Hubert H. Humphrey Building
Room 705A
200 Independence Avenue, SW
Washington, DC 20201
TABLE OF CONTENTS
Call to Order and Review Agenda - Mr. Blair
First Panel - Health Data Quality:
Second Panel - Health Data Quality and Users:
Third Panel - Users of PMRI:
Briefing Letter to the Secretary
MR. BLAIR: Let me begin here. My name is Jeff Blair. I chair the Work Group on Computer-based Patient Records. The purpose of this work group is defined by the HIPAA legislation, which is to study uniform data standards for patient record information, and the common exchange of that information. We are still in the data gathering phase. We will report the secretary of HHS will be the August of next year with respect to our findings and recommendations.
I'm going to ask the other members of the committee and staff to introduce themselves.
[Introductions were made.]
MR. BLAIR: The last thing I'm going to mention before we begin our testimony is just to put things in perspective, I'm kind of addressing those folks that are listening in on the Internet. Our work group was formed, as I indicated, to study uniform data standards for patient medical record information, and the electronic exchange of that information.
Back in December of last year we agreed upon a set of focus areas. Those focus areas include: the study of message format standards, medical terminologies, the business case issues, the differences in state laws, how we relate to the emerging national health information infrastructure.
We have had hearings in March and May and June so far. And besides today and tomorrow, we will have some additional data gathering hearings in October as we begin to gather the information on all of these areas.
Our first two panels this morning are primarily focused on the areas of data quality, and how that affects uniform standards for patient medical record information. May I invite please, our panelists to introduce themselves?
MR. LAMB: Good morning. I'm Stephen Lamb. I'm the vice president of public policy with the National Committee for Quality Assurance.
DR. BUCK: Good morning, I'm Al Buck. I'm the executive vice president of the Joint Commission for standards, performance measurement, and research.
DR. JESSEE: I'm Bill Jessee. I'm president and chief executive officer of the Medical Group Management Association.
MR. BLAIR: This can be a self-selection process. Can you go from left to right, and go ahead and begin your testimony. You do have a little more time than we had indicated to each of you. So I think that you could probably feel comfortable maybe doing about 20-25 minutes apiece.
Michael, is that a correct timeframe for each person?
DR. FITZMAURICE: I would guess about 15 minutes plus or minus. Nobody feels bashful about taking less, and if you need a couple more minutes to fully explain your positions, we would welcome that too.
MR. BLAIR: And then we'll have time afterwards for questions.
Agenda Item: First Panel - Health Data Quality, William Jessee, M.D., MGMA
DR. JESSEE: Mr. Chairman and members of the Work Group, the Medical Group Management Association is pleased to have the opportunity to present our testimony to the National Committee on Vital and Health Statistics Work Group on Computer-based Patient Records. As I mentioned earlier, my name is Bill Jessee. I am a physician. I am the president and chief executive officer of MGMA.
Just a brief word about our association for those of you who may not be familiar with it. MGMA is the nation's oldest and largest medical group practice organization. We represent some 8,300 physician group practices in which just under 211,000 physicians practice medicine. Our membership reflects the full diversity of physician organizational structures today, including large world renowned integrated delivery systems -- Ford, Mayo, and others; multi-specialty clinics; small single specialty group practices; hospital-based clinics; academic practice plans; as well as management service organizations; and physician practice management companies.
We are very pleased the NCVHS has invited MGMA to testify on one of the most important issues facing medicine today, the computer-based patient record. The computer-based patient record offers far more than simply a new method of maintaining information. For those practices that have the technical capability and the financial resources, electronic record systems have the capability to integrate information from multiple sources, and can pave the way for improved clinical decision-making, reduced administrative cost, and enhanced research opportunities.
In my testimony today, I will be outlining some of the issues, concerns, and expectations that medical group practices have with this promising new tool for patient care.
Let me very briefly begin trying to identify some of the key components of the patient record as we see them. Obviously, the record includes information about an individual's health status and health care, as well as a variety of administrative data. This information is characterized by several features: (1) data originates from multiple sources, for different uses, by many different users; (2) data many be in various forms including text, voice, and image; and (3) data cover a range of time spans from a single medical encounter or patient's lifetime chronology health information.
A comprehensive record should include at least the following elements:
First, the general information including demographic data: patient address, phone numbers, age, occupation, social security and other identifying numbers; medical history; genetic history, social and family history; and legal information.
Secondly, referral information, which includes the name of referring or consulting physicians; diagnoses or reasons for referral; insurance authorization information; appointment information; and follow-up notes.
And finally, diagnosis and treatment information, which is what I think most of generally consider as "the patient record," including history and physician findings; laboratory, radiology, and pathology reports; medications; allergies, immunizations; reminders; diagnosis; treatment procedures; and follow-up.
In addition, we think it is very important that any electronic patient record have the capacity to be linked to patient financial information, including insurance data, charges, procedure codes, procedure charges, number of relative value units, authorized visits or procedures, number of visits left in a treatment protocol that may be authorized by a health plan, and insurance expiration data.
While these financial data are not part of the record, and should be stored and maintained in separate files, it is essential to insure that linking elements are included in both medical records and the financial files to permit analyses of effectiveness and efficiency to be conducted for purposes of practice improvement.
The limitations of the traditional paper-based record are well known. Among the many concerns with data quality and completeness in paper records are such things as: lost or misfiled records or record components; illegible handwritten entries; unrecorded observations or interventions; or inaccurately recorded observations or interventions.
As a consequence, physicians and other providers find that maintaining accurate information on such important care elements as current medications, active problems, immunizations or preventive services status is almost impossible. This problem has been compounded by a highly mobile patient population, which may be frequently forced to change physicians as a consequence of insurance requirements, coupled with a lack of portability of the paper record.
An electronic record would contribute greatly not only to improving the quality of information in that record, but also the patient's ability to assure physician access to that information whenever necessary. I would also indicate how important it is that we not ignore the issue of user interface issues.
One of the greatest barriers to the use of electronic records right now is that they are not well integrated in the care processes. One could make an argument that that's one of the greatest problems of the paper record as well. It is not well integrated into the care process. As a consequence, it is not well kept, and the quality of the data is quite poor.
If we are not to simply substitute an electronic problem for a manual problem, we need to look very carefully at issues of standardization of the approach to integrating the user interface into the care process to the greatest extent possible.
Clearly, the growth of computer-based records has addressed many of the mechanical problems inherent in paper records. Data are less likely to be lost or misplaced. Legibility has been enhanced, and access for both patients and their physicians has been improved.
But the absence of standardized formats and data definitions has created a continuing cacophony of data, little information, and has inhibited the utility of the computer-based record for quality and cost studies, research, and patient care improvement.
Obviously, there are a number of advantages to standardization of data, data elements, data definition, data format. Uniform medical records format stored electronically can make records and record keeping, and by extension the analytic data derived from those systems more uniform and more accurate. Standardizing patient record data will enhance the ability of practices to benchmark their performance both internally and in comparison with others.
And finally, standardization of patient records will allow the full integration of clinical and administrative information. With this integration will come the ability to greatly enhance practice productivity in the protocol-driven processes that may range from patient schedules to diagnosis, treatments, and so on.
Let me spend a few moments talking about the areas in which from our association's perspective, there is a need for standardization. Obviously, the first of those areas is for clinical functions. Although medical tradition has encouraged some degree of standardization of the major components of the medical record, I think every medical student is taught about chief complaint, history of present illness, social and family history, review of systems, physical examination and so on, there remains in practice great variability in the content of each component.
This variability occurs in both paper and electronic records, and is a major barrier to interchange, benchmarking, research, and a variety of other activities that require pooling of data from multiple sources.
One possible approach to achieving consensus on the content of each, which is a critical first step toward standardization might be through the use of what David Eddy has termed "clinical logic." For the most common presenting complaints, or common chronic illnesses, there is a base of evidence sufficient to construct algorithms that depict the logical flow of the clinician's evaluation and decision-making process. Where the evidence may be weak or equivocal for a particular intervention or decision point, it can be supplemented by expert opinion.
Once such clinical logic tree has been constructed, it lends itself to the construction of a series of standardized queries that can then produce a standardized record while still accommodating the variability inherent in patient-specific differences. Electronically, these logic trees can then be readily converted to point and click menu-driven tools for collecting standardized data on most patients much of the time, and I emphasize most and much.
It is imperative to emphasize the importance of focusing on conditions of high prevalence for which there is good evidence on appropriate management. Any effort that attempts to standardize all the data in patient records all the time is likely to be unsuccessful. I think if you will, we need to pick some low hanging fruit here before trying to move forward on a broader front.
In addition, electronic records offer the potential to incorporate real time recommended treatment options, and highlight potential treatment conflicts, generate reminders for patient follow-up, prescription renewals, and so forth.
Just as a sidebar, and a reflection on the work that the agency is doing, after spending a number of millions of dollars in developing about 18 sets of clinical guidelines which are now accumulating dust on many bookshelves around the country, AHCPR has shifted the focus to try and increase the access to those guidelines. The National Guideline Clearinghouse is moving in the direction of doing so.
A logical next step would be integrate access to the guideline clearinghouse, and the guidelines that are contained therein through electronic medical record systems. So at the time the patient is being seen with congenital heart failure, and the clinician is going through a point and click menu to try to document the history, they have the option to quickly go through a hyperlink to the most current guidelines on care of that condition. That's where we are going to begin to see the return on investment that has been made not only by AHCPR, but by a variety of others in guideline development.
A second area in which standardization is obviously important is for administrative functions. The patient record is a critical tool that drives the referral process both inside and outside of a given practice or a health care organization. Uniform records and standardized administrative data sets would greatly facilitate the referral process, permit accurate tracking of resource use, and perhaps most importantly, result in the information promptly being in a single repository where there is access both to the referring physician, as well as the specialist to whom the referral has been made.
Linking the patient record to resource use will facilitate analysis of various clinical processes, including the use of resources for specific conditions and diseases, and the resulting patient outcomes.
Standardizing the record should also permit the individual recipient of health services to become a direct source of input into their own record. I would just briefly refer to some of the work that Shelly Greenfield has done with diabetic patients, and has found that interviews with diabetics are more likely to produce accurate information about what was done during the physician visit than is an analysis of the patient record. The lessons there are the patients are a good source of information, and if we can capture information directly from the patients into the patient record, that will enhance the quality of the data.
User friendly interfaces and patient monitoring devices are already available that can readily permit patient to maintain electronic log books of their treatment, thus reducing staff workload and improving data quality.
Under controlled circumstances patients should also be able to access their own medical records to verify the accuracy of the data, and to update administrative information. One of the most interesting innovations I've seen recently is an Internet-base ENM coding tool that produces a visit note.
But in this particular product the patient has access to their own information over the Internet, and they have hyperlinked that to health information about conditions that the patient is being treated for. So if you are an asthmatic and come in for an office visit, and your physician uses this particular system, you can go back home, access your record, review the information about your care, and then hyperlink to patient information on how to best manage your own illness.
That is the kind of innovative approach that I think will be a great step forward, not only in the quality of the data, but in the utility of that record as a tool for improved patient health.
A third area in which standardization is essential is for research functions. As noted earlier, standardized patient information is extremely valuable both for individual medical practices seeking to benchmark, and for larger research and public health analyses. In addition, the standardized patient record is a potentially powerful research tool for management and administrative research purposes.
Large data sets on quality, outcomes, referrals, hospital use, variability in health care use, and cost, and the overall process of care can be derived relatively easily. These data can then be the foundation of accurate and timely provider profiling systems, and the basis for expanded evidence-based health services delivery research.
Let me very briefly identify some of the concerns that MGMA and our member organizations have about the development of computer-based patient records. Unfortunately, many current so-called electronic health care information systems are simply data repositories that may store scanned documents, and may also contain a limited amount of structured data.
To improve on these systems, it is important to limit the collection of redundant information coming into the system from multiple sources. In order to achieve this goal the data collection system must implement a standard data dictionary built upon common data sets with uniform data definitions.
Creation of such a standard data dictionary is a critical priority, and one in which we feel government must play a leading role. The introduction of universal provider, employer, payer, and patient identifier while controversial is another important priority for building a useful patient record system. MGMA, as you are probably aware, has been a member of the Work Group for Electronic Data Interchange, and the National Uniform Claim Committee boards of directors, and has been an active participant in the development of HIPAA standards, including those on individual identifiers.
Another critical area for government leadership is in standardization of interchange protocols from external databases. Unless we achieve some standardization of formats and data definitions, it will be impossible to begin to pool data from multiple sources. If we can't pool data, the utility of those data for population-based analysis is very limited.
We also believe that it is important that there be sufficient flexibility built into patient record systems to allow for diverse practices and different elements within organizations to have different operational processes for the delivery of their services. For example, it is clear that a format for a radiology department would have to utilize different care processes than those used in a laboratory, which are different still from those used by a speech therapy clinic.
So we must be able to accommodate the fact that the processes vary tremendously within departments within an organization, and the format must be flexible to accommodate that diversity.
Computer-based record systems must also integrate health information with data from external knowledge bases in order to apply rules-based logic-driven clinical decision support; as I mentioned earlier, the potential for linking with guidelines activities. Such decision support can have a significant impact on health care processes and outcomes.
One other issue that is of great concern to our members and to the public as well is security and the patient medical record. As everyone knows, there is a growing concern amongst the public regarding the protection of health information. While the issue of confidentiality did not arrive with the advent of computers and computer- based patient records, there appears to be a sense that information is more vulnerable in the electronic age.
A case can be made, however, that if properly instituted and monitored, a uniformly formatted computer-based record system can provide better protection of confidential health information than our current paper-based system. It is important to strike the proper balance between protecting the confidentiality of a patient's medical information, and not imposing undue burdens on medical practice.
It is critical that any computer-based system include standards for certifying the user's identity and authorization levels. Authentication of identity provides assurance about who is getting access to what information, and can be accomplished through a variety of techniques, including multi-level password assignment and entry, biometric means of identification such as retinal scan or voice recognition, and other means utilizing sophisticated technology as public key/private key encryption.
To assure that access by all authorized users is permitted, the record system must maximize the use of open technologies and architectures. Here too, standardization is important. At the same, however, users of the system should only be granted access to that information that they are required to access in order to perform their functions.
Reasonable safeguards must be incorporated to prevent unauthorized entities from logging into the system and retrieving data. We anticipate that many of these security issues will be addressed in the HIPAA final rule on security and electronic signature standards.
As we saw last summer with the firestorm over the HIPAA-mandated individual patient identifier, the public is extremely concerned about confidentiality of personal information, and those developing computer-based record standards must be cognizant of the fact that only must the health care industry be educated about the purposes of and need for standardization, but there is a huge public education task as well.
Cost to groups is obviously for our constituents, a significant concern. A computer-based patient record is rapidly becoming recognized as a requirement for building a truly integrated health care delivery system. However, it is clear that the high cost of developing a comprehensive system is still a major impediment for many medical groups, as well as for many large, integrated delivery systems.
Many practices continue to be wary of extending the capital for large scale patient record systems as they have seen their colleagues experience little or no return on that investment. A long, phased-in approach for implementation may provide the best avenue for these practices seeking to add a computer-based patient record system. We expect that uniform data standards leading to industry interoperability will be an excellent incentive for many practices to move forward in implementing a computer-based record.
The role of the federal government in this process is quite significant. It goes back to my earlier question about geographic or political position. Despite numerous efforts, the private sector has not yet been successful in developing or implementing comprehensive standards for computer-based records. There are simply too many innovative firms in an industry with rapidly changing technology.
While this competition and innovation has certain advanced the state-of-the-art, the lack of standards is clearly a barrier to widespread implementations of computer-based records. Under these limited circumstances government must serve as a focal point for standardization. However, great care must be exercised to insure that standardization does not stifle technologic innovation. Further, the process by which standards are developed must be broadly inclusive of the users of computer-based records, and the authors of patient records, not just technologic experts.
Further, it is critical than any proposed standard be subjected to a rigorous what I would call "field review" and pilot implementation before widespread adoptions. Despite the urgent need for standards, the consequences for error are so high that a very cautious, carefully evaluated trial period, followed by gradual implementation will be serve the public interest.
The leadership that the government has shown in other areas of HIPAA standardization should also be applied to the issue of computer-based records. This leadership should include not only the production of data exchange and data vocabulary standards, but the delineation of certification procedures and security standards as well. Finally, the government must take an active industry educational role to smooth the transition to these new standards once promulgated, and to insure widespread cooperation.
We also believe that federal funding of demonstration projects and incentives to private industry to investigate new avenues for the streamlining of administration, and grants to research entities utilizing patient data to improve practice administration are all strategies that the federal government should employ.
Let me just make seven specific recommendations that MGMA wishes to offer for consideration by the Work Group. First, obviously the federal government must show leadership in the development of standards for computer- based records content and format.
Secondly, standards to insure the functionality of computer-based patient record systems should include: (1) standard data definitions and formats; (2) standards for data storage and retrieval; (3) standards for data security; (4) standards for processing the data in a uniform manner; and (5) standards for transmission of the data.
Third, MGMA and its members are very concerned about the confidentiality of transmitted data. Physicians have a duty to their patients to insure that medical records are held in confidence, and are disclosed only in appropriate situations. We support the highest level of security that can be achieved without imposing onerous administrative burdens on practices. Further, we would contend that no administrative simplification provision should be implemented prior to the enactment of comprehensive standards for health data privacy and security.
Fourth, large, well established medical groups create and store many thousands and sometimes millions of patient records. To scan existing paper records, convert the data to a standard format, and store the finished product would be a daunting task for these groups. Accordingly, any standardization must be prospective, and a transition period to the new standards must be lengthy enough to allow groups to move to the new system without incurring undue hardships.
Fifth, we encourage the federal government to continue working with EDI standards developing organizations, committees, and advisory bodies to assist in the development and implementation of standardization.
Sixth, mechanisms must be identified to assure wide input for medical record users into the standards development process.
And seventh, the health care industry and the public at large must be fully educated regarding the need for standardization, the implementation process, and the security measures adopted for safeguarding patient information.
In concluding, successful development of computer-based patient record systems depends very much on the ability of the federal government to provide leadership, foster widespread support from all sectors of the health care industry, and create an industry-wide implementation strategy. Managing all components of the record system within a practice and ultimately across enterprises will require considerable effort from both government and industry.
The health care system is likely to go through various stages of implementation prior to achieving a comprehensive system. We are highly supportive of the development and use of national standards for patient medical records. These and other standards for the collection and transmission of electronic health data will improve the quality of health care, while at the same time lowering the cost of providing care to our communities. We are confident that national standards will ease administrative burdens, facilitate improved data interchange within the industry.
We very much appreciate the committee's interest in this topic, and thank you for the opportunity to present our views.
MR. BLAIR: Thank you, William Jessee.
Our next person to testify I believe is Al Buck from JCAHO.
DR. BUCK: Thank you, Jeff.
I'm Alfred Buck, executive vice president of the Joint Commission for performance measures and research. On behalf of our organization, the Joint Commission, we thank the committee for their interest and the opportunity to be here today and discuss this issue.
I would mention in passing that the Joint Commission currently is a national accreditor, now accrediting approximately 20,000 health care entities across the country, representative of all types and sizes of such entities, and is very much, as you know, involved in wrestling with the issues of quality assessment and standards compliance assessment, disclosure, public reporting, and so on. So all these issues and questions and options are of very central interest to us.
I have presented to you, and each of you should have a copy of a written statement which is organized as a series of responses to questions that were posed in the invitation by Mike Fitzmaurice. Rather than repeat this or be redundant to prior testimony, which by the way I would say that we certainly concur with, I would emphasize in passing, going through this document, a few points for the committee.
The first is that while I wouldn't present a definition as a definitive statement either from us or from this committee, I have attempted to frame for you a draft definition of a patient medical record. You will see that in these phrases are there are contained, major issues. Data or information documented and retrievable easily from one or more sources; hard copy or electronic; pertinent to an individual; used to describe and measure their health status and health care interventions over time and across settings.
To facilitate their health care management by qualified professionals and other personnel. We could focus and discuss any one of those phrases, and the whole raft of issues that go along with such a simple statement. I would hope, however, that over time there would be a growing consensus and ultimately, however dynamic, consolidation to a definition that I think would be of assistance to the industry.
Secondly, to pick up on some of the themes that Dr. Jessee just presented, I wanted to come at this from a little different perspective, a slightly higher level of what the uses of such a capability of patient medical record should or would be now and in the future. Not to minimize the importance of the central, absolutely irreducible requirement that such a record be useful for the management of an individual's health care, I have selected additionally for emphasis, the other uses which I don't think can be forgotten such as: organization management, a very legitimate use of that information, I think with appropriate safeguards.
Second, process improvement and assessment, also important whether for the individual's benefit, a group of beneficiaries' benefit, or a societal benefit. A very legitimate concern about oversight and accountability. And we could have a whole meeting just on that alone as you know, I'm sure. And then importantly, research and I don't want to mention that alone without education, because I believe that some of the information in that record must serve both purposes, research and education, whether it is focused for the individual or society.
I would like to pass on to page 2, and a question that I found to be particularly significant I think, which is where we all want to get to, but I felt it was important for you to hear it at least. There are two levels of needs I think. One is a higher level of need, that is a larger, perhaps a more governmental level. But if someone -- and people did, and I thank you -- asked what you see as not necessarily the low hanging fruit, but the issues that absolutely have to be resolved, I would list them as follows.
First, I think we must have confidentiality legislation. I know that that has proven already to be a singularly frustrating topic, but I just don't see us able to obtain the benefits that we either can demonstrate, or that we intuitively feel we can achieve without some kind of a confidentiality legislative umbrella within which this whole thing is put together.
Second, it's been mentioned by Dr. Jessee, but because it is so important, however controversial, it's another wicket we must get through, the so-called universal patient and provider identifier issue. Whether that is a specific number, and ID if you will, however clever and secure and sophisticated it might be, or a technique that would generate the same capability through various linkage capabilities. One or the other, we have to have a patient and provider identifier system basic to the linkage that Dr. Jessee was alluding to earlier.
Standardized chart format is one of those dry, dusty, boring things that I guess you would have to either be brilliantly insightful at a young age, or come to realization as you ease into the twilight of your career. But regardless of where we are all at in that particular spectrum, it turns out I think that chart formats and the standardization of formats, whether we are talking about the whole patient record, or whether we are talking about forms, specific needs addressed within the patient record, a standardized chart format is another one of those things that we have to simply address to get consensus on, move out, and work with and refine. It's essential. What are the components we are going to look at? How do we achieve it and so on are all related issues.
The standardization of data element fields and definitions is something that I have personally -- let's say a windmill that I have tilted for the recent years. It's very difficult, no question. Core data sets, refinements in the front sheet, and my old friend the "universal" billing form all need constant update and integration into this effort.
Then at this relatively high level, aggregation rules, coding and coding audit methodologies, and methodologies to accurately follow and aggregate reimbursements as opposed to billings, which fit I think very nicely within the context of the remarks already presented.
Now at a different level, perhaps at a more granular level, there are a series of data quality issues which I would like to list, only to identify that I think they too must be resolved. These are often either unrealized, and sad to say unappreciated by many of our colleagues. It has been my experience that need something of partly an explanation, and also partly a hard sell. I apologize to you people, because I know I'm preaching to the choir in this regard, but it needs to be emphasized.
Under data quality I'm going to list a few things, the issue of data element definitions, data field determinations, and technical standards, which has been emphasized in the previous statement. Such other things include the categories and domains of measurement, reliability, validity, interpretability, collection and calculation methods, and the whole business of proprietary versus non-proprietary status. And I would vote for non-proprietary in practically every case as we try to get through these issues.
At a slightly higher level, when we talk of aggregating single patient data elements, we are wrestling now with defining categories and episodes of care coding and coding accuracy. Then in a slightly higher aggregate the issue of core measurement sets, risk adjustment methodologies, perhaps at this level the single most frustrating but important issue facing us nationally. And you emphasized in the past audit methodology, aggregation rules, cohort generation, edit standards, management of missing data, and we could go on.
Now I want to leave that data quality issue point with a comment that comes from our work on the HL7 special interest group, of which I'm a member on quality, accountability, integrity that we don't forget narrative entries. I have had a comment here that I would like to repeat to you in this testimony, that at the individual patient record level, non-measured elements, that is narrative entries, need to be addressed through standard vocabulary and terms, standard entry categorization, format rules, and so on and so on.
Applicable standards and criteria for accountabilities, the identification and validation of source, completeness, time of entry, and integrity, meaning honesty, completeness, wholeness, and permanence is non-alterability must apply to data elements, as well as to the narrative. Again, I apologize, because I know I'm preaching to the choir, but I will emphasize, and I'm sure you are aware, that many of our colleagues have not thought through these issues, and do need to be appraised of their importance.
Now given all that, what is the role of the government? I think it's the vital question and each of us will probably address this in different ways. But clearly there has to be a role. There is a role in partnership with the private sector. I have listed two. One is a facilitator, that is in identifying significant activities and consensus development, and identifying and evaluating expert opinion.
But then as a consolidator as well. A consolidator, and even occasionally a driver through regulation. And I think in this context I would leave it that I concur with the recommendations previously made, and I think they strongly supported a combination of roles between the government and the private sectors.
Lastly, I want to emphasize a couple of points on where I think this all needs to go, and offer some comments perhaps that may seen a little at a general level, not quite on the data quality issue, but I think certainly derivatives of those issues to keep a sense of direction.
The patient medical record and the data quality issues that we are trying to resolve must be supportive of addressing and communicating aspects of quality and value at the patient or various aggregate levels related to health care. Quality, costs, and improvement are the key factors that define value. And data and data-based products must be capable of depicting these aspects in meaningful ways, and in supporting analysis of their relationship and impact.
It's true I agree that that's perhaps an expansive view of what a patient medical record is all about or should be all about, but I hope that expansive view could be captured and maintained as a specific goal. Until data and data management infrastructure are sufficiently refined to allow the depiction of probabilities related to clinical outcomes resulting from various interventions, including preventive measures -- and you recall that probabilities are the core of the well accepted definition of health care quality provided by the IOM a decade ago -- an interim approach should be defined that would be workable, with near-term capabilities, but that would also build to address fully the IOM definition in the future.
Key components of an interim proxy definition for health care quality would involve in my opinion, compliance with science as the evidence-based piece, consistency with wisdom-based processes and outcomes and goals and measures. I think that's at least partly the piece that Dr. Jessee was referring to with the interesting work of David Eddy.
Customer -- and I have to tell you, I have mixed feelings about the word 'customer' but nevertheless -- satisfaction and perception. And selected aspects of oversight and accountability, including the performance evaluation of accreditation. And in passing, because it's not generally always understood, I would tell you that the accreditation process of assessment most significantly is an education-based change engine, as opposed to a sanction-based change engine, the two processes both being essential. I believe in the oversight and accountability framework for the country.
