Towards an Integrated Platform for Improving Hospital Risk Management

2009 Sixth International Conference on Information Technology: New Generations Towards an Integrated Platform for Improving Hospital Risk Management Jawed Siddiqi, Babak Akhgar, Fazilatur Rahman and Sally Atkinson Informatics Group, Sheffield Hallam University, UK Alberto Savoldelli, Stefano Arici and Paolo Bertele' Consorzio per l'innovazione e la Gestione delle Imprese e della Pubblica Amministrazione, Italy Brian James and Mike Pinkerton The Rotherham Foundation Trust, UK internal and external audits. Examples of national information sources available to hospitals include Institute for Health Care Improvement (IHI) US, National Patient Safety Agency (NPSA) UK which among other tasks has the role of acquiring, contextualizing, investigating, reporting and disseminating risks reported by hospitals. This means that events that may be relatively insignificant in risk or frequency or negative outcome at local level can be brought together at national level and true population judgments made regarding their importance in terms of warning providers, intervening with manufacturers and influencing national policy, particularly so in the UK. In many cases, these data are collected after the occurrence of an adverse event. Due to the complexity and numerous actors and variables involved in a clinical process, it is difficult to determine the root cause retrospectively. This situation has been confirmed by preliminary results of the socio-economic framework survey that was conducted in several European hospitals during the initial stages of the REMINE project: analysis of existing risk management strategies shows a lack of proactive phases (i.e. risk identification and evaluation), while reactive phases are quite developed and diffused (i.e. adverse event reporting and analysis) [6]. As with the vast majority of large industrial organizations, it is understood that acquisition and interpretation of information in hospitals is fundamental to evaluating risk and building new processes to manage, reduce, or avoid risk. However, the dynamic hospital environment and demanding practical responsibilities of staff often leads to inconsistencies in reporting. Furthermore, collection of data may also be dependent upon compliance of staff to report near misses. The subsequent lack or gaps in availability of data can make it extremely difficult for risk managers to build new recommendations for patient care. Abstract REMINE Project aims to develop an integrated system to manage risk associated with patient safety. The system is a computer based platform that facilitates integration of diverse multimedia data, providing an interface between clinical activity and risk management. The project objective is to develop a framework whereby data is extracted from a heterogeneous source. This framework is designed to make it possible to efficiently handle risk management data and to trial, validate and evaluate new risk management processes to improve on patient safety. 1. Introduction Improvement of patient safety is a cornerstone to the healthcare agenda in the UK, where it is currently estimated that 1 in 10 patients admitted to hospital will fall victim to error during their stay. In UK NHS hospitals £2 billion a year in is spent in extended hospital stays alone as a result of adverse events involving patients. Current studies predict that 50% of these adverse events could be avoided if risk management strategies had been devised and implemented through evaluation of similar previous incidents [1]. Patient safety is a critical issue in other countries also: in the USA it has been estimated that between 44,000 and 98,000 patients die every year as a result of medical errors [2], in Australia 16.6 percent of patients admitted to hospitals experience adverse events [3], in New Zealand the figure is 12.9 [4] percent and in Italy it is 4 percent [5]. Risk management in hospitals is certainly integral to good clinical management and practice. Currently, the information resources available for developing risk management strategies largely consist of incident reporting schemes; focused clinical area studies; complaints and litigations data; patient statistics; and 978-0-7695-3596-8/09 $25.00 © 2009 IEEE DOI 10.1109/ITNG.2009.204 1079 Healthcare risk management would benefit substantially from the development of comprehensive information repositories integrating data from a heterogeneous source. In order to achieve the highest quality of risk management, it is clear that data should be collected prospectively. Data acquisition should be automated and it should cover both horizontal and vertical viewpoints of clinical pathways. The REMINE Project entitling high performance prediction, detection and monitoring for patient safety and risk management was conceived out of this understanding. REMINE is an ICT-FP7 Project concerned with the development of a risk management platform targeted toward managing risk against patient safety (RAPS). The platform design encompasses data description, collection, analysis and provision of response based on relevant RAPS and currently available information resources. This definition of framework architecture, enables prospective collection and analysis of relevant data, is anticipated to provide an important turning point in healthcare risk management. This paper describes a study employed in defining existing risk management procedures in relation to the patient pathway. According to Wilkins and Holley [7], a risk management system must possess five distinct components such as establishing the context, identifying the risks, analysis and assessment of the risks, treatment of the risks, and finally monitoring and reviewing it throughout the process of risk management. In the case of the REMINE Project, the organizational context is based on management of clinical processes within the hospital. In the study presented here, clinical pathways mapping and socio economic framework surveys have been used to identify risks and to provide understanding of the key stakeholder needs and objectives. This methodology not only helps to provide understanding of the organization and its main requirements, it also generates a detailed and logical account of the strategies important to ensuring patient safety. With this information, it is possible to determine how the REMINE system should facilitate the policing of risk management strategies and generate information that may be used to enhance risk management overall. • • • • • This list, although not exhaustive, demonstrates to some extent how the system design must incorporate many stakeholders, as well being integrated with a number of software, hardware and other inventory devices. Figures 1-3 provide examples of some of the information collected in the REMINE platform development study. 