Hubert H. Humphrey Building
200 Independence Avenue,
S.W.
Room 801
Washington, DC
Call to Order and Introductions - Dr. Cohn
DR. COHN: Good morning, and for those of you from the West Coast such as myself, a very early good morning. I want to welcome everyone to the first set of hearings of the Work Group on Computer-Based Patient Records for the National Committee on Vital and Health Statistics. I'm Simon Cohn, a member of the NCVHS, and co-chair of the work group. On my left is Jeff Blair, who is also a member, and fellow co-chair.
The charge of this work group is related to language in the Health Insurance Portability and Accountability Act of 1996. Let me just quickly quote the section that applies to what we are doing. Section 263 of this act requires that the NCVHS shall study the issues related to the adoption of uniform data standards for patient medical record information, and the electronic exchange of such information. And then shall report to the secretary no later than four years after the date of enactment of the legislation -- in other words, August 2000 -- recommendations and legislative proposals for such standards and electronic exchange.
Meeting the requirements of this subsection is really the primary focus of this work group, and this is really the first set of hearings designed to explore the issues. Over the next year we expect to having a number of hearings exploring various aspects of the issues pertaining to uniform data standards for patient medical record information as we prepare our recommendations on the subject.
Now Jeff and I want to thank the testifiers who have made themselves available, starting with panel one, over the next two days. We certainly appreciate your participation and willingness to provide input.
With that, I would like to ask the NCVHS work group members and staff to introduce themselves, followed by those in the audience.
MR. BLAIR: I'm Jeff Blair, vice president of the Medical Records Institute. I am co-chair with Simon, and a member of the National Committee on Vital and Health Statistics.
MS. FYFFE: I'm Kathleen Fyffe. I'm with the Health Insurance Association of America, and I sit on the National Committee, as well as this work group.
DR. FITZMAURICE: I'm Michael Fitzmaurice. I'm senior science advisor for information technology to the Agency for Health Care Policy and Research, and with Bob Mayes I'm co-lead staff on the computer-based patient record work group.
MR. GARVIE: Jim Garvie. I'm the deputy CIO for the Indian Health Service, and I'm staff to the work group.
DR. FERRANS: Dr. Richard Ferrans with LSU Medical Center, and also with the VA, and I'm staff to the work group.
DR. YASNOFF: I'm Bill Yasnoff. I'm associate director for science of the Public Health Practice Program Office at the Centers for Disease Control and Prevention, and I'm staff to the work group.
DR. BRAITHWAITE: I'm Bill Braithwaite at the Department of Health and Human Services, and I'm staff to the work group.
MR. MAYES: Bob Mayes, Health Care Financing Administration, staff.
MR. GELLMAN: Bob Gellman. I'm a privacy and information policy consultant here in Washington, and a member of the committee.
DR. STEAD: I'm Bill Stead. I'm associate vice chancellor of health affairs at Vanderbilt University Medical Center. I am editor of the Journal of American Medical Informatics Association. I'm here to testify.
DR. KORPMAN: I'm Ralph Korpman. I'm chairman and chief executive of Health Data Sciences, and chief scientists of Medaphis and Per-Se, professor at Loma Linda University. I'm here to testify for some of those people.
MR. WAEGEMANN: Peter Waegemann, Medical Records Institute, representing NCHISB, and the tact for ISO TC215.
MR. QUINN: I'm John Quinn. I'm a principal in Ernst & Young's health care practice, their Center for Emerging Health Care Technologies, and I'm technical chair of HL7, and I'm here to testify.
DR. GREBERMAN: Mel Greberman, FDA's representative to the group.
DR. COHN: And the audience, if you would.
[Introductions were made around the room.]
DR. COHN: Well, thank you all. Let me take a moment just to review the agenda for the work group and others. We are going to start out this morning with two panels, with obviously a break in between the two of them, and then have another two panels after lunch.
Tomorrow morning we will be having our final two panels, and then spending the afternoon tomorrow afternoon talking about next steps, what we what learned from these panels, and how they impact on the work plan of the work group.
I would also ask that all members of the staff and work group come tomorrow with their calendars, because we are going to need to schedule some hearings for next year. We're probably going to be looking at the March and May time frames, but we'll need everyone's input to come up with some dates before the end of the meeting tomorrow afternoon.
With that, Jeff, do you have any opening comments you want to make? Kathleen? No? Okay.
With that, let us begin the first panel. I want to thank you all for coming. As I think we have indicated in our letters to you, we were hoping that you would limit your comments to 10 minutes, plus or minus, so that we can have some time for questions and discussions afterwards.
Which one of you would like to testify first?
MR. WAEGEMANN: Thank you for the opportunity to provide input into the role the government can play with regard to uniform data standards for patient medical record information and electronic exchange of information. I'll be speaking for NCHISB, as well as for the Medical Records Institute.
Let me go right down to the questions. The first one is, how do you interpret the congressional instruction? I really see this as two parts. The first part is in regard to uniform data standards for patient medical record information, and should not be limited, from my perspective, to coding systems or to data standards.
It should include all standards in regard to medical records, particularly also: uniform information content standards, uniform structure and format standards, uniform information representation standards, uniform presentation standards, uniform data capture standards, uniform data processing standards, uniform information storage standards, and uniform data integrity standards.
The second part for the electronic exchange of information is probably moving toward electronic health records, and this goes into messaging.
In regard to the second question, what factors and issues are preventing or delaying the widespread implementation of uniform standards for patient medical record information? I would say that there are two categories of impediments. The first one is subject matter impediments. These are issues which need to be resolved. The first one is the inconsistencies in clinical documentation techniques, styles, and formats. This includes lack of sufficient uniformity in content of information. It also means lack of clinical specificity in regard to coding system terminologies, and so on.
The third one, which is quite often overlooked, is the adherence to the principles of documentation. If you look today at systems which are labeled as computer- based patient record systems or similar, they do in the majority, not comply with the principles of documentation. For instance, they do not have requirements stated by state laws and other organizations for medical documentation, such as signatures, data integrity, storage of information so it can not be changed, and so on.
The second impediment is lack of interoperability. Many organizations are working on that. I cannot resist to mention here that part of this is the bridging between various providers of health records. This has to go toward an personal identifier. I personally believe that the legislation in regard to confidentiality will not solve the problems with the unique patient identifier.
We have developed an alternative of a universal identifier, which provides more efforts to the patient, and more support and power to the patient and the provider. If you would like to hear more about that, I would be willing to come back and testify at a future point on that.
In regard to the second category, I think the impediments are really organizational. We do not have sufficient professional involvement at this point. Here I'm talking about the medical specialties and the medical professions who are not involved in the standards process at this point. There are some beginnings there, but the majority are not there.
The last one is a lack of national empowerment or legislation that enables electronic health records. We have to remember that a good number of states do not allow paperless electronic health records. What we really need, and what I recommend, as you will have many people having different opinions here, that this committee focuses on defining on the minimum requirements for electronic health records in regard to confidentiality, security of indication, data integrity, retention, and permanence.
Secondly, to override inconsistent state laws on electronic signatures, retention, permanence of patient records.
Thirdly, to formally recognize an organization such as NCHISB as the center for standards coordination.
In regard to the third question, is the private sector able to address these problems sufficiently? I would say that in regard to this there are many efforts, and have this in my document various standards organizations in regard to interoperability, in regard to clinical specificity, in regard to inconsistencies in format and content.
However, what we really need is to overcome these states laws I mentioned, and secondly, to have an introduction of some kind of a personal identification system which is acceptable to everyone. And thirdly, to create uniform document routes. These are the three major issues, which I think should be focused on.
Question number four, which standards related to patient medical record information and its electronic transmission would add the most value? Well, from our point of view it is the uniform documentation standard, and standards for interoperability. They would increase the quality of care, save money, and at the same time provide a platform for interoperability, which is badly needed, and accelerate the introduction of electronic health records.
In regard to the business and goals of our organization, we generally promote electronic patient records, and what I just have said is true for this question too.
The business case for more rapid standards development is really the increase in quality of care. The way of saving money, and really to make the health care system in the United States more efficient.
Number five, do I agree with the emphasis on the four areas listed? And I can summarize that I do no agree with any of them. I think that it is more important, rather than looking at messaging, where many standards organizations are working on that at the national and international level.
In regard to focus area two, we have to realize that rather than looking at data models and code sets, where many organizations are working on that, it is important that we here are looking at the opportunities we have at the legislative levels.
So therefore, I would recommend rather than focusing on the four areas mentioned, that future hearings be focused on national legislation that defines the minimum requirements for implementing paperless electronic health records. That legislation should replace various existing state legislation.
Secondly, to use the position to draw professional organizations who are not at this point part of the standards process into the standards process.
Thirdly, to give NCHISB a mandate to accelerate and coordinate the standards process for electronic health records.
Thank you.
DR. COHN: Dr. Korpman?
DR. KORPMAN: Good morning. We went through the titles. I'm speaking from the context of someone who has been electronic health record systems, whatever they are, for a long, long time. We have about 20 million patients running on our systems at the moment. We have 150,000 professional users -- doctors, nurses, whatever. So this is the school of hard knocks, or screw you graduation speech, or whatever you want to call it, based on what we have learned.
Simon heard me talk about this two years ago. I think this is the most important thing that this group will do. Many other things are interesting, but I really could care less about X12 transactions when all is said and done. They have nothing to do with patient care.
The whole point here is not how to send better bills. The whole point is not do we need electronic medical records. I'm perplexed when I go to institutions that say, we need a computerized record. I say, that's nice. Why? Well, everyone is getting one. Yeah, so? Well, it will screens. So the whole point here is to make patient care better; that's the only point. Now if you also make it cheaper, and all those other things, most of those things seem to work in lock step, and we seem to lose sight of that all the time.
This is a rare opportunity where there is enabling language to slide in and do something about it. There haven't been very many, as someone who has been up before groups like this for longer than I can count, and shame on us all if we don't take proper advantage of it.
I agree with almost everything -- in fact, with everything that Peter had to say here. First of all, how do I interpret the congressional instruction? There has now a wedge that has been opened to actually talk about the only important thing worth talking about, which is how you put informatics into what is clearly an information industry, and has clearly the worst set of information tools of any of the information industries, except maybe higher education.
But you see, in higher education a student usually doesn't learn as much as they should have, or you don't get another grant in health care, with improper use of information, you die. So the short-term consequences are clearly different.
As far as gross enablers, even the other parts of the HIPAA legislation all involve what you do with data which is derived at some fundamental sense from the upstream electronic record, or upstream some record. So even looking at the downstream transaction is still a garbage in-garbage out issue if you can't fix the upstream issues, and these are the upstream issues that we're finally talking about here.
So I think this is a broad tunnel through which as many trucks as possible should be driven while the door is open, because we don't get this opportunity very often. When it says uniform data standards, I think as Peter said, that should be construed in the broadest possible way. This is not let's have a coding scheme. People say, patient name is 20 characters and we'll use SNOMED, which we should anyway. That's a separate issue.
The issue here is there are the issues of integrity and accountability and auditability, and all of those things that pull through, that really represent the data standards you need to make this make sense. There is a whole stream of these, and I've got pages of them in this testimony, which you can read at your heart's content.
I actually tried to deliver this time that could be useful, and we'll come back to that useful as a working document, as opposed to useful as a testimony document. So we'll come back to what's in there.
Why haven't we gotten there? Almost everybody around this table has been working on the standards for a long, long, long, long time. They are better now than they were when we all started, but they are not very good. I'm always intrigued by those of you who would like to repudiate and forget about it, but there was an early picture at HL7 that had this puzzle seamlessly coming together, and ENY was here and HBO was here, and all these companies. It just made a perfectly square puzzle, and everything fit perfectly, voila, out emerged this computer-based patient record. Of course that was absolutely dead wrong.
Part of the impediments here are that every week there is a new announcement about something that will fix the problem, and everybody gets on that bandwagon. It is always the wrong bandwagon. Not that any of these things are bad. They all serve a useful purpose, it's just the wrong purpose.
We come down to what this committee can do that uniquely needs to be done, which Peter certainly alluded to. I think that's the area in which it can be of most use. So we have a lot of efforts going on in a lot of small pieces. But sort of what we have been doing all this time is -- I have to give you analogy here.
I just watched the space shuttle get up. They are building the space station, and I think that's pretty exciting. But what we have been doing in health care, in the standards world for years is saying here is the idea guys. You go build the seat. You build the engines. You build the wings. You build the heat-proof tiles. I want you to build the best tiles, the best wings, the best seats.
Then when we are all done, we're going to bring these pieces, and we'll see if they happen to fit. Then we're going to glue this together with bubble gum and string, and then we'll see what happens when you take it off. Of course what happens is it doesn't take off.
The piece that nobody has been willing to take on is not what the best vocabulary is, or what the best interchange method is, or the best lab standards, or whatever. It's the piece to what does the space shuttle look like if you want to design an infrastructure that allows you to support the best patient care?
Those are the standards that no one has been willing to take. They are too big for any assembled commercial group to sit there and scratch their noggins over, and they are the only ones you really need.
Now I am not a great fan of the government putting its nosy face into everybody's private business, but there are some things that only the government can do. I think this is one of them. I have testified to this over on that side of the street as well, more than once. I am always loathe to do that, because people hear that and say, ah.
But what needs to be done, and I don't mean setting up pretty much the system and do this thing, because that will never happen and shouldn't. But the standards that say this is what the broadly contextualized electronic health records, data model, whatever you want to call it, needs to be, and how the pieces need to fall in order to comprise something that is HIPAA compliant is probably the single best opportunity this group will have to do something really useful.
That includes the issues of integrity and trust and accountability and auditability, sequencing and merging, and things that the other standards just don't even touch. You can glue a lot of systems together with HL7. We have 250 operating HL7 interfaces in the field, but they are always 10 percent of the interface. All of the things we have taken on have been least common denominator approaches to the world.
You take the five things upon which people agree, and then you announce you have a complete success. The trouble is it's the other 95 things that were important, and you have to take care of them, and you do. But you do that on a one up basis over and over and over again. Nobody profits, least of all patients.
So the private sector can do many things. I think we have made enormous strides in vocabulary standards. I think we have made enormous strides in messaging standards. I think many good things have been done. I think the important thing is now left to be done, and we laid out a host of things in the literature, which I won't review. What needs to be done? Standards for broad-based, comprehensive architectures.
If you are going to make a thing, here is what it needs to look like. Standards for end-to-end data flows. Protections for individual identifiable information. You are not going to get away from the trust factor issue. I realize it's not the subject of this discussion, but it's not going to ever go away.
Completeness -- completeness is the bane of every standards architecture that is out there. Yes, we can transmit stuff, but we don't know if anything ever got done. And if it got done, did the right guy see it? If the right guy saw it, did they do the right thing? So there is a whole host of these, and they are all known to people around this table. I'm just suggesting that we put them on the table and do something about them.
The business case is compelling. Nobody around this table needs to be convinced of the business case, although I've got several pages of business case points here, if you want to read them. This is an information business. At least 40 percent of the health care dollar is spent on information processing; not on computers, on people shoving little pieces of paper around, almost all of which become immediately useless after they are shoved.
We gather more data and use it less well than any other industry by orders of magnitude. It is an absolute embarrass -- except maybe higher education. I'm sorry, Bill just took over all of the higher education informatics(?) at Vanderbilt. We were having a sad discussion about this earlier.
So the four focus areas I think you've laid out are fine. I would pay them cursory attention, but I agree that a lot of people are working on those. The big thing that isn't being worked on is can we put in place a set of rules that say if you are going to build a space shuttle, if you are going to build a space station, it's being built in a lot of different pieces from a lot of different places, but they started out with a plan that says here is what the space station is.
Here is what each piece does, and here are the parameters that say this is the integrity, this is how they hook. Those are the standards that are most needed, and most lacking. So given our druthers, if you want to make a contribution, and I'm sure you do, knowing the people around this table, here is a contribution that the door has been open to be made, that no one has been able to make so far, that probably requires a kick in the behind to do, because it's been too difficult and too unwieldy for the private sector to do, that the private sector could run within once it was done.
Thanks.
DR. STEAD: Vanderbilt is a medical center that brings together schools of medicine and nursing, an integrated delivery system with a hospital, a variety of joint ventures, clinics, medical group, health plans that have Medicare/Medicaid and commercial populations. My responsibility for the operation/decision infrastructure for that complex gives me a very pragmatic view of the questions that this group is dealing with.
So from that view, as I look at the congressional language, it seems to me what we are discussing is how to improve the efficiency and effectiveness of the health care system. That's the fundamental purpose of the law. I think the language is actually quite interesting in that it directs the establishment of standards related to administrative data. And it directs of the study of issues related to adoption of uniform data standards relative to patient records.
So I share the feeling of other people here that you have a fairly broad charge, because "study" is a broad term. I really think you should begin by asking yourself a simple question. Will a concentrated effort by government and private sector on the development of uniform data standards for patient medical information have a reasonable chance of translating into desired increases in health care quality, access, and affordability in the near or immediate term?
Only if the answer to that overarching impact question, if you will, is yes, does it make sense to focus just on the data standard implementation issues that are raised in the working group's four focused areas. If the answer to that question is no, then the committee should conduct a broader inquiry to identify the set of initiatives related to electronic patient records that can be completed on a widespread basis, and have the desired effect.
Question two, in terms of barriers. I've been working on structured computer-based records for 30 years. I believe in them, but don't misread me. For the long haul, I support the vision of an electronic patient record, one that is recorded by the providers at the point of care, or by the patient or their family using uniform data standards.
Such standards, together with safeguards for privacy will facilitate access from different facilities, linkage to knowledge-bases for decision support, and aggregation to document cause and effect of clinical outcomes. Work on these standards should be a high priority, but there are a number of reasons why I feel that the work will not be sufficient to have a significant impact on the health care system in the short-term.
Most successful standard setting efforts have involved physical systems. Physical systems are well defined, and they can be reproduced exactly. Quality can be assured by building one factory that does the same process over and over again. A physical system can often be characterized by a single equation, and an instantiation of that system may be specified by a few data points.
Unlike physical systems, biological systems are inherently variable. They are compartmentalized. It's the only way we can allow change between high and low energy states, with minimum generation or consumption of heat so we can stay alive. Each person has many factories, each with its own copy of genetic material; material that mutates and evolves randomly. Because of this variability, the number of formulas and data points required to document a biological system increase several fold over anything we have experienced in the physical world.
I think this variability has several implications for what you are thinking about. First, the number of conditions that would need to be handled by uniform data standards for patient records are several magnitudes greater than that covered by non- medical standards. We simply do not yet have the tools that would be needed to manage such a large base of concepts.
Second, each records requires significant data entry. Data must be captured at the source, using a non-ambiguous data standard if it is to be mapped to other standards or exchanged between systems. Our existing data capture techniques are not rapid or accurate enough to permit widespread deployment into practice environments that are already overburdened by demands for increased patient volume and response to administrative requirements.
So what's the role of government given these realities? I think there are several areas where the government action is essential to moving the country, both in the short-term and toward our long-term vision for the electronic record. First, we need to continue to support basic medical informatics research. Key areas include: structures for representing data and information that make relationship between terms and concepts explicit; methods for data mining and limiting retrieval to context; work flow; change management; and human-computer interface issues.
Second, we need to provide incentives for the use of electronic records and practice, such as requiring that the clinical reports for billable services be accessible electronically in an open format; and such as requiring practices to show how they incorporate best practice or evidence in the practice.
Third, we need to provide support for demonstration projects such as case studies to demonstrate effective implementation strategies, or to develop metrics that can be used to bench mark and compare strategies.
Fourth, I think we should come up with a way of certifying a product's functionality regarding the use of data exchange standards; its ability to export patient data into non-proprietary, open formats without loss of information; provision of full function through application programming interfaces; and linkage of external information resources for operation and decision support.
Fifth, we should look at providing a standard way of identifying people, and the facilities that give or receive care, and the products that are used in clinical evaluation of care.
Sixth, we should require government or commercial payers to make contract terms and authorizations available in a standard way as profiles that can be easily referenced by practice management and billing systems.
Which standards then will make the most difference? In the near-term I would focus upon standards that allow any report or block of clinical content to be treated as a package. The key is a header that identifies the patient, provider, facility, type of report, et cetera. The content behind the header should be in an open format so that a human can determine its content with a viewer, and so that it can be processed through data mining.
An example of this type of standard is provided by a paper by Dean Bidgood et al, that will be appearing in the January 1999 issue of JAMIA, outlining the framework for representing image acquisition content for images that are themselves represented within the DICOM standard.
In parallel, I would promote development in three areas that are essential to the long-term vision for the electronic patient record. First, I would promote development of tools such as interlinguas and other data modeling techniques following the lead of the National Library of Medicine's Unified Medical Language System project. These tools will be needed to manage content rich standards.
Second, I would promote development of standard software architectures such as CORBA, to support services as contrasted to a data view of the problem.
Third, I would promote use of emerging E-commerce technologies to support patient (as contrasted to facility) ownership of the record.
In terms of commenting on your four focus areas. For the short haul, I question whether we should place development of an electronic patient record that is coded according to a uniform data standard in the critical path to evidence-based medicine and disease management. At Vanderbilt, we have been able to obtain 40-50 percent reductions in the cost of procedures, while documenting improvement in patient satisfaction by helping care teams work with data to identify where there is variance in practice, and to then look at specific cases in the literature to agree on best practice.
Our severity adjusted charge per case for Medicare patients based on a statistical sampling of DRGs is now below regional, non-academic market, despite inclusion in our charge structure of the costs of providing over 40 services that are not provided anywhere else in the region. This change in practice was accomplished with data from the billing system alone.
In a similar vein, we have used brute force to approach the problem of providing access to patient care information for direct care. Any computer readable text, whether generated by a system such as the lab system, or by specialty clinical applications such as a pre-op work-up application, or by simple transcription is imported into an indexed text repository that uses massively parallel computation to construct patient records on the fly from a retrieval mass.
The information can then be accessed by a browser over our secure intranet without querying proprietary systems. With this technique, we were able to implement a computer-based record for our inpatient and outpatient complex in less than a year, for less than $500,000.
So in closing, I suggest that the working group direct recommendations to areas that will let the country reap the benefits of this type of low hanging fruit in the near-term, in addition to making recommendations for efforts that should be carried out in parallel to lead toward our long-term vision for electronic patient records.
Thank you.
DR. COHN: Thank you. Will we have a copy of your testimony available?
DR. STEAD: Yes, I have it on disk.
DR. COHN: Thank you. John Quinn.
MR. QUINN: In relationship to question one, how do you interpret the congressional instruction? I interpret the congressional instruction as a general request by Congress to the secretary of Health and Human Services to provide information and advice to Congress. I interpret that Congress is somewhat frustrated by the efficiencies of computer science, when looking at the U.S. health care industry, are not being applied to reduce overhead, and increase the access for caregivers to the medical record.
Congress is asking the secretary for recommendations on what, if anything, it could or should do to move the U.S. health care industry to make better use of computer technologies.
Question two, preventing or delaying the development, what are the factors? As many of you are aware, this U.S. health care industry relies heavily on a set of software and systems integration vendors that are typically referred to as health information systems vendors or HIS vendors. In fact, the health care industry segment makes overwhelming use of packaged solution sets from these vendors to meet specific functional business and clinical automation needs.
The health care HIS market has self-selected the demarcation lines for functionality between these packaged solution sets. There is no consistent grand design for the specific solution sets that are offered. It is instead a clear case of vendors producing what the customers have asked for or require.
Historically, vendors first found a market for business automation systems that streamlined the ability of a provider organization, for instance, hospital physician group practices, home health agencies, et cetera to bill for and receive reimbursement for services. The use and need of these billing and practice management systems accelerated dramatically during the seventies as the federal government, and specifically the Health Care Finance Administration introduced DRGs and prospective payments, et cetera, and the other payers similarly increased the complexity of what was needed to get paid.
Recent developments over the last 10 or so years in this area has produced an environment where one could safely say that it is probably not a viable economic proposition for anyone to attempt to produce a health care insurance claim for an acute care episode without a computer system. Parenthetically, that is actually one of the bigger concerns about Y2K. Is there a manual system and one could fall back on at this point?
At the same time, clinical automation systems that assisted in the communication of information, for example, order for diagnostic tests, and the storage of information, for example, the record keeping associated with those same orders, and the diagnostic results also took hold. Both types of systems, business and clinical, have their own peculiarities that are tightly related to the business, science, and art of health care that make them significantly different than similar analogous software applications in other industries, for example, invoices for manufacturing orders, et cetera.
This can best be understood by looking at the collection of vendors that support health care information system needs. Functional areas that have little to no distinction have a significant overlap with other U.S. vertical businesses. Examples of these include general ledger, accounts payable, human resources, and payroll. Current vendors such as Peoplesoft, Lawson, and SAP, and many others have a significant presence in all industries, including health care.
Functional areas such as patient billing, diagnostic services such as the pathology lab, radiology, et cetera, and the entire area of the electronic medical record have few, if any, parallels in other industries, and there is no significant market penetration of these software solutions into other industries.
In short, the U.S. computer software applications industry has not been able to leverage its successes from other industries to the health care industry. Among the reasons for these include:
(1) The health care industry, as mentioned above, has application needs that are unique, and do not allow for easy reuse of applications from other industries.
(2) The vendors in this industry are significantly challenged by a significant lack of standard processes in the health care industry. All packaged software vendors profit from a business formula that says built it very well once, at significant expense, and then sell a huge number of copies, for example, office product software such as word processors and spread sheets. The initial applications development is a significant capital investment. The reproduction costs should be a fraction of a percent of that initial investment.
Health care has few standard processes among buying institutions, and even fewer between physicians who rely on personal knowledge, experience, and in some cases, intuition to determine what to do, and how to do it. I don't want to overdramatize this point, but I do need to point out that any vendor who produces health care industry software faces the variability that applies to this industry alone.
Clinical health care data is extremely complex, and the challenge is great for anyone trying to build a database to hold clinical information. More importantly, clinical information needs to be updated and retrieved quickly in order to not slow down the physician. As a high level example, the HL7 reference model has about 830 complex data items or attributes at last count. The similar model produced by some of the same modelers for X12N contains about 160 of the same data items. There is clear overlap in some of these items, but even if there was total overlap, which there is not, HL7 has about four times as much data as the non-X12, or primarily demographic and administrative portion of the model.
The fact that a significant part of this information is codified adds another burden to the database designer, who must come up with a database design or schema for this information, while managing and mapping the time varying coding sets which are most likely different sets for each participating institution, even when they share the same information systems.
(4) Technology is complex, and changes rapidly. The HIS vendors require a significant amount of time to design and implement solutions for processes as complex as medical records. The information industry has recently gone through an explosion of new technologies over the last 15 or so years, that has been driven by a huge expansion of the general computer market to the consumer (that is, read PC here).
In short, the technology industry is based on relatively simple consumer applications. The HIS vendor community has not been able to capitalize on these changes, and get them to market before the next wave of changes hits. This next wave causes buyers to stop and say for example, let me wait for the thin client version of this application, a possibility that may be several years away from delivery in everyday use.
(5) The technologies that HIS vendors use are general information industry technologies, for example, relational database systems, programming tools, et cetera. There is insufficient evidence that some of these technologies can support the scale of application deployment that a ubiquitous medical record will require. In fact, one can arguably state that the largest single commercial available electronic medical record system today, which actually has some people here from that institution, is only supporting about 4,000 devices at one of Kaiser Permanente's smaller regional groups, Kaiser Northwest.
The largest custom built academic electronic medical record system that I am aware of is supporting about 20,000 devices at Partners Health Care in Boston. The message here is in spite of any claims, no one vendor has demonstrate the technology and the ability to support the electronic medical record at a large health care provider, or a truly large group practice. This must be done before we can consider the technology for the amalgamation of medical records across a region, state, or the country as a whole.
Question three, is the private sector able to address these problems satisfactorily, and the role of government. First of all, the private sector is probably doing as well as it can. Given enough time and patience, it will eventually get there, but probably not in my natural lifetime, at the rate things are going, and I'm 49.
The role of government then should be to increase the rate. Maybe I'm being selfish here, but I figure sooner or later I'm going to need this more than I needed it in the past. The role of the government then should be to increase this rate. In the past, this has been by primarily funding academic research and large government projects, for example, the JCBR(?) project, in fact the ability of information technology to meet health care's needs must be seriously assessed, and incentives should be given to the development of commercial systems that can be implemented across the industry instead of just one situation at an institution.
Finally, the government needs to come to grips with the contradictions of standards. Standards are only good if everyone is required to adhere to them. For example, DRGs are standard because HCFA won't reimburse you unless you use them correctly, and there is little, if any consensus that there is an open group for development there.
Open consensus standards take a long time to create. It's the nature of having consensus-driven standards. If this is not serving the country's needs, then another less consensus-driven process should be endorsed.
Question four, which standards related to patient medical record information and electronic transmission would add value, et cetera? First of all, use standards that are accepted and work. The HIS vendors have, for the most part, widely adopted HL7, X12, and DICOM. Where these messaging standards meet the domains' needs, they should be used and required.
In addition, however, the codes for use within these standards needs also to be standardized, the process at least semi-standardized, whether there is no demonstrated need for process variation, and a process set up to fairly review and update those areas that reflect emerging areas of knowledge through academic review and periodic publication from mandatory when applicable use.
My organization consults to providers, payers, pharmaceutical companies, medical equipment, and supply manufacturers. This pretty much covers the health care industry. Someone asked the question does it meet my organization or my customer's needs, what is good for the industry helps my organization, and my clients. Without more standards development and implementation, we are not going to enable the common and inexpensive electronic exchange of health information.
Do I agree on the focus areas? Yes, I do.
Thank you.
DR. COHN: Questions from the work group?