How do we know we achieved success? If all this really comes together, we realize the benefit that I think intuitively we think should be realized. I'll leave you with this summary statement on that. The patient medical record initiative efforts are fundamental to the achievement of more substantial database resolution for health care challenges and refinements of policy.
These include things like: definition of health care benefits; the structure of access; the allocation of resources; the alignment of research and education; priorities for efforts to evolve these policies at various levels, not the least of which is the federal level.
As we can understand and see data and derived information used in such development and refinement, we can be reassured, and I think the public can be reassured that we are making real progress toward achieving the expectations we currently hold for our expanding technology in the near-term and in the future.
I would offer that those are rather hefty, rather lofty goals, but I hope that in this process we could begin to start to address them. We certainly thank you for the opportunity to be here today.
MR. BLAIR: Thank you, Dr. Buck.
Our next presenter is Stephen Lamb from NCQA.
Agenda Item: Health Data Quality - Stephen Lamb, NCQA
MR. LAMB: Good morning. As I said earlier, I am Steve Lamb, and I'm the vice president of public policy with the National Committee for Quality Assurance. One of the dangers of testifying after Drs. Jessee and Buck is that you are tempted simply to say I concur, and move quickly to the questions and discussion. So I'm going to abbreviate my remarks somewhat, because much of what has already been said is what I had planned to say.
And also because both in their current positions and in past positions, NCQA has worked closely with both of the individuals on these issues. I would just this opportunity to call your attention to the work on the confidentiality program that Joint Commission and NCQA collaborated on, and I'll make sure the report gets to this group.
By way of echoing what Dr. Buck said, I think we are entering a very, very sensitive period in which issues around confidentiality are ones that could, in my opinion, torpedo this whole effort if they are not properly tended to. Just by way of brief departure let me say we have had some brief experience with that at NCQA, and would be happy to share with you some of our experiences as we have gone about evaluating the performance of managed care health plans.
What we found is that the media and the public are often only interested in the worst performers. Our communications department has begun an outreach effort to members of the media to try to discuss with them the importance of reporting on a range of performance, but not simply zeroing in on those poorly performing health plans.
I think that is a lesson that might be helpful for the folks around the table, because in educating the public, the media is going to be absolutely critical. And I would be happy to offer you some of the information on the lessons learned on our experience as it relates to the public reporting of health plan performance information.
But let me return to the work of this group, and first let me offer both our support and our encouragement for this group. When NCQA began the journey in 1979 with evaluating the performance of health plans, it was with a very rudimentary accreditation program that looked simply at the processes in place at health plans. Today we have a very robust accreditation program that is being merged with our performance measurement efforts.
What that means on a practical level is that when go on site to a health plan now, we are looking not only to see if the structure and processes are in place to deliver good care, but also is the health plan actually delivering care? Now the way we do that is through a set of performance measures called HEDIS, or the Health Plan Employer Data and Information Set.
In addition to our accreditation program, which is required by most of the Fortune 100 companies, the nation's public and private purchasers also require that performance information from their health plans be reported according to HEDIS specifications.
Now that work has gone along very well, but two years ago we realized that we were beginning to bump against some of the technical limitations in that some of the things that we wanted to measure with regards to health plan performance were simply not an option because of a limited information system infrastructure.
That led us to produce a document called, "The Information Systems Roadmap," which was part of HEDIS 3.0, and was released now a couple of years ago. It was in my opinion, really a call to arms for the managed care industry, and putting them on notice as to what kinds of IS systems would be required for the performance measurement expectations of the future.
It also vented some of our frustration, and spoke about those things that we wanted to measure, but could not because of the limitations in the information systems at the health plan level, and in other places in the health care system.
We look now at the work and the framework that was established by HIPAA as something that is extremely exciting, not only for what it could do for the country, but more selfishly, what it could do for the work of our organization, and the ability to increase the information that we would like to provide to purchasers, consumers, regulators, and others.
Let me give you just a couple of brief examples of the cost that we have seen, and the limitations that are currently in place, and the cost in the financial sense, and then the limitations in terms of the scientific, and in terms from a health policy standpoint.
Currently, when we go on site to a health plan, or when an auditor that has been certified by NCQA goes on site to a health plan to look at that health plan's HEDIS performance measures, what we find is that although health plans could gather this information from their administrative data systems, and certainly allowed under the HEDIS specifications, most of them have gone to what we call hybrid methodology, which is pulling both from the administrative data set, and more expensively, actually going out and doing an actual chart review.
Now what that means practically is that the health plan has to hire a nurse or a medical record abstractor to go out -- and in most cases these are network health plans -- go out to individual physician sites and pull the paper record.
Now what concerns is us one, the cost that is involved in doing this. So when we think about the cost that is imposed because we don't have a good computerized patient record system, and we don't have good information systems at the health plan level, that's the first thing. The cost of going out and actually pulling these charts because we can't pull the information from the administrative data set.
I would say just by way of branching off briefly there is another huge issue here. That is the problem of miscoding. We have recently pulled back a measure that was going to look at ACE inhibitor use, because we had such a hard time identifying accurately who had actually suffered from congestive heart failure. When we went to the coding and CPT codes, we found errors. We found an accuracy rate of about 50 percent, which was unacceptable for our purposes.
So when we think about the cost, the first thing to think about is this pulling of medical records to facilitate the performance measurements process.
On another side is the cost in terms of health policy. For an average plan right now when we look at them, and ask them to calculate their mammography rate, which is a measure in HEDIS, most plans have about 7,000 women in that age group. Now if a plan uses an administrative data set, or if they actually go out and pull medical records from a randomly selected group, which would usually be about 400 medical records, they can come up with a rate.
But if we have two plans, one with a rate of 72 percent, and one with a rate of 70 percent, the margin of error is larger than the difference between the performance of those two plans. So what we find is that we are unable to tell employers that are looking to select a health plan, that in fact there is a meaningful difference.
Now some might argue that the difference between 72 percent and 70 percent -- an employer or purchaser shouldn't base their decision on a health plan, but understand that this is one of 72 measures that add up. If we had an accurate administrative data system based on computerized patient medical records, those are differences that would be real. And so we could begin to make more of these granular distinctions in the performance between health plans. We can't do that right now, because we are pulling random samples of medical records for so many of these measures. That's the first story.
The second story that I would like to tell is about the measures that never made it into HEDIS. We do have measures that are in HEDIS, and as I explained to you, it could be very costly to gather this information. Once we do gather the information, there are issues about the accuracy of that information, and our ability to distinguish between the performance to health plans, both of which would be aided greatly by the advent of improved information systems, and more specifically a computerized patient medical record.
The second question is what measures never made it into HEDIS? There is a Committee on Performance Measurements that decides which measures make it into HEDIS. We are very fortunate to have individuals like Dr. Eddy and Dr. Mark Chassin and others to sit around in this group and decide which measures are going to be included in HEDIS.
Now one of the things that we learned very early on about the measures that made it into HEDIS was that they were in many ways driving part of the nation's public health agenda. We would hear back from the Public Health Service and from others that they would go to health plans and say we would like you to focus on this, and the response would be is it a measure in HEDIS?
It's sad but true that what gets measures gets done, and what gets measured gets improved upon. For that reason, we have reached out to the public health community and said to them, we now know that what we measure in HEDIS oftentimes dictates where quality improvement resources are spent at the health plan level. We want to make sure that this is done in line with public health priorities.
However, I can tell you that there are some measures that we would have liked to have had included in HEDIS and ACE inhibitor use is one of those, that were not able to become part of the measurement set, although they were in the testing phase, because of problems with the information gathering and the information systems.
So let me conclude by saying that there are costs that are currently being imposed not only health plans, but on the work that we do. What that means is that we don't have as robust a set of performance measures, and we don't have as robust a set of information that we can provide to purchasers, to consumers, to providers, to regulators, and to others to allow them to assess the quality of health care and the service of that health care that is being provided at the health plan level.
I should say that although our experience to date has been with the health plan level, these are problems that we hear from health plans, and that they are experiencing at the medical group level as well. These are not unique to the health plans, and in many cases, particularly on the West Coast, the health plan exists as an administrative over-structure, and it farms out much of the quality improvement and quality measurement work down to the medical group level.
The problems with getting good information are present at all levels in the system. And again, let me offer our encouragement and support for this Work Group, because we believe it has a tremendous potential to improve our ability to assist in providing purchasers, consumers, and providers with better information about the quality of the health care that they are receiving.
MR. BLAIR: Thank you, Steve Lamb.
First, let me open it up to committee members and staff for questions. And Dr. Cohn, if you could help me identify the individuals.
DR. COHN: Maybe I'll lead off with a question. I also wanted to welcome Kathleen Fyffe, a member of the Work Group.
I just wanted to check with our panelists. I think I heard a number of the panelists indicating that confidentiality needed to be resolved before any of this could really happen. I'm actually sort of thinking in my own mind that there is a letter that we are fashioning to the secretary, just giving her an update.
I just wanted to make sure that I heard it very clearly from all of you that we probably need to communicate to her that this confidentiality issue needs to be resolved before we can go ahead. Is that all of your consensuses and agreements?
DR. BUCK: I think I would speak for all of us, but certainly for our organization we feel very strongly about this. I would really appreciate it, Steve, if you could provide, if not maybe our two organizations could take it upon ourselves to provide our talk to Jackie Adler, and get a list of attendees, names, addresses, et cetera of those who were unable to be here today, and see to it that you get a copy of this publication.
I don't want to present it as kind of tooting our horn, but I think that the collaboration here was achieved between NCQA and the Joint Commission in this regard to bring an incredibly diverse and very, very experienced, insightful collection of views to this issue is -- I think it's a true statement, it was very good material.
The bottom line of this is that while it doesn't seek, nor could it give you an answer let's say in a paragraph at the end, I think there is preponderance of opinion that is very diverse, that the confidentiality issue, because of many, many other issues, the creation of state law, the emergence of the change rate of technology, the applications of data that we have all touched on here today is so complex, and yet pressingly urgent to address, that it must be a legislative initiative.
We'll commit -- is that okay? -- that we will provide the members of this committee with a copy. It's about a half inch or so thick. It's not overwhelming, and a wonderful reference. I think the bottom line answer is yes, and more so.
DR. JESSEE: I would just add to that though that this ain't going to be an easy issue. Confidentiality is part of it. Security is the other part. And great care has to be exercised in crafting any legislative or regulatory approach to confidentiality, because I think there have been some instructive experiences at the state level when we get more confidentiality than we might have wanted, without adequate input from the public about how much confidentiality do they really desire of their medical information.
And I use the state of Maine as a specific example. With great support from the health care community in Maine, Maine enacted comprehensive confidentiality legislation a year and a half or so ago. And after six months had to reconvene the legislature in special session to repeal the law, because of the public outcry over the fact that real simple things like a wife calling the physician's office to confirm her husband's appointment the next day, and being told by the receptionist, I'm sorry, I'm forbidden by state law from disclosing that information to you.
Florists seeking to deliver flowers to individuals in the hospital, being asked what room Mrs. Jones was in, and being told, I'm sorry, state law forbids me to disclose that information to you.
So sometimes I think there is a disconnect between what we in the health care field believe our patients want, with what patients in fact do want on this particular issue. Like many in public policy, it's going to be one that is going to require a very fine line to be walked to try to craft something to provide adequate protection for privacy without creating massive disconnects and massive consternation in the field.
MR. LAMB: I would just add one thing, and that is I think one of the biggest threats that we see that could come from this is the inability to close the quality improvement circle for our auditing work and for accreditation work. The ability to identify individual member records can be important, but it's probably something we can work around.
But if individuals can, as some people have proposed, opt out of the system, we may be forever handcuffing the plans and shackling them to the point where they can't identify members and reach out to them and deliver this care when it's necessary. If that happens, then it really calls into question what is the role of a health plan.
Our idea all along has been that a delivery system should be seeking those individuals, and identifying them, and delivering the care, oftentimes when they may or may not need it. The second most commonly cited reason for a woman not receiving a mammogram is that her physician didn't mention the need for one to her.
We know that identifying these individuals and closing that quality improvement loop, and actually performing the intervention can have terrific results. That's what our accreditation program is all about. But if forbid the health plans from doing that kind of activity, I think the result could be far, far worse than whatever cure might come about in any sort of privacy legislation.
DR. BUCK: Could I just make one additional thing, because what Steve just said jogs my mind, and what Bill mentioned. I guess I would put my oar in the water, yes, this should be pursued. It should be done soon. It's very tough, just as Bill emphasized.
But there is another emerging 900 pound gorilla out there that needs to be considered as well, not to abandon or slow down the commitment to resolving the confidentiality effort, but to shape it in such a way that it can anticipate a more or less evolutionary future.
I'm citing the increasing concern about errors and adverse events, and that whole spectrum. And yes, the minute you start this you are on the slippery slope to dare I say it -- I guess I can in this room -- tort reform. But the issue here of a rational, proven, scientifically sound way of reducing, preventing, eliminating errors within a context that has been well established in the engineering fields for I want to say decades, and moving that into medicine is also on the scope now.
The confidentiality piece, while I wouldn't present that I think it will solve in any way the raft of issues that go along with having a more rational error management capability in the country, and all the legal implications that may flow from that, nevertheless that's a piece, I think, looking forward, that also needs to be considered in the context of confidentiality legislation.
MR. BLAIR: Do we have some other panelists that have questions?
DR. COHN: I think Mike has a question.
DR. FITZMAURICE: I would like to follow-up on Al Buck's question and divert it to Bill Jessee, asking that as we face problems with errors in medicine, adverse events, the large amount of information that caregivers have to carry in their heads to deliver care, certainly accurate data and some computer-based patient systems can help out in that regard, but there are barriers to the use of this information to reduce potential errors.
Some of them are wrapped up in tort reform. Some of them may be wrapped up in as you mentioned earlier, the interfacing. Do you have some thoughts on what the barriers are from a physician standpoint of being faced with potential errors in medicine and ways to address them? I would refer to an airline pilot who has performance reviews. Errors have been made and corrected, but they don't wind up in the court as much as physicians do, so that's bothering me. If that's a good model, is there something wrong when we apply it to medicine?
DR. JESSEE: I think it's like many issues, it's one that has got a lot of threads. And clearly we have, as a nation, resorted to the courts as a means of trying to redress errors in health care. And the downside to that has been there is a backlash in efforts to not wind up in court if you can avoid it, which is probably conducive to trying to understand why errors occur, and what are the systems issues that lead to those errors.
The airline is such a stark contrast, because FAA has bent over backwards to try to produce a risk-free environment for pilots to report their own errors. If you report, you are protected. If you do not report, you are at risk. You can have your license lifted. So they have taken a totally different approach in the airline industry to try to encourage reporting of error when error occurs, because that's the only way you can find out what's the frequency, and what are the systems factors that lead to those errors.
We are in a situation today in medical practice that first of all, most of our knowledge about errors was hospital-based. Almost all the research that has been done on errors is on errors occurring in hospitals, and yet over the last 15 years a huge amount of care of increasing complexity, and therefore increasing risk has moved out of a hospital setting, which is probably the most highly structured.
As bad as the data quality is in the hospital, it's a heck of a lot better than it is in most office-based practices. We are seeing more and more high risk procedures moving from the inpatient setting to the outpatient setting. That is provoking state level concern in a number of areas, and we have already seen California, I believe New Jersey and Florida all have passed state level legislation to put more regulation on the performance of procedures in an office-based practice out of concern over the fact that those are largely unregulated settings.
And again, the purpose there is to try to reduce errors. But we really have almost no research base in the ambulatory setting about what kinds of errors occur, what is their frequency? Our organization has just recently -- I floated this before my board in July, and they have said you are making sense. Bring us back a final plan for approval in October.
Don't hold me to this yet until I take it to my board in October, but we are planning to launch a major initiative on patient safety in office-based practice that will include a research component, an educational component, a communications component, and an advocacy component, because no one is doing that right now. We feel that it's an important issue to be faced and addressed.
I don't mean to use a whipping boy here, but the growth of managed care has produced new pressures on office-based practice for throughput. When there is more pressure for throughput, there is an increased likelihood of error. There is an increased likelihood of systems breaking down, and in keeping with the concerns of this group, there is a reduced likelihood of data capture, so long as that data capture is not integrated into the care process, because it becomes one more "administrative burden" on a practice where time has become at a premium, not just for the physician, but for everyone in that practice.
I think they've got the confluence of a number of events that are going to probably increase the frequency of errors in office-based practice, that are going to produce more and more cynicism amongst professionals about if you want me to do more in less time, and you also want me to reduce the likelihood of error and capture data. There is only so much I can do.
The challenge is going to be in my mind, to try to find ways of using technology that will alleviate some of that burden, improve the ability to manage throughput, provide the computer-based reminders and so on. Clem McDonald is missing from that corner, but how many years has it been since he demonstrated what you can do with simple computer reminders to improve care of diabetics?
And Steve and his group are still trying to increase the frequency of hemoglobin A1C determinations in populations. Those are simple kinds of things if you have used the technology appropriately. I think all the experience is somewhat interrelated.
There is a huge task that I think we have before us to try to increase the understanding of how these threads relate, and try to produce some recognition that from a public policy perspective, if we really are serious about improving quality, reducing risk of error, reducing costs, we have to come to grips with some of the issues. You can't sue everybody every time you turn around and accomplish that objective.
DR. FITZMAURICE: I just wanted to follow-up with a comment. What you described is really what I say is the state of health care in the United States is errors are being made, often through faults of systems rather than faults of individuals. So much is loaded on them, and so much pressure is put on them, there is no environment where it can be studied well without the threat of legal activity.
I'm not sure what an answer would be. I'm glad to hear that you are coming out with an initiative. I can't imagine two better organizations than JCAHO and NCQA to advocate such an initiative whereby we might gather more information about what can work better to bring information to reduce errors in medicine, and help people do the right thing.
MR. BLAIR: The question that I have would be for Al Buck and William Jessee. Steve indicated that the lack of standards that enable us to have more confidence and specificity in quality has impeded HEDIS's ability to be able to measure performance. I think intuitively we understand that that lack of data has impeded us in research, in being able to have more meaningful measures of outcomes, a whole host of areas.
So I would like to ask Al Buck, in your capacity the JCAHO, where you are accrediting and to some degree measuring outcomes in performance for health care organizations, and William Jessee, where your members wind up providing clinical care, if there is maybe one example that comes to your mind of you are frustrated or you have been frustrated.
Let me get specific. Al Buck, something where you have wanted to measure outcomes or performance, and have not been able to because of lack of quality of data standards for data quality. Let me ask you first, Al.
DR. BUCK: Great question, Jeff, and probably will open up several Pandora's boxes here, but let me address as best I can. Speaking for myself, but I think it's pretty close to the mark, we have had I think a fair amount of frustration in attempting first to define measures and get the data, and make an analysis in as constructive a fashion as possible, with an educational thrust that does in fact get to the root of either preventing or minimizing the incidence of what we are calling sentinel events.
These are the errors. These are the safety issues. These are the scandalous stories that you read about in the newspaper.
The bottom line of that to me is monumentally a consensus piece, less a mechanistic piece, because as we heard in this wonderful meeting some of us were at just a few days ago, one of the great thinkers that I have had the privilege to meet said nothing needs to be discovered here. Everything exists technically that we need to have to use to solve these problems, and I absolutely agree with that. I absolutely believe it. I know it to be true.
What we are lacking is the consensus, and we're lacking experience, we're lacking data. Let me put this in very concrete terms, an example. We have reviewed the tip-tip-tip of a huge iceberg of sentinel events, 512 specific sentinel events in the country. We have reviewed the exhaustive analyses done by organizations, health care entities that have in turn tried to create a so-called root cause analysis right out of engineering textbooks. This is nothing new here folks.
What are the two themes that come out of these analyses? Staff training, and staffing levels. Wow. Think about it. Think about the implications of any health care entity trying to extract from patient medical records, from administrative data, from chart abstractions, from other data sources, national benchmarking efforts, et cetera, et cetera, data that constructive assists either an organization, a group of organizations, an oversight entity, a policy body, a government entity, you pick it, but assist all those organization in producing some kind of a rational, economically sound, constructive maintenance framework that supports and nourishes constructive assignments, economic commitments to achieve FTE appropriateness, and then given all that, a training success achievement on an ongoing fashion.
That is a long winded answer to your question, Jeff, but I mention it, because I think it brings together the issues of safety concerns, the issues of integrating data sources, the administrative, the clinical, the course, the resources, and so on. In a way, I think it does complement what has been said here to give you a kind of state-of-the-art statement of where we are, however lamentable we may see it.
MR. BLAIR: Let me pass this same question on William Jessee. Where have your constituents struggled because of the lack of standards to define quality in health care? Not just simply having good data that they have confidence in?
DR. JESSEE: I think the best was to answer is it where have they not, because the list of where they have is quite long, and the list of where they have not is quite short. Just to give you some very specific examples, in one of my previously lives I sat in Al's chair. One of the activities I was involved with was pilot testing performance indicators of trauma care in hospitals.
We got a group of trauma surgeons together, and they all agreed that if you had blunt trauma to the abdomen, the amount of time that it took from the time you hit the emergency department until the first incision was made was a very important measure of organization performance in diagnosing blunt trauma.
We started then by trying to see if we could identify patients who were demonstrated to have a perforated discus, a lacerated liver, spleen, whatever. So you started with a population of people that you knew by diagnosis these were the ones that were found. You knew that those individuals had a significant abdominal trauma.
You then tried to back up and find out what was the length of time they hit the emergency department until the time of first incision. Guess what? It was thrown out because it proved to be an impossible measure to evaluate reliably even within an institution, much less across institutions simply because times are so inaccurately recorded in patient records.
Even in high visibility areas like the emergency department, where time is of the essence, we found just in the recording of time, the capture of time of events was so variable that it was simply an unreliable measure.
Another issue that is much more pertinent to my current membership, drugs from time to time are recalled. Some of you may recall seldane, for example, which is widely used as an allergy medication and then began to be demonstrated as being associated with hepatic problems. Imagine how you feel if you are physician, an allergist, where all your patients were on seldane, or a general internist or a family physician. How do I even identify the patients in my practice for whom I have prescribed seldane?
If I am lucky, if I've got a "sophisticated" system, I can go into my billing data and at least identify the patients in whom I put down some sort of a diagnosis related to allergy. I should be able to identify that through my billing system. Now I have to go pull all of those charts and try to left through them to find out did I write a seldane prescription. Hopefully, I wrote it down, so at least I can inform them.
It's further complicated by the fact that I don't know who else may have seen my patient, and may have their own record that they may have prescribed seldane, and I would never even know about it. The list goes on and on.
Retinal exams and diabetes mellitus, when I was at the AMA before I joined MGMA, one of the areas we were attempting to develop was performance measures for evaluating individual physicians through the American Medical Accreditation Program, AMAP. Retinal exams and diabetes mellitus, everyone will agree that patients with diabetes should have retinal exams. But trying to find out if they did, when the retinal exam may well have been performed by an ophthalmologist, not the endocrinologist or the internist who is responsible for the primary management of the patient's diabetes care; different office, no linkage of the record.
There may be an indication in the record that the patient was referred for retinal exam, but there is no evidence of whether or not they really had that retinal exam. I think every one of us who has had some clinical experience can go on and on and on with a list of things that are really quite important for understanding what we are doing, and how it impacts on patients. But you can't measure it, because you can't get the data.
MR. BLAIR: Dr. Cohn, do you have a question.
DR. COHN: Actually, first of all, I want to thank the panel, because I think it's been a fascinating set of discussions. Dr. Jessee, especially some of your comments, both verbal and written I find fascinating. I actually wanted to follow-up with one question.
We talk a lot about business cases. And recognizing that standard exist out there in the ozone, but trying to develop a business case about why we should be promoting or recommending things becomes critical, because there is a cost usually associated with things.
I saw that you had mentioned cost to groups. Your testimony tended to indicate that you thought that the issue of greater uniformity of standards, while maybe not the magic bullet, might be a major incentive or people go to road where -- once again, my view is I guess I'm trying to avoid adverse events. I practice medicine still sometimes, and I'm very committed to measuring quality, but I really want to reduce adverse events. I think the public really wants to reduce any possible adverse events. So you do you think standards fit into this as being a piece to help us with that?
DR. JESSEE: My comment about standardization increasing the likelihood of adoption was based upon the value equation. What is happening now is that in many practices, if you get a couple of good techies who close to the geek end of the spectrum, they may be able to persuade their colleagues, you know, it's silly for us to keep doing this stuff on paper, and be able to get them to do it because it's cutting edge stuff.
The business case analysis of the return on investment for our practice is pretty shaky. That applies perhaps more in ambulatory care than it does in inpatient care. At least there is a larger base of experience in inpatient care. At places like Intermountain, which have made huge investments in information systems have actually now had enough experience to begin to quantify the return on investment in terms of reduced use of antibiotics is just one example, and therefore reduced patient risk and patient injury.
It's a long tale on trying to determine what expense have you avoided by virtue of having identified the potential for an adverse drug reaction before the drug was administered? You could quantify what the cost is if you had the drug reaction, but quantifying the savings is not that easy.
There is much more of that experience in inpatient care than there is in ambulatory care. So what you've got right now is if you look at the bell shaped curve with innovation, you've got a few early adopters way out on a limb, who are in some cases convincing their colleagues, and they are also convincing the practice administrator to make an investment that is probably not going to have, at least over the short-term, a return for them.
If the value were increased by standardization and saying not only would we be able to, by making this investment, have better records, we will be able to pool data. We'll be able to find out if we are doing as well as we think we are compared to the other internist group down the road or against national norms.
If you can add some value back, creating opportunities for pooling of data, enhanced comparative information, I think you are going to see more adoption. Plus, it's the typical barrier that every one of us faces every time you think about buying a piece of computer equipment. If I buy it today, Bill Gates is going to announce tomorrow that the next generation has come along, so I'll just wait one more day.
I think there is a huge amount of that going on right now. I don't want to make a capital commitment to a system, whether it's the software or the hardware, because I know that as soon as I do -- this really hits the business manager a lot in the practice -- as soon as I convince my physicians to spend $35,000 on this wonderful electronic record system, I know I'm going to go to a meeting the next week and there is going to be one that does four times as many things, much more functionality, and is $15,000 less, and I'm going to look like a real fool.
I think there is a lot of that going on. I believe standardization won't be the magic bullet, but will contribute to eliminating some of that.
DR. FITZMAURICE: Simon, could I follow-up on your question to any member of the panel who wants to respond. Bill Jessee mentioned the work of Clem McDonald and also Bill Churney(?) has worked closely with Clem McDonald. They published an article in JAMA in 1993 showing that if physicians used computers to do orders, drug orders, test orders, that information got around faster, and patients were discharged like 0.89 day reduction in stay, saved $550 worth of costs.
