Table of Contents Foreword Acknowledgments Introduction: Lessons Learned from Eight Case Studies by Brant E. Fries and Charles J. Fahey The RAI and the Politics of Long-Term Care: The Convergence of Science and Politics in U.S. Nursing Home Policy by Catherine Hawes, Bruce C. Vladeck, John N. Morris, Charles D. Phillips, and Helene Fredeking From Anecdotes to Evidence: Complex Continuing Care at the Dawn of the Information Age in Ontario by John P. Hirdes, Duncan G. Sinclair, John King, Paul Tuttle, and John McKinley Toward Informed and Evidence-based Elderly Care: The RAI Experience in Iceland by Pálmi V. Jónsson and Hrafn Pálsson MDS (Minimum Data Set) Assessments for Policymaking and Evaluations in Israel by Jacob Gindin, Sara Levi, Orna Intrator, and Jochanan Stessman The Implementation of a Common Assessment System and Care Model in Italy by Roberto Bernabei, Marina Panfilo, and Giuseppe Panio Muddling through LTC Policy: The Japanese Government's Approach to the Introduction and Dissemination of Resident Assessment Instruments by Naoki Ikegami and Masanori Nishiyama The Challenge of Dependency in a Changing Society: New Tools for a New Era at the Health Care–Social Care Interface in Spain by Elaine Duncan and Enrique Castellón A Thread from Many Strands: Assessment, the InterRAI MDS System, and Health and Social Care Policy in England and Wales by G. Iain Carpenter and David Challis The Authors

---

Foreword

The authors of these case studies describe the use of unique instruments, devised by cross-national collaborators, to array and assess data about persons who receive long-term care and related services in eight countries. These data are of particular importance for policy to improve the quality of care—either voluntarily or through regulation—and to allocate resources more effectively and efficiently across a continuum of services. The instruments also make it possible to compare the cost and quality of long-term care services in different parts of a country and among countries.

The Milbank Memorial Fund, an endowed philanthropic foundation, collaborates with decision makers in the public and private sectors to develop and implement policy that maintains and improves health. The Fund and its constituents among decision makers are eager to learn about the best evidence available to inform choices among alternative policies.

The federal government in the United States commissioned researchers to design and test the instruments that later became the basis of the international project in order to improve formal regulation of the quality of care. Researchers from 22 countries subsequently devised a family of related instruments and helped to adapt them to conditions in particular countries.

Many people participated in writing and reviewing these case studies. They are identified in the Acknowledgments.

Brant Fries requires special recognition. A founder of interRAI, the organization that devised the family of instruments, he was central to planning the case study project and essential to encouraging authors to draft, revise, and revise yet again their manuscripts. Fries and Charles Fahey, a program officer of the Fund, convened a meeting of the authors in Helsinki to review the initial drafts of the cases in the difficult month of September 2001. Fries and Fahey collaborated in writing the introduction to the report, in which they describe the history of the project and the significance of the experience reported in the cases for clinicians and policymakers.

Daniel M. Fox
President

Samuel L. Milbank
Chairman

---

Acknowledgments

The following persons participated in meetings and/or reviewed this report in draft. They are listed in the positions they held at the time of their participation.

Sergio Ariño Blasco, Consultant Geriatrician and Director, Health/Social Care Department, Fundació Hospital Asil de Granollers, Barcelona, Spain; Christine E. Bishop, Schneider Institute for Health Policy, Heller Graduate School, Brandeis University, Waltham, Mass.; Charles Botz, London Health Sciences Centre, London, Canada; A.M. Clarfield, Sidonie Hecht Professor of Gerontology, Ben Gurion University of the Negev, Beersheva, Israel; June Crown, London, U.K.; Michael J. Donnelly, Reader, Epidemiology and Public Health, School of Medicine, Queen's University Belfast, Belfast, U.K.; Luigi Ferrucci, Clinical Research Branch, National Institute on Aging, Baltimore, Md., and Laboratory of Clinical Epidemiology, Istituto Nazionale Ricovero e Cura Anziani (INRCA) Geriatric Department, Florence, Italy; Robert B. Friedland, Director, Center on an Aging Society, Georgetown University, Washington, D.C.; Aðalsteinn Guðmundsson, Medical Director, Hrafnista-DAS Nursing Homes, Staff Geriatrician, University Hospital of Iceland, Reykjavik, Iceland, and Clinical Assistant Professor, University of Wisconsin Medical School, Madison, Wisc.; Jack Habib, Director, JDC-Brookdale Institute of Gerontology and Human Development, Jerusalem, Israel; James K. Haveman, Jr., Director, Michigan Department of Community Health, Lansing, Mich.; Shuichiro Hayashi, Chief, Division of Health for the Elderly, Ministry of Health, Labor, and Welfare, Tokyo, Japan; Miriam J. Hirschfeld, Director, Long-term Care CCL/NMH, World Health Organization, Geneva, Switzerland; Katsunori Kondo, Associate Professor, Faculty of Social Welfare, Nihon Fukushi University, Okuda, Japan; Sandra D. Lang, Deputy Minister, Ontario Ministry of Consumer and Business Services, Toronto, Canada; Ariela Lowenstein, Head, Department of Masters in Gerontology and Center for Research and Study of Aging, Faculty of Welfare and Health Studies, University of Haifa, and Chair, European Behavioral, Social Science and Research Section, International Association of Gerontology—European Region, Haifa, Israel; Niccolò Marchionni, Professor of Gerontology and Geriatric Medicine, Department of Critical Care Medicine and Surgery, University of Florence, Florence, Italy; José M. Martin-Moreño, Director, Spanish Agency for Health Technology Assessment, Institute of Health "Carlos III," Madrid, Spain; Olivia P. Maynard, President, Michigan Prospect for Renewed Citizenship, and Regent, University of Michigan, Flint, Mich.; David R. Nerenz, Director, Institute for Health Care Studies, Michigan State University, East Lansing, Mich; Stuart G. Parker, Clinical Director, Sheffield Institute for Studies on Aging, University of Sheffield, Sheffield, U.K.; Susan Reinhard, Co-Director, Center for State Health Policy, Rutgers University, New Brunswick, N.J.; William E. Reynolds, Public Service Professor, School of Social Welfare, and Clinical Associate Professor, School of Public Health, State University at Albany, Albany, N.Y.; Otto Christian Rø, Project Manager, Norwegian Board of Health, Oslo, Norway; Leocadio Rodriguez-Mañas, Geriatric Service, Hospital Universitario de Getafe, and Spanish Geriatric Medicine Society, Madrid, Spain; Maria Pilar Salinas Bujan, Regional Manager for Elderly Care, Health/Social Services Coordination Program, Galician Ministry of Health, Santiago de Compostela, Spain; Martin W. Shreeve, Programme Director, Better Government for Vulnerable People, Wolverhampton, U.K.; Sigurveig H. Sigurdardóttir, Director, Iceland Red Cross, Reykjavik branch, Reykjavik, Iceland; Sigurbjörg Sigurgeirsdóttir, doctoral candidate in Health Politics and Policymaking, London School of Economics and Political Science, London, U.K.; Robyn Stone, Executive Director, Institute for the Future of Aging Services, American Association of Homes and Services for the Aging, Washington, D.C.; and Marco Trabucci, Professor, University of Roma Tor Vergata, and Geriatric Research Group, Brescia, Italy.