MR. BLAIR: I have a question at this point for both Peter Waegemann and for Ralph Korpman. Peter, you wound up indicating that we need to give greater focus especially to data capture, data quality, and documentation standards. Could you elaborate a little bit on the kind of standards that you feel the government can constructively facilitate or encourage, and why our role might be needed in that area?
DR. KORPMAN: Peter is going to fob this off to me for a second, so I can talk about something I forgot to mention.
I mentioned this, but starting on page 21 of my testimony, I've got 20 pages of things that we have learned in our site -- some of which have 7,000 devices -- that we have learned need to be handled. They fall into broad categories like characterizing what represents the medical/legal record. There are broad discrepancies from state to state in terms of what you have to do to characterize that.
What are the points of authority and reference? And this gets into the trust issues as well, which have to be dealt with at some point, whether people like that subject area or not.
The stewardship issues; who has to look after the data, and how do you have to look after it, and how do you chronicle the looking after the data?
The wholeness and integration issues; you have to put together a record which represents what happened to this patient. There are completeness issues. There are time line issues that are unique to health care that have to be tracked. These have all become intersystem issues. They are the ones that really vex you when you are trying to put things together and make them work.
There are a whole series of accountability issues; who did what, when they did it. What was backed out? What data did they have at the time they did it, as opposed to some other time when they did it?
There are all the typical security issues. There is a whole set of scalability issues. It has to work for more than one person in one place, and John alluded to some of these. These are serious technical issues where if you define the standards, you can then figure out what technologies to apply.
There are availability and reliability issues. Can you have a legitimate record that goes down? These are serious issues that somebody should say no, you can't.
MR. BLAIR: Or else? There is only one word that you have referred to that I didn't have a real understanding of what you were talking about, which was trust. I'm not sure how to separate that out from authentication or data security. Tell me what you are thinking of there.
DR. KORPMAN: I'm not sure it separates. It's the issue of how good is the data? How can you be sure that no one else is getting it? I've got a whole table in here of what makes sense. It's on the floppy disk. You can look at it later. The issue is unless people really believe this record, and believe it is right, and believe it is complete, they will always be scribbling notes on little pieces of paper, which defeats the whole purpose.
We have gone through institutions just taking the paper away from people, and saying, no, no, stop. You can't use that. Somebody had to do it every day for a month until they finally stop, because you have to really believe in it.
We've had people who put something in the paper and then immediately race around the corner and try another terminal and see if it's there. There are a whole set of trust issues that have to be there for practitioners -- they know their patient's life depends on it. There are persistence and permanence issues that are very significant.
I'm just giving you the broad brush areas that really haven't been dealt with in the standards area, that have to each be dealt with if we are going to deal with electronic records intelligently, because solving little pieces of this puzzle will never get the shuttle into space.
John mentioned process issues. There is a whole set of process integrity issues in terms of integrated scheduling and integrated lists, and integrated this that are absolutely fundamental to delivering health care. That's how health care is delivered; most of them are ignored.
There are knowledge base issues, there are personalization issues, localization issues, at least how you contextualize that, definition issue. Anyway, there are 20 pages in here. I'm not going to go through them all, but what I wanted to do was say these are the things we have learned. We have to deal with all of these to make this work. You can't say, I'll just forget about persistence. Ninety days after discharge it will disappear, but don't worry about it. It will be printed out. Those things just don't work.
We'll guarantee this will never go down. That's our premise. It won't go down. You'll always have it. If you don't always have it, you print everything out every shift just in case it will go down, and then you kill the rest of the trees in the United States and you have no record.
So there is a whole set of issues, both technical, personal, physical that need to be dealt with. They are not the here's the name, followed by the number, followed by the address. I think we have made good progress in a lot of those. They are the broader issues in terms of what it takes to put the pieces together so that you can improve patient care, as opposed to what it takes to put the pieces together so you can make a system.
I think Peter was alluding to basically exactly the same stack of things that are broader than the things that we've been dealing.
MR. WAEGEMANN: To summarize, I really believe that it is easy for this group to focus on standards groups and what can be improved there and so on. We have to keep in mind that about 5,000 people are involved in various standards manifestations(?). Also, it's not perfect, but we have to realize that we are just at the beginning of the third generation in standards.
The first generation was before HL7 was formed, were things were really slow. Then we really had about 10 years ago, the second generation. We are now getting to the end of it, where messaging and all of this is coming. People are looking for interoperability or plug and play. The next generation is really focused by, for instance HL7 coming out with 3.0 and object oriented approaches and so on.
So what we really have to see is that many people are working at standards, but where are the areas where we need help? And we is the standards community. It is really in regard to creating a legal system for a paperless electronic health record, which has the following features. First of all, data integrity. There are many issues that information is being sent, that information is being stored without any change. Secondly, we can have data integrity authentication. That we, at any time, have clearly an accepted minimum requirement for authentication for any electronic health record system.
The same applies to storage. Information is being stored on computer systems. Do we know that this information is available after 30 years? Do we know that information is available after -- if you take just a minor plus 30 years or so on, what requirements are there for the average provider to make sure that such information is stored properly?
These are just tiny examples of a whole range of minimum requirements which need to be set forward in order to have national legislation which empowers providers and industries at the same time to create and implement electronic health records. That's where the standards community needs help, because this cannot be done in a voluntary standards process. It needs to be done in a way to overcome the state laws, and overcome the various existing laws. That goes into processing.
We could go into all the details. I have quite a bit in my paper, but it's in general, those areas that I would like to focus on.
DR. COHN: Peter, thank you for that comment. I'm actually wondering if -- I think you and Ralph are both focusing on some of the same sets of issues in terms of the focus of what needs to happen. I'm curious on what Bill Stead and John Quinn have to say about these as high level items that the work group should be focusing on. Can you both comment?
DR. STEAD: My concern is that we, as a group, try to come up with 3-5 things, that if we do today, five years from now will have made a palpable difference across the health system, and that Congress won't be frustrated, because whatever we do has not resulted in change. Everything that anybody here has said I resonant with as being part of the problem. My concern is given the variability, where do we focus?
I think in areas such as authentication, we shouldn't have to solve -- if the basic computer science industry cannot solve that for the health industry, we're licked. We have flunked your process of being able to build it once and reuse it.
So I think we've really got to figure out what are the few things we can do that are not stymied by the variability, and that let the variability continue under the covers or within the wrapping of those few things we can do. I simply would argue that if we could get every container of clinical information in an open format, labeled with the right set of meta-data, that we would have made a giant step forward, and I think we could do that across the country within three years if we decided to do it.
MR. BLAIR: Could you explain a little bit what you mean by an open format?
DR. STEAD: ASCI text is an open format. The DICOM standard is in large part an open format.
MR. QUINN: I guess I would have to fall back on two issues. One, you are asking for how can the government make something happen quickly, and two, I'm also looking at a set of problems that are largely systemic to both the health care industry, and another set of problems that are how well does the information technology industry apply to health care.
I think if you really want to demonstrate value, you better focus on a couple of narrow things that you can do, that Congress could do, to produce a quick return, at the same time of looking at you've got some really broader things that need to be addressed. This issue of process variability, the ability of not being able to leverage the software business model into health care is a huge barrier.
Without getting too particular in the names, I would love to have a dollar for every time I have met someone from outside of health care who is extremely smart, probably made millions of dollars in a start-up somewhere, tens of hundreds of millions -- I've seen them all up into the billions -- who have said, we're smart. We know how to fix health care. We've done it in e-commerce. We've done it in manufacturing. We have done it with Internet browsers. Whatever it is, we've done it, and we can fix health care.
So they come in, and they just grind themselves to a pulp. Now they are wealthy enough that they just walk away from it, and their egos are big enough that they just basically never admit that they failed. But there are no examples, to the best of my knowledge, of people that have succeeded in doing that. So that should be a good indicator that there is a problem here.
That these really smart people that go out and form these really large, successful companies in the general information technology industry, when they see health care as the great unused, untilled field that it is when they look at the percentage of gross domestic product that is spent on health care, and how little information technology is used, and what small percentage of the gross revenue of a hospital is spent on information systems.
They look at these numbers and they say, wow, we can make a huge difference. There is a lot of money to be made here. And they all fail. So there is something systemically wrong. It doesn't meet the model. Work on that as a long-term goal as to how do we align that.
MS. FYFFE: Thank you very much. Your presentations were all very informative.
Dr. Stead, I really appreciated your comments about the human-computer interfaces. In my deep, dark past I installed computers into various hospitals, and can take the exact same system in software and install it in one organization and it works very well, and you go to another organization and it is an absolute disaster. So I think that's a lesson for all of us.
Moving ahead a couple of years, if you were to be faxed some proposed legislation on the computerized patient record or standards for such, what would make you the most nervous in terms of pernicious provisions in such a piece of legislation?
DR. STEAD: Well, I guess a couple of things. First is legislation that doesn't reflect an understanding of requirements that it may impose on work flow. A lot of our trouble with the human-computer interface stems from the fact that we have systems that are designed to do a task, and we have got people who have roles. Those roles include doing several tasks.
We need a way of interacting with the electronic infrastructure in a way that supports our roles, and also gets our tasks done. That's one of the things that has not been well worked out with the software industry, and therefore if there are requirements that are put in front of us that will not fit with our work flow, we in end, created a work around that makes us sink.
The second thing that would onerous to me would be a requirement without the information base or standards to back them up. When a payor requires that we get patient okay to provide a service that the payor is not going to pay for, and they don't give us an electronic form, what they are going to pay for or not, they have effectively created a mandate that we almost cannot meet. We just all have been recently living through this.
So those I think are the two ends of the spectrum that would worry me the most. I want to package our variability, so I tend to live at the two ends of the spectrum. Let the variability exist, but give it a very standard package. Most people tend to want to try to get to where there is a standard that will work across the variability, and that would worry me.
MS. FYFFE: Dr. Korpman, your company's software is currently being used for about 20 million patients. What would it take in legislation to increase that number three-fold? What would you like to see?
DR. KORPMAN: It's interesting who uses the kinds of software we have, which is pretty complete, and we standardize the processes, even though we let each organization standardize their own. Bill is right, and John is right, there is enormous variability. What we try and do is make it look variable, and eventually reduce it to a standard set of things internally that make it work, and we've been pretty successful at that.
So we do for instance, all the public health care in New York City, all 16 hospitals and 100 clinics, and all that stuff. They all look very different, but they all congeal into exactly one database that if you show up in the Bronx or you show up in Coney Island, it doesn't make any difference. It knows who you are. It knows exactly what is going on with you.
We do little 200 bed community hospitals out in the middle of nowhere that just were inspired to do this right, and there are 100 doctors in town, and that's all there are, and they are lashed together. It doesn't matter where you show up in this little town, it will all work. We do giant HMOs.
So you look at who purchases these kinds of things. They are basically people that meet two criteria. One is they eventually figured out it wasn't about labs, x- rays, pharmacies, bills, X12s or whatever. It was about patients. You would be surprised who few people get to that point at an administrative level. They talk about everything else -- doctors, nurses -- anything but patients, because it is so hard the way the health care system is put together.
There is no person who says, okay, everyone will do this. This is the main reason, like John said, why it doesn't work like other industries. The CEO could say whatever he or she likes, it doesn't make any difference. The doctors will do what they want. The nurses go on strike. It goes on and on, and people just ignore the rules.
In a private company or government, the president says you do this, you do this, within reason. Let's put it this way, in health care there's no pretense it works that way. In most organizations there is pretense at least that it works that way. If you want it to work that way badly enough, it will work that way. The CEO of an organization realizes that most of the time he or she says something, it sort of goes off where it goes. But if you really want to make sure it happens, you'll make sure it happens.
MS. FYFFE: Yes, there are consequences.
DR. KORPMAN: Right. So you have to deal with that. So we look at who buys, and it's the people who have gotten over that issue, and it's the people who have gotten together enough internal will to get over the change process, which in health care is really horrible, for exactly the same reason.
So that's why I said the government has a limited role to play. I'm not a great fan of government intervention, because it comes with so many strings attached that usually it ends up hurting, rather than helping. But it can play a role, and I thought the DRGs were a good example. Say look, we don't care how you managed to this point. What you do must meet the following criteria.
So that you could put enough people on the same page, and say, okay, I can't just go randomly buying systems and hope I end up with something and go on. What I buy will now have to authenticate and have the following data integrity, and the following process allowability, whether it's packaging. It's a whole second level order as to what it looks like.
But right now, people really need a kick in the rear to look beyond the lab screen to get into a lab system issue, which is where most of them are. Or my only important thing is to get bills out, and so I'll do whatever it takes to get bills out, which is difficult enough. Have a nice life.
What you want to do is establish enough infrastructure that says okay, you must do this. We don't care how you do it. Find 50 different solutions. That's the American way, and it works very well. But to put that amount of homogeneity so that the CEO or whatever can say, I don't care if you guys want to listen to me or not, unless we do this, we won't get paid, and that's the only big shoe there is to drop.
No one would be using DRGs if you didn't have to use them. If you could come up with things that say this is it, you must do these things. That would put people on the page I see. I need to get a system that is oriented to how I care for a patient, and that assures quality of data, and assures that it is useful for care. And now have to look for something different, and all these people are screaming that they won't pay attention, I'll just show them that they won't get paid for this. That's about the only way you congeal everybody.
It can be done. The government does it from time to time very successfully. But unless it is done, for our 20 million, which we have gathered over 15 years -- it's not like they all raced up -- it's just the onesies, twosies that have eventually reached this realization themselves. What you really want to figure out is how can you write some law that speeds it up, so that within our lifetime everybody is getting on closer to the same page? How you implement, where you get -- a different issue. But just to change that focus would be of enormous value.
MR. BLAIR: As all of you are well aware of the fact that I'm blind guy. So I am not able to go through the detailed handouts that you have given me, and I wind going through the process and synthesizing things. I want to know if I am getting the right message from your testimony, so I'm going to pull this stuff together, and then I'd like each of you to comment, tell me where I'm off base, because I'm going to try to simplify this stuff.
This is kind of what I think I'm hearing. I believe that I'm hearing from all of you that for the most part the private sector, while it is imperfect, and it is not moving as quickly as we would like, is making good progress on message format standards, on pulling together medical terminology, and working closer on a number of current standards activities. The message that I think I'm hearing is any role the government should play should support, facilitate, be a catalyst for that.
The other thing that I think I'm hearing is that there are broader issues than just this more narrow focus on electronic health records, computer-based patient records, whatever term we use. That we need to look at it more broadly in terms of the health care information infrastructure. There are many aspects when we broaden that, the infrastructure elements, that do need more attention.
Peter Waegemann pointed out the fact that there are different state laws that impact our ability to wind up having a single solution across the nation for authentication, for electronic signatures, for retention of records, permanence of records.
I'm hearing from Ralph Korpman that he is indicating there is a whole array of issues in terms of accountability, in terms of trust, data integrity that the standards community hasn't been able to focus on as intensively, and if the government can play a role from a broader infrastructure aspect, that that would help us.
I believe Bill Stead, that you mentioned that tools to be helpful to be able to assist with medical terminology, however, I also heard Dr. Stead that you cautioned us if we focus too much or only on health care information infrastructures and facilitating that, we might come up -- not that we shouldn't do that -- but that if we only focus on those areas, those are things that will take us a long time to show results.
I think you are cautioning us that from a political standpoint, we probably also ought to be doing things to encourage the existing standards efforts, to get them broadly accepted and implemented.
Tell me, on each of you, where did this summary miss the mark?
MR. WAEGEMANN: I think it is correct, except I was trying to say a little bit more. I really fear that it is not a question of what standards to pick for instance for authentication. I really fear if Kathleen Fyffe would have asked me the question, what is needed to get the vendors to be more successful, if I look around at the 289 vendors we have on our database at the Medical Records Institute, Ralph's company is one of very few companies where the use is almost down to no paper, or have very little paper.
I think the one issue we have to address is how can we make it possible that there is a transition from the dual paper-based and hardly electronic record to a fully electronic record. So I am saying where we may need this at this point is to focus on what minimum standards can be put into what I call an abling law, which will say any clinical hospital can have paperless electronic health records if they comply with certain standards.
I don't think that these standards need to be created or enhanced. I think we do have them. I think it is a question of picking them, and putting them into legislation, and enabling our providers and vendors to do the major step which will save tremendous money, and will make the whole system much more effective. Therefore, this really is the main emphasis I have.
MR. BLAIR: Bill Stead, where did my maybe overly brief summary miss the mark?
DR. STEAD: First, I think your summary of all of it was amazingly on the mark. I would sort state what I said a little bit differently. I would say that first all of have said that the health industry is variability in how it does business. I tried to make the point that part of the reason for that is because of the fact that people are variable.
So when people from outside of health come into health and get in trouble, it's because they don't understand the difference between a biological and a physical system. It is not just that we are all crazy in how we do business. It's partly that, but people are inherently variable, and we're talking about how we get an information resource that let's us manage the complex interaction between genetics, the environment, the brain, and habits. That is the ultimate set of variability.
Therefore, I do think politically I feel very strongly that we've got a chance. As everybody said here, a door has been opened. If we let this door close with something that would be great in paper, and maybe even make the difference for 20 years from now, but doesn't deliver a change in 3-5 years, I think we are going to have a very hard time getting sustainable support to do what we are going to have to do over a 10 or 15 year period, to have an optimal way of dealing with this variability.
So I want a political win. I want one very badly; you're right about that. I think the way you get that is to instead of mandating things that require more overhead and more systems, that we actually pick a few processes that we can show have worked, and that we help health organizations use those processes to use the data that we have already got in our systems to drive out the variability in process that we don't need.
We can get rid of variability in process if we can handle variability in humans. That's the tightrope that we have got to walk, if that helps at all.
MR. BLAIR: Thank you. Ralph?
DR. KORPMAN: Actually just to follow-up on Bill's comment, I think one of the most interesting future sessions of this committee might be a discussion as to what kinds of standards -- because I have laid out pages here of interesting things that Bill would like to reduce it to carrying packets around, which is an interesting concept.
I think it would be interesting to discuss what things could be rolled out that could be effective within a 3-5 year period. It think that's too much to discuss here in the next couple of days, but I think that topic by itself brooks major interest.
I think a political win here, because this is a decade long process. My great fear is that if we go for the easy win and nothing else now, that's all we'll ever get, because it will be 20 years before someone opens a hole again. We need more infrastructural boot in rear than that.
But I think the major shares of what I was talking about are compatible with a fairly short kind of thing. I think investigating how one sequences that stuff would be an interesting future endeavor of this group, because I think that would allow you to get the win, and also do the right thing.
Other than that, I thought your summary was pretty good. I think the compelling thing is to put enough things in place. Sometimes people have to be told, you will do this, especially in health care, where they typically won't do it, whatever it is. That's where you see an effective role of government. When you say you will do this, and it doesn't have to be the world's onerous thing.
These are things that are important to patient care that you will do. So somebody can stand there and say we must do this, because the government said so, I think is an important part. Coming up with that list of things, and it doesn't have to be very long or very big to be very important and very effective, I think is probably the most useful thing that this group can do. It's the things that people won't motivate themselves to do.
MR. BLAIR: John Quinn, what did my summary leave out?
MR. QUINN: Well, I think Bill hit the process variability issue. You can't leave here without understanding that compared to other industries, the process variability is a real inhibitor to enabling information systems technology in this industry.
Secondly, I would not walk away from here with the assumption that the information technology industry is capable, ready, and properly aligned to do this job in health care. You should examine what it is that is missing from the general information technology industry that becomes an inhibitor when you try to apply it to health care.
Some here are nodding, but your organization would probably amplify that, and be one of the best demonstrations of where that problem can be best exemplified when you get to the very large organizations.
Process variability will work itself out over time if you just want to wait long enough. As the provider organizations become larger and larger, they will impose their own standards of process upon their caregivers because of the economic efficiencies, but that is going to take a long time at the rate mergers are going on.
DR. COHN: Clem, you had a question?
DR. MC DONALD: Actually, there are two. I apologize for the airline not keeping their promise of when they are supposed to arrive.
It really has to do with sort of tension and a ying-yang thing between say the government force and kind of good. That is the democracy and dictatorship and all stories. I would like to get specific about what people think government maybe should do. Government hasn't been hesitant to put in regulations or strong things.
A recent one that is a lot of fun is the ENM regulations, which specifies exactly how we should do things and gets rid of the variability, but it forgets the fact that every patient is not -- the only thing it reflects is a sophomore medical student's view of how medicine is practiced, but the variability of a patient doesn't show up in it.
So should we be dealing at that level of specifying non-variability? Should we be specifying things -- and I'll bring in another discussion we have had internally about fields such as ethnicity and race being included in our registration records. Well, we can't get people in our hospital to record race, because it's felt to be an offensive thing to ask patients. That's a real simple one. Should we, the government sort of say, yes, you should collect that data?
MR. QUINN: I'm not sure I would dive and start picking -- if you want to see real value in ethnicity and race. Yes, those are politically charged --
DR. MC DONALD: Well, I didn't mean politically charged, but that's a simple one.
MR. QUINN: Think about something a little more systemic like what do we mean by an encounter, by a visit? Those kinds of things that all the vendors spend huge amounts of time every time they put a system in, trying to come to some agreement with the buying institution as to what do you mean by an encounter, or what do you mean by a visit?
DR. KORPMAN: Clem, if I might take it even a level further, because you have mixed two issues here. We all know on a public health basis you should have race. There are some things that affect some races more than other races, and it's important to know, but I'm not sure, as this committee has discovered when they got into privacy, confidentiality, and all those other issues, that those are issues that can be solved technically.
There are issues, high level issues that say your data must be end-to-end authenticated in the following way. Your data must have the following persistence. Your data must have the following auditability. Your data must have the following process connections. We don't care what process you use, but you have to be able to trace it from beginning to end.
There are high level things that you can put in place that would make a major definitional change to how people right now were randomly sorting their data, go out and do things with it. That's where I think the government role is. Right now, people aren't willing to tackle those issues for a whole set of reasons, mostly political.
But if someone stood there and said, oh no, we have to do this, I think it would make a big difference. So I'm with John on staying away from the let's discuss the race card, or whatever it is.
DR. MC DONALD: I wasn't meaning to push that. The question is, can we get that level of detail?
DR. KORPMAN: No, I think you should be saying there must be end-to- end assurance. For instance, we're dealing with a lot mediators today or a lot of transliterators. Well, how do you know that what went in is what came out? Those are real important issues, and people don't trust system where you put it in as A and it comes out as B, and you're counting on the system to make sure that A is B is A, because sometimes it isn't.
The worst case is non-lawsy(?) stuff, but the picture guys understand that well. In the data stuff there is usually not a lot of understanding until you see something come out that wasn't what you put in. Those are the kinds of things. They don't have to be very specific, but they can be very focused and very important.
DR. STEAD: I think what we need to do is we need to come up with a way of identifying who a piece of information is about. The identity problem is something that we've got to solve at the government level. We don't necessarily have to do it with an identifier, but we've got to do it somehow.
Second, we've got to make sure that any container of clinical information is open, accessible outside of the system, any system. Those are two things that we can do.
The third thing that we need to do is to focus on a way of organizing concepts. When you ask, should we get to that level of detail, and then John answers, let's get people to agree on what an encounter is. That's what scares me when I listened to Bill's presentation in AMIA(?) about where the other HIPAA standards were, because people are arguing fundamentally about imbedding business rules in the use of the data field. That's not scalable. It will, in the long run, not work.
Let's get agreement on the 20 things that could be an encounter. Let's figure out a non-ambiguous way of mapping the relationships of those, and let's hang that out in information as effectively a dictionary. Then you can link into those terms in any way you want to, and we have a way of disambiguating it.
It's like at Vanderbilt, people want to get agreement on what a faculty member is. They want a data field for what a faculty member is. Well, is it what the provost's office says? Or is it the clinical affiliate working through -- we need to know the attributes. We need to know the set of granular attributes that anybody can pick from to classify somebody as a faculty member. It's a totally different question.
If we could get that concept, I think we need the research to support that concept mapping. That's what I was saying. If we had that concept, we can then let people use it however they wish.
DR. COHN: Bill, can I just follow-up on that question? What I'm hearing is some feeling that there is a need for almost a single data model, or at least for some of these concepts, a high level way of saying this is the representation we will use. It may be the encounter or the 30 different kinds of encounters, but let's at least agree that there are 30 different kinds of encounters. Is that what I'm hearing from you?
DR. STEAD: What I'm saying, there needs to be a meta-data concept model. That's very different than a data model to me, because a concept model is -- basically, we've got to in the end get to where a term can only have one meaning, and where we have a way of linking together and externalizing explicitly the relationship between terms that only have one meaning, that gives you complex meanings.
That's the only way we are ever going to solve the information content problem. That's the kind of thing UMLSs tried to do. People have criticized UMLS because UMLS is not a good data recording tool. It was never meant to be a data recording tool. What it is meant to be is a way of showing the relationship between the various data recording tools. It's very useful for that.
DR. KORPMAN: Just to add for a second here, what you need to do -- I think if you are going to put the pieces together, because you are hearing some pieces here that actually fall together, there are concept representation issues. I don't think you should come up with a list of encounters, I agree with Bill, because everybody has them differently. We just have a data type called "encounter" and we let every organization establish as many as they can.
But once you have done that, you have to be able to say a whole host of things about that encounter. All data about that encounter is traceable to the originator. All data about that encounter is authenticated. All data about that encounter is end-to-end processed in the system. Those are all very important things, otherwise the encounter data itself becomes fairly useless, whether you have five encounters or 500.
So there are a series of things in terms of representations, in terms of what you do with the representations that are fairly high level, that don't require you to fill eight races or 25 encounter types or whatever, and still will make that data far more useful than it is now, where it is completely useful for practical purposes.
MR. WAEGEMANN: Let me respond to Clem's question, and summarize what I was trying to say at the beginning. I really think that we should not focus on detailed standard actions, particularly which are not really helping in general ways.
We should secondly, not really try to duplicate work which is going on within standards organization. If we look at terminology, we are moving towards some consensus there, and it might take some time. And I don't think to come in and cut the knot, and then really use the power of the government will help in any way.
I think, however, that there are three areas where the government can help. The first one is what I mentioned, enabling national legislation to replace the state laws. To really describe the minimum requirements for data integrity issues, and for principles of documentation, as we have them right now defined at the state level for paper records. This needs to be translated into the computer environment. It is not a question of creating standards. We have those standards. It just someone to pick them and say this is the national minimum requirement for computer-based patient record systems.
The second item, what can be done, is to empower some of these standards organizations in coordinating issues such as NCHISB.
The third one is really almost the most important one from my perspective. It is that the clout NCVHS has can bring the professional organizations to the table who have been missing so far in regard to some of the data content standards and so on.
These are the three issues I see, and I think I would recommend to focus on those issues, and not to get lost into details.
MS. FYFFE: Peter, what professional organizations?
MR. WAEGEMANN: I'm talking about the medical specialties. I'm about having the current work say by the cardiologists on data sets and so on, all of it reported and looked at. And bring them to the table as partners in creating data content standards.
DR. FITZMAURICE: In hearing you, I'm convinced of the complexity of the medical model, and the need to be able to describe it. My problem is, when I go home, I have to explain what I do to my wife and to my kids, and I'm not very good at it. Or to get down to a basic level of granularity, when I discuss with congressional staff what is the need for health care data standards, eyes glaze over. It takes something in your blood I think, to stay awake during a standards meeting.
So I want to ask the panelists, if I were to start at a very high level, and to describe the need for health care data standards, there has to be some kind of model that we can start hanging things from or building into. You start with the model, here is a patient. The patient is good. The hospital is good. The doctors' offices, and they get things, and we need to describe these things.
Is there a model in your mind or that has been on paper that you would recommend? Or would you recommend the committee to devote any time to look at modeling? I have devoted a lot of time to modeling. I don't understand it. There are all different kinds. There are data models, information models. There is this model and that model. I can't explain it well enough to my wife and my kids.
Is there something that as a starting place, when new staff comes on board, they can talk with old staff, and they can say, well, yes, there is this model. Start with this picture. Start flipping pages. We don't tell them how many pages there are underneath that, but we start describing the system.
Let me start with Bill and then go to John, because you people are paid to either develop systems and put them in place, or to advise people who are developing systems and putting them in place. They have to do something. They have to do functions, and they have to do it efficiently, or you don't get paid for very long. So let me start with Bill.
DR. STEAD: I really do live in a paradox, Mike, between the fact that I know that without any models that we do not already have, and without any systems that our billing systems don't give us -- just stick with those basic tools we've got -- we can create a significant improvement in the health care system. I know that. We're doing it at Vanderbilt, and I would like us to do it across the country. So one of my challenges is I'm arguing for the low hanging fruit.
The other half of me is more relevant to the question you are asking, if I understand it. That is that if we could build up a lifetime record of a patient over time, with each piece of data being put in as it occurs, then we would have an ability to look at what of the variance is necessary and not that we don't have now.
So I view that as for medical informatics being like the Human Genome Project. In the end, we've got to get at the genetic map of the health care system, but that's a decade or longer project. I really see us as having flailed a lot in the last ten years trying to do things that assumed we could do that big project in the short-term, without ever sequencing the data that underpins our processes, without ever doing that basic work.
So we've been in a halfway zone, and the halfway zone doesn't get us there. So I'm trying to tease us apart into those two very different ends of the spectrum. But I think the big picture of what we are trying to do with data standards is intuitive. It's what I've been trying to do for 30 years. It's what we did with the Duke Cardiovascular Disease Databank in the early seventies.
I think the simple message that I give is that if we could have data about what we do, together with long-term outcomes, the computer would provide a time lapse camera that would let us see which of those process variables made a difference, and which did not. That's my simple message to my wife when answering that question, but I don't think as a country, that's where the big pay backs are going to be in the next five years.
MR. QUINN: I look at your question as basically saying we need a way of being able to communicate to people, and some of those people may be others like us coming up that are going to be doing the same thing we are doing, and others may be your wife or my wife, and basically I'm trying to wave my arms around the air and say this is what I do, and I've given up actually. She just knows I have something to do with health care and I travel a lot.
