Skip Navigation U.S. Department of Health and Human Services
Agency for Healthcare Research Quality
Archive print banner
Future Directions for Residential Long-Term Care Health Services Research

This information is for reference purposes only. It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work. Persons with disabilities having difficulty accessing this information should contact us at: Let us know the nature of the problem, the Web address of what you want, and your contact information.

Please go to for current information.

Expert Meeting Summary (continued)

Medicare Current Beneficiary Survey (MCBS)

The MCBS, sponsored by HCFA, has existed in the field since 1991. The purpose of this survey is to develop reliable information on Medicare and non-Medicare covered services and costs, to study transitions in beneficiaries' lives, and to monitor the effect of change on the Medicare beneficiary population. Data are collected for beneficiaries living in the community and in health care institutions, and are supplemented with Medicare claims data.4

There are three rounds of data collection per year; data are collected using CAPI methods. Persons are followed for 4 years, from setting to setting, to capture 3 calendar years' worth of health care use and expenditures, and transitions in care. The design is an overlapping panel with an over-sample of the disabled and oldest old. Persons are sampled from the Medicare rolls and after the first round; cost and use of services questions are asked about the period since the last interview.

An MCBS "facility interview" is conducted for persons who live in a residential health care setting; it contains no over-sample of residents in these settings. The "facility interview" collects person-level data as well as facility characteristics. Since the 1997 redesign, the MCBS facility instrument has closely mirrored the MEPS NHC instrument, collecting data on health care use and expenditures, use of services and prescribed medications, and transitions in care and MEPS/MDS health status information, except for the year-round measures for infections, fractures, pressure ulcers, and hospitalizations. There are no staffing data; however, linked Medicare claims data are available. The potential also exists to link MCBS persons (residing in a HCFA-certified facility) to their MDS health status data.

The sample size for 1997 was approximately 12,000 beneficiaries: 10,400 resided in the community for the entire year; 1,000 resided in a long-term care setting all year; 200 made a transition between the community and a residential setting and were administered both the community and facility interviews; and 240 had short institutional stays, but only were given a community interview since the long-term care stay occurred in the time period between two community interviews. Therefore, the same level of institutional information is not available for the latter group, but these persons can be linked to skilled nursing facility (SNF) claims data.

4. See Adler GS. A profile of the Medicare Current Beneficiary Survey. Health Care Financing Review, Vol. 15, No. 4 (Summer 1994):153-63 for additional details.

National Nursing Home Survey (NNHS)

This NCHS-sponsored survey, periodically in the field since 1973, is currently being assessed and may incur many changes. In 1999, it included a sample of certified or licensed nursing homes (about 1,500) and a sample of residents and discharges selected in those facilities. Two components have been consistent—the current resident and the facility characteristic surveys. Usually a discharge survey is done, although it was not done in 1995 because an expense component was conducted to capture revenue and expenses of the nursing home. This was fielded with a monetary incentive to improve expense questionnaire response rates; however, no differences were seen across the incentive treatment groups. The resultant expense data had a response rate lower than 70 percent and was produced for public use but not analyzed by NCHS. Earlier iterations of the NNHS have included a nursing home staffing questionnaire and next-of-kin community data collection, although these have not been in the field since 1985.5

Census Bureau interviewers collect the information by talking with staff who rely on records; the instrument is a paper questionnaire. In each facility, information is collected on six current residents and six discharges. The survey includes information that is collected essentially at one point in time. Therefore, the survey provides distributions of resident health and facility characteristics but is not designed to provide outcome measures, which need more than one point of measurement.

