Draft eHealth Use Case Definition Stephen Kay Executive Summary This paper describes an initiative to produce the ‘eHealth use case template’ standard that would meet the business need of eliciting, specifying and communicating requirements to and between various stakeholders. In particular, it is seen as a way of bridging the gulf that can exist between policy-makers and implementers of eHealth interoperability by constraining the meaning of ‘use case’ and specifying content appropriate for, and relevant to the domain. The on-going preference for ‘use cases’, ‘the use case approach’, and ‘use case driven projects’ in eHealth initiatives is arguably a ‘good thing’ but its value is undermined to a large extent by the lack of consensus as to what a ‘use case’ actually means. This paper describes an early attempt by CEN/TC251 to produce an eHealth use case template standard that would meet this communication need. To date, the draft ideas have been used/modified by two EU projects (Antelope and Trillium Bridge), and one UK based project which was intended to provide interoperability guidance for commissioners. The latter has been proposed (not yet adopted) for use by the a new EU project which has just started in 2015. If it is, the feedback received will be used to produce a further iteration based on good practice. The implementation experience is intended to refine the initial idea so that a useful and usable eHealth template can be shared by the community in this specialised domain, based upon sound principles. The critique of this template will further input so as to improve its quality, and achieve the long-standing goal of a shared use case description standard that works for eHealth. 1 Introduction The on-going preference for ‘use cases’, ‘the use case approach’, and ‘use case driven projects’ in eHealth initiatives is arguably a ‘good thing’ but its value is undermined to a large extent by the lack of consensus as to what a ‘use case’ actually means. There are a range of descriptions, which are at different levels of granularity and specificity, and because the content too is often described in very different ways, the use and application of the term is in danger of becoming meaningless; a hindrance rather than a help to effective communication. These differences make it difficult to assess or compare different use cases and, that difficulty in turn, significantly complicates any attempt to derive a repository of use cases for subsequent prioritisation, comparison, and testing. As well as making a complex domain more confusing, it also leads to duplicate effort and hinders the re-use of existing work. Ironically, in the EU context, ‘use case driven projects’ are often funded to advance ‘interoperability’ which fundamentally requires ‘consistent agreements, specifications and systems’ if it is to succeed at all! As might be expected, what is meant by ‘use case’ is often coloured by the stakeholder that uses the term. Although ‘Use cases’ originated in the technical, modelling area of information systems and computer science, its simple, sometimes simplistic portrayal of complex requirements, has seen the general idea being taken up by the non-technical, business oriented stakeholder in order to express their ‘system’ problems. The ‘use case’ is attractive for that very reason; it potentially offers a way of bridging the communications gap that can exist between the non-technical and technical stakeholders. This potential, however, is rarely realised because the stakeholders have very different interpretations of the same concepts, and the two sides misunderstand, almost, by design. The problem-holder and the solutionprovider fail in their serious (and often costly) attempts to specify and meet the ‘known’ requirements. A further frustration is that the use cases generated by a single project are hardly ever reused by any other, 9th July /2015 Page 1 of 11 Draft eHealth Use Case Definition Stephen Kay resulting in both a loss of learning and of investment. A use case approach, like a messaging approach can become uncontrollable with instances increasing dramatically at the expense of management and safety. 2 Previously… Appendix 1 has a diagram from the Interop Report. The report was developed by a collaboration of the European Standards Organisations led by CEN/TC251 (principal author was Melvin Reynolds) as preparation for the Mandate 403 on ‘Interoperability’. The diagram shown was created back in 2009 but I believe it is still relevant to the on-going work on ‘interoperability’. In particular, I believe the top, right hand side rectangle, labelled 1, is particularly relevant to this work. As CEN/TC251 WG1 convenor, I initiated a technical report (TR) proposal to create an European Definition of a use case for eHealth, addressing what I perceived as being the most important, and tractable part of the whole process. I did some preliminary work as the initiating project lead, and the TR was approved by the member states. However, M403 did not go the way that CEN expected and no resource/funding was received, so the TR faltered. The