IBM Almaden Research Center

From Regional Healthcare Information Organizations to a National Healthcare Information Infrastructure

Abstract:

Recently there has been increased focus on the need to modernize the healthcare information infrastructure in the United States[1,2]. The US healthcare industry is by far the largest in the world in both absolute dollars and in per cent of GDP (>$1.5T – 15% of GDP).  It is also quite fragmented and complex. This complexity, coupled with an antiquated infrastructure for medical data, leads to enormous inefficiencies and sources of error. Driven by consumer, regulatory, and governmental pressure, there is a growing consensus that the time has come to modernize the US Healthcare Information Infrastructure (HII). A modern HII will provide care givers with better and timelier access to data. The launch of a National Health Infrastructure Initiative (NHII) in the US in May 2004 – with the goal of providing an electronic health record for every American within the next decade- will eventually transform the healthcare industry in general just as I/T has transformed other industries in the past.  While such transformation may be disruptive in the short term, it will in the future significantly improve the quality, efficient, and successful delivery of healthcare while decreasing costs to patients and payers and improving the overall experiences of consumers and providers.  The key to this successful outcome will be based on the way we apply I/T to healthcare data and to the services delivered through that I/T.  This must be accomplished in a way that protects individuals, allows competition, but gives caregivers reliable and efficient access to the data required to treat patients and to improve the practice of medical science.

Introduction:

Improving the state of healthcare I/T has become a bipartisan top-priority issue for the United States.  In his State of the Union Address on January 20, 2004, President George W. Bush stated[1]:

By computerizing health records, we can avoid dangerous medical mistakes, reduce costs, and improve care.

Also in 2004, Congressman Patrick J. Kennedy (D-RI) officially introduced the "Josie King Act" (also called the "QUEST Act")[3]. This act is named in honor of an 18-month old child who died as a result of preventable medical error.  The bill proposes to create an I/T backbone for the American healthcare industry by 2015.

The largest challenge we face in realizing a national information infrastructure is to define an implementation model that is<> simultaneously consumer-centric, cross-institutional and supports the longitudinal healthcare and health record standards emerging around the globe[4-9]. Furthermore the model needs to support the private healthcare provider model of the United States[4]. The system should be scaleable and capable of supporting (in the near term) the Regional Healthcare Information Organizations (RHIOs) and networks that are now emerging at the county and state levels[10,11,12].  As the system grows, RHIOs will become interconnected through a larger NHII network.  In a national system, the RHIOs must have access to pointers to Patient data regardless of where the records are located. Through a data federation mechanism, the RHIO will be able to bring together, on the fly, all required patient’s records and present them to a clinician at the point of care.

This new infrastructure would enable providers to cut long-term archiving costs and to provide better care as patient medical histories would always be available. Insurers would benefit, too, as quality of care for their customers improves. Patient safety would be significantly improved and unnecessary testing would be avoided. Privacy would be better protected, as no global patient identifiers would be needed.  Most importantly, a true EHR with summary and topical parts could be available at the point of care, based on processes that would constantly reconcile and summarize the incoming temporal data received from all providers who have seen a given patient.

Towards a National Health Information Infrastructure

There are many complex regulatory and policy decisions required to establish an NHII system[5-9]. We must meet critical technical requirements while satisfying a Social Contract for medical care appropriate to the United States. Fortunately the technical challenges are greatly reduced by recent advances in Web Service technologies[13]. The most important of these recent advances is the development of the Enterprise Service Bus (ESB). The Enterprise Services bus is a scaleable integration architecture that permits incremental integration or addition of new data sources and data services driven by business requirements. Each new integrated component can be wrapped as a web service so the system is not limited by legacy standards or constraints. In IBM Research, we are working with IBM’s Healthcare/Life Sciences Division to prototype a multipurpose HII test bed based on an ESB architecture. The goal of this testbed is to

• Demonstrate a web service architecture that provides for scaleable integration of healthcare information across the distributed healthcare enterprise.
• Illustrate a replicable methodology for healthcare data and application integration from the RHIO to the NHII scale

The architecture of the HII testbed will utilize the new Service-Oriented Architecture (SOA).  In this model, participants in a regional network will make use of JServices running on a common Enterprise Service Bus. SOA enables flexible connectivity of applications and resources by representing them as services with standardized interface that exchange standardized structures (Service Data Objects SDOs). This allows rapid integration of new and existing services and a staged development of a full National Scale system. 

Figure 1: A Services Oriented Architecture for a Model Healthcare Organization

An important goal of our Research effort is to demonstrate cross institutional EHR as depicted in Figure 2.  In a regional system, organizations including local hospitals and ambulatory networks will continue to rely on local and unique infrastructures of PACS system, etc.  Such enterprises include providers, regulatory agencies (e.g., the CDC), payers, and individuals (personalized medicine portals).  Each of these will have their own enterprise service bus with appropriate services attached, while an additional cross-enterprise service bus will operate the cross-enterprise services.  Data that moves across enterprises (red dashed line in Figure 2) will need to be de-identified according to HIPAA unless patient consent is secured. To understand the requirements for an NHII system, let us first consider the integration across two regional systems or RHIOs. Each institutional system emulates a community, and as such represents much more than a single hospital or healthcare enterprise.  It is the enterprises service bus that takes a heterogeneous institution and transforms it into a single integrated (virtual) healthcare organization. A fully interconnected national system is represented by a network or hierarchy of RHIOs and satellite institutions (Figure 2).

