EXECUTIVE  MANAGEMENT      •      MARKET STRATEGY      •       VENTURE CAPITAL      •       CORPORATE DEVELOPMENT

Introduction

Healthcare is undergoing a revolution because of technological, political and sociological changes. In order for these changes to result in cost effective, quality healthcare, a fundamental change in the way healthcare providers and payers manage patient information is needed throughout the nation and around the world. The change requires treating patient data in a completely different way than has been considered until recently.¹ In particular, it is useful to think of the patient's medical record as an entity that exists on the network in its aggregate form simultaneously populated from multiple locations. We describe in this whitepaper some of the motivation for this change, what we mean by a virtual patient record, the potential impact of this change, what is required to support this shift in medical information and some results of some early implementations of a virtual patient record.

Background

Information technologies have begun to have profound impact on a variety of business and social applications. These technologies both improve the quality and lower the costs of business processes. Healthcare is no different in this respect. As an example, the increased mobility of the patient populations and changes in healthcare providers and payers has resulted in a patient's medical information being accumulated in a variety of locations -- hospitals, HMO's and doctor's offices -- with little or no linkage between them. Because of these multiple points of entry of patient information into the healthcare system, both healthcare provider and payer get a fragmented picture of the health history of a patient, particularly if she has some kind of chronic illness such as asthma or diabetes. This fragmented view can occur over a regional network of clinics as well as over the entire country. We call this requirement for multiple entry points into the healthcare system "distributed healthcare." Because of distributed healthcare, the patient frequently is the individual with the most complete historical information as to how her clinical illness has progressed.

In addition to the trend towards distributed healthcare, there is a rapid movement to computerized patient records within hospitals, HMO's and even over CHIN's (Community Health Information Networks). However, even electronic access over a region may not be sufficient to track a significant number of patients as job mobility increases. This increasingly wide diffusion of the population requires patient data to be accessible in an organized manner on a national and even global scale, independent of the healthcare provider or payer. To deal with this information explosion, there are a number of organizations working to standardize healthcare information and communication² including the CEN (European Committee for Standardization), UN/EDIFACT (United Nations rules for Electronic Data Interchange For Administration, Commerce and Transport), JWG-CDM (Joint Working Group for a Common Data Model), the Health Level 7 (HL7) group, the CPRI (Computerized Patient Record Institute) and recently the Object Management Group (OMG).

The impact on the healthcare industry of making healthcare information available over wide areas in a secure manner will be quite profound. Such availability could potentially allow for "data mining" of information. This information could then be used to discover and analyze associations between disease entities and previously unknown risk factors (recorded in the patient history), to test hypotheses regarding putative risk factors, or to study disease distribution using demographic data. Applications of "data mining" could also include enabling a physician to do a comparative analysis of a particular patient's symptoms with the symptoms of other patients with similar or different diseases. Having wide area access to healthcare information would allow for more intelligent video consultations. During these consultations, along with the video, specialists in multiple locations could simultaneously see and annotate a patient's record. HMO's could do a better job of outcomes analysis, physicians would have access to better decision support information, and patients could be better educated to manage their health. All of these applications require advanced pattern matching techniques beyond simple database searches.

Virtual Patient Record Concept

For a long time healthcare providers and payers have realized that electronic records have real value and that moving this digital data around to areas where it is needed is highly desirable.³ There has been considerable success at the Veterans Affairs Hospitals and other locations implementing a decentralized patient record system, but these systems have not addressed the issue of doing so over a wide area network between different domains.⁴ With legacy systems, moving data between disparate databases has been a real problem that has been well addressed by the HL7 effort. However, we believe that for ease of access by end users such as physicians and patients, the patient's information must appear to the user as a unified set of data even though it may be spread all over the country. The user's view, of course, might access only a specially tailored subset of the records in order to handle issues of displaying the information in an intelligible manner. With distributed object technology, which can hide much of the vagaries of accessing information, such a view of data is now possible.⁵

This virtual patient record is virtual in that it is a view of the data that might be configured differently at different locations, but that is mapped into a common format at the time the record is required. Creation of the virtual patient record must be done with minimal compromise in the integrity of the data while maintaining high accessibility. For example, simple store and forward systems have potentially serious difficulties because data is copied to multiple locations and then edited and amended locally. There typically is no mechanism to integrate new information entered into any local copy back into the primary record and all other local copies without considerable human effort.

Through a virtual patient record distributed healthcare data is made available through references (analogous to hypertext links of the World Wide Web) and is only brought together (or created) on demand by the end user. Since users generally access components of a record rather than the entire patient record, data movement is minimized. In the distributed system, reference counting capabilities and distributed transaction processing maintain the integrity of the data. Thus, full asynchronous access of the record that enables multiple physicians, other healthcare providers and healthcare payers to update the patient record is supported. Many of these capabilities have already been specified, for example, in the new CORBAservices developed by the OMG.