---

Introduction: Lessons Learned from Eight Case Studies

by Brant E. Fries and Charles J. Fahey

The case studies presented in this report tell the stories of researchers and their policy colleagues from eight countries who are committed to the concept that collecting accurate information in a common format within and across service sectors and countries enhances both the well-being of frail persons and the efficient and the equitable distribution of public resources. The cases are the work of researchers who in 1992 formed a nonprofit consortium known as interRAI (RAI = Resident Assessment Instrument) to promote the use of assessment instruments and of cross-national comparisons as a foundation for improving care. InterRAI's membership currently consists of 45 persons in 22 countries or regions: Australia, Canada, Czech Republic, Finland, France, Denmark, Germany, Hong Kong, Iceland, Israel, Italy, Japan, the Netherlands, New Zealand, Norway, South Korea, Spain, Sweden, Switzerland, Taiwan, the United Kingdom, and the United States.

Each of the cases unfolds within the cultural, social, and political realities of a particular country and regions within it. Because of the uneven deployment of these systems across most of the countries involved, the cases do not seek to compare data. Moreover, full implementation of the range of available instruments across long-term care domains has yet to be realized in any of the countries. The U.S., which first implemented the RAI, is presented first, to provide background and context. The other studies proceed alphabetically by country: Canada, Iceland, Israel, Italy, Japan, Spain, and the United Kingdom.

The Development of the Resident Assessment Instrument

The executive branch of the U.S. federal government commissioned the development of the first of what eventually became the RAI "family" of instruments in 1987, in response to mandates in the federal Omnibus Budget Reconciliation Act. These instruments are now routinely used in long-term care settings and, despite some ongoing criticism of them, have received the endorsement of the major long-term care provider trade organizations, the American Association of Homes and Services for the Aging (nonprofit organizations) and the American Health Care Association (investor-owned organizations).

Since 1992, interRAI has developed and tested other instruments in this "family," in some cases with partial funding through government contracts or from foundations in the United States and other countries. InterRAI licenses its assessment instruments, without charge, to any governmental or caregiver organization, because its members are committed to encouraging and assisting all stakeholders to use the instruments across a range of settings and services.

Each RAI instrument serves varied purposes for particular clients, providers, payers, and regulators. The most important of these are:

• Assessing and determining eligibility for services
• Contributing to care planning and management by clinicians
• Establishing a structure for allocating resources within an agency or organization
• Providing a basis for provider reimbursement
• Developing a sound basis for quality indicators
• Assisting in determining who uses which services
• Contributing to an understanding of which services and settings are appropriate for persons with various conditions
• Tracking frailty in individuals and specific populations over time
• Rationalizing services and systems
• Contributing to the projection of need and attendant costs

The RAI has had extensive development and testing for validity and reliability, including review by hundreds of geriatricians, nurses, clinicians, and other practitioners. The authors of the instruments provide definitions and training materials. Many commercial vendors supply computer software that supports the RAI. Because the Minimum Data Set (MDS) is used in every nursing home in the United States, there is considerable experience with the use of aggregate data for various purposes. A research archive at the University of Michigan contains more than 12 million assessments from 13 states, with multiple, longitudinal assessments on millions of residents. The national mandate in the United States has produced a federal database that is substantially larger and rapidly growing; these databases are increasingly in demand.