To hide the complexity of the diverse and heterogeneous data definitions in use today, cross-institutional functions will be enabled based on the recently approved "HL7 EHR Functional Model" definition[14]. A key component required to accomplish this is an institutional extraction/translation service within each RHIO. The job of this "extractor" is to translate data from individual (ISV) data sources converting proprietary formats into a common standard schema (e.g., CDA R2).  By converting to a standard schema, RHIOs can exchange data nationwide. Data received in the standard schema can in turn be converted to the specific data definition used by particular data stores in the receiving RHIO.  Other services and data adapters will support the most important existing standards including HL7, DICOM, HIPAA, LOINC, SNOMED, thereby addressing the whole spectrum of biomedical care including healthcare, public health, clinical trials, and the life sciences[14-18].

Examples of some possible cross-enterprise services are:

• Patient Healthcare Information Location Service - a service to locate for an identified consumer, the actual location of his data. It includes an index of consumers and references to their data.
• Characteristics Healthcare Information Location Service - a service to locate data by characteristics such as locate data with a specific observation. It includes an index of de-identified consumers data and references to the original data.
• Person Identification Service - a service that correlates the various ids of the same consumer as well as assigns an anonymous patient id (AGPI) when needed.
• Clinical Terminology Service - a service that enhances the data by adding codes from controlled vocabularies such as LOINC, SNOMED, ICD9.

The composite of the core components and services will form the infrastructure for a Universal Health Information Integrator.  Extraction, transformation and aggregation services designed to operate on the electronic health records in the Healthcare Information Systems will allow the NHII offersuch innovative functions as:

• Virtualization/federation: the ability to assemble a composite health record from the partial records contained in disparate RHIO’s.
• Distribution: the ability to move a record among systems (through a standard data exchange format).
• Windowing: the ability to reveal/transmit only relevant portions of a health record for a given purpose and to a given user controlled by record owner policies.
• Delegation: the ability for a record owner to delegate windowing authority to a trusted expert such as a primary care physician.

Figure 2: Cross Institutional EHR Platform

In the future, individuals will require the ability to control and manage their own healthcare. The ESB architecture supports the creation of various web portal based applications to support personalized medicine as well as more efficient physician access to patient data. This paradigm change will also enable new service and utility models for the healthcare industry.  Once individuals can access and control their own medical data, they can make better informed decisions. The availability of more complete medical data will also lead to the creation of new medical and health services, giving individuals greater choice and control in planning their own care. Electronic data sources will provide the information necessary to understand the real outcomes of accepted medical treatments, advance medical research, and advance the science of medical care.

Table I: Services enabled by a modern NHII

With the HII test bed, we can explore many of the advanced services that will be enabled by a nation wide health information infrastructure. All of these I/T services can be built around reusable technology components per the framework depicted in Figure 1.  Some - but not all - of the services detailed below are also referenced in the figure. Some of the many future services are also listed in Table I above.

Conclusion:

The time is right to begin the work to transform our national healthcare infrastructure through the proper application of IT. However, unlike IT for finance, which evolved over several decades, the escalating cost, demand for, and error rate of our current health care infrastructure dictates that a significant change in healthcare IT must take place this decade as the baby boomers retire. We must achieve this rapid change in a cost affective way that does not violate the social contract between individuals and care givers, and we must get it right the first time. To guarantee success, and to act quickly, we can take advantage of the latest web services technologies and enterprise service bus architecture to create a scaleable system wherein new data sources and new I/T services can be added incrementally as required by an emerging new infrastructure. With proper application of I/T, we have the opportunity to turn a revolution in healthcare into a renaissance of healthcare that provides better, safer, and more efficient care for all.

References:

1. See: http://www.whitehouse.gov/infocus/healthcare/  and references therein
2. http://nihroadmap.nih.gov/clinicalresearch/index.asp
3. The Josie King act of 2004 (H.R. 4880),  http://www.afehct.org/library/pdfs/HR4880section-by-section.pdf
4. http://aspe.hhs.gov/sp/nhii/
5. Health Information Technology Report http://www.hhs.gov/news/press/2004pres/20040721.html,
6. and http://www.hhs.gov/ocr/hipaa/
7. NY Times (7/21/04) http://www.nytimes.com/2004/07/21/technology/21record.html
8. Center for Health Transformation (7/21/04) http://www.healthtransformation.net/news/chtnews.asp
9. iHealthbeat (7/26/04) http://www.ihealthbeat.org/index.cfm?Action=dspItem&itemid=104473
10. Brailer, D.  Augostinos, N.,  Evans, L.,  and Karp, S., Moving Toward Electronic Health Information Exchange: Interim Report on the Santa Barbara County Data Exchange,  http://www.chcf.org/topics/view.cfm?itemID=21086
11. http://www.taconicipa.com/
12. Berman, J., NY Presbyterian Inks Big Multi-Year Deal with GE Medical,  Health-IT World, http://www.imakenews.com/health-itworld/e_article000183666.cfm
13. IBM Healthcare and Life Sciences, http://www-1.ibm.com/industries/healthcare/
14. HL7 functional model http://www.hl7.org/
15. DICOM standard, http://medical.nema.org/
16. National Standards to Protect the Privacy of Personal Health Information, http://www.hhs.gov/ocr/hipaa/
17. Logical Observation Identifiers Names and Codes, http://www.loinc.org/
18. http://www.snomed.org/