This model does require ubiquitous network connectivity and accessibility, but high bandwidth transmission is not necessary unless large amounts of image or video data need to be moved. Such an infrastructure is rapidly coming into existence, even in rural areas (For example, in rural areas the virtual patient record can be assembled over the POTS [Plain Old Telephone System] for the price of a long-distance telephone call). The model also requires a robust security infrastructure to support authentication, confidentiality, and data integrity so that there is no single point of failure that, if compromised, would give access to all the information.⁶ This security model has been slow in coming, but with recent developments in electronic commerce, such security systems are imminent. In order to provide robust data access even when larger numbers of users are attempting to access data and a universal but secure way to identify and locate patient information, various replication servers are also required.

We believe that it is now possible to implement the virtual patient record concept if all stakeholders -- government, public, and private -- cooperate in making it a reality in the everyday practice of medicine. Many of the underlying standards are being put into place, but more standard representations of medical objects are needed. This is one of the goals of the new Healthcare Task Force created by the OMG (and also known as CORBAmed⁷).

For example, the virtual patient record will depend heavily on the ability to quickly and securely identify patients and their respective healthcare providers and payers.⁸ This requirement can be met by a Master Patient Index (MPI). Besides the basic architecture to enable healthcare objects to work interchangeably and together, in order to avoid the chaos caused by using existing naming conventions, the virtual patient record system will have to be adopted by a large portion of the healthcare community in a short period of time. This requires use of the standards model similar to that used by Internet or World Wide Web model. We also believe that, in order to provide a cost effective approach to healthcare, the healthcare community must build on the much larger networking and software infrastructure that is being put into place by the nations business community.

Telemedicine or, more specifically teleconsultation systems have seen expansive growth from federal, state, and local initiatives. Although the data on the success of these telemedicine networks are meager,⁹ there are indications that many successes stem from using video technology in the referral process. A number of trials have concluded that the ability to have access to the full electronic medical record greatly enhances the teleconsultation and enables the evaluation of the effectiveness of the resulting treatments.¹⁰

Implications

There are a number of events that must occur in order for a virtual patient record system to become the normal mode of operation of the nation's healthcare industry and there are a number of events that will follow from the extensive use of such a system. One can think of this model as the Internet approach to healthcare in which there is essentially no central control but rather a loosely coupled set of database systems with agreed upon standards for exchanging information. The MPI, for example, has some similarity to the Domain Name Service (DNS) of the Internet and the security model is closely related to DCE (Distributed Computing Environment) security¹¹. We believe that, in order for there to be a real standard way of healthcare information to be exchanged and interleaved, some easily adopted base implementations must be made available to the healthcare community. A few organizations need to demonstrate that this new environment, even on a small scale, can result in substantial cost savings as well as improved quality of healthcare. Others will then follow. It may be argued that a distributed system cannot work because of the enormous number of transactions that must occur. However, we believe that precisely because of the large number of nearly independent transactions, such a system can scale, or be expanded from the local to the national level. The same conclusion applies to security; it must be handled in a distributed manner so that single points of failure don't compromise the entire system. A centralized system, in fact, cannot possibly manage, either technically or economically, all of the information that will be contained in a National Health Information Infrastructure. The challenge to create such a scalable, distributed national system is very great and will require cooperation of private industry, healthcare providers, and federal, state, and local governments.

A virtual patient record system cannot be constructed unless reusable healthcare data information components (i.e., objects) which can be assembled in a variety of ways to meet the highly varied needs of the healthcare practitioner are made available. This will require application of some of the most powerful software tools for enabling object reuse including adaptable, dynamic objects inheriting object behavior.¹² The virtual patient record will be quite dynamic and new data entered must be self-describing so that the computer patient record can evolve with changes in the industry. We believe that the healthcare software market will undergo a major revolution in the next 5 years to accommodate these requirements. The healthcare business model will radically change because of some of the new deployment technologies coming into the marketplace.¹³ Businesses that help set the stage for the future will have a distinct advantage in this new marketplace.

With such a virtual patient record system in place, patients will be able to review their medical records and more actively participate in managing their own health. In addition, when combined with financial information, the system will enable the use of powerful metrics that measure the delivered quality of healthcare while minimizing waste, fraud, and abuse. Such metrics are an essential component of a Clinical Decision Support System (CDSS). In addition, it should be possible for healthcare providers and payers to determine health trends on a national scale and to identify the most effective means of improving healthcare. Regulations may impact the deployment of such a system, but we envision the system primarily augmenting the existing referral practices rather than changing them.

Prototype experiences

Only recently have people attempted to build systems along the lines described above. One such effort has been at the Technical University of Berlin under the direction of Dr. Fleck and supported by DeTeBerkom.¹¹ This BERMED system has similar goals to the TeleMed effort described below, although it is not built upon an object infrastructure. Additional efforts are going on at the West Virginia University at the CERC (Concurrent engineering Research Center) in the ARTEMIS project¹⁴ to make patient information available over public networks in a secure manner.