InterRAI has also devised parallel assessment systems for home care (RAI-HC), post-acute care (RAI-PAC), palliative care (RAI-PC), assisted living (RAI-AL), acute care (RAI-AC), and mental health (RAI-MH; for psychiatric inpatient care). InterRAI works to balance the goals of making these systems specific to their applications yet compatible, so that individuals can be compared across types of care.

Why These Efforts Matter

The dramatic growth of the population of the very old, with the concomitant increase in the number of frail people, poses challenges to individuals, their loved ones, those who care for them, and policymakers. These challenges can be met adequately only if good information is available to those who must make decisions at the personal, clinical, managerial, and public policy levels.

A standardized assessment tool is a foundation for addressing a range of needs. Clinicians need information that will help them care for patients more effectively. Managers need information to most efficiently use the resources available to them for managing existing services and developing new ones. Frail persons and members of their families want assurance that services are being provided honestly and in a safe, preferably homelike, environment. Policymakers have responsibility for allocating scarce resources and ensuring public health.

A common assessment tool is also a useful first step in sorting out differences among and within countries. The characteristics of the frail are similar from country to country and from jurisdiction to jurisdiction. But there is considerable variation in how countries define needs and organize services. Even within the same jurisdiction there are often divergent understandings of who is being served, where, and why.

The Resident Assessment Instrument (RAI)

InterRAI's instruments derive from the Resident Assessment Instrument (RAI) for nursing homes. (In the case study texts, the assessment systems for other modalities of care are differentiated with suffixes, such as the RAI for Home Care [RAI-HC].) The RAI system consists of two parts, the Minimum Data Set (MDS) and the Resident Assessment Protocols (RAPs).

1.

The Minimum Data Set (MDS) contains the core assessment items necessary for a comprehensive assessment of a nursing home resident. The MDS includes individual assessment items and specifies definitions, time frames, and exclusions for the items as well as the response codes needed to ensure accurate assessment. It covers a wide range of functional domains, such as the resident's status in terms of cognition, communication, activities of daily living (ADLs), continence, psychosocial well-being, disease diagnoses, and health conditions.

2.

The Resident Assessment Protocols (RAPs) are guidelines for additional, more sharply focused assessment and care planning. The MDS triggers use of the RAPs by identifying residents who need more specific assessment. The specific "triggers" are an individual resident's MDS-item responses that suggest the presence of a problem condition (e.g., mood, incontinence), a risk factor for decline (e.g., dizziness, wandering, use of trunk restraints or certain types of drugs, all of which elevate the likelihood of a fall), or the potential for improved function (e.g., the resident believes he/she is capable of increased independence in function or has a hearing problem that, if helped, could improve communication). Each of the 18 nursing home RAPs is a structured framework for developing plans that address the targeted problem by identifying treatable causes of the problem or risk factor or by effecting management interventions. The intent of the protocols is educational rather than prescriptive. Each protocol helps organize information from MDS items that can be used to inform the care planning process, that identifies additional information that may be needed, and that provides a context in which information about residents and their strengths, preferences, and needs is linked to the care plan options. Slightly different terms are used for the care planning protocols in the other RAI instruments: for example, there are Client Assessment Protocols (CAPs) for the home care instrument and Mental Health Assessment Protocols (MHAPs) for the RAI-MH.

Lessons from These Case Studies

These cases offer evidence about (1) the RAI's effect on patient care, (2) its importance as a policymaking tool, (3) success in RAI implementation, and (4) obstacles to the instruments' adoption.

Patient Care

• The RAI can contribute to improved direct patient care. It provides a means for an accurate, comprehensive understanding of the person in need of care; points to areas requiring special attention; and contributes to an ongoing care management plan.
• The RAI can be a powerful means to encourage coordinated, interdisciplinary activity on behalf of the resident.

Policymaking

• The RAI can be an important tool in assessing the intensity of need and in planning the efficient use of resources to meet need at both the provider and societal levels.
• The RAI can contribute to reimbursement policies that correspond to needs of the residents in a given program and, at the same time, offer incentives to care for those in greatest need.
• The RAI can contribute to developing quality indicators.
• Because preliminary, but substantial, data analysis confirms the perception that people with the same needs may be found in different settings and programs, the RAI can help in assessing whether resources are being used appropriately.

Successful Implementation of the RAI

• The clinicians who are the primary, hands-on users of the RAI must be convinced that its use contributes to good care and is worth the opportunity costs involved.
• Education and training are essential components in developing clinician acceptance of the RAI.
• Program managers must see the utility of the consistent, accurate use of the instrument. Strong incentives for them to use the RAI instruments include the possibility of having more sophisticated systems for paying providers and for comparing quality across care programs.

Obstacles to Implementing the RAI

• The RAI can be seen by persons in various professions as unnecessarily burdensome, less precise than existing instruments, too medically oriented, and an encroachment on the responsibilities of particular professions.
• It is difficult to introduce compatible, consistent assessment instruments across different systems and domains, each of which has a culture, history, and current approach to gathering and using information.
• The RAI can be the basis for reimbursement based on nursing home residents' levels of acuity and thus disrupt the status quo in reimbursement.
• The introduction, maintenance, and full exploitation of a data system can only develop over time and with a significant expenditure of resources.
• Staff turnover impedes the consistent, continuing use of the instrument.
• The use of the RAI is affected by such contingent factors as perceived crises that claim public attention.
• The RAI can assist but not determine the judgment of policymakers.

