The National Library of Medicine recently acquired the Patient/Problem Oriented Medical Record System Archives, a collection of materials related to the development of an early computer system for organizing patient data and diagnostic decision-making. Jan Schultz generously contributed archival materials from his work with Dr. Larry Weed at the PROMIS lab at the University of Vermont. NLM archivist John Rees asked Jan Schultz about his perspective on the early history of electronic medical records.
John Rees: Tell us a little about yourself. Where are you from? What’s your background?
Jan Schultz: I grew up in Chicago in the 1940s and 1950s. I attended the University of Illinois, studying mathematics and very early computer science. I stayed at Illinois for a master’s degree in math and also worked in an educational psychology computer lab developing SOCRATES, a very early computer-based teaching machine.
I moved to Cleveland and started working on a PhD in Mathematics at Case Western Reserve University. I didn’t last very long at Case. It turned out my main interest was computer science and not mathematics. Call it serendipity or maybe fate, but someone steered me to Dr. Lawrence Weed at the Cleveland Metropolitan General Hospital (CMGH). This was 1966 and I was 24 years old.
JR: You worked with Larry Weed for many years. What was he like? Who else worked with you? What was the lab trying to accomplish?
JS: Larry was brilliant, driven, and dedicated to providing good medical care while educating physicians. As the Director of Outpatient Clinics, Larry taught residents many things, including how to organize medical records. I worked in Larry’s office with the head resident, Dr. Charles Burger, and saw firsthand how they managed the voluminous paper records of patients with numerous complex medical problems.
In the late 1960s, Larry and I worked together to envision a computer-based patient-oriented medical record (POMR). I developed a sample printed output for a ‘theoretical’ electronic POMR, and Larry, Charlie and I reviewed it and discussed how to begin getting the data into the electronic record. We decided that the ‘information-originator’, whether a physician, nurse, ancillary staff or patient, could enter the record directly using a computer terminal. Glass teletypes were just coming into existence, so this decision was radical for the time.
Robert Masters of the Research Division of Control Data Corporation had developed an early system using a touch-screen CRT terminal, the Digiscribe, and a software menu editing and paging system. In 1967 we used a Digiscribe to prototype PROMIS and as a concrete example we showed that with properly structured branching-logic displays, a complete cardiovascular problem could be defined and then stored in what would become a patient’s electronic POMR.
In 1968, Larry and I were Co-Principal Investigators on a grant from the National Center for Health Services Research entitled “Automation of a Problem Oriented Medical Record” to develop a prototype system called PROMIS (Problem-Oriented Medical Information System)
Then, in 1969, Larry accepted a professorship in the medical school at University of Vermont (UVM) in Burlington, and our problem-oriented group moved with Larry to Burlington. We set up our computer room and offices at the Medical Center Hospital of Vermont (MCHV) and called the new computer lab PROMIS Laboratory. Larry and I worked together for 12 years at UVM. Over the years, the PROMIS lab had numerous visitors, so many that taxi drivers started calling us the “Promised Land.”
JR: The PROMIS/POMR system was one of the first electronic health record systems. What was your role in its development? Can you describe how it worked? What problems was it trying to solve?
JS: It became clear that we needed to be able to manage a patient’s complete problem list and for each problem be able to retrieve all of the associated data. All medical record data, once the initial workup was done, was problem-oriented. Larry wanted to be able to retrieve data from the electronic medical record (EMR) in many different ways: problem-oriented, source-oriented, time-oriented (chronological and reverse chronological) and be able to create dynamic flow-sheets for any data in the EMR. In1970, we set up the initial operational POMR at MCHV on the Ob-Gyn floor.
Larry described the PROMIS system as removing dependence on user memory, capturing user logic using problem-orientation, coordinating care among all of the providers, and providing feedback for medical research. I described the PROMIS system as responsive: allowing 70% of all selections to be processed within 250 milliseconds. Because the system needed to be available 24/7, it had to be reliable, and, since a single system could not handle all of the terminals needed for a hospital, the system needed to be scalable. Here is a video demonstrating PROMIS in action. PROMIS was also fortunate to be identified as a landmark in the history of personal computing with a chapter in A History of Personal Workstations, 1988.