5. For details on the NNHS, see Garbel CS. An overview of nursing home facilities: Data from the 1997 National Nursing Home survey. Advance data from vital and health statistics; no. 311. Hyattsville (MD): National Center for Health Statistics; 2000. Strahan G. An overview of nursing homes and their current residents: Data from the 1995 national Nursing Home Survey. Advance data for vital and health statistics; no. 280. Hyattsville (MD): National Center for Health Statistics. 1997. Gabrel C, Jones A. The national nursing home survey: 1995 summary. National Center for Health Statistics. Vital Health Stat 13 (146), 2000. Also go to

Online Survey, Certification, and Reporting System (OSCAR) and Minimum Data Set (MDS)

OSCAR data are collected, at least annually, by State agencies to license and certify nursing facilities. The data are entered into databases at the State level and then transmitted, in a standardized format, to HCFA. Information collected includes facility characteristics (including special care units), staff information (employee and agency), and aggregate health status resident information and deficiency information. Data are kept only for 1 year, except for aggregate resident characteristics and deficiency information, which are kept for four inspection surveys.6

MDS data come from the resident assessment instrument that is mandated for all federally certified nursing homes.7 Nursing homes electronically transmit this person-level health status data to State licensing agencies where the data reside. The States transmit MDS data to HCFA (in a standardized electronic format), where they are also kept in a national repository. Data are used to monitor quality and validate payments. MDS data is not collected for persons residing in non-certified facilities, although assessments are required for all residents in a certified facility. MDS assessments are made upon admission—quarterly, annually, and when there is a significant change in health status.8

HCFA is performing some validation work on the quality of the MDS data, but this is still in a developmental stage. In addition, HCFA is developing plans for standard reports and linkages to other data, such as to the MCBS and other public use files. Also, a facility ID that remains the same from year to year is being developed to help identify facilities over time. This would facilitate linking MDS persons to OSCAR and claims data. At some point, prescription drug information will be available for residents in certified nursing homes.

MDS data is individually identifiable and therefore Privacy Act protected. Access to data in the National Repository (which now contains over a year's worth of data) requires that the principal investigator submit a completed Data Use Agreement and study protocol to HCFA for approval prior to data access. A users' guide for the MDS resident assessment instrument is available at the HCFA Web site,

In the discussion of the MDS data, a participant cautioned that it is difficult to match Medicare numbers when trying to match MDS data to enrollment files and claims data.

6. A large number of the OSCAR data is available for annual cross-sections that are released by HCFA on the Provider of Service Public Use File. Additional detail on the OSCAR system is available at

7. Estimated to be 95 percent of the nursing home universe (Rhoades J, Potter DEB, Krauss N. Nursing Homes—structure and selected characteristics, 1996. Rockville, MD: Agency for Health Care Policy and Research; 1998.) MEPS Research Findings No. 4, AHCPR Pub. No. 98-0006 or go to

8. For details on the MDS assessment instrument see Morris JN, et al. Designing the national resident assessment instrument for nursing homes. The Gerontological Socity of America, Vol. 30, no. 3; 1990.

Return to Contents

Institutional Consumer Assessment of Health Plans Survey (CAHPS®)

CAHPS® (Consumer Assessment of Health Plans) is an easy-to-use kit of survey and report tools that provides reliable and valid information to help consumers and purchasers assess and choose among health plans. The kit contains a set of questionnaires with questions that ask consumers about their experiences with their health plans, sample formats for reporting results to consumers, and a handbook to help implement the surveys and produce the reports.

The CAHPS® tool goes beyond statements of overall satisfaction by measuring and reporting on consumer experience with specific aspects of their own health plans that are the basis of satisfaction. AHRQ, HCFA, and the CAHPS® Consortium are in the process of developing a CAHPS®-like tool to measure quality of life and satisfaction information from residents in institutional settings.

The CAHPS® institutional initiative will focus initially on nursing homes. The institutional CAHPS® questionnaire is in the developmental stage, and an expert meeting was held on June 2, 2000. The meeting focused on how, and whether it is feasible, to collect quality-of-life and satisfaction information for residents in a nursing home setting.

