The final project report is due in September 2009. This interim report summarises progress to date and was posted in June 2009. More details about the final report can be found here
Interim report executive summary
Knowledge is the enemy of disease; the application of what we know will have a bigger impact than any drug or technology likely to be introduced in the next decade. Sir Muir Gray, 2008
Many science-based fields are facing a “knowledge crisis”, in that knowledge is expanding explosively while economic resources and the human capacity to apply it remain finite. Medicine is perhaps the most prominent example; the field is the focus of a vast and productive research effort, but new tools and techniques are not always either quickly disseminated or effectively used. In the last thirty years, clinicians and computer scientists have developed technologies that could hold the key to managing that ever-growing complexity. Research on computerised Clinical Decision Support (CDS), Workflow (CWF) and Knowledge Management (CKM) systems has blossomed into a global movement, with applications in most areas of medicine.
Given the right push, these technologies will be ready for deployment across the National Health Service within the next 5 years; this report presents out vision of how that goal might be achieved. The Safe and Sound final report to be published on this site in September 2009 will present in detail our vision of: the motivations for this push, the challenges to be faced, some technical options for achieving the objectives, and a road map for making the transition.
Chapter 1: Motivation and opportunity
The pace of medical research has delivered treatments that might have awed an earlier age, but the price of that power is service complexity, which can tax even the most dedicated and capable clinician; medicine is becoming "a humanly impossible task". Although healthcare brings enormous benefits to us all, there is compelling evidence that avoidable errors are common and patients are frequently harmed. CDS systems apply formalised (machine-interpretable) medical knowledge to patient data in order to arrive at patient-specific treatment recommendations in a way that ensures best practice. CWF systems can ensure that clinical decisions and other tasks are made and followed up in a timely way.
The evidence that these technologies can have the desired effects (improving clinician performance, patient outcomes, and the cost-effectiveness of health centres, among others) is substantial, but at present no one knows how to deploy these systems in a safe, sound, and scalable way. Furthermore, the use of CDS technology is not widespread and the necessity for developing it is not widely understood. Many reasons have been advanced but in our motivation discussion we will focus on two of the most persuasive arguments: the evidence that healthcare is not safe and the challenge of coping with the knowledge explosion in healthcare.
Chapter 2: The technical landscape
Much has been written about CDS systems in the last four decades. Recent systematic reviews have shown many to be effective at improving patients' health outcomes and increasing compliance with best practice. Despite the growing evidence CDS systems are not yet widely deployed. The challenge identified in this project is to develop an application development platform which will make it practical to routinely engineer (design, implement, test, deploy, maintain and in due course decommission) ICT services for supporting effective and safe decision making which have a high probability of clinical success and practical adoption.
We are not trying to achieve this in an empty landscape. While the clinical takeup of CDS systems is still slow development and deployment of ICT in healthcare generally is moving rapidly, in the establishment of infrastructures on which other services will depend (e.g. standard terminologies, electronic patient records, PACS, messaging, order entry systems etc). Major commercial systems are appearing in the UK which offer many such functions, though to date experience using them is mixed. The review of current trends reported in this chapter shows that these developments represent both major opportunities and constraints on the deployment of CDS and CKM services. In the full report we set out and give the rationale for a modular solution, on which advanced services including decision support, workflow and other knowledge based services are "stacked".
A major challenge is how to integrate and deliver support for a variety of such services at the point of care, in the complex, unpredictable and ever-changing context of healthcare delivery, and how to do this in a flexible yet scalable and safe way. One way of achieving integration is to manage all clinical, patient and research services with a supervisory control system which forms the top layer of the stack. Safe and Sound has demonstrated two complementary approaches to the service management problem, which are described in the next two chapters. One is based on a relatively simple, centrally managed but relatively inflexible approach; the other is a more experimental but potentially much more flexible peer to peer architecture, which is scalable and open to inclusion of services offered by 3rd party providers.
Chapter 3: Centralised service architecture ("orchestration")
Our baseline approach is the "orchestration” model, in which all services are under the sole control of a single provider such as an NHS agency in the UK (e.g. NICE or the Do Once and Share programme), an individual hospital, or a commercial healthcare provider. On the centralised approach decision support and workflow services are integrated with patient record, order entry and other systems and deployed for all categories of user on a central server. All clinical decisions, care pathways, research trials etc are coordinated through a central patient record. The orchestration prototype developed in Safe and Sound demonstrates a range of decision support and other services for surgeons, pathologists and radiologists who are looking after a patient with suspected breast cancer.
The prototype was developed using Oxford University's PROforma technology and the OpenClinical.net web server. Click here for demo after September 30th 2009
Chapter 4: Open service architecture ("choreography")
The second approach developed by Safe and Sound facilitates run-time identification and use of 3rd party services as well as those “core” services which must be centrally managed by the NHS or other healthcare provider. A core breast cancer service for example might be based on a NICE guideline or Do Once and Share care pathway which is supported on a national or regional basis. However clinicians who use the core services can also make use of non-core 3rd party services such as Adjuvant Online!, services offered by a research charity, or a patient support network. A service choreography (noun) is a protocol which defines when and how services from 3rd party providers can be accessed safely from within the core.
The choreography prototype we have developed demonstrates a more flexible implementation of the services provided by the orchestration prototype. Here the services for surgeons, pathologists, radiologists etc are dynamically located and contracted by the automated service manager. Access to accredited third party services and Direct-to-patient services can be choreographed in a similar manner.