At any rate, when I look at it from an engineering perspective, in reality you look at two things. You look at things that are static and things that are dynamic, the static things here are the data. The dynamic things are the processes. Whether that be processes and technology, or in this case processes and data.
So you need to be able to describe both, that is what it is that we are talking about. What are the data items? That's the actually easy part, because they are relatively static. And they are also relatively constant actually percentage.
But the real question is how do you adequately describe the variability of what do we do with it, because that changes. It doesn't just change with time, it changes with institution. How do you adequately model, map, describe, make a blueprint? You don't tell somebody, here is a 747. You want to figure out how it's put together? Go dig through it. You give them a considerable stack of papers, microfiche, CD or whatever and say, here are the blueprints for the 747. This is how it's put together.
Now that's maybe an extreme model, but if I was looking at the genome, it's actually a simplified model of what we are talking about. So that's the purpose of model. The purpose of the model is to communicate. It needs to be in various levels. It needs to be a simple model for people who don't need to really understand the depths of what we are talking about, but need to understand what it is we are doing. There needs to be a far more complex set of models for people that are actually doing the work.
DR. FITZMAURICE: Any additional comments from Peter or Ralph? You don't need to make any.
MR. WAEGEMANN: I believe if you focus on modeling, we need modeling. Most of you who know me, I have been spending years in modeling. It is possible for this committee to empower another organization, and to say here some funding should be made available, however, I warn you from my experiences that this is a long, outdrawn exercise. It you look at the modeling in various countries, probably Australia is the only one where there is more than 50 percent success.
If you look at the money spent in the UK, I would say they spent lots of money on it, and it's very controversial at this point. So it's an issue where on the one hand, it is needed, and maybe the best way to go about it would be to empower another organization with funding and getting the right experts, and have this done in a parallel way, but I would not want to dig in here and this group to the questions of data modeling or processing modeling, or any of the types of modeling we need.
DR. KORPMAN: I'll just add, I have beat my chest about modeling for so long, I won't bother. You must have models. I agree with Peter, I don't think that's the role of this committee. But if you set enough high level standards about what the data need to look like, and how it needs to be able to be intercommunicated, and how it needs to be authenticated, the models will arise, whether it's from a committee or a commercial consortium or something, because you will have to have congruent enough models to make this work.
You cannot make the data really congruent unless you have data in process models that are congruent. They don't have to be identical, but you have to have thought through the issues to make it work. So they have to emerge. They will emerge.
Just one other comment. I'm glad someone brought up the genome project. Here was something that people thought was -- I'm depressed when I hear people around the table, this is all going to take 100 years, so let's do something less good than we should. The genome project, that was something people thought would take the rest of their lives. Now they are going to finish three years early.
It can't be any more complex than putting down a few standards for health care information systems that might help make the process better. So I think whoever brought that up, that was a great analogy. There is something very difficult, very complex, that when people finally decided to do it, it turned out they got it done significantly faster than they thought it would.
DR. COHN: Ralph, we are already running late. Clem wanted to ask one final question, which hopefully will be brief.
DR. MC DONALD: I didn't promise it would be brief, but I will. I'm just not clear what you mean by content standards, and whether that is a standard to say what you must store, or how you can store it or record it.
MR. WAEGEMANN: For interoperatability we really need to get into minimum must, but that 10 or 20 percent, but at least some guidelines as these are being created in the field of specialties and so on. I think it is worthwhile to look at it. I would like to come back to some data, or lead you to some data.
Why is it in the south of this country, there is 60 percent more documentation than at the east or west? Why do we have that in some areas it has shown at ASTM some documents do not have a major point included? So you take an operative report, and what you really are looking for hasn't been entered in 50 or 60 reports they looked at. All of these are issues of uniform documentation.
DR. MC DONALD: But there is a "must" side to it.
MR. WAEGEMANN: There is a little bit of a "must" side, yes. The Joint Commission will probably talk about that later.
DR. COHN: Other final comments from panelists? I want to thank you all for what was really a wonderful first panel. We're going to break for 15 minutes. We'll get back together at 11:10 a.m. Once again, my thanks to all the panelists.
[Brief recess.]
MR. BLAIR: It's after 11:00 a.m., so we are going to convene the second panel. As you may have noticed from your agenda, the theme of this second panel tends to be focused on quality and value. We have had to do a little bit of adjusting on each of the witnesses that are in panels in order to accommodate individual schedules. We felt that all of you who are witnesses have testimony of great value for us. We felt it was much more important that you be included on the panels, than necessarily fit within a particular theme. So while we are going to try to focus primarily on quality and value, don't feel as if you are constrained from that.
Now Simon has just indicated to me that the agenda is slightly modified. In order to make that accommodation, Dr. Rick Peters -- is he here?
DR. FITZMAURICE: Jeff, it's modified further again. Rick Peters called and said he switched his airline reservations, and he can make the original agenda, which would have him speaking tomorrow after Dan Russler. So I apologize for not being able to get the agenda today to reflect that. It happened just yesterday afternoon.
MR. BLAIR: Thank you. Then may I ask the witnesses to introduce themselves, please?
DR. SCHYVE: I'm Dr. Paul Schyve, the senior vice president of the Joint Commission on Accreditation of Health Care Organizations.
DR. SCHUTT: I'm Dr. Dave Schutt, the associate medical director for the MEDSTAT Group.
DR. WEBMAN: Dr. Dorothy Webman, an independent consultant with Webman Associates, and also the chair of the Work Group for Computerization of Behavioral Health and Human Services Records.
MR. BLAIR: Dr. Schyve, could you begin with your testimony?
DR. SCHYVE: Sure, thanks, Jeff.
First, I would like to thank the work group for inviting the Joint Commission to testify. As was pointed out by one of the speakers in the earlier panel, the charge of this particular work group, I think is the most exciting charge of everything coming out of HIPAA. It is focused on the goal of health care, which is to improve health for individuals and populations, rather than on the business of health care.
Most of you know what the Joint Commission does, but I would like to point out that we current accredit 18,000 organizations in the United States, only a third of which are hospitals. I point that out just so you realize that as I make comments from the Joint Commission's perspective, that they are relevant to all health care settings, not only hospitals.
The Joint Commission has viewed information, including specifically patient health information as a key resource for any accredited health care organization. It is in fact at the center of health care, even at what the doctor and patient do together. Therefore, it has to be managed effectively and efficiently in order to provide patient care, and to continuously improve its quality.
Every accreditation manual of the Joint Commission contains a chapter of standards entitled, "Management of Information." While these standards do not at present require the use of an electronic patient record, they were intentionally drafted to facilitate and encourage computer-based management of personal health information.
For example, there are four management of information standards that I could cite that are particularly relevant to the task of this work group. That is a standard that requires uniform data definitions and capture methods be used; one that requires that minimum data sets, data definitions, codes, classifications, and terminology are standardized; one that requires the format and methods for disseminating data are standardized; and one that requires that comparative information and data information are defined, collected, analyzed, transmitted, reported, and used consistent with national and state guidelines for data set parity and conductivity.
It is very interesting, however, that three of those standards, and with the phrase "whenever possible." That phrase goes to the center of this hearing's focus, the general absence of universally recognized standards for personal health information content, format, or data elements. So while we can focus on whether the organization is attempting to standardize within itself, we cannot at this point, expect that it is using national standards.
The Joint Commission concurs with the thrust of HIPAA of 1996, and thinks that the questions that were outlined in the work group's letter are well crafted to address the charge. In the written testimony that you all have, I have identified six factors that have hampered the adoption of uniform standards for patient medical record information and its electronic transmission. In the interest of time, I won't repeat those at this point.
I would point out that these barriers to the adoption and use of uniform standards for electronic medical records cannot be overcome by either the private sector or the public sector acting alone. It requires a partnership in which each sector brings to the table its strengths in order to complement the those of the other.
I have included in the written testimony six actions the federal government could take to facilitate standards development in the private sector. But I would point out that for the private sector and the public sector, the ability to overcome the barriers that have been listed, and the willingness of the federal government to invest in supporting these efforts is ultimately dependent on the value of the uniform standards, and of the computer-based patient record.
So I would like to focus my oral testimony on the Joint Commission perspectives on the value of uniform standards and the electronic patient record. I'd like to focus specifically on four issues. First, the roles of the standards in computer-based patient record in improving the care of the individual patient. Second, the role of them in continuous improvement of the effectiveness and efficiency of health care delivery. Third, their role in assisting consumers and purchasers in value-based choices. And finally, the role in supporting the quality oversight system in the United States.
The first function of the medical record is to serve as a conduit of data and information among multiple clinicians who care for the patients -- the doctors, nurses, emergency room physician, et cetera. The current lack of standardization of the data in the medical record exchange between health care organizations, and frankly even often between departments in the same organization inhibits the effectiveness and efficiency of these communications.
For example, when new practitioners in the emergency room see a patient, information that is critical to the safe and rapid care of the patient may not be readily located or recognized in the medical record; things like allergies. Or information that is located and recognized may be misinterpreted, for example because a non-standard abbreviation is used.
Many physicians and nurses can relate personal experiences in which not being able to find information rapidly or to interpret it correctly led to less than optimal care, sometimes to harm to their patient. In addition, quite frankly when the same practitioner looks back through his or her patient's lengthy record, the inability to quickly identify previous medications, treatments, et cetera can have the same adverse effect as miscommunication among multiple practitioners. The human memory is fallible, and the medical record is used to compensate for this fallibility.
These potential problems of non-standardization have been recognized by clinicians, and have led to standardization of formats and abbreviations adopted within health care organizations or their departments, however, as soon as these physicians or other clinicians provide services in more than one department or organization, variation in the standardization itself can be the root cause actually of errors.
The clinician may review the record with the expectation that meets a different standard, and therefore, the clinician may be less vigilant for information that appears in some other place or format than he or she is used to. Unfortunately, the demand for increased productivity in health care no longer permits the practitioner to carefully review all the previous entries in the medical record.
I should also add that the linkage with expert systems, decision-making systems that assist in decision-making is the ultimate promise of the computer-based patient record. That linkage is often dependent upon standardized entries that the expert software can link to.
With regard to the second issue, the use of standardization to help improve the quality of care, the performance of the organization. The record is the source of data information, that usually in aggregate form provides the basis for assessing and continuously improving the performance, both the effectiveness and the efficiency of the organization or health plan an integrated delivery system.
This performance of course has always included the provision of quality care to individual patients, but the performance of an integrated delivery system or health plan also includes the improvement of the health status of the population the system or plan serves. Today it is recognized that the performance at all levels of the system is also measured by the efficiency with which they achieve their health care goals, as reflected in resource utilization and cost.
The lack of uniform standards across all practitioners and provider organizations is a significant barrier at all levels of the health care system to their improving their performance. For example, the collection of clinical data within a health care organization is often costly because it is dependent on manual charge abstraction. This cost of data collection results in less collection of what otherwise could be good and useful data to support both ongoing monitoring of care and quality improvement efforts.
In addition, the collection of clinical data by an integrated system or health plan is not only often dependent upon expensive chart abstraction within each provider site or practitioner office, but ostensively the same information from different sites or offices may be actually quite different, because of the use of different standards, for example, definitions of data elements by the different sites. Aggregation of these data may lead to the misleading combination of apples and oranges.
With regard to the third issue, which has to do with other stakeholders who in fact want standardized information. Standardized information as derived from medical records serve two key stakeholders in the community, consumers and purchasers, both of whom want standardized information on the quality of care to help them choose among practitioners and providers based on both quality and cost.
While this interest in standardized data is to meet their own needs, their choices, if it is based on valid, useful, quality information will stimulate improvement among providers in a market-driven health system. So their own personal, selfish needs actually will also contribute to the quality of health care.
The lack of uniform standards for format, classification systems, codes, including for data element definitions mean that health care organizations' responses to these consumer and purchaser demands are expensive in both time and dollars, and often yield data which are not neither accurate nor comparable across health care provider organizations.
Under such conditions, reliance on the available data could lead to poor decision-making by consumers and purchasers, and to ineffective improvement efforts by practitioners and providers.
An additional problem that faces consumers and purchasers, and therefore the health care organization that is trying to respond to their request for data is that consumers and purchasers have not, and I suspect never will, achieved consensus on a limited set of performance measures that will meet all their needs.
For example, consumers may find different measures to be most relevant to their decision-making depending on their age, their sex, whether they have children, the acute disease for which they are at risk, whether they are healthy or chronically ill, et cetera.
As a result, the data which are requested, are likely to change over time. A non-standardized medical record system that requires redesigning, and often manual data collection methods for each new request is not only cumbersome and costly, but is unlikely to be fully responsive to the consumers' and purchasers' needs.
Finally, with regard to the fourth issue, the importance of uniform standards and the electronic patient record for the private-public partnership for oversight of the quality of care in the United States. That oversight is dependent increasingly on good outcome and process data, a major source of which is the medical record.
Thus, the Joint Commission, as part of that partnership, has its own specific reasons to advocate the adoption and implementation of uniform standards for health information and the electronic medical record. The Joint Commission would, in the future, if such standards were available, move organizations to use of uniform national standards and electronic patient records.
Because they would be powerful tools, first, in the hands of clinicians for improving the care of individual patients; second, in the hands of health care organizations, delivery systems, and health plans for improving their performance; third in meeting consumers' and purchasers' needs for information to aid in their decision- making, all of which in fact are central to the Joint Commission's mission.
Specifically, the implementation of OREX(?), the Joint Commission's initiative to incorporation performance measures into the accreditation process, and that I would the same type of things are occurring in all the other accrediting bodies, HEDIS with NCQA; URAC(?) is looking at measures at this point. This is dependent upon the identification of common core measures to be used by every similar accredited organization that provides similar services.
Each core measure will be carefully specified with respect to its numerator, denominator, data element definitions, algorithm for calculation, and if applicable, risk adjustment data elements and algorithm. Uniform national standards for electronic medical record formats, classification systems, and codes would simplify this definitional process for the Joint Commission and other accrediting bodies, and the data collection process for the health care organization. It would also facilitate revisions in those core measures as knowledge, technology, and the needs of professionals, consumers, and purchasers change over time.
In summary, while uniform standards would be advantageous to the Joint Commission in conducting its accreditation process, their real value stems from the contribution of uniform standards for patient medical record information and the electronic patient record to achieving the Joint Commission's public mission, improving the care provided to the public, which is obviously a goal that we all share.
MR. BLAIR: Since I can't see you, I'm not sure if Dr. Schutt is next. Is that correct?
DR. SCHUTT: Yes, I am indeed. Good morning. You have in front of you some copies of the presentation slides that I will use to support my testimony. The MEDSTAT Group's mission is to apply health care information strategically to improve the value of health care. We do this for a variety of clients, which number over about 1,000. They span the range from the states, to purchasers of care, to providers, and to managed care organizations. So the comments I will be making this morning will be based on the experience we have with this breadth of clients.
What these clients tell us they want is the ability to demonstrate value. They want to demonstrate value in two ways. They want to be able to demonstrate that the value in the case of a health provider organization, that they actually are doing something worthwhile for the patients that they see.
The second part of the value equation is they have to be able to justify to their CEOs that the information systems that they purchase are indeed providing value to that organization, a burden which now has been turned on organizations like the MEDSTAT Group, where when we go into organizations, they frequently want to know first off, tell us how this is going to help us with the business that we're in.
Lastly, they also want to be able to measure simultaneously, cost, use, and quality. They have, over the last few years, measured each of these independently. They have tried to make assumptions and guesses about the impact of performance in the other two domains with information contained in just one of the domains, a relatively hazardous undertaking I would caution. So they now want to be able to look at all of those domains simultaneously. As Dr. Korpman mentioned earlier, the unit of measurement here is the patient, and we will talk about that a little bit more.
Lastly, they would like to be able to have some way to compare their performance, some way to compare themselves to somebody like themselves. As Dr. Borwick(?) frequently says, collecting data and reporting it, and comparing yourself to a benchmark and sharing that benchmark basically creates a new reality. It basically says that people who are in the same business as yourself are doing what you do better, and it motivates that individual or that organization to perform in an improved manner.
On slide number four you see a very lengthy definition of what clinical performance is. This is not MEDSTAT's definition, this is a definition we borrowed from Dr. Ann Lauthers(?) from Harvard. The reason I'm showing it to you is to show you how long the definition is, and to show you how difficult it is to develop, as Dr. Schyve just talked about, performance measures.
If you look at the adjectives and adverbs in that statement, each of those cries to be measured, and it is not that easy to take the available that I'll talk about in the next couple of slides, translate it into performance measures that sheds bright light on clinical performance.
The data we have available in the next slide, on slide number five, are those that are available in the public and private sector. First are the data that are available through reimbursement records in the inpatient and the outpatient sector. You see in the parenthesis there the sources of those data. In most instances, those data are standardized, for instance, looking at ICD-9-CM codes or CPT4.
We also are asked to use employer records to look at information on absenteeism and disability days, worker's compensation claims and so on. What some of our large purchaser clients are now very interested in is not only tell us about the medical care that is being provided to our employees, but help us measure whether it is doing anything to improve their productivity and their presence at work.
So we now have been asked to look at additional sources of data, which also get into the arena of non-standardized data when you look at employer records on absenteeism and disability.
We also are using survey data. There is no standardization for surveys that we have been able to find, although that hasn't stopped us from being in the business of generating surveys, and collecting the data, analyzing them, and reporting that. We collect data from both employees, providers, health plans, and from households, again, to try to round out this picture of performance.
On the next slide we have a bullet called emerging data availability. We have some clients that are leaning very far forward in this area, that are making available to us their pharmacy data and laboratory data. The reason this is important will be shown in the next couple of slides, but I want to point out the small comment at the bottom.
With the data that is currently available, and as good a job as we think we do in collecting and analyzing it, and reporting it, we still are only getting very small snippets, or in the words of some of our technical folks, pixels on the entire picture of how a patient is cared for, and the impact of that care on their overall well being as they transverse the various settings of health care. They are small little flash bulbs that go off that we capture and try to weave together in a moving picture, which is very clunky and cludgy at best.
Two of the products that we have embedded the current data sources in I'll show you in the next two slides. One of them is called disease staging, which is a methodology which allows you to severity adjust at the individual patient level. The reason I'm showing you this is to look at the far right-hand column under diagnostic findings. What you will see there is a very generous clinical description of a particular stage, in this case of a patient that has diabetes, that help us describe the level of illness that the patient brings to the care setting.
You will see that that clinical definition is ripe with lab values. Now some of our clients actually use this in a very manual and slow and difficult way to stage their patients, looking at the manual, and trying to figure out matching the patient's lab data that they get out of the medical paper record with what we have described in our manual, however, most of clients won't tolerate that. That takes a highly motivated person to do what I've just described to you.
What the rest of the world does is they wanted us to translate these clinical findings into ICD-9 codes, which we have now done, which allows us to electronically process and therefore classify patients by severity. The problem with that is you go from a very elegant clinical description of a patient's stage into a very inelegant description using ICD-9 codes. There is an awful lot of clinical descriptiveness which is left on the editing room floor.
As a second example, we are able to tell -- and this is now getting closer to looking at clinical performance -- we are able, using claims data, to get a sense of the services that are provided to an individual patient or to a group of patients. In this case, what we have done is taken national guidelines and operationalized them.
In other words, take the recommendations that are found in certain guidelines, in this case the recommendation that hemoglobin A1C tests be done for diabetics at a certain rate over the course of the year, and run those findings using the client's claims data. What we are able to tell the client is how many of their patients did or did not have the test done in a certain time period.
As nice as this is to be able to provide for them in 1998, what they now are pressing us for is what was the actual value of that hemoglobin A1C? It's nice that we know that a test was done yes or no, but tell us what the value was, because that would be very helpful in identifying these people for case management, disease management, demand management programs.
And not only identifying them, but tracking whether the programs actually are doing anything. In other words, we put in place a disease management program for diabetes, and are their hemoglobin A1C values actually coming in line?
This gets to this business on return on investment. Is it worth our money to invest in a program that is going to improve in this case, the lab tests for a particular patient or a group of patients? What does it take to invest in that, and then what return are we going to get?
On the next slide, which is slide number ten, we think there is great value to be gained in collecting clinic data. As we have just shown, we already have clinical data embedded in many of our products. It has been there for several years. We are just waiting for the opportunity to exploit the availability of those clinical data.
Getting the specific questions that the committee has asked, our interpretation of what the congressional instruction was is they are asking basically for guidance on who will lead in the effort of standardization.
Question number two was what factors are preventing implementation? From our view, the biggest factor is that there is no universally accepted standard. Number two is from our view there is no clear leader. We are willing to follow an industry leader or a government leader, it doesn't matter to us. We would just like to see one.
Thirdly, the issue of confidentiality, privacy, and security is a big issue for us as a health care data user. Internally we have a very detailed and aggressive policy on confidentiality and privacy and security, but these issues at large we are finding, are making the going slow.
Fourthly, this business of return on investment. Our clients are now asking, after a decade of investment in information informatics infrastructure for their organizations, tell us how what you do MEDSTAT, is going to make our business better. Show us that in buying what you do is going to return in investment, theoretically on the positive side for us.
Fifth, it was mentioned earlier by the first panel, the complex of the care process makes it extremely difficult to capture everything that needs to be done. We are here today to say that we are not sure that everything needs to be captured in order to begin to move ahead in this arena.
Lastly, the complexity and imprecision of the nomenclature in medicine, Dr. Schyve again mentioned, the imprecision and the misinterpretation of abbreviations that occur.
The third question is what is the role of government in all of this? We think the role of government is to facilitate a closure on a standard. We think that what government can do is narrow the corridor to help correl and guide this random motion that is occurring in the industry as we see it. To basically present a perspective, so as we go down the road, the corridor gets more narrow and more narrow, and the standard is eventually reached.
To basically promulgate one standard, and as government is good at doing, to prepare and put in place either mandates or incentives that motivates the industry toward the use of a standard. Our suggestion is that this has been done successfully in other arenas, and that you look to those for examples.
Slide 13, the business case. I would submit to you that MEDSTAT is the embodiment of the business case. We exploit everything that is available to us for our clients. We have a foot in the academic side of looking at what the actual missions are proposing in terms of measuring clinical performance, and we have a foot in the business side of translating those into a going and profitable enterprise.
We have to do what the clients find valuable. The clients tell us what they find valuable by what they are willing to pay for. What they are willing to pay for is usually something which is not too expensive, and it's not expensive for us means it is something that is standardized.
When clinical data becomes available and meets the above three bullets, I will tell you that MEDSTAT is willing to use it, as you have already seen. It's already in our products.
Lastly, do we agree with the four areas of focus? The first bullet, there is a question mark, because it's not clear to us when we read those, exactly what was meant by that. But that did not stop us from offering you advise.
We would suggest that you start narrowing, that you focus on the doable. I think Dr. Stead mentioned that in the first panel. That you focus on the common, and that you focus on the quantifiable, because there is usually more agreement around the quantifiable than around the non-quantifiable. I have given you some examples of what, in our view, quantifiable means.
Lastly, Mike, you had mentioned that you were looking for a model or a format to explain to people outside of the arena, what you do. The thing that circulates around MEDSTAT is a description of how a patient traverses through the health care system today is that as a patient it is the same as if every day you woke up and you left the house without your glasses, without your wallet, without your birth certificate, and without your social security card.
When a patient makes contact with the health care system, or traverses from one setting to the next, there is amnesia, and we feel that that impacts on the value and quality of the health care that can be provided.
Thank you.
MR. BLAIR: Thank you very much. Our next witness, please.
DR. WEBMAN: I want to thank you in the panel for actually answering the very first request that the Work Group for Computerization in Behavioral Health Records made to HCFA when the proposed transaction set came out. That request was to please seek input from data standards, data exchange, privacy, and consumer groups from the behavioral health community.
That is why I'm here today, to talk a little bit with all of you about the needs of the behavioral health and human services industry in this picture of talking about health care records.
MR. BLAIR: Could you please give us your name.
DR. WEBMAN: I'm sorry, Dorothy Webman. I chair the Work Group for Computerization in Behavioral Health and Human Services Records, as I stated. That group's primary mission to create and promote equitable standards and guidelines for ownership, privacy, confidentiality, quality, and accessibility of behavioral health and human services records.
It's a non-profit, voluntary organization, and represents an alliance between members of the public and private behavioral health and human service systems, and family and consumer advocacy communities. We are dedicated to insuring that the transition to a computerized record evolves in a way that is beneficial to children, families, consumers, providers, payers, researcher, and regulators.
The executive committee of the group is composed of a diverse group of individuals who range from chief information officers to providers to consumers. Leading organizations in the public and private behavior health and human services industry are represented, and the committee has worked on projects of national scope with the U.S. Center for Mental Health Services, as well as large scale projects within our respective organizations.
The group is committed to advancing the use of state-of-the-art affordable technologies toward the end of improving the quality of all health care-related services, and the well being of people qualified, and the group feels qualified to offer recommendations.
In preparing this testimony, I didn't feel comfortable only speaking with the executive committee of the work group, and contacted colleagues at the Substance Abuse Mental Health Administration, the Mental Health Statistics Improvement Program, and the American Managed Behavioral Health Care Association, and members of a national project that is housed at Apt Associates working on the development of a paper prototype for the behavioral health record. In doing so, it's fair to say my testimony becomes a representative sample of their collective thought.
I described in the handout that I gave out just now and the Power Point slides a little bit about each of those groups and their background. All I can say is that they are standard setting groups that are particular to the behavior health and human services community, and have been working long and hard at that. We would urge this committee to seek their advice on issues in the near future.
In terms of interpreting the congressional instruction, I would interpret it, as many of our colleagues have, regarding the Health Care Portability Act, however, an additional piece that we want to place some extra punctuation on here is the incorporation of behavioral health information in the administrative simplification activities.
It seems as though when the proposed data standards came out, that many of those directives seemed to have fallen at the wayside. We are extraordinarily concerned about this. Dentistry and education seems to have found its way in, and behavioral health not with adequate attention.
I would call your attention to electronic commentary that came directly from the Center for Mental Health Services in reaction to the May 7 statement, which says, the document is seriously deficient with respect to essential information on functioning mentally ill persons required for enrollment data, and is totally non- responsive with respect to encounter data on the classification of mental health services.
This opinion is shared by us, the Mental Health Statistics Improvement Program, and our group on automated human services records, the group I speak of. We strongly recommend that the Center for Mental Health Services and its technical consultants meet with representatives of HCFA and ASPE to address these issues. In its current form we cannot recommend the standards for adoption.
When I then asked other colleagues about the --
MR. BLAIR: Excuse me, you referred to the document?
DR. WEBMAN: The May 7 proposed data transaction standards. Proposed rules.
So with regard to the field, many of the people in the field are unaware that these rules came out. So lots of people sent me back e-mail saying, I don't know about that congressional information. I have no idea about the congressional instruction. Tell me about that please. How did that affect me? Then there are others who are much more aware, and their concerns lie in more of the parity issue.
When I then speak and talk about cross-system needs, many people who experience behavioral health care episodes, also have needs in the areas of special education, justice, and other human services. This is probably not new information to the committee, however, we want to draw your attention to the higher viability of looking at a medical record that would incorporate all of these silos, rather than continue to keep them silos.
We would like to point out respectfully that of the 1,346 data elements in the proposed rules, that only 34 of them overlap with mandated child welfare data enrollment sets, with voluntary Mental Health Statistics Improvement Program data sets, or with the TANIF(?) enrollment encounter data sets.
This is a very bare minimum number, and in the mental health area there are only six. It is quite interesting that of the six, that three have to do with unique provider identifiers, unique patient identifiers, and other unique identifiers.
Which brings me to the issue of privacy and confidentiality as it relates to the behavioral health care community, and the deep concerns related to that that evidently are shared with this committee and those creating the proposed rules, and the behavioral health community.
Many people here spoke about an infrastructure. The group that I chair has actually gone about the business of starting to look at what an infrastructure might look like across systems, and in your handouts I would call your attention to the very back of the handout, that lays out real quickly what we hope will be a bit of a prototype for an infrastructure that would connect human services, education, housing, health care.
In addition to that, on the flip side of that, the APT project that I spoke of about the prototype, starts to lay out what we think are ideal content in the areas of population data, enrollment data, encounter data, cost data, clinical guidelines, system guidelines, consumer outcomes, system performance, and record cards. And in that project are spelling out what the unique elements and indicators are specific to behavioral health and human services.
With respect to your questions -- I'm going to leave our other comments on the standards alone -- but with respect to your questions, and the factors preventing or delaying the development of widespread implementation of standards, essentially we would break this down to the lack of dissemination and knowledge within the behavioral health and human services community, the lack of consensus on standards, which has been mentioned here, but not articulated again, in terms of behavioral health and human services.
So if I give one small example of that, I would talk about that there is actually quite a bit of consensus in the area of outcome measures for children and families. Children should be in school. Children should be at home. Children should be in their neighborhood. Children should be functioning positively with good behavioral health and good health.
But then when we go down from the indicator level to the measure level, we get into lots of difficulty, because another one we have consensus on is something like children should be out of trouble with the law. When we go to that, where do we go but to normal administrative data sets to look for that, and we find that those administrative data sets like probation data sets are extraordinarily biased, and children would be misperceived at the least if those were used.
The lack of money in the mental health and behavioral health fields far exceeds that in the health care field in the underwriting of infrastructure development, advances in behavioral health outcomes, and performance measurement, and report card areas. That leads to a lack of available technology.
I giggled to myself when I heard one of my colleagues earlier talking about Y2K, and whether or not there might not be paper records around. What I'm thinking in the behavioral health world, a lot of it is all paper records still. So they don't have those kind of Y2K problems just yet.
