Q and A With Dr. Clem McDonald

PDF File: 

HCT3_wp_mcdonald.pdf

Regenstrief Institute, Inc.

The director of Regenstrief explains how the longest-standing continuouslyrunning EMR system was developed.

Healthcare Technology: Can you tell us a little bit about your work at Regenstrief?

Clem McDonald: The Regenstrief Institute has invested in medical information systems since 1972, when we began to build an electronic medical record (EMR) called the Regenstrief Medical Record System (RMRS). It grew over the years, continues to operate today and is one of the largest EMR systems. Regenstrief has organized a consortium of the five major hospital systems (15 separate hospitals) in Indianapolis called the Indiana Network for Patient Care (INCP; all of five of these systems contribute information to the central INPC database. However, it’s not a single database in the center; it’s a federated database, so each of the institutions that contribute the clinical data has its own separate physical file system.

For a number of purposes, the data can be combined for one patient across many institutions for display and care. We have stored 700 million rows of discrete clinical observations, 14 million narrative reports (like discharge summaries) and 45 million separate radiology images from these five different systems. One hundred and twenty message streams deliver 90 million Health Level 7 (HL7) messages per year. For at least five years, ER physicians have been able to access the combined data from all INPC hospitals for use in emergency care. The access by physicians in any one ER is limited to the patients who have been checked into their ER for 24 hours from the time of checking. So the access is controlled very tightly. We now have agreements in place to provide the same kind of access to hospital physicians while patients are in the hospital. So while a patient is in their hospital, they will be able to access data, e.g., laboratory tests, radiology reports, operative notes produced at any of the participating hospitals and view it as though it came from a single medical record. But a physician at hospital A will only be able to see the full INPC medical record for patients who have checked in to hospital A.

The INPC also assists public health activities. For example, it examines laboratory results as they are delivered to the central database, looking for reportable cases according to predefined rules.When it finds evidence for such a case, e.g., a positive culture for cholera, it ships information about this report to the public health department in an encrypted form. Finally, INPC assists many kinds of important medical research.

HCT: Regenstrief was ahead of its time.

CM: Regenstrief has a strong track record in medical informatics. The RMRS is the longest-standing continuously running EMR system. Regenstrief investigators published the first randomized trial of computer reminder systems in 1976; deployed the first smart order entry system in 1984; and the first (and possibly only) randomized trial of computerized physician order entry versus manual order entry in 1993. We began investing in standards in 1984 because we realized that the cost of interfacing electronic data sources to electronic medical records would prevent the widespread adoption of such systems. We wrote the first consensus clinical message standard (ASTM 1238) and contributed to the first and succeeding HL7 message standard. (We wrote the chapter for observations and for orders.)

This effort in standards development has had a direct benefit to us. Without the HL7 message standard we couldn’t have connected the hospitals in Indianapolis and created the INPC. We have also been developing standards for naming and coding observations, measurements and reports since 1994 — the result of which is the LOINC database of universal observation codes. The LOINC codes are a lingua franca for reporting clinical observations so that when a test such as white blood cell count arrives in an HL7 message, the receiving system can store that white blood cell count in a flow sheet combined with white blood cell counts from other institutions.

HCT: It sounds like the biggest challenge has been the standards issue.

CM: Well there are challenges everywhere. However, HL7 was up and running at all of the hospitals when we began the INPC project — a windfall to us. Such standards are a sine quo non for success. Challenges do remain. One is the fact that observations (e.g., laboratory tests, pulse rates) are identified in the HL7 message by local and idiosyncratic codes. Most institutions have not yet adopted LOINC codes so our biggest effort is to map the local codes that identify tests and measurements in HL7 into a common code system — LOINC — so that a glucose from one laboratory can be equated with a glucose from another.

HCT: Do you think that we’ll have EMRs implemented nationwide in five years?

CM: Well first I would like to make some distinctions between the functions included in the rubric of EMR. The first is the repository function. The repository gathers information from many sources, systematizes and unifies it and provides tools for retrieving and displaying information about a single patient — for care purposes and for searching across patients for management or quality assurance or research purposes. These kinds of functions, though not trivial to create, are universally welcomed by clinical users and managers.

The second class of functions might be called work flow and workstation functions. These can demand complete re-organizations of the practice, and major changes in work flow and work assignments, i.e., physicians may be required to enter their orders and notes into the computer. Depending upon the kind of practice, the maturity of the system and the degree to which the computer demands more data gathering than the manual approach, these systems can require more physician time which can hurt their productivity and be considered a detriment by the physicians.

It is important to separate these two classes of functions, and be sure you have a repository function before implementing physician workstation functions. You get much value from the repository. Physicians always like having access to a repository — everywhere. It helps them to do a better job because they can always find the information they need with the computer. It takes too long to call for the chart or call the lab. The repository can also deliver reminders to improve quality and loads the database with outcomes management and research. So, I think ideally we should be thinking about EMRs as two parts, with implementation done in two steps. Put all the energy in the repository function first, and then take on the other physician workstations things — in steps. Today too often, the system is thought of as one thing.

Now to answer your question. I think you could easily have the repository function deployed almost everywhere in five years. Most hospitals, at least the hospitals and large practices, are already building repositories. But when you talk about the physician workstation functions that require retreading the office and all of the users and having everybody being online, that will take longer. Realize that it takes less physician time to shout an order or write it on their chart than to log on to a computer and navigate to the place in the computer and then enter the order. Further, regulatory demands on physician time are increasing, as are the demands to see more patients per hour. So it will take longer for this more demanding functionality to disseminate — maybe 10 years.

One last point: We should not be thinking in terms of a national database or national system, but a series of independent communitywide systems.