Medicare: New Directions in Quality Assurance Proceedings (1991)

Thomas G. Morford

I am pleased to represent the views of the Health Care Financing Administration (HCFA) on the Institute of Medicine (IOM, 1990) study. We at HCFA are used to being studied, and I always delight in having the opportunity now and then to give the agency's reaction to some of these efforts. I would like to lay out a brief picture of where we have been and where we are going, at least from the HCFA perspective, with the Peer Review Organizations (PROs).

At the beginning of my involvement with the PRO program four years ago, we identified three major problem areas that were of sizable proportion.

The first was that we lacked fundamental stability in the program, both within the HCFA and within the PROs themselves. Our policy development was somewhat haphazard; our communications were limited and really more focused on internal operations.

The second was what I viewed as a lack of capability in all the PROs. At that time we simply did not have the talent—the capacity administratively or medically—to run some 50 organizations throughout the United States in the manner that we had set out to do and in the manner that the Congress intended.

The third major area that is critical to the issues in the IOM report was the waste and inefficiency of the case-by-case review process. It is not systematic; consequently, it is subject to wide variation and a waste of resources. We pay $300 million a year for nurse reviewers to look at unending numbers of medical records, which in turn have been xeroxed by

the many hospitals and health maintenance organizations throughout the United States. That nurse reviewer, with a massive set of instructions from HCFA, goes through each medical record, makes appropriate notes, and makes all sorts of initial judgments on quality and appropriateness of care. Because there are 50 PROs times "X" number of nurse reviewers, you get variation in review. In addition, no data are captured. When the nurse reviewer reviews a case and the finding is no problem, the review is wasted in terms of clinical information captured for future use, and the case is shredded. Only those cases with a problem or a potential problem really begin to move along in the system. Consequently, we preserve data for only a brief time from the problem cases and pay little attention to good practices.

Seeing those three major problems, particularly the last in terms of the waste and inefficiency, I have been fond of telling the PROs for the last two years that we are not going to spend $300 million a year to catch a few bad guys and process gobs of medical records. The same issue that faced us four years ago faces us today. At HCFA we believe that we have resolved it and are moving the PRO program in a specific direction.

We have developed a very concrete, very particular strategy to change the approach to peer review in the United States. It had to be done in a very directed, cautious manner and over some reasonable amount of time. First, obviously, we had to stabilize our internal operations and to some extent improve the accountability of both the PROs and ourselves. Three or four years ago we did not know what kinds of quality interventions PROs were taking. We had no data system. If a PRO found a problem, we did not know the severity of the problem. Was it a little problem—was it dotting the i's in the medical record? Was it somebody who had actually done considerable harm? We had no way to get at those issues. One of the primary concerns was to bring some order and accountability to the management of the program.

Second, we wanted to shore up some of the PROs. We knew that they needed technical assistance. We did not have the capacity to run the program as intended, and it took some time. I am also candidly not satisfied yet that in all 50 PROs we have that capacity. Nonetheless, from my parochial perspectives, we have done a great deal to accomplish these two objectives: improving our own management and improving the capability of the PROs.

The other thing that we have begun to do is to develop within the PROs the capability and the data bases to accomplish some of the fundamental changes that we have talked about: to focus on the movement of the program from

case-by-case review with its wasted resources to an examination of outcomes and of patterns of care based on clinical data. This is not simply a "blue sky" notion. This is a very concrete plan we laid out about three and a half years ago, and we have been moving steadily on course to accomplish it.

Plans are great, but you have to fasten the nuts and bolts to get there. There is nobody but HCFA able to accomplish this in the United States. We have seen the ashes of the PSRO¹ program; we have gone through difficulties with the PROs; and we have learned that fundamental steps have to take place for these processes to work. The PROs must have the capacity to analyze the data. They must have the infrastructure, the data systems, and the staff capability. They must have reasonably stable management and an ordered scope of work to be able to make these kinds of changes. Otherwise we have another health services researcher's dream, and a great plan with little hope of ever really accomplishing it.

So we have now fastened the nuts and bolts, and we are moving the program. We are developing and testing a variety of analytic tools, data bases, and hardware configurations to support the kind of transition into a more systematic approach to the assessment of quality of care. The objective of all these efforts is to equip the PROs with the skills and the tools to examine patterns of care and outcomes, to draw inferences about performance, and to share that information with the medical community and with the consumer community.

