National Invitational Conference on Long-Term Care Data Bases: Conference Proceedings. D. Other Long Term Care Data Sources


Aurora Zappolo, Health Care Financing Administration

MARY HARAHAN: We thought it would be useful to have some discussion about other data bases which may not be so large or so complicated in-scope, but which do have relevance for long term care policy.

AURORA ZAPPOLO: The data bases that are being discussed or at least the three primary national ones, are specifically designed to address long term care issues from a national perspective.

They are of special interest because they provide current information on functionally impaired elderly people living in nursing homes and in the community.

A number of important questions can be answered by these data bases, especially relating to the magnitude of the populations that we are looking at.

However, with recent policy debates on various aspects of long term care, I am increasingly hearing that these surveys do not answer all of the questions. In particular, the kinds of topics that people say they can not find answers to are duration of nursing home stays, outcomes of such stays, the likelihood that people living in the community will need nursing home care at some point in the future, and the point in a nursing home stay at which a person depletes his or her assets and becomes a Medicaid recipient.

I have two basic points to make in answer to these concerns. The first is that some questions are better addressed at state and local levels, and can never be answered adequately by a national survey. The second is that there are studies that are already done that we just need to become more aware of. The gap is often in our knowledge about data bases, rather than in the data bases themselves.

Some of what I am going to refer to are data bases that are in progress right now and are not available for that reason.

There are a wide range of what I am calling "sub-national studies" that have been conducted at various levels, whether state or local. There are also studies in other countries that we should be looking at.

I do not in any way intend to give anything more than examples of some of these data bases that you should be aware of. I do not know where to look for a complete inventory.

Let me start with national studies. My talk originally was going to be only about sub-national studies, but it became clear as we developed the agenda that there were just some studies that should be mentioned. Those that you have heard this morning are important examples of other national studies to be aware of.

In particular, there are questions of duration, outcome, likelihood of institutionalization, Medicaid spend-down, and utilization of health resources.

We have talked at various times about the need to model long term care insurance options, for example, and measuring the magnitude of federal responsibility for nursing home coverage.

The common characteristic of many of the questions is that we need repeated measures of the same individual over time. Jennifer Madans made reference to the increasing attention being given to longitudinal surveys.

One of the major efforts going on right now that you should be aware of is NMES. NMES is collecting a full year's data on the functional status, health services utilization and expenditures not only for persons living in households, but for those living in nursing homes and in facilities for the mentally retarded.

Repeated observations of the same individuals over time will give us measures of change in functional status, admission to nursing homes and hospitals, discharge from those places and the length of time in a nursing home before someone spends down to Medicaid. Another unique aspect of NMES is in its collection of data simultaneously from persons in households and in institutions, which allows comparisons based on the same time period and using identical or similar questions.

NMES is going to be a rich data base, and although it will not be available for a few years, I think that most or all of the current questions will still be with us at that time.

The ILTCP, which Curt Mueller referred to, is the sampling frame for the institutional part of NMES, and it is important as a data base in itself. It broadens the universe of long term care facilities that we are looking at one point in that it includes facilities for the mentally retarded, as well as nursing and personal care homes.

It does not include all long term care facilities, however. Facilities for the mentally ill and the chronically physically impaired are not included. Nevertheless, it provides valuable information on the supply of services that are available according to standardized characteristics. An important point to keep in mind with the inventory is that it is a census, and because it is a census we can look at state and local level data without the limitations that come from sample surveys.

The importance of data bases on facilities used for long term care is in the analysis of supply issues. Studies which examine utilization of health resources, for example, without regard to the availability of those resources, the kind of picture you get on a national scale, can be misleading. This is especially true for those studies that are conducted in areas in which the population in that area has few alternatives for the services that they need.

Another national survey that is being developed which you might want to know about is the National Mortality Follow-Back Survey (NMFBS).

This survey, which is being done by the NCHS, includes information on utilization of nursing homes during the last year of life.

