1997 Partnerships for Networked Consumer Health Information Conference

Transcripts of Plenary Sessions and Breakout Sessions

Customizing Information 2: "Patient Records for Patients"

Tuesday, April 15
3:45-5:15 PM

Moderator: Margaret Amatayakul, Executive Director, Computer-Based Patient Record Institute, Schaumburg, IL

Speakers: David Forslund, Ph.D., Deputy Director, ACL, Los Alamos National Laboratory, Los Alamos, NM

Ross Fletcher, M.D., Chief of Cardiology, Department of Veterans Affairs Medical Center, Washington, DC

Peter Szolovits, Ph.D., Professor of Computer Science, Department of Electrical Engineering/Computer Science, Massachusetts Institute of Technology, Cambridge, MA

Amatayakul: I am the moderator for this event, and as a member of the health profession, I consider myself a health advocate like the other professionals before you. As we get more into information technology, we are losing sight of the patient's needs and records. I am here to set the tone for the following discussion relating to computer-based patient records.

The Computer-Based Records Institute was formed by varied recommendations and is a nonprofit organization trying to solve the problems attributed to information technology. We have coined the term at the Institute for "computer-based records" to mean giving the patients care by computer record. It is a pretty broad use, and as the organization has been working to do so over the last 5 years, we have established criteria to further explain this definition.

As we begin to take advantage of information technology that has become integrated into our lives from multiple sources, it can now serve as a primary source of patient care. Many of these new innovations have wonderful components, but we still have a way to go to make the total package exist for everyone.

We recently had a summit and invited Don Detmer, Chairman of the Institute's Study Committee, and currently Chair of the National Committee on Vital and Health Statistics. He spoke on a study released in 1991, stating that computer-based records may be established and in control within a decade.

As we move toward use of computer-based records, we move toward a focus on people. Today our focus is on the element of patient "centeredness" of the record. But you, the audience, are challenged to bring us step by step to a higher level so we can better reach and serve people. A lot can be done to improve the patient monitoring process, as well as to influence legislation that is already underway that does not discuss patient access. In aiming for these goals, we are trying to promote interaction from patients facilitating their contribution to their records.

I would now like to introduce the panelists. Our first speaker is Dr. Dave Forslund, a known wizard of technology in the medical field who sits here with his ThinkPad. Dr. Fletcher is the Chief of Cardiology at the National VA Hospital and first to use and be recognized for the use of computer-based records. We also have Dr. Szolovits, from MIT. Hopefully, we will have some time after presentations for a brief question and answer period.

Forslund: I want to talk about various issues surrounding patient records. Certainly security is an issue, and in trying to deploy solutions to the problem, I will give some view graphs from the Patient Center Access to Secure Systems Online (PCASSO) project, mentioned earlier this morning. Most users are accessing the Internet and could be doing so out of their homes with secure user IDs to access their own records.

I would like to call attention to the fact that we need a security infrastructure to give confidentiality and data integrity. For example, Jay Sanders mentioned e-mail contact with patients and doctors, but we need to secure confidentiality. Other problems are information overload created by too much data, accessing data, and having a master patient index. This index would report data location, health care changes, and where the patient moves. Translation is another problem in that users cannot always understand information as it is applied in variable contexts. We need standardization. How do I provide this information so I can understand, as a consumer, my situation? I certainly don't have all the answers, but I am searching.

We need to look at the integrity of the data as well as at human safety. Patient privacy and intellectual property also hang in the balance. How do we handle anonymity on the Internet? We must establish a method of authentication. Access control is also necessary, which authorizes the right users to the system. In other words, confidentiality should be maintained, and integrity of the information content should be unable to be altered, except under authorized circumstances. And there should be some guarantee that the actions taking place are reliable.

Some of the building blocks of the intranet and Internet security are firewalls; IP addresses; user authentication; various hardware such as fingerprinting, crypto buttons, and smartcards; and session-specific encryption (both symmetric and asymmetric -- private and public keys). An example is that, in New Jersey, welfare checks are printed in fingerprint manner. For safety, smartcards could be used. For data integrity, there is the Universal Resource Name (URN), which is similar to a URL but safer, as it can't be moved around or broken into. The URNs provide mapping between a URL and a name space. Our servers and browsers today can actually handle it, but we aren't told that. Secure end-to-end transactions require trusted code at both endpoints. One of the unsolved problems is the Trojan horse attack. An example of this would be, every time you download your medical record, it downloads to someone else as well, if you don't set your browser's security features.

The name of the project is PCASSO. The goal is to secure the use of the Internet for a specific purpose. The roles of the system relate to the primary caregiver, secondary caregiver, patient, research administrator, and so on. They basically have multilevel security in the PCASSO system, accessing the secure components of records. Increased security is still necessary, but not efficient as of yet. We can still give out our information without knowing we have breached confidence. For example, employers and employees could print out a record and get a copy while someone else could get a copy. There is now a market for selling your data. It is noted that 80 percent of medical data are retrieved without the patient knowing it.

In terms of data accessibility, the master person index is necessary, a lexicon for translation of information into a form we can all understand.