In order to remove the dependence on user memory, PROMIS lab managed an ever-growing library of displays of medical knowledge. The final library had over 60,000 displays, and PROMIS lab had 10 ‘medical-content’ domain experts managing and updating the library. Papers written by Mel Conway, a consultant hired by the National Bureau of Standards, and Larry describing the national library are included in the donated materials.
JR: How do today’s EMR systems resemble the PROMIS system?
JS: The PROMIS system of touch-screen workstations connected to a scalable network of mini-computers using a hypertext interface for all clinical and administrative functions was in operation at the Medical Center Hospital of Vermont (MCHV) in 1977. At the time, touch-screens were very rare: most observers did not fully understand the complexity “behind the scenes” needed to make the system so easy to use. Many visitors went away after seeing PROMIS and tried to reproduce it. A group at Carnegie-Mellon University experimented using PROMIS to enhance the human-usability of their ZOG hypertext architecture, but as far as I know, none were successful. Today, every mobile phone has a touch-screen with browsers using HTML hypertext-pages, and the major Silicon Valley hi-tech firms use services with multiple computers tied together in a high-speed network with very large distributed databases. Included in the PROMIS archive are notes related to two patent cases that I participated in, one relating to early hypertext systems and the other to distributed databases.
JR: Software preservation is a challenge for archivists and there is not a working PROMIS system anymore. Can you describe the materials in your donation and how they might contribute to the history of medicine and computing?
JS: While much is being written currently about physician EMR use, I am struck that contemporary EMRs are not looked at as clinical tools to manage a patient, but as an administrative and billing system that many clinicians feel that they are forced to use. PROMIS was a very early electronic medical record that was oriented around a patient, not around the hospital’s billing office. The materials I contributed to NLM can provide many of the technical and clinical details needed to create an electronic patient/problem oriented medical record that is focused on clinical care for a patient. The materials may be quite valuable if future software architects want to understand how Weed and the PROMIS group structured the electronic medical record and the computer system to support it. I still believe strongly in the soundness of the PROMIS approach and did not want to see the materials lost.
JR: We connected with each other almost accidentally before I came to Burlington to appraise the collection, learn about your work, and meet some of your former colleagues—everyone was so happy to see each other this mini-reunion seemed like old home week to me. Have you ever worked with archivists before? What was the donation experience like for you?
JS: This was my first time working with an archivist. It was an exhilarating experience. As I sorted through the PROMIS papers, I reconnected with several old PROMIS colleagues and the energy that got the project going. I remembered Larry at his visionary best. Larry was focused somewhat like a bulldog, and when I met him, I was a twenty-four-year old bulldog, so we worked well together. Over the years, I’ve become mellower. I don’t think the same can be said for Larry Weed!
Julie McGowan is a medical informatics expert who at one time headed UVM’s medical library. It was she who suggested I talk to the NLM. When you visited Burlington, assessed the materials, and met several PROMIS folks, your thoroughness and enthusiasm made it clear that the NLM was the right home for these materials.
Explore the finding aid for this collection and learn more about our Archives and Modern Manuscripts collections. For questions about this collection please email archivist John Rees or contact the History of Medicine Division Reference staff at NLM Customer Support or call (301) 402-8878.
Not mentioned here is the Interact Network of teaching hospitals at Vermont, Dartmouth, and Central Maine and that was based around Dr Weed’s electronic patient information system to record and cross-communicate patient data among them. It was the pioneer of todays telemedicine. So this little lab made first-use of what we now call ‘file servers’, ‘touchscreens’, and ‘telemedicine’. The headscratcher why Dr Weed’s Problem-Knowledge Couplers software never got wide acceptance among clinics. It was and is brilliant. I bumped into it last week at an orthopedic clinic when they were taking down information about my case.
PROMIS was a clinical system organized around a patient’s list of problems and progress notes using Subjective, Objective, Assessment and Plan sections. It managed the patient’s medical record using the problem list as an index to much of the data in the EMR. Inspired by Weed and popularized by use in the PROMIS system, aspects of the problem list and the SOAP notes organization have been incorporated into many paper and EMR systems. However, unlike PROMIS, the data are not associated with a specific problem but with a data source (e.g., physician note or lab test result) or with a date. The clinician must infer the clinical rationale for the data in the patient’s chart.