We also should become more aware of those studies that are done by the Veterans Administration (VA), because they have done a great deal of work, especially in terms of projections of the elderly.

In addition to surveys there are two special categories of national data that I just want to mention. They are relevant to long term care, even through they are not designed to be long term care data bases.

The first of these is the Decennial Census. Although the census is not designed as a long term care data base, every census has data on people living in institutions. Just as NMES on a sample survey basis will allow us to compare those in institutions with those in households, every census lets us do that.

In addition, Census is considering or has decided, I am not sure where we are at the moment in this, that in the 1990 census there will be a pair of questions that will collect the functional status of the population.

The two questions being considered are global. The first is like ADL, asking whether a person needs help in bathing, dressing or getting around inside the house. The other is an approximation of IADL, asking whether he need help in shopping, housework or getting around outside the house.

Although these items are general and they do not differentiate the kind of impairment, they are going to be the first data available on such a large scale on functional dependencies. They will provide bench-mark data from which we can access changes in the proportion of functionally impaired in different settings.

The final category of national data that I want to mention are the various administrative data systems at HCFA.

These are not available in the form of data tapes. But they are disseminated in statistical reports for the agency.

The Medicare statistical system is composed of three administrative record systems: the Health Insurance Master File (HIMF), the Provider of Service File (PSF) and the Utilization File (UF).

The HIMF contains a record on each person enrolled in Medicare, showing age and other basic information. The NLTCS was designed based on that file, or it was sampled from that file. The PSF describes hospitals, nursing homes, home health agencies and other providers who are approved to give care to Medicare beneficiaries. The UF is based on the billing records, which includes dates of service and amount billed. Since the advent of the Prospective Payment System (PPS), we also have diagnostic detail on 100 percent of the hospital bills. The combination of these three files allows both population-based and provider-based analyses. Like the Decennial Census, these data are important as overall measures of the population.

In the case of the Medicare files, they represent most of the elderly in the country and their use of covered health services.

I would like to move into the area of state data sets. Although we have national data on the Medicare population from administrative records, there is no similar data set on the Medicaid population. Consequently, the project known as Tape-to-Tape was developed at HCFA.

This project is a major effort to produce person level data on the Medicaid population; and it is in five states, California, Georgia, Michigan, New York and Tennessee. The data will eventually cover 8 years of enrollment claims and provider information which can be compared from one state to another.

One of the projects that is currently underway is directed specifically at the spend-down question, the point at which someone becomes a Medicaid recipient. SysteMetrics is developing tapes which will be used for a series of papers comparing the date of admission and the date of coverage on Medicaid. Data from the spend-down study will be available this year. The data tapes will not be available generally, in this case, because the agreements with the States require prior approval. Another part of the Tape-to-Tape analysis, in fact, that came out this week in the American Journal of Public Health, focuses on the oldest old and you might want to take a look at that article.

The development of the Tape-to-Tape project is important in health services research not only for the data it produces but for its methods, which brings me to the next part of the talk.

One of the difficulties in using most national surveys for studies of the Medicaid population is that the sample designs do not permit making specific estimates for specific states.

The cost of data collection on a national basis simply precludes the inclusion of the number of cases which would be necessary. Often, the fact that state differences are smoothed over is overlooked in analyses. In studies which focus on Medicaid recipients, the use of national estimates is actually less meaningful than the use of state estimates. Since Medicaid is a state-administered program and, in fact, varies by state in the services that are provided, it is important to recognize these differences.

Such studies should be based on the smallest geographic unit on which the data are available. Of course, even state level data can smooth over differences between urban and rural areas.

In recognition of the need for state level data another agency is emphasizing this--NCHSR.

Currently, they are assisting selected states in the development of data bases to examine spend-down. It is part of an overall project to encourage a public/private partnership in the development of financing option for long term care.

Many states have already collected and analyzed data on their long term care populations either through administrative mechanisms or occasionally through survey mechanisms.