2.1. Exploring a REMINE Use Case The pathway illustrated in figure 1 shows the high level mapping of the clinical pathway associated with the medical admission ward. This is a clinical area selected as a focus for one of the four hospital pilot sites involved in the REMINE Project. The medical admissions ward is a key entry point to the hospital: patients enter this ward either as a direct referral from a GP or from the accident and emergency room. The pathway represents the stages which the patient normally goes through from the point of admission to the ward, through to their discharge or transfer to a more specialist clinical area. From this horizontal view it seems that the patient pathway is quite simple. However, by moving to a more detailed view of the admissions process, the patient pathway is shown to be composed of a series of events, involving a number of stakeholders and hospital resources. Figure 1: A basic Clinical Pathway for a hospital medical admissions ward 2. Identify the key elements of a Hospital Risk Management System As each step of the admission process is examined in more detailed, the full route that a patient takes becomes increasingly complex. Figure 2 shows the outline processes related to just the first stage (patient admission) of the patient pathway that is illustrated in figure 1. There are many considerations to be taken into account in defining REMINE: one of these is the hospital environment with which the platform will interact. Main stakeholders and other important concerns are: • • Professionals allied to medicine – physiotherapist, speech and language therapist, radiographers and pharmacists etc. Technicians Administrators Patients Infrastructure – Software, Hardware and Consumables Doctors Nurses 1080 Figure 2: Process map for admitting a patient to the hospital Figure 3: Infection control risk management procedure during patient admission The process of admitting a patient to the medical admissions ward of the hospital is shown, with events associated with risk management protocols highlighted (RM). This process not only involves the patient and administrative staff, it also involves hardware, software, and a database of existing patients. There are also risk management strategies in place for administrative staff to identify patients and to assess risk management content within the patient file. By examining the risk management procedure associated with retrieving a patient file (figure 3), it is apparent that more stakeholders may be introduced into the admissions process as a result of suspected infection. In the event that a patient's file contains information to indicate that he has had previous high-risk infections, the administrative staff informs the infection control nurse to carry out a preliminary infection assessment. This must occur prior to sending the patient to the waiting room, or to the main ward to see a consultant. If the patient is not considered a risk to other patients, then the normal admission process will continue. If it seems there is a risk to other patients, then the new patient will be put into isolation before continuing on the patient pathway. When a patient file is loaded, administrative staff will be alerted if the patient has previously been admitted to the hospital with MRSA or other such high-risk infections. This results in a risk management procedure being implemented to avoid transmission of infection to other patients. In comparison to the processes in many of the more specialized clinical areas, the patient admission process in the medical admissions ward patient pathway is a very simple process. However, from these examples it is clear that even a simple process such as this is dependent upon co-ordination of a number of key factors. If one of these steps is missed or carried out incorrectly, there may be adverse consequences to the patient, or other patients in the ward. As we begin to look at more complex processes, such as the ordering of blood tests or prescriptions, there is the additional complexity of involving the laboratory and pharmacy departments, both of which have their own IT systems, processes and risk management procedures. A key factor in the successful application of the REMINE platform is the ability to predict or provide alerts about a potential adverse event. There are often numerous risk management strategies in place to prevent or reduce the occurrence of a particular adverse event. For example, in infection control there are: • • • Alerts for new admissions with previous records of high-risk infections; Traffic light system to assess level of risk to each patient; Patient bed mapping to avoid spread of infection Failure to properly implement one or more of these risk management procedures could result in an outbreak of infection within the hospital. In the event that outbreaks increase above a set threshold, REMINE data 1081 mining functionalities could identify an eventual flaw in the ‘as-‘is’ risk management system. As a result, the system could provide the risk manager with patterns of data that show strong correlations with the risk indicators. For instance, it could be that many of the infected patients who seem to be the sources of the outbreaks are from different regional areas: the hospital doesn’t have a previous record for these patients and so there was no information about previous infections to register an alert. On the basis of this, the risk manager may redesign the admission procedure by making all patients admitted to the hospital without previous medical records being treated automatically as a high risk case for infection control. From examples such as these numerous use cases have been developed, analyzed and synthesized to propose an architecture, the details of which are outlined in the next section. 