Recommendations for the Future

Each of these case studies carries an implicit sense of urgency about the need to collect good data in essentially compatible formats; to introduce systems of collection across service domains, regions, and nations; to make these data appropriately accessible to clinicians, providers, policymakers, and researchers in manners that are relevant and usable and that maintain confidentiality; and to increase and sustain the public investments that will ensure that appropriate data systems are maintained and adapted to the particular needs of clinicians, provider organizations, and regulators.

---

The RAI and the Politics of Long-Term Care:
The Convergence of Science and Politics in U.S. Nursing Home Policy

Catherine Hawes, Bruce C. Vladeck, John N. Morris, Charles D. Phillips, and Helene Fredeking

EXECUTIVE SUMMARY

The Resident Assessment Instrument (RAI) is a uniform assessment system used to assess and plan the care of residents in virtually all U.S. nursing homes. It became a federally mandated system as part of a comprehensive set of nursing home reforms passed by the U.S. Congress in the Omnibus Budget Reconciliation Act of 1987 (OBRA '87).

The creation of Medicare and Medicaid in the mid-1960s provided the first significant influx of federal funds as payment for nursing home care in the United States. Federal regulations accompanied these funds. This watershed policy event was followed by rapid growth in the supply of nursing home beds and by the transformation of the U.S. nursing home industry. What had been a sector dominated by almshouses, county nursing homes for the poor, church-related nonprofit facilities, and small, individually owned proprietary homes rapidly became a predominantly for-profit industry dominated by multistate firms (nursing home "chains"). In addition, the first 20 years of Medicare and Medicaid were plagued by scandals about seriously substandard care in nursing homes, conditions attributed to a largely ineffective regulatory system.

The nursing home provisions of OBRA '87 resulted from these concerns about quality and constituted the most comprehensive set of federal reforms addressing the regulation of nursing homes since the passage of Medicare and Medicaid. While OBRA '87 mandated many changes, the RAI was one of the most fundamental. Beyond its use in resident assessment and individualized care planning, the RAI provides resident-level data used to determine Medicare eligibility, to generate quality indicators used in the inspection of nursing homes by government agencies, to plan quality improvement activities by facilities, and to adjust nursing home payment rates to reflect differences in the amount and type of care residents need.

Since the implementation of the RAI and most other OBRA '87 regulations in 1990, researchers have found significant improvements in both process and outcome quality measures. Further, a number of other countries have indicated interest in using the RAI for assessment or reimbursement purposes. Despite these developments, the RAI and the rest of the federal regulations mandated by OBRA '87 came perilously close to being eliminated during the mid-1990s and remain somewhat vulnerable to the vagaries of state and national politics. This paper examines the factors that led to the adoption of the RAI and the other OBRA '87 nursing home reforms; it explores the factors that make quality improvements through federal regulation essential but politically vulnerable; and it summarizes the lessons to be learned from the process of developing and implementing the RAI.

The enactment of the OBRA '87 nursing home reforms and the mandate to use the RAI faced considerable barriers in the United States—barriers that differed significantly from those faced in other countries. First, the U.S. nursing home industry is largely private and proprietary, and much of the industry views regulation as interference with business, even when the goal is quality improvement. Second, many lawmakers share this view, and disputes over the appropriate role of the federal government, particularly in the regulation of industry, are highly politicized. Finally, in political conflicts, the nursing home industry has traditionally had a significant advantage over residents, family members, and consumer advocacy groups, and the industry has been a more consistent and effective advocate for its positions than have groups representing these other constituencies.

Despite the traditional barriers to regulatory reform, several factors combined to make the OBRA '87 reforms possible. First, an attempt by President Ronald Reagan to reduce nursing home regulation at the federal level was met with strong opposition from consumer advocacy groups and congressional opponents of the president. The result was a commissioned study by the National Academy of Sciences, Institute of Medicine (IOM). The IOM's research activities—and the organization's prestige—lent scientific credibility to its recommendations. Moreover, the IOM recommendations provided a core around which supporters of quality improvement could coalesce.

A second factor facilitating the adoption of the reforms was the formation of the national Coalition for Quality Care. This coalition included consumer advocacy groups, associations of health care professionals, and nursing home providers. It reached consensus on most IOM recommendations and supported them before Congress. As a result, the Congress, with strong bipartisan support, adopted nearly all of the IOM recommendations, including the requirement for a uniform, comprehensive functional resident assessment.

The Centers for Medicare and Medicaid Services (CMS, formerly the Health Care Financing Administration) and the research team that developed the RAI also made a number of important decisions that facilitated the RAI's implementation. To maintain support from the Coalition for Quality Care, the team involved all key stakeholders in the process of RAI development. Consumer advocates, providers, and health care professionals all bought into the system, deflecting initial industry attempts to limit the scope of the RAI. In addition, CMS and the development team were committed to making the RAI clinically relevant, feasible for providers, and politically defensible.