Lack of adequate attention to privacy protection was our last and most important one. We have deep concerns, because there was a meeting recently on the computerized patient record here in Washington where testimony that was given to this community by mental health consumers was referred to as testimony by a crazy person -- by a person incompetent to stand trial. A person unable to offer useful information to the committee.
We know this committee heard that testimony. We know that the committee took it seriously. It was just terribly disconcerting to hear in a public forum that someone would take consumer information that lightly, to put it mildly.
We ask this committee in the role of government here to not allow that to happen. And in addition to creating protections for privacy and confidentiality, to create protections along the lines of the consumer rights bill, specifically mentioning the needs of the behavioral health community with regard to that.
In terms of the role of government, we think that government should protect the public trust. We also believe that government needs a business plan. That when you develop technology, my colleagues referred to building the space shuttle part by part. That we have done this in the information technology area, building part by part.
Even this prototype I showed you from the Center for Mental Health Services will be built part by part. We're looking for the places where it exists part by part. But the government needs a business plan for the health and human services industry. It needs to take HHS and CDC and pull them together, and have a business plan that can somehow fit into something like the graph we show you here around the electronic infrastructure.
Then with that business plan, to behind that build a technology plan that will follow, not the reverse. We have plenty of examples in the federal government, SCQS(?) not being the least of it, to buy hardware before we buy software, to buy software before we decide what needs to be in the software. We need to start to do things the way the rest of industry does, and then perhaps we could buy and build models that will be in fact replicable and salable later as the vendor community seems to be calling for.
Last but not least here, with regard to quality and adding quality, we are absolutely sure that particular to the behavioral health community that MISUP(?), the work group, and the Apt project offer you plenty of examples of ways to do that, and we are happy to work with you on an ongoing basis in that area.
We want to punctuate that. National co-morbidity studies certainly show the overlap of behavior health information on health care in general. And that we would no doubt improve the efficiency and effectiveness of health care if we are adequately covering the behavioral health care needs of individuals in our development of the record.
In terms of the four areas, for the most part I can tell you that my colleagues agreed, however, they all believed for the most part, and I was surprised by this, the business case had to come first. And that all of the rest of your number 1, 2, and 3 really fell under the development of the business case.
And that the business case needed to be created vis-a-vis a public-private partnership. This is probably the most radical thing I have to say, and it's not in my handout, I regret. People were really calling for public-private partnership in the sense that we may need to look at some significant changes to federal regulation to allow for ownership of code.
And not to repeat over and over again the use of large scale consulting firms et cetera, to create one after another data systems, state by state, neighborhood by neighborhood, federal agency by federal agency that is owned by somebody, and antiquated five years after it is built for the 80th time. So we would really like to see people consider a public-private partnership that allows for full private investment.
MR. BLAIR: Do we have Charles Beradesco? Okay, only these three individuals. Are there questions from the committee members?
MS. FYFFE: Thank you all for your good presentations. Dorothy, you have really caused my ears to perk up. You said something about the administration standards overlooking mental health, but including dentistry and education. Can you explain that?
DR. WEBMAN: Well, there is just a significant number of data elements listed in that proposed rule that focus on dentistry, focus on some school-based health clinics more specifically. So you have general mention, and then at the code level there is more mention of behavioral health, but not at the general data element level. You get things like outpatient, inpatient, with no description of what that means, or no attention to things like social functioning on functionality scales.
So these are the kinds of things we are talking about. Actually, here I have actually provided a newsletter that lists and itemizes a few more of these things. We will be providing electronic copy of testimony that will be a little bit more extensive, with more specifics and recommendations.
MS. FYFFE: I would like to explore this a little bit, please. Please, the other folks jump in. My perception rightly or wrongly is that in the development of the standards that went into the proposed regulations that there was several years of effort, probably since the late 1980s in trying to figure out a way to automate or put into an electronic transaction claims for professional services, institutional services, and payments for those, as well as insurance eligibility and verification and so forth.
So I guess my question to you is how are the professional services you are talking about for mental health being billed today? Also, please let us know if the DSM3 codes, which I think are part of ICD-9, fit into that.
DR. WEBMAN: DSM4 as we move on. In terms of the billing, on the public side you are doing your standard Medicaid billing system, and with a lot of the same broad brush categories, and because of that, not capturing a fair number of important services that are provided in the behavioral health community, and being to describe their value.
So when we talk about wrap around services, for example in behavioral health, they are coded differently, and maybe placed under outpatient care, on an outpatient option, or case management option on Medicaid billing. They are not captured. So then when we go to justify their value to the behavioral health community, we have a tremendous difficulty.
DR. FITZMAURICE: Should there be more research that focuses on how you describe the services or how you count them, so you then can relate the services to changes in outcome, and therefore give them value? Is that one of the pleas that you would make?
DR. WEBMAN: Absolutely. That would be one of the many things that have not been adequately financed by the federal government that we feel is a significant role that the federal government can play.
MS. FYFFE: I think -- and I cannot speak for the dental community -- but my observation of the dental community is that they were very aggressive in developing standardized data elements and codes and so forth for classifying the services they performed. You know you have to begin at some sort of starting point with electronic data interchange, and the fact that ICD-9 codes were being used, CPT4 codes were being used, and dental codes were being used, those were quite legitimately put into the proposed regulations.
I don't mean to be defensive about the many years of work that has gone on, but the law also included a mechanism for changes to code sets over time. There are several standards bodies that are working on that, and will continue to work on that. My suggestion would be that you all get involved. I've been around this for a few years. This is the first time I've heard of your organization. Please don't be discouraged. There is a vehicle for you to talk to the data element organizations and standard setting organizations to get what you would like included.
There are some other parts of the industry that do not have mature data sets either, for example long-term care being one of them. How do you classify how people who are chronically in need of long-term care and get home services, there does not seem to be a widely accepted data set for them.
So there is still quite a bit of work over the years ahead, and again, I would encourage you get involved. There are several of us here that could talk with you about that in terms of names and addresses and so forth of organizations you should get in touch with.
DR. WEBMAN: We absolutely appreciate that, and in fact people who sit on the Mental Health Statistics Improvement Programs ad hoc group, and people who sit on the work group would love to do that. Again, I could only point to how important resources are in that area, because all of these groups are voluntary. Most people, if they are public employees, can't travel out of state on government money, and private employees can't afford to go on their own.
So the resources to join groups and attend meeting just simply have not been there. The Center for Mental Health Services has been -- and I have to in fact name a person here, Ron Manderschid -- have been phenomenal leaders and supporters to make possible any involvement that these groups can have at a national level with the other data standards groups.
We are well aware of the groups, and have been trying our very, very best to keep abreast of them. Peter Waegemann's Website is enormously helpful in that sense. Others here have been helpful over the years. It's really been a struggle because of the absence of resources to participate.
MS. FYFFE: Sure.
DR. COHN: First of all, Dorothy, thank you for your comments. I actually was sort of taken as I was looking at your comments on the four focus areas, where you commented about the issue of the business case. I think one would observe that even though it is number four on the list of focus areas, it's actually where we are starting out, because that's really the focus of today and tomorrow, is looking at the business case issues.
I guess I would ask for the three of you to perhaps let's talk a little more about business case issues. The work group itself and the committee has a fair amount of discretion about how it approaches these issues around comparable patient medical record information, and the areas that we have talked about.
I have heard from many of you, the issues of doing the doable, not having too long of a view of the future, but trying to focus on three, five, seven years potentially. From all of your views, where would you suggest would be the really high value areas that we should make sure not to miss as we begin to move forward in these areas? Paul, do you want to start?
DR. SCHYVE: Sure. This is going to sound I think actually a bit strange to this group, but I would actually suggest starting at the level of trying to push toward more uniform national standardization of such things as the medical record format, the content issues, and Peter had responded to that issue that some is mandated. If it's going to be there, this is how it should be kind of thing.
Coding systems that are uniformly used, including specifically whether they are being used well. One of the things we discovered as we did all the research on developing a performance measurement system is that people would be using what was the same coding system, but in fact when you went back and looked at the data, they were actually using it quite differently. You still actually had apples and oranges as you tried to compare. So part of the issue is do we have good classifications, good codes, et cetera.
I would also suggest that we have to focus -- and this is tough -- but at the atomic level you might say. The reason for that is that classification systems are based on a model of the world. As soon as you move into the classification system, you are already moving into a model of how things work.
A little piece of information comes along that tells us that our scientific model turns out not to be correct, and your classification system, if that's what has been built into your system, you've got a problem. So at the point at what I would call the sharp end of dealing with patients, where the greatest variability has to be dealt with, we have to have systems that let us in fact in a very standardized way record very detailed atomic kind of observations. We can't just rely on when it all sums up into some classification.
The reason why I started these comments by saying that what I'm about to say may be a little surprise to you is because in fact I believe that though standardization would make a great advance in paper medical records, the examples I used of the doctors and nurses making mistakes, literally making medical errors, or the inability of a health care organization to collect information that lets them measure a number of different variables for improvement purposes, those same issues are there even if you are using a paper medical record.
In fact, the examples I used of the errors that the clinicians make are errors primarily based today on paper medical records, not on electronic records. If you think about the role of the clinician, people have described the system as being an inverted triangle. At the blunt end, and down the sharp end -- and the sharp end is what hits the patients -- and the people working at the sharp end of this system are the clinicians. They are the ones interacting with the patients.
What you would like to do is the systems they work in, which are the blunt end of this triangle, the upper part, the big weighty part of the triangle, the systems they work in need to be as standardized as possible, because their key job is to adapt standardization to the needs of a varying input, which is the patient.
It is exactly the same model that goes on in other what are called high performance, high risk industries. Probably the best example is what goes on in the flight deck of an aircraft carrier, but it's the same thing that goes on when you talk about the nuclear power industry, manufacturer of dangerous chemicals, et cetera.
You have to have standardized systems that the people working in the systems always know where their feet are. Because their job is to anticipate what is changing that has to be adjusted in a way that a standardized system can never make the adjustments. That's the difference between machines, including computers and software, and human beings.
So in fact some people have described this as the role of the human being in this system is to be constantly creating safety. Creating the right thing for that patient, because of the patient's variation being dealt with an increasingly standardized system.
Whether that standardization is in the format of the written record, or ultimately because of all its advantages, in the format of an electronic record, I think we have to take into account what it is that we standardize, and how we deal within that standardization with the role of the clinician at the sharp end, to constantly deal with that adaptation.
I think that was part of the theme that I heard this morning about trying to deal with the variability in health care. Maybe one of the clues is to think about standardizing the systems that people work in, which includes the medical record system, the codes they are always working with, the classification systems they are always working with, while at the same time letting that system accommodate to the constant changes that the human has to make, the clinician has to make to create safety at that point.
I would just close this by making one comment. A great quotation I heard was from Kathleen Sutcliff(?), who is at a business school, and studies these kind of high risk, high performance organizations, who said, "The old definition of reliability is minimal variation." That's the engineering definition.
It is the definition we usually think of when we think about computers. The new definition of reliability in these kind of high risk, high performance places -- and I stress the high risk -- is that reliability is the ability to identify the changes that require interaction in order to prevent unintended consequences.
Those are two different definitions of reliability. Computers are great for the engineering definition of reliability. The human is the one that fits that new definition of reliability, and they both have to work together. I think where we struggle with the computerized medical records that interface between the engineering definition for the computer system, and the need and the role of the individual.
MR. BLAIR: Paul, thank you. I'm reminded of Bill Stead's discussion this morning. It's a very interesting different take of that whole thing.
David, do you have a comment?
DR. SCHUTT: People that function in health care are well educated, well intended, and they do the best they can with what they know. It is that last piece that MEDSTAT is in the business of improving. That's the discussion we have with our clients when some contention develops about well, you're telling us we're not performing well.
The answer is we think you're performing as well as you know how to perform. Or you are organized as well as your results show. Once again, every organization is organized to get the results it gets, and health care is no different.
I happen to agree with Paul about using the information we have well. One of the first things we do for our clients is go in and clean up their data. It is kind of jaw dropping enlightenment that they think they are ready to take on and meet the Joint Commission's requirements for ORAX(?). They think they are ready to take on the mandates from the states in terms of demonstrating how well they perform. They think they are ready to respond to their large purchaser clients about how well they are taking care of the employees.
When we go in and look at the data, in spite of the fact that they are using very standardized ways of coding, it is not done very well. Simple things are not done well. We get very enthralled with very sexy kinds of technology. In my previous life we were working on some very exciting ways of transmitting radiologic images from navy ships to shore facilitates, and it was spectacular.
The young sailor said, I'm very impressed, Dr. Schutt, but how come every time I go to see the doctor, they can't find my record? It's simple stuff that the guy who is doing the job is interested in. This is very exciting, but when he and his wife and his child go to be seen, they would like to know that what happened to them last time is available to the person seeing them this time. And they are not the ones who are paying the price of having to have the lab test done again, which is fraught with not only costs, but discomfort, as well as Paul suggests, potential for error in doing something again.
The other piece that Paul mentioned, and I happen to agree on is where we're playing this game is at the level of the patient, and at the processes of care that are provided. Measuring outcome is okay, and MEDSTAT indeed does do that. We indeed to sell that service.
But when you talk to the physicians or the providers, when they go to work in the morning, and they are on the elevator going up to the operating room, you don't hear the discussion of, gee, I think today I'm going to lower my mortality rate. That's not what they do. They are in the business of applying processes to a patient in the hopes that the sequencing and orchestrating of these processes result in a lower mortality rate. They don't say, gee, my rate is low, I think I'll kill somebody today and get back up to the norm. It doesn't happen that way.
So the argument here, Simon, to answer your very specific question, is the patient focus, the patient level, process oriented. And if I can be as clear as possible, the things that shed the most light to help complement the data that we now have available to us are pharmacy and lab. That is going to add 20-30 percent descriptive ability to our methodologies to measure what was done to the patient, so that you can get to discovering best practice, so that you can get to measuring outcome, and so that you can get to evaluating and articulating value.
DR. WEBMAN: I just want to start with emphasizing that we would agree around patient-consumer, but also family. Especially when we're talking about coding, it is extraordinarily important to note that in both Medicaid and Medicare there is not a family code. There is an extraordinarily large number of dollars wasted on families with multiple case managers that do not need to be there if only we knew that these people were related, and that we could work with them together.
So this is a very, very important piece, and is one of the things that we asked for in our recommendations that the committee pay attention to.
So we would agree that from the consumer perspective, you look at the business case. It's a supply and demand thing. So you've got consumers. You've got family members who are dying for reliable, coordinated, useful, privacy protected information that they can use, and that they can share at their will. And that there is available electronic infrastructures to make it be that way. That consumers can go to electronic kiosks and draw down information about providers on an electronic smart card, or with biometric recognition.
I'm sorry I'm using these fancy things that you say aren't necessary, but it's possible to build that trust that my colleagues have talked about impeding the system by making the system useful to consumers to on the front end.
And that they are able to extract information from the system, correct the information in the system, write on the record in the system, and use the information in the system on a regular basis before they are asked to give up private information.
In addition to that, clinicians in our field, in behavioral health, the smart ones, are leaving by the droves burdened by the demands of managed behavioral health care companies and other payers, with multiple payment forms and their standard forms, and are dying for these standards.
We talked to AMBHCA, the American Managed Behavioral Health Care Association, members there would probably tell you that they are dying from the state burden, from the Medicaid regulations, and from the different state interpretations of the mental health statistics improvement program enrollment encounter data sets, and that's varying in very single state, and they are not having a standard to adhere to. What is a Medicaid data set?
Then when we go to the state level, we talk about -- Dr. Bob Ross(?) will give you the example, gave a very compelling speech. I wish I could take credit for it myself, at the families and health conference in June in Tennessee. He brought the vice president two gifts. He brought him the Nashville Yellow Pages, and he brought him a volume larger than the Nashville Yellow Pages. He brought him an 850 document that was the eligibility forms for the health and human service agencies in the county of San Diego.
This is a sad commentary, folks. But I am happy to tell you that the county of San Diego has done just what we are recommending in the work group here, that the government create itself a business plan. The county of Sand Diego has a business model it contracted to get developed. After it did that, it articulated its business model in terms of the values public sector holds near and dear to its heart -- family focused, community-based, culturally competent, community driven, coordinated care.
After that, it built a technology plan using the Internet and intranet to connect vis-a-vis common intake referral assessment system. And then movement toward a coordinated CPRI. So that system folks, is going to cost in excess of $24 million they are estimating at this point in time, and it's up and running and it's called Synergy. So we wish the county good luck, and that it really does exist out there. That people are starting to look to contract for that.
Spokane just released an RFP in Medicaid managed care, asking for two things, an ASO, and an MIS system for Spokane County wanting a decentralized IS system, asking for that intranet thing, and then letting the managed behavioral health care companies off the hook during the bidders conference, saying yes, you can deliver that in two years. Because why? The industry didn't have it to deliver.
There is a need for partnership. When we talk about public-private, private is a big word. Private means non-profit, small agency. Private means technology company. Private means big managed behavioral health care company. Private means small company. We can't forget that it means all those things when we think about public-private partnership.
Thank you.
MR. BLAIR: Dr. McDonald.
DR. MC DONALD: You mentioned, Dorothy, in your original presentation about a presentation to here you were saying? This committee?
DR. WEBMAN: No. The one where the person was not appropriate around the consumer issue? No, it was a public meeting where they referenced involvement.
DR. MC DONALD: David, you described sort of the ultimate low hanging fruit in a sense.
DR. SCHUTT: On the ground.
DR. MC DONALD: Well, you mentioned laboratory and pharmacy data. Is there something the government could do to -- when you are talking about it being very fragmented, you meant very few people are doing that, or very little of that is around? What would the government do to help that lowest hanging fruit?
DR. SCHUTT: My comment was, I think if I'm communicating here, that there are some of our clients that have available to them pharmacy data that is in an electronic form. Some of them are purchasers of care. Some of them are providers of care. The easiest example to think of is an individual hospital that is generating requests for medications both on the in- and the outpatient side.
The ability to get in there and get that kind of data gets you so far down the road in understanding the care process, and being able to get at what I had shown on one of the slides, this measurement of clinical performance. Were things done on time? Were they done appropriately? Were the right things done to the right people? Were they done safely?
Just using pharmacy as an example, it allows you to find out when was the medication given, what was the medication, what was the dose, who was it given by? You now have information also about the patient's diagnosis. So you are able to get a clearer picture of being able to evaluate the quality of the care that is being provided, short of having to mine through the paper record.
On the lab side, that is less clear. We have a client who has contracted with a lab organization, a lab company that has available all of the lab transactions electronically. They are asking us to come in and help them hook that information to the individual patients.
What we have found is that this return on investment piece is somewhat of show stopper. In other words, competing on quality isn't there yet. We are still competing on price. Until that gets kind of level, the ability to compete on quality is going to have to wait.
In order to compete on quality, we now get to what Paul was talking about, the ability to measure things the same way across diverse organizations, so you have some sense of who is performing well, and who isn't. It gets specifically to the question of what government can do. My sense is to push this issue of competition on quality, to push this issue of performance, and to be able to help people and disseminate the information so that they can make informed decisions, and that they know they should make those kinds of informed decisions in terms of choosing the care.
The irony of all of this, those thousands of clients I showed you all want the same thing. I hate to tell you this as a business secret, but we package the same data for different clients and charge them different prices. But they all want the exact same thing, whether it is me trying to decide who I want to see, or whether it is me running a surgery department, trying to decide whether I'm performing well, or it's me as the CEO of the hospital trying to decide whether the organization is performing well, and explain to the community what I'm doing to them, or me explaining to the FedEx that they should send their people to me because we're doing a good job. It is all the same.
MR. BLAIR: Are there any other committee questions? If not, I have one. Dr. Schutt, you said something that rang in my ears. You wound up indicating that in terms of the business case, your clients, which is basically health care providers that come to you, by virtue of the fact that they come to you to analyze data to determine the value of the health care services that they are performing, that your existence indicates that there is a business case, because MEDSTAT exists as I'm assuming a profit making organization.
Then one of the other things that you said in addition to that is that one of the first things you have to do, a major task you have to do before you can even perform that service is you have to clean up the data. The data quality is something that you have to do, and of course that means that that is built into your business case.
So I have kind of series of three or four interrelated questions with respect to data quality, data capture, documentation, and related this panel, because in the previous panel one of the areas that was mentioned by more than one individual was that standards are not very mature for documentation, and for data quality.
If you are going through this process, number one, I'm wondering if the first time you do this for a client, whether they pick up the clue, and you create de facto standards for documentation, where your clients begin to wind up improving their documentation practices.
Number two, if you are doing it, there are others like HCIA and some other companies that perform similar services. Do you work with them and any standards organization to standardize these documentation requirements?
Number three, the last piece, if there are emerging standards for documentation that you see, does this mean that they are likely to proceed without needing government facilitation, enablement, encouragement, or do you feel like this is an area where the government could play a role to help you?
DR. SCHUTT: Yes, yes, and maybe. When we work with clients to clean up their data, we are in essence, modeling for them how they should behave in the future I think in two ways. One is from an academic pure way of saying, look, this is really how you should be doing it. We who have these numerous clients, we can tell you how other clients are doing this, and therefore, we advise you on a consulting basis, to behave in a similar way.
The other thing, if that doesn't get there attention, is a discussion of how much money this is costing them to be collecting data which is totally unusable. If the first doesn't get them, the second one gets them.
You can also argue from the high moral ground of how much is it costing you to continue to run your organization in an unenlightened way? That's a harder thing to get your arms around. But the answer to your first question is yes.
Part B to that first question is data doesn't get better unless you use it. So what we see is as the data gets better, and people trust it, it gets spread throughout the organization, and it gets used and counted on, and people trust it. But the very first question they ask is, is this stuff believable? Can we believe that what we have actually done inside of our own building is worthwhile, particularly from the provider point of view.
The second question was, since there are other organizations doing the same thing, and the answer is yes, of course they are, is there some place that we get together? The answer is yes, but I'm hesitant to say that it is done in an organized way.
MEDSTAT belongs as an organization to many health informatics groups. We attend a variety of meetings that are put on both by private and public organizations. We present at those meetings. From just plain brownie in motion, you can't help but get some common stuff on you. Now I don't know that that's the most efficient way to do it, but it is the way that MEDSTAT is currently doing it.
Your third question about government, I think that MEDSTAT had an experience back in the eighties where we were collecting claims data for a variety of organizations, mostly insurance companies, and because we were sort of the only game in town at that point, we ended up driving the standard. That's not the same thing in the late nineties and the beginning of the new millennium. When we look around and look at other than claims data, we don't see on the horizon, somebody who is like us back in the mid-eighties.
So I think the government's role is to do what I mentioned in my testimony, and that is we need to make the sidelines a bit narrower. The game is being played in a very broad community. I think that it will take a very long time before there is any coalescence of any kind of a standard.
I will just entertain you with one little vignette, which is in an article that Rosana Coffee(?) authored, which is in Health Affairs of August of last year, where she talked about the Baltimore fire of 1904, and the fact that numerous fire departments responded to the fire. But in spite of that, some 1,500 homes burnt down. Later that year the Bureau of Standards had a fire, of which the night watchman saw it, woke up his other fellow night watchmen and other federal agencies. They came running, and the bottom line of this story was, none of the hoses would connect together. There was no standard.
In the city of Baltimore there was no standard. In fact, across the nation there was something like 400 standards for thread size and diameter of fire hoses. So that unless you had a fire burning very close to a hydrant that your hose fit, you were in trouble. This feels very much the same way.
Now the bad news, if that wasn't bad enough, is that it took decades before there were standards reached on the diameter and thread count of fire hoses. That, I would submit, is nowhere near the complexity of collecting medical data.
So I think, Mr. Blair, to answer your question, is there a government role? Yes, it is to catalyze this, and to make the sidelines more narrow, so that this random motion gets channeled, and we get something done that is relatively small, relatively doable, and I think I've talked enough about the things that we think are important.
Did I get all three of them?
DR. SCHYVE: If I could make one comment about business case. Almost all the discussion that we have had, it seems to me centers on the business case for the health care organization, which is very understandable, because it's the organization that ends up purchasing and paying for it.
But the people who are going to be most affected by whatever we do in terms of standardization in the medical record are in fact the doctors, the nurses, the practitioners. They are the ones who are going to be entering data. They are the ones who are trying to get data from the system, et cetera.
I think it is critical that the business case be made for them. I think not only is it critical that the business case be made for them -- there was a discussion this morning about the issue of the difference between the roles particularly of physicians, but I would argue other professionals in health care, and how they relate to the organization even when they are an employee of that organization. There is a different kind of professional relationship.
I think if we fail to make that business case about why this is good for their doing their work in terms of helping patients get better, that in fact we will have a constant battle. That is another reason why I think this has to be done despite it's a slower process to get consensus standards, because the people who are actually going to be implementing this and using it in their daily lives, those clinicians have to have buy in to the process. That's less likely to occur if it is simply imposed from the top.
DR. WEBMAN: The issue of the data quality, data capture, document piece, I just want to add that supporting and holding public purchasers more accountable is an invaluable piece of this puzzle also. Public purchasers not only need education about how to purchase for quality, but they need reinforcement, requirement, monitoring to do that, because they are compelled by their local communities to buy on cost, not on quality. They don't even know what the parts of the engine are that are quality, never mind what it costs. They know what the lowest price was from the bidder.
This also has its way of playing itself out whether you are talking about a public or a private system, because of the inability to monitor cost shifting across public and private systems. In between public systems and across to the private system.
In the behavioral health field, where many persons who are seriously mentally ill, and children with serious emotional disturbance wind up exhausting their private benefits, and moving to the public side of the fence. We have no way of documenting what the cost of that "episode" of care was.
I would suggest to you there that the movement toward a virtual record, and the creation on the part of the government for security standards, privacy standards for the use of the Internet is going to become more and more important, because the way we probably will integrate care and determine the cost and quality of care across public and private systems will be in a virtual way, it will not be in an absolutely integrated way. So with that, I close.
DR. COHN: After this discussion, I feel like I'm asking a very mundane question. David, I actually wanted to follow-up with further clarification of Jeff's question, just to jog my understanding. Jeff had asked about issues of documentation standards and all of that. I first of all admit that I'm not exactly sure what those are, so I'm at a bit of a loss.
But as I listen to your description about the data and the data quality, which is something I think is critical. I think as we talk about comparable patient medical record information, we can have all the codes, we can have everything in place, and if people use them poorly, we don't have comparable patient medical record information. So you have brought that to my attention.
Recognizing, at least to my knowledge, you primarily use things that are off of HCFA 1500s, UB92s and all of that. I was actually wondering from you view, when you go in and you look at poor data quality, is it a documentation issue? Is it a poor quality coding issue? Are they messing up putting fields in? Where does the problem lie?
Because if it's a coding issue, then certainly there are a lot of standards out there put out by all the coding organizations, that are already sort of standards on how you are supposed to code things, and that's one solution. If there are other things, there are other sorts of solutions. So when you see poor data quality, what do you see?
DR. SCHUTT: It's all of that. It's all the way from the fact that the medical record is -- the documentation of what actually happened from a series of providers that are captured in let's say on a paper record is not there. Things happen to the patient that just didn't get written down. The things that did get written down were perhaps written down imprecisely, or illegibly.
You then have that kind of start to a process where now you have somebody who is looking at the record and trying to decide what actually happened, and then translate it into an ICD-9-CM code for instance, assuming that that went well. Assuming that they interpreted the record, and it did get done correctly. So all of those, Simon, are opportunities for things to unravel.
The other thing I'd like to say which overarchs all of that to some degree is why are they coding in records to begin with? It's for reimbursement, not for improving quality of care. Those are two -- although you would hope they were the same, they are not.
When we go in and say, look, it would really be nice to support the Joint Commission's effort on finding sentinel events and help Lucien Leap(?) with his effort at reducing errors in medicine, and how about capturing all of your complications? They go, we're doing that. We go, really? Well, let's take a look and see.
They capture the complications, which will mean that the care was more expensive, so the get reimbursed more. But they don't necessarily capture the complications which make the care more expensive, and therefore, they won't get reimbursed. Do you understand the difference between those two?
So there is this overarching sense of why am I doing this begin with, on top of a very shaky foundation of medical record documentation which is done by tribal myth, habit, local influence. And then asking this records person to interpret and jam into a pigeon hole, although there are thousands of ICD-9 pigeon holes, what they think actually happened to that person. Us then capturing that, and returning it back and saying, this is a correct interpretation of how you guys actually performed.
So that's what we find, Simon, is through all of that. You hear people argue that the gold standard is the medical records. Claims data is worthless. It depends on what you want to know about. If you want to know -- or that the patient, taking on Dorothy's comment, is an unreliable source. If you want to know whether the encounter went well, and the provider was caring and listened to what I had to say, it's not in the record. It's not in the ICD-9 code. You have to ask the patient.
If you want to know whether a certain service was provided, then the ICD- 9 code can capture that very nicely. If you want to get a sense of what in the world was this provider even thinking, then you need to go to the medical record, assuming you can interpret what was said.
So there are gold standards, and it all depends on what it is you want to ask, but specifically, Simon, to your question, the error is in the system of how we capture what happened to the patient, and how it gets recovered.
DR. SCHYVE: And because so much of the data currently is gotten through manual abstraction from paper records, another big contributor to error is the very lack of standardization of the format of the record. So when we went back, for example, to re-check data that was sent to us, we would find sometimes data was there, but the abstractor had not been able to find it, because it wasn't simply in a standardized format. So even if the data is accurate, you may not be able to find it in the record.
DR. MC DONALD: I always get nervous when standards have always been really effective when they are enabling. Everybody loves them. But they call them laws or regulations when they are kind of required, and they don't always love them. Back in the seventies there was a big activity in quality research, and what invariably happened, it started out with some experts who decided what was necessary for quality. Then they review the charts and they find a 5 percent level quality. It was awful.