original TR’s scope statement in 2010 was: “‘Use case’ methods are common place in the development of healthcare applications. However, although the methodology is mature, the actual practice is varied and, as there is no standard specific to the healthcare sector, the usage is problematic. This Technical Report will use existing taxonomies of ‘use cases’ to differentiate one type of ‘case’ from another. This report’s focus will be directed at the business level, and is concerned with ‘use cases’ intended to inform people of what is being done. The creation of this type of ‘use case’ precedes, and is different from, those which are aimed at a more technical level, intended to guide design and implementation. There are however overlaps in what is documented for each type, but beyond the taxonomic positioning, this latter type is explicitly out of scope. It is the purpose of this report to achieve the desired objective of clarity by constraining what is documented and, in the longer term, to facilitate reuse of ‘use cases’ for business requirements within healthcare. The immediate intention of this report is to present a generic framework that can gain wide consensus as to how to deploy the use case technique for high-level appraisal and reuse. The specific target was the ‘Interop’ application, as part of the M403 mandate initiative, but the broader ambition is to benefit the healthcare sector in general by providing a common way of presenting high level requirements.” In terms of the ‘Interop Model’ in Appendix 1, this piece of work enacts the missing feedback link between the [virtual] Best Practice Forum (5) to the Definition and Prioritisation of European Use Cases (1). The work to be carried out will also ensure that the template can be improved by other projects and relevant standards to make it relevant and to facilitate reuse. One goal will be to agree how such a template can be standardised to meet the interoperability needs of business and stakeholders in this domain. 9th July /2015 Page 2 of 11 Draft eHealth Use Case Definition Stephen Kay Building profiles (usually linked to technical use case) is recognised as requiring base standards to be in existence (Appendix 1 shows a graphical view of how they relate). Nevertheless to get a domain-wide, principled approach to interoperability, rather than an ad hoc or partial one, it requires a specific type of standard to provide a reference or ‘big picture’ to make such an approach coherent and effective for eHealth. This specific type of standard is also not shown or made explicit in the simplified diagram in Appendix 1, but it may comprise several instances. A single information model for interoperability will not work, even if it was possible to produce such a mammoth blue-print. Co-existing standards are what we have now, concurrent use, however, will need a different approach if interoperability is to be realised. 3 Rationale for producing a Use Case Template Our understanding of use cases has evolved a great deal since Ivar Jacobson first described them to the public in 1987. Hailed as a great breakthrough, they have nevertheless remained controversial in Software Engineering and most of the debate on usefulness has been to do with how formal or informal their representation should be. “Jacobson’s original usage scenarios were intentionally informal. He found out that people did — and still do — resist writing them whenever they become more formal.” [1] If too formal is too hard, the converse of too much informality results in confusion as to their purpose and use. Achieving the correct balance for any representation of a use case is probably impossible as it will remain a subjective notion and therefore a matter of personal preference. Achieving a consistent application, however, suggests that it would be better to err on more formality for the greater benefit but to make it as easy and relevant as possible to the would-be user so as to encourage use and minimise resistance. The idea of ‘templates’ for many similar tasks has long been recognised as being one solution and, unsurprisingly, many use case templates have been proposed/developed over the years. It makes sense because automated support is often provided via templates and if a use case is simply a prose description then it is amenable to basic tooling such as that found in most if not all of today’s word processors. It has the added bonus that no expensive or complex software is required to produce such a template-supported description and therefore there is no access problem… so why develop a new template? If ‘degree of formality’ has been a technical problem for uptake, then the problem of how general, or how domain specific a use case description must be, is a socio-technical one. Jacobson originally invented use cases to describe system architectures and system behaviour. This is very abstract and in such Software Engineering applications the objective is always to be generic so as to be operationalised as widely as possible. The formality question would necessarily be to the fore. However, the evolution of use cases