff write out lists of patients and their details with markers on an office whiteboard. Any changes to this list (for example, when a recipient received an organ, or when a patient died) – involved the list being erased and manually rewritten.
Second, the programme team wanted to test and prove new technologies in the context of a process with a relatively low-volume flow of work and decisions before applying lessons learned to larger-scale opportunities. If, for example, the team had started with liver allocation, the resulting software would have required a much larger and more involved testing effort; and it would have been much more difficult to convince stakeholders that the new technology could deliver the needed results with the right performance.
From the start, NHSBT realised that in replacing its Oracle system, the complexity of the rules it wanted to encode in new applications meant a specialised rules engine would be required alongside a workflow component. However it found that many of the common Java-based business rules engines (BREs) available were aimed principally at specialist software developers. However the team also realised that it needed to be able to get the medical statistician subject-matter experts who design NHSBT’s organ allocation schemes directly involved in the application implementation projects, rather than imposing an extra layer of interpretation in projects (with software developers interpreting the knowledge of the subject-matter experts). Everyone agreed that involving the allocation scheme designers directly would increase transparency, and increase business confidence in the technology as well as making it easier to understand the impacts of changes.
NHSBT’s shift to a cloud-first IT strategy was part of a broader recognition: that although NHSBT has a unique mission and offers a set of unique services, its operational needs are not really unique. From an operational perspective its needs are similar to those of many commercial and third-sector organisations: it needs to manage marketing to and interactions with ‘customers’; it needs to manufacture and deliver products; and so on. It doesn’t need to custom-build unique systems to support these functions, and it certainly doesn’t need to run its own custom-built data centre.
Today, NHSBT uses IBM BPM and ODM largely as application development technologies, within the context of its applications refresh work – rather than seeing business process management more broadly as a strategic operational management approach.
At the business process architecture level, NHSBT uses IBM’s Blueworks Live as its business process and decision specification definition tool and repository. Workshops are led by business analysts from the NHSBT core team.
From a technical architecture point of view, the T-Impact resources working as part of the NHSBT team (see below) are working to well-established architectural principles, including some specific to IBM BPM and ODM. Specifically:
- As far as possible, the team is working with the out-of-the-box capabilities of IBM BPM and ODM; treating customisation as a last resort.
- System performance is designed-in right from the start, with user responsiveness as the key metric under management.
- Service and process definitions are reused as far as possible, through the creation and use of IBM BPM and ODM ‘toolkits’. T-Impact is developing and maintaining common toolkits that implement REST-based integrations; UX features; and data objects.
Organisation and people
From the start, NHSBT knew that in order for it to be successful, it had to embrace an agile, collaborative solution design and development approach. With help from T-Impact, an IBM partner with deep expertise in digital business architecture, IBM BPM and ODM, it assembled a team comprising a couple of analysts and four developers, and paired them with four NHSBT internal developers (to facilitate knowledge transfer over time). It augmented this team, in the initial stage of the first project, with two IBM product specialists. The team was co-located at one of the NHSBT’s main offices in Bristol, UK.
NHSBT and T-Impact have implemented a ‘scaled agile’ framework for application development using two scrum teams. Resources are planned a quarter-year in advance; interim releases are chained together through each quarter. The team conducts all the familiar agile ceremonies: including conducting daily standup meetings, tracking progress through burndown charts, and so on. The initial implementation of the heart allocation scheme represented NHSBT’s first serious attempt at running an agile software development process.
Crucially, this programme of work is not formally delivered by NHSBT’s IT organisation. A leader from ‘the business’ is accountable for programme delivery, and responsible for making important decisions about implementation priorities. The programme leader manages a small group of product owners, again from within the business, focused on different areas of work (for example individual allocation schemes, workflow implementation, and so on). Here, NHSBT has been influenced by the experiences and advice of the UK Government Digital Service (GDS).
As scheme implementations become part of ‘business as usual’ and as new advice becomes available from transplant advisory groups, this advice will be used to create change specifications, with the results prioritised by advisory group chairs. The resulting changes are fed into the three-month SAFe release train.