As you think about that, why doesn't everybody do it, if it can save a hospital maybe $3 million a year? The answer seems to be in part that the business issues that Simon is raising, if a physician spends the extra time -- in that study it was 5.5 minutes per patient, in a ten hour shift to do all this data, he sells less product. The hospital he or she works for sells less product. And the gainer is the health plan or the insurance company who had diminished bills coming up.
And yet it doesn't get back to compensated time of the people who put the information into the system that makes it work more efficiently. So I think that along with standardization, there must be some consideration of incentives rewarding the people who make the system work better, in addition to addressing some of their legal problems.
That's not a question, but I guess it's a statement that I would look for qualifications, modifications, or you've got it wrong, Mike.
DR. BUCK: I'll speak to that. I think you've got it exactly right. And in paraphrase I would say that the issue here, and there is early evidence, very early evidence, very fragmentary evidence from organizations like Intermountain and a few others that quality may pay in the sense that there can be some closing of the loop to administrative personnel, physicians, staff, public, et cetera. That an investment of infrastructure has a pay off that is of value in a variety of ways to the users.
I would emphasize one thing, however, and it is additional to what Bill just said, and that is that in creating the value for the user, whether that is the professional health care provider, whether that is the organizational provider, whether that is the patient or consumer or public, there has to be a meaningful translation if you will, of whatever the data product is.
And that's still another whole area that has really only I guess now is beginning to get its fair share of attention. Maybe it's the same data, and it is presented in different graphic formats, in different language, and so on for different audiences, becoming kind of a science in itself. I think the power of all this to persuade professional providers has been established. The costs are I think, causing every bit as a realistic an obstacle as Bill has said.
There is one other thought I want to leave on this subject with you, and that is that Bill, I don't think he knew it, but he touched on one of my hot button issues. This is the value equation. We know the value equation I think as quality over cost equals value. I want to leave you with a thought that that is inherently incorrect. I say that just as a personal bias, and this is looking downstream.
Because I believe that there is a piece here that we need to talk more about, and that is improvement, and somehow weave that into what is the traditional equation of value. I say that, because down the road, if we ever live that long to see it, when we have some more infrastructure, we have better consensus, we have all these standards in place, or at least some of them will build us to a better functional reality.
What is going to make it for us, and the providers, and the public in situations where everything in fact does go well, is done to the best knowledge, is done to the best evidence, and yet you have a tragic outcome. And we have all seen that. What makes that work at the bedside is the understanding by the patient, by the family, by the community, by the organization, by the staff that the organization has proven its ability to achieve improvement in its commitment.
Think about the tragedies that most of you have seen at the bedside in oncology centers or in hospice areas or in other health care settings, where despite the best of everything, things did not go well. So I just leave you with that. The improvement here, that experience marching over time that has demonstrated, known, communicated, educated the providers, administrators, patients, the public, et cetera is a very important piece of the value equation. I don't think we talk enough about it.
DR. FITZMAURICE: I see people looking at their watches. I have looked at my watch. It may be a little past your time. It's almost 10:45 a.m. I would ask one short question of Steve Lamb and NCQA. I read the Roadmap. I helped review the Roadmap. I thought it was good. NCQA followed up with an education effort of putting out information about these measures would strongly benefit from information system improvements.
Has NCQA looked at those measures and said we need to get a word to some standards developing organizations to bring uniformity to these particular measures, because they would add value to our product? Are these on the agenda at NCQA?
MR. LAMB: There are conversations that are taking place with the AMA with regard to CPT codes. There is also an effort, and we are receiving some pressure from the employer community I should say. Certain employers have said you should at some point set a date and say we are no longer going to accept hybrid submissions for certain measures. And if you try to submit anything other than an administratively drawn measure, you will be listed as non-reporting in an effort to force the investment at the health plan level.
I should say it's a very controversial proposal, and one that we certainly haven't made any decisions on, because I think there is a more constructive way to get at it. But I think NCQA could probably improve our work in terms of reaching out to those communities. Right now we focus on the coding; improvements we would like to see in coding to allow us to do better measurements, but we could also do a better job as it relates to the actual data standards.
Let me say one thing about what Dr. Buck and Dr. Jessee have said. In a previous life Dr. Jessee has a very compelling anecdote that he would tell about a physician in private practice that received notifications from two of the 14 health plans that he contracted with telling him that on the one hand his c-section rate was too high according to one health plan, and another health plan was notifying him that his cesarian section rate was too low.
Now both of them were only looking at the enrolled population that was being served by that doctor. It's very frustrating. I think it's also illustrative and tells us a little bit about why it is that some providers are reluctant to make the investment in office-based system when they are contracting with at least in Portland, Oregon as I understand it, at least 11 different health plans.
Even with a very large health plan like Pacific here in California, I don't think there is any one medical group there that has more than 10 percent of their overall membership in one medical group. So it's a tough business case to make, particularly when you've got to figure that you've got to try to coordinate that investment with the information systems of 12 or 14 different health plans that you might contract with.
So I wanted to draw your attention to yet another challenge that is faced. I think it's certainly understandable on the part of the provider community, before they make a big investment, they've got a lot of folks, a lot of payers they need to coordinate with.
DR. JESSEE: If I may very briefly follow on to that, because I think it relates back to Mike's question about incentives. While I certainly am a believer in pluralism, I think we have to recognize what the cost of pluralism has been in terms of really trying to build incentives into the system.
The average physician in the United States is in 11.9 different insurance products. That's an average. There is a guy in Wyoming, who is in one, who we know is pulling the national average way down. In some urban areas with high managed care penetration it's not unusual to have 25 or 30 different insurance plans. And for certainly highly specialized practices like neurosurgery for example, the world record holder that I have met is a neurosurgeon from Southern California who told me he participates in 92 different insurance products. No two them pay him the same way.
I think that's an important part of the incentive. If you've got one plan that is incentivizing you to do more, and another plan that is incentivizing you to do less, it produces a great amount of schizophrenia. And trying to figure out how to build systems to incentivize people to collect data, you have to go back and ask the question, what plan.
So it's going to be very, very difficult to address some of these issues without really understanding the financial drivers. You would like to be able to appeal strictly to professional motives for excellence, but the reality is that when that conflicts with where the money come from and goes to, it is very difficult to appeal to the altruistic motive.
MR. BLAIR: I need to exercise my role here to keep us a little bit on schedule. So I think we need to take a break at this point, a 15 minute break, and then our next panel will reconvene here.
Thank you all for your excellent testimony.
[Brief recess.]
MR. BLAIR: If we can convene here. This will be our second panel that is focused on data quality. And we have is it William Jenich?
MR. JENICH: Herman Jenich.
MR. BLAIR: I'm sorry, Herman Jenich. And Dr. Stanley Griffith. Herman Jenich, may I ask you to introduce yourself, and maybe you could go ahead and start on your testimony.
Just so we can open it up a little, I know that we have a started a little later than we thought we would. Intuitively, Michael, let me check with you -- I didn't have a chance to do that in the break. I'm assuming that we do have the freedom to go until twelve. Is that right?
DR. FITZMAURICE: Yes, we do. We're scheduled to go at 11:30 a.m. We can go to 12:00 p.m.
MR. BLAIR: So if we did that, if each of you have 15 minutes of testimony, and then that would be a good 20-25 minutes available for questions afterwards. So Herman Jenich, please begin.
Agenda Item: Second Panel - Health Data Quality and Users - Herman Jenich, IPRO of NY
MR. JENICH: Good morning. My name is Herman Jenich, and I'm the associate vice president for managed care at IPRO. For those of you not familiar with IPRO, IPRO serves as the Medicare peer review organization for New York state, holds Medicaid oversight contracts in several states, and serves a variety of private customers as well. Through its work with this variety of customers, IPRO has gained a lot of experience not only with the quality of health care, but with the quality of health care data.
I would to say I'm delighted to be here with you today, to share with you some of the lessons we have learned about data quality, and to provide you hopefully with what you think is useful information as you embark on your efforts to develop electronic patient medical record information.
I am responsible for the implementation of a wide range of IPRO studies regarding health plan processes and outcomes. What I'll be presenting for you today is some of IPRO's work basically in the managed care area. And in general I will be drawing from three different types of work that we have done.
The first is HEDIS auditing. I'm not going to talk a lot about HEDIS. Mr. Lamb this morning did a very thorough job of telling you about what HEDIS is. IPRO is one of the organizations that audits HEDIS data. We have overseen hundreds of HEDIS audits and probably reviewed tens of thousands of medical records that have been used to support HEDIS rates.
I'm also going to draw a lot on our work in encounter data validations. More and more states and now HCFA as well are requiring that health plans report claims encounter data to either the state government or HCFA. However, because of the limited analytical utility of this data up until now, the data hasn't really been used very much. I will tell you a little bit about what we have learned in that arena.
Finally, I'm going to tell you a lot about quality improvement studies that IPRO has conducted. There are a wide variety of clinical quality improvement projects. These projects are scientifically designed and conducted with an emphasis on examining patterns of care and patient outcomes.
The material I'll present today will be based on these three areas. I won't repeat a lot of what Dr. Jessee and Mr. Lamb said this morning. I think they did a really good job of laying out what should be components of the medical record, what problems with the patient records are, confidentiality issues.
I'm just going to go ahead and talk to you most as a user of data, with the focus on the nitty-gritty limitations in our current systems, and hopefully provide you with some important items for consideration as you move ahead to develop computer-based patient medical record information standards.
In my testimony today I would like to address three key issues for you. One is why do we need comparable patient medical record information? I would like to talk a bit about what are some of the key system limitations that currently hinder health care data quality. And then finally talk some about what are some of the issues that government and industry together may want to address collaboratively in the development of patient medical record information standards.
So let's get right down to it. Why do we need comparable medical record information? This morning we have already heard a lot about the need for this information, who were able to fully affect health care organizations' care time limits, access to care. I'm not going to repeat all those. I'm just going to concur with all of our previous speakers.
The only other issue in this area that I might add a little bit on is one of the questions the committee did pose was thinking about this information, what is an acceptable level of bias. How close do we have to be before the information is really meaningful? I can say that currently even when health plans take great care to actually process administrative data and consistently review medical records, the resulting performance rates as of now may still not be accurate or consistent across health care organizations.
Determining the maximum amount of bias depends largely upon two things. One is your needs and the needs of the users of data, and the other is the ability of the validation techniques that we currently have to detect differences among health care organizations' data. For example, NCQA has as part of its HEDIS compliance audit processes, requires that audits be biased by no more than 5 percentage points.
While this level of specificity is generally acceptable to users of data, although we heard today that people would like to be able to detect differences at 2 percent levels, even validating at 5 percent given our current data infrastructure is quite a monumental task. While we would like to be able to say on an audit, every last decision we make when we make a 5 percent distinction is based on complete and full data, oftentimes it is based on judgment, based on whatever we could get. So even getting down to 5 percentage points right now is quite a task and quite an expense if you talk to a health plan that has undergone a HEDIS audit.
Like I said, everything about why we use patient medical record information, I completely concur with all the previous speakers.
I think I would like to spend most of my presentation on talking about some of the system limitations that I have seen out there. Like I said, IPRO has done probably about 150 audits on its own, in addition to the last two years we have overseen for HEDIS 1998 and HEDIS 1997, all of the Medicare HEDIS audits, and that was 284 audits for 1997, and about 354 for HEDIS 1998.
So we have seen a lot of different health care system limitations. And what I have done is I have categorized them in five areas: clinical data, membership data, provider data, vendor data, and then data integrity, pulling all that data together.
In developing the patient medical record information standards, I doubt that any committee in the next few years is going to be able to fix every single one of these issues, but what you do I think it's fair to say, will at least impact all of these issues. So let me begin by sharing some of the key system limitations we have seen in clinical data. When I say clinical data, I'm really talking about the claims and encounter data that is flowing into health plans.
The first several all relate to coding. We talked a little bit about coding today. I'm going to talk in a little more nitty-gritty about some coding issues. One is certainly the use of proprietary coding schemes. Some health plans still require that that provider use homegrown proprietary diagnosis or procedure codes in place of standard CPT and ICD-9 codes, which hampers performance measurement across health plans.
In addition, individual providers working for multiple plans may be faced with the difficulty of keeping track of multiple coding schemes for diagnoses and procedures across health plans.
The second issue is modification of standard codes. Many health plans do say they use only standard coding schemes. But when we arrive on site for a HEDIS audit, we realize that some of these standard CPT codes and ICD-9 codes have been reinterpreted. For example, in the area of preventive care or well woman care, you may take a code that simply means a woman came in for a health check up and all the sudden a mammography and pap smear is all attributed to this code, when indeed that is not a standard process.
Many of these changes are done to accommodate billing and payment needs. In large part coding interpretations often occur because of provider fee schedules. But when you turn around and try to use these codes similarly across organizations, you find out that the underlying usage of these codes is very different among providers in different health plans.
Another important issue is limitations of current coding systems, as well as the standard coding systems, particularly those designed for billing purposes don't always capture health care data as needed for performance measurement. We had some good examples this morning regarding ACE inhibitors. We have tried to study ACE inhibitors with health plans and find that to try and identify the right population often results, given our current coding scheme, of having to throw out 40-50 percent of the records.
Another is the retinal exam; the exact same thing, trying to identify what diabetics should get retinal exams, what exactly is a retinal exam. Using our current coding schemes requires pretty complex combinations of codes and administrative data that still don't quite give you exactly what you need. Indeed, many health plans do have to go to the medical record to be able to do this.
Use of proprietary forms. Not is there variation in how codes are used, but then for the next step in data collection processes, okay, how do you take all these codes and capture them? Many of you will be familiar with the standard HCFA 1500 UB-92 forms. Not many providers use them, but some health plans use their own proprietary forms to help facilitate the data capture process.
The problem is oftentimes these forms don't capture all the relevant clinical data that providers actually are performing. Or you may have to distinguish important distinctions in the data. For example, it's very typical for proprietary forms not to look at distinctions between primary and secondary diagnoses, or primary and secondary procedures, which when you try and do certain types of quality improvement efforts and performance measurement, it becomes very relevant.
The amount of variation in the rigor of data editing. Some health plans have very sophisticated systems to insure the quality of their data make their data useful for both billing and performance measurement purposes, while others very simply do not. And this again results in a lot of variation across health care organizations.
Many health plans suffer from the problem of multiple, hard-to-integrate data systems. Some health plans, particularly those who have undergone mergers, acquisition, and or major system upgrades, and as you see the ripening of the managed care industry, that is becoming more and more everybody, are finding that their data now reside in multiple systems that are really a legacy of their history. Trying to combine data across those systems with their different coding, their different identifiers is becoming very difficult.
Finally, I think one of the most important and difficult problems we struggle with right now with data is data completeness issues. Many health plans, particularly those paying capitation to their providers, don't receive anything from providers on encounters or claims for services provided by their practitioners.
This results in either one of two things. One, typically going back to the medical record to get the data that you need. Or two, systemic underreporting of rates of performance, and huge variation across plans that are good at capturing data versus those that are not. I know Mr. Lamb earlier talked about efforts to sort of improve data capture and NCQA's information roadmap.
And also the effort more and more to move health plans away from using medical records. But for many measures, the move away from medical records means rates that are reflective of data capture issues, and not health care performance. At least that's what I think we as auditors have seen out in the field.
Membership data. By and large membership data is about the best quality data we see. Clearly membership data has a lot of effects on payment in a per member per month environment, so you see relatively good member data. However, there are still a few things with membership data that are problematic for quality improvement and performance measurement purposes.
Patient identifiers often vary across health plan systems, even within a health plan, making it difficult to consolidate all of the clinical membership and provider data for a particular member. Especially in health plans that have purchased or acquired other health plans, those two plans had different member identifier systems.
Oftentimes you sort of are starting from ground zero on the transition date, and you really can't link what happened to a particular patient before the acquisition and what happened afterwards. For some of the measures that require longitudinal assessments, two and three year time frames, you really are losing the ability to measure performance.
And again, data being maintained in multiple systems. Many health plans capture enrollment data in multiple systems, particularly if they serve more than one population. It's not uncommon for Medicare, Medicaid, and commercial data to reside on different systems.
Performance measurement efforts are sometimes hampered by the inability to link members who transition from one population to another. Medicare is always wanting to know what happened to the members that are aging in from the commercial Medicare population, because it's often a real break in your ability to see what is happening to someone when they turn 65.
Also, many health plans have difficulty tracking dependents who change relationships to the primary subscriber. Membership numbers are often driven off of who is paying the bill. So if you are just in a relationship with the person who is paying the bill, and the relationship changes to the person is actually "insured," you may often get lost in the shuffle when trying to do performance measurements.
Provider data. I think as HEDIS has evolved to more and more complex measurements, we've been taking a closer and closer look at provider data. I think provider data is often right now the most complex and difficult to sort out data out there right now in health plans. There is a lack of consistent provider identifiers.
Providers often vary in how they are identified within different health plan systems, making it difficult for the health plans to consolidate all of the data that they have on providers. Also, even within a single system at the health plans, providers may have different identification numbers for each location in which they practice, or each provider group that they participate in.
Provider data is very often housed in separate systems. One good portion of information that relates to contracting and billing is in one database, and data related to a provider's credentials or specialties are often kept in different databases, and sometimes incompatible databases.
Finally, the fourth kind of data is vendor data. When I say vendor data, I am talking about one of two groups of data. One is the wide variety of ancillary service providers, those such as pharmacy, laboratory, behavioral health, home health, eye care. The health plans rely on a huge percentage of the data that a health plan uses from vendor data.
Vendors are also sometimes even used for basic health plan functions such as processing claims and encounters data. The key problem is often health plans are unable to use vendor data for performance measurement purposes simply because: (1) the vendor doesn't collect the data; (2) so okay, the vendors get the data, but the health plan does not collect the data from the vendor; (3) the health plan is so differently formatted that they simply cannot use it for performance measurement.
Last but not least are issues of data integration. For a variety of reasons mentioned above, health plans often have difficulty consolidating the claims, encounter, membership, provider, and vendor data required for performance measurement. Health plans that have multiple systems for each of these data types, as well as those that have recently undergone a merger, acquisition, or significant system upgrade, have the greatest difficulties.
The record right now is a California firm both for acquisitions and mergers, and I think the count was approximately 1,200 databases pulled together report a rate of X percent.
So what do we do about all of this? I don't have any perfect answers for you, but I have a few things that I might want to recommend that you think about in your efforts to develop standards for patient medical records. I think the first thing is to in your development process, really take the opportunity to address some of the shortcomings of our current health data infrastructure.
We talk about automation and computerization, claiming that automation ought to be fast and cheap. But the new automated systems simply codify, and sometimes even magnify the existing problems. The electronic standards should, for example, address the variations and limitations in existing code schemes and the need for unique patient and provider identifiers for linking multiple databases.
Second, any patient medical record system must consider a variety of potential users, including the individual practitioners, health plans, hospitals, vendors, government and private purchasers, and even patients. A set of standards that meets the needs of only some of these constituencies is less likely to succeed than standards that address the needs of most or all.
Next, the success of any electronic patient medical record information standard really relies on the commitment of those individuals who will enter the data into the system. This includes not only clinicians, but also a variety of administrative staff that support clinicians in their documentation and billing efforts. Clinicians and their administrative staff are most likely to pay attention to how they enter information into a system that provides them with timely information on their performance, with comparisons to peer groups that they believe are meaningful.
The patient medical record information standards development process should address how individuals will be trained to accurately code and enter medical record information. As mentioned earlier, some health plans and individual practitioners currently use non-standard coding schemes or data collection forms, and others have difficulty properly interpreting the current standard coding systems. Individual practitioners and the administrative staff that support them must fully understand patient medical record information coding requirements for any electronic patient medical record information initiative to succeed.
The PMRI standards should include guidance on how to edit and audit electronic PMRI data. Health plans vary int he level of rigor that they use to insure the accuracy of their administrative data. Setting minimum national standards for automated edits and system audits will help insure the comparability of health care data across all organizations using the PMRI standards.
The PMRI standards should be flexible to account for changes in health care delivery and information system technology. A process should be created to continuously review and improve the standards.
Finally, but importantly, the standards for electronic PMRI should include sound requirements for data security and patient confidentiality. One unintended benefit of today's disjointed health care data infrastructure is that it makes it more difficult for unauthorized users of data to piece together all of the clinical information related to a particular patient or provider. Because the electronic PMRI standards will likely make it easier to link data from multiple sources, the need to protect patient and provider confidentiality will become even more important.
Finally, I just want to say that the development of standards for electronic patient medical record information will assist both health care purchasers and providers in their efforts to measure health plan performance. In addition, the PMRI standards will likely encourage the development of fully automated medical record keeping and thereby eliminate the need for costly abstraction from paper medical records to support performance measurement.
Finally, for any electronic PMRI effort to be successful, government agencies and the private sector will both need to be active participants in the standards development and implementation process.
That concludes my comments for this morning. I look forward to responding to any questions that committee members have.
MR. BLAIR: Thank you very much, Mr. Jenich.
For those of you listening on the Internet, let me just make sure that our mission is to study standards for patient medical record and the electronic exchange of that information. We don't envision ourselves developing standards ourselves.
Dr. Griffith.
Agenda Item: Health Data Quality and Users - Stanley Griffith, M.D., Indian Health Service
DR. GRIFFITH: Thanks, Jeff.
A quote I just heard two days ago I think is a most appropriate lead in to my testimony. I believe that Mark Twain said that a man walking down the street carrying a cat by the tail has ten times as much experience as someone just watching. So what I bring to you is the collectively wisdom of innumerable of my colleagues who have been trying to actually enter data into these systems and to use them.
I have divided my testimony into five sections, and I'm very quickly going to go through the relevant background. I'll give you a brief overview on the Indian Health Services' health information systems. I'll address a couple of questions very quickly -- how important is data quality is to our work, and what has IHS learned about that? Finally, I'll give you my conclusions and recommendations.
First, my relevant background so that you know I've been a practicing family physician since 1980. I was the medical director of a typical IHS facility in Washington state for ten years. I've been both a clinical and administrative user of our clinical information systems for over 15 years now. I've helped lead both outcome measurement initiatives at the agency, regional, and local levels, including HEDIS measurement systems.
I have participated as the functional lead in the design of one of our major clinical applications. And I have served a number of years now as our functional lead to our clinical information system at the Indian Health Service. Most recently I have served as the functional lead for our agency to the Government Computer-based Patient Record Group, which you heard about in June.
PCC is the clinical component in the suite of applications of our entire health care information system. It's a system that is like most clinical information systems. There is an electronic abstract of the patient chart. It does not replace the paper chart. Most data are entered by data entry clerks from a written encounter form, although there is some point of service data entry that is available in this system.
From the beginning of the design of system, over 20 years ago it was designed around one single, comprehensive, central data repository. While that certainly is not unique today, 20 years ago it was incredibly innovative and far-sighted, and won the system a Smithsonian Award.
It integrates information across the entire enterprise, rather stove-piped or department or very specific programs. It is implemented at over 200 facilities, ranging in size from small rural, isolated, one provider, one day a week facilities in rural Alaska to tertiary medical centers. Of those, I would say that many of those systems have been in place for over 15 years, and most have been in place for over a decade now, so we have a wealth of experience about what it's like carrying a cat by the tail.
It's not just a government systems. Presbyterian Healthcare Services, the largest managed care organization in New Mexico state has PCC as its outpatient information system. It is also implemented at the Clemson University Nursing Wellness Center, Saipan Trust Territory Hospital, and various other sites.
Finally, today, the PCC interfaces through various HL7 messages services and various other COTS applications such as the Viking pharmacy package, and various billing packages. So that gives you just a quick flavor of what our information system is like.
I don't have time to go through it, but in my written testimony I give you an idea of some of the outputs the system provides.
How important is data quality? Well, I think I'm telling you something you already know, but clearly data quality is critical for individual patient care errors and since it can have serious adverse consequences to the patient. IHS has been a public health organization since its inception in the early 1950s. We fully understand, recognize, and utilize these systems to monitor our health care access, the delivery of preventive health care, monitor disease trends, monitor the effectiveness of various interventions. That just cannot be done unless we have reasonably accurate data.
Bad data leads to bad policy, and bad policy can significantly adverse affect an individual's health care just as bad information can affect an individual health record.
We also have come to realize that there are other administrative uses of this clinical information, and our clinical information advisories also very important. Poorly managed care, poor billing and cost recovery, inadequate legal documentation all reduce the resources we will have available to provide care to the populations that we serve.
Very quickly going through this, what have we learned? We have gathered our information on data quality at the Indian Health Service in a number of different ways. We have done it informally. I have used the systems for 15 years, as have many of my colleagues about the quality of data and how it varies, and where it's good and where it's bad.
We use certain standardized audits. There are standardized audits at each of our facilities that use this. The use of intelligent error checking -- the program itself does intelligent error checking. This slide I just wanted to show you, we are also doing some fairly rigorous research studies on the quality of our data, and this slide just shows you under the written testimony you've got a little more detail about some of the recent studies too that we have conducted. One I'm participating in right now. What I'm going to tell you about comes from some of these studies.
We have found that data quality problems in our systems vary. We see things like erroneous data, where blood pressure is recorded as 180 when it's actually 120. We also see missing data. We see that a lot more. Data where the information system will accommodate the data, but because of problems within the structure -- lack of resources, lack of time, lack of training, lack of terminals -- the data never gets into our system.
We have problems with lack of specificity. A typical example of that would be diabetes, a patient with uncontrolled Type II diabetes being coded in the ICD coding system as 250 rather than 250.02, which would provide us further information when we go back to try to review that information.
We have problems in our system about data getting there in an unstructured form. And the obvious consequences as well of being able to retrieve that information in any meaningful way.
Finally, we have problems with data that is just not contained in our system. We see all of these kinds of problems.
We see that something we have observed over the years, the data quality in our systems varies. It varies between facilities, and between departments. In my experience, this is largely because of structural reasons -- because the hardware is not available, the training and technical support is not available, the computer applications themselves that a facility or department is using is not as user friendly as it should be.
The second category though I think is more interesting. And that is the data quality varies between elements in a patient's records. The reasons for this I want to take some time to explore and get into a little more detail.
We find that certain elements in our data system vary from high quality, medium quality, low quality and extremely low quality. The high quality data in our systems would be medications. Medications are recorded in one place in our paper records. It is picked up by the pharmacy and entered into our system in a point-of-service data entry system.
When it is entered, it saves that pharmacist time, because automatically they print out a label for the bottle, another label to be put in the chart to show that you actually filled the medicine, and check that label to show it was picked up. They can print out in automated fashion the various reports the pharmacist needs -- product reports, drug utilization reports. It makes the pharmacist's job easier. It is recorded one place in the chart at a point-of-service data entry.