Although several researchers found significant improvements in both process and outcome quality measures as a result of the implementation of the RAI and the other OBRA '87 nursing home reforms, the OBRA '87 reforms were nearly repealed in 1995 as part of a larger attempt to change the nature of the Medicaid program, which pays for the care of about two-thirds of all nursing home residents. At that point, part of the nursing home industry supported repeal of the OBRA reforms, particularly the enforcement provisions. But consumer advocates, aided by researchers, were able to use the empirical evidence about the positive effects of the RAI and other OBRA provisions to effectively oppose the dilution of federal regulations. Once consumer advocates redefined the issue as one of quality of care, opposition to the repeal spread from congressional Democrats to moderate Senate Republicans, and the OBRA '87 reforms were saved.

The passage of the OBRA '87 nursing home reforms and the subsequent implementation of the RAI, as well as their near-repeal, reveal important lessons about attempts to improve quality in American nursing homes. It is clear that because of the highly politicized nature of regulation in American politics, all regulations remain politically vulnerable, even those aimed at improving quality and even when originally enacted with strong support from both consumers and providers. But it is equally clear that consumer advocates can use research findings to identify practices and policies that promote quality and can redefine political conflict over regulation to focus on patients and their quality of care. Regulations embodying policies that both consumer advocates and providers view as clinically relevant and useful are more sustainable over time.

INTRODUCTION

Since the passage of Medicare and Medicaid1 in the mid-1960s, nursing home policy in the United States has largely been a product of two factors. First, nursing homes and their residents have been and continue to be affected by policies aimed at the acute care sector, particularly those policies intended to reduce hospital use and costs. These have had significant spillover effects on nursing home policy and operations, including financing, utilization, and resident case mix. In addition, nursing home policy has been episodically affected by scandals about poor care in nursing homes (Hawes 1987; Vladeck 1980). Indeed, for decades, major reform efforts have often been tied either to a series of tragic events involving multiple deaths² or to public reports of seriously substandard care that have captured the public's attention and, at least temporarily, generated the necessary political support for new legislation or regulation (California 1983; New York 1975, 1976; Ohio 1978, 1979).³ Occasionally, however, scientific endeavors have fundamentally affected nursing home policy. These endeavors have, at times, influenced both the process of policy development and debates about whether to retain or alter key federal regulatory policies. This paper recounts one of these episodes.

The development, implementation, and retention of the nursing home reforms contained in the Omnibus Budget Reconciliation Act of 1987 (OBRA '87) represent key examples of the effect of science and politics on policy. The purpose of this case study is to illustrate the interplay of science and politics by focusing on one particular aspect of the OBRA '87 nursing home reforms: the federally mandated Nursing Home Resident Assessment Instrument (RAI). The paper also seeks to illuminate the politics of long-term care and the conditions under which resident-centered reforms can be enacted, implemented, and sustained, as well as the way in which research findings may be used to structure policy and policy debates. In addition, this case study illustrates important differences between nursing home reform in the United States and other countries.

BACKGROUND: THE POLITICS OF LONG-TERM CARE

Before the mid-1960s, state licensure laws governed nursing homes. Federal funding was quite limited, and federal standards were mainly tied to the use of federal funds (e.g., Federal Housing Authority) to finance construction of facilities. Two dynamics, however, generated greater federal involvement in nursing home regulation. First, a series of federal reports were highly critical of the shoddy conditions and abysmal quality of care in nursing homes (U.S. DHEW 1958; U.S. Senate 1956, 1957, 1960a, 1960b). Second, the introduction of federal payment for nursing homes with Medicare and Medicaid funds generated a demand for greater federal oversight to ensure adequate accountability for and control over these large public expenditures (IOM 1986; Vladeck 1980). Federal standards were initially limited to extended care facilities (ECFs) intended only to provide skilled nursing or rehabilitative care and to facilitate shorter hospital stays for Medicare beneficiaries. Within a few years, however, federal standards were extended under the Medicaid program to cover intermediate care facilities that provided long-term "custodial" care.

Medicare and Medicaid, which fostered the emergence of and serve as the financial foundation for the current nursing home industry, also fostered the development of the politics of long-term care. There is often a clear nexus between politics and the content of policy (Lowi 1964; Schattschneider 1960). Indeed, in some cases, policy creates programs that essentially generate their own "politics." Certainly this has been true with respect to nursing home care. Public payments essentially created an industry that, if it has not dominated the subsequent policy process, has had the most sustained effect on policy over time. These new sources of funding for nursing home care and their cost-based reimbursement policies attracted Wall Street investors and real estate developers to a sector that had previously been populated by not-for-profit, church-related, and fraternal organizations; county poor farms; and small individually owned, "mom and pop" proprietary homes. The diverse and locally controlled pre-1960s nursing home sector metamorphosed into a largely for-profit industry dominated by "the chains" (multifacility systems that own and operate nursing homes in several states and whose stock is publicly traded) with increasing concentration of ownership nationwide in relatively few corporate entities.4

Each industry has its own economic interests, and the nursing home industry is no exception. To protect its interests, the industry has become a major contributor to political campaigns at both the state and national levels (Pear 1997; Vladeck 1980). One of the chief lobbyists for the proprietary nursing home industry—himself a nursing home owner—chaired the Democratic Party's Business Council, a group of donors who each contributed $10,000 or more per year to the party during the Clinton administration, and he became the party's finance chairman. At the same time, nursing home interests contributed more than $1.1 million to Bill Clinton's 1996 presidential campaign (Weisskopf 1997).