NHSBT plans to retain a small internal technical team that will deal with small-scale change requirements. Larger-scale functional changes will involve the use of partner resources alongside the internal core technology team.
Technology and infrastructure
As mentioned above, NHSBT is implementing new applications to support its organ allocation schemes using IBM’s Blueworks Live, together with IBM BPM and ODM running on IBM’s SoftLayer cloud.
Initially, the new platform managing organ allocation schemes is integrated in a point-to-point fashion with the legacy Oracle system, via a REST API. This API is used by the new platform to retrieve and manage data held on donors, recipients and transfer units. Over time the team plans to introduce an Enterprise Service Bus (ESB) as an integration intermediary, enabling NHSBT to replace existing systems more flexibly.
Today, NHSBT uses the out-of-the-box IBM BPM Coach framework to present application user interfaces to allocation staff. Coach customisations are limited to the use of custom widgets and layout specifications.
Although it’s still relatively early days for the NHSBT organ allocation scheme digitisation programme, there are already clear benefits being realised – thanks to the hard work and dedication of the combined team.
One of the starkest improvements seen so far is that in heart allocation, process steps have been reduced from around 100 (many of which were manually conducted) to around 60, with an attendant reduction in organ allocation time. More broadly, the key benefit is a huge improvement in the level of control that NHSBT now has over allocation processes. Further, the liver allocation scheme currently under development is expected to lead to around 50 additional lives saved each year, through faster and more accurate organ allocation.
As well as continuing to implement systems to support other organ allocation schemes, NHSBT also has plans to continue to improve the process and systems it’s built so far. One such example improvement is to shift from static approaches to allocation, to dynamic schemes – that take into account the context that exists at the time the evaluation is made, including the length of existing waiting lists, up-to-date information about trends in organ availability, and so on.
Of course, there have been challenges. Although staff on the ground have found the changes involved in introducing the new platform relatively painless, there have been technology challenges relating to the integration of the new cloud-based platform and the on-premises Oracle-based legacy system (principally relating to the implementation of single sign-on). The team has also found that it’s had to carry out more development of its older systems than it expected, in order to facilitate integration with the new platform.
Recommendations for adopters
As part of being interviewed for this case study Aaron Powell, Chief Digital Officer of NHS Blood and Transplant, shared the following advice for other organisations pursuing similar kinds of initiative.
- Firstly, make sure you think through the integration implications of what you’re building as early as possible, so you have a clear view of what the integration task looks like in terms of cost and complexity.
- Secondly, make sure you start your implementation work with a scenario that’s big enough/complicated enough whereby improving the situation will return a significant business benefit; but also, a scenario that’s not completely business-critical (so if there are delays or complications there are no existential business implications).
- Thirdly, don’t be tempted to expend a lot of effort on analysing your current system and process state; with modern process and decision platforms, you have the opportunity to apply technology to real-world problems – this will help you understand things much better than doing a lot of work on paper. Don’t be afraid to get your hands dirty!
Best practice insights
Through the implementation of its organ allocation schemes using BPM and Decision Management technology with IBM and T-Impact, the combined team has demonstrated three particular aspects of best practice that you should think about in the context of your own implementation – that hold true regardless of your own industry or business domain.
Firstly, when it started out NHSBT was very careful to select an implementation that was well-understood in terms of its scope and potential impact; but that was modest enough in terms of scale and ‘business’ impact to mean existential risks would be avoided if the implementation were to go awry. Through this first implementation the organisation managed to get to grips with a number of thorny issues and educate itself about the potential of BPM and Decision Management technologies – while also solving a huge pain point.
Secondly, the team has exhibited real “platform thinking”. The teamhass architected its application implementations to maximise component reuse, and avoided customising platform functionality wherever possible to minimise potential risks associated with future cloud platform upgrades.
Lastly, NHSBT made sure to co-locate its implementation team. Particularly when starting out on a BPM implementation programme, as far as possible co-locate all the members of your team – software developers, business analysts and non-technical businesspeople helping to provide subject-matter expertise. Physically co-locating a diverse team is the fastest way to get the mix of skills you need to combine, working at full strength to address the opportunities and challenges you’ve identified in your programme’s terms of reference.