Vital signs are recorded one place in the chart, and only in one place. It is used by people who are entering the data.
Immunizations and paps are still of very good quality, but not quite as good. This is the information that the providers use. They record this information. They are the full spectrum, by the way, of providers. They are the ones who actually need this information. So you've got the people entering the information who are the same people who are using it.
Low quality is patient education and smoking history. Smoking histories and patient education, because they are recorded in multiple places in our chart; patient education in the one place where data entry will pick that. The provider has got to remember one of 100 different codes, and if they don't code it, it doesn't get entered. I have a choice of either remembering the code or referring to a manual which may or may not be in the clinic at the moment I need to actually record the data.
Smoking history is not quite that bad, but very similar. You do have the people entering the data, using the data though, and that's a plus. But data quality, because of the reasons I just mentioned, start to drop off.
And finally, cause and place of injury. That gets recorded in two different places in the chart. The place where data entry picks it up is not the typical place where I would tend to record it. It is recorded in part by a check box. But most crucial, I believe is I'm not the one if I record it, that's going to be using that data.
It is used by epidemiologists. It is used by our engineers, our sanitarians who are worrying about injury control. As important as that is, when I've got 40 patients waiting, and everything is going crazy, that's not my highest priority to check that box.
I'm going to come back to that in my conclusion, but I wanted to make this comment in response to one of your questions. Clearly, data quality problems -- I don't need to tell this group that quality problems impair our ability to deliver care, to perform research, to do outcomes analysis, all to manage our programs to do all the things we need to do.
But beyond that, and that's because in large part of data quality problems we have in the data that gets entered into our system. But a more profound problem for us with an abstracted EMR is the data that just doesn't get into our systems right now. We have a large amount of data in the written chart that just does not get into our systems.
In my view, one of the primary reasons, not the only reason, is because of a lack of consensus coding systems, data dictionary, lexicons, the full spectrum of infrastructure we need to actually support getting this information into our systems.
Now this Work Group has gathered a lot of testimony. We and the GCPR project have gathered a lot of information. It is clear to us that there is a lot of private and public consortiums out there that are dealing with various aspects of infrastructure. But it's also clear to us that there are of holes.
One of recommendations will be, and I'll repeat at the end is one of the most appropriate roles in my judgment for federal government is to identify these holes, and to use the powers of the federal government in the various mechanisms it has available to it to deal with some of these holes. If we don't deal with these holes, we're not going to be able to move on to the fully functional CPR.
Conclusions and recommendations. In my judgment, and from the perspective of somebody who actually enters the data, and is being asked to enter the data, and then is actually utilizing the data, it is very clear to me, and I suspect it will be very clear to you all that data quality is going to improve when we accomplish the following:
When those we are asking to provide specific information are the same ones who are actually going to use it. They need to be the only users, but at the point where we are asking the people to enter the information, if they are going to be the ones who are using it, we are going to get better quality data.
If I know that the diagnosis I'm entering in my progress note is the diagnosis I'm going to see when I get a call at night, and I pull up a health summary on the screen so that I know something about that patient, I'm going to have a vested interest to make sure that that diagnosis is recorded well and accurately.
It's going to improve when all staff depend upon the electronic, rather than the hard copy. If we've got our clinicians and other providers depending upon a clinical information system, and billing depending upon a written super bill, and administration depending upon other systems to deal with work loads and cost accounting, it shouldn't surprise anybody that we don't have the systems in place to make sure that the data quality in our clinical information systems, and our more broadly described health care information systems are high quality.
Data quality will improve when we made data entry non-redundant. It should surprise nobody in the room when I have to record one diagnosis on my progress note, and I have to record the same diagnosis on an x-ray or lab requisition, and then I have to record the same diagnosis on the super bill, and then I have to record the same diagnosis on the referral sheet or an admission not, then one of those times it's not going to be done right. When we make our data entry non-redundant, the data quality inevitably is going to improve in my judgment.
We will find data quality improving when we get data more fully structured and codified in a consistent manner so we can implement some of these things which I believe clinicians will find very useful -- decision support, intelligent alerts, automated billings. These kind of things we are not going to be able to do until we begin to standardize and codify the information that we are putting into those records in a consistent manner.
When more parts of our organization depend upon the same information. Practically, I will tell you we've got more attention in our agency on data quality now that our organization is dependent upon money for billing, and is more dependent upon some of the numbers we can generate for outcome analysis than we ever got when we were just using these systems as clinical information systems.
That wasn't because people were bad and they didn't care. It's just that it become a vested interest of the billing departments, administration, and others to make the data quality was good, and therefore suddenly we got attention to the infrastructural needs we needed at the local facilities -- the training, the hardware, the technical support -- to actually get the data in there correctly.
We will improve our data when we have efficient and accurate point of service data entry. When we have a robust and accurate voice recognition system, which I will tell you right now we do not have, when we have that, and have one that can actually enter data in, in a codified manner, which is another step, then I believe when we have true point of service data entry, that our data accuracy would improve.
That's one I would probably slightly disagree with our previous testifiers. I do believe there are some technological advances we still need to see before we're going to be able to improve our data quality. And it probably has to do with point of service data entry. I don't think we're there yet.
Finally, probably the key point of this entire thing. We will get better data quality in our systems when our systems are designed to make it easier for our staff to do their jobs well, and data is just collected as a byproduct. If we can do that, we will get good data quality.
Our systems will never, in my judgment, be effort neutral. There is always going to be some added effort. But when we're talking about an extra 10 minutes per patient, when we are allowed 15 minutes per patient to get the data in there, it just isn't going to cut it.
It is interesting to note by the way, and I'll make this comment right now, if you look at those criteria I have just listed for improving data quality, those are the criteria that I think most of us would include if we were to design a fully functional CPR. What we are talking about here is to improve data quality, we need to promote and design a true, fully function CPR. That's my point in this slide right here.
Interestingly enough, data quality is going to be absolutely critical if we are going to have a useful, fully functional CPR. Interestingly enough, it's that very goal of achieving the fully functional CPR that could be the single greatest step we could take to improving the data quality we have in those very systems.
And of course the benefit of that, as all of you are aware, is improved quality of health care for individuals, and improved public health care, more effective service delivery, improved financial well being, and improved legal risk management. That all derives from that, but we've got a bit of a great Mandala here.
So therefore, my recommendations, my final slide. I believe this Work Group and its single most important step it could take to improving data quality would be to focus on the promotion of the development of a fully functional CPR. That's the single greatest step that you could take. Our health care organizations don't have this luxury. Our production systems right now are abstracted EMRs. We cannot afford to wait until a fully functional CPR is developed before we deal with data quality.
But this Work Group has, I think, the luxury to look to the future, and I think that's really what you ought to focus on. And everything and anything you can do to promote us getting to that future quicker is going to have the biggest impact in my judgment, in improving data quality. It's going to be far more effective than trying to beat people over the heads to do better.
Secondly, and a point I previously made, I'm going to re-emphasize it. I believe the most appropriate role for the federal government is to identify which aspects of the necessary infrastructure are not being adequately addressed by current ongoing efforts. That's the information models, the coding systems, the data dictionaries, communications, secure aspects that are not being dealt with adequately now, and identify the holes. And then to move to fill these holes through the various mechanisms the federal government has available to it.
Many of us are absolutely convinced that if a critical mass of a standardized infrastructure is developed to allow us to move to that CPR, that private industry is going to be ready and willing to move forward in creating the CPR applications themselves. I strongly feel that should not be the focus of the federal government.
Thank you for this opportunity to share IHS's experiences with you, and for me to share with you my recommendations.
MR. BLAIR: Thank you both for excellent, excellent, informative testimony.
Dr. Cohn, could you help me identify the individuals that have questions?
DR. FITZMAURICE: I wanted to ask Herman Jenich questions about -- often it is useful to have anecdotes about how hard it is to get data for HEDIS measures in the hopes that spreading those anecdotes will improve the system capabilities to give us that. We have heard about ACE inhibitors, how difficult it is to identify the populations that should have them, let alone look at the actual rates of use.
Also with the retina eye exams, difficulties in determining the populations that should have them, but also what is a retina eye exam, we heard in testimony. Have you a couple of other anecdotal information based upon your audits that could give us ammunition to improve systems? The anecdotes may be focused on a particular diagnosis or condition, but they let us generally look at coding system to see what improvements need to be made in coding systems. What additional detail need to be grabbed to make the data that is already being collected valuable?
MR. JENICH: I could probably give you some examples that probably cut in a variety of different ways. The broad issue of needing to capture all the information that is there, talking about secondary diagnosis and procedures, especially in behavioral health is particularly important.
Behavioral health is an area where it's very tough to capture administratively, but it's also very tough to capture medical records, because the patient confidentiality and security issues are even tighter around there. So when your health plan is trying to measure behavioral health issues, those are among the most difficult.
In terms of confidentiality, right now from the health plan's perspective -- and this applies to every clinical condition -- more and more providers, states, societies are encountering tighter patient confidentiality restrictions. So health plans more and more as they try to do HEDIS measures, are getting less and less access to patient medical records. So rates are going down.
That sort of stymies the effect. On the one hand you could say it would encourage quicker development of electronic systems, but at least in the short-term of becoming very problematic.
Something Dr. Griffith mentioned, for many clinical conditions the specificity of coding is very important, particular in identifying the right people. For any record there is denominator and a numerator, and I could contend the denominators are harder to figure out than the numerators. Who is a diabetic is a tough question, and that really goes across a lot of measures right now as identifying who is eligible for the different populations. Those are the things that most come to mind.
MR. BLAIR: Anyone else with some questions? If not, let me ask my own. I don't want to jump in before somebody else.
I think both of your testimonies were very informative. You are both involved in great detail in capturing data, and using it in one form or another. And Mr. Jenich, you have tremendous experience in auditing data, so it's very helpful to have your perspective on this.
I picked up a difference in perspectives, and I'd like to see if I can get clarified a little bit. Because the impression that I got from Dr. Griffith's testimony was based on his experience with I think he referred to it as a partial patient record, that does the abstracts, but at least captures some things at the point of care.
And based on that experience you are envisioning that if you are going to improve data, you seem to feel as if unless the system does the coding, that it is going to be too onerous for a human being to wind up improving specificity if you are going to expect a clinician to look things up in a book and look up codes.
So if that is correct -- let me just verify if that's a correct perception, Dr. Griffith.
DR. GRIFFITH: That's the crux of a real problem, yes.
MR. BLAIR: Then if I were to take that back to Herman Jenich's comments and his experience, do you feel like that is kind of a reality, that while it would be helpful to have greater specificity and completeness in data capture, that we cannot impose that on an individual to do coding? Would you say that that is correct?
DR. GRIFFITH: I believe we're going to have to come up with better systems that are going to allow us to code eventually at the point of service. I don't think we can do that now. Just to give you an example, last week I tried to look up a CPT code for a routine coronary angiography. I'm not a coder, but I'm not that unproficient in coding. I spent 15-20 minutes trying to figure out what the right one was. I couldn't find it.
I eventually called up a coder. They took another 15-20 minutes to figure it out, because there is some fair complexity about if you this it makes it that, it makes it that. It's very convoluted.
We have got to get to the point where we get to point of service data entry. We cannot have with current technology currently available applications, have our providers do the coding at the point of service.
You're absolutely right, clearly we've got to get there at some point through some sort of intelligent system that eventually will gather enough data that it can do the coding. An example would be if we get to a fully functional CPR where we gather in codified fashion, a review of systems, a fiscal examination, and a diagnosis with some sort of taxonomy that tells us the level of risk there, how complicated a diagnosis is, we should be able to do automated E and M coding through an intelligent system.
I'm hopeful that if we get closer to a fully functional CPR, that we're going to do similar things for coding. That we are not going to ask the providers to remember that the code changes if a simple thing they should remember, if it's type I or type II diabetes, or if it is controlled or uncontrolled. But we should have systems that will pick that up and actually give us the increased specificity.
No, I don't think it should be the providers. I don't think we can train all the providers to be expert coders. We have to have systems that do it. But it can't also efficiently occur if we send all this information back to a person to go through and run through and code later on. That may be an interim solution, but it's not an eventual solution.
DR. KOLODNER: This is Rob Kolodner. One of the things that your statement reminded me of Stan were ones that Chris Chute talked about in terms of the level of granularity in that a lot of our coding systems are aggregate systems. And that what we are looking for, for the front line clinicians are ways of capturing the clinical data level specificity that we talk about when we are talking about patient care.
And then the system then would roll that up to a higher level of aggregation, although most likely there would need to be some interaction, that intelligent system that asks the two or three yes/no about the data elements that might not have been captured that distinguish between coding.
My sense is that when you talk about coding, you tend not to distinguish in the conversations, the kind of thing that you were talking about, what the clinician needs versus this other reported activity.
MR. JENICH: And just to say on this issue we concur. I think that ultimately it has to be done at the point of service, and it have to make it easy for providers, because my God, to be able to code, you have to send someone school for years. And then you take five of these people and put them in one room, and to get them to agree on what one chart's principal diagnosis is, it's really quite an effort.
But I think one important point more from an "auditors" perspective is that someone facilitate what the coding system is going to be. Many health plans, for example, have embarked on super bills, often presumably to make it easier for providers to code, but every super bill is different. And it's very hard to do comparable information to plans.
One that for example, I went to, it used the super bill. The way their system worked is they had super bill. You checked off from asthma to well child. We thought it was odd that they had asthma rates through the roof, and well child rates that were one-tenth of what they should be. You realized that well, in their development efforts they didn't distinguish the principal diagnosis, so they just alphabetized them. So if there is more than one, asthma was always principal, and well child was always not. So there is a lot of technology that still has to come around to make these systems really functional.
DR. COHN: Kathleen Fyffe has the final question.
MS. FYFFE: Thank you, Dr. Griffith. I don't know much about the size of the Indian Health Service. Is there any volume you can give me about the number of patients seen?
DR. GRIFFITH: Sure. We've got over 500 different facilities. We serve about 1.5 million users. We have sites I believe in 39 different states. Does that give you a feel?
MS. FYFFE: Is that within the ball park of let's say the Kaiser system in California, 1.5 million users?
DR. GRIFFITH: Users. Patients that have used the system in the last three years. We include people who have not used the system in over three years, but are eligible for care, and may have used it four years ago. It's more than that.
Just to give you a ball park, I was just at a meeting with the VA and they said that their ambulatory users were between 3-4 million. So we're not as large as them, but we're not actually as small as many people would assume.
MS. FYFFE: Right. Okay, thank you.
MR. BLAIR: Okay, we need to take our lunch break at this time. Let's convene at 1:15 p.m., please.
[Whereupon, the meeting was recessed for lunch at 12:15 p.m., to reconvene at 1:15 p.m.]
MR. BLAIR: We have had a few additional committee members able to join us since we introduced ourselves first thing this morning. So let me go around and let everyone on our committee and our staff introduce themselves.
[Introductions were made.]
MR. BLAIR: This is our third panel, and it is focused on users of standards. Just to put this in perspective a little bit, in March we had testimony from developers of message format standards, and a little bit of testimony on data quality. In May we had two full days where we heard testimony from developers of medical terminologies.
We're now trying to get a little bit different perspective. And we are looking both for positives and negatives in terms of people's experience with using standards that relate to patient medical record information, whether that is message format standards or terminologies, or any other experience.
As you can see by the agenda we have Dr. Jeff Rose from Kaiser Permanente, and Gary Arvary. I'll tell you what, I'm going to ask Gary, would you like to go first, Gary?
DR. ARVARY: Sure thing.
MR. BLAIR: Why don't you introduce yourself a little and go ahead.
I should have given you a little bit of a framework here in terms of time. I think if each of you takes about 20 minutes for your testimony, and that would allow us to have the 30 minutes afterwards for follow-up questions.
DR. ARVARY: Great. I just wanted to present this basic crude slide to demonstrate my expressive view of what the public has demanded from the government, and what they are attempting to get from providers in order that they provide standards and information to people in the health information system who require it.
Right now there has been a big public demand for more efficient and effective health care. They and the government agree that the way health care is handled right now, it is inefficient, and it does not get health care to people who need it as quickly as possible.
So what happens is the federal government assigns federal agencies like yourself to look into the problems of standards, to look into the problems of the development of the health information infrastructure. They have a positive aspect in the development of electronic medical record systems, standards of those systems, and standards which would allow for the transfer of this information to people who really need the information.
Those people being insurance carriers, academic institutions and the federal government. They need information primarily for getting accurate data, so that they can make accurate determinations of actuarial analyses for accurate risk determination. Academic institutions need information so that they can develop best guideline practices so providers like myself and people like me can do a better job with patients.
And the federal government of course needs as much structured information as possible so that they can do population studies, epidemiologic analyses, and also determine where the money is going to, and where best to put the money. What processes to improve so that it can be most effective in the health care system. So those are all positive things. The public wants those. And they are influence providers in a positive way.
Unfortunately in this time there is a very strong negative influence on the providers from the same public. The public says it's paying too much for health care. Because of that, the federal government agrees they are paying too much for health care. We can't keep track of the money. We have to keep better track of it.
As a result, the federal government has assigned to other federal agencies the task of retrieving bad money that has been taken fraudulently or in error. Also, they have also allowed insurance carriers to enter the free market with HMO products with managed care products. What has created is primarily a managed care environment for providers.
What a managed care provider does to a private practicing physician in the private sector is it creates a bigger administrative overhead. On the average, one family physician will take one full-time employee more to manage a managed care practice.
As an anecdotal remark, in 1995 I was 50 percent managed care. I had an adequate amount of disposal income for technology. Right now I am 95 percent managed care, including 60 percent capitated, the rest discounted fee for service. The idea of the government was to reduce the salaries of physicians. Unfortunately, physicians aren't going to do that that easily.
What they have done is destroy their disposable income. Physicians are very apathetic towards the electronic medical record right now, because they can't afford it. In 1997, the AAFP through a survey determined that there were 4.3 percent of primary care physicians that used the electronic medical record. In a study done this spring that number was done to 3.2 percent.
This all means in the last two years the negative influence is winning over the positive. Now what are the positive influences over the health information infrastructure? What could happen for instance, if all providers used electronic medical records and entered structured information and data? Where would it go to? Do you think insurance carriers, academic institutions, the federal government would all want that structured information for accurate analyses? You bet they would.
What isn't commonly know is so would providers. Providers have been shown to want structured information. We want problem lists, allergies, medication lists, all structured, all available to us easily so that we make less mistakes.
Why do you think primary care physicians would do a better job even with poor hierarchical structured standards like CPD and ICD as Dr. Griffith mentioned before? Because they do a better job of making choices. If we do it ourselves, if we enter the data, it's much more accurate. Less mistakes are made in medications, less mistakes are made in diagnoses. Right now, that's not happening.
Part of the problem stems from the difference in definition of what patient medical record information is to the insurance carriers, academic institutions, and the federal government. As an oversimplification I'll say it is a sum total of all the health care information of a person. That's an oversimplification, but it's something that they expect to be quantified, codified, and sent to them so that they can do studies on it.
For a primary care physician or a provider that definition is extended to include all the information that accurately described the interaction of the patient with the physician. That is the big problem with electronic medical records. That's one of the reasons physicians have not been willing to use the electronic medical record.
Physicians would like to have several areas where standardized information can be transferred, and these areas are, when they are not in their own office, they would like to be able to pass some form for transferred information to analog phone lines to their home, to their cars, to remote areas. An emergency room physician would love to have structured information, even if it is only a chronologic problem list, an allergy list, and a medication list. Most of them would give their eye teeth only for that small bit of information.
The whole ball of wax does not have to be transferred. Standards can be developed so that electronic medical records can give that small amount of fragmented information in a meaningful way. The entire package could be transferred through high band width lines, DSL lines, et cetera, through a central hub or regional entered infrastructures type of set up that I will describe briefly later.
I have a brief comment on coding. In the outpatient setting in primary care we deal with two codes, ICD-9 or ICD and CPT. Both have poor hierarchical structure. Both cause a lot of misunderstandings, and both have a lot of problems. Unfortunately, they are the de facto standards right now.
If you are attempting to develop a new health information infrastructure to substitute these with something else that has not been tested, to map these to something like even ICD-10 results in lost diagnoses, obsolete diagnoses, difficulty in interpreting previous records. To avoid a lot of hysteria, probably the standards for diagnostic and procedural codes should be maintained.
My greatest area of concentration, however, is the medical office. What things are important to a physician in his office? It's not just how he does the record. It's how the electronic medical record affects his staff. Unlike popular interferences in the literature, most physicians don't use transcriptionists. We don't dictate our notes. We don't type our notes. Less than 20 percent of physicians type or care to.
Most of try to do the notes in the exam room. In a managed care environment we have 12 minutes to do an examination. How much time do you think we have to do a note? Twelve minutes. Why? Most of the time we do the note at the same time we're taking the history. It's very difficult to do with a desktop computer.
Studies have been done that show that physicians are mobile, and need to have the ability to move around in the examining room. They can do so with a pen base slate-type computer. Desktop computers are an environment unto themselves. They do interrupt the patient-physician interaction. This has been studied. It's been shown.
Physicians don't really care one way or the other. You can go in there with a shovel and some coal and write on it. They just care that they get the time that they want. Physicians care a lot. Most of them would not accept a keyboard in the exam room.
More important is how it affects their staff. We don't have transcriptionists, as I mentioned before. Telephone messaging, in which we do 20 percent of our work, is done by a $9 or $10 an hour clerk in the northeast region. These are accepted standards for salaries for this position. They have to be taught to triage. They have to be intelligent, and they have to be fluent in language.
Language is the thing that is missing from most electronic medical records. It cannot be duplicated with templates, drop down boxes, or any kind of other structured device. Language has to be represented as it is.
Unfortunately, language represents shared experiences and implied conventions that computers or any type of data dictionary could never, ever duplicate. For that reason, the attempts to make data dictionaries that are all inclusive are foolish. Why not leave a certain portion of the medical encounter as unstructured, and leave a small portion of it for codable data?
What has worked in the exam room? Physicians have been successful in research only -- and I agree with Dr. Griffith, there are no proprietary softwares that do this -- the research has shown that several types of factors are effective in the exam room. One is a mobile computer, and basic computer so the physician feels comfortable and moves. He should have one environment with the patient, not with the computer.
Another factor that works is a standard structured, coded medical vocabulary that remains the same no matter what check box is used, no matter what area of the medical record you are in. The sameness that makes it easy for the physician to remember the location of modifiers.
Likewise, the scroll bar should be removed in electronic medical records. They slow down the process. They do not adequately take advantage of something called spatial memory. It has been shown that the speed of the completed note in a pen-based computer with paginated, coded medical vocabulary is just as fast as a handwritten note.
In a study done this spring I asked 411 primary care physicians in northwest New Jersey to tell me what they would prefer, and what they were using. In that study I got a response rate of about 38 percent, not unlike a typical study done of that type. What these 156 individuals in primary care, including family practice, general practice, internal medicine, pediatrics, and obstetricians, and gynecologists indicated was 63 percent would agree to use a computer in an exam room.
Of those 99, 92 percent would prefer to use a pen-based type of computer; 7 percent would choose a desktop computer. As far as handwriting in the medical record, one of the advantages of using a pen-based computer and using unstructured text that does not have to be transferred, does not have to be codified, it just has to be present so that the physician has a note that he knows accurately represents his interaction with the patient. It is a legal document, and most importantly, it avoids misunderstandings. I'm going to give you an example, and this is in your workbook on what that would look like.
Basically what we have is in a normal electronic medical record that presently exists, we have several choices of modifiers we use to describe a particular part of the history of the present illness. When we pick out certain modifiers, they can appear in a prefabricated note, although that is really unnecessary.
Unfortunately, when it comes to giving more exact modifiers, or expressing language in the note, it has to be typed in. That's the second thing that is seen. In that case, we are substituting the modifier. The injury is aggravated by movement, with this more accurate statement, lifting kids and hanging clothes. That is then transferred to the note, and replaced that one word that is not totally accurate.
An alternative would be in a pen-based electronic medical record the one level of complexity or the aggravating factor could be left in as moving. What could be added is a more accurate state underneath without any loss of time. That would be, the pain is made worse by lifting kids and hanging clothes. I recognize that data is not easily codified, if at all. And I recognize that it is not easily transferrable over analog phone lines. But basically, physicians do not need all the information in a chart all the times.
In a centralized information structure health information network what would be possible? What would be possible would be to transfer all the information, structured and unstructured in the chart through high band width lines. This could be done because the cost of this communication is coming down.
My recommendations at this time based on anecdotal experience, and based on research in surveys that have been and available technology are several. One, because of the present state of the health care system in the private sector and lack of disposal income, subsidies should be made to developers of software so that they can standardize this record and include certain aspects that make the note easier to produce in the exam room.
These things would include a medical vocabulary that is not all inclusive, but is standardized. This should paginated. It should be as easy to use with a pen-based computer as it would be desktop computer. Likewise, parts of it should be able to be fragmented and transferred over analog phone lines if necessary, while the entire note can be transferred over high band width lines like T1 or DSO lines.
Second, physicians should also be subsidized to some degree to purchase these systems in order that the health information infrastructure can begin. If they are able to purchase these systems and begin to use it, they begin to see some of the improvement in office processes with Internet activity. There will be reduced chart flow. There will be reduced traffic in the office. There will be less stress because of endless contact with your colleagues in the office.
Finally, as far as the structure of the health information infrastructure, probably the best overall picture for it would be a regional one. Why regional? Primarily because cultures are regional. Referrals are regional. Insurers are regional. Even Medicare is regional in administrators. The administrators of these networks should be in the private sector. They should be small groups and small IPAs who already have formed as a result of managed care trying to reduce cost.
They have already seen the need to have Internet activity for communications, and avoid the cumbersome, expensive nature of courier service. They would be likely candidates to begin these small centers of activity for the infrastructure. Once the infrastructure is in place in the small groups, larger numbers of members could join. Subsidies should be given to the use of communication lines, including T1 lines.
After several years, and up to 10 years, I think that a centralized regional health information network, as part of a national health information infrastructure could be self-sufficient. It could exist as a not-for-profit and a for-profit segment. The for-profit segment could take advantage of the information technology personnel that it hires to run the system. And they can offer directory services for regional vendors, tertiary care centers, CME programs, and avoid all the mailings that we presently get right now. They could also offer Website services, Web development for their subscribers.
In conclusion, providers are the biggest portion of the health information infrastructure. We have to provide all the data to insurers, to academic institutions, and the federal government with adequate information, or the health information will be basically worthless.
I thank you for your time in allowing me to present my diverse view.