has led to ‘business use cases’ being developed, and these even have a different notation, but more importantly, seek to operate in a different environment to engineers and computer scientists. It would seem that the internet boasts many different templates that claim to succeed, and do so precisely because they are tailored or targeted at a particular business domain, often with a different, yet focused audience in mind. The traditional use case, enhanced by business-oriented notation has not succeeded for many business users. The new notation applies to the model, and the graphical model comes with the connotations of 9th July /2015 Page 3 of 11 Draft eHealth Use Case Definition Stephen Kay being part of the technical UML approach. The stylised stick men do not scale for complex requirements. The need to link multiple drawings/models to satisfy a set of requirements is a daunting task and time consuming, whereas the trend is towards ‘agile’ solutions and the favouring of user stories to drive development, in a faster and more expressive way. User Stories, like the sketched Rich Pictures, can be very expressive for an individual communication need but at a potential cost of being too specialised and difficult to generalise, and rely too much on the skillset of the communicating author/artist. Such stories and pictures are not easily reusable. In addition to finding an appropriate way to express concepts in an audience friendly way, it is also a real challenge to identify a relevant set of common concepts for a particular domain. Why is it that health care system people dismiss ‘banking success’ stories as being irrelevant to eHealth? If the success story of business use-case templates is proven for specific domains, however, then why not develop a standardised one for eHealth that takes the context and content of the healthcare domain into consideration? To the author’s knowledge, it was true in 2009, and is still true even now, that no such agreed template for eHealth deployment exists or is being widely used. 3.1 The requirements for an eHealth Use Case Template The table following provides a starter set of requirements for such a template: Reference 1 2 3 4 5 6 7 8 9 Short Description Domain Specific Domain Wide Standards-based Policy friendly Sufficiently Formal Implement bridge Easily referenced Reusable Extensible Longer explanation of requirement Unique concepts and characteristics of/for eHealth Broad, coherent coverage for interoperability across eHealth domain Consensual and stable product, providing confidence in deployment Communicate business concerns and considerations effectively Enough to ensure consistency, concise enough to assist compliance Clear path from high-level of abstraction to fine-detailed, concrete use Ensure would-be users can find relevant use cases in a timely way Components to minimise fragmentation but allow local customisation Link to Agile methods and/or UML model approaches as required These initial 9 requirements will be used to critique the development and use of the templates used in the projects. One concrete result from this will be a proposal for an eHealth use case template for standardization. It will take into account the relation to the ISO TC215 WG1 work item on re-usable components. 4 Framework for the eHealth Use Case Definition There is a significant range of meanings for what a ‘use case’ actually means. Amongst other possibilities, it can mean anything from a general area of concern, to a simple example, to a fully specified technical model, spanning many pages. Given that new initiatives should both respect and prioritize the stakeholders’ representation of requirements, then it is a reasonable strategy to develop a framework which can facilitate understanding between the existing representations and thereby derive a more formal use case specification. The figure in Appendix 1 has the clear merit of emphasizing the primary role of stakeholders to make interoperability a feasible proposition. It suggests too that stakeholders will have their own individual 9th July /2015 Page 4 of 11 Draft eHealth Use Case Definition Stephen Kay versions of ‘use case specification’ at local, national and regional levels which is then somehow mapped to the European definition. This work acknowledges the reality of that situation. However, this work also emphasizes the need to actively reduce the gap or mismatch between the stakeholders’ independent versions and the standardization community’s desire to harmonise the use case definition. This will be done by providing a usable framework that defines a stakeholder-oriented eHealth use case template; one that can be accepted then productively used by both parties to facilitate interoperable solutions. The following points show why this approach is a desirable one: 1. to reduce miscommunication of requirements (introducing cost, error, and safety concerns), 2. to avoid the unnecessary and repetitive work in mapping case by case, and 3. to facilitate re-use of use cases for the benefit of multiple stakeholders. The requirements in 3.1 were derived for a generic use case template. However, most of the requirements can also be applied to the idea of a wider ‘Framework’ in which the use case template will be defined and subsequently deployed. 