I want to add a point of emphasis here. We should not forget that the consumer movement is a very important part of this. Although we embrace the notion of continuous quality improvement and our plans for the PROs will clearly foster it, there needs to be by statute and by common sense some kind of concern about patient protection. Although we clearly acknowledge that there are only a very few cases of aberrant providers, that most physicians in the United States are good practitioners, and that most hospitals are good, Congress demands, the consumers demand, and common sense demands that we have some protection against the aberrant provider. It need not be the whole program or cost $300 million a year to catch a few bad guys. That $300 million a year needs to be spent in an orderly process to move toward the collection of data and toward the analysis of data and the kinds of outcomes that we have talked about.

We have available summaries of the numerous activities we have undertaken to change the course of PRO review (see appendix to this chapter).

These activities, currently underway within the PRO program, encompass essentially all the IOM recommendations other than those pertaining to oversight and advisory mechanisms.

I want to emphasize one activity that I think is the key to the future of the program: the Uniform Clinical Data Set. Contrary to some popular rumor, we have developed it with extensive advice and assistance from academicians and health services researchers, as well as practicing physicians. This project was started two and a half years ago, and we are in the final stages of field testing it right now. We will begin to implement it in the PROs in the fall. This project forms the basis for fundamental changes in the PRO program. PROs will abstract all of the hospital records they review under current program requirements according to well-documented rules and procedures. That is roughly 20 percent. We are going to capture the data from the medical records.

The data that are collected have two applications. First, to achieve greater consistency and accuracy in PRO review, the abstracted information will be screened by a computerized screening expert system that will identify cases to be reviewed by PRO physicians. These are two critical points. We are going to make the review of the cases consistent and stable. We are going to try to eliminate much of the independent decisionmaking by the nurse reviewers in interpreting our massive instructions. We are going to computerize this front-end process, but we are not going to have the computer make the medical decisions. The purpose of the computerization is to have a uniform, consistent application and then to take the cases as they are appropriately identified to the peer reviewers.

Second, and obviously at least equally important, when linked to currently available claims data the abstracted clinical data will establish an epidemiologic data base that will enable the PROs and HCFA to characterize patterns of care, adjusted for sociodemographic characteristics and risk-adjusted patterns of outcomes, and provide information to the medical community and to the consumer community. This is the data base that we have talked about. This is the data base that everybody would have loved to have had for the last 25 years, and I guess, to put it bluntly, we are going to do it because I am sick of talking about it, and I am sick of listening to the health services research community say, "Gee, it would be great to have the data, then we could analyze the outcomes." It is not going to be perfect by any means, but we are going to use the PROs to abstract the data and develop these sizable data bases. We will be able to share them across the country, with specialty societies for standard setting, with practitioners, and with the health services research community.

In this context, two principal advantages of having the PROs do this are (1) they are reviewing about 20 percent of 12 million discharges in the United States, and we are not going to let that data simply be wasted; and

(2) they are uniquely situated around the country. They are state-based. If they are managing their organizations properly in accordance with the scope of work, they will have strong local roots and be able to interact with the medical and the consumer communities within those states. We are going to provide them the data to be able to do that.

Just to name a couple of other projects, we have engaged with seven PROs to design an approach to the assessment of quality of care in the noninstitutional setting. We are supplying the PROs analyses of small area variations across the country. We have contracted with the Medical College of Wisconsin and with the Wisconsin PRO to develop a data management and analysis infrastructure to enable the PROs to analyze existing large Medicare claims data bases as well as those other data bases that are emerging. We are currently negotiating a project through the New Hampshire PRO with the Dartmouth School of Medicine to develop the analytic software and supporting hardware for PROs to use in assessing quality of care provided in the area of internal medicine. Those are just several examples, I say with some pride, of an extensive laundry list of the kinds of things that we are in the process of developing.

We are moving the program to where we think it ought to be; to where we have heard the research community say it ought to be; to where we have heard organized medicine, as well as practitioners, say it ought to be. I conclude by saying that I appreciate the Institute of Medicine study's having endorsed the plan we unveiled about three and a half years ago. I think it will continue to supply us momentum and the kind of support we need to make these kinds of improvements.