MR. BLAIR: Thank you, Dr. Arvary.
Dr. Rose.
Agenda Item: Users of PMRI - Jeffrey Rose, M.D., Kaiser Permanente, Colorado Region
DR. ROSE: While the projector is warming up I must tell you that Mike Fitzmaurice has kind enough to send me 39 of the hardest questions in medical informatics, with the idea that I would answer them in 20 minutes. So I'll do the best I can. They were very difficult questions that you have asked.
DR. FITZMAURICE: You're going to have blame Jeff Blair and Simon Cohn and the rest of the committee for the difficult questions too.
DR. ROSE: Is that right?
DR. FITZMAURICE: Yes.
DR. ROSE: I'll give them all credit.
MR. BLAIR: I don't mind at all if there are certain things where you feel they are more meaningful and relevant, and you pick and choose that you want to include in your testimony. I also think it might be helpful since you are in a unique situation that you have created Kaiser Rocky Mountain Division, maybe some of the other committee members would need to be familiar with the basis for your testimony and recommendations based on the environment you have created there.
DR. ROSE: That's exactly what I intend to do, Jeff. I'm the director of clinical information systems in Colorado. If you need to know more about my CV, you can go to the Website that I have listed, where I keep that updated. But I'm not going to into my own personal characteristics at this time.
What I wanted to do was to tell you what I see as the main problems facing health care now; how we solved those with the system we have; what our system does; what it's like; and how that is based upon standards. Then I want to specifically answer some of the other questions very briefly that we were asked. Then you'll have plenty of time to ask me more hard questions, which I'm sure you'll think of.
Here is the problem. Between half and two-thirds of the potentially controllable rising costs in health care are due to increases in the volume and intensity of medical services or the content of care that is completely determined by clinical decisions, not by salaries, not by administration, not by research, not by Y2K, and not by fraud and abuse. This is an enormous number. I can send you a lot of references for that fact if you are interested.
We cannot as clinicians, begin to get a handle on any of these costs without adequate information tools, which of course exist today. Einstein said, and I agree, "We can't solve problems by using the same kind of thinking we used when we created them," and it seems to me that's what we've been doing over and over and over again in health care.
I agree with Dr. Arvary that we practice in an environment that I call informaciation. Unfortunately, there is not enough time to draw forth this metaphor for you, but I liken it to a labyrinth from which no one can escape. The informaciation consists of three things: inadequate patient information when we're making clinical decisions. This is largely because of the paper medical record.
Inadequate evidence to guide diagnostic and treatment choices, and to maximize effectiveness, which has already been brought up by my colleague. And something that people often don't think of, which is paradoxical data toxicity. That is an overload of redundant, inaccurate, uninformative, or confusing "facts" that lead us to incorrect conclusions.
Now it is largely because of these rises in health care costs, and the paucity of reliable clinical information that we have seen utilization management techniques of very questionable character arise. The point that I want to make is that price and cost control have no meaning without the determination of quality and value.
So what we have seen is an enormous myth. Managed care is not managed care. Let me feed back to you some of the HCFA actuarial estimates. The average rate of health care spending growth has been 5 percent since 1993. It will be 6.5 percent between 1998 and 2001, and 7.5 percent between 2002 and 2007.
What that means is that fueled mainly by the private sector, expenses are going to increase to $2.13 trillion dollars, which is nearly 17 percent of our gross domestic product by the year 2007. I would submit we can't afford that, and it also is unnecessary.
I like to put things in Latin, so I will say obesa cantavit(?) as far as managed care, which basically means the fat lady has sung. We are managing costs, not care. If we are going to manage care, then we need information, and we need information technology to help us do that.
I'm going to talk a little bit about where we need to be. This talk is derived from one called Ariadne's thread. Maybe you remember that Ariadne gave the hero Theseus a thread to unwind as he found his way to the Minotaur, so we could find his way back out of the labyrinth. So based on this metaphor, I have drawn what I think clinicians need as their Ariadne's thread.
Notice that it's not an electronic health record. It is something more complex than that. Dr. Arvary already implied how complex this is. It is a process facilitation, meaning that it makes the process of care more efficient; real time, meaning that it happens as we practice; education, meaning that it gives us alerts and guidelines, and also is an educational tool for our patients, as well as for us.
Physicians simply cannot remember. We're no better than anyone else at remembering the huge amounts of information that are coming at us all the time, and we need something that will assist us and support our decision-making.
And finally, it's a documentation tool, because it is the legal record of care. It's the way in which we communicate with people, including the patient ultimately, about the care that they are getting. So the way to remember this is quite easy. The P is silent. This is PFRED.
Here is what we wanted to create in Colorado, and spent I would say eight years built. We built Ariadne's thread, and I'm going to talk about what we built, and how well it is working. Ariadne's thread connects clinical processes, documentation, guidelines, and reminders. It manages a coded vocabulary. We chose SNOMED. We also chose the First Data Bank NDC vocabulary, because SNOMED was actually fairly weak in the area of medications. And also the medications tend to change far more rapidly than most of the other substance and terms in SNOMED.
We adhered to standards, which was risky. We picked HL7 and DCPIP, which was a very fortunate decision, as was the SNOMED decision in my opinion back in 1992. The system that manages this vocabulary allows us to unobtrusively collect clinically accurate comparable data. This is key, clinically accurate comparable data.
It allows an evolutionary path our health enterprises medicine changes. Medicine is changing almost as fast as technology is changing. For those of you who wonder how fast that is, it's about 10 million times the rate of biological evolution. It is scalable. It's open in architecture. It's amend to emerging Web technology. It is patient-centered, and flexible across provider specialties and patient populations.
Here is what we have. We have an ambulatory electronic health record/PFRED in 25 facilities: 15 medical offices, 4 mental health offices, 3 administrative offices, 1 emergency room, and 3 hospitals. It is an ambulatory system. We have workstations in the hospitals for continuity of care purposes, but we have not interfaced with hospital system per se.
The total population of physicians using this is about 500 doctors and 100 students and residents who come through. We have a total of 3,000 users of this system. This is not a physician workstation. It's a clinical information system. It's a clinical workstation. It supports team care. Anyone who needs to look at or use the medical record is equally capable of doing so in this system within parameters that we set in terms of authorization and security.
There are 3,400 workstations. These are PCs. I can talk about what kind of PCs they are. They are in every exam room, and they are in every office, and they are every place. They are pervasive wherever the medical record is needed. There are 175 printers. It's interior client server architecture with distributed replicated data, and I can describe that for those of you who would like to more know about that later.
But the point is to make sure if any server goes down you can still get to the patient's record, because remember, we are now totally paperless. Medical records have gone to the graveyard for paper medical records. We no longer have them in our clinics. They have all been sent away. Now you get them if you want them, but all of the work of every provider of all of those 3,000 people I mentioned is now done completely online.
All of the network components of course are very complex, and everything is installed. We have the charts for all the Rocky Mountain Division members on these servers in a distributed replicated fashion. It's about 350,000 active charts. Of course there are over 1 million charts, because just because a patient comes and goes, doesn't mean that you get rid of their medical record. So right now there are well over 1 million medical records in the system.
This is an example of the ancillary traffic that runs on one server. We do all of our ordering, all of our results review, all of our desktop medicine on the same units, and in the context of the patient's chart. So you can see the rather vast numbers of signals that have to go back and forth in an HL7 format to support our practice.
And all of this is managed actually from Boulder, which is about 45 miles north of Denver, so the whole system is actually managed remotely. It's really quite an interesting thing to see. And that is where the single point of contact help desk is.
This then is what a clinical information system is in our view, and where we should be headed as a country. We all have ancillary operational systems that need to be linked to what we call the clinical information system. The clinical information system consists of the four blocks that you see. Although people have pulled off one or two of these blocks in some cases, no one has achieved all four. The unfortunate thing is this is like firing a nerve. You either have a threshold and you have all four of these and you fire the nerve, or nothing happens.
The hardest part of this is to get all these things, the user interface, which is amenable to people who are relatively computer-phobic, meaning for the most part clinicians. It needs a standard medical terminology. We chose SNOMED. The electronic chart needs to be stored and transmitted, because the paper chart is only available about 60 percent of the time, even for appointed patients in the best of systems.
But the difficult thing is that all of this clinical information system process has to fit into the care delivery process moment to moment. This is a new tool. This is a revolution in health care. It is a revolution that we haven't seen since antisepsis. And it will be resisted with the same vigor, and that where the governmental role comes in.
The whole point of doing all of this is not only to improve the care process and the care information for the provider, but to collect as much comparable data as we can, so that we can begin to understand what it is we are doing in health care. About I think 80 percent of what we do is not evidence-based, and we need to change that. That's what results in a huge amount of waste.
Here is how the clinical system is structured. I will be happy to come back to this slide. I don't want to overstep my time, and I'm not even looking at my watch. But notice that everything is drawn from the top square, which is controlled medical terminology. We have a patient-centered medical record, which means all information on every patient, everything that is ordered, everything that is documented, everything that comes from ancillary systems gets back into that patient medical record.
It is a physician workstation. It is not a OB-GYN workstation or family practice chart, it is a patient medical record. We do allow some specific tailoring and personal preferences which allow different specialties to use the system in a flexible way without having information hidden from them.
They also do all of their work within the in basket. All of their signing of all of their results, sending copies of notes, all of their consultations are also done in the paperless world. And all of their orders and results are returned to them and to the chart via HL7 interfaces.
Two very important features of this that I would like you to notice. The first is that there is multi-provider access, which is critical. Multiple people need to be able to look at the chart at the same time, and in fact work in the chart at the same time. This is one reason why a lot of electronic record systems have failed.
The other is that there is the concept of lock up and transfer, which means that you don't have to log on or log off of the system as you move in and out of the exam room. You can simply lock up the workstation. Then when the physician comes in after the nurses record the information, they simply unlock the workstation, all of their personal preferences and formularies and lists and terms are loaded, but they are put right back into the note where the nurse left off. So it's a very natural flow and supports team care.
How did do or how are we doing? This is a chart that shows you the number as we began. We began actually our full roll out in November of last year. It shows you the number of progress notes that have been generated with coded information in them. I will give you more details about that by month.
And it shows the number of chart accesses by month. Every single chart access by the way is logged and audited. We know who has looked in the chart, not just who has done something in the chart.
I would like you to take a look at the number of times that clinicians accessed the medical record for information a month -- 880,000 times. Now if you compare that to the number of times they would have liked to have looked at information in the paper record that wasn't available, and start to get some idea of the magnitude of the problem that we are dealing with, and the problems that we have with the quality of care.
Here is a chart that shows you the percentage of visits which are actually charted completely within CIS. And it is almost 99 percent. I'll show you some other numbers. And the number of coded assessments -- these are SNOMED coded assessments -- is at approximately 84 percent. Now recognize that not every time you see a patient do you actually make an assessment. So this is an enormously high use.
In fact, I think it's safe to say everybody uses this system, everybody. There is no alternative. M.D.s, D.O.s, nurse practitioners, P.A.s, our physical therapists all use the system; 99.1 percent in the audit at the first clinic that went up after about six months, 99.1 percent of the visits were within CIS, and 89 percent of them had coded assessments. In addition, the allergy tabs were populated in almost 99 percent of the cases. The health risk tabs were populated in 95 percent, coded to clients in 46 percent.
Now the last number is very important and very pertinent to what Dr. Arvary, and that is that completely structured text is untenable. There is no way to do it. That's why there is some text or some dictation in almost 100 percent of the medical records. It's impossible to be completely expressive using only coded vocabulary. So we support both the entrance of free text and dictation in the system. Then below are listed the number of specialists.
Well, I hope you understand this system. I have some agreements and some disagreements with my colleague about the use of mobile computing and what's important, but we can discuss those and debate those during the question and answer session. I think what is most important, and what the clinicians want most is a system that allows them to practice efficiently, and gets them on information they need, not only on the patient, but about new ways or better ways of treating the patient.
Here are some of your questions, Mike. Is the current state of data quality, integrity, comparability, and accuracy impairing our ability to do valid outcome, quality, performance, and research? You can see what a simple question he asked. My answer is simpler, absolutely, profoundly.
What are the specific limitations? First and above all, vocabulary; code sets, as they relate to vocabulary. Billing requirements, which are based upon billing coding sets, which are inadequate in terms of documenting clinically relevant care.
There are many cultural limitations. Dr. Arvary referred to some of those. I can go into more detail about those, because I sort of like them, later.
Information sharing standards. How do we share information with each other electronically, and how do we represent the data? So this is a data representation and messaging issue in the electronic world.
Is the private sector making satisfactory progress? No.
Is there a role for government? I believe there is, particularly in supportive standards. I would suggest supporting a standard vocabulary which is clinically accurate, and it looks to me like SNOMED is the one that should be chosen as a reference terminology.
I would suggest the government mandate representation of the terms in some form that is Web-enabled such as XML, and I know there is a lot of activity going on, and that communications be standardized, HL7 and TCPIP being examples of what has enabled us to get where we are. I think there is a great role for government in saying this is the way we're going to do this.
I would like to see the government move away from these inadequate codes and toward clinically accurate terminology. Pay us for being clinically accurate.
How comparable, precise, complete, and accurate must the PMRI be? We have a saying in radiology, and I actually am a radiologist by training, and had to take the physics boards. And in having to do that, there is a term called ALARA which is referring to the amount of radiation that it is okay to zap somebody with. The ALARA means as low as reasonably acceptable. In other words, nobody would put their foot down as to the precise number of rads and renkins; as low as reasonably acceptable.
So I hereby introduce into the formal vocabulary AMARP, which is as much as reasonably possible with respect to this question. When we open a chart, we expect to get as much information as we possibly can about the patient. I agree that if we could get limited bit and pieces, it's better than nothing. But in terms of where our target it, it's as much as reasonably possible.
What are the consequences if don't achieve this kind of PMRI? I would like to state that there will ongoing, horrendous iatrogenic injuries. I'm not going to go into the details. You can read about these. They cost us billions of dollars a year. More people die as a result of adverse drug reactions for example, than die of cancer of the breast every year.
We have a problem that's an ugly, ugly secret. Now it doesn't need to remain a secret anymore in American health care, because there are solutions, and IT information systems are those solutions. It's time to bring these to the surface and deal with them directly.
We will find it almost impossible to rapid advance to evidence-based practice without systems such as these. The pace will continue to be the pace that it has been, which is just about one notch above zero. We will have ongoing and increasing waste, and ongoing and increased costs, which are already untenable.
We will have continuing inappropriate use of pharmaceuticals and technology, uncontrollable quality, and therefore uncontrollable expense. We will have uninformed decision-making. We will have damaging managed costs inflicted upon us even more so than they are now, rather than value and quality-based health management, which should be our goal.
What type of standard vocabulary coordination? I believe that we should converge to a single reference model, with common definitions and contextual meaning. This means the use of SNOMED rather than ICD or CPT for all reporting, all reimbursement, all research, and particularly in government funded programs. I think that we have to encourage the private sector to create application terminologies that are based on the reference terminologies.
The reference terminologies themselves are not usable by the end user, the clinician. The application terminologies, however, are. I would like to use a metaphor. What we need is a standard dictionary. If one group of providers wants to write a novel, and another group of providers wants to write a poem, and another group of programs wants to write a screenplay, so be it. But the words and the meanings of the words must be the same, so that we can derive the context for research and for outcomes.
I am not in favor of everybody doing things the same way. I'm very much in favor of collecting the same data in context. We cannot do that with ICD and CPT. They are not clinically accurate, complete. It is not accurate terminology for that purpose.
Now we of course can't just stop doing that. I think that would create tremendous hysteria. First of all, physicians aren't about to stop billing. But we can set a reasonable path and timeframe for migration from current clinically inadequate coding sets, which actually were never intended for clinical accuracy, to the more accurate one. I think that if that is mandated and there's a timeframe, we can move from the ICD and the CPT into much more clinical terminology.
Government versus private sector. The government absolutely has a role in standards and infrastructure, no doubt about it, and in terms of privacy. And please, exercise care when you are looking at this privacy legislation. Mike scared me to death the other day, and so I'm telling you, you will add enormous costs with some of the privacy legislation that is coming up and being paraded around.
You need a health record number. No one, please, no one accept the fact that that can be killed. Until we have a unique health record number, until we have one place where we can put all the information on a patient reliably, we will never get there. So we cannot allow that to die. That has to be carried forward, and we have to do it in a way that the public accepts.
Government, for culture and change management, to practice value rather than cost. This is critical. At the present time we reimburse and reward our clinicians for doing more, not for doing better. That has been a tradition in American health care forever. To change that is going to require guidance from the very top.
Only when we begin to reward practice value rather than simply the number of procedures done, or the amount of time spent, or the intensity, et cetera, et cetera, only when we get to value rather than cost, will we really get to where we have optimal quality health care. The private sector, on the other hand, can develop the applications for the use of the standard, of transmission and representation of clinical information.
Now there is a very popular among some circles and unpopular among other circles patient's bill of rights. I would like to introduce to the record the clinician's bill of rights, which is equally important:
To have access to any and all important about the health of individual patients, communities, and populations at any site, and at any setting is a right of a care provider. To be financial renumerated based on proof of value, rather than on the extent of intervention. To have access to information from a variety of sources to support ongoing real time education, evidence-based care, best practices, and highest attainable quality.
Now quality is a loaded word, so let me tell you what I mean by that. Quality, according to the JCAHO is the degree to which the process of care increases the probability of a desired patient outcome. I'm going to make it simpler. It is all interventions that have demonstrable positive impact on the health status of individuals and populations. No interventions that have a demonstrably negative impact on the health of individuals or populations. And no interventions that have no demonstrably positive impact on the health status of individuals and populations. When we achieve this, we will have achieved maximal quality in health care, and we will also be at the lowest possible cost.
Continuing the bill of rights: To collect, share, analyze, and use standard, reasonably secure, clinically accurate information about our patients for caregiving, billing, and research without disruption of the clinician-patient relationship. To engage in shared decision-making with patients based on evidence and experience. And to be guaranteed evidence-based legal and political protections for those judgments.
Now there are no rights without responsibilities. Here is just one slide of the clinician's bill of responsibilities:
To protect the privacy, confidentiality, and security of the information and the health of our patients. This is entirely possible technically and administrative today. It is far superior to the present safeguards and methods we have, and this is what people need to understand, and this is what the public needs to understand. The fears are tremendously trumped up in this regard. We can do this; we can protect this information.
We have the responsibility to practice in accordance with these evidence-based guidances and directives. Having the information alone will not be enough. We have to act on the information that we have. That's a responsibility.
And we have the responsibility to exercise judgment in the interest of highly quality medicine -- remember how I defined quality please -- regarding the health of patients and populations. And in so doing, manage those costs of care over which we have control. Two-thirds of the controllable costs I already mentioned, many billions of dollars. Finally, we have the responsibility to avail ourselves of information technology tools and standards to promote these practices of highest quality.
I want to conclude with a quote from the Book of Common Prayer, and ask you how far we have come from this statement. "We have done those things which we ought not to have done, and we have left undone those things which we ought to have done, and there is no health in us."
Thank you very much.
MR. BLAIR: Thank you, Dr. Arvary and Dr. Rose, both for informative testimony and humor in the testimony, which is well received.
Are there members of the committee that have questions?
DR. FITZMAURICE: I'll go, Jeff. A question for Dr. Arvary. A lot of your argument was posed upon the premise that the disposal income of doctors is declining, and therefore they don't have extra capital with which to invest in computer equipment and decision support equipment. But they might be encouraged to make the investment if it were subsidized, and they might be encouraged to give up their own data if it were anonymized. Have I described some of your argument correctly?
DR. ARVARY: There are a lot of issues involved in the utilization of the electronic medical record. Part of them are lack of expression. This is a well known document of fact. Nurses, clinicians, and physicians really have not been able to express themselves adequately, and they feel unsafe using the present desktop computer's fragmented statements, narrative text. Physicians don't type well. There are a lot reasons that they don't feel comfortable with that.
But the lack of disposal income is real. I was in private practice from 1983 to 1997. In 1994, I swore I would never go into a group. By 1997, I was helping to form a group. Part of the reason was that -- and one of the good things about managed care is it forced physicians to get together to cut costs. Administratively it just cost too much for a primary care physician in private practice to manage a managed care environment. So the disposable income does go very quickly. And apathy is out there in the private sector as far as electronic medical records at this time.
DR. FITZMAURICE: Now to follow-up, would physicians be more likely to adopt a pen-based system. You have done some experimentation with this. You have perhaps used it in your own practice. Maybe you have helped develop a pen-based system. Are there strong advantages to a pen-based system that you don't see in a desktop? One of them you mentioned was interaction with the patient. Perhaps it interferes with patient care. Another one is that perhaps it would take longer to make the notes for a physician using a desktop computer than a pen-based system.
DR. ARVARY: You don't have to take my anecdotal report. This is done in research -- Stanford, they have done it in the Netherlands, and in Japan. They have all shown with clinicians in clinical practice that utilizing a pen-based digitalized type system is faster than having a clinician type in narrative information. It's just been found.
Also it has been found at Stanford that after study of a lot of physicians using computer, physicians tend to be mobile. They like to move around. Even if they don't normally, they like to have that ability. Most physicians can't afford a desktop computer wherever they feel like writing a note.
DR. FITZMAURICE: If we can make the leap then from a pen-based system to data that are usable for research, for quality assurance, as well as for billing -- it has to be billing. That's the number one.
DR. ARVARY: I think that you can put as much structured information in a note with a pen-based computer as you can with a desktop computer. The question is what does the federal government, insurers, and the academic institutions want from us structured? They haven't said it.
I agree with Dr. Rose that we definitely need a codified medical vocabulary that is standardized. I don't know if SNOMED is the answer. Maybe it's something else that is in development, but we do need a codified medical vocabulary that is standardized, that does not change unless everybody changes it at the same time.
But I do believe that it's very easy at this point in time now through either a desktop environment or a pen-based environment to give ICD or CPT or whatever reference terminology you want to use, can go through a pen-based computer. You can use one level of modifier complexity with a desktop computer using check boxes and templates, just as you can with a box.
My contention is that you can add unstructured information a lot quicker with a pen-based computer, and this information is important only to the physician who is doing it. So he recognizes that as an accurate expression of his interaction with the patient, including the language of a patient and himself, a very important point in documentation of record.
DR. FITZMAURICE: Let me ask Dr. Rose, how do you stand on physician data entry? That is, how do physicians in your system enter data? And how easy is it to get them to adopt a common medical terminology?
DR. ROSE: Well, the answer to the first question is that they enter information largely derived from a controlled medical terminology that is actually quite natural for them to use. If you ask the majority of our clinicians when they are entering coded information if the information is coded, they have no idea. So it's a very natural use of the coded information.
They would disagree I think in large part with the idea of mobile computing, so let me present another view. I don't disagree that most clinicians may not be able to afford desktops. I don't that they can afford laptops or expensive mobile equipment either, which is easily dropped and stolen, and has its own set of problems.
We actually have a full PC workstation in our small exam rooms. When we finally found laptops that would run this program, and it's an enormous program, and put laptops in some of the exam rooms, the doctors and nurses did not like them, and insisted that they be removed and replaced with the PCs, the reason being that they interact with the keyboard, the interact with the mouse, all of them. And they interact actually with the PC to do their dictation, even if it's a matter of pushing a button, and then dictating directly into the PC.
So that's the way they interact. They like to see a lot on the screen, and they would prefer a lot of functionality and the ability to look at a lot of information over the portability. That's been my impression.
Let me just add to that, I've seen the studies in the Netherlands and Japan. I don't know exactly what their programs are. I think you really have to know what the program is that is in the exam room. You also have to know how the clinician uses the equipment. There is a good way to use computers in the exam room with the patient. And we have found by and large our patients absolutely loved this, which is a contradiction to the studies that have been published. And there is a very bad way to use it, in which case the patients react very negatively.
DR. FITZMAURICE: Just a follow-up, does it impact on productivity?
DR. ROSE: No, our clinicians are equally productive at this point. In fact, those who have really become facile at using the keyboard commands instead of using the mouse are seeing patients once every 10 minutes. They are getting out at the same time or earlier. In fact I agree with Dr. Arvary, it's just as easy to enter a rapid, legible, structured, codified note in our system as it is to write a sloppy, illegible one in the chart.
We do not force our clinicians to finish their charting in the exam room, although they certainly can. I can't give you an exact number, but I would say very many, probably almost half, finish all of their charting before they leave the room. This seems to be a matter of style. It's the same in the paper world too. A lot of doctors finish their charting over lunch, and they have stacks and stacks of these things. Others will finish all of their charting while they are in with the patient.
MR. BLAIR: Dr. Cohn?
DR. COHN: First of all, I want to thank our speakers I think for some very interesting presentations, and Dr. Arvary, thank you very much. I especially want to thank Dr. Rose, who is one of my colleagues, and I really appreciate the education.
Dr. Rose, you had made some comments about application terminology and reference terminology. This Work Group has some previously about the concept of reference terminology. How do you see them working together? What do you think this Work Group should be recommending in relationship to application terminology, reference terminology, et cetera to really get to comparable data?
DR. ROSE: I think the major role and the major recommendation ought to be around reference terminology. And that the private sector, the people creating the electronic health record, or the PFRED that I talked about are the ones who are responsible for new and creative interactive techniques that utilize either through mapping or directly that reference terminology.
And there is a big different. For example, SNOMED needed a lot of work for us to get it ready to be able to use it as an application terminology. That's what we spent a long time doing.
MR. BLAIR: As an application --
DR. ROSE: As an application terminology. It is both. I think we are one of the very few applications out there that uses the terms directly within our application. But there certainly needs to be some standard to which any kind of application terminology or billing terminology actually gets mapped.
An example is, the reason we chose SNOMED is it is a relatively atomic vocabulary. The terms are well structure, well thought out. They are in good taxonomies. They can be managed. There is a process for updating them, and for managing them. And the atomicity of those terms is very important. The more you combine the terms, the more fuzzy the terms become, and the more difficult they become from a reference terminology standpoint.
This is the problem with ICD for example, where you have diabetes with nephropathy, but without retinopathy, and no foot ulcers. That's like a thousand terms in one statement that has a single code. That's not useful. On the other hand, you can choose those terms individually and catenate them in SNOMED, and actually store them and have individual terms stored.
DR. COHN: Can I ask one other question around terminology. I think you made an impassioned plea for the migration from administrative to clinical terminologies. I just have a question really for both of you, since you are both familiar with the ICD, CPT, as well as I think the needs of clinicians for more expressivity.
One of the things that I have observed personally as I have looked at administrative terminology is that there are problems with administrative terminologies. Then there are a lot of problems that have to do with billing policy and payment policy, and the fact that certain things get paid for and certain things don't. I think we should see -- we probably should not say it in this room -- that sometimes that influences how a final code gets assigned.