Then they looked at the experts, and theirs was 5 percent level. They asked them why, because these were different cases. It wasn't 5 percent, it was that the standards stunk. They didn't know what they were doing. We still don't know a lot of what we are doing about what's necessary.
So what's the process if we do require that it be collected, what's the cleansing process? What's the process of understanding scientifically which are the necessary ones? Because you get experts around the table, I don't know that we have much evidence that that's really a good way to do it.
DR. SCHYVE: I think, Clem, that the question you are asking goes back to the point I made earlier about the difference between collecting data at the atomic level versus at the classification system level, or even beyond that, you are talking about when you start to put together a bunch of data elements to create a performance measure, or a clinical indicator.
DR. MC DONALD: If you're going to say what, say in a surgical note, how do you decide if they are required, and the guy has to spend the time on them, that these are cost effective, they are useful? Because they will come out really out of -- if you get the compulsive-obsessive guys on the committee, you could have 500 items. If you get the loosy-goosy, you'll get six. We still don't know what any document -- I'm not clear, except for the regulatory activities, of any evidence that any documentation helps care.
DR. SCHYVE: I think the theme of your question is very much along the line of what you had asked Peter. The argument I'm making is there is kind of data. Forget about whether we require that it be collected and recorded, can we at least start at the level of saying if this is the piece of data, we all collect the same thing and record it in the same way?
Secondly, in terms of formats, if we want that piece of data, where will we put it so that people can find it. We know whether it's a written record or an electronic record, what field or area you would in fact record that piece of data in. We're not even at that level at this point.
The reason why it becomes difficult to go beyond that level is for exactly the reason that you have described. It is also the reason why I think there needs to be both research and consensus in deciding what's going to be the minimum piece of information that you have for example for operative procedures, so there is some consensus about what that will be.
My point is though that that consensus, as we get more information, will in fact change. The science behind it will change. Can we in fact develop a data dictionary for data elements, in a format of where we put information that as the science changes of what information we want, we can easily regather that information out of an electronic patient record to turn it into the new operative report?
If all of the sudden we discovered something recorded over there, that really should be part of the operative report, or the new performance measure in terms of how well are we doing, whether it's in terms of processes or outcomes.
DR. MC DONALD: Let me just clarify. I just think we ought to always have in these processes that are required, some cleansing process, to be able to throw them out when we learn they aren't, or some way to really be sure we're right in the first place.
MR. BLAIR: Let me interrupt at this point. It's just about 12:45 p.m. We're about 15 minutes past time. This has been another excellent panel. If I've cut short some of the questions and comments, please, I would hope that our witnesses would be available during lunch, and we could kind of continue some of those conversations on a one-on-one basis.
Since we are at 12:45 p.m. now, we will convene at 1:45 p.m. That will give everybody one hour to get through the lunch facilities here.
Thank you everyone.
[Whereupon, the meeting was recessed for lunch at 12:45 p.m., to reconvene at 1:45 p.m.]
MR. BLAIR: Welcome back. Why don't we start with each member of the panel introducing themselves and your affiliations?
DR. CHIN: My name is Homer Chin. I'm the medical director for clinical information systems for Kaiser Permanente of the Northwest, based in Oregon and Washington.
DR. MATTISON: John Mattison, the national director for Kaiser Permanente of clinical documentation.
MS. NARCISI: I'm Jean Narcisi. I'm with the American Medical Association, director of the electronic medical systems.
DR. ORIENT: Jane Orient. I'm a private physician, and also the executive director of the Association of American Physicians and Surgeons.
DR. SPACKMAN: I'm Kent Spackman. I'm associate professor of pathology and medical informatics at Oregon Health Sciences University, and I'm the scientific director and chairman of the editorial board for SNOMED for the College of American Pathologists.
MR. BLAIR: Homer, would you like to start?
DR. CHIN: Yes, as I mentioned, I'm Homer Chin. I'm the medical director for clinical information systems for Kaiser Permanente of the Northwest. Today I am going to be talking from this handout. Most of you should have it on your desk. It's a Power Point handout, two slides per page.
I'm going to do something a little different from the previous speakers. I'm going to talk mainly about the business case for standards, and really key in on that. Then I'll talk about some of the prioritization from my viewpoint as to which standards should be prioritized over others.
The outline of my presentation, I'm going to talk a little bit about Kaiser Permanente, and how Kaiser Permanente is different from some of the other health care organizations in the sense that it is integrated, and it is comprehensive. It provides comprehensive care in an integrated fashion. This gives us some strategic advantages in building information systems.
We in fact do have standards by virtue of the fact that we have only a few systems that de facto become standards. Then I'll talk about data standards in health care, and the ability to prioritize the different data elements, and to really look at which data elements should be prioritized above others.
The second page shows Kaiser Permanente. The organization overall is the largest not-for-profit group practice health maintenance organization in America. We are in 18 states and the District of Columbia. We provide care for 9.1 million people in the United States. We call them members. These are people that seek care through Kaiser Permanente in a capitated fashion.
Importantly, we're an integrated health care system, where the insurance, the administration, as well as the provision of care is provided under the same roof. It is a comprehensive and integrated system.
The next slide shows Kaiser Permanente of the Northwest, which is the region that I represent. It is in Oregon and Washington. We cover 430,000 lives, and again, we're an integrated health care system, which is very important, and I'll elaborate on the subsequent pages.
On page 3, the slide on the top really shows you how Kaiser Permanente is integrated. We have Kaiser Health Plan and Hospitals that runs the insurance, as well as some of the operations. And we have the Permanente medical groups. There is an exclusive arrangement between these two groups to provide care for our members. We really do act, even though we are legally separate entities, we really do act in concert as one organization to provide care, with the full responsibility for providing the medical care, the full responsibility for the medical decision-making rests with the medical group.
In terms of information systems, we have some significant advantages over other organizations. We have a single member identifier, so that if a member goes to the outpatient clinic, to the lab, to the pharmacy, to the inpatient hospital, if they go for dialysis, if they get a kidney transplant, whatever it is that they get in terms of medical care, the single member identifier follows them, and we're able to then get that information and consolidate it.
We're a single organization delivering comprehensive health care, which means that for every segment of care that a patient goes to receive, they receive it under the Kaiser Permanente umbrella, unlike some of the other organizations that are more loosely affiliated. You might go to a hospital to get care, but then go to your primary care physician, and then go to a specialist, and those three entities don't talk to each other. Within Kaiser Permanente they are all one entity already.
We have single organizational ownership of systems, which obviously makes it easier to build interfaces between these systems, and to get these systems to talk to each other. We don't have the political problems where you have one health care organization saying well, this is our information. We're not going to give it to you, and another organization saying we've got our information, and we're not going to divulge it.
So to a certain extent we have de facto standards by virtue of the fact that for the most part, we have single systems. We've got a single laboratory system. We have a single pharmacy system. We have got a single radiology system. Whatever data standards those systems are using, are essentially the standards for the organization.
Then again, we have comprehensive health care, so we have complete clinical information on our members. So when we go into our systems, and we look at the information, it is for the most part, complete, as opposed to other organizations where if you are in the hospital you will see the inpatient information. If you are in the clinic, you might see the information from that particular clinic's vantage point. Within Kaiser it is comprehensive, and it is integrated.
The next slide shows were we were in 1992. We had essentially these islands of automation. We had a pharmacy system, a lab system, an appointment system, a tumor registry, transcription system, and membership system. None of these systems really talked to each other. These were systems that were built to support the function of the individual department. The problem is that the clinician was not able to access this information at the point of care.
So in 1993, the next slide shows we consolidated this information. We built interfaces from all these systems into a single system called results reporting. Results reporting is then accessed by the clinician, and the clinician has access to this complete, comprehensive information at the point of care.
Now there is another piece on that slide, which is outside reports. Around this time, we started to affiliate with some of the outside community hospitals. So one of the things that we did is we built an interface from their transcription systems into our systems. Now one of the reasons why we were able to do this is because of the unique patient identifier.
When that information comes across, we know which patient it belongs to, and we are able to insert it into the electronic chart for that patient. But at the same time, the standards of those transcription systems are not completely conforming to the standards that we have for transcription, and that does pose a slight problem, which I will go into in the subsequent slides.
The slide on the bottom shows the opening screen, the main screen for our results reporting system, where you can see that information is coming from pharmacy, from reports, from laboratory, from the hospital, from visit information, immunizations, demographics. We have different views of the information.
One of the advantages of electronic information is you can really define different views. With a paper record you are stuck with the way that the paper record is put together. With electronic systems you can build different views that look at the data from different vantage points.
The next slide shows all events, which is basically everything that has happened to that patient in reverse chronologic order. So whether it is a prescription, an appointment, a lab test, a report, all of that is able to be consolidated and displayed in reverse chronologic order so that you get a good idea of what has happened to the patient, no matter what it is that they obtained. Again, the advantage that we have is that this information is complete and comprehensive.
The following slide shows that if you just wanted to look at the medication list, you could go ahead and hit a button and look at the medication list.
The next slides shows dictated reports that are available in the system. The first three items you will see have the label "dictated by" and it says "unknown doctor." These are the reports that are actually coming over from our affiliated hospitals. These are not reports that are coming from within Kaiser Permanente. If it was coming within Kaiser Permanente, we would know who the clinician is and put it in.
But because we have information that is coming across from our affiliated facilities, and they do not conform to the exact same standards that we do, we don't know who the doctor is, and so we have to put in unknown doctor.
On the slide at the bottom you can see the emergency room report that is coming across from our affiliated hospital, Providence St. Vincent Medical Center in Portland. You can see that it's an emergency room report, and the doctor's name is Tom Caverley(?), but at the tope of the report it says "unknown doctor." Our systems are not able to parse that name out to place it into the summary title for that report. That's one of the issues with standardization, is the ability to do those sorts of things, and insert information into the record.
By having the unique patient identifier, we are able to consolidate the information, but in order to get further down into the information, you need standards to further specify those reports.
The next slide shows a laboratory display, which shows the headers for each of the laboratory tests. Also in reverse chronological order you've got diabetes studies, lights, lipid studies, quantitative urine chemistry studies, stool occult blood. This gives you the titles of the tests. Then if you wanted to look at the body of that test, you could hit a button, and the next slide shows the results of the diabetes test. Then if you wanted to look at a trend of diabetic results, you could hit a button and the following slide on the top of page 9 shows the trend of laboratory tests.
Now many vendors can do this. This isn't unique to us by any means. The thing that distinguishes us is that when you do these sorts of things, and you look at trends, and you look at information, for us the information is comprehensive on this patient. It includes specialty lab tests, primary care lab tests. It contains lab tests from two and three years ago. It is really a patient-centric view of their information, as opposed to a provider-centric.
For many other organizations if you're on the inpatient side, you would just be seeing inpatient labs. If you are on the outpatient side, you would be seeing labs from that particular office. Because our organization is integrated and comprehensive, this information is comprehensive information on these patients.
The bottom of page 9, that slide did not print out very well. It's a summary printout from results reporting. For every patient visit, we print out a summary that comes out of results reporting. But on the following slides I will show you the detail that comes out of that summary. So the fact that you don't have the summary displayed is not a major concern, because on the following slides this shows you the different sections that are in the summary.
The first section shows health screening information, the last mammogram, the last pap test, and the last flexible sigmoidoscopy, as well as immunization information on this patient, as well as the guidelines that go along with those tests. So you can look here to see whether the patient is due for a test or an immunization. Again, because of our comprehensive information and database, we are able to pull this information together, and for the most part, it is complete.
The slide on the bottom shows a cholesterol trend, as well as a diabetes trend. If the patient is diabetic, the diabetes information is assembled, so you can see exactly how they are doing in terms of their diabetic labs, in terms of their kidney function, and the last time they saw an eye care physician, with the annual recommended screening for diabetic retinopathy being once a year.
Again, it is this ability to view the same information in different ways that really makes this powerful. The fact that we have comprehensive information makes a big difference.
The next two slides on page 11, the top shows a lab summary section. What we do is we print out the last time they had any type of lab. So the last time they had a diabetes study, the last time they had a CBC is printed out. Then at the bottom, if any of those lab tests -- the last three results from those lab tests are abnormal, we print out a flow sheet. So in just a few lines we can really summarize the laboratory status of that patient.
The report on the bottom shows all the reports that exist for this patient. As you can see, the reports date back to 1991. When we went live in 1992 and 1993, we actually backloaded a lot of the transcription information that existed, so the information really goes back to 1991.
Then at the bottom of that slide it shows any hospitalizations, any emergency room visits, and the last six outpatient visits for that patient. So in one report we really do summarize in a sense, the entire medical history of that patient.
The next slide shows what we are able to do with some of our information. Because it is complete and comprehensive, we are able to do population management of our diabetic members. These actually happen to be the diabetic patients in my own panel. I still practice about 20 percent of the time. I have about 600 patients in my panel.
What we do on a monthly basis is we go through the list of patients, and we list out the patients. The first four patients are patients that have not had visits to primary care in the last year. The next three patients are patients that have not had appropriate diabetes testing in the last year. Then there is another section, if I had any patients that more than four admissions or ER visits within the last year, they would be printed out there. Then you've got a section with patients who are under poor control in terms of their diabetes. Then at the bottom, the patients that are under moderate control for their diabetes.
We have care managers that go down this list, and they actually contact these people who haven't come in for the last year. They know exactly how they are doing through this printout, and they ask them to come in to be seen, or to get the appropriate laboratory tests. They follow them to make sure that their diabetes control does improve.
The effects of this are shown at the slide at the bottom. We were running at about 45 percent in terms of members with good or excellent control of their diabetes in 1992, and through the use of these systems, both the population management, as well as the tools we have in results reporting and now epi care, we have brought up the percent of members with good or excellent control from 45 percent to 65 percent in 1997. That's a very significant and dramatic improvement.
The next slide on page 13, shows that we do similar things with mammograms and pap smears. We are able to identify through our systems, all the patients who are due for pap smears, who haven't had pap smears. We have been able to increase the pap smear screening rate from 69 percent to 80 percent over the course of three years.
We have three components to our computer-based patient: results reporting, and the prevention and disease registries, both whom I've talked about, and then finally I'm just going to touch very briefly on epi care. Epi care is our comprehensive, computer-based outpatient record, which we have implemented in 1995- 1997.
The following slide shows you what we do with our epi care system. It's not only a computer-based patient record, but just as importantly, if not more importantly, it really computerizes all the processes that occur in the outpatient arena. So we use it send prescriptions to the pharmacy, to order labs and radiologies, to refer to specialists, to message each other.
The other important thing is it does integrate decision support into the process of care, because we have clinicians using this system to actually do the work of medical care. It's quite easy to integrate decision support into the process. The slide on the bottom shows that when you order an upper GI for a patient, a guideline will pop up, guiding you as to the appropriate use of upper GI imaging.
The next page shows that we do a similar thing with pharmacy. If you order a medication, you get a guideline if it's appropriate, that guides you onto the use of these medicines. The thing that I would emphasize is that the final decision in terms of what to do is really left up to the Permanente physician. As I said, health care is controlled by the medical group, it's not controlled by the insurance company. So the physician is allowed to use their best judgment, with these guidelines helping them to decide what the best course of action is in a particular situation.
The slide at the bottom shows laboratory test utilization. The number of outpatient lab tests per member per year has dropped from 17.4 tests in 1993, down to 16.1 tests in 1997. That is a drop of about 8 percent. We believe this is happening because clinicians have access to information, and they are not redundantly ordering tests that have been ordered already, simply because they have access to the previous information on that patient.
The next slide on the top of page 16, shows the number of outpatient visits per member. The number of outpatient visits per member per year has gradually been increasing from 4.4 visits per member per year in 1986, up to 5 visits per member per year in 1995. Since 1995, with the advent of results reporting and epi care and our other systems, the number of visits per member per year has dropped to 4.8 in 1996, and 4.6 in 1997.
There are a number of reasons we feel this is happening. One is when people call up over the phone, because of our immediate access to their chart, we are able to handle many more situations over the phone. The other thing is that when they do come in to receive care, because we have comprehensive information on them, we are able to treat them more effectively and appropriately.
The bottom slide shows that we do, to a certain extent, have de facto standards at Kaiser Permanente. We already have a single patient identifier. We have single systems, that by default, are the standards. We have complete information on our members. The results of this is that we are able to consolidate information. We are able to improve quality and the appropriateness of care. WE are able to decrease unnecessary variation.
I've got slides that demonstrate both the improved quality adherence to guidelines, as well decreased utilization of certain tests. We have decreased the outpatient office visits. We are able to do population-based prevention and disease management. The bottom line obviously is improved overall quality, with less cost.
The next slide shows our view of the prioritization of health care data standards. Number one on this list has to be the unique patient identifier. If you don't have a unique patient identifier, and you get information from different systems, without that identifier it is very difficult to consolidate that information together. So that would be the number one thing, is the unique patient identifier.
The second piece in terms of prioritization would be to have simply dates, labels, and bodies of information. So that if you get some information, it might say well this test was done on December 6, 1997. This is a CDC. Then you've got the body of the information. That allows you to assemble the information and put it together, so that you can at least display it to the clinician.
The next thing would test and procedure naming standards so that you can see on a summary level, what information it is that you are looking at. Then finally, the results standards, so that you can then compare results between systems.
Diagnosis code standards are obviously important, and other standards are helpful as well. Our view is that it is not necessary to code everything. It is helpful to code those elements with known benefit in both decision support and outcomes analysis. If it is a piece of information that you are going to use for something, then it is helpful to code it. If it's a piece of information that is not going to be used in any decision support or outcomes analysis, then it becomes much less useful.
One of the things that has emerged as a technology is a technology called XML, which allows you to basically structure and transfer a chart from one system to another at various levels and degrees of specificity. My hope is that one day I might be able to hand somebody a floppy like this and say, this is your medical record. They can take it and go to a different system, and say here is my medical record, and insert that information into the other system's medical record.
Now with XML, depending on the level of specificity, you should be able to do that. There are a couple of levels of that. If it just has a unique patient identifier, it might a unique patient identifier, and the rest of it is just simply ASCI text. If you have the dates and labels of the different events, it would be the unique patient identifier, and then with the dates and labels you could integrate it with the other information in reverse chronologic order.
If it is dates, labels, and the naming standards for the different events, then you can integrate and at a summary level, look at exactly what has happened to the patient, at that level. So I think there are different degrees of standardization, and the unique patient identifier really is the important one.
That's the end of my testimony. This is sort of a glimpse into the benefits of data standardization, and I think to a certain extent we are ahead of the curve, and we are demonstrating some of the benefits that exist with data standardization in health care.
Thank you.
DR. COHN: Homer, thank you very much. I guess I should remind all of our speakers to keep it to ten minutes, plus or minus, so we have time for discussion at the end.
John?
DR. MATTISON: My background is I'm an internist. I was director of one of the critical care units of Scripps Clinic. I was in the fee-for-service world for five years with them. I have been with Kaiser Permanente for 10 years, and managed the clinical systems for Southern California during the last six. I'm now participating on our national integrated approach to clinical information systems. I have also been actively involved in either the design, development, or implementation of five different electronic health record systems over the past ten years.
What I would like to do is walk through what I believe are some of the outstanding problems and issues that need to be addressed from the perspective of an integrated delivery provider. The first is really the statement of the problem. What is the problem that needs to be solved? I think to summarize the first slide, I was tempted to retitle it, "From Fire Hoses to Space Shuttles," since those both seem to be the metaphors du jour.
We need an overarching architecture to put the components of the space shuttle together. Some of the problems are as simple as fire hoses; many of them are not. Putting together a space shuttle, you cannot simply put out RFPs for each of the components and expect that when they are all built you can put them all together and the thing will fly. So it requires quite a bit more I think, than some of the simple standards that would still be very helpful.
The second aspect of the problem is that there are a lot of excellent initiatives underway to develop the standards for pipe threads and pipe diameter. What is difficult is to coordinate some of those activities so that the gaps and conflicts between and among them can be identified and managed to create this overarching architecture.
The next slide refers to what I view as the strategic objectives from the business standpoint in an integrated delivery system for managing electronic health records. The first is mom and apple pie. Create a unified longitudinal record to support coordinated care delivery and work flow between institutions over time. This is the antidote for amnesia that was mentioned earlier today.
The second strategic objective is to create clinical data as a substrate for advanced decision support, so that when we do find out what needs to be done, we can implement it on data that allows us to have a specificity in the match between a rule and the clinical context. If, every time you pull up a diabetic patient a rule fires that says, ah ha, a diabetic, check a foot exam, check a flutosamine(?), check their eye grounds.
Providers get overwhelmed with alert desensitization or sensitization and then desensitization. So that there has to be a high level of specificity between the rule and the clinical context. It is important that the clinical data be specific enough that they serve as a good substrate for the rules base.
Third is to create clinical data as a substrate for population-based outcomes analysis. And finally, combining the last two elements into support for continuous quality improvement cycles. So as Clem was referring to earlier, we really don't know what to recommend in a lot of circumstances. We first have to determine it, and then be able to implement it. As a byproduct of that, it supports both internal and external reporting, which is a lot of what the governmental role would seem to be.
So I believe there are three prerequisites for fulfilling those four strategic goals. There are different models. The first is a user interaction model. What we need to do is to be able to allow the clinician to document clinical encounters with a level of specificity that provides machine readable information that can be processed, or either outcomes analysis, or real time decision support.
The second is a terminology model, and Kent Spackman will get into more detail on that later. We need to support automated encounters and somatic normalization to know that heart failure means heart failure means heart failure, and that asthma means asthma means asthma across institutions, and across applications and platforms. That's a fairly simple issue, but a fairly difficult solution.
The third is an information model that supports the transfer of the full semantic richness of notes across applications, platforms, and institutions over time. I'll come to that a little bit later. I'm going to review each of these in just a little more detail.
The user interaction model must be at least as efficient as a paper-based data entry. Paper is incredibly efficient for capturing information. It's lousy at representing that information, at sharing that information, at repurposing that information, but it is great for capturing. So the gold standard is pretty high in the aspect of capture.
It must also support structured entry to high level clinical detail, and it must support automated encoding linked to standard reference terminologies.
The terminology, there is one key point I would like to make here. We need to distinguish between the reference terminology models, and how that is implemented in a particular application, and Keith Campbell, who works on our team, has published some very nice papers making this distinction.
We know we need at least 300,000-400,000 terms to really be able to clinically represent the diversity of clinical encounters. It has been estimated there are 8,000 new terms added to the lexicon each year. But we need to be able to develop and maintain this rich set of terminology in a reference environment that is optimized for that purpose. Yet when we go into a specific application to document what is going on, we cannot be browsing through 300,000 terms. They have to be represented in the relevant clinical context.
An example, in your reference set you need to have color as an entity, and you need to be able to represent all of its variations across all of clinical care. However, if you are documenting the color of appendix, you don't need nearly the selection set, nearly the number of color options that you would do for example to represent the color of a melanoma, or the color of the iris. So those configurations of the reference terminology need to be very much adapted to the clinical context and the specificity.
The third is that the reference terminology must be able to be deployed in different interface configurations across multiple applications within a single institution, and between institutions.
Finally, the information model must be based upon an industry standard information model. A great example of that is the reference information model of HL7. The information model must support the rich semantic context within a hierarchy of contexts. Let me use FUNDUS(?) as an example.
If you say FUNDUS to an ophthalmologist, they know exactly what you mean. If you say FUNDUS to a gynecologist, they know exactly what you mean, but they don't bear any resemblance to each other. So FUNDUS in the context of the eye is a very different thing than the FUNDUS in the context of the uterus.
Furthermore, there is a different context in past history versus current findings, and furthermore, there is a difference between the context of it being in a progress note or an operative note. So being able to represent that semantic richness is critical to really understanding what you are talking about. Otherwise, if you seek a report of how many FUNDUS exams have you had in the last three years, you are going to get the gynecologic and the ophthalmologic exams, and that is not exactly what you are after most likely.
So you must be able to support variable levels of detail based upon the business needs of either end of the transaction. What Homer is referring to the in the XML standards activity, which is something that I launched about three years ago, it allows you to be able to specify detail at varying levels of detail. So the business needs on either end of an information transaction can determine the level of specificity required for that transaction, and you don't have to have it all solved before you can start down that pathway.
Finally, you must enable these clinical data transmissions between applications and platforms over time. An XML, one of its virtues is that it is application and platform independent, so that when Homer says I can hand you a disk, it doesn't matter what kind of platform or application or institution; that is readable by the same parser.
I would like to shift a little bit in the next slide to some of the major shifts that I see coming as we succeed down this pathway. First of all, the task of coding is going to shift from coders to clinicians. Currently, there are armies of coders out there wandering through paper notes. That was referenced in the talk this morning.
The ability to do some coding behind the scenes as well defined terms are entered in a clinical context is going to cause a great shift. It's not going to eliminate coders. It's going to cause a shift to clinicians.
The second is a shift in the responsibility for care from a provider-centric model to consumer-centric model. Homer referred to that also. The consumers want to know what is going on with them. They want to know what the knowledge base is, and they want to participate in decision-making.
Finally, there is going to be a shift in the universe of knowledge. Currently, the universe of knowledge that is applied to every clinical decision resides in the head of that individual clinician, and that knowledge base is going to shift the network and all the decision support providers. These are really huge shifts.
The next slide talks about some of the business incentives that are in the marketplace to achieve these goals. I have sorted them by health care provider, software developers, payers, employers, consumers, and so forth. I would like to focus on the bottom two bullets, where the incentives are convergent and where they are divergent.
The convergence is on low cost, high quality health care, and on a short- term return on investment. The divergent incentives have to do with commercial product differentiation versus adoption of standards. Those are in conflict.
The second is time to market, versus high quality health care. Rushing things to market does not get you very close to the target, and in many cases, paints you into blind corners.
The third is rapid return on investment, versus size and cost of the task. This is a concern I know for HCFA. And speaking to Bob over lunch, the true size of this as a space shuttle, as opposed to a fire hose is quite large, and we need to be careful not to rush it, or we may find ourselves in some blind alleys.
So I'm going to review some of what I see as the major outstanding issues. Privacy, confidentiality, I heard the admonition to keep that out of bounds, so I'll just skip that.
The mobility of consumers is one that absolutely requires that we be able to provide information in ways that are representable across institutions. One of the driving forces for me in starting the XML activities within HL7 has to do with the fact that in practice in San Diego, the chances of my seeing another patient who is from a different region of Kaiser Permanente is far lower than the probability of my seeing a patient who came from across the street at Sharp Health Care or Scripps Clinic or Restaly(?).
So if we solve all of our standards problems within Kaiser Permanente, we haven't addressed the 10-20 percent annual turnover of patients as they change jobs, and shift between employers, and therefore shift between health care providers. And the ability to have application platform independent representation of their information is essential to be able to operate on that information for decision support and population- based outcomes analysis.
The next issue has to do with changed management. The changed management -- and this is a lot of where I live on a daily basis -- the changed management required to implement these systems is enormous. It is directly related to the impact you have on people's work flow. As we move from a paper-based paradigm to an electronic paradigm, we enable a lot more team care, a lot more coordination of care. And as we do that, we need to know what each of them are doing, and we need to support that work flow.
Of necessity, we need to have a pretty major impact on the work flow of individuals, and so the resistance to change is quite high. Because of that, if federal regulation prescribes systems that have specific impacts on work flow, the ability to sell that to the individual, and to implement systems, and manage the change associated with that is going to be hampered.
So my prescription from a regulatory standpoint in that respect is to set what the objectives are, to set time tables, but if they become too much more specific than that, and have direct consequences for work flow, it will be a significant impediment to the implementation of change management required for these systems.
The last major outstanding issue has to do with the gap between the magnitude of the task -- the space shuttle versus the fire hose -- and the need for incremental demonstration of return on investment. The task of developing the architecture, deploying an infrastructure that allows us to put the components together into the space shuttle is quite daunting. We need to be able to build that so it supports decision support and population-based outcome analysis.
The need to demonstrate rapid ROI in the health care industry today is very high. Long-term horizons have shortened, given the financial dilemma that health care finds itself in today. So the business cycles that are driving investments are much shorter than the implementation cycles for a lot of these systems. So we need to be sensitive to that.
The last slide just has to do with some general thoughts on opportunities for regulatory influence. First of all, I think it's important to identify from among the existing standards initiatives where there are conflicts and gaps in the existing initiatives. Where there are unleveraged opportunities. To identify what business drivers and incentives for the information infrastructure need to be supported, encouraged, funded.
The third bullet is to endorse or fund specific solutions, only after there is a moderate level of consensus on this issue. What I would like to do is use terminology as a example of this. I was reading on the plane on the way out here yesterday the current issue of Harvard Business Review has a briefing on the euro problem. It's the first I've ever heard of it.
They have a problem probably as big as Y2K that has to do with their conversion to a single currency. The problem is that they are incrementally implementing the euro, and so if you imagine the currency conversion issues between the various currencies and the euro, and the implications for database storage, and the rounding errors, as you go from one currency to the other, they set a standard of at least three decimal points for the rounding error.
But if you translate from the franc to the mark and back again, you don't get the same number. They are just beginning to understand the complexity and the severity of this whole issue for information systems for things as simple as general ledger profits and losses, and all the things associated with that.
I think there is a tremendous parallel with what we need to address with terminology. If we continue going forward with multiple different terminology sets, and we're triangulating back and forth between the sets, there is a rounding error, if you will, with each transformation. If we are going back and forth with these sets, there is degradation of the semantics at each step of the way, and I think there is a very compelling case to choose.
This, I believe is as simple as pipe diameter and thread size. There is a tremendous need to establish a central reference framework to do this in. It does not imply the elimination of the other existing sets, but we do need to have a central reference point.
The fourth is that in defining the market incentives that will help accelerate the adoption and implementation of standards, and to really address those incentives wherever possible through regulatory activity. And finally, with the privacy issue, to engage responsible consumer advocacy groups to participate in the policies related to privacy and confidentiality.