National Long-Term Care Survey

This survey, funded most recently by the National Institute on Aging, has a longitudinal design and collects cross-sectional health information. Data were collected for 1982, 1984, and every 5 years since. The institutional sample includes about 1,600 residents (although the 1989 sample included about 1,300). Interviews run for about 30 minutes, and information is collected from staff and sometimes next-of-kin. Health status and source of payment data are collected. Persons (in the community and health care institutions) are sampled from the Medicare Beneficiary rolls, and those with functional limitations are over-sampled. Recently, the survey expanded to include genetic aspects of aging and disease, including the collection of blood and saliva samples and information about next-of-kin.9

9. See Manton KG, Corder LS, Stallard E. Estimates of change in chronic disability and institutional incidence and prevalence rates in the U.S. elderly population from the 1982, 1984, and 1989 National Long-Term Care Survey, Journal of Gerontology, 1993: 48:S152-S166. Also see "Informal Caregivers Supplement to the 1999 National Long-Term Care Survey" at

Survey Integration

Survey integration is a Department priority, and integration of long-term care surveys is part of this process. Survey integration activities have been encouraged by the HHS Data Council, a group made up of representatives from each of the Department's agencies who meet regularly to discuss data and survey issues. The Reinventing Government Initiative, Part II, and the identification of data gaps and data redundancies during health care reform efforts also have heightened the importance of survey integration.

The general goals of survey integration are to reduce redundancy, make data more current, improve the ability to link data, identify critical data gaps, reduce respondent burden, and reduce conflicting estimates between national surveys. Another focus is the dissemination of results to policymakers, providers, and consumers.

An overall plan to attain this goal is being developed. Accomplishments of HHS Survey Integration include the integration of the MEPS household component and the National Health Interview Survey (NHIS) sample frames, and coordination of questions across surveys.10

10. For additional information, see the HHS Data Council's Web site,, and the HHS Survey Integration Plan

Summary of Discussion About Data Needs

After the formal presentations, the meeting turned to an open discussion of future long-term care residential data needs given the research priority areas discussed on the first day, as well as the need to develop quality indicators for monitoring the quality of long-term care. An analysis of those needs indicated that the following types of data should be collected:

  • Demographic and health characteristics of residents.
  • Facility characteristics including services provided and hotel-type amenities offered.
  • Services received by each resident.
  • Health outcomes.
  • Payment information including costs and revenues.
  • Market information and prices.
  • Longitudinal information about the transitions of residents to and between health care settings.

The following sections highlight these data needs with respect to the specific research categories enumerated by the experts. For most resident-level studies, basic demographic and health data are necessary. For studies conducted at the facility level, basic facility characteristics (e.g., size) are necessary. These items are not repeated below.


  • To assess cost-effectiveness, it is necessary to measure outcomes of care. To measure outcomes of care, health status needs to be collected for at least two points in time for the same individual.
  • To evaluate service effectiveness, resident service use needs to be linked to resident characteristics.
  • For facility-cost studies, in contrast to studies of the expenditures of residents, average daily census and wage rates need to be collected. Costs for certified facilities could be derived from cost reports.
  • Facility characteristics are also important. Variables highlighted in the discussion were: special care units, staffing levels, and ownership type.
  • To assess the cost-effectiveness of different settings, it is necessary to follow people across settings and compare outcomes and costs. The consistency of measures across settings is important.
  • Data need to be linked: resident to facility, facility to market (State, county, ZIP code).


  1. To study the financial burden of long-term care, data are needed on family income and assets, expenditure information for long-term care service use, and payment sources. Race and ethnicity data also are needed to study subgroups.
  2. To study access to long-term care in general, the study population would include not only a residential population, but all persons with long-term care needs. Nevertheless, for persons in long-term care facilities, access to other types of facilities (e.g., hospitals) could be studied. Updated information on the supply of nursing beds by county is needed, as well as other supply variables such as supply of home care slots, etc. Data on market area characteristics also are needed.

Financial and Market Incentives

In the area of financial and market incentives, additional data needs include:

  • Information about variations in State reimbursement.
  • Information about linking providers to residents.
  • Information on insurance coverage at the resident and market levels.
  • Ability to track revenues and costs.
  • Ability to link people to markets to determine the influence of market factors on incentives for efficiency and quality.