If government policy has essentially created a provider industry and the conditions under which that industry has flourished and attained political power, it has done little to support similar access to influence by consumers. Nursing home residents are often naturally disadvantaged by the very disabilities that cause them to need nursing home care. Moreover, long-term custodial care is the province of Medicaid. Thus, elderly people who need long-term custodial care, such as those with Alzheimer's disease, must exhaust their incomes and assets, paying for needed care out of their own pockets, before becoming eligible for public assistance. For many frail elderly in the United States, public policy has essentially added poverty to the burden of the multiple chronic diseases that force them into nursing homes.

The imbalance in resources between consumers and providers of long-term care has always been striking. Medicare and Medicaid payment policies accept as allowable costs the nursing homes' dues to organizations that represent industry interests before Congress, state legislatures, and administrative agencies, but no similar, sustained public funding has been available for groups representing residents. Resident advocacy groups whose sole focus is long-term care policy, such as the National Citizens Coalition for Nursing Home Reform (NCCNHR), are usually shoestring operations. Unlike provider trade associations, such as the American Health Care Association (AHCA), consumer advocacy groups do not have political action committees to lobby for them. Similarly, the long-term care ombudsman program, which is often thought of as an advocacy group for residents, is barred from lobbying, and limited funding has constrained its efforts to educate policymakers about consumer needs and interests (IOM 1995). Well-organized and well-financed consumer advocacy groups that focus only on nursing homes or long-term care are a rarity. As a result of their structural and fiscal limitations, groups representing consumers of nursing home care, such as NCCNHR, rely on the media to focus occasional attention on quality problems and depend on the good will and skills of the small number of policymakers who champion nursing home reform issues.

Some consumer advocacy groups whose focus is more general and whose resources are much more substantial than the NCCNHR's, such as the American Association of Retired Persons (AARP) and the Alzheimer's Association, do support nursing home reform. But because these organizations represent constituents who have many other pressing interests that must be addressed in the policy arena—including Social Security, basic biomedical research, and Medicare policy in general—they rarely put nursing home issues at the front of their primary policy agenda.

This environment, with its strong industry sector and its weak consumer and advocate sector, is not unique to the politics of long-term care. Nonetheless, it was this environment that ensured that during the first 20 years of Medicare and Medicaid's involvement in long-term care, policy changes were mostly incremental, and legislation tended to focus on nursing home reimbursement issues. The industry tended to dominate legislation addressing reimbursement and was the most consistent and effective influence over the development and implementation of other aspects of regulation (Hawes 1987; Vladeck 1980). The result was a regulatory process that all too often failed to protect residents or ensure quality.

Between the passage of Medicare and Medicaid and the enactment of OBRA '87, a series of state reports documented abysmally poor care and failed regulatory processes (Arkansas 1978; California 1983; Colorado 1978; Connecticut 1976, 1980; Florida 1981; Illinois 1984; Maryland 1973; Minnesota 1976; New Jersey 1978; New York 1975, 1976; Ohio 1978, 1979; Texas 1978; Virginia 1979). Reports by the federal agencies regulating nursing homes (U.S. DHEW 1971, 1975) and by congressional agencies and committees (U.S. GAO 1971; U.S. House of Representatives 1985; U.S. Senate 1970, 1971, 1974–75) essentially echoed the same findings and criticisms. These reports found unsafe and unsanitary conditions, including failure to meet state fire codes and federal life safety codes; abuse; neglect; malnutrition; overuse of physical and chemical restraints; medication errors; and failure to provide prescribed medical treatments and therapies. Moreover, the studies determined that such substandard conditions were widespread.

The state and federal reports almost universally cited the same factors as causing or contributing to substandard care.5 Most of the studies made the following arguments:

• Despite the existence of federal standards, there was considerable variation in quality across facilities and substantial differences in the performance of state regulatory systems.
• Federal and state standards were inadequate to protect the health, safety and rights of residents.
• The inspection process focused on paper compliance with standards and failed to capture the realities of the day-to-day life of residents.
• Federal regulations did not provide states or the federal government with an adequate range of compliance mechanisms.
• When states adopted a range of sanctions, most failed to make effective use of the available remedies.
• State and federal regulatory systems allowed nursing homes with a history of seriously substandard care to remain in operation year after year.

The genesis of these failures in regulatory policy was not wholly mysterious. Many observers attributed it, in whole or part, to the power of the nursing home industry (Mendelson 1974; Moss and Halamandaris 1977; Townsend 1971; U.S. Senate 1974–75; Vladeck 1980). Despite reports about widespread problems and systemic regulatory failures, however, comprehensive reform did not occur. Those reforms that did occur were, for the most part, extraordinarily limited. For example, despite reports of widespread and serious quality problems, the only major new federal rule imposed during the first 15 years of Medicare and Medicaid was the federal life safety code. First applied solely to skilled nursing facilities, it was later applied to all certified facilities as reports of unsafe conditions and multiple fire deaths persisted.

Similarly, the U.S. Senate Special Committee on Aging (1974–75) published a series of shocking reports entitled Nursing Home Care in the United States: Failure in Public Policy. The only piece of federal legislation enacted in the wake of these revelations created the federal Office of Inspector General (OIG) in the U.S. Department of Health and Human Services and mandated development of Medicaid Fraud Units in the offices of the state attorneys general to prosecute cases of provider fraud. More far-reaching efforts, such as implementation of the Patient Appraisal and Care Evaluation to assess residents and develop an outcome-oriented survey system, as well as subsequent reforms proposed during the Carter administration, fell into limbo in the face of opposition from the nursing home industry (IOM 1986).