Institute of Medicine. Medicare: A Strategy for Quality Assurance. Volumes I and II. Lohr, K.N., ed. Washington, D.C.: National Academy Press, 1990.

Listed below are seven current activities that support a transition from the current Peer Review Organization (PRO) program to a more systematic approach to assessment of the quality of care. The objective of these efforts is to equip the PROs with skills and tools to enable them to characterize patterns of care and patterns of outcomes in their jurisdictions and, from their correlations, draw inferences about patterns of performance of providers of medical services. For PROs to be able to perform these functions they need an adequate and cost-effective data management infrastructure and the epidemiologic and biostatistical skills to make effective use of the data.

1. We have developed, with extensive advice and assistance from the academic and practicing medical communities, a Uniform Clinical Data Set. This project has been underway for two and a half years and forms one basis for our proposed changes in the PRO program. PROs would abstract, according to well-documented rules and procedures, all of the hospital records they review under current program requirements. The data so collected have two applications. First, to achieve greater consistency and accuracy in PRO review, the abstracted information would be screened by a computerized screening "expert system," which would identify cases to be reviewed by PRO physicians. Second, the abstracted clinical data, when linked to currently available claims data, will establish an epidemiologic data base that will enable the PROs and the Health Care Financing Administration (HCFA) to characterize patterns of care, adjusted for sociodemographic characteristics and risk-adjusted patterns of outcomes.

2. We are currently engaged in a complementary project involving seven PROs to design an approach to the assessment of quality of care in the noninstitutional setting. Here also, the objective is to develop a methodology to characterize patient populations, the patterns of care they receive, and the effects of the care on their health, by recognizing the need to span periods of time and constellations of services. The project focuses on 16 medical areas and will enable PROs to evaluate trends and variations in effectiveness among interventions and among providers to offer useful feedback and stimulate continuing improvement in care. This project began in May 1989 and will last three years.

3. In 1987, we began a $2.6 million project with the American Medical Review Research Center and 12 PROs to use the small-area analysis tool developed by the Codman Research Group and John Wennberg. The project provided PROs with data and software to analyze variations in service use and outcomes in hospital market areas in their states. The objective of this

project was to begin to develop in the PROs the capability to feed back data on patterns of care and of outcome to the provider community and to evaluate the impacts of such an educational activity on practice.

4. We have contracted with the Medical College of Wisconsin, through the Wisconsin PRO, to develop a data management and analysis infrastructure to enable the PROs to analyze the existing large Medicare claims data bases as well as the emerging clinical data bases. The hardware configuration has been designed, the software is being written at this time, and the testing of the system in four PROs will begin shortly.

5. We are currently negotiating a project with the New Hampshire PRO and the Dartmouth School of Medicine to develop epidemiologic and analytic software and supporting hardware for PROs to use in assessing care provided in the area of internal medicine. The products here, as in the previous project, would be made available to PROs and would be in the public domain.

6. We are currently in the process of producing analytic information for PROs on 38 different procedures and/or diagnoses. The information would be analyzed nationally and by state, Metropolitan States Statistical Area, county, and hospital market area. We will present mortality, readmission, and expenditure information. The information would be risk-adjusted for a variety of demographic, socioeconomic, and patient characteristic variables. This is another example of the kind of analysis that PROs could eventually conduct for themselves with software that HCFA would provide.

7. We have undertaken several collaborative analytic efforts with academic medical centers making use of the HCFA claims and clinical data bases. The intent is to stimulate greater academic involvement with the PRO program and bring to it the biostatistical and epidemiologic skills necessary for a successful transition of the program. These efforts have involved the University of Maryland School of Medicine, the University of Pennsylvania College of Medicine, the Dartmouth School of Medicine, Boston University Hospital, and the Medical College of Wisconsin. The latter two have resulted in a joint effort with the PRO of Wisconsin (project 2 above). In addition, proposals have been received from several PROs to conduct evaluations of the quality of care in collaboration with Northwestern University, the Harvard School of Public Health, the University of Minnesota School of Public Health, the University of Michigan School of Public Health, and the RAND Corporation.