Consumer Issues

With regard to consumer issues, additional data needs include:

  • Ways to link people to market factors so that the influence of factors such as competitiveness and HMO penetration per capita income on patient decisions can be studied.
  • Measures about housing quality and other amenities that are important to consumers.
  • A way to measure consumer preferences, although this is difficult.

Quality Assurance

Experts made the following points with regard to quality assurance:

  • To develop risk-adjusted outcomes that facilities could use to assess their own outcome results, national representative data that include outcomes and health status measures are needed (for a sample of residents in residential settings).
  • To evaluate the effectiveness of quality assurance approaches in facilities, certification status and process measures of quality assurance are needed.
  • To assess quality of life, more research is needed to develop reliable and valid quality-of-life measures for the long-term care population.

Return to Contents

Current Data

The following two sections summarize the main points made during the afternoon discussion of current data.

Pros and Cons of Person-Based and Facility-Based Sampling Designs

  1. Most recently, facility-based national data collection efforts have focused on nursing homes (and people in those homes), but not on assisted living, group homes, etc. (and residents in those homes). Since the market has changed greatly, this is a major limitation.
  2. A national frame does not exist for the entire continuum of long-term care facilities.
  3. Cross-sectional designs, like the current NNHS, are unsuited for measuring health outcomes, which require at a minimum health status measures at two points in time for the same person.
  4. Surveys (like the MCBS) that have a person-based sampling design and are longitudinal, can follow people into all types of residential- and community-based settings. Thus, these data can be used to study transitions across long-term care settings.
  5. The 1996 MEPS collected more than one round of health status and therefore can be used to study factors affecting health outcomes for both nursing home admissions and current residents.
  6. The sample size (of persons and facilities) in facility-based samples (like the NNHS) is generally larger than the size of person-based samples (like the MCBS), which have not typically included over-sampling of persons in residential care settings. The MCBS sample is too small to answer many questions (especially among sub-populations). For example, it cannot compare persons in assisted living facilities to persons in nursing homes with respect to outcomes, access/equity, and financing.
  7. The MCBS data are routinely linked to HCFA claims data. Theoretically, MEPS data can be linked to HCFA claims, but this has not occurred yet. At this time, NNHS cannot be linked to HCFA data based on current data collection agreements.
  8. Since MCBS is based on a sample of Medicare beneficiaries, long-term care residents not covered by Medicare, like most persons under 65, are excluded from the study population.
  9. It is not possible to make facility-level estimates with the MCBS (although facility-level data can be linked to persons to adjust for facility-level risk factors).
  10. The MEPS NHC (but neither the MCBS nor the NNHS) includes a rolling sample of persons admitted to a nursing home any time during the year. This sample, when linked with the expenditure and service use information, allows for better understanding of the cost and service use of sub-acute care. It also furthers our understanding of health status and sources of payment for residents at admission.

Overall Limitations of Current Data

  1. Business arrangements are evolving as the long-term care business environment becomes more complex. In general, data are missing on the intra- and inter-organizational relationships, the networks, and vertical and horizontal structures of long-term care businesses. Understanding and collecting data on different types of facilities is important for determining how facility types react differently to market incentives. While some of this is captured in the MCBS, it is not possible to make facility-level estimates.
  2. MDS and OSCAR data are available for certified nursing homes and their residents. For these residents, MDS data theoretically could be obtained (electronically) from HCFA and might therefore serve as an alternative to in-person data collection (for some data items). Since this has not occurred at a national level, this approach needs to be evaluated to determine the feasibility and implications for data quality and data collection costs. Some limitations to this approach were mentioned. First, in-person data collection of health status data may be preferable to using records only, since MDS/medical record data could be supplemented by talking to staff to improve the quality of the data. Second, it appears, based upon a HCFA report presented at the meeting, that it is not possible to sample from the MDS universe.
  3. MEPS NHC, NNHS, MDS, and OSCAR data are available only for nursing home residents.
  4. The 1996 MEPS is one of the few surveys that collects, for a calendar year, detailed person-level, long-term care expenditure data. MCBS collects similar information for persons in all types of long-term care facilities, but sample size is small, limiting analysis.
  5. Administrative data for the non-elderly disabled is State specific. It would be very difficult to build national data on health status for these persons except through a survey. For those under age 65, Medicare data generally do not exist to link information.
  6. Income and assets data were collected in detail in the 1996 MEPS NHC.11 The MCBS long-term care instrument contains an income variable and an assets variable. The 1999 NNHS did not collect income or assets data.
  7. Medicaid pays a substantial portion of the bill for long-term care in a residential setting, yet no national long-term care survey links persons to their Medicaid claims data (although the 1997 Balanced Budget Act mandated that Medicaid claims be electronically transmitted by the States to HCFA in a standard format).