The Safe and Sound choreography prototype uses PROforma to model specific decisions and care pathways but the peer network is implemented using Edinburgh University’s interaction language LCC (Lightweight Coordination Calculus) and the Open Knowledge software kernel. Click here for demo after September 30th 2009
Chapter 5: Knowledge sharing and service provision in the digital economy
The Safe and Sound team takes the view that societal pressures and growing technical capabilities are driving us towards an increasingly open ICT service model for healthcare. This has received independent support from an influential international source, the World Economic Forum which has sponsored a project to anticipate future trends in the provision of integrated IT, Telecom, Media and Entertainment services. Forum discussions with many commercial and government organisations have led to the concept of future “digital ecosystems” which consist of the users, companies and governments who combine to deliver and consume IT services in a particular sector.
Although healthcare is only mentioned in passing this sector seems to be an area where the analysis seems highly relevant to the future evolution of knowledge sharing and knowledge based services. The WEF’s digital ecosystem concept distinguishes three scenarios for the future development of ICT services (http://decommunity.net/).
(1) The first scenario is "vertical integration of services by centralised providers supported by governments which tend to be closed to external providers". The centralised integration of NHS healthcare services, using NICE clinical guidelines, running on a BT patient record system and communication platform could be said to be an instance of this scenario in healthcare.
(2) At the other extreme “grass roots communities” create many knowledge-based services and offer these in an open marketplace. We already see the emergence of such an open “ecology” in personalised healthcare services in which the countless web sites offering patient information, patient networking and support, and some primitive decision support capabilities are a precursor to this. In the WEF scenario personal digital agents will find services according to individual user requirements and preferences, though service access by means of such agents are not yet visible in the healthcare sector.
(3) In the third, middle ground scenario “service aggregators” compete to offer services in relatively specialised but large market sectors, e.g. cancer, diabetes or the catchment of a large NHS Trust). According to the WEF concept this is only likely to be practical if deployment infrastructures support openness and service interoperability, within a standards-based and regulated environment.
Our vision of the medium and long-term trends that will impact the provision of IT services in healthcare is consistent with the WEF’s view of the digital ecosystem. Safe and Sound has demonstrated the combined power of PROforma and LCC for formalising clinical decisions and processes, and delivering practical services on an integrated management platform using OpenClinical and Open Knowledge. The result is a practical mechanism for supporting a range of digital ecosystems such as those anticipated in the WEF discussions.
Chapter 6: Technology assessment and success criteria
Clinical decision support and workflow systems are well within our reach. Given the urgent need that they can address and the evidence of their clinical value these technologies should be a high priority for deployment. Although many technical challenges have been met, however, CDS and CWF have arguably not yet developed to the point where they are routinely used and acceptable in everyday clinical settings. Embedding these services in clinical processes requires further development so that they mirror and support the complex and distributed nature of clinical work itself. However, there is also a host of wider psychological, ergonomic, organisational and political factors that determine whether a technology that works in the laboratory will ultimately find a place in the clinic. Consequently we need to understand the range of criteria on which CDS systems will be judged and develop an approach to evaluation that addresses these criteria.
The ICT services we discuss in this report are complex interventions; they can have many interacting components, require extensive tayloring for a local environment, and have multi-dimensional impacts on multiple groups at multiple levels in healthcare organisations. That complexity makes it difficult to understand (or predict) why some clinical services are (or will be) more successful than others. Our goal must be to develop an assessment tool that considers the wider impact and utility of the technology for the many groups who may encounter it and helps designers understand what has been or will be clinically acceptable and successful.
An enormous variety of evaluation frameworks have been proposed for the evaluation of ICT systems in recent years – but though different in detail, these frameworks have much common ground. Given this emerging consensus, the Imperial Healthcare team in Safe and Sound has developed a pragmatic framework which associates the assessment methodology directly with known CDS success factors and different stages of the Software Development Life Cycle.
Chapter 7: Road map
The overall conclusion of this report is that we recommend that the development and delivery of future CDS, CWF and knowledge management services should be implemented in at least two waves - a low-risk approach we call "Safe, Sound and Secure" based on the orchestration approach, and a more flexible but relatively unproven approach we call "Safe, Sound and Open" which is based on choreography techniques. This road map is summarised here - full details will be provided in the final report.
Safe sound and Secure (1) Develop clinical & patient services using a centralised content management model; (2) Engineer an orchestration prototype to support application trials based on a conventional client server framework; (3) Deploy and trial applications in appropriate (e.g. NHS) hospitals and (4) support evaluation by independent clinical teams.
Safe sound and Open (1) Develop advanced peer to peer platform which dynamically choreographs CDS and CWF services (2) open up access to services offered by 3rd party providers; (3); Deploy applications in safety and quality trials and (4) support evaluation by independent clinical teams;
Clinical evaluators should report to an appropriate group of stakeholders independent of the technology developers (in the UK these could include Connecting for Health, NICE, NIHR, EPSRC, MRC and other research sponsors, other research groups and representatives of commercial companies in relevant sectors).
Review and report. The benefits of the orchestration and choreography approaches should be compared on comparable applications, and assessed by the stakeholder group on service quality, safety, scalability etc. The goal should be to arrive at recommendations to government and commercial healthcare providers. A key outcome of the work would be a pre-standard for developing, deploying and integrating decision support, workflow and knowledge management services in a routine way which promotes the growth of the digital economy in healthcare.