In a joint effort with the National Jewish Center for Immunology and Respiratory Medicine (NJC), we have constructed a prototype of such a system described above. This TeleMed system¹⁵ enables physicians at multiple locations to simultaneously see, edit and annotate a patient record at remote locations. It handles multimedia data including CT imaging and audio annotations. It is uses Object Request Brokers (ORBs) that abstract away the distributed databases that provide the persistent object storage of the multimedia data. It has object-level security implemented to provide authentication and encryption for confidentiality. It is built with the idea of providing easy-to-use access to complex information while providing advanced data-mining techniques accessible to an end user. It uses a simple ID-server to identify which databases have particular patient data so that the patient record which is graphically displayed is a combination of all the information in the multiple databases. Thus, TeleMed is an early implementation of the virtual patient record described above and demonstrates that the concept is achievable. It has been deployed at the NJC, the National Institutes of Health, and at the Texas Medical Center for early testing and evaluation. Physicians at these three institutions can simultaneously view, edit, and annotate the patient data stored at multiple locations while each physicians can see the data the other physician has entered. To the physician using TeleMed, it appears as if all the data resides on her own desktop computer; there is no indication that multiple databases are involved. TeleMed supports basic data mining by providing abilities to compare images with "similar" features and to visually navigate through a image data base. We have also implemented, where the available bandwidth permits, the ability to support video teleconferencing within the TeleMed system.

Opportunities

One virtue of the virtual patient record system is that it can be extended to a wide variety of areas. For example, the concept can be used in engineering where one deals with a designed or a built system and needs to track its history over a wide region or a period of time. We believe this approach will become predominant over the next decade as the telecommunications and computer hardware and software infrastructure become more robust. We believe that the cooperation of all stakeholders working with together to build a common infrastructure will not only develop new business opportunities but also will make a positive impact on the healthcare delivery nationally and worldwide. The opportunity exists now to invest in a new healthcare infrastructure that will significantly enhance the delivery of quality of healthcare at a reasonable cost.¹⁶

Footnotes

1. Dick Phillips, "A Network-based Distributed, Media-rich Computing and Information Environment" at the Digital Media and Electronic Publishing conference sponsored by the British Computer Society in Leeds, UK in December 1994.
2. cf.: http://www.acl.lanl.gov/sunrise/Medical/standards/standards.html.
3. Richard Dick and Elaine Steen, eds. "The Computer-Based Patient Record, An Essential Technology for Health Care," Institute of Medicine, National Academy Press, 1991.
4. "Decentralized Hospital Computer Program", Department of Veteran Affairs, August, 1995. See also: http://www.va.gov/vama.htm
5. Robert Orfali, Dan Harkey, Jeri Edwards, "The Essential Distributed Objects Survival Guide", John Wiley & Sons, Inc. 1996.
6. Molla Donaldson and Kathleen Lohr, eds. "Health Data in the Information Age: Use, Disclosure, and Privacy", Institute of Medicine, National Academy Press, 1994. and U.S. Congress, Office of Technology Assessment,"Protecting Privacy in Computerized Medical Information", Washington, D.C., 1993. (See also: P. Solovits and I. Kohane, "Against Universal Health-care Identifiers" Journal of the American Medical Informatics Association 1:316-319, 1994. )
7. OMG's CORBAmed Task Force: "http://www.omg.org/corbmed.htm";
8. U.S. Congress, Office of Technology Assessment, Bringing Health Care Online: The Role of Information Technologies, OTA-ITC-624 (Washington, D.C., U.S. Government Printing Office, September 1995).
9. Office of Rural Health Policy, Health Resources and Services Administration, Public Health Service, Department of Health and Human Services, Reaching Rural, (Washington, D.C.: 1994), P. 11.
10. Various presentations at the First Annual American Telemedicine Association conference in Albuquerque, NM, Feb 23-25, 1996.
11. E. Fleck, ed. "Open Systems in Medicine" IOS Press, 1995.
12. Wes Rishel and John Quinn, "Software Components, the Clinical Workstation and Healthcare Networks: How HL7 is Helping You Get There", HL7 Special Group for Object Brokering Technologies, 1996.
13. Arthur Van Hoff, Sami Shaio, Orca Starbuck, "Hooked on Java", Addison-Wesley Publishers, 1996. (see also: http://java.sun.com/whitePaper/java-whitepaper-1.html).
14. Ramana Reddy, et al, "Collaboration Technology for Real-Time Treatment of Patients", http://www.cerc.wvu.edu/nlm/nlm.html
15. David Forslund, et al. "TeleMed: A Graphical, CORBA-based, Virtual Patient Record System for Clinical Use" , Object Management Group, First Class magazine, March/April 1996. LA-UR-96-647. See also http://www.acl.lanl.gov/sunrise/Medical/overview.html.
16. This document is accessible at http://www.acl.lanl.gov/telemedicine/virtual.html

[ Return to Top ]

Bridging the Gap between Concept & Execution