A legal and social issue that is raised is, who owns data -- the patient or the people who create the information? There are bills in Congress right now addressing this. But there will inevitably be multiple choices, so we think managing the data from point of origin is best. I mention the "extranet" as well, a kind of summation of Internet and intranet technology that will increase bandwidth requirements. Who will pay for those increases as well as pay for a means of access? Will this system be separate from our normal health care management? Another issue is whether people will accept the security system on the Internet.

Also, who reimburses the consultation appointments? What is the role of the Food and Drug Administration in all this? Going beyond the World Wide Web, we need better ways to manage the complexity. The system can be used for consultations as well as direct patient care.

We need to move the data as necessary. We need security and data integrity. We need to manage the high, complex volume. We do have some telemedicine deployment from the National Jewish Hospital, as well as from some hospitals joining us in New Mexico. In New Mexico we have frontier medicine -- not rural but frontier -- that can lead to a 4-hour drive to receive medical care. There are 15 rural clinics in New Mexico using telemedicine, and teaching people how to use the system has been the challenge.

Here is a sample medical record form Los Alamos via Internet, modified to protect the patient. We can bring up drug interventions and bring up various images, and this occurs over a plain old telephone line. There is a CAT scan on the screen. This is basically what we are deploying in northern New Mexico. The web site in Los Alamos is http://www.acl.lanl.gov/telemed.

Fletcher: That was a fascinating discussion, and at the National VA hospital in Washington, we are now thought to be the underground, initiating the computer-based patient records. I am a cardiologist, and following patients with pacemakers is getting harder. We now are in 1997, and the principle of taking care of patients so they think that they are being cared for in a personal way is getting more difficult as we manage cases quicker through information technology.

We have made 460,000 house calls through telemedicine, including interview and electrocardiogram (ECG) over the telephone. We have had records pulled together and have been completely paperless since 1988. The initial data points are sent electronically. I currently follow all patients east of Mississippi, while other doctors do the rest, completely following all patients across the country. We have follow-up visits, and ECGs are in the computer and can be brought up instantly. Here is an example of the form that we have sent in by mail and now put into the computer. Here is an example of an ECG. We get excellent transmission so it is very clear -- better than what I can get at their bedside. We do not see errors of the hookup by the ECG as causing a problem, because we just call down and say you need to hook it back up.

The center started with 3,000 patients and now is up to 9,000. As you can see, it is growing. The business is growing, and as it is semivoluntary for VA hospitals, resources have increased thanks to Government funding. Caseload per caller is well over a thousand now; workload is now double; and there has been a safe reduction in visits using international generator failure rates to monitor the generators. With what we are monitoring we never saw generator failure, only lead failures, which are random and rare. Survival and use of the device was developed by nurses' suggestions, based on patterns of failure in those who used their pacemakers a lot. And note that the failure rate was not because of the generator itself. The leads followed as well; some were failing. We were following the newer ones, which were not failing.

Here is an example of the screen that comes up with every caller, with the different patient notations, which cite problems of the specific generator as well. This report comes up, we send it to the doctor, and we look for changes in pulse rate. When we see changes, we recommend replacement. We like to use the units for as long as possible, but not until they fail. We have experienced no failure without rate change -- in other words, no battery failure. Lead replacements were 21, while generator replacements were about 40.

At the medical center, all statistical and generator survival curves are available. One can log in and get the same information for the referring institution from the VA hospital. In addition, patients are instructed to give notification if their rates are going down. They are given an 800 number where they can report any symptoms.

We have downsized the clinic as well -- the patient does not know this -- but we have gotten the clinics down to one physician at a time. With call schedules, we will individualize each call, noting any previous problems, giving immediate feedback if necessary, and asking if there are any new symptoms. We will then give advice, if needed, to the patient. In an example of how this system works, we faxed an ECG to a vacationing expert. We then had the patient admitted, changed the lead, and the problem was solved.

Immediate care is available at any site. Another person on vacation suffered from dizziness. The patient carried the unit, reading the transmission, and was told to go to the nearest hospital where the lead was reinserted by ECG observation.

Many of us have attempted to humanize the computer. We sent out birthday greetings, and patients responded positively. Direct interviews with clinicians were also helpful because they give immediate feedback to patients, who can get us anytime through emergency numbers -- and it isn't very expensive for us to have the 800 number. Here is the card on the screen, giving numbers and so on, while wishing patients a happy birthday. We received letter after letter of appreciation.

Patients have the information about their generator and their lead carried on them at all times. The patient summary can also be produced, which includes the lab work, medications, discharge summaries from before, and clinical work. And all this can be pumped out of the computer in an easy format at any time. All of the medical information is on network and can be seen at any of the 900 work stations in the hospital.

Szolovits: I wanted to talk to you today about Guardian Angel. We have been working on the project since 1974, building expert systems, centering our work on care institutions to create lifelong, personal, active, health information systems. Based on a lot of experiences, what I was doing -- building computers for doctors -- was not the primary importance, for it is the patients who need them. We see the main characteristics of the system as the following:

It is lifelong. How many different institutions in your life have cared for you? Many changes are most likely; many times your information is lost in the move. With this system we could track all data collected, as well as record your own observations.

