TESTIMONY OF NELDA McCALL
Thank you for the opportunity to testify on the important issue of health information. My name is Nelda McCall and I am a health services researcher. I have been involved in using health data information systems for more than 25 years.
Besides being a user of federal Medicare data systems, I have also worked with many state data files, including Medicaid fee-for-service data and encounter data, private insurer's data, and participated in the development of a uniform long-term care insurance database for the Robert Wood Johnson Partnership for Long-Term Care, which collects uniform data across the four participating states for all participating private insurers on purchasers, assessment, and service use and cost.
I am going to focus my remarks today on:
USES OF STANDARDIZED DATA
Despite the difficulties of negotiating and collecting uniform data it is of paramount importance. It is important because it is critical in five areas:
All of these are critical to a well-functioning health delivery system. I am going to say a few words about each of these areas in turn.
Utilization and Access Monitoring
Analysis of the utilization of medical services is a critical component in understanding how a medical care program is performing. Service utilization has a direct influence on cost and an equally important relationship with quality. Several tools are available to help assess if there is appropriate utilization of services and access to care by analyzing the specifics of what services are provided and investigating potential problems with access to care.
Analyzing the Specifics of What Services are Provided - Services delivered can be monitored for a particular time period for covered beneficiaries by any of the data elements coded or by any data elements that can be linked to these data elements. These include analyses by type of service, by provider, by characteristics of the beneficiary, and by characteristics of the provider.
Monitoring the Adequacy of Ambulatory Access - Data from hospitalizations can be used to examine whether there appear to be problems in access to care. In the last few years, researchers at New York and Columbia Universities (Billings, Teitel, Jukomnik, Carey, Blank, and Newman, 1993) have developed lists of hospitalized conditions that should occur or should occur infrequently if access to ambulatory care is adequate.
Financial Analysis and Rate Setting
Data are of critical importance in the financial analysis and appropriate rate setting. Utilization is a substantial element in costs and data concerning utilization is important to improve market competitiveness. These data are also required for the establishment of appropriate rates as they can be used to set rates which take account of the health status of the enrollees. Several mechanisms currently exist to control for risk selection, a critical concern in public programs. The two most prominent are those put forward by researchers are Johns Hopkins and Boston University. Ambulatory Care Groups (ACG) is a case adjustment system developed at Johns Hopkins University (Weiner, Starfield, Steinwachs, and Mumford, 1991). Diagnostic Cost Groups (DCG) is a system developed at Boston University (Ash, Porell, Gruenberg, Sawitz, and Beiser, 1989).
Quality Assurance
Assuring that high standards of quality are maintained requires the ability to analyze data in detail. These analyses require individual service specific encounter level data. Quality assurance analyses which are of particular importance include:
Analysis to Detect Underutilization of Services - The use of data permits analysis which can pinpoint underuse of services. One important measure which can be examined is the percentage of individuals not having a physician service in a year by age group. Percentages can be compared across providers and with external data. Rates falling substantially below comparison data highlight areas for review. Certain procedures/hospitalization rates can also indicate underuse of services. In addition, data can also be examined to identify differences in mortality and morbidity which may be associated with system problems. These include deaths which are possibly avoidable (Dubois and Brook, 1988) and hospitalizations for conditions which may be preventable (Weissman et al, 1992).
Review of Treatment Patterns by Diagnosis - Utilization data can provide profiles of the treatment given for specific diagnosis. These can be aggregated for the program by provider and compared with internal or with external data. Providing appropriate treatment for medical conditions is the core of medical care. Review of the actual type of services received for particular conditions and the outcomes achieved is of substantial interest. The formulation of standards of appropriate treatment begins with a review of patterns in place.
Monitoring of Selected Procedures - The frequent occurrence of certain procedures may signal areas of quality concern. Utilization data permit this kind of review. Data can be arrayed for individual providers, for aggregations of providers, or program-wide.
Fraud and Abuse - Review of data can point out areas of concern for further investigation concerning fraudulent or abusive activities.
Physician Profiling - Utilization data can also support the development of physician profiles. Physician profiles can be used by plans to measure and reward productivity as well as to provide the basis for the development of quality standards.
Program Planning
Data of this type can be used to satisfy external information requests, estimate the costs of program modification, estimate future costs, identify activities which promote long-run cost containment, and research areas for general study.
Societal Uses
Data are of critical importance more generally to develop information which can
These are perhaps the most important reasons for data collection as these data provide the source of information for outcome analyses important to chart the future course of health care delivery.
THOUGHTS ON PROMOTING STANDARDIZATION
With respect to the standardization of collected information, there are some very important considerations. The current standards are generally based on data processed by the Medicare program and to a lesser extent by the Medicaid program (where many of the standards are not national) and these provide some standards related to services covered by those programs. What appears to be needed is broader thinking about a logical structure for all health data which looks beyond provider types while at the same time obtains the buy-in of all the effective parties (i.e. consumers, providers, and users).
As with so much of the health delivery system the problems of data integration are related to the ways we pay for care that provides one set of rules for care delivered in the hospital and that delivered on an outpatient basis, for care delivered for acute services and for chronic long-term care. Integration of service delivery systems could substantially improve our ability to standardize collected information.
In thinking about how to implement these standards, determination needs to be made of the balance between building upon existing standards as compared with defining new data elements that would provide better, more outcome focused, and more clinically focused data. While it is important to continue working on the definition of new and improved data elements for collection, it is also important to consider that the standards we have in place do come at least part of the way in satisfying societal needs. This is especially the case in the area of the collection of encounter data in public programs where some have argued that we should give up on collecting utilization data in the form that it is already available in the fee-for-service system (UB 92s and 1500s), while forgetting that the data currently being collected at least by one managed Medicaid program (Arizona) goes at least part of the way in satisfying the societal needs. Not that improvement in data elements collected should not be promoted but throwing out existing data collection standards should be approached carefully as it can result in no data, rather than improved data collection.
The standardization that should be taken on should be broad based and include standards for long-term care insurance policy data collection. The sooner that discussion can begin on standards in this area, the sooner that such data collection can begin. Of importance in this area in addition to knowing information on the characteristics of those who have insurance (enrollment) and on service use and cost, data should also be collected on assessment for benefits process.
In the next decade more and more of our health resources are likely to be spent on beneficiaries in need of chronic long-term care, and the standards developed to monitor how and when care is delivered is of critical importance. We have been part of the development of such a system for the Robert Wood Johnson Partnership for Long-Term Care and would be happy to provide more detailed information on that system's structure and operation.
Thank you for the opportunity to testify. I am happy to take any questions.