Now in a sense we have protected the clinical record from that perhaps the term is perversion of the codes. If we dispense with any sort of a middle layer there, how are we to protect our clinical terminology from that sort of once again, perversion, or is that a concern in either of your views?
DR. ROSE: I think that the perversion has occurred because of the billing system. It has not occurred because of our desire to be clinically accurate in the medical record, and I think that's where the problem is. I think we should be able to do things once accurately. That is the information that should then be used for billing. That gets rid of all the perversion in terms of the billing as well.
The problem then is the ownership is back to the payer. Are we going to pay for a SNOMED code, or a catcatonated SNOMED code? If so, how? But I don't think that's nearly as big a challenge as it seems right now.
There is no chance of up billing or up coding. Well, there is always a chance of doing that in the record if you are going to be unethical. You can do it with clinically accurate terms as much as you can do it with the billing terms, so let me retract that statement. But I think the likelihood of that happening is much less.
DR. COHN: Dr. Arvary?
DR. ARVARY: As far as SNOMED goes, I think Dr. Rose can talk more than any of us, because very few have licensed SNOMED, particularly for an application terminology. I could only find it with pathologists, quite frankly.
But as far as the CPT-ICD situation goes, part of the problem that we have, and I do this when I enter from the exam room, if even if the physicians are the ones who do the ICD-CPT coding, it will be 100 percent more accurate than what we have now. Because once you send a route slip to front desk and a receptionist cannot find -- no, does not want to look for that code, they end up giving a wrong code. They give a code that they are familiar with. This is common place. It's well known that physicians do a much better job when they do the entry themselves even with the present ICD-CPT coding system.
I think the problem is with the electronic medical records devices that the search engines just aren't strong enough. I think that if a physician wants to see diabetic nephropathy with neuropathy, he should be able to do that using a search engine that is adequate. Most of the diskettes that come with electronic medical records, or the programs that have been set up really have poor search engines. So you are forced to get a few diagnostic codes. It really doesn't search for you using your language or expression of what you want.
DR. ROSE: I agree, that's a limitation of the applications that have been developed to date. When you overcome that limitation, people seem to be, in my experience, quite happy using codified information. In fact, they don't even know they are using it.
MR. BLAIR: Other questions?
DR. FERRANS: I wanted to thank both of the panelists. I think those were excellent presentations. You had mentioned something about telecommunication costs, and I agree. One thing that we really haven't talked to is communications costs are certainly a hidden part of the total cost of ownership that people need to look at when they make decisions.
I would like to get, if you will, to sort of simply things, the outpatient clinician version, and the health system version of if you take the factors of what I think are the barriers to having comparable patient medical record information, or a computerized patient record application, those barriers being cost, user interface, and lack of standards, how would you weigh each one of those as to why people aren't getting those systems now?
And I want to add the caveat that I think certainly you all have done an excellent job, and clearly both of you were early adopters. So how are those barriers to the people who fall in the middle of the bell shaped curve. What's it going to take for them to start investing in the system so we can have everyone with the data?
DR. ROSE: First and foremost, there has to be something in it for them that is directly and immediately tangible. If it's something that takes longer or requires pain to use, it is not going to happen. So there has to be an application first and foremost which people can be taught, which we take adequate time to teach them, which we reward them for, which is in and of itself rewarding.
In fact, in our system I believe that what has driven the acceptance of this system, we did a good job with training, we did a good job with change management, but the most powerful change management in the world is when the complex patient comes in on Saturday morning when you already have a full schedule, and you know this patient was seen two days ago or last night in the emergency room, and you know it's going to take you all afternoon to figure out what happened.
You say, well, I'll try CIS. You pull it up and there is the note. It takes five minutes to do a much better job with the patient the second there is that realization your change issues are gone. Doctors want the information. So that's number one.
Number two, in terms of cost and the infrastructure, I think we have an enormous opportunity now with the Internet, and with the new next generation of Internet that is being developed with potential of actually using thin client applications, not only to get over most of the communication cost problems, but to deal with the privacy problems, and to deal with the uniform representation of the data problems.
If we can get doctors thin client machines that cost less, and we give them a very familiar way to interact with the patient record, just as we do with the Internet, we have solved all of the problems basically that you have mentioned, except for the cultural ones.
DR. FERRANS: As just a quick follow-up to you, before you answer, do you think that application service provider model is going to come into play?
DR. ROSE: Yes, I think there aren't many systems like Kaiser Permanente out there. So either there will be more of those formed, and they take advantage of these kinds of client server systems, which I think is unlikely. Or people will subscribe to something and people will keep their medical records on a secure -- I don't know if the government will house it, or some entrepreneur will house the medical records.
But the point is they will all be represented in the same way, and if you are an authorized user, you can see the patient's medical record.
DR. ARVARY: I would answer your question up front, all of the above. You can't isolate one. If you throw a lot of money at physicians and give them something they don't want to use, they are not going to use it, because they are very in tune to processes, office processes. If you can show that there is something of value to them to run the office better, to handle their patients better, to empower them.
If a physician is standing in front of a patient who has a disease he is not familiar with, and is able to get instant Internet access and go through six or seven abstracts in seconds to have the information that he or she needs, that empowers him. That's a revelation. If his secretary is able to take a note, and eliminate getting up, putting a note on a chart, putting another pile of things next to him and creating that tension that occurs when 20 or 30 charts are piled next to you constantly, if they eliminate stress in the office, if they reduce traffic, and if they reduce chart pulls, that's something of value.
Unfortunately, now that's not enough. Money is necessary because of the environment that I talked about with Dr. Fitzmaurice. We need a starter. We need good money now, instead of bad money now and good money after bad later to get it started. I think that a regionalized information network would be the most appropriate, since medicine is culturally, insurance-wise, and referral-wise regional.
DR. ROSE: If I could tag onto that. I agree with the stress issue, but I think there is also another underlying stress issue which clinicians deal with all the time that they are not even aware of until it goes away, and that is they are making difficult decisions that are highly impactful without enough information, and they are used to doing it.
So they live at a stress level that they don't even realize they are living at. And you can feel a palpable drop in that stress level when you work with a group of people that have all the information on all the patients all the time. That's another piece of something that's in it for the clinician.
DR. FERRANS: I just want to close your comments by saying I think you all have really crystallized what in my mind, is the value of the CPR. I'm very glad we're having these hearings today. I think it's really important that we get this perspective.
MR. BLAIR: Are there other individuals with questions? I have one if there is not.
DR. FITZMAURICE: I have one, but you may go first, Jeff.
MR. BLAIR: Dr. Rose, we have had many people who have testified to us that the path to the future which would involve technology helping the process of clinical care enter the information age -- I'm going to phrase it that way -- needs standards. Among the standards we often mentioned are HL7 and the SNOMED reference terminology. And among the drug knowledge base ones First Data Bank has been mentioned.
You mentioned those three. Let me ask you thoughts on a few things. SNOMED reference terminology right now is just beginning beta test site for commercial vendors. Thirty vendors have signed up. It will probably be a year or two or three before we see many vendor systems that are able to utilize that clinically specific medical terminology.
When we look at HL7, they are going to Version 3, which is based on their reference information model, which will enable them to update HL7, make it more useable. They are using use cases to make it more appropriate. It will probably be a year or two or three.
When we look at drug knowledge bases we wind up looking at attempts by standards organizations to develop a common, open drug knowledge base. And you had to go forward before all of these were in place. You did a lot of this work yourself. In fact, my understanding is that Kaiser has turned at least the SNOMED-RT information over to the College of American Pathologists so that they could share it with the rest of the world.
My question to you is in light of the fact that -- let me add one other last piece. I'm dumping a lot on you here. So there are more pieces to this now. Another concern is that some people feel like SNOMED-RT is proprietary because they have a fee, and First Data Bank is proprietary. HL7 is not. There is not a fee, although you may have to become a member in any event.
What advice or wisdom do you have for us in terms of the role of the government to facilitate, encourage, help people, because clearly not all these things are available for vendors and users to use today.
DR. ROSE: Well, that's a good question, Jeff. The point is that right now we're using inadequate systems today. And hopefully five years from now we will be using really good clinically accurate, standard, well developed systems. I am not under the illusion that this is going to happen over night.
But I think that if we fix in stone that we are going to stick with ICD and CPT for example, and mandate that for the future, we have taken a tremendous step backwards. There is no reason that we should not, as a government, and as clinicians in general, proceed on the path to get to SNOMED-RT.
It may take SNOMED-RT two to three years to become a fantastic reference terminology, but in point of fact we are using it now, and there are application vendors who will in the very near future, be able to serve up those vocabularies based upon SNOMED. I know of at least two companies based upon SNOMED that will allow any vendor to "lex-enable" their application.
So I'm very hopeful that we can get to a standard. Now SNOMED is going to take a long time to finish, and it will be constantly an evolution. That's just the way it. Vocabulary evolves, new diseases appear, things change, and there needs to be a lot more work done on SNOMED even than we have done on it to make it a complete vocabulary, but we should try, and we should begin.
HL7 is a wonderful effort. I'm glad it's in the public domain, that's wonderful. But I don't think we can expect that after all the hard work and the money that's been expended on SNOMED and some of the other vocabularies, for that to happen. So I'm okay with people needing to license that kind of vocabulary.
I think the one that you did not talk about is the representation of the data. I think both HL7 and ASTM are working on representation of medical data. I try and keep up with what Margaret already knows. I read her articles. I think XML may well become that standard. Whatever it is, let's pick it and then let's move toward it. Let's not let the amount of time that may lapse between now and getting to the ideal cement into the inadequate present.
So let's see, what did I miss? We hit HL7. We hit SNOMED.
MR. BLAIR: First Data Bank.
DR. ROSE: Oh, First Data Bank. That's a very interesting one. I think that's going to be a very difficult one to get, but I'm not uncomfortable with that. I don't think that has to be an open, non-proprietary thing. I think that's almost an impossibility because of the nature of the pharmaceutical industry in this country.
However, First Data Bank now owns both the major vocabularies that have been used, and it is their intention, and the government should encourage them to proceed with the project they are doing, which is to merge the Medispan and the First Data Bank vocabularies into a single one, or at least to map them to each other. And they already have that particular process in mind.
The reason that is difficult, I think I pointed out earlier is that the medications and the medication names and brands and doses and generics change so rapidly that I think it would impossible to have any single organization actually try and manage that.
MR. BLAIR: Let me follow-up with one other question. We had an individual, Tim McNamara from Multum testify. It was very compelling testimony, indicating that they were willing to make their drug knowledge base available without charge on the Web, and that it had characteristics of a clinically-specific medical terminology a la similar ontological principles you find in SNOMED-RT. We hoped to get someone from First Data Bank here too. You picked First Data Bank.
DR. ROSE: No, actually we picked Medispan, because that was the vocabulary the American Hospital Formulary Association, through Medispan, because that was our pharmacy system. We did not pick First Data Bank. At this point now that First Data Bank has purchased that vocabulary somehow or they have agreed to merge, we lucked out big time. It was another example.
For instance to mandate NDC, which I read somewhere, would be a horrible mistake. NDC is not that good. That's not the kind of thing -- I don't know all of the details about Multum. I do know where it was grown, which is my home state, and I know Bob Schrier(?) very well, who was the person who started the company, but I don't know if that particular vocabulary is one that would meet the clinical or clinicians needs.
However, the important thing about Multum and about First Data Bank is that both of them have the capacity, because they have the standardized vocabulary, to shoot off alerts and drug-drug interactions, and reminders, and drug allergies, and we need these. We hurt people as a result of this. So what Multum has is an example of what you can do if you are willing to work with a fixed vocabulary. Whichever one it turns out to be, so be it.
MR. BLAIR: Michael, you had another question.
DR. FITZMAURICE: I had a question. Jeff raised a problem about privacy and the cost of privacy. Gary posed an information and infrastructure solution perhaps, and I think Jeff also picked up on it. It was the notion of a thin client.
I can imagine there is a doctor's office seeing patients. The medical record is not on the premises. The medical record is connected with a T1 line, a phone line, whatever to a trusted repository. It is trusted to keep the patient care data private, to keep the line secure. It is trusted to return it back to the physician or physician's office whenever it is called for. Maybe you call for it the day before the appointments for all the patients coming in.
It is trusted to be able to strip identifying information off if the provider and the patient say yes, it can be used for things like research. Or if there is a public health need that overweighs that and is permissible by law.
In the department, we are working on a health privacy regulation. It is part of HIPAA, mandated. People look at it and say, well, we're not sure that it adds to our bottom line. But if the push toward a thin client model, as you described, can provide the security, then it might also provide some cost savings. Not that privacy and confidentiality will contribute to revenue without increasing costs. It may be a new way of practicing.
It comes down to Johnny Carson said many years, "Whom do you trust?" You remember that quiz show with Ed McMahon. So I ask both of you, whom would you trust to maintain a repository of patient information, and to provide the integrity and the rapid response that you need to take care of real live patients. Let me start with Gary.
DR. ARVARY: I agree with Jeff when he said that there is adequate knowledge and technology now to maintain adequate security. As in with the paper record, the place where all these records are going to disappear to are not by a hacker, not with alligator clips, not by somebody getting into the central repository, but out the front door, which is where they go now with the full expressed consent of the patient, under duress, to their employer for fear of loss of employer, to their insurer, for fear of loss of insurance, or to an attorney who is getting information from the chart that he has to no business to, because it has nothing to do with his case.
Unless the government starts to enforce laws that limit the amount of information that these agents can get from a chart, even with full expressed consent, all the security in the world is not going to stop charts from walking out the front door.
DR. FITZMAURICE: And you would say then physicians would not want to trust their records to a repository that is remote from their offices until that problem is handled?
DR. ARVARY: That's not what I said. I think they would be perfectly safe in a repository. I think that they are lost because the patients want them released because it is requested of them from their insurance carrier, from their employer. Whenever a patient comes in who is ill, and if an employer says I want to know why you were sick. I don't want a note just saying you sick from your doctor. You tell me what was wrong. You release that information to me, or you will lose your job. As far as I know, that's against the patient-physician trust.
And likewise, insurance carriers get information from patients with their full expressed consent that patients I believe, get it under duress, because they are afraid of losing their insurance.
DR. FITZMAURICE: My question was whom would the doctor trust with keeping the repository of records? Would they trust XYZ Company? Would the doctor trust the state health department?
DR. ARVARY: As I mentioned before, I think, and this is just a vision statement, because I need a vision statement to start with, I think that medical groups who already have to establish Internet networks should be the ones to handle a data repository. I think that IPAs would be fine as well. I think that it would have to be done by the physicians who have something to lose if that information is lost.
DR. FITZMAURICE: Dr. Rose?
DR. ROSE: I think you came perilously close to asking whether we trusted the government there.
DR. FITZMAURICE: No, I was not promoting the government as a repository. I was posing that as maybe one end of a member of society, and then the other end is a private company who says I will do this.
DR. ROSE: I think if a private company is subject to the same legislation that the government and everybody else is subject to, the private company may well do a better job than the government. I don't know. I think the government is going to have its hands full keeping its own repositories on all of the work that it does as a government-base entity, keeping those things secure.
But I think there are lots of examples in industry where privacy and confidentiality and quality and everything else is very highly upheld, because the companies assume the risk of loss of data, or for breach of data. So to me, it doesn't matter really where the repository is, as much as what are the rules around the repository.
MR. BLAIR: Richard?
DR. FERRANS: I have another question. Actually, I wanted to address what Mike said, because I think it's very interesting. I think that there can be trusted third parties that are for-profit. It's not whether they are for-profit or not, or whether they are private or public. It's the rules with what they actually do with the data.
I would refer back to Claudia Tessier's testimony in one of our last sessions as some of the troubling things that are going on out here. I know that information can go out that front door. It's all those stops in between on the back sort of claims processing end, and all the intermediaries or the infomediatries the patients don't know about that are very concerning.
Dr. Rose, we have heard a lot of testimony about SNOMED. And it has all been on the theoretical construct of why this is a better, smarter, faster vocabulary. And from an informatics perspective, I certainly buy a lot of those arguments. What I'm really to hear from you, since you all are actually using the vocabulary, could you sum up why SNOMED is helping you deliver better health care, that using other vocabularies wouldn't let you do?
DR. ROSE: Well, no, I couldn't answer that question. That's a very insightful question. I don't know that using that specific vocabulary is what allows me to deliver care moment by moment. But I do know that it allows me to gather accurate, comparable information that down the line is going to be very valuable, and very important to me in launching alerts and reminders, and doing outcome analysis, and doing meta-analysis of the data.
So what we were looking for was the richest, most expressive, most natural and complete, unambiguous vocabulary we could find at the time, and this was 1991. And the one that we could find that seemed the most atomic and unambiguous and all of those other problems was SNOMED. Quite frankly, it wasn't very good, and we spent eight years working on it.
Now let me give you a couple of examples. The pathologist will have malignant neoplasm in SNOMED. When we gave this to the users, and the users said I want to be able to say cancer, we suddenly realized that the term "cancer" was not in SNOMED, nor was the term "fever." You could pick elevated body temperature, et cetera, et cetera.
So it needed a lot of work and a lot of modeling, and a lot of enrichment, which is what we have done. A lot of creation of synomomies so that people can use the term they wanted, but it still mapped to the same concept. It's not easy work to do, but that's work that we had to do to have something functional for our users, and I think we learned a lot from it.
So it allows us to order quickly, but we could use another coding scheme to order quickly. It allows us to express our assessments, our complaints, and in many cases our physical examination or other historical items in a comparable way.
We also developed the things that Dr. Arvary talked about, which are very powerful search and display engines that allow you from any place you are within the vocabulary, to get the term that you want. If you don't do that, you wind up with people just texting in information, and then you lose data.
So you know I'm giving you a yes or a no. It's kind of a lousy answer.
DR. FERRANS: I think what I also hear you saying is that five years from now you would be able to give me an answer as you build in all that other functionality, and look at outcomes. Those are the things you have planned I take it?
DR. ROSE: Hopefully, it won't be five years. We are collecting huge amounts of data right now. If you look at just in the one year since we went up, there are 1,000,300 progress notes with coded assessments. All of that data is codified. Everything that is ordered is codified, everything.
So when that goes to the data warehouse after some period of time, you can begin to analyze it. You can do the data mining, and you can begin to fire the rules and the logic. So, yes, there is a delay for me to be able to prove that. But until I start putting really good stuff in, I have no absolutely no possibility of getting good stuff out. So that's where we chose to start.
DR. FERRANS: Thank you.
MR. BLAIR: Other questions? Let me just give one other opportunity to both Dr. Arvary and Dr. Rose. Is there anything else that you feel the committee should be aware of as we go forward trying to understand what recommendations we should make for uniform data standards that will facilitate patient medical record information, and the use of that information before we close?
DR. ROSE: Jeff, I'm kind of a simple guy. I don't think it's that hard of a problem. I think the recommendations need to be for a controlled, common medical terminology of some kind. For a standard way of representing that data, which is going to be at least amenable to Web browsing. And of course we all have the problems of keeping up with technology. And for a standard way of communication information securely.
I think those are the standards the government belong in that we can't do it without. And I think those are absolutely critical to us. Then having the basic infrastructure that Dr. Arvary talked about to begin managing the change, managing the culture, getting control of things, providing applications either by subscription service or other ways that are of value to our patients, and are of value to our clinicians.
Our patients are on the Web. Our patients are coming in with good information, and they are coming in with bad information. And we are literally powerless unless we are enabled with the same sort of IT, and all the information about the patient.
Without those standards, the example being the Internet. There would be no Internet without TCPIP. The only thing they needed to do was say we're going to use some standard mark-up language and TCPIP, and look what happened. If we do the same thing in health care and back off, the industry will take it forward as long as they know what standards to go for.
You look at the standards in the electrical industry. I can plug in a toaster, or I can plug in a projector, or I can plug in a computer, but by golly, it's the same plug and the same prongs. That's the point. The innovation can occur in many different ways. The government helps through the standards organizations that already exist, to push just a few of the really core issues, and then let industry take over.
MR. BLAIR: I have one more question, Dr. Rose, that I think it would be helpful for the committee to understand. What you did at Kaiser Rocky Mountain Division was more than just a technological change, and it was more than a functional change. It probably has the breadth of even being a cultural change, because from what you just told us, you have 500 physicians, 3,000 folks that have made a complete transition to using electronic health records all times, three hospital, I can't even remember the number of clinics, an entire community that's on this virtually.
And it took you eight years to do this. It doesn't matter how many dollars, but obviously it was expensive. You probably had to help these people navigate this change. I think it might be helpful for the committee to understand what it took for this community of caregivers to adapt to this new environment.
DR. ROSE: I think that's a great question. We recognized from the outset that not only did we have a technical issue that we had to deal with, but we had to understand our business, and we had to understand our capabilities, and that's what took eight years.
Once you understand that and develop something that is of immediate and intrinsic value, then you are faced with a purely cultural change management issue. And at that point I think it is important for the committee to understand that we cut our clinicians' schedules back by half for a month after we trained them. They had three and a half days worth of training. We literally paid them to participate in the design, and then we paid them to learn the system when you think about it, because we back filled. We couldn't afford to cut half the appointments and let the patients take the hit.
So we were extremely careful about the change management, and what it would take to get the doctors over what I call the energy of activation, so they could slide down that curve and stay there at the endpoint. That is not going to happen in the private world without some help.
I think Dr. Arvary made the point, and on my own I should have jumped on that. He is right, there has got to be in some way, some not only immediate reward for using the system, but some way to compensate the clinician who is will to embrace the new technology, knowing that the pay off down the line is in the billions of dollars.
There is no way that the busy clinician is going to stop seeing a third of their patients. They can't and continue on, unless we somehow subsidize that or reward that or do something. That be yet another area where the government, maybe only the government is potentially able to play a role.
If you use an electronic medical record system, a lot of malpractice carriers are lowering your premiums. Similarly, I think if you are using SNOMED, and you are using whatever is decided upon as standards, and your information is complete and legible and accurate, and you are following guidelines and have access to this information, perhaps there should be some reward in that financially for people who put forth that effort. But the training and the change management is huge.
MR. BLAIR: Thank you.
DR. ARVARY: The only thing that I can leave you with as I have completed here today is that it is definitely a technological marvel what you done out at Kaiser in Colorado. However, in the northeast we are caught in change history. We have no big employer that hires a lot of physicians. Physicians make a lot of individual decisions about their offices.
If any standards are to be developed, in fact the majority of the physicians in this country are part of that cottage industry, and make individual decisions based on what processes in their offices are going to be improved. If all I can leave you with is this, it will be enough. Whatever standard is devised, whether it is a limited data dictionary, whether it's a transfer standard, anything at all involving a medical record, keep in mind how it is going to affect the micro environment of the medical office. If that's all I leave you with, then I've done my job.
MR. BLAIR: Thank you both very, very much. Can we adjourn and reconvene in 15 minutes? Thank you.
[Brief recess.]
MR. BLAIR: I wanted to mention a couple of things here. I know that people are sometimes available on one day and not another. Our next topic is to go through and update the progress letter to the secretary of the DHHS. Michael Fitzmaurice will lead us through that.
I would like to bring your attention to a few other items. We are planning tomorrow morning to go through the timelines and pick a date in late January or February for when we will have a meeting for our Work Group to decide on our recommendations. That's an important meeting, so if there is anybody that won't be here tomorrow morning, please get with me so that I know some idea of your availability or non-availability for late January or February.
Lastly, there is just a piece of information. There was a survey that my organization happened to have conducted, and a number of folks were saying that they thought the committee would be interested in seeing the results of that survey. That is in your handout. We are not going to be reviewing that in our meeting. So it's just for informational purposes.
I think on that note, Michael and Rob, are you ready to step us through?
Agenda Item: Discussion - Briefing Letter to the Secretary
DR. FITZMAURICE: If you can't hear me out there in Internetland, please raise your hand. Okay, I'll talk a little bit louder. I've had feedback from the people in Internetland. One of the feedbacks was some of their basements are danger of filling up with water, even up as high as Pennsylvania; not water as high as Pennsylvania, basements who are in Pennsylvania.
What we want to do is go through the letter, particularly the text of the letter, which is the first two pages, and wordsmith and look for meanings. What are trying to communicate to the secretary? If there are places that we need to lengthen to be more explanatory, we will. If there are places we can be more terse, more concise, we would really prefer to do that too.
DR. COHN: Mike, can I just make a comment about process? What we are trying to do here, just to make sure everybody understands, is to develop a draft letter that will then be considered and discussed at the NCVHS meeting, with the opportunity for further revision and hopefully final adoption at the meeting in two weeks. So this doesn't have to be perfect, but it has to be good enough that we feel comfortable sending it to all the members of the National Committee for consideration as a draft letter.
DR. FITZMAURICE: Good point. And the letter is a letter to Donna Shalala. It is to brief her on the activities of the Computer-based Patient Record Work Group, and more particularly on the progress that we are making toward fulfilling a commitment that Congress put upon the National Committee to produce a report on standards for patient medical record information by next August, that is August 2000.
So let's start with the first paragraph which begins, "On behalf of the National Committee on Vital and Health Statistics (NCVHS), I am pleased to inform you of our progress on the next report we are preparing for you as mandated by the Health Insurance Portability and Accountability Act of 1996."
I say the next report because the letter is from John Lumpkin, the chair of NCVHS. Any questions about the first paragraph?
DR. COHN: And this also includes opportunities to wordsmith?
DR. FITZMAURICE: It includes opportunities to wordsmith, Simon.
The next paragraph, "As you may know, HIPAA directs the NCVHS to 'study the issues related to the adoption of uniform data standards for patient medical record information (PMRI) and the electronic exchange of such information, and to report to the secretary not later than four years after the date of enactment of the Health Insurance Portability and Accountability Act of 1996, recommendations and legislative proposals for such standards in electronic exchange.'"
Now I have in front of me the changes recommended by Jeff. One of the changes he recommends is to put a period after the first quote.
MR. BLAIR: Actually, it's not so much that I'm making that recommendation. We have a choice. That phrasing is right out of the law. There is a virtue in keeping that phrasing intact the way it is now, even though it is a very, very long sentence. I just wanted to get the sense of the committee, is it more important for us to go ahead and retain the exact words of the law, or is it more important for us to break that into two sentences to make it more readable? I don't care either way, but I think the committee should give a thought on that.
DR. KOLODNER: Well, you could do both. Use the exact wording, and just start another sentence with the second part.
DR. COHN: I actually had another comment. I don't know if it supersedes it, and may be a little tangential. I actually thought at the beginning where it says as you may know, because I think that's completely unnecessary. I think it should start out with HIPAA directs the NCVHS.
MR. BLAIR: I agree.