3.3. Risk Management Process Support RAPS management support is linked to the RAPS data capture and data processing via a semantic interface. This component includes the RAPS taxonomy and RAPS ontology. Both are integral to the RAPS management support, which provides users with the ability to build and assess new scenarios for risk management. This component allows the REMINE platform to be used as a decision support tool for developing new 'to be' processes to reduce or prevent risk. 3.4. Info broker patient safety framework The “info broker patient safety framework” is connected with the Hospital Information systems. It is implemented for integration, data mining and data assessment purposes. It is also fundamental to supporting distribution of results, alerts, and other system response. The underlying ontological structure, which determines response, is designed to evolve based on the measured effectiveness of the suggested risk management procedure as applied to the identified RAPS. 3. REMINE Architecture As a first step toward the development of an integrated platform for hospital risk management framework, REMINE architecture has been developed. Each part of the Risk Management process is selfexplanatory from its name. The most important part of the process is the monitoring and review, which must be continuous and should feed back into each of the other processes. The REMINE platform employs a semantic approach that allows fast, secure extraction of data and correlation of information across several domains. The logical schema of the REMINE architecture design is composed of the following components: • • • • 4. Novelty of the proposed architecture Much work has been conducted in the European research community in the field of eHealth to provide Health Care professionals with advanced tools facilitating the treatment of patients, the identification of diseases and the overall management of the process. These works have often led to the creation of tools and procedures that are able to improve the management of all the knowledge that is involved in hospital assistance. 4.1. Related Projects Data Capture and RAPS alerting; RAPS data process integration; Risk Management Process Support; InfoBroker Patient Safety framework. ARTEMIS project developed a semantic web services based interoperability framework for medical information systems. Doc@Hand aimed to create a set of IT tools in order to reduce the time and associated costs to collect the information and in making the best use of it for a more informed decision making (diagnoses, therapies, protocols). PALLIANET created knowledge driven collaborative practices in order to minimize risks in the context of palliative care by providing advanced solutions that support the needs of city professionals by combining a Community and Knowledge Management service – through advanced human-computer interaction features that makes access for caregivers easy and natural – and Coordination Facilities. PIPS improves current HealthCare (HC) delivery models while creating possibilities for HC professionals to get access to relevant-updated medical knowledge and the European citizens to choose healthier lifestyles. TACIT supports the entire clinical process across primary and secondary care by specifying prototype and test Ambient Knowledge 3.1. Data Capture and RAPS alerting The elements responsible for capturing digital information from the hospital environment provide input data to enable formation of the 'as-is' scenario within the REMINE platform. Data collection is in co-ordination with clinical guidelines, hospital staff procedures, and standardized clinical and legal processes. 3.2. RAPS Data Process integration Integration of RAPS data from the heterogeneous source is performed via transformation and filtering to generate a common base format for the information repository. Knowledge extraction and data mining are the core functionalities to enable provision of the RAPS meta-database. 1082 Elicitation integrated with an Expertise Browser and expert locator. stakeholders with easy interaction capability for RAPS reductions. 4.2. Major REMINE Innovations 5. Conclusion In relation to the above described projects, REMINE introduces several advances. Firstly, REMINE deals with the modeling of all procedures and models relevant to security, designing a comprehensive approach to adverse events prevention. It deals with the acquisition of data to be analyzed and matched with the above mentioned procedures. This match allows the identification of risks associated to the different situations that health care professionals face in their everyday working life. Professionals receive alerts about arising risk and can further explore the issue by simulating the impact of different choices the may make, finding out the best possible solution. This process is iterative and leads to a continuous improvement of procedures. Current approaches against RAPS early identification and effective prevention suffer from two major problems: lack of RAPS information at the right time in the right place, and absence of standardized procedures easily accessible and usable. Proposed REMINE architecture aims to promote a detailed analysis, an early identification and an effective prevention on RAPS when there are significant mass of inhomogeneous data sources, stored in multimedia databases, and distributed environments with different care professionals contemporarily involved. RAPS management has been modeled in the project as