Even in states that issued highly critical reports, reforms were seldom far-reaching. For example, an Ohio legislative study commission's reports provided a troublesome, and all too familiar, picture of conditions in many nursing facilities and offered a comprehensive analysis of the regulatory failures that allowed such conditions to persist (Ohio 1978, 1979). As a result, legislators introduced two bills incorporating the study's recommendations. One bill, which addressed reimbursement and discrimination against Medicaid recipients, was enacted, but the other, an omnibus quality improvement bill addressing standards of care, the survey process, and enforcement remedies, failed. It was not successfully revived until a decade later, following the passage of the OBRA '87 nursing home reforms.

Nationally, the nursing home industry was able to limit reform efforts in two basic ways. First, it was often able to appeal to legislators who had an antipathy to government regulation by defining the issue as one of overly burdensome regulation and arguing that nursing homes were "the most regulated industry" in the country. Second, the industry often defeated reform by arguing that proposed regulations would impose substantial additional costs. For example, the Carter administration argued that its proposed regulations would cost about $80 million a year. Industry-hired consultants estimated the cost at $586 million for the first year of implementation and $435 million annually thereafter (Applied Management Sciences 1980)6. As a result of the dispute and the industry's power, the proposed Carter reforms went nowhere until the final hours of the administration, when one part of the proposed reforms, a provision making adherence to a set of residents' rights a condition all nursing homes had to meet in order to participate in Medicare or Medicaid, was published as a new regulation (IOM 1986, 247). Even that rule was immediately rescinded by the incoming Reagan administration as part of its "regulatory reform" initiative.

All these attempts at nursing home reform were played out in the context of an ongoing debate—one that lies at the heart of much political conflict in the United States. The nursing home industry in the U.S. is predominantly proprietary. As a result, the industry tends to view regulation as government interference in the operation of its "business." The more refined version of this argument is that the industry, left to itself and the forces of competition in the marketplace, could provide a better "product" than a nursing home industry constrained by the fetters of governmental regulation. Such regulation, this argument contends, perverts the market, stripping it of its purported ability to generate better products at lower costs. Any attempt at new regulation, even regulation aimed at improving residents' quality of care and quality of life, was immediately embroiled in this debate.

Issues involving the allocation of public resources are inherently political. Even in countries where long-term care and nursing home services are provided by local or national agencies, questions regarding who has the authority to allocate resources and who should receive them cause political conflict. Outside the United States, however, these conflicts rarely inspire such ideological fervor. This country is unusual in the array of powerful, well-funded political interests that become involved and in the tendency, during eras of deregulation, for enormous quantities of money to move from the public coffers into the private sector.

DESIGN: GENESIS OF THE RAI AND OBRA '87 REFORMS

The passage of the OBRA '87 nursing home reforms seemed to herald a new era of politics and policymaking in long-term care. Several factors contributed to the enactment of what were arguably the most sweeping set of reforms to the federal regulation of nursing homes since the passage of Medicare and Medicaid.

The OBRA '87 reforms were based on the recommendations of a panel established by the prestigious National Academy of Sciences Institute of Medicine (IOM). Democratic congressional reaction to the Reagan administration's proposal to reduce nursing home regulation was swift and negative. In an agreement worked out between Congress and the administration, the IOM was charged with evaluating the regulatory process and making recommendations on how to improve quality of care and quality of life for the nation's nursing home residents.

The commission conducted a two-year study that included public hearings, a review of state reports completed during the preceding decade, commissioned papers on specific aspects of nursing home quality assurance, and other attempts to understand the existing systems of federal and state regulation and how they could be improved. In 1986, the IOM committee issued its report, Improving the Quality of Care in Nursing Homes, which contained a wide array of recommendations aimed at improving the standards governing nursing homes participating in Medicare and Medicaid, refocusing and enhancing the survey process for monitoring the performance of facilities, and expanding the availability and use of a range of enforcement remedies (IOM 1986).

With strong bipartisan support, Congress adopted nearly all of the IOM committee's recommendations. The broad support enjoyed by the legislation resulted from a number of factors. First, the IOM used research findings, whenever possible, to craft and support its policy recommendations. Some recommendations, such as reductions in the use of physical restraints and reliance on the principles of geriatric assessment, were based on empirical evidence about good clinical practice, with high-quality, innovative nursing homes in the United States and other countries serving as a model of what was possible. Similarly, the IOM carefully studied the best state regulatory models, drawing on their example of what could be achieved.

Second, the 1986 IOM report clearly documented severe and widespread quality problems and significant deficits in existing regulatory policies and practices. This placed the issue squarely on the public agenda, giving impetus to the formation of a coalition of consumer advocacy groups, unions, health care professionals, and nursing home industry representatives.

Third, this diverse coalition, called the Campaign for Quality Care and organized by the NCCNHR, had a major impact on the legislative process. The coalition met weekly during the period following the release of the IOM report to develop a list of priority areas for legislation. The coalition's members ultimately reached consensus on recommendations and presented these recommendations, as a group to Congress (Edelman 1997–98). The formation and action of this fragile coalition—the first to meet at the national level and to vote and reach consensus on key issues to take before Congress—was a major factor contributing to the passage of the OBRA '87 reforms.