DR. COHN: And I think since we are sending this letter to the secretary, maybe we want to say and report to the secretary. Maybe we want to get rid of the report to the secretary, and say report to you, and quote after that.
DR. FITZMAURICE: I could agree with that as making it more readable. But I also find favor with telling the secretary and anybody who reads the letter here is what Congress told us to do in Congress' own words, and I think we are following Congress.
DR. COHN: Oh, I think otherwise the quotes are fine. I'm just changing the quote marks a little bit.
DR. FITZMAURICE: Now I would make a suggestion too, and that is what do you think about combining the first and the second paragraphs since they all have the same chain. Here is HIPAA and here is what HIPAA directs us to do.
DR. FERRANS: I sort of like the way it is laid out.
DR. FITZMAURICE: So Richard likes the way it is laid out already. Any other comments? Leave it like it is? Okay, leave it like it is.
So we have taken the 'as you may know off.' Jeff now has on the table putting a period after the quotation marks, and then starting another sentence, NCVHS is also required to, and then begin the quotes again, 'report to the secretary,' and that leads to four years after the date. That keeps the integrity of what Congress said. Any thoughts on it?
MR. BLAIR: It makes it more readable, but I don't know if you want to do that or not.
DR. FITZMAURICE: Well, do we want to take a straw vote?
DR. COHN: Actually, with the changes we just made shorten it a little bit. I think leaving it the way it was is probably fine. We can take a straw poll if you want.
DR. FITZMAURICE: That's what I favor. How many people favor leaving it the way it is? Five. How many people would like to have it changed? Two. Okay, if the committee agrees, we'll let the majority rule that one.
Jeff has also suggested a change, that is to take out legislative proposals for such standards, and replace such with these. That would change the congressional wording. Any quibble on that?
MR. BLAIR: I'm sorry I didn't realize that out of the law. Sorry, I withdraw the suggestion.
DR. COHN: Can I make one other request for a change? I think maybe we need to do a straw vote on this one also. I personally don't believe that there is a need for us to create and then promote throughout this letter an acronym known as PMRI. As I look at the times that it is used throughout the text, I don't think it increases understanding. I think the times where it has been used time after time that can be handled by calling it patient medical record information. I would speak in striking all references to PMRI.
DR. FITZMAURICE: And replace it with patient medical record information?
DR. COHN: Well, there are times where it is listed redundantly as both points, where it could be handled by putting into the upper body of the piece. And I would say if there is place where we find that we wouldn't need it, understanding we can reconsider it. But it's not a normal acronym that everyone uses. I think that there is a tremendous amount of value to be adding that to our acronym vocabulary.
DR. FITZMAURICE: All right, so does everybody agree that we will strike the initial PMRI wherever they appear and replace it with patient medical record information, and where possible we will put patient medical record information at the top of a list of bullets so we don't repeat it in the bullets below?
DR. COHN: As I said, if it turns out to be -- not work, I'm willing to reconsider it. As I was looking through I didn't see any place that I thought that it was going to be a problem.
DR. FERRANS: There may be places where we don't even have to name it. We can say like on the second page, so far testimony regarding these standards, since it's been mentioned ten times before. You can probably simplify it.
DR. KOLODNER: Was the abbreviation in the legislation?
DR. FITZMAURICE: No, and so it would be struck there too. And I have both after PMRI in the third paragraph, patient medical record information, to raise the point just that Simon raised.
MR. BLAIR: I have a different thought. It not only shows up when we wind up talking about the focus areas, but it also shows up again in several other paragraphs. And it also shows up again in attachments. If we are spelling it out every time, some of the sentences get awful long. So I thought it was helpful if we, at least within this letter and the attachments, so that we were not every time referring to patient medical record information. Otherwise, some sentences that were already long -- so anyway that's my thought.
DR. FITZMAURICE: So then let's decide as we go along.
DR. COHN: We can reconsider. As I said, I looked elsewhere throughout the document, and I thought that it could easily be removed. But if we come to a different decision --
MR. BLAIR: Whatever the group decides.
DR. FITZMAURICE: I must admit in typing though, I just type through the acronym just to save time. But it can always be searched and replaced very easily.
MR. BLAIR: What is the interest of the group?
DR. FITZMAURICE: Why don't we just as we go along say it's not needed here, here it may be needed.
DR. FERRANS: Could we at least on that single time, patient medical record information (PMRI), perhaps reference it once, and then sort of see how it shakes out down the document? Certainly on that one point on the first page we can simplify and move something.
DR. FITZMAURICE: I might take it out of that, because it is in the quotes, and it was not part of the congressional language. So maybe we would do that in a paragraph below just to maintain consistency with what Congress said, rather than put it in parentheses.
DR. COHN: Well, why don't we hold and see if there are ways where we can leave it.
DR. FITZMAURICE: If we don't need it, we'll take it out wherever it is. In the meantime, we'll put it in brackets rather than parentheses in the quote. And if does appear elsewhere, we will take it out.
DR. FERRANS: If we do put it there, I would certainly put it after the definition, rather than prior to the definition, as is customary.
DR. FITZMAURICE: You mean before the words, 'patient medical record information' or after?
DR. FERRANS: I would put it with the acronym after it. But right now it's before it.
DR. KOLODNER: It's usually customary to do it the first time it appears, but not in the quotes.
DR. FITZMAURICE: But if you put it in brackets then that's understood that it wasn't part of the quote. If you put it in parentheses it's assumed to be part of the language.
DR. FERRANS: No, I'm looking in the following paragraph. It says the report will define patient medical record.
DR. FITZMAURICE: That was just to raise the issue. The bolded print was not there. I added the bolded print for discussion purposes. I do agree.
DR. COHN: Why don't we read the second paragraph and see what other changes we might want to make?
DR. KOLODNER: Do you want to put it in brackets?
DR. FITZMAURICE: The answer is no, strike that.
Now we move to the third paragraph? The third paragraph says, "The Computer-based Patient Records Work Group within NCVHS's Subcommittee on Standards and Security has the responsibility for studying the issues and preparing the committee's report by August 21, 2000. The report will define patient medical record information (I'm now replacing the acronym by the full phrase) as information about a single patient. This information is generated by health care professionals as a direct result of interaction with a patient, or with individuals who have personal knowledge of a patient, or with both."
That last phrase starting with 'information about a single patient,' is all in quotes. The explanation is that this definition "was chosen after receiving input from the health industry and other developers and users. It comes from the 18 month study of the Institute of Medicine, which generated the report entitled, "The Computer-based Patient Record: An essential technology for Health Care."
The bold comments first of all suggest using patient medical record information instead of the acronym. Secondly, I would recommend striking "the 18 month study of the Institute of Medicine." Just call it comes from the Institute of Medicine's report --
MR. BLAIR: My thought was this letter is going to the secretary of the DHHS. I'm kind of looking for guidance and advice here, because I haven't been writing letter within the Department of Health and Human Services, and you have far more experience than I do on this. I'm wondering, does she really care about where the definition came from? Do we need to have that in the letter? I would almost rather hold her attention with the relevant points, and try to give her a little bit of a feeling of what the major issues are.
Or maybe you are aware of the fact that these are the types of things you put in for credibility. So what is your thought or the thought of the people on this?
DR. FITZMAURICE: Simon has a thought.
DR. COHN: Actually, I do have a thought on this one. I was wondering whether it's important to her how we define patient medical record information at all. Really there is a very good first sentence which says that the Work Group is responsible, and when we move to this report will be important to you and the nation, because, which I think would be a much more pertinent next sentence.
Now I think what I'm thinking of is that this ought to be maybe in one of the appendices that this is our definition of patient medical record information.
DR. FITZMAURICE: It could go above the focus areas as here is how we define patient medical record information. Here is where we are focusing.
MR. BLAIR: As an attachment? Or you mean in that attachment?
DR. FITZMAURICE: In that attachment.
MR. BLAIR: Very good.
DR. FITZMAURICE: One of the first paragraphs. So let's pick out what Jeff has in --
MR. BLAIR: Let's just make sure that the rest of the group agrees with that.
DR. KOLODNER: What part?
MR. BLAIR: Would you mean describing it again?
DR. FITZMAURICE: We would keep only the first sentence which says, "The Computer-based Patient Records Work Group within NCVHS's Subcommittee on Standards and Security has the responsibility for studying the issues and prepared the committee's report by August 21, 2000."
Then I think Simon's suggestion was to merge that with the next paragraph, which starts, "This report will be important to you."
DR. FERRANS: Take that part and put it in the appendix, what you highlighted.
DR. COHN: It goes in the appendix. And actually it might be that we might consider putting that previous sentence as part of the previous paragraph, and starting out, "This report will be important to you."
DR. KOLODNER: I think it's useful to have the definition in there, and it's useful not to have to go searching for it. We don't need to have how we arrived at other things, but just the definition so that somebody reading in the letter knows what we are talking about, and the boundaries of it.
DR. COHN: So you think it ought to be left in?
DR. KOLODNER: Well, I think just that one sentence with the definition; not all the explanation that was chosen and where it came from. But just that definition. If we want to do a footnote, we can, but I think the main thing is here is what we charged with. Here is the definition that we are using, and then go on to those two tight sentences that we like.
MR. BLAIR: That's acceptable to me. What about the rest of the group?
MS. FRAWLEY: I don't really think that we need this at all in the letter. I'm not even sure it needs to be in an appendix, because our report is going to lay out in our report what our definition was, what our methodology was, what are findings and conclusions are. So I'm not really sure in the letter, which is really to let the secretary know what we are doing, and what will be coming, that we need to get that specific about a definition. So I would argue we don't need it.
MR. BLAIR: I guess we have three options then?
DR. KOLODNER: Kathleen, the only thing would be if that definition is for some reason not acceptable. This is what we are focusing on. Then we would want to get that feedback early, rather than spending the time writing the report.
DR. FITZMAURICE: That was the original thought, Rob, you are right. That was the original thought for us putting it in there.
DR. FERRANS: Because I'm even less familiar with the Dear Secretary format than Jeff, that sort of brings into a broader question of should this be a one page letter without appendices and everything else? What is the level of detail that is required for a typical letter? I mean maybe there is too much information in it. I just don't know. I just wanted to sort of ask.
DR. FITZMAURICE: Let me go back and suggest that the purpose of the letter is to let the secretary know the progress we are making. And also to begin putting out information so as to get feedback from the secretary and those around her as to are we going in the right direction. So by putting the direction out there, this is our first chance to get feedback.
DR. COHN: I guess we have about four different choices now. Just running through the options we have, just to see if we can come to a consensus. We could leave it in there as it is. We could leave it in there, but strike the sentence about how it is that we came to this definition. We could put it in the appendix, or we could put it as a footnote, and just footnote the next time we have patient medical record information.
I tend to agree with Dr. Kolodner that it probably needs to be there somewhere. I'm not sure that I think it needs to be in the second or third paragraph. But beyond that, I don't have a strong feeling of where it should be.
DR. FERRANS: I guess my other point is that if we were expecting interaction, then certainly we would want to bring a lot of detail. It's just with Bill on the panel, and I mean he's not here today, I'm just saying is it a reasonable expectation that they would come back and look at the details and say, you need to focus more on vocabulary, or this way is off or that? I just don't know. I just think that sometimes less is more at the executive level.
DR. FITZMAURICE: You're right if the purpose is to say we're working on this, and we're making progress. But if the purpose is to say here are some of the things that we plan to look at, then the secretary or someone around her may say, you know there's a gap. This hasn't been looked at. Or this was taken care of somewhere else. You have too much or too little, but it's a foreshadowing.
The sense of the group is what would govern in this case as to what the purpose of the letter is; what information do you want to get up to that level. You can always then talk with somebody and say do you have any feedback to the letter? What letter? And then they have to run around and grab it.
MR. BLAIR: What is our vehicle for choosing here? Should we articulate each option and see how many people support that? In my case, to be honest with you I'd vote for two different options --
DR. FITZMAURICE: Let me interrupt, Jeff, because I think this is really something that is the call of the chair, maybe the chair of the Standards and Security, because it's very personal. It's what you personally want to have John Lumpkin say to the secretary. So I would go more with the chain of leadership on this one.
DR. COHN: The chair of the NCVHS or the chair of the subcommittee or the chair of the work group.
DR. FITZMAURICE: The guidance we have here today, and then as it gets up to the chair of the NCVHS, he may change it.
DR. COHN: I'll make a proposal just to see what other people think. I actually like the wording. I think it actually ought to be a footnote that references the first time that whole phrase is used, maybe not in the legislation, but down further, and the whole thing ought to be referenced. Because I think it's important if we are going to give a definition, we need to explain why we came to it. I think the secretary would want to know that we just didn't come out of this just brainstorming.
MR. BLAIR: Would the footnote have just the first sentence?
DR. COHN: I would have the whole thing.
DR. FITZMAURICE: Could I suggest different wording? As I move it around, suppose after August 2000 it said after researching several definitions of patient medical record information, we adopted a definition of the Institute of Medicine used in, "The Computer-based Patient Medical Record and Essential Technology." The definition is, and then what's in quotes, information about a single patient.
That might work well in a paragraph. It might work well in a footnote. It just says after researching several definitions. We don't have to talk about the health industry and others; that's part of our research. But it shortens that.
MS. FRAWLEY: That actually is the part that I like. It shows that we obviously got input from other individuals. It's wasn't the five people on the Work Group pulling something out of the air. We actually got input.
DR. FITZMAURICE: I think that's an important point.
MS. FRAWLEY: I don't think this language is bad. I think it could be in a footnote. I just don't think it needs to be sitting in that paragraph.
DR. COHN: I agree.
DR. FITZMAURICE: It's hard to have a letter with a footnote.
DR. COHN: Is that right?
MR. BLAIR: Then I do have a hard time envisioning -- that's such a big footnote. What about the other option of having a reference at that point, and having it in the attachment?
DR. FERRANS: Put it in the attachments, and then perhaps let the chair decide how much of the attachments that they want to send. That might be appropriate. So we just give everything, and move it to the chair. Perhaps that is proper that the Work Group should be very inclusive.
DR. KOLODNER: It's a pretty small attachment.
DR. FERRANS: There are two or three pages of attachments.
MR. BLAIR: So you are putting it into the beginning of the first attachment?
DR. FERRANS: Yes. I think the definition should go first.
MS. FRAWLEY: I would put it in an attachment too.
DR. FITZMAURICE: I would put it in an attachment too, because that leads into the focus areas. That's the work that we did particularly to the report.
DR. KOLODNER: And we'll leave the wording as is, without the underlining?
DR. COHN: Yes.
DR. FITZMAURICE: We'll move on now to the fourth paragraph. Simon, you would suggesting linking the third paragraph, making that part of the paragraph above it? I think that was your suggestion.
DR. COHN: That was a thought.
DR. FITZMAURICE: It was a good idea. It holds together well.
MR. BLAIR: The second paragraph begins with what words?
DR. FITZMAURICE: HIPAA directs the NCVHS to study and then it gives the charge. And this would add to that, right after the charge, "The Computer-based Patient Record Work Group within NCVHS's Committee on Standards and Security has the responsibility for studying the issues and preparing the committee's by August 21, 2000."
Now you had suggested, Jeff, in your notes that studying these issues, as opposed the word 'the,' but since we haven't introduced the issues yet --
MR. BLAIR: Fine.
DR. FITZMAURICE: Okay. The fourth paragraph, which will now become the third paragraph reads:
"This report will be important to you and the nation --
DR. COHN: No, the sentence before it goes up with the previous paragraph.
DR. FITZMAURICE: The Computer-based Patient Record Work Group, that sentence goes up into the second paragraph, correct? Oh, I'm sorry, you're talking to Rob.
MR. BLAIR: That was the piece that I was really trying to clarify with my question. Is the sentence that starts with, "This report will be important to you?" I was hoping that would be the beginning of a paragraph.
DR. COHN: Yes, I think we're agreeing on that.
DR. FITZMAURICE: I think we're all set now. For those of you in Internetland, we're working on the screen Rob Kolodner has it projected, and all of our wordsmithing changes are going on this document.
Now the third paragraph reads, "This report will be important to you and the nation, because it will recommend standards that will 'significantly improve the efficiency and effectiveness of the health system for the delivery of high quality care.' This reflects the first of ten guiding principles for selecting patient medical record information standards (see Attachment 1). These principles are based on the guiding principles for selecting standards for HIPAA."
DR. KOLODNER: Do you want PMRI to be spelled out?
DR. FITZMAURICE: I think we have agreed to spell it out wherever it exists, unless somebody says we ought to have PMRI there.
DR. KOLODNER: And using capitals?
DR. FITZMAURICE: I would leave it as lower case all the way.
MR. GARVIE: Is the 'significantly' really an appropriate part of the quote?
DR. FITZMAURICE: That's the first of the ten guiding principles that are in Appendix 1. Oops, it's not in the first principle. Then let's take out 'significantly.' You're right, good catch. 'Significantly' is not part of the direct quote.
MR. BLAIR: Could we leave it in the sentence, but have the quote start after it?
DR. COHN: Yes. Actually, I have a question for the Work Group. I'm mulling about the statement that we will recommend standards that will, because I think our recommendations are going to be beyond just necessarily recommending standards. So I'm not sure if I have the right words, because it will make recommendations that will make specific proposals that will.
DR. FITZMAURICE: Make recommendations about standards?
DR. FERRANS: Recommend solutions.
DR. COHN: Actually, make solutions about standards.
MS. FRAWLEY: What we're supposed to do is our charge was to study the issues related to the adoption of uniform data standards. That's what we were supposed to be doing.
DR. KOLODNER: Making recommendations and legislative proposals. That's where I got the proposal language.
MS. FRAWLEY: The standards should be in there, because that's what we were charged to do.
DR. COHN: I think recommendations about standards.
MS. FRAWLEY: I don't think we're going to recommend standards.
DR. COHN: So recommendations about standards.
DR. FITZMAURICE: That will significantly, "improve the efficiency and effectiveness of the health care system for the delivery of high quality care."
MS. AMATAYAKUL: Where is this quote from? It doesn't give the source, and usually when there is material quoted, you can identify the source.
DR. FITZMAURICE: It is the first of ten guiding principles found in Appendix 1.
MS. AMATAYAKUL: If somebody is reading this letter and they don't where that came from, they could assume it came from the legislative language. They could assume it came from -- you really don't identify where it's coming from.
DR. COHN: Well, actually, I guess the question is, is this -- maybe it's already changed up there.
MS. AMATAYAKUL: Maybe you just don't need the quote, because that's what you are going to do.
DR. FITZMAURICE: Or we could put see Attachment 1 just right up by the quotes.
DR. FERRANS: It's also this is going to important to you because, blah, blah, blah, instead of because "blah, blah, blah."
MR. BLAIR: Good point.
MS. FRAWLEY: Do we need to have the word 'see' in that bracket? See Attachment 1? Shouldn't it just say, (Attachment 1)? We should tell her to see it.
DR. KOLODNER: Where do you want the Attachment 1?
MS. FRAWLEY: It's fine. I just wanted to get rid of the word 'see.'
DR. FITZMAURICE: As long as we're not referring to the quote, then nobody has to see it. It's there if they want to see it.
MR. BLAIR: It's being sensitive to blind people anyway.
DR. COHN: Is this exactly the same as all the principles in the HIPAA standards? It is, isn't it?
DR. FITZMAURICE: I would guess that it.
DR. COHN: So I'm wondering whether or not we should just say -- we say this reflects the first of ten guiding principles for selecting patient medical records information standards. And then say these principles are whatever on HIPAA. I don't know if we should just say this reflects the first of ten guiding principles for the HIPAA standards or something.
DR. FITZMAURICE: Except that we modified them to develop guiding principles for promoting PMRI standards.
MR. BLAIR: Yes, that's right.
DR. COHN: Okay, that's fine.
DR. FITZMAURICE: I did with consensus of the group.
MR. BLAIR: Plus, we also had reviewed an ANSI health care informatics standards board as well.
MS. FRAWLEY: It's the guiding principles for selecting standards for HIPAA. Who developed those? You did?
DR. FITZMAURICE: Yes, with group.
DR. COHN: They were included in the NPRMs.
DR. FITZMAURICE: Every NPRM has the guiding principles. But it was with significant input from outside groups, as well as inside the government. You could have a host of authors if there is any trouble with it. That's a joke out there in Internetland.
Are we ready to move on to the next paragraph, or do people still want to look at the third paragraph? Let's move on.
MR. BLAIR: By the way, have we missed anything where Clem was making comments on these?
MS. FRAWLEY: We haven't got to that yet.
MR. BLAIR: Okay.
DR. FITZMAURICE: The next paragraph leads into a set of bullets. It reads:
"This report will address this subject this the following six focus areas:
DR. FERRANS: If we change this subject, strike that, and put in patient medical record information, then we can delete it from all the bullets.
The other one is should effected be affected?
DR. FITZMAURICE: That should be affected, yes.
MR. BLAIR: I think we could do that. I would just like one kind of little validity test before we actually take the action. A lot of those are very general type of activities that are listed. And we want them each to be specific to PMRI. Could you just visually glance down that list with the concept in mind, PMRI would be introducing those bullet items and not be in each one.
DR. FERRANS: How about if we changed it to this report will address the following six focus areas as they relate to patient medical records?
MR. BLAIR: That's a good strong introduction.
DR. GRIFFITH: You might consider also rephrasing that last one to something like effects of inconsistencies and contradictions among state laws. It's more consistent.
DR. FITZMAURICE: So do you have a wording for that?
MR. BLAIR: I had thought of getting rid of contradictions and just saying inconsistencies. Does that help?
DR. FERRANS: The effects of inconsistencies among state laws to keep it short and consistent with your previous bullets.
DR. FITZMAURICE: So if would be --
DR. FERRANS: Effects of inconsistencies among state laws.
DR. FITZMAURICE: Is the patient medical record inconsistency or is it the state laws that are inconsistent? I think it's the state laws that are inconsistent. We're not saying the PMRI is inconsistent. As they relate to PMRI effects of inconsistencies among state laws.
Jeff, let me ask you, would you like a corresponding change in the sixth bullet on the attachment, which talks about the patient medical record information focus areas? The bullet there reads PMRI effects of inconsistencies and contradictions among state laws. So they should be consistent, whichever they are.
MR. BLAIR: I really don't have a strong feeling. What are your thoughts on that?
DR. FITZMAURICE: I took them exactly as they were and put them into these bullets. So for changing the bullet, I would advise changing the label on the last one to read --
DR. FERRANS: Could I ask sort of a subject question? Is the focus area going to address consistencies among state laws? Or is it going to address the effects of the inconsistencies? That's the question in my mind.
MR. BLAIR: Our report has to address the effect, and whether or not the effect is serious enough that we are making recommendation to address those effects.
DR. FITZMAURICE: Let me read the paragraph that described that under the focus area. The one that talks about PMRI effects of inconsistencies and contradictions among state laws. I'm reading from Appendix 2, Attachment 2. It reads:
"These laws can discourage or prevention nationwide consistency in the creation, storage, or communication of patient medical record information. Inconsistencies include laws pertaining to record retention, document authentication, and access to records."
DR. FERRANS: So it is the effects.
DR. FITZMAURICE: So it's the effects of the laws on the creation, storage, and communication of PMRI.
DR. FERRANS: For accuracy I would leave effects in there. The focus is not the laws.
DR. COHN: One other wordsmithing. Is it among or it in state laws?
DR. FITZMAURICE: Among, because it's across state borders, as opposed to inconsistencies within each state law.
DR. COHN: Okay.
DR. FITZMAURICE: So the question I come back to is in Attachment 2 where we have changed the sixth bullet, should we also change the sixth focus area?
DR. KOLODNER: The wording of the sixth bullet I thought is in fact exactly this PMRI effects of inconsistencies. It's really the effects on the PMRI.
DR. FITZMAURICE: We took out contradictions. That might be worthwhile taking out 'and contradictions.' Did we agree to take out 'and contradictions.' Maybe I jumped ahead of myself.
MR. BLAIR: Yes, I thought so.
DR. FITZMAURICE: Then we'll take it out also in the focus areas, right? And talk about PMRI effects of inconsistencies among state laws, and not talk about the contradictions, because a contradiction could be an inconsistency.
MR. BLAIR: I'm sorry, did I get confused?
DR. FITZMAURICE: No, it's not a matter of confusion. The focus reads, "PMRI effects of inconsistencies and contradictions in the state laws."
MR. BLAIR: And I thought we were going to take out the word 'contradictions.'
DR. FITZMAURICE: Okay, so we will take out and contradictions, both in the sixth bullet, and in the attachment, which describes the focus areas.
DR. KOLODNER: And which otherwise doesn't have the word 'contradictions' in it.
DR. FITZMAURICE: Right. So that paragraph looks pretty good. We're telling the secretary here are the focus areas of the report. Now we get into a paragraph that describes the process briefly. It says:
"To gather information for the report on patient medical record information standards we have held four hearings over a total of seven days in 1998-99, and have planned for three more days of hearings before the end of the year."
Now by the time she receives this letter, we will have had two of those days of hearings today.
"We have heard from over blank testifiers during this time period." Now I don't know if we need to say we have heard from over so many testifiers. It probably runs up about 50 or 60.
DR. COHN: Well, that's fine. I think we can just say to gather information for this report, first of all.
MR. BLAIR: My thought was that it would be sufficient in this letter to just simply say X days and so many testifiers. Just make it one short sentence.
DR. FITZMAURICE: Okay. So to gather information for the report, deleting on PMRI standards.
MR. BLAIR: Let's see if the rest of the group agrees with that.
DR. COHN: I disagree though. I'm sort of mulling over this. You're saying that we should only reference the hearings that we've already done, as opposed to reference the full number.
MR. BLAIR: My thought was I don't think it's all that important that we indicate over the period, how many days, and how many testifiers. I think you wind up with two sentences there or more. I was just trying to make it more concise. I think the only thing you want to convey to her is a lot of testimony has taken place so far. I think you could do that probably by saying we've had six or seven days of testimony with 120 testifiers. That makes the point. I thought that might be adequate, but I don't know what the rest of the group feels.
DR. FITZMAURICE: I think to be accurate we would say nine days of testimony counting today and tomorrow as days of testimony. We can certainly count up the number of testifiers. You would suggest that we not talk about hearing that we may have in the future?
MR. BLAIR: Oh, no, I think you could wind up saying --
DR. KOLODNER: We have held or plan to hold.
MR. BLAIR: Plan to hold additional testimony, additional hearings.
DR. FITZMAURICE: Without being specific?
MS. FRAWLEY: Actually, I think it's important that we make sure that we show how much we have put into this, the number of testifiers. I think it's important if we are planning more days, which I know we are, that that should be in the letter. That needs to be wordsmithed a little bit, that sentence.