5 The Anatomy of the eHealth Use Case Template and Framework From a simple classification of the given requirements it is possible to describe the main elements of a business-friendly, domain-specific eHealth use case framework. The whole framework will provide the means to reference all the use cases developed to facilitate wide-spread re-use and to make links between similar use cases for comparison and assessment. Its focus is on the back-office management of the eHealth use cases in an accessible repository. It will be addressed later, but suffice to say it relates to requirements 6-9 in the above table. In this section the focus is on the ‘template’ to describe the proposed use case which attempts to be a fusion of a normal business case and a simplified business use case offered in UML. The five block anatomy diagram will be explained before a full example is given. Its primary audience is the policy/decision maker/manager rather than the technician. It embodies the requirements 1-5 in the above table, and also 9 as it can be extensible to new emerging requirements. 5.1 Summary Each use case will have a name. In addition there will be a ‘user story’ (ala Agile Development), which is a requirement formulated in everyday business language. Together these two make up the Use Case Summary. The Use Case Summary provides the short-version public handle of the use case described by the template. 5.2 Goal As with formal Business Cases, a goal is paramount to a Use Case’s success. What is intended to be achieved is 9th July /2015 Page 5 of 11 Draft eHealth Use Case Definition Stephen Kay fundamental and must be stated clearly, upfront. As Cockburn says, “Linking use cases to actors’ goals is so very significant because it shifts the writer’s attention away from the function lists that most programmers worry about and puts it back on the users: What the users really want to accomplish with the software are their goals when using the software. If the software supports those goals, the software will yield the greatest business value.” Without a ‘goal’ the use case is useless, with a Goal present, a major criterion is automatically provided to measure success or failure. Goal driven behaviours are also part of our domain business and complement the other parts of the Anatomy related to the business model. 5.3 Participants This is the easiest, yet hardest block in the anatomy to get agreement on (or at least the most controversial if the various trials are to be considered.) Everyone agrees that ‘PEOPLE’ are at the heart of use cases, one way or another; the ‘stick people’ on the use case diagrams have made this seemingly obvious. ‘Participant’ is perhaps the most neutral term (suggested in the Antelope project). However ‘Actor’ and ‘User’ are the original and usual modelling terms for use case characters. There is also the matter of the ‘organisation’, the ‘device’ and the ‘system’ and how these stakeholders are managed. It will be important to make it clear in practice, certainly, but at least for the anatomy block, ‘participants’ defers the discussion! Participants (of various sorts) occur in several places within the proposed template. 5.4 Domain Sensitized Business Model This is the unique part of the Template, differentiating it from other domain-specific templates, and satisfies requirements 1, 2, 4 and 5. The need to be both domain-specific (1) and domain-wide (2) points to the specialised concerns of Health Informatics. By way of brief background, alongside the traditional use case, there is also now the more recent ‘misuse case’ which specifically describes the threats to the system, either deliberate or accidental, as well as introducing the ‘misuser’ who initiates the action or causes the problem. These ideas have also been woven into this proposed template. Personal (or patient) safety is a core principal for Health Informatics, and some authorities devise ‘safety cases’ as part of the system specification. Security, privacy and governance are also core. It is appropriate that any business model used in this template should explicitly represent or document these concerns when relevant to do so. Domain-wide reflects the use of the WHO definition of ‘Health’ in Health Informatics; the template must therefore be appropriate and capable of being used for any of the domain systems and must be able to refer to an individual or a cohort as well as to an organisation, be it local, national, regional or global. The template must be able to communicate business concerns and considerations effectively (4) but also needs to permit decision-makers to express their ideas in a formal way, sufficient enough to ensure consistency, and concise enough to assist compliance (5). This is also about usability; UML business use cases do not seem to be popular, partly because the extended notation is not widely adopted. We know too that more formal representations seem to