Connecticut is one that I became aware of that has a very interesting data set. The Department of Health Services has a comprehensive longitudinal data base showing the characteristics of nursing home residents. Data are available at admission and on a fixed report date, which is the discharge date, when that is appropriate.

Additional data on prior nursing home stays and whether discharged to another health facility for each individual allows more accurate analyses of length of stay, outcomes and episodes of institutionalization than is possible with any national data base that is currently available.

Furthermore, the identification of source of payment at admission and discharge data allows analysis of Medicaid spend-down in the context of Connecticut's Medicaid program.

There is a directory of agency heads and contact persons for designated state statistical agencies. The directory was put together by NCHS as pan of its ongoing efforts to exchange information with state governments. You might want to contact the state government that you are working with, or that you are interested in collecting data on, to explore the availability of data bases.

Another source of state level data is, of course, university research. A study I would like to mention that is going on right now and has been going on for a while is at the University of Maryland. The study represents all licensed nursing homes in the state and it is a project that is funded by the National Institute on Aging (NIA). The data base is stratified to represent nursing homes at all levels of care. Data are collected from patient records within the sample nursing home similar to how the NNHS does it.

Let us look at local areas. A variety of federally-sponsored studies are available on local areas. One that is of particular prominence are the studies at NIA, especially the Established Populations for Epidemiological Studies of the Elderly (EPESE).

Richard Suzman from NIA has a handout describing all their various studies.

Let me just mention what the EPESE study is about. It was begun in 1980. These epidemiological studies were developed in three communities originally, East Boston, Massachusetts, Washington County, Iowa and New Haven, Connecticut. In 1984, a fourth community was added in Durham, North Carolina.

The project is designed to produce estimates of chronic conditions and impairments among the institutionalized elderly and eventually, over a period of time, to develop predictors of mortality, hospitalization and admission to nursing homes.

The project, I think, is going to be an important one in years to come. At this point, the only data that are available are the original cross-sectional data. If you do want to find out what is in that data collection activity, you can get a copy of the baseline data book from NIA.

This is another case in which data tapes are not available, but there is a mass of cross-sectional data that is published already.

At HCFA, the Office of Research and Demonstration is another federal agency that collects or conducts studies in smaller geographic areas.

Currently, there are over 300 studies going on that are research evaluation or demonstration projects related to the federal programs. Many of them relate to long term care, not necessarily under that name.

There are studies of nursing home case mix, home health, hospice services, the impact of prospective payment on nursing home utilization, and beneficiary awareness studies regarding their health insurance options.

The Health Care Financing Status Report, a large red volume, describes all of the extra-mural projects and many of the intramural projects that are currently under way. If you want a copy of that report, you can purchase a copy through the NTIS.

You might also want to look at the Health Care Financing Review. One of the issues, in particular, includes an article by Candace Macken describing the 1982 NLTCS.

The NCHSR is another federal agency that sponsors studies in smaller geographic areas.

They have both extra-mural and intramural research. Some of their intramural research activities have focused on such topics as the cost and economic implications of informal support, the size and sources of long term care expenditures, and the feasibility of alternative financing mechanisms, such as long term care insurance.

Other federal agencies produce data related to long term care. The important thing that I want you to know is that the place to find out about all of them is NTIS. NTIS is an archive of federal statistical information and it includes both data tapes and reports.

I wanted to also mention data bases from other countries.

Data from other countries can fall into one of two categories, and, in a sense, it is how you choose to use it. It can be cross-national, that is, comparative, or it can focus on a particular country.

The concern for the increasing size of the elderly population is not limited to the U.S. It is a worldwide concern and consequently there is a lot of research activity going on in many of the western nations.

Two examples of countries that have produced extensive relevant analyses of these populations are Canada and Sweden.

The Canadian study that I want to mention is a longitudinal study of nursing home admissions in the Province of Manitoba. It was initiated in 1974. There are four levels of nursing home care that are available there. Three of them represent levels that seem to be like our skilled and intermediate care. The lowest level of care, though, provides service that, as I interpret it, is more of a personal care service.