a process made up of three distinct steps such as risk identification, risk analysis and alert provisioning. Support for advanced business process modeling is another requirement for RAPS management process. REMINE a framework architecture aims to bring significant improvement and create innovations in different step of the RAPS management process. Risk identification step of a RAPS management process will be improved through integrated multimedia data acquisition (including RFID) as well as data enrichment through semantically developed metadatabase, RAPS taxonomy and RAPS ontology. Risk analysis, which is known as 'clinical audit' will be enhanced by knowledge extraction algorithm and methodologies from Data Mining & Pattern discovery. Thus, REMINE will create a new modeling system for RAPS using prediction, detection and monitoring techniques. Alert provisioning the final step of RAPS management process, will be provided using RAPS infobroker patient safety framework using advanced business process modeling. The REMINE approach represents the need of the users in defining the best way of organizing the RAPS process management system as well as empowering the health care professionals and the major All of the processes within a patient pathway must be coordinated in an organized way to ensure that time lapses, administrative and/or human or technical error do not occur to the detriment of patient safety. From these data the REMINE project has to define the multimedia content that will form the core information repository of the REMINE system. The process maps will form the basis of the next stages of the project concerned with building the ontology, defining the semantics of the system and integration of data. Clinical process and risk management procedure mapping as well as socio-economic framework surveys carried out in this study have been useful in defining the risks associated with the patient pathway, their significance to stakeholders, and their importance to the organization as a whole. These studies have been a useful exercise in creating greater understanding of the presence and avoidance of risks and how risks are often dependent upon different stages and resources (i.e. actors, information and equipment) in the process. These connections can be very complex and hard to identify without computation tools, thus a system that provides advanced functions of data mining, taxonomy, ontology, design and simulation can significantly improve patient safety in the hospital. Literature review and preliminary results of the socio-economic surveys carried out in the REMINE project have shown how medical errors are not only common and costly, but also preventable. They appear to occur in the setting of three major forces: human/systems errors, information-seeking behavior, and clinical communication. It is possible to model this domain with an ontology engine that extends the concepts already contained in a defined taxonomy and defined in the ontology process building. The REMINE ontology engine will provide a tool that, based on the defined ontology, will support via means of resolving coding disagreements, clarifying the role of communication in medical errors, developing a project database, targeting interventions, and promoting hypothesis-generation. REMINE will develop an ontology tool representing the intersection of medical errors, information needs and the communication space. The ontology engine will be used to support the collection, storage and interpretation of project data. Moreover, the rules based engine will work directly connected to the ontology engine in order to define the semantics of the concepts and relations. Ultimately, the cause-and-effect reasoning will use mainly concepts derived from the ontology and taxonomy defined in the REMINE system. 1083 The application of ontologies in medical informatics in relation to computer based medical systems is no longer considered an innovation [9]. However, currently most of the contemporary systems only consider applying semantics and ontologically represented medical information over a very limited domain. Furthermore, the systems do not represent the information in a way that is representative of the needs of all the stakeholders and generally it is noted that these systems suffer from significant interoperability issues. In this case, the necessity to take into consideration the needs of all stakeholders is a hallmark of the REMINE System. 6. Acknowledgements The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 216134. 7. References [1] Department of Health Expert Group (Chairman, Chief Medical Officer), An organisation with a memory, The Stationery Office, London, 2000, available at http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/ PublicationsPolicyAndGuidance/DH_4065083. [2] L. Kohn, J. Corrigan, and M. Donaldson, To err is human: building a safer health system, Committee on Quality of Health Care in America, Institute of Medicine, 2000. [3] R.M. Wilson, W.B. Runciman, R.W. Gibberd, B.T. Harrison, L. Newby, and J.D. Hamilton, “The quality in Australian health care study”, Medical Journal of Australia, 1995, Vol. 163 No. 9, pp. 458-71. [4] Davis et al., Ministry of Health, 2001. [5] Cineas and Makno, ASI, 2002. [6] REMINE consortium, D1.1 – User needs and requirements, 2008, available in the “Deliverable” section at http://www.REMINE-project.eu. [7] R.D. Wilkins, and L.K. Holley, “Risk management in medical equipment management”, Engineering in Medicine and Biology Society, 1998. Proceedings of the 20th Annual International Conference of the IEEE, IEEE, 1998, Vol. 6, pp. 3343-3345. [8] Drafting Group of the Project to Develop the International Classification for Patient Safety, The Conceptual Framework for the International Classification for Patient Safety, 2008, available at http://www.who.int/patientsafety/taxonomy/en/. [9] Olivier Cure, “Ontology interaction with a patient electronic health record”, Proceedings of the 18th IEEE Symposium on Computer-Based Medical Systems, IEEE, 2005, pp. 185-190. 1084