One of the things that I would just highlight on that, as was discovered in Chicago last summer, is that you cannot isolate the privacy and confidentiality issues from a lot of what we are trying to accomplish. They are pretty pervasive, and engaging these responsible groups I think is essentially through the whole process of building this architecture and infrastructure.
Thanks.
DR. COHN: Thank you very much. Jean?
MS. NARCISI: My name is Jean Narcisi, and I would like to thank you for the opportunity to testify on behalf of the AMA. My statement summarizes the views and concerns of the AMA on issues related to uniform data standards for, and electronic transmission of patient medical records information, as well as the appropriate role for the federal government in addressing these issues.
The AMA has a policy on the computer-based patient record, which you can see in your handout. Briefly, it just outlines several steps which will allow the AMA to access the source of physician input, and the AMA will provide leadership on many of these issues. We will work with the state and specialty associations. We'll define the characteristics of an optimal medical record system. We will plan to focus on the CPR aspect of human-computer interaction, and work with software vendors.
We will continue to be involved in the forums related to subjects such as medical data control, access, security, and confidentiality. And of course we will continue to work to insure that issues of patient confidentiality and security are continually addressed prior to the implementation and use of the CPRs.
In response to the questions we were asked to discuss, how do you interpret the congressional instruction? It is important to note that the NCVHS's responsibilities on this issue are fundamentally different than those for HIPAA administrative transactions. There is no federal legislation requiring implementation of medical record standards. Moreover, it is not at all clear if the HIPAA for administrative transaction standards, which is intended to address interenterprise communication, is fully applicable to medical record standards that involve a large intra-enterprise focus.
As you know, HIPAA is intended to improve the efficiency and effectiveness of the health care information system through the establishment of standards and requirements for the electronic transmission of certain health information. Since the NCVHS was asked to study the issues related to the adoption of uniform data standards for patient medical records, the AMA believes that the NCVHS should also continue to study and identify privacy, security, and access measures to protect individually identifiable health information in an environment of electronic networking and multiple uses of data.
These measures must be in place to assure a secure and ethically principled electronic environment for patient data. Because the secretary is likely to rely on the advice of the NCVHS, it is essential that your recommendations be carefully evaluated and thoroughly deliberated with these precautions in mind.
In response the second question, a recent AMA survey indicated that only 14 percent of physicians have electronic patient record systems in their primary practice. This survey didn't define an electronic patient medical record, therefore, this number could actually be much lower. These data imply that there is a great potential for the health care community to acquire health information systems in the future.
Although there are numerous efforts underway to standardize key elements of an electronic patient medical record, we cannot overemphasize the critical need for physician input. Without this input, the electronic patient medical record system will be modified in an attempt to fit physicians' applications, rather than developed specifically for the profession that is key to their creation and use.
We realize that one of the biggest barriers is the current status of standards. It has been mentioned several times in the previous panels regarding medical vocabulary, identifiers, data exchange, and privacy and confidentiality. But we also believe that standards are lacking in the categories of authentication, which has been discussed, system interfaces, and interoperability.
But this latter deficiency is especially troublesome, because of the underlying technology and infrastructure of an electronic patient medical record system, and these must incorporate the ability to communicate between one system and another.
Although many organizations are addressing standards and issues related to electronic patient medical record systems, no organization is focusing on the necessity for vendors to incorporate the specific needs of the practicing physician in the ambulatory care setting. Since there are no agreed upon standards for electronic patient medical record systems, the vendor is creating a variety of electronic patient medical record systems that are often incompatible. This situation leaves little guidance for physicians in selecting systems.
This incompatibility is especially acute, because clinical efficiency often demands that multiple programs be available on a single work station, and unfortunately, vendors have not effectively dealt with this need for multiple systems in the smaller physicians' offices, and physicians must switch from program to program.
The AMA believes that a true private sector approach to standards development is needed, with the role of the federal government much more sharply limited than has been the case with the development of administrative transactions. The federal government could play a useful role, however, in helping to identify any gaps in standards, as well as helping to facilitate and prioritize standards development. At the same time, we state emphatically that the AMA does not believe that federally mandated standards are likely to be an appropriate or productive course.
Cost may also be an obstacle for delaying the widespread implementation of electronic patient medical record systems. In addition to the purchase and installation expenditures, physicians also expect an initial increase in administrative costs, as well as a decrease in productivity.
The response to question number three, it is clear that benefits can be derived for electronic patient medical records. At the same time, the AMA recognizes that there are important issues related to potential impact on physicians' time, cost, patient care, and confidentiality.
As indicated previously, the AMA believes that standards will evolve, and should be allowed to do so in the private sector. Automation of clinical data will occur at different rates throughout the health care industry. Larger health care delivery systems such as hospitals, integrated health networks, and managed care organizations will continue to lead the trend. This trend will be slower in the ambulatory setting, with the larger practices automating first.
Although the requirements of institutional systems often overlap with those of non-institutional and physician systems, they are not identical. Therefore, electronic patient medical record systems should be continually evaluated in order to meet the specific needs of physicians and patients.
Based on the current environment and obstacles, the AMA believes that it is really premature for the NCVHS to suggest specific standards for electronic patient medical records systems, particularly in the context of potential privacy and security federal legislation in subsequent regulations.
At the same time, there is a need for accelerate progress in the electronic patient medical record systems by the vendor community. This progress should address the areas of interfaces, interoperability, and authenticity, as we discussed previously. It could focus on developing a framework for the specification of clinical data architecture, which was discussed earlier this morning, in looking at some of the concept models.
In response to question number four, the AMA believes that standards are deficient for electronic patient medical records in a number of areas, as I mentioned earlier, in particular, medical vocabulary, identifiers, data exchange, and privacy and confidentiality. The AMA believes that there is no single solution to the vocabulary needs for computerized patient records. We see, consistent with previous NCVHS discussion, the need for interlocking code sets, with each code set addressing different aspects of the vocabulary issue.
As you know, we are working on the development of the next generation of CPT, and have taken initial steps with the National Library of Medicine and the College of American Pathologists to map CPT4 to SNOMED. CPT5 is explicitly designed to meet the HIPAA criteria for a longer-term procedure code set for administrative and financial transactions, and a promising and integral element of the coding sets needed for computer-based patient records.
The CPT5 project is examining several different needs: making additional changes to CPT terminology to reduce the ambiguity, and accommodate the needs of non-physician users; explore how managed care organizations can make use of the CPT code set; make changes to CPT terminology as a result of moving towards the computerized patient medical record; make changes to CPT terminology as a result of the increase interest in developing and using practice guidelines, record cards, uniform data sets, and other quality measurement devices; and make changes to the editorial process to allow for the open exchange of information.
The CPT5 project is structured to capture input from a variety of sources. Participation on the project work groups include coding, payment, and health system experts from a variety of backgrounds, including clinical, data management, and research from both the public and private sectors. Representatives from the National Center for Health Statistics and the Agency for Health Care Policy Research, as well as a liaison from the National Library of Medicine also participate in the work groups.
There are currently topic-specific work groups involved in the CPT5 project including: management care, research, non-physician practitioners, maintenance and education, sites of service, and structure and hierarchy. The AMA recommends that the CPT5 code set be designed to meet the diverse needs of its various users.
In addition, the AMA believes that data content standards for claims encounters and attachments should adhere to the recommendations of the National Uniform Claim Committee, the National Uniform Billing Committee, the National Council on Prescription Drug Programs, and the American Dental Association.
As we have discussed in previous testimony, it is essential to recognize that the data formats are separate, and are viewed as messages, as compared to the data content. Therefore, we strongly believe that the groups such as the NUCC, NUBC, and NCPDP, and ADA should make content decisions.
Since no standards exist for the computer-based patient records, claims data will continue to be used to capture some of this detail. The NUCC has done this with their data set, and is currently revising its protocol and expanding its membership, and they are looking to include public health and the research community, as well as the HL7 standards developing organization. The AMA is highly supportive the NUCC's activities, and the efforts have resulted in the development of the standardized data set that defines the set of data elements that physicians and other health care professionals should submit with claim and encounter transactions.
It actually could provide a blueprint for vendors to build into their electronic patient medical record systems for administrative transactions. It clarifies to providers what is required on a claim for all payers. It provides guidelines to clearinghouses and payers on the standardized components of a claim.
Besides recommending the CPT5 code set, the NUCC data set for the current and future HIPAA applications, the AMA realizes that HL7 is widely used in current health care software systems, especially in the hospital environment. Therefore, the AMA recommends that HL7 should be a key standard used by vendors when building electronic patient medical record systems, however, as stated previously, the AMA believes that standards will evolve, and should be allowed to do so in the private sector.
In response to question number five, the AMA believes that the four focus areas identified by the NCVHS are extremely critical areas to be studied, however, as stated previously, in addition to studying the issues relating the adoption of uniform data standards, the AMA believes that the NCVHS should also continue to study and identify privacy, security, and access measures to protect the individually identifiable health information in an environment of electronic networking and multiple uses of data.
Thank you for the opportunity to present the AMA's views.
DR. COHN: Jean, thank you very much. Jane.
DR. ORIENT: I'm Jane Orient. I am a physician, an internist, and I am here representing the Association of American Physicians and Surgeons, which is a group of private physicians nationwide. I am also the editor of a textbook for medical students called, "The Art and Science of Bedside Diagnosis," which tries to teach medical students how to interview a patient, do a physical examination, engage in clinical reasoning, come to a diagnosis, and record things properly on the chart.
I'm not going to read to you my testimony. It would take too long, and you can do that for yourselves. But I would instead like to comment, bring some reflections to you on some of the things that have already been said today.
In my testimony, I did not directly address the questions that were asked, because I would like to point out that if you ask the wrong questions, you are going to get the wrong answers. If you were ask me, how should we go about cutting the heart and soul out of the medical profession, I would say to you, well, maybe you shouldn't do that at all. If you are going to give advice to Congress on how to impose a nationwide standard for the electronic record, maybe you should say, well, maybe there shouldn't be a nationwide standard for the electronic medical record.
I am here representing the end user of a number of the products that we have been discussing; a very frustrated group of people with some of the vendors who are here, because we don't want to buy your products necessarily. As previous witnesses have pointed out very well, the role of the government in this whole thing is to use brute force to make people do it, to make people use the system if they don't choose to do that themselves, because it all very good for them of course.
I would also like to point out to you that only people like me, the independent doctor out there don't want to use your product, but our patients don't want to either, because for one thing, they don't want to pay two to three times the cost of my charges now, because it costs a lot of money to support an infrastructure like that.
Also, they do not really want every detail about every medical visit that they have to go into a computer. They do not trust the computer. They have had experience with the impossibility of correcting errors, plus there are many things that they don't necessarily want to be ineradicably in a system that is accessible to the government, third parties, their employers, and a lot of really unknown and uncontrollable entities.
Additionally, when they come to see a private physician, they are deliberately making the decision to consult the unique and somewhat idiosyncratic computer that resides between my ears, which admittedly is not very good, but in some contexts unfortunately may be as good as it gets. Some patients may still choose the option of consulting an individual physician who does not need a unique patient identifier, because he knows their name, rather than the network brain. One thing that the American public still finds quite horrifying about the movie, "2001," is that computer named Hal.
Really, we are getting into some very basic issues about philosophy. As Shakespeare put it in Hamlet, "What a piece of work is man." As has been alluded previously here today, a human being is not a fire hose. He is not a space shuttle or an aircraft carrier, or even just a genome or a biological system. A human being is something really quite unique. As Shakespeare also said, "There are more things in heaven and earth, Horatio, than are dreamt of in your philosophy."
Much of what we are talking about today has to do with the difficulty that we face in a problem very much like that of squaring the circle, which mathematicians worked upon for centuries. The problem with squaring the circle is not that you don't have a big enough computer, or a big enough hammer, or a government to rush things along, but that the problem really cannot be done, because circles and squares cannot be made congruent, because there is a little irrational number in there called pi. It just doesn't reduce to a rational number, no matter what you do to it.
It's the Harvard law of animal behavior that you put an experimental animal under carefully controlled conditions, and the critter does whatever it pleases. As you all know, human beings are the very worst experimental creature in existence.
We have based a lot of what we are trying to do here on some assumptions. One is that quality is the same thing as compliance with a set of criterion and standards. It sort of assumes that we know the best way to take care of every clinical problem, and I submit to you that we really don't. In the course of my rather brief career I've seen very dramatic changes in what was the acceptable standard of care, and we don't necessarily want to set in concrete, what we are doing now, because it really may not be right.
The other thing is there is always the assumption that when we are measuring quality, we really are measuring something. Well, actually, we're measuring what we can measure. Many times we are not even getting the root of the problem, which is, did you make the right diagnosis? Well, to do that, sometimes you have to do an autopsy, or at least you have to have some sort of gold standard that is not going to be found necessarily in the medical record.
The second is that one of the big, prime movers behind this whole electronic medical system is that we're enforcing a system of price controls -- DRGs, the codes that are paid for under Medicare and Medicaid -- and we're assuming that we should have a system of price controls. The DRG system was brought into effect nationwide after a miserably unsuccessful but small test in New Jersey, and now everybody is trying to figure out how to implement that system.
Maybe the problem is with the system that you began with. Maybe the problem is that everything is being paid for through third party payers, which should not be paid for in that way at all, because it disconnects the normal regulatory mechanisms in a free market when the customer is spending his own money according to a set of his own personal values, and the trade offs that he, himself sees.
We also have the assumption that we need to have a top-down command and control economy in the medical system, which is called the health care system now. Maybe that isn't really what we want to do. Maybe we want to have a decentralized system where patients primarily consult the private physician of their choice, or at least that option is preserved for those who choose to exercise it.
It is also assumed that we will have some top-down rationing system that will allocate the resources, instead of leaving that up to the individuals to choose what medical services they want to receive.
Another potential answer to the question of why are we doing this, well, some many entities want to do it simply because there is a huge amount of profit to be made from so doing, and the unavoidable bias that comes into things, because people will make money if their system is adopted needs to be very carefully considered by committees such as yours when you make your recommendations to Congress.
I would like to suggest that the role of government should be to protect individual rights against those rights being violated by others. It should not be to dictate the standards of medical care, or to dictate how medical records shall be kept, which will eventually amount to having a very great influence on how medical shall be practices. There are very serious adverse consequences to doing something like that, some of which are outlined in the testimony that I provided to you.
Briefly, it violates people's constitutional rights and protections against unreasonable searches and seizures, against privacy. It destroys the patient-physician relationship, impedes scientific progress by trying to put everyone in a straitjacket. It will drive independent practitioners out of the medical field, because many of us simply cannot afford the overhead required for implementing such systems, and we believe will seriously degrade of the quality of medical care.
Probably the briefest summary is at the end of my testimony. If you have mandatory computerization, you will probably end up with what Dr. Herbert Fred called "elephant medicine." "In elephant medicine, each doctor, like a circus elephant, metaphorically wraps his trunk around the tail of the elephant in front of him, all marching in a line, looking neither left nor right, exactly in the trail of the first elephant." Which might all be very good if the first elephant knows where he is going.
I think that right now we are in the situation like a number of blind men trying to describe an elephant, each by feeling a part of the elephant. To are all blind as to exactly where we want to go, and exactly what the medical system should look like, because it involves a bunch of very unique entities, each one of the being a human being.
DR. COHN: Jean, thank you very much. Kent Spackman.
DR. SPACKMAN: Thank you. I'm Kent Spackman, association professor of pathology at Oregon Health Sciences University, and as I said, I'm here representing the College of American Pathologists, where I serve as the scientific director of the SNOMED international division. On behalf of the CAP, I thank you for the opportunity to appear here.
I'll first provide some important background on the CAP, and its interest in the issues currently under consideration by the committee, and then turn my attention to the questions.
CAP is a not-for-profit medical society, serving nearly 16,000 physician members in the laboratory community throughout the world. The CAP is the world's largest association composed exclusively of pathologists, and is widely considered the leader in laboratory quality assurance.
The CAP is an advocate for high quality and cost effective patient care, and accordingly, the CAP has invested in several initiatives to insure the delivery of quality clinical services. SNOMED is one of these. SNOMED, the systematized nomenclature of medicine, is a comprehensive, controlled reference terminology designed to encompass all of the terms used in medicine, including procedures and diagnoses.
CAP has more than 30 years of experience in the development and ongoing maintenance of this controlled clinical terminology, and it extends well beyond just the interests of pathologists. The CAP committee on nomenclature and classification of disease spawned this initiative in the 1950s.
While the CAP recognizes that no system today captures all clinical concepts, independent studies have demonstrated SNOMED to be one of the most complete reference terminologies in use today for the clinical environment. Thus, we believe that SNOMED, the comprehensive controlled vocabulary covering diseases, clinical findings, etiologies, therapies, procedures, and outcomes is the central component of the terminology solution.
The detailed granularity of concepts contained within SNOMED, which now number more than 157,000, can easily be linked to the broader classifications used for statistical and administrative purposes, but its applications and benefits extend far beyond these.
The compositional nature of SNOMED, and its hierarchical, systematized structure transforms the notion of coded vocabulary into a powerful tool that can be used for telemedicine, outcomes analysis, cost effectiveness studies, knowledge-based practice guidelines, decision support systems, and comparison studies of health care plans.
As a part of its long-term commitment to the quality of care, the CAP will premiere the next generation in nomenclature in June 1999. SNOMED-RT, under development in collaboration with health care providers in the Kaiser Permanente for more than a year is now ready for beta testing by selected system vendors and end users. SNOMED-RT, a reference terminology and knowledge base underscores SNOMED integral foundation for the computerized patient record.
It combines the granularity and comprehensiveness of SNOMED terms and term codes, with improved clarity of meaning. In SNOMED-RT non-ambiguity of meaning is made possible by multiple hierarchies and explicit semantic definitions. So the ability to represent a vast number of additional concepts is made possible by compositional syntax.
The uses of SNOMED are extremely diverse. Today, SNOMED is used for mapping terminologies, for standardization of clinical reports, and coding of medical concepts, transmission of normalized data, article search and retrieval, decision support, outcomes assessment, coding adverse drug reactions in clinical trials, surveillance reporting, and assessment of physician utilization patterns. Applications continue to be identified as more people become aware of and knowledgeable about the benefits of SNOMED.
Overall then, SNOMED can contribute to improvement in patient care, reduction of errs inherent in data coding, facilitation of research, and support of compatibility across software applications.
So with this background, I will now turn to the questions. First of all, how do we interpret the congressional instruction? As you know, the HIPAA legislation expanded the charge of the NCVHS to include advising Congress about adopting standards to facilitate the electronic exchange of health information. Today's hearing focuses on the two-fold HIPAA mandate to NCVHS: first, to study the issues related to the adoption of uniform data standards for patient medical record information, and the electronic exchange of such information; and secondly, to report to the secretary recommendations and legislative proposals for such standards and electronic exchange.
The CAP believes that the legislative directive to NCVHS is fairly broad, and therefore allows the committee the discretion to focus its efforts on what is currently happening in the marketplace, including recommendations of clinical code sets for those purposes. The environment in which health care data is generated and used, and applications for its use is constantly evolving, as noted by Congress in the HIPAA conference report.
"The conferees recognize that technological innovation with respect to electronic transmission of health care related transactions is progressing rapidly in the marketplace. The conferees do not intend to stifle innovation in this area. Therefore, the conferees intend that the committee take into account private sector initiatives."
The language appears to be recognition on the part of Congress that initiatives related to electronic exchange of health information have occurred primarily in the private sector, and appears to be in support of such initiatives.
In the United States, with the exceptions of the initiatives such as the National Library of Medicine's UMLS, and the current collaboration between the DOD, VA, and the Indian Health Service to develop the government computer-based patient record, the public sector has not taken the lead in this area.
In contrast, several private sector initiatives have been implemented that advocate for and provide research to support the concept of widespread adoption of standard clinical terminologies for the computer-based patient record.
To its proponents, standard clinical terminology holds the promise of creating consistent communication across clinical specialties and sites of care, thereby facilitating clinical decision support, research, and efficient care delivery. By providing a common basis of understanding, reference terminologies streamline communication, and help to improve patient outcomes by eliminating miscommunication among health care providers.
A reference terminology allows easy transmission of patient-related data across often diverse and incompatible systems. SNOMED is such a clinical reference terminology. It is a multidimensional, structured nomenclature for electronically coding all elements of the patient medical record, including pathology reports.
The SNOMED terminology is comprehensive, with content coverage including: diseases, clinical findings, etiologies, therapies, procedures, and outcomes. As a reference terminology, SNOMED enables the computerized recording, storage, retrieval, and analysis of clinical information essential in order to compare information related to patient conditions and events.
As far as users go, the pathology community continues to be among the strongest supporters of SNOMED, however, SNOMED users are expanding, and this a testament to the comprehensive and multidisciplinary nature of the terminology. More and more systems developers and vendors, hospitals, group practices, managed care organizations, medical libraries, insurance companies, research entities, and government agencies are recognizing SNOMED as an essential component of their total information needs. Globally, SNOMED is used in over 25 countries.
So the CAP believes that HIPAA has served as a catalyst to promote public sector leadership in the development of standards for the electronic exchange of health information, and can be the fuel to provide the continued support of such efforts.
Question number two, factors or issues are preventing or delaying the development? There may be several factors delaying the development and widespread implementation of uniform standards for patient medical record information and electronic transmission. First of all, lack of understanding of different types of terminology. Until recently, there has not been a clear understanding of the differences and relationship between user interface terminologies, administrative terminologies, and reference terminologies.
This construct, I think is a useful one that provides us a way to move forward synergistically. An interface terminology is defined as sets of words or phrases used to describe patient-specific information in the words common to the recorder of the information. So structure data entry screens, clinical-specific sheets with check off lists, and so on are examples of interface terminologies.
On the other hand, administrative terminologies ordinarily specific groups of clinical concepts that have specific purposes. These are typically based on broad categories of diagnoses or disease states that have been established for epidemiologic purposes, or based on collections of medical procedures and services that require similar resource allocation or resource utilization.
Typically, administrative codes are recorded and reported for reimbursement purposes, and are utilized secondarily for statistical analysis. So from this perspective ICD-9-CM and CPT4 are considered administrative terminologies.
In contrast, a reference terminology such as SNOMED, supports the detailed coding of all relevant clinical activities and concepts based on the observations and activities of clinicians, patients and/or their families. The reference terminology also provides what might be called a semantic scaffolding onto which the various terms used in health care placed, making the concept meanings and relationships explicit. This facilitates interoperability between a wide variety of systems and clinical records.
Reference terminologies support detailed clinical documentation for charting purposes, analysis of data at higher levels of abstraction for outcome determinations, and also clinical decision support and other purposes.
Reference terminologies are designed specifically to support both clinical and research applications. More detailed SNOMED concepts can be linked to user interface terminologies, and can be combined to create precoordinated terms for ease of use and linking to the administrative terminologies.
This can be done without compromising the integrity of meaning. Perhaps an example will help to illustrate the difference between an administrative terminology and a reference terminology. CPT code 58150 represents total abdominal hysterectomy with or without removal of tubes, with or without ovaries. That CPT code is appropriate for reimbursement purposes, because it groups procedures with similar resource utilization.
However, whether a patient has one or both ovaries or fallopian tubes is a detail that must be made explicit for charting, and for research purposes. It is not apparent from the CPT code. By virtue of its granularity, SNOMED enables systematic documentation at such detailed levels. In addition, it can be linked, if desired, to locally accepted user terms for these procedures.
Without a clear understanding of these three types of terminologies, a logical approach on how to proceed in the development and widespread adoption of uniform standards for patient medical record information and electronic transmission has been virtually impossible. Slowly the three types of terminology, their differences, and their relationship to one another are being articulated and better understood. Each has its place, but only a clinical reference terminology can adequately support the clinical detail required for the computerized patient record.
The current health care environment has also changed a tremendous amount, and has fostered the need for standardized terminology that can create internal consistency within the electronic record, facilitate information sharing, as well as support population-based studies at a very detailed level.
But the development of uniform standards for patient medical record information and its electronic transmission has been a slow process. Until the passage of HIPAA, the private sector had been hesitant to expend additional resources, time, and dollars to move towards a standard terminology that ultimately may not have been accepted by the government.
Other barriers include the problem that scale is hard. Scale is extraordinarily difficult. Most people have experienced how easy it is to make a list of 10 or 15 or 100 terms, and then extrapolate from that to the notion that making a list of 100,000 terms may be just as easy, but the scale is a major barrier, and there has been a lagging recognition of the resources that are required to handle scale.
Question number 3, is the private sector able to address these problems satisfactorily? The private sector, in particular, the academic community and medical associations and the vendor market have embarked upon initiatives to overcome the obstacles preventing the development and widespread implementation of standardized terminology. But these efforts have been limited, and narrowly focused.
For more than 30 years, the CAP has worked on developing and enhancing SNOMED. Although SNOMED was initially developed for a unique purpose, pathology applications, its comprehensive nature has enabled it to evolve into a terminology used by multiple specialties for care management, quality assessment, clinical research, public health reporting, and so on.
After more than 30 years SNOMED continues to meet the changing needs of the health care environment. It continues to serve as a bridge between administrative and clinical environments.
One of the advantages of SNOMED is its ability to stimulate convergence of terminologies. In other words, to enable clinicians, researchers, and patients to communicate across medical specialties and sites of care. For example, SNOMED has been selected by the DICOM standard, digital image communications in medicine, for representing anatomical and other concepts in the reports that accompany images in electronic messages.
In cooperation with the American Dental Association, SNOMED has been revised and enhanced to incorporate dental terminology proposed by the ADA, integrating among other things the ADA's CDT nomenclature with the SNOMED structure. To further insure lack of overlap with the medical terminology, the ADA will be adopting a wide range of SNOMED terms to document co-morbidities that are not specific to dentistry, for example, diabetes.
This cooperative effort also permits the ADA wide latitude in determining the content of dental diagnoses within SNOMED. This is just one example of the kind of convergence that we see as extremely important to foster.
Through its formal liaisons to the SNOMED editorial board, the CAP continues to work internally and with professional organizations to improve the detail and clinical content coverage of SNOMED. The CAP is committed to continuing these many efforts to meet the needs of the users of clinical vocabularies.
Development, maintenance, and coordination of terminologies is a costly process. In September 1997, the CAP committed more than $17 million over a five year period of time to the continued success of SNOMED. This financial commitment is a testament to the CAP's overall dedication of the development of SNOMED as a scientifically validated reference terminology.
However, it's not certain if the CAP's financial commitment alone will be adequate to insure the continuous evolution of the terminology to meet the ever changing needs of contemporary clinical practice. The development and implementation of reference terminologies benefits both the public and private sectors, and therefore, should receive support and resources from both sectors.
Without adequate government support, widespread development and implementation of uniform standard terminologies will continue to be delayed. The CAP believes that there should be a public/private partnership in the development of an ongoing maintenance. The private sector has the appropriate and necessary expertise required to develop the standards, therefore, the government may not need to play a large role in the development of standards.
Government involvement, however, is crucial in testing and funding for ongoing maintenance, either directly or indirectly, by providing these financial resources and mobilizing use, and evaluation of standard reference terminologies.
For example, the government could adopt core leading clinical terminologies for use in the government-related projects by the National Cancer Institute, Centers for Disease Control and Prevention, the Veterans' Affairs, the Department of Defense, and so on. Through this mechanism, the government could be using its influence in the health care industry to encourage movement in the marketplace toward greater uniformity.
Question number four, what standards related to patient medical record information would add the most value? We believe that SNOMED itself is an important component of a solution to the needs to patient record information and its electronic transmission. Other existing standards that have already been mentioned, such as HL7 and DICOM can also be very much a part of the solution.
We also believe that existing administrative terminologies such as ICD-9- CM and CPT4, and the terminologies that are developed and/or adopted to replace them, such as ICD-10-CM and CPT5 will continue to be a part of the overall solution.
Do we agree with the emphasis in the four areas? CAP believes the four focus areas are appropriate. Of the four focus areas, we would argue that three, coordination, maintenance, continuous availability, interoperability, accountability, and clinical specificity is a key issue, and perhaps should receive the greater attention.
I'm going to skip down to a brief comment about the editorial process that SNOMED has embarked on. The scientific content of SNOMED and the editorial direction is governed by an editorial board, which has recently expanded beyond just pathologists, to include a wide variety of other organizations.
Liaisons from the following associations and government agencies have been employed to serve on the editorial board: the American Veterinary Medical Association; the American Dental Association; the American Nurses' Association; Centers for Disease Control and Prevention; the American Academy of Ophthalmology. This depth of representation and involvement on the editorial board reflects the CAP's commitment to expanding and integrating terminology that is useful for all clinical specialty areas.
SNOMED is currently cross-mapped to ICD-9-CM. Additionally, the CAP has recently embarked upon an collaboration with the AMA and the NLM to cross- map SNOMED to CPT4.
As part of its longer-term strategy, the CAP recognizes the importance of international collaboration and convergence, to enable data transfer and accurate communication across the globe. To that end, the CAP is in very active discussions with the National Health Service in the United Kingdom, developers of Clinical Terms V.3, formerly known as the Reed(?) Codes.
When brought to fruition, this partnership will further expand SNOMED's scope beyond national borders, and maximize utilization of resources available to develop a truly comprehensive and internationally accepted clinical terminology.
So in conclusion, the strengths and benefits of SNOMED as a recognized leader in clinical reference terminology have been demonstrated over the span of many years. The CAP is clearly a strong advocate for high quality, cost effective patient care, as shown through its long-term commitment to the ongoing development of SNOMED.
That commitment has not changed, and will not change. As an organization, the CAP envisioned the importance of standardized terminology to support quality patient care, and therefore, it has continued to enhance the structure and expand the scope of SNOMED to insure that it is scientifically accurate, and representative of the practice of medicine.
The CAP is also committed to widespread availability and access to SNOMED, with minimal restrictions and nominal cost. This has been done without external influence. These efforts continue to evolve in response to the dynamic changes in health care.