I think that fact alone could make it an interesting comparison for us to look at, suggesting the implications of an additional care level.

Some comparisons that have been noted by the Canadian researchers are that Manitobans enter nursing homes later than their American counterparts and they stay longer after that.

Turning our attention to Sweden, there are a number of important epidemiological studies that have been going on for some time. Most of the research that I am aware of is more in the area of epidemiology rather than health services research. However, they are moving in this direction. As you know, Sweden has a very generous national policy regarding services for the elderly and chronically impaired.

A recent law requires local jurisdictions to reduce the population in long term care institutions and prevent further institutionalizations, all under a very tight timetable. This has resulted in dramatic changes in their delivery system.

Many of these changes are still going on, and a number of studies are focused on looking at those changes. In Sweden, it is interesting to note that there is much less emphasis on national estimates than we have here.

A study which some of you may have heard of is the Longitudinal Study of the Elderly in Gothenberg, which was begun in 1971. The study collected information on personal and health characteristics of a cohort of elderly using personal interviews and physical examinations. A somewhat similar study is being developed in Lund this year.

I was fortunate last year to have the opportunity to visit long term care providers in Sweden and I think that one of the most important and interesting lessons I learned is that international research can tell us a lot about our own system.

I was asked frequently to explain some aspect of the U.S. long term care system, and usually it was something that was either difficult to explain or that gave me an insight into our own system once I thought about it.

We can learn both from other nations' views of our system and from a fresh perspective based on distancing ourselves from old habits and patterns of thinking.

I might add the same kind of fresh perspective can be gained from looking at other aspects of long term care, other than nursing homes and the elderly in the development of the part of NMES that is on the facilities for the mentally retarded.

Again, there were insights into our long term care system for the elderly as we learned about what is going on in the service system for the mentally retarded here in the U.S.

I told you at the outset that the examples I am giving are only examples for illustration. The key point to remember is to look beyond the large national data bases for other sources of information.

I have noted some of the directories. I want to repeat the key ones that you should know about. NTIS, as I said, houses most studies sponsored by the federal government.

There is the Health Care Financing Status Report which describes current research projects. There is no particular vehicle for showing all of the completed ones. That is why I draw your attention to the current one, because then you can track them, You can see when they should be completed and then find out what is available on them.

There are two others. The HHS Data Inventory, which identifies data bases throughout the Department, including a brief description and a contact person.

Again, you might want to look at those that do not particularly focus on long term care or the elderly to see where there is information that we can use in regard to long term care research.

The other collection of studies is the National Archive of Computerized Data on Aging. This is conducted by something called the Inter-University Consortium for Political and Social Research. They are located at the University of Michigan, and they have funding from the NIA to have an ongoing archive of data bases.

Universities that participate in the Consortium get data tapes free from the Consortium. However, anyone can contact them and I do not know what the charges are, but I gather that they are nominal to get information on these studies.

This particular archive includes not only national data bases but sub-national data bases. It also includes not just federally-funded projects but from any funding source.

There is no single entity, as far as I know, that identifies all long term care projects without regard to sponsorship. Clearly, we need a complete inventory of existing data bases, including published research reports and identification of those for which data tapes are available. Until such an inventory is available, it will be up to each of us to discover the less recognized data bases.

I think another important need is to synthesize and integrate the information that we can get from these various data bases.

I would like to close by repeating my theme that the answers to many of our questions are available if we increase our awareness of existing data bases and recognize that national studies are not necessary for all purposes.

With the abundance of long term care studies that have been done, we have a wonderful opportunity for researchers to discover new relationships by integrating the findings from different studies.

CHRISTINE PATTEE: I am the long term care person from Connecticut and I realized after Aurora gave this nice little summary of our data system that we are not listed in the Connecticut Data Center Contact. So if anyone would like to hear about the Connecticut data system, and we would be happy to share information with you.

View full report


"87cfproc.pdf" (pdf, 2.52Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®