DR. FITZMAURICE: You say the more days should not be in the letter?
MS. FRAWLEY: I think they should be. I think that we're trying to make the point that this is going to be a credible report. It's going to be based on a lot of work, with a lot of input from a number of individuals. I mean we've heard from tons of people. I don't even know at this point how many.
MR. BLAIR: One point five tons.
MS. FRAWLEY: I think so.
DR. COHN: More than that.
DR. KOLODNER: Because this letter is actually dated next week, there will be one more day of testimony that we haven't accounted for, since these will be behind us. So we have held at that point it will be five hearings over a total of nine days. We can decide whether we want to have the duration 1998-99, and have one more day planned, or something like that, but we need to put the number of testifiers in there too.
DR. FITZMAURICE: Which we'll be able to count after tomorrow.
MR. GARVIE: Can we say we have heard from a number of testifiers during a number of hearing over a number of days, and then go on?
MR. BLAIR: We're trying to make it more concise, that's all.
DR. FITZMAURICE: So you're saying delete the 'over' and just say we have heard from blank testifiers during this time?
MR. GARVIE: No, you can still have the number of days. But you could make it a little more concise.
DR. FITZMAURICE: By combining the sentences?
MR. GARVIE: Yes, we heard from blank testifiers, whatever that number is.
DR. FITZMAURICE: Semi-colon.
MR. GARVIE: I don't think you need one. Just you heard from the testifiers during five hearings over the number of days.
DR. FITZMAURICE: Okay, so we've heard from blank testifiers over 5 hearings over a total of blank days, 9 days in 1998-99, and plan for one more day of hearings.
MR. GARVIE: Right, and then the only change I think, Mike, was during the number of hearings.
DR. KOLODNER: What was the last change?
MR. GARVIE: In other words, we heard from the testifiers during the hearings.
DR. FITZMAURICE: We have heard from 60 testifiers.
MS. BURKE-BEBEE: Could I make a suggestion here? You could put a period at the end of 1998-99, and then be more specific and say our final hearing date is such and such date if we have it.
DR. COHN: We don't have it. I guess the thing we're sitting here wondering about is how we reference the fact that we may actually have formal hearings for two more days, certainly not one. But we're also having meetings in December and meetings in January and February. Do we need to reference that? That's not really hearings. I think Jeff was making that point.
MR. BLAIR: Again, the only other thing that I'm thinking of here is the only other thing that I'm thinking of here is the only point we want to make to communicate that I think at this point is that we have got information from a lot of diverse people. I sort of feel like once we have made that point, how many times we meet beyond that, I don't know if we need to put that in the letter, but whatever.
DR. COHN: How about to gather information for this report we have so far heard from.
DR. FITZMAURICE: Good.
DR. COHN: And then just forget the future.
MS. AMATAYAKUL: We could say as an additional sentence, we plan for additional hearings between now and the end of February or something like that. We plan for additional hearings as well or something.
MR. BLAIR: During the next few months. Why don't we make it ambiguous?
DR. COHN: A separate sentence there.
DR. FITZMAURICE: It seems like a lot of sentence for one more day of hearings.
DR. COHN: Well, it may or may not be. We're planning for additional hearings during the next few months.
DR. KOLODNER: Do you want to read it for Jeff?
DR. FITZMAURICE: To gather information for this report, so far we have heard from blank testifiers during five hearings over a total of nine in 1998-99. We plan for additional hearings during the next few months. That's good enough to move forward.
The next paragraph talks about what we have learned from these hearings. It reads:
"So far testimony regarding standards for patient medical record information has shown some areas of great accomplishment, and other areas where the federal government might play a constructive role to advance progress."
DR. COHN: Can I make a first comment, and then a second one? I think we can say so far testimony has shown, rather than maybe we already know what we're holding testimony about.
The other piece is we need somewhere to put here, and it may be here or after the next paragraph, to make some comment about confidentiality. Kathleen, I look to you about that one. This morning we were hearing pretty strongly from people, just reminding us that without some sort of confidentiality protection, we're not going to be able to go really forward with the implementation of all of this.
So there is maybe a place here that maybe we need to put a sentence or two just reminding that is one of those things we heard. Now it could here or it could be the next.
DR. FITZMAURICE: While confidentiality was said to be important by many testifiers, this is actually be addressed by the subcommittee somewhere else in NCVHS?
DR. COHN: No, what we are saying is that Congress hasn't acted yet.
MS. FRAWLEY: I think that we would make a strong argument for the legislative proposals.
DR. FERRANS: Witnesses have expressed concern.
MS. FRAWLEY: And Congress has failed to act.
DR. COHN: And we urge the secretary to fulfill her role.
MR. BLAIR: I agree with you. My only thought is could we add that on after we finish this paragraph, because there is kind of a continuity in what's in this paragraph. One of the other thoughts is we did have one of our focus areas being data security, with a reference to the fact that another committee was working on that, and we removed that. Despite the fact that we removed that, I still agree, Simon, that this is really an important, and it really ought to be added. But let's go through the paragraph first.
DR. COHN: And see where it fits. I'm just putting that it. I don't know whether it's here or the sentence itself needs to be changed a little bit, which I think is the point that you were making, Jeff, because you're right, it doesn't go along with accomplishments and a constructive role. It's a question of where we think confidentiality sits. I would say it's before the constructive role piece.
MR. GARVIE: Why don't we say so far testimony has shown some areas of great inconsistency?
MR. BLAIR: Could the read the paragraph through in its entirety, and then go back and wordsmith it, just so we have things in context?
DR. FITZMAURICE: Now Jeff has proposed linking the short paragraph with the long paragraph beneath it, so let me read them both together to show the continuity. It reads:
"So far testimony regarding standards for patient medical record information has shown some areas of great accomplishment, and other areas where the federal government might play a constructive role to advance progress. For example, health care communications standards and terminologies that facilitate the reimbursement process have relatively strong support for development and implementation. However, health care information standards and terminologies that enable clinically specific documentation, facilitate performance measurement, and improve clinical efficiency and effectiveness have less support for development, and are not yet widely implemented. The impediments to the development and acceptance of clinically specific standards and terminologies include the need for better information models to promote faster development and coordination, equitable fee structures for the use of these standards, improved coordination among standards developers, and additional incentives to accelerate development in implementation. For additional information the environment and trends for health care information standards and terminologies see Attachment 3."
DR. FERRANS: I have a problem with the word 'support' as it is used there. It's very confusing. If you say that -- you sort of implied there that the clinical standards have less support for development. People could interpret that as saying that people don't want it. The lay reader would interpret it that I think. So we need to rephrase that.
MR. BLAIR: I think Clem reacted to that as well. So do you have a suggestion on wording on that?
DR. FERRANS: I think what we are really trying to say is that they have been widely adopted.
MR. BLAIR: It depends on which ones you are looking at. I was trying to keep it concise there. SNOMED-RT has not been widely adopted. And a lot of the clinically-specific medical terminologies haven't been yet. And even HL7 when you look at it, it depends on what context you are looking at, because some HL7 message formats have 40, 50, 60 percent adoption by vendors, and others less than 10 percent.
DR. FERRANS: If we said something like for example -- and I agree with Clem's comments about health care communications standards -- if we said for example, administrative standards have been widely adopted or something like that, however, health care information standards and terminologies that facilitate all these things are not yet widely implemented.
DR. FITZMAURICE: As I read through Clem's comments, I think he would not make the distinction between administrative and clinical standards, and would have us substitute we need more areas covered by standards, and more widespread adoption of existing standards to be able to solve many of the health care system problems including clinical efficiency, effectiveness, performance measurement, and research. And he says you could go on, government could find ways to influence this process.
So I think he wants to link we need more of this, because it will achieve that, and not get into the difference between the administrative and clinical.
DR. FERRANS: What if -- and maybe this is too much detail, I'm sort of thinking as I go there -- but what if we said something like I guess he's right, and we're talking about regarding standards for PMRI, areas of great accomplishment. One could say for example Health Level 7 is widely used.
MR. BLAIR: See the piece that I have a problem with, with that --
DR. FERRANS: And messaging. I don't know.
MR. BLAIR: The thing that I'm getting here is that in an ambulatory setting or acute care setting, everyone uses the standards and code sets for reimbursement purposes. They have to. In the areas of HL7, there is nowhere near 90-95 percent, except I think it's good acceptance. I'm not critical in any shape, manner, or form of HL7 and what they have done. I think for what they have done without having it being required for reimbursement purposes, the private sector has widely accepted it. But that's a big phrase that I just wound up saying, and it was in a certain context.
But 40 or 50 percent acceptance by vendors or even less, 5, 10, or 15 percent by other HL7 message format standards, it depends which ones you are talking about. The thing is that I don't want to get into an HL7 reference here. I think the point of that sentence is a little different.
The point is from Donna Shalala's standpoint, there are certain things she wants to do to drive the health care system. She wants to measure outcomes, and measure performance of health care systems, and improve clinical processes. From that standpoint, the standards are not well accepted yet. Now apparently I failed to say that well. Different people had read different things into it.
DR. COHN: I just say that some other people like me have some comments. I'm going to make a comment after Jim Garvie, because I don't know what the answer to this one is.
MR. GARVIE: Jeff, would it be acceptable to eliminate a number of words in that sentence and say for example, standards that facilitate the reimbursement process are widely implemented? And then your next sentence draws the distinction that I believe you were referring to.
DR. FERRANS: Aren't there great accomplishments though? I guess that's the thing. We talked about areas of great accomplishment and then said, for example. Is that an area of great accomplishment?
DR. COHN: I'm just going to make a general statement about this paragraph, which I am very uncomfortable with, because I think it goes this is the substance of the recommendations, and we are identifying where the problems are, and where the good things, the bad things, the post analysis work, what we think needs to happen.
In three or four months I would be happy to delve into this, though not quite in this fashion. I don't have an answer to how we need to do replace this, but I think that this is too deep and too specific. I think Dr. McDonald is reacting to that. He doesn't like the word 'model.' We have heard certain things, and I'm just trying to think of what it is we have heard that we can, at a high enough level, say.
I guess what I'm bothered about is we have heard relatively uniformly that there is a role in government in relationship to this area. That's something we have heard again and again.
DR. FERRANS: It's listed as a question. It's interesting, because I think on the conference we had, I think that everyone agreed to that.
MR. BLAIR: Yes, and I failed to get that in, and I apologize. I think we need to get that in.
DR. FERRANS: Perhaps what we could say, so far testimony regarding standards for this have shown, and we can list if there any sort of no-brainer analyses that we have done, and that's the one that stands clear. There seems to be agreement among, and then we list some of the different groups. So we show it has broad support like among developers, health care organizations, blah, blah, that there is a constructive role for the federal government to play.
DR. COHN: Yes, and I think we can even follow that piece up by saying that we've heard that, and that we are now undergoing discussion and analysis in preparation for final recommendations to identify how best government can help promote and facilitate this area, or something like that. I don't know. Mike, help us.
DR. FITZMAURICE: I'm sorry, I was reading Clem's paper, and you woke me out of my reverie.
DR. COHN: I was actually saying let's avoid -- I think this is too specific.
MS. FRAWLEY: If you go back to Clem's last little paragraph --
DR. FITZMAURICE: It reads, "We need more areas covered by standards, and more widespread adoption of existing standards to be able to solve many of the health care system problems, including clinical efficiency, effectiveness, performance measurement, and research. The government should find ways to influence this process."
MS. FRAWLEY: See I think that's what we need. What he put is really what we want to say, without getting into all of this detail. I think that's really what we want.
DR. FERRANS: The only thing I would modify from that is rather than saying the government should, it should be the witnesses or the consensus is that the government should, rather than us telling the government.
DR. FITZMAURICE: We might also say the witnesses testified that we need more areas. If we are putting this all on the witnesses, it should reflect that. The 'we' should be the witnesses, not us.
DR. KOLODNER: First sentence we leave?
DR. COHN: Let's get this new one on, and then we can change the above. So the first one is fine, but we just need to modify it a little bit.
DR. FITZMAURICE: Do you have in front of you Clem's paragraph? I can give you mine.
DR. FITZMAURICE: Want me to read it while you type it?
DR. KOLODNER: Yes, that would be easier.
DR. FITZMAURICE: We need more areas covered by standards and more widespread adoption of existing standards to be able to solve many of the health care system problems, including clinical efficiency, effectiveness, performance measurement, and research. The witnesses believe the government should find ways to influence this process. Although we don't refer to a process, do we?
DR. COHN: That was the next sentence.
DR. FITZMAURICE: The witnesses believe that the government should find ways to influence the standards process.
DR. COHN: I'm not sure what this process is.
DR. FITZMAURICE: But process wasn't referred to anyplace else.
DR. COHN: That at least gets it up there. I think going up higher, we say, "So far testimony has shown some areas of great accomplishment." I think we mean great accomplishment by the private sector, don't we?
MR. BLAIR: Yes. The only thing is that maybe it's not just the private sector.
DR. FITZMAURICE: If you look at DEEDS(?) from the CDC and the immunization standards, ICD-9.
DR. COHN: I know what you are saying. I'm just not sure where --
DR. FITZMAURICE: Whom to pat on the back for what we have accomplished so far.
MR. BLAIR: Maybe we ought to acknowledge both, in both the private and the public sector.
DR. FERRANS: Should that be significant instead of great accomplish? The problem with great accomplishment is that it really just says well then why do we need to be doing this? The other thing is where the government might play, or where the government should or could play.
DR. FITZMAURICE: We could drop that and move that to the sentence that Clem added to his. Because in the paragraph we don't talk about what the constructive role might be. So it might be good to end with something like that.
DR. COHN: That's move that down.
MR. BLAIR: Could I ask whether, and I don't know if this is going to be strong enough -- we may need a separate sentence or two -- but Clem listed those areas where we need additional attention. Should we add to, within that sentence, the area of privacy and confidentiality? Or do we need to make it an even stronger statement, and have that as a separate sentence?
DR. FITZMAURICE: I would make it separate. Our focus is not on confidentiality and privacy. I would make it, if we mention it, separate, saying that they also mention confidentiality and privacy, but of course that's the focus of another area of NCVHS, or that is not a major focus of this report.
DR. COHN: We here so far testimony has shown some areas of significant accomplishments by both the private and public sector. The witnesses believe that we need more areas covered by standards, and more widespread adoption of existing standards to be able to solve many of the health care system problems, including a variety of things.
Witnesses believe the government should find ways to influence process. Or maybe this is believe that the government should find ways to play a constructive role to progress.
DR. KOLODNER: So pick up the phrase down below.
DR. COHN: Yes, ways to play a constructive role to advance progress.
DR. FERRANS: Should that sentence start with however?
DR. COHN: That's possible. Which sentence? Oh, there, okay. I think we need to also to say maybe as a final sentence there that the Work Group is now beginning to consider how the government -- what activities the government might undertake.
DR. FERRANS: What specific activities.
DR. COHN: That's sort of the reasonable next sentence.
DR. FERRANS: One thing about the sentence, 'the witnesses believe we need more areas covered by standards, and more widespread adoption,' it's sort of poorly constructed. If it was something like we need over coverage of different areas by standards, more widespread adoption in the English language sense. I don't know. I know what we are trying to say. There's a better way to say it. And it's important, because that is the message right there. We need to say it clearer.
DR. KOLODNER: But standards need to cover a broader area or broader areas.
DR. FERRANS: And that standards should be adopted.
DR. KOLODNER: One is the breadth, the other is how well they are used. So coverage of areas, and then degree of implementation.
DR. FITZMAURICE: The witnesses believe that we need both more areas covered by standards, and more widespread adoption of existing standards.
DR. COHN: That sounds good. It probably isn't perfect now, but it may be good enough for us to then review it tomorrow.
DR. KOLODNER: We need for additional areas to be covered by standards.
DR. FITZMAURICE: If we come down to standards and put a period, then we could start the next sentence, these standards are needed to be able to solve many of the health care system problems. It breaks up a long sentence.
DR. KOLODNER: So what do you want to do with this sentence?
DR. COHN: Period after existing standards.
DR. FITZMAURICE: After adoption of existing standards period. These advances are needed to be able to solve.
DR. KOLODNER: Are we happy with the phrasing? Should we read the sentences now?
DR. FERRANS: Not to be able to solve it, just to solve. Be able is redundant.
DR. FITZMAURICE: Good suggestion.
DR. KOLODNER: So far testimony has shown some areas of significant accomplishments by both the private and public sectors. However, the witnesses need both additional areas to be covered by standards, and more widespread adoption of existing standards. These advances are needed to solve many of the health care system problems, including clinical efficiency, effective, performance measurement, and research. The witnesses believe the government should find ways to play a constructive role to advance progress.
DR. FITZMAURICE: You know one thing that seems to be missing is patient care or quality in that list. You know, medical efficiency, more productivity, effectiveness.
DR. KOLODNER: This is the business side, rather than the individual.
DR. FERRANS: I just want to make one other thing. Where is says these advances are needed to solve. I think people had a much stronger opinion. They were saying that these advances are essential to solving. Without it, it's not going to happen.
MR. BLAIR: Could we hold on to Michael's comment there? Michael had suggested that we didn't have anything in there about improving patient care.
DR. FITZMAURICE: The quality of patient care.
MR. BLAIR: I would think that's very important to the secretary.
DR. FERRANS: Making that first for clinical efficiency; first and foremost.
DR. KOLODNER: We were working on some wording, and we do need to get that. We want these advances are essential for solving?
MR. BLAIR: For improving patient care.
DR. FERRANS: That was a different thought.
DR. KOLODNER: For solving many of the health care system problems, including --
DR. FITZMAURICE: Improving patient care quality.
MS. FRAWLEY: But you are putting patient care in with health care system problems, and it's not. So what you want to do is do another phrase.
MR. BLAIR: If we just change the word 'problems' to issues, then we could wind up saying that all those things are issues to improve the quality of care, to improve efficiency, to improve effectiveness. Does that give us an umbrella to work within?
DR. FERRANS: I think the objection had to do more with the fact that these are like health care system problems.
MR. BLAIR: I don't even know if we should characterize it as saying that what we are focusing on is process.
DR. FERRANS: What if it said, however, the witnesses believe that both this and that, if we said first and foremost this would improve individual patient care. Saying in addition, these advances are essential for solving, something like that?
DR. COHN: Sounds good.
DR. FERRANS: This would improve individual patient care. We probably want some adjective before improve.
MR. BLAIR: Significantly, dramatically.
DR. FERRANS: I have heard people say dramatically. I'm big on big adjectives. I would defer to the group.
DR. COHN: Let's try dramatically.
MR. BLAIR: It is dramatic.
DR. COHN: Could we say improve individual patient care and reduce adverse events?
MS. AMATAYAKUL: Is the quality of care in the United States so bad that we have to dramatically improve it?
DR. FITZMAURICE: You mean bad compared with what it could be, or compared with the rest of the world?
MS. AMATAYAKUL: I mean to me I think we want to promote improvement, continual improvement. We want to insure quality. This makes it sound like --
DR. FITZMAURICE: I do have a problem with the 'dramatically.' It makes it sound like we are really negligent because we don't have these standards in place already. It is a good point.
DR. FERRANS: I would argue that the data on morbidity and mortality from adverse drug reactions and deviation from evidence-based --
DR. COHN: Maybe we need to relate that to adverse events then, as opposed to patient care. Maybe these would improve patient care, and could dramatically reduce adverse events?
DR. FITZMAURICE: Now to what is it that these refer, these what?
DR. KOLODNER: The witnesses believed that we need both.
DR. FITZMAURICE: These both? These areas? I'm having trouble.
DR. KOLODNER: This being what? Because we say we need both this and this, first and foremost.
DR. FITZMAURICE: First and foremost both would improve? No, that's confusing. These areas. These two whats?
DR. KOLODNER: Achieving these. Achieving the coverage of areas by standards, and more widespread adoption.
DR. FITZMAURICE: Doing this. Doing what was above.
DR. KOLODNER: Okay, because we're into the grammar, but the essence I think is -- we've got the introduction of we've got some significant accomplishments, but there are two things, the broadening of areas covered, and the widespread adoption. Those are important, and doing those would improve patient care dramatically. Could dramatically reduce the adverse events. These advances are essential.
The question is which advances there, because we are actually going back two sentence. Are essential for solving many of the health care system problems, including da da da. Witnesses believed the government should find ways to play a constructive role. We probably need a transition to that last sentence too.
DR. COHN: Transition to the?
DR. KOLODNER: To the last sentence, where we do this, this, and this. The witnesses also believe or something.
DR. COHN: And I think we need one final sentence on this paragraph that says the committee is currently developing recommendations for your consideration on how best to proceed.
DR. FITZMAURICE: To advise the government?
DR. COHN: Yes, on how best to proceed.
MR. BLAIR: Did we ever get privacy and confidentiality in?
DR. COHN: No, we haven't.
MR. BLAIR: Maybe a stand alone sentence?
DR. FITZMAURICE: Although many witnesses mentioned the importance of standards for confidentiality and privacy, you will note that this is being addressed in the appropriate subcommittee.
MR. BLAIR: Could I try a different phrase? Because what you kind of tacked onto the end was, but it's not being pursued by this committee?
DR. FITZMAURICE: That's right.
MR. BLAIR: What if sort of reversed the sentence, and we start the sentence out by saying although it was not included within the scope of this work effort, it should be noted that many testifiers pointed out how important privacy and confidentiality legislation will be to progress.
DR. FITZMAURICE: Beautiful.
DR. FERRANS: Legislation and/or regulation?
DR. FITZMAURICE: I would skip regulation.
DR. KOLODNER: Jeff, do you want to try that again? I've got although it was not included in the scope of.
MR. BLAIR: Let's see if I can say it again. Although it was included within the scope of the work effort, many testifiers expressed or indicated the importance of legislation to protect the privacy and confidentiality of health care information as a foundation or prerequisite for further progress with health care standards.
DR. COHN: Good.
DR. KOLODNER: As a prerequisite for -- I don't want to make it too strong. Why don't we say as a foundation. I don't want to say we can't make progress without it.
DR. COHN: We can say the importance of legislation and regulations.
DR. FITZMAURICE: The HIPAA regulations don't have enough scope. And if they pass legislation, there will always be regulations to enforce them.
MR. BLAIR: I wanted federal laws.
DR. COHN: Well, we're stuck in the sense that we're saying to the secretary that we can't do anything until we have legislation, even if somebody comes up with regulations. Maybe, Mike, you can explain to me how the scope is so limited that it won't be protective for this.
DR. FITZMAURICE: Suppose we just put a period after health care information? Many testifiers indicated the importance of legislation to protect the privacy and confidentiality of health care information period.
DR. COHN: Stress the importance.
DR. KOLODNER: Okay, we'll go back up to that part.
DR. FITZMAURICE: That gets us out of the box.
DR. KOLODNER: Are we satisfied with this sentence, although it was not included in the scope of this work effort, testifiers stressed the importance of legislation to protect the privacy and confidentiality of health care information. Now where does that sentence go? So it will be after the other paragraph?
Now do you want to go back up to these advances which?
DR. FITZMAURICE: I like advances.
DR. KOLODNER: But the advances in standards? The problem is that we interposed a sentence.
DR. FITZMAURICE: These advances in standards.
DR. KOLODNER: We actually say first and foremost. So it would accomplish the -- however, the witnesses believed that we need both additional areas to be covered by standards, and more widespread adoption of existing standards. First and foremost, doing this would improve individual patient care, and could dramatically reduce advance events. These advances in standards are also essential?
So just these advances in standards are essential for solving many of the health care system problems. We can debate whether it is one or two words, including and then the list that we had before.
DR. COHN: Maybe we can get rid of the word 'doing.'
DR. KOLODNER: And that's where we were debating this or these.
DR. FITZMAURICE: There are two things above, you are right. Obtaining these needs or satisfying these needs is what it really refers to, but a long phrase. First and foremost, satisfying these needs would -- lengthen an already long paragraph.
DR. KOLODNER: The question goes back to the start, which is good enough versus perfect.
DR. COHN: We'll probably have to come back to it tomorrow. Let's stop the wordsmithing here. It may drive us crazy. And I think maybe Mike could help us or others fresh.
I think that actually this ends the letter. Let's see if there is anything else.
MR. BLAIR: I have another suggestion, Simon. Do you want to not bring that up at this point? That was kind of streamlining the role of the government.
DR. COHN: I actually thought that this may replace all of that. I think that rather than asking a bunch of questions about facilitating or whatever, I think this paragraph actually handles that, because it talks about a possible role for the government. It tells them that we are considering substantive recommendations coming down. So I thought just basically end the letter there.
DR. FERRANS: I do think though, Simon, I agree with you. I do think though if we keep these questions somewhere else, because we are going to have to answer these questions, I think we need to rephrase the questions now to how should the government, and what role. We need to redefine the questions for our own internal process.
DR. FITZMAURICE: But not for them as part of the letter?
DR. FERRANS: Right. But now the phrasing has changed.
MS. FRAWLEY: Also, could we ask the secretary for any feedback? Would that be appropriate in the letter?
DR. FITZMAURICE: That would be appropriate I think, if you want to say we welcome any feedback..
DR. COHN: Questions.
DR. FITZMAURICE: There's got to be a way to say that. We welcome your feedback as we move forward.
MR. BLAIR: Actually, could I suggest that since we are giving her so little information to react to, rather than saying feedback, what we might ask her for is advice and guidance.
DR. FITZMAURICE: But that's what we are supposed to give her.
MR. BLAIR: I was thinking a different kind of advice. She might have advice and guidance from the political position that she deals with in terms of how we might craft or address or focus the report; something like that.
DR. FITZMAURICE: We welcome any comments you would care to make. Let it open-ended. She can not respond or respond as wishes.
DR. COHN: That sounds fine.
DR. FITZMAURICE: We welcome any comments you care to make. If there is a better way of saying it, I will find it.
MS. FRAWLEY: We have one sentence at the end of our letter that says the NCVHS looks forward to delivering a cogent report on time with the recommendations and legislative proposals.
DR. COHN: That's fine. Do you like that?
MS. FRAWLEY: We're going to finalize the report.
DR. COHN: That sounds fine. Let's put that in as the last sentence then.
DR. KOLODNER: So we will take out Appendix 3, which we don't refer to anyplace.
MR. BLAIR: A cogent letter and a cogent report.
DR. COHN: I think the only changes I would make in the attachment are once again to get rid of PMRI throughout it. It's called patient medical record information record areas. Thank you.
MR. BLAIR: What time are we convening tomorrow morning? Is it 8:30 a.m.
DR. COHN: It's 8:45 a.m.
MR. BLAIR: And Betsy will be testifying.
[Whereupon, the meeting was recessed at 5:38 p.m., to reconvene the following day, Friday, September 17, 1999, at 8:45 a.m.]