Safe and Sound is a collaboration between three leading research groups with a common interest in the delivery of healthcare services, but with very different backgrounds and research experience. A key task at the beginning of the project was to establish a consensus on the motivation, rationale and success criteria of the project. This consensus is summarised here.
Project motivation and opportunity
1. Medical error is a significant cause of suboptimal care and avoidable patient harm. “Errors” include individual mistakes in clinical decision-making and lack of critical knowledge but also include miscommunications and organization failures.
2. ICT has a major contribution to make in reducing medical error and mitigating its consequences. Clinical decision support (CDS) and Clinical workflow services (CWF) at the point of care have significant benefits for improving service quality and patient safety.
3. First generation CDSs (e.g. alerts and reminders, prescribing aids) have been clearly shown to improve practice. The evidence for more sophisticated CDSs (e.g. diagnosis, treatment selection) and CWF management (e.g. care pathways, care planning) is less extensive, but comparable or greater benefits are possible.
The technical challenge (“grand challenge”)
4. The challenge identified in this project concerns the need for an application development platform which will make it practical to routinely engineer (design, implement, test, deploy and maintain) ICT services for supporting effective and safe decision making and patient care which have a high probability of clinical success and practical adoption.
5. A key subsidiary challenge, on which all such services will depend, is the need to formalise, disseminate and share medical and other knowledge that is relevant to an individual’s care, in a standardized, scalable and interoperable form.
Technology assessment and project success criteria
6. Technology assessment (TA) must be multi-dimensional, covering clinical, technical, organizational, ergonomic (“usability”), health-economic, social policy and other criteria.
7. A key deliverable is an instrument (e.g. a structured checklist) for assessing CDS and CWF technologies against key TA criteria. The instrument should inform service design prospectively (to determine whether a proposed application is likely to be successful with respect to the criteria) and retrospectively (to assess whether an implemented application satisfies the criteria in practice).
Client server model
8. CDS and CWF services can be delivered using a variant of the client-server approach, in which clinicians, patients and other users have access to role-specific views and services layered on top of a centralized patient record and quality-managed knowledge base.
9. Despite its relative tractability the centralized model still has significant issues as a model for delivering useful and maintainable services in realistic clinical settings, including compliance with standards, service compatibility and interoperability.
An advanced platform
10. The client-server model is a practical stepping stone to a more advanced approach to delivering CDS and CWF services which is more flexible and robust, and more appealing to clinical professionals. Features of this model are
a. Applications will support care pathways in which clinical services are physically distributed and used by clinicians with different roles and skills.
b. Services are conceived in terms of clinical goals, roles, responsibilities, and professional commitments, based on an explicit medical rationale, rather than preprogrammed tasks and business processes.
c. A multi-agent, open knowledge perspective in which service agents have specialist roles and different specialist knowledge is a promising platform for demonstrating this model.
Knowledge representation and knowledge service delivery
11. Safety, ethical and other requirements will dictate that in the short term medical knowledge bases will be created and maintained on a centralized architecture, such as a client-server model.
12. In the longer term the centralist model of content development will give way to a more flexible but also more complex architecture in which providers can compete to offer content and services, on both "open source” and proprietary bases.
13. A realistic solution to the grand challenge will need to offer both the simplicity and safety of a centralized model where this is required, while being able to evolve into a scalable, open knowledge model as the content market develops.