turn off would-be users. Business use cases should be independent of technology, concentrating on events and processes rather than products; they should be in the language commonly used and should have familiar concepts. In an attempt to encourage policy 9th July /2015 Page 6 of 11 Draft eHealth Use Case Definition Stephen Kay makers and the like to use the template, it was decided to incorporate a business model within the template to bridge from their ‘world’ to a more formal one. [Note 1: The Antelope project adapted the original template and specified another lower level of use case description in order to map specifications and standards for use; this may be useful for describing the content of the ‘bundle’. Note 2: Trillium Bridge also adapted the original template, and work will be on-going to synthesize the best practice.] In theory any well-known/well used business model would do. The candidates are numerous, but I went with the SWOT model, which is perhaps the most, widely used and earlier attempt to understand the business concerns in a semi-formal way. The Strengths & Opportunities fit well with the so-called happy day scenarios in Use Cases, whereas the Weakness & Threat quadrants can satisfy the concepts from the misuse case scenarios. The ‘pre and post conditions’ of the UML models are also accommodated and the Internal/External concepts are relaxed to suit the domain wide requirements of granularity. 5.5 Key Events The more process-oriented aspect of the business use case fits well with the importance of events. There is the initiating or ‘trigger’ or ‘starting’ event, as well as the event sequences recorded in the different type of scenarios. It is common to have a primary or ‘happy day’ scenario where everything proceeds as intended, and one or more adverse event scenarios that show contingencies etc. This can be made as extensive and as elaborate as required and various modelling diagram methods (UML again) can be introduced to represent the various paths, timing etc. My suggestion is that the trigger event is highlighted and the happy day scenario are the only events presented within the template; the other material whilst obviously important is likely to be too much detail and therefore too much effort for those prepared to use the template. A further suggestion is that this other material is part of the extensible Use Case Framework and it can be used as reference material. The same is true with all the diagram notation/ rich pictures and the like. The use case template described here should be common-use, easily understood prose, ideally limited to 1 to 3 pages at the most. A template could be set up in a word processor, with no esoteric software for models required; this means that the template can be made readily available and accessible to all in the community. The following gives two illustrations of the template. The first is the original template destined for the CEN report; it provides the outline and description of the content as I saw it then back in 2009. The second is one that is filled in as part of an education guide intended for commissioners of interoperability in the UK. This was done in the last part of 2014. I intend to analyse the uses made and the adaptions made by the Trillium Bridge and the Antelope projects, before providing an update and revision for the CEN report. This is on-going. This current document will also be passed to Gary Dickinson who is doing some complementary work in ISO. Hopefully, this might be of use here and if JIC buys into it then requirement 3 will have been satisfied and we will all know what we mean by ‘use case’ in our domain. 9th July /2015 Page 7 of 11 Draft eHealth Use Case Definition Stephen Kay 6 The Original Template with descriptive rules The grey background refers to the proposed Use Case Framework, which needs to reference each use case for re-use purposes. The greyed out parts at the bottom, are optional ways of extending the description with more detail. The suggestion is that these too are part of the Framework, not the template. 2 9th July /2015 Page 8 of 11 Draft eHealth Use Case Definition Stephen Kay 7 The Latest modified version (sections labelled), and filled in This is an example created for a clinical commissioning group in the UK. Some details have been omitted, e.g. some steps in the scenarios; CCG is the acronym for a clinical commissioning group. Sections Use Case Description #2 A. Name Rebalance non-acute services B. Stakeholder story CCGs are facing a deficit because of an excess of inpatient admissions due to demands from changing demographics and lifestyle. The CCG and with the Health and Wellness Board the CCG determines with its main providers and social services that attention needs to be directed towards re-balancing support in the community. However, resources are below budgeted manpower levels and non-acute based services have not worked in a coordinated manner in part due to geography and use of limited IT administrative systems. C. Starting event The movement to establish integrated care budgets and pioneers combined with the CCG’s