A clinical reference terminology captures the detail necessary to document care, retrieve data to perform studies, and assess patient outcome. The CAP believes that SNOMED is such a reference terminology, and is consistent with the standards requirements for a comprehensive clinical code set necessary to collect and analyze data more effectively, compare the quality of health care being administered, develop effective treatment guidelines, and conduct important outcomes research.
Thank you.
DR. COHN: I realize we're all running a little late in terms of timing. I would suggest we go for about 15 minutes with questions, and then we'll break at that point. Are there any questions from the panel?
MS. FYFFE: Thank you, everyone. Dr. Orient, if you were in our shoes, what would you do? We are required to make recommendations to the secretary - or pardon me, Congress.
DR. ORIENT: Well, I would point out to the secretary and to the Congress all of the potential adverse consequences of mandating nationwide, a system that has to be deeply and fundamentally flawed simply because of the nature of the beast. And to recommend that certainly electronic systems can be a value in many circumstances, and those who find such, should be free to buy them, but others should not be forced to invest all kinds of money, resources, and time in a system that can only degrade the practice of medicine.
So much of medicine depends on doing your own history and your own physical examination, and relying on the electronic record is very destructive, because there are so many errors in it.
MS. FYFFE: If you would support a voluntary type of standard, if you will, can you comment -- I'm going to ask you a big question here. It's going to take some thought, and we may not be able to do it today. What would you like to see in order to improve what we have right now?
DR. ORIENT: Well, I think you can see that when a piece of technology is beneficial, people buy it quickly. I mean you don't have to force me to buy a fax machine or a personal computer. Or you don't have to force doctors to use magnetic resonance imaging, because it is so valuable to them, but if you're having trouble getting them to buy a clinical information system, that should tell you something.
Plus, to say "voluntary," sometimes we say things are voluntary. In a way they are. Sure, you don't have to practice medicine, but if you do practice medicine, you have to do it this way. Sure, you don't have to get paid for practicing medicine, but if you want to get paid, you have to do it this way.
So I think by voluntary, I mean truly voluntary; that you can do something else if you like. That's the only way that we are going to have real innovation and real problem solving is to permit people to have the freedom to follow their own judgment.
DR. MC DONALD: There is such a nice, good, long spread of speakers, I almost can't remember yesterday. We ought to make them in little smaller packages. I like all the comments.
John, I just wanted to sort of give you special acknowledgement, because you have said so many things that I thought. You said them better than I thought them, and I just really like that.
Dr. Orient, I come from a different world, and I guess I have kind of made my life in medical records systems, so I should be upset, but I'm not, because you said it well, and because I think there is some deep truth in this business about forcing people. I don't think anyone is really talking about -- even the billing standards, you have to use them if you send electronically.
I don't think there is anything in the bill that suggests that the medical record will be forced on anyone. I don't think there is any conception that it will be the same system throughout the country, or that it will even be a national medical record. I hope it isn't, actually.
So I think your statements will counterbalance anybody who gets that notion, that the goal is to force people to use it. I think it simply can't work. I think John made some points about the difficulty of implementing it. If we force things on, it will screw up everything. It will all break, and we will never get it. Although physicians may feel they are weak, we do have some power. We can change things. I think we should work on some things the ENM rates, but that's another story.
I like what you had to say, and everybody had good things to say, but I won't wander on, given the time.
DR. MATTISON: If I could just comment on the change resistance to implementing the systems. As I mentioned, I have implemented five different systems in different groups of physicians. You invariably find a very high level of resistance initially. It starts to taper off at about three months. At a year of use of one of these systems, it is very hard to find a physician who won't say, I don't know how I ever practiced without one.
What happens is they begin to realize that computers are very good at doing the same thing over and over again; the mundane things that tend to lead us to errors of omission. What it allows people to do is to really focus on the judgment side of issues, the art of medicine.
So rather than take the art of medicine away, what I have experienced is watching people appreciate that as they learn to use these systems, it allows them to focus on the art of medicine, and take the mundane out of their face.
DR. FERRANS: Yes, this is for Jean Narcisi. I wanted to thank you for reminding everyone about the problems of the user interface for clinicians. I think that there are people in this business who deal a lot with user interface issues, and there are also people who deal with data feeds from huge numbers of hospitals. It is very important to understand that there is that significant barrier to capturing granular information. At this point, it still remains a major problem.
I think speech recognition is going towards that. I think mark-up language in particular will go a long way. But as I think Dr. McDonald said at our last meeting, if you are delivering different pieces of information to clinicians, like all the lab results, and summarizes, and dictated reports, you can have somewhat of a record without reducing everyone to a typist.
Having said that, I did want to sort of get to what you said about the AMA's position about the appropriate role of government. As I was reading through here, I was struck that you said, well, computerized patient records will have benefit, or are a good thing. The AMA does not recommend that the government get in and do regulation.
At the same time, I think you stated in here that there are deficiencies in standards, and that you recommend accelerated private development, or accelerated private progress in the private sector. Is there any way that the AMA would suggest in terms of how to facilitate that? It seems like that is something that everyone is interested in. My sense is a lot of people don't want the government to get in and issue regulations. What they really want is for the results.
I think we are all interested in the results for a convergence upon terminology standards, upon messaging standards, and things like that, and just your thoughts on that.
MS. NARCISI: Well, what the AMA is opposed to, of course I stated was mandating any kinds of standards. We think that they should be allowed to evolve in the private sector. Although there have been so many barriers to this, what was discussed this morning, if we focus on just a few things, for instance, the administrative transactions. We have a lot of work to even do on those, before those are actually implemented and usable.
If we kind of allow ourselves to evolve that way, and perhaps the government can maybe prioritize some of the things that you found in your own areas, some of the activities, the coordination between the DOD and the VA in those kinds of projects. Report to us how those are working, how those have evolved. Probably, eventually some of that will all fall out, and we'll be able to use it more in the private sector.
The government participates in all of the standards activities, and they should continue to do that. But just mandating any kind of standards is what we are opposed to.
DR. FERRANS: I guess my last question was to John Mattison. With your involvement in HL7, if somehow everyone agreed that they were going to use HL7, and everyone agreed to use 2.3 or 3.0, whatever, what would be the effect upon health care, if everyone agreed to use the standard, and was using it, and using it appropriately?
DR. MATTISON: There are a lot of benefits that come to mind, but I think from the standpoint of the individual consumer, what they would benefit by is that when a decision was made about their health care, that whichever clinician was making that decision would have the benefit of all the information that that particular consumer had shared with the health care delivery system, and that it would be a well informed decision.
From the standpoint of vendors, it would allow them to focus on a single standard that would make it easier to support multiple different implementations in different institutions, so that they would be able to produce better software at a lower cost to support it.
It would also enable, to a great extend, the normalization of data across different applications so that the ability to do population-based outcomes analysis would accelerate. That would ultimately enable a faster quality improvement cycle. The axiom of quality improvement is know what you need to know, and measure it. If we can measure it and normalize it, and develop knowledge out of that information, that would go a long way towards accelerating our understanding of illness, and how to treat and prevent it. I think HL7 could play a huge role towards all of those goals.
MR. BLAIR: Dr. Spackman, I'm going to zero in on something where some months or years ago I thought that government enabling or facilitation might be able to play a constructive role. I don't know if it's true any more, and I'd really like to get your feeling about it, because from what I understand, a number of things have been happening in the private sector which I think, makes it not necessary for the government to assist in the development of reference terminology or control medical terminologies, and the interoperability and clinical specificity of terminologies.
Let me get at it three ways. The College of American Pathologists and SNOMED has matured where you have many other code sets that are beginning to align themselves with SNOMED in various ways. SNOMED-RT has been announced, and I understand that you are also having dialogues with other terminologies on an international basis. So it sounds like from the standpoint of being able to maybe a core or a nucleus of many different medical terminologies, from that standpoint you seem to be doing fine just on your own.
The other piece is that you are also working with message format standards. You and other code set developers, I think of the HL7 vocabulary seg, where they are beginning to embed LOINC and SNOMED, and probably will wind up using other code sets. So that integration is taking place without government assistance.
Now the other area does have some government assistance, because the National Library of Medicine funds the Unified Medical Language System, which basically maps between coding systems, but that is already in place.
So since the purpose of this hearing is to find out what we should focus on, what we should address, where the private sector might ask for assistance. I'm beginning to question whether or not you need -- the medical terminology environment still needs government assistance. If you think we should continue to look at that area, could you please help us focus and understand where assistance would be needed?
DR. SPACKMAN: I think the government can play a role there still. It is around this whole notion of convergence. Obviously, we are doing the best we can to convince people to collaborate with us. We talk to the LOINC committee and say, well, how do we work together, and then we talk to the Reed folks and say, how do we work together?
But there may be an additional role, and it has to do with this issue of scale. I don't think by any means we have linked up with all of the potential developers or users of terminology who have an interest in this area. Given enough time, maybe we will, but there could be an acceleration of that through perhaps some government initiatives.
One might be to help fund some analysis or some testing of the tools and processes for convergence. In other words, looking at how well do our current tools and processes work for integrating terminologies, and what else can be done there as a way of incenting medical specialty organizations and other groups who are interested in terminology to collaborate with the existing reference terminology.
Of course the other direction which I mentioned in my testimony is for the government, if there is some way of organizing or coordinating government effort, to begin to adopt and use some of these standard clinical reference terminologies within government projects, within the National Cancer Institute, the GCPR, and so on. So I think those are a couple of the ways that I would recommend.
The third thing that I would recommend is beyond the representation of clinical detail, which is what the reference terminology does, we need some kinds of studies of the whole encoding process and the reliability and useability and testing the actual application of the terminologies in various environments. I think research funding for those sorts of studies would be extraordinarily helpful as well.
We can provide the codes and the meanings, but there is a significant step from there to actually having an implemented electronic patient record where you can rely on the data that is encoded. You can rely on the codes, because you know what they mean, but getting it into the record is a big step.
MR. BLAIR: Did you elaborate on this, by any chance, in the hard copy handout you gave us?
DR. SPACKMAN: I believe that's in addition to what is in the handout.
MR. BLAIR: I would appreciate whatever descriptions you could give us, so we could understand it as well as possible.
DR. SPACKMAN: Okay.
MR. MAYES: I would like to just elaborate a bit on what Kent brought up. I think it's important to realize we are talking here as if the government's primary role is mandating things nationally. I would point out that the administrative simplification aspects of the HIPAA legislation are unique within the health care sector in the government. We have never in the past -- actually, beyond the health care sector -- mandated standards nationally.
The government is in the business of health care. It is the largest single payer of health care. It's the largest single provider of health care. It's the largest researcher in health care in the United States. The vast majority, in effect all but HIPAA of regulation put out by the federal government are response to business needs; the same type of business needs, albeit on a very large scale, as all of the presenters that have presented to us this morning.
When we talk about the role that the federal government has, I think that Kent rightly points it out, that yes, indeed we could contemplate an approach that we took with the administrative simplification and financial standards, which are these sort of broad, mandated standards.
But far more likely is going to be the adoption or the use within the federal programmatic requirements of standards. It would be of most interest to hear how that might be accomplished. I mean I represent HCFA. Obviously, if HCFA adopts a particular approach, right now we think up things. ENM is a good one. Obviously, there is great potential here. If HCFA adopts a particular approach, it tends to have influence far beyond just HCFA's primary business. The DRG is a perfect of that, and that's been brought up several times.
I did want to make sure that we are clear that when we talk about "the government," the government actually plays a number of roles, many of which are identical to the roles that each of you play in the particular organizations and sectors that you represent. It's not just a sort of big brother, broad mandating on a very general area.
DR. COHN: Before the next person I guess I would comment that recommendations can be carried out in many ways.
DR. FITZMAURICE: I, along with everyone else in the room, wish to thank you for the expertise that you have brought to the table, and the time and distance you have spent to give us a better education on patient medical record data standards.
I have two questions, one for Jean Narcisi, and one for Jane Orient. For Jean, I applaud the resolutions that have been passed by the house of delegates in the past year dealing with the computer-based patient record, and with the policy that among other things, has the AMA working to define the characteristics of an optimal medical record system for its members.
So I ask, I assume that an optimal medical record system would be one that gives you access to uniform accurate patient care data. I'm making the assumption it would also give you access to medical knowledge, perhaps through the decision support rules.
Is the AMA undertaking research, or taking a look for measuring the magnitudes of these benefits for the members? Because there are significant costs. There are significant barriers, and they need to be balanced with the value of benefits in order to make a good investment decision. Is there something going on to look at that, that we might expect in the next year or so?
MS. NARCISI: No, we're not to that level yet. What we are trying to do is pull together some advisors to help us through this problem of trying to just consider what should be in a medical record, just the components that should be in a medical record.
Then from there, try to develop some guidelines to help physicians select systems, because there are so many different systems out there. I think this morning it Peter Waegemann who mentioned there were 249 systems. I think that there are really about 2,000 systems to choose from. So that's the level that we're out right; to really try to develop some guidance for physicians to select their systems.
DR. FITZMAURICE: I sure it will be very welcome.
For Jane Orient, I noticed in your testimony -- I'm looking at page 5 -- a conclusion, "There is no a priori reason to expect that a national network of computerized patient records will significantly affect patient care in any positive way."
I may be looking at a straw horse, but I don't know of any push to get a national network of computerized patient records in the private sector. Maybe with regard to the Department of Defense, maybe with regard to VA, but I don't think that this is going on. Is there an effort that you know about, that I don't know, or is it just something to watch out for?
DR. ORIENT: Well, I think it's going on as far as immunization records. I think that's the impetus for building the infrastructure. One certainly does read about intimations of that idea. There are certain states that have passed laws, such as Maryland and Wisconsin, that make it point of law that every medical visit is supposed to be in a database somewhere. I think that that's a prelude toward national things.
It has been our impression that HIPAA is the entering wedge for that. First, we establish the standards. Then we say, okay, you have to transmit things electronically. Medicare is beginning that with HCFA. They mandate that you should have to do that, or if you don't, we'll punish you in some way. We'll this little set of things that are supposed to be guidelines, but if you don't do them, then there are very punitive consequences.
Sometimes physicians are really terrorized, because they know they are supposed to abide by the rules under pain of prison time and other things, but they can't even figure out what HCFA's rules are.
DR. FITZMAURICE: Moving away from a national network of computerized patient records, but to computerized patient records in general, would you agree that if they are a good thing, that we do need to measure the benefits, as well as these costs, and to show where the balance is?
DR. ORIENT: Well, I think it would be a very good idea to have a pilot project carried on by unbiased observers, that goes on for long enough that you can see exactly what is going to happen, and that you have the data available to everyone, so that you can get objective viewpoints from a variety of different perspectives about what the results really are.
Unlike the DRG process in New Jersey, which was supposed to be a pilot project, but it really was just the first thing that immediately was implemented nationwide, without actually seeing how bad the results were.
DR. FITZMAURICE: You're not recommending we put a computerized patient record in New Jersey, are you?
DR. ORIENT: I think New Jersey can be first.
DR. FITZMAURICE: What you said the research is fair enough. Thank you.
DR. COHN: Clem, a final question.
DR. MC DONALD: One of the tensions about standards group is the fact that the cost of standards and the cost of the groups, yet they are voluntary in nature. In Europe the groups are funded. I'm not espousing that. I think it really works pretty well because of the cleansing process. It keeps the interest level right, and everything works.
But I wondered what the group would think about if we could find a way to reduce the cost of the standards products to users, because that is a barrier to a more widespread adoption. And whether one could imagine some ways to subsidize the documents, and that would indirectly subsidize the groups. X12 actually did get some subsidization for their document from HCFA.
SNOMED, for example, could maybe be subsidized in terms of getting the document. Does that sound like a bad thing or a good thing? Does anyone know of any examples that would make it not work?
DR. ORIENT: Well, there is the CPT code example, which is a great expense for physicians. They are forced to buy these to comply with HCFA's regulations, but the AMA has an exclusive license for that. So it's not like a government regulation you can just download from the Internet or get from the superintendent of public documents. You have to buy a book, and it's maybe several hundred dollars a year just to get the standards.
So I think that what's been done so far is to give a monopoly to a private group to produce something that the government forces people to use, is not a good way to keep costs down; quite the opposite.
DR. SPACKMAN: I guess the comment that I would make is that the real cost in clinical terminology has to do with the integration and the scale. That's something most people don't recognize. So distribution of the document doesn't -- obviously, helping to fund the distribution of the documents is a good thing, but that's not where most of the cost in producing a standard.
DR. MC DONALD: I wasn't suggesting that the subsidization would be limited to the document reproduction costs. But that's how some standards groups sort of fund themselves, is through their documents. Now some exorbitantly. I think I spent $40 for a 3 page ISO document once; and some fairly reasonably.
But still in all -- any other comments on it? I don't know if it's really valid or not. CPT4, they could subsidize with the book.
MS. NARCISI: I was just going to say that I think a lot of groups are having problems dealing with this whole thing. Even the American National Standards Institute is at a point in time where they have been so paper-based with selling of standards. Now things are more computerized. They are having a hard time figuring out where they are going to go next, if they are going to eliminate boards or whatever.
But I think that a lot of that cost will decrease once we can get the vendors to get together and incorporate the standards into the system, so it is not left for individuals to have to purchase those standards and those kinds of things. So this interoperability thing I think is really going to make a difference.
DR. MATTISON: I think there might be a principle that you could apply to that question. That is that if there is some sort of federal subsidization of standards initiatives in some fashion, with the intent of lowering cost, the risk is that there is never a free lunch, and that would be an invitation to influence in some content fashion, the standard itself, so that the principle would be that if those sorts of funds were solicited or offered, that they be earmarked in one and only one way, and that is a reduction in the cost of deploying the standard, but that no earmarking in terms of content would occur. It would just be given as a lump grant effectively to the standards agent.
DR. COHN: I would suggest that we save this as an idea to begin to talk about the work group needs to do.
Now I want to first of all thank the panel. We have kept you a half an hour beyond the time we had advertised. I thank you very much. I want to apologize to the people on the work group and staff who didn't get a chance to ask their question, as well as the people in the audience who haven't had a chance to ask any questions either, or make any comments.
We will take about a 15 minute break, and come back about 4:00 p.m. Hopefully, at the end of the day we can take a couple of minutes for any sort of comments from the wider audience.
Thank you.
[Brief recess.]
MR. BLAIR: Is our fourth panel assembled here? Yes? Can you please introduce yourselves?
DR. GARDNER: Reed Gardner.
MR. BLAIR: And you are with?
DR. GARDNER: I'm with LDS Hospital, University of Utah in Salt Lake City.
MR. BLAIR: Okay, thank you, Reed. Are the other people at the table?
MR. ARGES: George Arges, American Hospital Association.
MR. BOZEMAN: I'm Tommy Bozeman, vice president and CIO at North Mississippi Health Services.
MR. BLAIR: Is that everyone, just three? Reed, I gather you are standing there, because you're ready to give your presentation.
DR. GARDNER: I am indeed.
MR. BLAIR: Proceed.
DR. GARDNER: I couldn't sit and read testimony, sorry. So I'm going to take you through what I have lived through for the past 30 years. You have handouts for most of it. Some of it I will do very quickly.
I have worked for the health system in Salt Lake City. We have gathered data from bedside monitors, from IV pumps, from a whole variety of devices with the thought in mind if we collected those data in a coded form, we could use them for decision-making.
I'm not like others and think that you have to have a paper-free system to have an electronic record. I say let's get an electronic record. If you want to print it and look at it, fine. I learned how to use paper when I was a little boy in kindergarten, and I still use it pretty well.
So my objective is to get data into an integrated database so that I can do decision support, and have done that in the hospital. The example I have shown here is from a critical care unit where we have hooked up IV pumps, ventilators, and monitoring devices. I can just tell you I spent 30 years building, rebuilding, and even today I bought a brand new monitoring system, and the monitoring manufacturers all have their own standards. They say if you just take ours, it will be wonderful.
Well, I can't buy HP and interchange it with Marquette. I can't buy and IV pump from somebody. I have worked literally 15 years to get standards for medical information BUS, and that's a very, very difficult process. It's been very expensive. The manufacturers say, well, it's easy, you just do this. Well, you spend all your time in the infrastructure, rather than the important things you should be doing. So I think we need standards.
This just shows in the example in a critical care unit, the IV pumps are all interconnected in the system. The monitor is. The ventilator is an old analog device, which will be replaced within a year or so; that will be. And there is a computer terminal at every bedside. We have found the computer system to be very helpful for our people. We haven't necessarily forced it on people, but in order to use a system like this in a critical care unit, it certainly has to be reliable and available all the time, especially in critical care units in the emergency room.
We have done studies, and the question has been asked, do these systems really do any good? The answer is yes. I think Clem McDonald, his group and several others that have presented here today have shown that. We have used computer-assisted antibiotic management, and we have found that overall antibiotic use has gone down by about 23 percent. We have decreased mortality rates. We have decreased adverse drug events, since antibiotics are one of the major causes of that. And the cost of antibiotics has gone down 42 percent.
So these types of systems can indeed change the practice of care. They can improve the practice of care. Advising systems, which the physicians can use or not use as they wish, have been very successful, and very well accepted by our private practice physicians. We have at LDS Hospital maybe only 10 physicians who are employed by the hospital. The remainder are in private practice in the community.
So I think we can the use of computers has helped reduce inappropriate antibiotic use, and adverse drug events.
Then I have to quote Clem, just to make sure he is awake. He said this about four years ago. "When it comes to gathering information, medicine has not evolved much from the hunting and gathering stages of social evolution." It is still true in 1998.
A really wonderful little article came out on IIIA Computer about a year ago. They looked at the ten potholes in the road to information quality. This is the first; multiple sources of the same information produce different results. They looked at a whole variety of places, including health care organizations, parts delivery people for the aircraft industry, and so on.
The example is the hospital uses two separate procedures to assess severity of illness. We still can't determine severity of illness for patients in our hospital. It is very, very difficult to do. We have a reasonably highly computerized system, and gathering the right data to do that process is very problematic.
Information is produced using subjective judgments leading to bias. Again, when medical coders use their own independent judgment to select codes is the example.
This is one that continues to haunt us as we are part of Intermountain Health Care, which has 23 hospitals, now has 10 clinics, and want to share data across those institutions. We have worked very hard to make standards. Here it says distribute heterogeneous systems lead to inconsistent definitions, formats, and values. As far as I was concerned -- maybe I just read this thing far too narrowly, but that's what I saw from this. Then each division enters a different diagnostic code is the example.
Then this one is one that is for certain true. As I sit in my position at the university, we have just recently bought 40 clinics. We literally went out and bought them with our money, and now we are trying to integrate those into an enterprise. In addition, capitation has come along, and those sorts of things. And so the complex systems really make it so we have to be able to change, and we have to have some flexibility in that change.
Wouldn't it be nice, I think, if those 40 clinics had had the same coding system, and some of the same structure that the university hospital and its clinics had? It would have made the job tremendously less expensive, lots easier, and lots less problematic.
Well, I'll just take a little adaptation from Genesis and say, let us go down in unconfounder language that they may understand one another's speech. That's the position we are in now. I kind of got these out of order as you know if you are good Bible scholars. And the whole earth was one language, and one speech. I don't expect we'll come anywhere near close that, but we ought to move in that direction a little bit I think, because we are really in a Tower of Babel situation.
So what does it call for? I think it calls for a language and a structure so that we can exchange clinical and administrative medical data within a clinic, hospital, health care enterprise, and eventually community. If I am admitted to the university hospital, because I work over there two and a half days a week, and my record is at LDS Hospital, I ought to be able to share those records. That's my expectation. Why can't we do that? We can't do it for an infinite number of reasons. Basically, the best thing we can do is well, we'll call over and see if they can find the record and fax it over to you.
I have a little kid that I grew up with in southern Utah who runs a trucking industry. He has 160 trucks. I was talking with him the other day, and I said, well, are you using GPS technology? He says, oh, yes, we've got it on every one of our trucks. I know exactly where that truck is. He said, in fact, people won't ship equipment with me unless I have it on my truck.
I can't tell you even in my hospital, exactly where the patient is all the time. We've got a long way to go and other industries are far, far ahead of us. I mean I didn't think truck drivers would ever be ahead of physicians, and I think they are in some cases.
So that's what I think we are about. How should we standard a language? How should it be established? I thought of two possibilities. Development by private standards organizations, and development by government sponsoring standards organizations. We have had a little bit of this, and part of that was the discussion with the last group.
The advantage of the development by the private sector is it's the current existing methodology, and I think it's a good one. I've seen it work. I've seen it not work too. It uses the consensus methodology, which I think is absolutely crucial.
However, there are some disadvantages. It takes too long. Literally, I have waited 15 years to get a standard so that I can get data from IV pumps. I can tell you, based on experiments we have done, that the nurses only record half the IV drip rate changes. And by the way, they do it minutes to hours late, on the average of about an hour and a half. Is that good? I don't think so.
I have nurses who say, Dr. Gardner, we know you are a computer guy. Why do we have to read data off one computer screen, and hand enter it into another? So I think we need to move this thing along. It is too important to wait.
A problem that we talked about a little bit earlier, standards development would be controlled and owned by the organization that develops them. Now I think that is fine, because we have heard SNOMED, they spent $17 million on it. Surely they should get something out of it. But sometimes there is a conflict of interest I think in the development group, and the perpetuation of, and updating of standards aren't guaranteed. Let's say they hadn't decided to invest that $17 million. What would have happened then?
The advantage I think of a government sponsored standards organization, I think it can occur more quickly, though with some of the discussion here today, I'm not quite as sure about that as I was when I prepared these slides. The ownership and cost would be clear, and I don't think be too expensive. I think there ought to be a board of standards or something similar to that, that could establish and perpetuate and modify the standard language as needed. Clearly, this has to be a growing, living thing that will change with time.
So the disadvantages of it are that it would require establishment of a standards board, with funding and authority. In my viewpoint, I think there are a few other problems that exist with this approach.
So I would recommend, and I haven't cleared this with the AMIA board and so on, so this is Reed Gardner's personal recommendation, would be that the government agency be set up to establish and maintain a standard data dictionary, which would be free for public use.
Now the first strategy would be that some responsible governing agency take on the task by themselves, with an oversight of a public expert board. That would be one pathway that could be used. A second strategy would be some responsible government agency could contract with a private organization or firm to do the task, but they would also have oversight with some public expert board.
I think both of these things are doable, and I would highly recommend that the committee take a look at that. That's my presentation. Thank you.
MR. BLAIR: Thank you. George Arges.
MR. ARGES: Thank you, members of the National Committee on Vital and Health Statistics. I am George Arges. I am the senior director of the health data management group of the American Hospital Association. The AHA represents nearly 5,000 hospitals and health systems, networks, and other providers of care. We appreciate this opportunity to present our views on the development of a patient medical record system and its electronic transmission.
As many of you know, hospitals and health systems have a very strong interest in this area, particularly since patient records are at the heart of this information system development. The AHA is supportive of a well planned, adequately financed, and incremental approach for the development, adoption, and use of computerized patient record systems.
As many of you are aware, in November 1991, the AHA assembled a work group to identify practice steps towards the implementation of computer-based patient records. In April 1993, the work group issued its report to the secretary of Health and Human Services. The report mentioned ways that the medical record could improve the quality and efficiency of patient care. It further explained how computers can be used to manage patient information across time and place, and how they can be linked to practice guidelines and other decision support tools.
To do so, however, requires a health information infrastructure composed of several important components: a computerized patient records; standards definitions, codes, and formats, as we heard earlier; and high speed communication highways.
Since the issuance of this report, we have seen vast improvements in computer processing, storage, and communication capabilities. The good news is that these advancements occurred with a corresponding decrease in the cost for acquiring this technology. Despite these improvements, providers are generally finding it difficult to obtain the necessary financing in order to undertake such a massive information system investment.
One of the factors influencing their decision is the corresponding benefits also remain relatively small. Additionally, the passage of the Health Insurance Portability and Accountability Act of 1996 will soon require providers to undertake new administrative simplification provision requirements. These changes will impose additional costs on the institutional provider to upgrade their information systems, without corresponding benefits for the provider sector.
In terms of the questions in terms of interpretation on the congressional intent, clearly, we believe that the administrative simplification provisions were aimed at reducing the administrative costs through the adoption of standards, whether the involved the nine routine financial and administrative transactions as specified in HIPAA, or whether it would require implementation of other transactions that would be required at a later date.
The factors preventing or delaying the implementation still, at least from the computerized medical record is still the lack of a universally accepted standard for defining, coding, and storing the transmitted medical data. Another important reason for the delay is the cost to implement the system, which as I mentioned, was enormous. There is still little scientific evidence today that the implementation of such a system will reduce costs, or whether such a system will always lead to better quality and/or management of patient care.
In terms of the private sector, it is important to remember that the private sector can help define the patient medical record in terms that are more concrete. The primary purpose of the medical record is to document the course of the patient's illness and treatment received. Consequently, the patient, physician, and health care institution are the primary beneficiaries of its development.
Others also find the patient record as a useful reimbursement supplement, however these varied uses challenge us to define what additional information should be documented for inclusion in the record, and what can be maintained at the departmental level. Today providers are faced with the challenge of how to design an information infrastructure that can utilize the patient medical record for purposes other than what was originally intended.
We believe that it is important for the entire private sector to come to agreement on the type and amount of data and documentation that it is necessary to include in a patient record of any kind. Such an activity would be a good starting point for further computerization of the patient's medical record.
Doing so requires not only the financial commitment from a variety of organizations, but also a willingness to work through a lengthy process. Ultimately, the process should produce a national model that clearly defines the content and documentation to be routinely maintained in the patient medical record.
The government obviously has several important roles. First, government health programs like Medicare and Medicaid should actively be engaged in helping to define a core set of data standards pertaining to the patient record. Secondly, it is important for the government to establish a fundamental framework for the creation of a safe and secure national electronic infrastructure.