11. However, high levels of non-item response were observed for some questions.

Return to Contents

Possible Strategies to Improve Data

There was considerable discussion about the pros and cons of person-based and facility- based sampling strategies. Participants generally believed that both approaches have merit and unique strengths.

Build a Broader Residential Frame

There was consensus that a facility frame should include the full continuum of long-term care facilities. The Office of the Assistant Secretary for Planning and Evaluation (ASPE) and Catherine Hawes have developed a probability sample of assisted living and residential care facilities, but not for all States. A number of sources may provide assistance in developing a national frame. The National Investment Council for Senior Housing is putting together a national sample frame of all senior housing, including assisted living, 202 housing, congregate apartments, and residential care facilities. The Census Bureau's Economic Census (beginning with the 1997 version) may be another source of frame data. Access to IRS lists of businesses would also be very helpful.

As part of this effort, consideration needs to be given to the operational definitions of residential versus community care, and to the implications these definitions might have on household survey data collection. One suggestion was to add a few data elements to a household survey, like the NHIS, that could be used to determine if the dwelling unit formally provides assisted living/personal care services to residents. This information could be used to improve current estimation methods on the number of assisted living and board and care homes.

Increase the Long-Term Care Sample Size of MCBS

Although the MCBS has many strengths, participants felt that the sample size for studying the residential long-term care population was too small to help answer many of the questions that warrant comparisons between persons in alternative settings. One approach discussed was to increase the sample of persons in long-term care facilities by screening persons over the telephone to identify those most likely to enter long-term care facilities. Participants mentioned a number of other strategies, but most felt that more methods development is needed.

Conduct Specialized Periodic Surveys

Another approach that was mentioned was to have an ongoing, sequential facility-based panel. In addition, periodic special studies could be conducted in alternative years; for example, a study on outcomes and quality.

Explore Use of Administrative Data

The MDS and OSCAR data were discussed throughout the meeting as a potential source of information either to be used alone or in conjunction with survey data. Concerns were expressed about confidentiality, data quality, and feasibility. Since HCFA is now receiving these data in electronic form, this could provide an alternative for some survey data, although the merits of using these data still need to be assessed.

Return to Contents


This meeting provided the Agency with a comprehensive view of the data needs for answering the major research and policy questions for residential long-term care and established a number of priorities. Experts conveyed some very strong messages:

  • We need information about persons in all types of residential long-term care settings.
  • We need to be able to understand what affects health and functional status changes in residential long-term care settings.
  • We need to be able to understand the transitions that people make in the long-term care system and to evaluate the transition decisions that are made.
  • We need comparable data across all long-term care settings.
  • We need to evaluate the usefulness of Federal administrative data for nursing homes.
  • We need to assess the sample sizes of residential populations in Federal surveys to ensure that analytical needs can be met.

In addition, experts identified a number of priority data gaps that cannot be filled unless better measures are developed:

  • Measures to evaluate health-related quality of life in residential care.
  • Measures to evaluate clinical management in residential care.
  • Measures that incorporate consumer preferences.
  • Measures of process quality.

The meeting was adjourned in the afternoon on October 15, 1999.

Return to Contents

AHRQ Publication No. 01-0007
Current as of October 2000


The information on this page is archived and provided for reference purposes only.


AHRQ Advancing Excellence in Health Care