Clinical Messaging Systems and E-Health Automation

By Robert Keet, MD, FACP

In the last decade, American businesses have come to understand the value of an efficient electronic communication system. Employees can quickly compose messages, attach documents for review, and distribute to individuals or groups across a large geographic area. Businesses using workgroup software programs bring together the skills and expertise necessary to solve complex problems while remaining "in the field." The impact of efficient communication systems on corporate America is dramatic and plays a significant role in the management efficiencies now possible.

The healthcare industry has been slower in embracing these new technologies. However, in communities where clinical messaging systems have been adopted, both the data deliverers and physicians have seen an improvement in efficiency and quality. They easily share information with other physicians and create teams of experts to solve the complex clinical problems that challenge today's clinicians. With the ability to transfer clinical data—such as laboratory tests, radiology results, transcriptions, prescriptions, and clinical orders—quickly from provider to provider, the previously isolated practitioner can deliver healthcare in a truly integrated fashion with a team of clinical partners.

Some healthcare communities assume that a clinical messaging system will be prohibitively expensive and, therefore, hesitate implementing a system. This turns out to be untrue. Costs of electronically delivering data and automating the workflow around the management of clinical data can be justified strictly on the savings accrued from the data providers' ability to "turn off paper." The automation of the physicians' messages including prescriptions and orders further justifies the initial capital investment. Finally, once a community of physicians is on-line, the process of care management, assuring proper preventative and disease management, can also be automated.

This paper addresses the methods and issues related to the implementation of such a communication system within a healthcare community. It is based on the experience gained from a project begun eight years ago in a community in California. Today in this community, two competing hospitals, a large commercial laboratory, a major radiology provider, two MSOs, and 165 physicians and 300 staff that are now "on-line" communicating electronically and automating the processes of managing clinical data. Physicians now receive data that can be stored or used for clinical automation. A master patient index maintains data on 90% of the population, allowing the unique identification of clinical data as well as the automation of clinical processes.

Until recently the implementation of such a system across a diverse group of physicians was limited by technical constraints. With today's more sophisticated, standardized, and less expensive hardware and software, technology has ceased to be a limiting factor. Capital, training, workflow reengineering, and politics are now the most important considerations. In the following sections, each of these considerations will be addressed with recommendations based on our experience in implementing a clinical messaging system.

• The architecture should be completely open and use industry standard coding, such as HL-7 and LOINC. Proprietary hardware and software would limit the acceptability and maintainability of the system.
• It should be Internet based available from any standard browser, making it easy for physicians and staff to access and use.
• The system should be compatible with modern local and wide area network technology supporting Internet and intranet communication.
• The data model must be distributed, with "ownership" of the data remaining at both the ordering and the providing ends. A community of diverse providers will require that the data not reside at or be owned by a single economic entity.
• The system must ensure the unique identification of patients. Without such capability, the electronic delivery of data cannot provide the ability to automate clinical functions. As with all data, this patient index must be distributed across the community and fed by multiple sources of data, not owned by any single entity.
• The user interface must be intuitive, inviting, and easily customized to individual user preferences.

The good news is that all these requirements are now achievable and available from several sources.

The bulk of the initial cost of setting up a clinical messaging system involves creating the means for the data sources to transfer their clinical data electronically. Hospitals, laboratories, and radiology groups, which incur significant expense generating and distributing paper reports, have found that electronic data delivery can be justified by a simple return-on-investment analysis.

Once the data is available electronically, physicians can progressively add tools to automate the management of that data. While independent physicians tend not to be well capitalized, these tools can be added gradually as efficiency improvements justify the expense. Where a physician management structure (e.g., an MSO) or an organized group of physicians exists, the physicians can be brought on-line in a more organized manner based on a strategic business plan. In a managed care environment, even modest savings in the efficiency of clinical care can justify significant capital expenditures.

The implementation can be facilitated by ensuring that physician office staffs have access to the clinical messaging system. Most physician offices have at least one person skilled in computer use who can lead the others through the process.

The most efficient training programs train trainers rather than the entire community. These trainers become champions of the system and facilitate the progressive use of automated processes. They remain as permanent resources in the physician offices.