own financial challenges has encouraged the CCG to put in place a strategy that will re-balance the scales in favour of greater community based interventions. D. Actor and users The CCG and the acute, mental health and community trusts, all GP practices and social services. E. Goal To re-organise community and home based services in conjunction with social services such that the use of existing and planned nursing home visitor resources can be maximised combined with a more effective coordination between community and mental health services. Additionally, the goal is to provide improved access to local GPs / community locality based minor injuries services so that pressure on A&E is reduced. F. Stakeholders In addition to D above, patients themselves will be directly affected by less hospital visits and improvements to accessing the appropriate healthcare professionals. G. Primary scenario 1. 2. 3. 4. Develop a multi-faceted approach that combines new initiatives in care access and pathway management that can be supported by a new technology infrastructure to underpin the sharing of information across all parties. Produce an inventory of the separate systems across all of the organisations concerned and assess the extent to which they can share patient information. Particular attention should focus on silos & duplicates of service user information. Extend the use of existing facilities by agreeing appropriate staffing regimes, increasing visibility and access to ‘minor injury’ services across the population. … 5. Provide a common view of client information across all organisations. 9th July /2015 Page 9 of 11 Draft eHealth Use Case Definition H. Strengths 1. 2. 3. I. Weaknesses 1. 2. 3. 4. J. Opportunities & safety context: 1. K. Threats & security context: 1. 2. 3. 2. 3. L. Assumptions, Additional details and context 9th July /2015 Stephen Kay Some systems already in place Vision to adopt a specific strategy aimed at reducing the pressure on acute services the CCG plans to create a unified working environment across the locality strongly supported by interconnected information architecture. Commitment to address culture change and provide more accessible facilities especially for minor conditions. Legacy systems in place; not a ‘green field’ site. Patients need to be convinced that quality care can be provided in the community High risk women with post natal depression and many younger to middle aged patients with a range of non-critical problems are being handled by acute services. When (3) do receive support from community, mental health and social services, they are often visited by three care workers when one appropriately skilled could provide the same level of support. Improved co-ordination of services with all parties able to access a consolidated history and status for each patient. … The potential exists for other agencies to become involved e.g. the police and education establishments in the future. No information standards yet established so that common terms could be used with social services required to identify service users by NHS number. … A growing movement to integrated health and social care as per the change in budgets £3.8 billion away from NHS to a combined NHS/social care programme Assumptions: None. Additional Details & Context: To achieve the goal will require a co-ordinated series of activities. One possible way of supporting this is by the creation of an integrated digital care record (IDCR) to provide, as a minimum, all patient continuity of care data for review. This is not necessarily a single database, but it may be a federated source of data. With an established, comprehensive and shared data source the various health agencies can create a more unified approach to those patients making greatest demands on care services through a combination of technology deployment and pro-active management of the patients. The health economy with which the CCG healthcare organisation does most contracting comprises a large number of GP practices each with EHR systems but not all of which are from the same supplier and operate largely independent of each other. Page 10 of 11 Draft eHealth Use Case Definition Stephen Kay Appendix 1: Figure from the Interop Report for M403 Stakeholders: contribute to European eHealth Interoperability activities and includes (e.g.) - National, Regional, or Local projects - Health Professionals - Citizens - Industry European eHealth Interoperability Activities: a European eHealth Interoperability Coordination Committee empowers and accredits the numbered activities below (1-5) Local, Regional or National eHealth Project Use Case Specification 1. European Use Case Definition and Prioritisation Portfolio of Base Standards Prioritised Use Cases 2. Base Standards Development and Maintenance Feedback System Implementation Interoperability Specifications Portfolio of Recognised Profiles Profiles Test Plans and Tools 3. Profile Development and Maintenance 4. Profile Quality Assurance Feedback Feedback Feedback Deployment of Live Systems 9th July /2015 Feedback 5. Best Practice Forum Page 11 of 11