This must include development of national standards protecting the confidentiality of patient information, along with adequate security provisions that will protect the storage and authenticity of each electronic patient health record. There must be a national approach that preempts state laws, and can impose severe penalties for violation or misuse of patient medical record information.
The goals outlined in the 1993 report also indicated that the time line to create a viable computerized patient record is a long one. It laid out a series of incremental goals that included an estimate of the cost needed to complete each of these goals, along with a brief description of the benefits associated with its implementation.
As you look to making nationwide changes, we must be clear about how these changes would affect an organization's resources. The examination should include an assessment of the job skills, management commitment, and other hurdles or obstacles that may lay ahead for a provider to implement. Such an examination should apply more scientific rigor to the analysis, and provide a blueprint to the steps that should be followed.
Clearly, the content standards that describe information about the patient's care are extremely important. Earlier this year, a notice of proposed rulemaking concerning the HIPAA transactions called for the migration to the ICD-10-CM for diagnosis coding of disease and illness, along with the adoption of a new procedure classification system.
We believe that these changes alone will require providers to make large investments in their clinical information systems, not only with respect to their information systems, but a corresponding effort to train the staff on their proper application of these new classification systems will be necessary.
We recognize that today the Internet is changing the way our nation exchanges information. As we look to enhance communications and promote continuity of care among physicians and other health care professionals, we need to look at how the new electronic standards should apply, and the availability of a secure, high speed electronic network. Otherwise, the full benefits of establishing an electronic patient record in itself will be lost.
Finally, the NCVHS should keep in mind the price tag associated with implementation of such a massive undertaking. Already, the fix to the Y2K problem will impose a larger price than what many within health care were willing to absorb as part of their operations. Upcoming changes to the clinical code sets will further strain a provider's financial operations. We need to be sensitive on how these massive information changes are affecting the delivery of patient care, since today financial resources that would be earmarked for patient services may have to be diverted in order to solve problems such as the Y2K.
Providers and others must be certain that plans that call for the implementation of an electronic patient record are the right ones. Moreover, it is important that the NCVHS give thoughtful consideration before any massive nationwide deployment takes place. In the meantime, establishing a fundamental framework for the future infrastructure needs to begin, but one that allows providers the flexibility to gauge when their operations should move in this area.
Again, I would like to thank you, and we look forward to working with you on this important endeavor.
MR. BLAIR: Thank you, George. Tom Bozeman.
MR. BOZEMAN: Thank you. I'm Tommy Bozeman. I'm vice president, chief information officer for North Mississippi Health Services, headquartered in Tupelo, Mississippi, birthplace of Elvis Presley, if you've not heard of Tupelo before.
I would like to give a brief background on health services, because I think it would help you to understand some of my answers later on in my testimony. We have some definite experience in developing the computer-based patient record. We have been at this for some 15 years now, and have learned a lot of things not to do, and things to do.
Our health delivery system reaches out 90 miles from its home base in all directions, and serves a population of about 600,000 customers and patients. We have had a goal for some time that we would create a computer-based patient record that would serve that population. That is, we would have the patient's complete clinical information no matter where in our delivery system that patient had received medical care, and no matter where in our delivery system that patient would next show up for care.
Our wide area network touches 125 different facilities or buildings from physicians' offices to nursing homes to hospitals to home health agencies to dialysis facilities in that region. That represents 28 communities.
Obviously, the computerized patient record has been of some concern to us for some time. The work that we have done in achieving the level of automation that we have achieved at this point in time gave us recognition in 1997 of the Davies Award, which is given by the Computer-based Patient Record Institute for excellence in computer-based patient records.
Also, in 1997, PC Week magazine recognized Health Services as number 31 in an international ranking for use of technology. Primarily, the technology was the implementation of that wide area network to serve all of our care facilities.
About five years ago we started the development of the computer-based patient record outside of our major facility, North Mississippi Medical Center, which is a 650 bed facility in Tupelo. That's the largest hospital in rural America.
Also, something else that we have done that I think is pertinent to some of the standards we will talk about is that we have developed and created a common patient identifier for our consumer base. The common patient identifier we think is basic in that you have to have some place to store a patient's data, and there has to be some identifying number for that doctor to insure that he or she is looking at all of the medical information or clinical information on that patient.
We also have implemented a magnetic stripe patient identification card. We have 750,000 of those cards out, which means that probably 1 out of every 2 consumers in our area has one of the cards. That card has the common patient identifier magnetically encoded, as well as printed on the card, and at any point in our delivery system all the patient has to do is present that card. It is swiped in through a swipe reader, and that brings up the patient's complete medical record.
Currently, we have clinical records on over 275,000 patients, representing 1,235,000 episodes. That is since March 1995. That represents about 64 gigabytes of data. A lot of the information we have is text, but also we have a separate database called Cost Information Decision Support that we take coded data into, that we use for clinical efficiency studies, and for research.
In our environment we do not chart with pencil and paper. All of our charting is done electronically via mouse or light thin(?). All of our nurses, all of our clinicians, a lot of our doctors are using the computer. Some doctors still are using the pencil and paper, however.
Now to the questions. The first question, how do I interpret the congressional instruction? My interpretation is that the committee was charged with the responsibility of studying the issues related to the adoption of uniform data standards, the patient medical record information, and the electronic exchange of such information, and to report to the secretary of Health and Human Services not later than four years after the date of the enactment of the Health Insurance Portability and Accountability Act recommendations and legislative proposals for such standards in electronic exchange.
The second question is what factors or issues are preventing or delaying the development and widespread implementation of uniform standards for patient medical record information and its electronic transmission? First, I would say generally you have a lack of top management buy in and support within the hospital industry to funding the creation of an electronic medical record.
The question administrators always ask is what is the benefit. They can see the benefit of an electronic exchange of claims data, because payments come in faster and they are more accurate, however, they have not seen the benefits of the electronic medical record. And we are just now beginning to see some of the newer clinical decision support software that is coming available. It is probably too early for them to understand that the newer generation software really can reduce costs, and can make us more efficient.
But until those systems are deployed, and there is more evidence of savings, there will be a lack of complete support by top management. Most health care organizations do not believe health care is an information business. We do. We believe health care is an information business.
Those that don't believe it's an information business don't allow their organization to have a lot of IT staff, and the amount of money that is spent as a percentage of revenue is not going to be the average that you might find in other industries. I think banking spends 9-10 percent of their revenues in information technology. In health care we are lucky if we are averaging 3 percent. A lot of places are averaging 1-2 percent.
There is a lack of buy in to automate of nurses, physicians, and other caregivers. It is difficult to get physicians to enter clinical data such as progress notes unless they are paid house staff or residents. Our physicians have told us that they would not embrace automation unless it makes them more efficient and saves them time.
There is also a lack of current automation. The hospital health care business or health care terrain is littered with millions of dollars worth of failed systems, delayed, failed implementations. That causes hospital administrators and boards to be a little slow in approving additional funding for new systems.
Cost and lack of funding, the Balanced Budget Act is taking its toll on health care's bottom line. This year for example, we received $10 million less from Medicare than we did last year. We took a big hit in home health. Inpatient revenue or the inpatient side of our business -- that is most hospitals' cash cow -- is projected to continue to decline until around the year 2002, 2003, and then because of the aging population, is expected to begin to increase. So we are facing tremendous pressures on our bottom line, going at least until the year 2002 or 2003.
One of the major problems in health care is the lack of integration of systems. I term it islands of information. There are a lot of those out there. That's one reason we need common patient identifiers, because in a lot of health care organizations the different departments or different systems within departments don't use the same patient identification number.
The health care industry is consolidating, much like banking, but the information structure or environment is beset with having lots of disparaged systems that are acquired through this purchasing of other health systems. And it is taking those IT organizations years, and will take them years to bring them into common systems. Integration is a lot more difficult to achieve than vendors would lead you to believe. It takes time, time, and a lot more time.
Health care is complex. I'm sure you've heard that today. We did a year 2000 or Y2K inventory of all of our software a little over a year ago. We had over 300 different brands, kinds of software in our organization. Some will have more than that. We have pretty stringent standards, yet we had over 300.
There is an issue of archival, historical storage of clinical data. Most health care facilities maintain some type of automated clinical data today, but most of the data is purged. It is not maintained or retained for any length of time. Then that brings the question of how far back does is the information important to a clinician, and how much should each individual facility retain?
Our physicians have told us that they are concerned about the accuracy or integrity of clinical data that they might get from another provider. They want to know who is responsible for the accuracy of that data.
The common patient identifier I have mentioned. To me, that's a problem, but it's a prerequisite for having any kind of electronic record that you can exchange with other facilities.
There is an issue of an on-line, real time clinical system containing lots of text that caregivers use in minute-by-minute concurrent care of a patient versus a database with standard definable fields of data that can be queried and analyzed retrospectively. The real time system is rich in clinical data, but lacking in standard data sets.
There is the issue of standardization of clinical data. For example, potassium normals may be different, depending on which manufacturer's test equipment you are using. That's a real problem to physicians. You can say a lot about technology, but there is not a system today that we can buy in health care that would achieve any type of total electronic record.
As far as standards, it appears to me, and I'm not an expert in this, but there appear to be many specialized interest groups pushing for their own standards. I don't think that is something that we're going to be able to go forward with if we are going to achieve one standard data set for health care.
Then there is a whole issue of clinical automation in the smaller providers; one and two doctor clinics, small hospitals. I don't know how we are ever going to get them automated. You are looking at mostly handwritten paper records in those institutions.
Security and confidentiality, I know there is a whole other work group working on that, but that is an issue.
Then I would also highlight that the remediation of the Y2K problems in health care have set IS back 6-18 months. Most of us have been focusing on that, and not on new clinical technologies as a result.
The third question, is the private sector able to address these problems satisfactorily? I believe the private sector can develop and implement the electronic record over some reasonable period of time. What I'm not sure of is the private sector being able to decide on standards in a timely fashion, and then the subsequent setting of some realistic target dates for the implementation.
The second part of question number three, what is the role of government for assisting the private sector? I have four things I would point out. One is coordinate development of standards, solicit input from the health care community to insure understanding of the complex challenge.
Secondly, coordinate with the health care industry the setting of reasonable time frames for implementing, however, it must be segmented into manageable and achievable components.
Thirdly, I would like to see incentives to get providers to implement these standards and these electronic records as soon as possible. Initially, if we could find some way to fund and publicize demonstration projects, I think that would be very valuable.
The fourth question, which standards related to the patient medical record information and its electronic transmission would: (a) add the most value for improving the quality and efficiency of health care for the nation? Why?
I think again, the common patient identifier has to come first, because you have to have some way to consolidate that patient information so that a caregiver, a physician can feel very comfortable that they are looking at a complete clinical record, and maybe not 75 percent of it, with 25 percent of it missing. There obviously is a risk factor to the physician if that happens.
There is some minimal data that would be very important in the care of a patient. I think we are talking about care of a patient here. That's why we are talking about spending millions and zillions of dollars for an electronic medical record. There are things that any physician or any caregiver would love to have if they are seeing a patient new for the first time: baseline lab results, baseline chest x-ray, baseline EKG, history physical, allergies, current medications, any significant medical event, like an open heart surgery, or is the patient in renal failure, and undergoing weekly dialysis procedures?
I think there is also some merit to initially utilizing the coded data that we already have. I think there have probably been some comments about the ICD-9, the CPT4, and HCPCS codes not being that accurate, and that people tend to code for billing. But I think given what I have seen in the last year or two, coders, physicians, caregivers are getting much more accurate in their coding, because they know they are going to be held accountable for it.
What would add value, and what would cause us, with our business goals, to do more to achieve the computerized patient record? I think there are two things that we would be looking for in our business. One is if we can use standards, if we can use data to achieve better clinical efficiency, that should allow us to lower our cost. Lowering cost is something we've got to do.
Also, the more we can standardize data, the more we can put this data in large databases, the more number crunching we can do, the more research, the more clinical efficiency we can get out of data that may look at a particular type of practice over several years, may look at a specific physician, that is the kind of thing that we are going to have to do going forward if we are going to remain competitive and be able to survive with less and less reimbursement.
The business case for more rapid standards for development and implementation; I really didn't have an answer for that. I'm not sure there is a business case for more rapid standards and implementation, but I do think we need to move forward, because if we don't, we will be here 10 years later, and we will be no better off.
That concludes my testimony. Again, thank you for inviting me, and letting me have a few minutes to talk with you.
MR. BLAIR: Thank you, Tom. Simon has pointed out to me that in addition to our time constraint, some of the people in the audience haven't had a chance to ask questions. So let us proceed where each person on the committee has one round of asking a question, and then we'll give the second round to people in the audience. Who is the first on the committee that has a question?
DR. MC DONALD: It's not exactly a question. I would just like to thank all the participants. That's a new idea I think. I don't think we heard Reed's idea specifically, and it's worthy of considering, thinking how that could really happen.
MR. BLAIR: Simon, will you help me identify?
DR. COHN: I'm next in line. Let's go down the row here. Tommy, I had a question for you, because I was a little confused, and I just wanted to understand a little better, your testimony. You had talked about -- I guess I got the sense that you had real questions about the value of your information system investment, or at least the perception of the CEO in your organization had some questions about your information system investment.
Yet I was observing that you had a very large data repository that had 65 gigabytes of information from 1995. Actually, I couldn't quite decide whether these were standard coded data. You had listed ICD-9 diagnosis codes, surgical codes, procedure codes, doctor codes, HCPCS codes. Is that what is in your data warehouse? Are those coded standardly? Tell me more.
MR. BOZEMAN: Let me answer the first question, and that is, is do we believe our investment is valid, and going forward, will we continue to invest? What I tried to say was I feel like most health care organizations don't feel the same way we do about the value of information, and about the value of a computer-based patient record.
We are committed to it. We believe it works. We are beginning to see some evidence that having a lifetime clinical record for the patients or consumers within our region is beginning to save us redundant testing. If a patient was in one of our rural facilities 50 miles away yesterday and got some lab work, and they are transferred by ambulance to our main facility, we have access to those lab results from that hospital 50 miles away, on-line. It is part of that patient's record.
Our doctors are telling us they are finding those examples that they are not having to redo tests, because they have the information. So we bought into that, and we continue to fund IT in our organization at $20 million a year; $10 million for operations, $10 million for capital budgets.
We have two databases, yes. The 64 gigabyte database is the on-line, real time clinical record that caregivers use every second of the day. We do over 500,000 transactions against that database every day on a 24 hour basis.
Now it maintains all of that coded data, but it has a lot of text too. It has everything that is dictated, is transcripted into that record, so there is a lot of text. Physician's progress notes are in there. But the coded data we off load to a research database called the Cost Information Decision Support Database. So we are maintaining the coded data in two databases. One is for the caregiver, the other is for research.
DR. COHN: So this list of fields was coded fields? In terms of the last page here, you talked about a list of eight or ten different things. That's in your cost information?
MR. BOZEMAN: That's in both places. We front end all of the data. There is a transfer of data. There is no redundant entry of data in our systems. All of that coded initiates somewhere with a caregiver or a coder or an employee.
DR. COHN: Let me just follow-up on the question here. Let me just ask the direct question then. Does having this data in coded form, is that valuable to your business?
MR. BOZEMAN: Absolutely.
DR. COHN: So some data is valuable to your business, and some isn't.
MR. BOZEMAN: The coded data that is valuable to our business in the Cost Information Decision Support System is used by a group or a department within Health Services called the Clinical Efficiency Department. We have actually published some articles in different journals on how we have reduced patient length of stay substantially over the last three or four years, how we have reduced the average cost for procedures.
What we do every year is we take the ten DRGs that we lost the most money on, comparing it Medicare reimbursement, and we go to work on those DRGs using that Cost Information Decision Support System to determine what we may be doing wrong clinically. Are we ordering too many tests of one kind? So we get to the best practice with those ten losers. One year we reduced our total patient care cost systemwide because of that Cost Information Decision Support Database.
DR. COHN: Thank you for the clarification. Jeff, do you have a question?
MR. BLAIR: Not right now.
DR. COHN: Kathleen?
MS. FYFFE: No, I don't.
DR. COHN: John Lumpkin?
DR. LUMPKIN: I don't know how to ask this question, so bear with me. I'm sitting here trying to puzzle over what has happened in the hospital environment with the consolidation, where there has been some major chains that are now divesting themselves of some of their holdings.
But it would seem to me that that environment would tend to engender some standardization in information systems. So my question is, what is happening in the marketplace? To what extent could government impact the marketplace either by incentives through HCFA or other places to create the environment where standardized systems would develop in the hospital, a nucleus to spread out into the communities?
DR. GARDNER: Let me try that. I think some incentives to do demonstration projects would be good. The second thing, somehow someone chose CPT codes. They became the standard, then everyone used it. Someone ought to decide somewhere along the way -- and I gave you my options as to how that might be done -- as to what we ought to do. What should we use for drug codes? What should we use for laboratory testing codes and those sorts of things?
I think there would be tremendous value -- even within our own organization there would be tremendous value in doing that. We could say, hey, this is a national standard, let's go do it.
DR. LUMPKIN: George, on this one, is this something that when the CEOs get together -- I know there are some notable ones who have really been committed to this, but is this the kind of thing where someone talks about it, and everyone else in the room glazes over? Or is this an issue that is coming into the fore?
MR. ARGES: It's an issue that is coming to the fore, but I also think too Tommy basically mentioned you have a situation or environment such as the Balanced Budget Act, which earmarked $24 billion in cuts. Those are very real for the providers, and they are looking to find ways to basically respond to those cuts, and whether to stay in a particular service sector or whether to partner with somebody else, or not to partner with somebody else.
These are getting in the way in terms of standards development. You can talk about standards, but when you have these very real, immediate things thrust upon you, you are basically going to have to try and respond to how to basically extricate yourself into the most favorable situation.
DR. FITZMAURICE: George, you probably heard me ask an earlier panel about the benefits and the costs of the computer-based patient record, the access to data, the access to medical knowledge. Let me ask you, are there any research activities underway at the American Hospital Association to try to tease out where the benefits of -- to measure the magnitude of the benefits of a hospital computer-based patient record, compared with the large costs of installing and maintaining one?
Is there any value in trying to obtain this data, and then feed it back to your constituents? Or is this something that they really would rather go alone on?
MR. ARGES: There isn't a formal process or study being done by the AHA on this. I do think though there are important activities underway by some of the members in terms of the computerized medical record. Pointing to what has happened in other facilities, I think that would be a valuable tool.
Bringing them I guess resource material that would help them understand how it is being applied by others is important. I think too, some of the components aren't quite there yet. The ability to have this in terms of real time, easy to access, easy to apply or in terms of the technology still requires money, and still requires high speed communication in standards.
Until you have all those three pieces, and you have the luxury of money to help fund it, I don't see a big thrust to move towards that area.
DR. FITZMAURICE: One question for Tommy. That is, I note that you have a common patient identifier throughout the system, randomly assigned, not a social security number or derivative. So the patient comes in and gets a randomly assigned number. The patient comes into another part of your community, forgets the number. How do you match up the number? How do you get the number back to that patient?
MR. BOZEMAN: If the patient does not bring their card, we go through the normal process of trying to identify that patient. Again, we have all the patients, the 270,000 some odd thousand patients. We have the demographic information sitting in there as well.
DR. FITZMAURICE: That would include the social security number?
MR. BOZEMAN: It does include the social security number. We do capture social security number. We just opted not to use it as a common patient identifier, because there seemed to be a lot of concern on the national level that that was not the thing to do, so we shied away from that, but we do have it, and we can cross reference to it.
DR. COHN: Any others around the table with questions, comments? We would like to open the floor for a moment, if anyone has a question or wants to make a statement.
MR. HEFRIN(?): This is a short footnote on today's presentations. What is interesting is the hearing today discussed issues that were really over the horizon of this committee two years ago. But now they are very much front and center, namely, the issue of even if people do use common codes, terminology, et cetera, the meaning of the people who enter that data can be different, and therefore we don't get the desired semantic consistency even if we use standards. There is something more that is needed, and that really hasn't been teased out; the discipline in organizations, and coordination.
For your information, over the past two or three years there has been an effort in the NCITS, National Committee on Information Technology Standards T2 committee to address this issue of pulling together the over 20 years of work of leading researchers and knowledge representation, artificial intelligence that has addressed that topic in artificial intelligence.
They have met. They have compared the work that has been done. In the beginning of the 1990s, DARPA recognized that this was a ripening fruit, as it were, and they put substantial money into the coordination of knowledge bases.
From early 1996, the T2 committee addressed how can we put this in a standard framework. The goal would be to have this electronic infrastructure for semantics, underlying terminologies, code sets, and all the rest, leveraging off this over 20 years of research among the best and brightest in the field.
They developed a plan of work. They recognized that the output would need to be non-proprietary, available to everyone. Essentially, the reversing of the Tower of Babel historically, for the whole human race. Now that is perhaps a little bit fanciful description, but nevertheless that was the goal. That's what they understood.
They developed a plan of action over the past two years. They know what needs to be done. They cannot find a funding source that would make this available. My suggestion to the committee is that you consider putting on your agenda looking into this opportunity, because if it were to go forward, the potential is that a decade of attempting to resolve all of these semantic inconsistencies between everyone using standard data sets, which hopefully we'll get to, could be resolved by having this common electronic open semantic infrastructure supporting all these systems.
So that's the suggestion. I will provide more information, and get it to you in electronic form so it can be part of the record.
DR. COHN: Thank you very much. Other comments, questions?
DR. MC DONALD: I wanted to go back to Mr. Ages about the AHA and your response about medical records. My world is that the hospitals all want computerized medical records, if we are talking about the repository level, if you are talking about gathering all the data in one place.
Wherever I go, like little places in Indiana -- Deaconess Hospital in Indiana, an LaPort Hospital, a 106 bed hospital up in northern Indiana -- they all are trying to build or are building what might be called a repository records, not so much the act of physicians.
Now have I just had a totally whacko life experience in that? I haven't met anybody in big hospitals that isn't trying to, or buying one, or ordering one. Every major hospital in Indianapolis has, or is buying some kind of a medical record repository thing.
DR. COHN: A data warehouse?
DR. MC DONALD: Not a data warehouse, because that is usually applied to the business side. I'm talking about active clinical data.
MR. ARGES: I would say if they haven't either already started one, they probably already have some form of computerized medical record of some sort. The question is, to what components, and how standardized are they across sites.
DR. MC DONALD: That actually reassures me that I haven't been hallucinating for the last five years.
MR. ARGES: I would say I think they do want more standardization in this process, and one which would be more of a turnkey operation if possible.
DR. MC DONALD: I think that maybe some members of the committee and the audience might have gotten the idea that nobody in the hospital cares or wants. I think there really is a high demand. We have residents come to our program using the word for like a substance like sugar, they have got the computer. Apparently that is a good thing. If they come and say they don't have the computer, then they elect to go somewhere else where they have the choice.
MR. ARGES: The only comment I will say is I was encouraged by the last panel in that you made the point that this would not be something that would be mandated. I think hospitals would be concerned about something being mandated, because what they may have invested in may be obsolete, and therefore they may be forced into investing in something different. To me, I think that would be problematic.
DR. COHN: Actually, maybe just to follow that question up, I know that there have been instances where you have consolidation of two hospitals and things like this. Is that not an issue for those hospitals that are acquired or consolidated? Is that an issue for the American Hospital Association?
MR. ARGES: It is an issue, and I know some of our members basically have talked about they need help in terms of helping to integrate. Clearly, some of the HIPAA standards would help towards that. Establishing I think some common identifiers would be important, some common structure around the code sets would also be an important component.
But again, I think a lot of them right now are really struggling with the business issues of integrating the organization as part of the operation. I don't mean it from a computer system, a clinical system, but really making certain that they made the right decision in terms of investment in return.
MR. BLAIR: George Arges, let me clarify something I believe you said. I thought you made the comment that the implementation of standards for electronic data interchange for financial transactions, which was in the NPRM that was released earlier this year, that you said that's going to be a significant cost burden on hospitals.
There is no doubt in my mind that it means that information systems will have to acquired and modified if a hospital chooses to send its claims in electronic data interchange format. The law does not require that they send it in electronic data interchange format. Is it perceived that it will be less expensive to continue to do it in some paper form?
MR. ARGES: No, actually what I was referring to is the fact that many of the hospitals currently submit electronically. They submit the flat file specifications. For them now to move to a different format would be a costly proposition. There really is no additional gain in terms of time. What they are incurring would be additional conversion costs or premiums to be paid to a clearinghouse to convert the data to the standard format.
DR. LUMPKIN: I think we have heard the message pretty loud and clear about not establishing standards for how these things operate within your shop. My question is that hospitals are not an island, separate unto themselves. That they operate frequently, rumor has it with physicians and other providers of care, who aren't based at those particular facilities. Not uncommonly, those physicians and other caregivers may be related to one or two hospitals if they are lucky enough to operate in an environment where they have choices, and aren't in a rural setting.
What would be the position on looking at not how a hospital keeps their data, maintains their medical records, their repository, whatever, but if government were to say if you send this medical information, John Jones' medical record, or information contained in there from your hospital to Dr. Smith, that that would be in a format prescribed.
Is that conceptually different than government saying this is how you should run your computerized patient record or medical record repository? That's assuming of course that we can find a consensus standard, much as we have done with the X12.
MR. ARGES: Is it different? Yes, it's different in terms of the requirement. Would it be more palatable in terms of accepting that format standard? Yes, I think we are looking for whether the electronic format standards and data content standards, we're looking for that sort of thing in terms of guidance, and I think that would be acceptable.
Then you know if that format will have the utility over a period of time, then you can recoup some of your investment costs. As long as that format has some stability, and is going to be universally accepted, then you are willing to make the investment, and it does have utility. So I think that would be acceptable.
You are then not prescribing more of the operational functions of what may make one facility slightly different in terms of how they network within themselves, and outside themselves as well.
DR. LUMPKIN: Let me ask a follow-up. Then if we were to do that, let's say because we've got all these new and different arrangements, it may be that a hospital and its partners, the physician staff organization, those kind of arrangements, where there are the new financial arrangements between hospitals and physicians, where they are actually captive, that would be considered to be inside the hospital, as opposed to a physician who may on staff, but is not part of those risk bearing arrangements.
Would you see those kind of arrangements would be concerned about communicating with their physicians in a standard format?
MR. ARGES: Yes, you raised really I think a number of other issues as well. I think there are also some legal issues that one has to be concerned about too, and it's the quality of the information that you are providing, how reliable is it. Would an organization be subject to I guess legal ramifications if the information is incorrect, and somebody relied upon it, and it wasn't part of your organization that did it.
I think it's important though that that information somehow be made available, at least for review, but it may be a disclaimer or something like that. Maybe I'm going off on a different side, but it reminds me a little bit of the Y2K problem. When you ask a vendor to give you assurances about whether their equipment is Y2K compliant, and the answer today is be silent on it, because it could open you up for a lawsuit.
DR. COHN: I think there was no answer there? A final question and then we need to stop.
DR. FITZMAURICE: When you start ranking hospital systems according to their excellence in use of information technology, you certainly have more than one in this room that are highly ranked. I want to address the question to Reed Gardner. Your hospital is often put high in that ranking. The question that inevitably comes up is not which system did you choose. How did you get the standards, but how did you get the cooperation of the physicians?
We heard John Mattison talking about the incentives to the different actors involved in making a system work. So let me ask that question of you. How do you get the cooperation of physicians at LDS to use the system and achieve the benefits that you showed us up on the screen?
DR. GARDNER: Physicians are principally data reviewers. Anyone loves data if they can go get it, and they can get it easier. In the antibiotic thing they showed it takes them five seconds to get what normally would take them 15 minutes to do. No one would take the 15 minutes. So it's an incentive. The data is there. They aren't the data grubbers. That's principally automated instruments and nurses.
They have come to rely on it. If we took it away from them now, there would be a lot of trouble. They wouldn't know how to do things. My own son just graduated from medical school about a year and a half ago. He said, Dad going to LDS Hospital is like going to the country club, because I can go over for rounds and be there 20 minutes before rounds, see my patients, get my data, and get the job done. Whereas, if I go to another hospital in the community, if I'm there two hours early, I can't get the data. So I think having access to the data has been very, very valuable.
There was some comment earlier about people being frightened about big brother and all of those sorts of things. I just finished a questionnaire to our physicians on that subject, and much to my surprise, there is not a great fear for that, that the administration is going to use this data to beat them over the head, or that somehow the government is going to get access to it and beat them over the head. This is from a very, very conservative, right-wing, Republican, whatever you want to say it, community. It really surprised me.
DR. COHN: Clem has one last question.
DR. MC DONALD: It's just a follow-up to a comment. I think the question is one that reflects -- a lot of people ask that question, and it's really two answers. When they are getting data from the computer, I don't think there is any place physicians don't love it. It just has been a universal experience.
This idea of physicians holding back is usually on the input side. The struggle that John Mattison had was getting them to put all their notes in. So I think you've just got to keep that distinction in mind.
DR. GARDNER: I agree with that. Putting data in is very troublesome, however, the antibiotics assistant, we have become smart about that. Physicians, when they write an order, they scribble out a little line, and from that in the past that then is expanded either by nurses, by clerks, by someone into a whole series of other processes. Somebody else takes care of all the details.
If we can be smart enough to do that with a computer, and we have done that with antibiotic assistants, where if they want to order it, they hit an asterisk key and an enter. If they want to change the dose, it is very easy to do. So we've got to be smarter about that process, and emulate more the manual system, which is a little scribble, and then someone else takes care of all the detail. We need to be a lot smarter about that.
DR. COHN: Well, with that, why don't we adjourn for today. Now we will start back tomorrow morning at 9:00 a.m.
[Whereupon, the meeting was recessed at 5:12 p.m., to reconvene the following day, Wednesday, December 9, 1998, at 9:00 a.m.]