A healthcare community can reengineer workflow by following gradual steps:

1. Implement electronic data delivery of clinical messages including laboratory, radiology, and transcribed results. Initially, the system can be set up to automatically print results, in sorted order. Physicians and staffs can use the electronic system to look up required results.
2. Use the initial data to build a distributed master patient index. This allows the unique linkage of patients to specific data, a precursor to more advanced automation.
3. Train the physicians and their staffs to electronically transfer clinical data to other physicians. They can also begin to use the system as an intra- and inter-office electronic communication tool with e-mail and groupware capabilities.
4. As physicians become comfortable using the system, they can begin to automate the management of incoming data. They can forward data to staff or other physicians with
5. Specific instructions for action and necessary annotations. In so doing, they will move away from the paper to the electronic management of data.
6. Add out-going message capability including authorization requests, prescriptions, and orders (laboratory, radiology, and hospital). Physicians will begin to see the immediate rewards of an automated system. As they write prescriptions, the system can check for drug interactions, allergic interactions, and formulary compliance. Patient instructions can be automatically printed. Hospital orders can be generated including specific care protocols and pathways, and laboratory orders can be checked against the diagnosis.
7. Implement appropriate care management reminders and alerts. Track to ensure that specific tests, such as Protimes and Glycohemoglobins, are performed at appropriate intervals for specific patients. Track mammograms, pap smears, and immunizations.
8. Add a full repository for all electronic data, thereby eliminating the need for the paper patient record.

Each step along the way will find physicians and staff members at various stages of workflow automation. The implementation accelerates, however, as a critical mass of physicians come on-line at each step. Just as e-mail is now joining the phone, letters, and fax as a form of general communication, clinical messaging will become a mainstay of physician communication.

Data ownership: As clinical data becomes electronically available, ownership of that data becomes an issue. Most would argue that the entity that creates the data as well as the provider who orders the data share ownership. Ultimately, of course, the patient is the true owner of the data. By initially implementing an automated system that simply delivers (rather than permanently storing) the data, much of this debate can be avoided. With the implementation of a messaging system, the data is stored at the natural owner's location. When central repositories of data are created, ownership becomes a critical issue.

Open versus closed systems: A single hospital or laboratory may attempt to capture market share by creating an exclusive system of data delivery. Physicians, however, are unlikely to accept any system not open to all data sources. Data providers can achieve a competitive advantage by entering the network early and by providing the most advanced use of the system.

Master patient index: The data providers must agree to feed and support a master patient index that automates the identification of patients and facilitates the integration of data from multiple sources. The index is distributed, has no specific ownership, and is maintained by software that manages data conflicts and provides automated procedures for merging duplicate patients and correcting data errors. A committee composed of representatives from the data providers must agree on the "rules" of what data can be updated by whom.

To implement a clinical messaging system, an interested healthcare organization follows these steps:

1. Create a steering committee of interested parties, including representatives from hospitals, clinical laboratories, radiology offices, and physician organizations. In the early stage, this committee oversees the choice of a vendor, sets up implementation schedules, creates standards where necessary, and so on; later they may be called on to adjudicate conflicts.
2. Choose a specific technology and vendor. Once a vendor is chosen, the vendor should assist in the process.
3. Choose a pilot data provider, such as a hospital or large clinical laboratory. Since these organizations have much to gain from automation, they represent the best source of initial energy and capital to start such a project. Implement data delivery for this pilot data provider.
4. Recruit several physician leaders to become champions for the physicians.
5. Once a single data provider is on-line and data is flowing smoothly, approach others to join in electronic data delivery.
6. Train physicians and their staffs to move away from paper to electronic data management.
7. Monitor the efficiency improvements achieved with automation.
8. Cultivate champions and mentors and create a process of intra-community support for progressive automation.

A successful implementation requires that data repositories be distributed rather than centralized and that competing healthcare organizations not be denied access to the system. Starting with an open, easily accessible system will assure the long-term success as well as true functionality for all parties.

Once a critical mass of healthcare organizations and physicians are on-line, the initial effort will be self-sustaining. The community can then begin to take advantage of the infrastructure that has been created, adding more advanced automation tools as they become available.

Dr. Robert Keet, MD, FACP has practiced Internal Medicine and Geriatrics in Santa Cruz, California for more than 20 years. He has served as Medical Director of a local IPA and President of a large group of primary care physicians. During the last eight years, he worked with physicians in his community to develop an automated clinical messaging network and served as chairman of the Steering Committee overseeing that project. He consuled with Axolotl Corp. and is now the Medical Director, assisting in the company's ongoing effort to provide clinical automation tools.