It is personal -- it is about the patient. When I need a system that knows the patient, I don't mean a smartcard, for it is actually quite dumb as it is a floppy disk. I want something sitting and running all the time, a computer in constant process running until after your autopsy. It is constantly asking itself questions, such as whether information is sensible, whether the patient is being cared for correctly, whether the medication is being applied correctly. And, if there is a problem, it could send an e-mail or ring bells to notify the proper people. This is an information system that is about health.

Surgeon General C. Everett Koop has made arguments for a system that knows the patient because of a concern for quality. This has since escaped the health care profession. Patients are highly motivated when something happens to them, but we have paid too much attention to the providers of health care as a source, rather than to the patient. By the ethos of medicine, we should take the patient's view into account. The benefits from involving patients as active participants will give us earlier detection of diseases, earlier notification, and enable us to educate patients about their health care in their individual situation.

The system must be patient-owned for security reasons, and it must be one that is lifelong and comprehensive, giving all information as well as allowing additions by the patient. Therefore, educating the patient is important.

Personal interaction with the health care system is important for the proper transfer of information. We could give the patient a medical encyclopedia as well so they know what to enter and so they understand test results. We must permit unobtrusive, continuous monitoring of the relevant heath-related activities and conditions. Portable intensive care is needed, allowing constant monitoring in an ambulatory fashion.

One example of a monitoring activity is -- you know those sneakers that blink? Well there is something we could add to the Guardian Angel that would give an indication of how you were feeling by your activity, depending on walking speed. There are other ways to get communication between the shoes and watch, too. Adding smarts to the system involves monitoring the progress of the patient, by their own knowledge, through different means.

Here is an example of the Guardian Angel onscreen, looking through the window, which makes the program much more user friendly. It is important to make it attractive. We also have education videos showing how to hook up equipment for testing purposes. Also, when we look things up on the web, we are now attaching to links on the Internet that could educate the patient as well, such as pages on diabetes.

Currently we have another project, which is interestingly enough about gestational diabetes and called Postpartum Guardian Angel. It focuses on communication between the mother, newborn, and pediatrician. The programs as you can see are given in "baby book" fashion, making them user friendly. Baby tips are given as well, advising on care of newborns. This example is used for Sudden Infant Death Syndrome prevention. The little icons take you to different places on the system. Here we note problems with breast feeding, recording the process and times, which are uploaded to the hospital for analysis.

Well, what I have shared with you today is just a sketch. There are the technical challenges as well. Creating a computer that stays up for more than a month is quite a challenge. So there is the problem of needing new hardware for the lifelong process. Standards are needed to overcome the "tower of babble" caused by different and nonstandardized record systems. We also need a noninvasive sensor that can track physical behavior relating to the physiological state. Overall it is difficult to foster a viable approach that is not very expensive. The idea grabs many people, and I hope to have positive experiments to push us forward.

Question: Dr. Forslund, or any others, I have two questions relating to the certificates of authority and moving away from the World Wide Web. Assuming there are some elements to the web, how valid do you see the certificates of authority, and how do we get patients to have validation certificates as well?

Forslund: I misspoke if I said the web will go away. It will not go away, but the key of certification is the question. We could move toward secure e-mail, and that is the first step. Management of the "keys" is the question -- the smartcard may work best as long as there is incentive not to lose it. You can make it invalid if lost, kind of like a credit card. There may be hardware coming along soon, as in Europe -- they all have smartcards, and we will have them shortly. We just don' t have the infrastructure to use them yet. These cards would be useful to commerce as well as health care. Knowing they could be used for money, people would give the card the extra protection.

Szolovits: I have been working on a study -- you can get it on the home page http://www2.mas.edu/cstb. It deals with computer science boards. Clearly with the technical mechanisms there are problems, but our report goes into the nontechnical problems as well.

Fletcher: The VA hospital has investigated the use of smartcards, and patients can carry up to 50 megabytes of information on their cards. From the physicians' point of view, this is very important.

Question: This is all neat, and I see "boys with their toys." But it seems like it will be very expensive to deal with the problems. I ask if the specifics of data are necessary -- or if there is an overkill and if a nurse calling would be better. Where is the value added in the process?

Szolovits: I can't say that each woman would do as advised, such as in the case of the breastfeeding example, but I have two answers: The attitude the mother would take in the case of the gestational problem, it was pointed out, depended on the background of the individual mother and where she was from. Some mothers are very excited about the work and using the computer. Some are not as technologically oriented and would not utilize the system. But our design criteria takes into account that the system must cope with varied levels of interest and knowledge in patients, and it will provide useful feedback.

I agree that a friendly call from a nurse would be preferable, but the health care system is squeezing this service out, and the naval hospital had one nurse calling all patients. She spent all of her time calling and trying to get in touch with all the patients. Electronic means could have done it much faster. Another project I didn't mention is in Boston, looking at behavior modification issues with regard to human contact and social aspects.

Amatayakul: I thought about that as well. But I thought of how we don't always see ways the space program could help us out. Also, the long-term effects are never seen. Thank you very much for your attention. I certainly enjoyed every bit of the presentation.

Back to Summaries and Transcripts page.