OBRA '87, rooted deeply in the IOM committee's recommendations, represented a comprehensive and innovative approach to regulation. First, it mandated simultaneous changes in three fundamental components of the federal regulatory system: (1) the standards, (2) the process for surveying or inspecting homes and determining whether facilities were in compliance with the standards, and (3) the enforcement system for dealing with those facilities that failed to comply with the standards.

All three components were to be resident-focused and outcome-oriented. For example, the new standards spoke to the process of care that was expected and to the requirement that care would promote "the maximum practicable functioning" for each individual resident. Further, OBRA '87 gave residents' rights and quality of life the same degree of regulatory importance as the provision of high-quality care. The new inspection procedures also focused on process and outcome quality. They incorporated interviews with residents, families, and ombudsmen about residents' daily experiences in the homes as well as direct observation of residents and the care they were receiving. Finally, the new enforcement system focused on outcomes, mandating that states adopt a wider array of sanctions to be used when unacceptable outcomes were identified. Their use was to be tailored to the nature and severity of the deficiencies found during surveys or complaint investigations.

The Problem: What Led to the RAI?

One of the central IOM recommendations was that the Centers for Medicare and Medicaid Services (CMS)7 should mandate use of a uniform nursing home resident assessment instrument, initially propounded by IOM member and committee chair Sidney Katz, M.D. This recommendation emerged from several concerns. Several prior studies that had identified significant and widespread quality of care problems had attributed them to nursing facilities' failure to adequately identify and treat residents' problems and care needs. Assessment information was often inaccurate, incomplete, and unrelated to the care plan (Hawes et al. 1995; Morris et al. 1990). For example, nursing homes typically underdiagnosed and undertreated residents' mood problems and delirium (Liston 1982; Rovner, German, and Brant 1991). Similarly, overuse of physical restraints and inappropriate use of psychotropic medications were widespread (Evans and Strumpf 1989; Harrington, Tompkins, and Curtis 1992; Kane et al. 1993; Ray, Federspiel, and Schaffner 1980). Other poor care practices were also common, including overuse of urinary catheters; deficient treatment of incontinence; inadequate prevention and resolution of pressure ulcers; inattention to nutritional problems; a lack of regard for hearing, vision, and dental problems; and inadequate psychosocial interventions, including a lack of effective behavior management programs (Gugel 1989; Himmelstein, Jones, and Woolhandler 1983; Howard, Strong, and Strong 1977; Marron, Fillit, and Peskowitz 1983; Michoki and Lamy 1976; Ouslander and Fowler 1985; Ouslander and Kane 1984; Ouslander, Kane, and Abrass 1982; Schnelle et al. 1988; Starer and Libow 1985).

The need for uniform and comprehensive assessment in long-term care was well recognized (Kane and Kane 1981; Katz 1983), and it was clear that one mechanism for solving these problems would be to improve assessment and care planning for residents. An assessment of each resident's strengths, needs, and preferences constitutes the foundation for developing an appropriate individualized plan of care. Moreover, the positive effects of comprehensive geriatric assessment on the health and well-being of older persons had been demonstrated in several studies (Applegate et al. 1983; Hendricksen, Lund, and Stromgard 1984; Rubenstein, Rhee, and Kane 1982; Tulloch and Moore 1979).

Finally, prior studies had suggested that the standards governing nursing homes were too focused on structural aspects of quality. As a result, the survey process had usually focused on the "capacity" of the facility to meet those structural standards rather than on the quality of care and quality of life experienced by residents on a daily basis (IOM 1986; New York 1976; Ohio 1979; U.S. DHEW 1975, 1976). The IOM therefore wanted a uniform assessment instrument that could provide data about the resident at admission and periodically thereafter.

The IOM recommendation on uniform assessment was intended to achieve the following objectives, at a minimum:

• To improve clinical practice in nursing homes, ensuring that all residents would have a comprehensive assessment that would inform the development of each resident's individualized plan of care
• To refocus the regulatory process on individual residents—the quality of their daily lives and well-being over time—rather than on facilities and facility capacity
• To develop quality indicators (from data about the resident gathered at admission and periodically thereafter) that could be used to generate resident-centered measures of process and outcome quality

The Opportunity

The conditions that led to the passage of OBRA '87 represented an opportunity for the development and implementation of the uniform resident assessment system supported by the IOM. As Weissert and Weissert (1996, 237) observe, "Nonincremental change sometimes occurs first with an incremental step that provides knowledge to policymakers and mobilizes public or group support. In a second stage, convergence toward comprehensive reform occurs." In the case of the OBRA '87 reforms, the "incremental step" was the IOM committee and the conditions that led to its existence. The committee, in carrying out its charge, did the very things Kingdon (1984) described as part of the "softening up" process that is essential to comprehensive reform. It held public hearings; members gave speeches; and the chair of the committee, along with two or three other members, met with congressional staff to brief them on what the committee was finding and to set the stage for legislative enactment of the committee's recommendations.

As noted earlier, the IOM committee's activities heightened the visibility of nursing home quality issues at the federal level, helped move nursing home issues onto the public agenda, and provided a comprehensive set of recommendations around which groups could coalesce. In addition, the IOM committee lent its expertise to defining and solving the problem, giving the task an aura of "science" and relieving congressional committees of the time-consuming and highly technical task of developing legislative proposals from scratch. Finally, the OBRA '87 reforms were the culmination of more than a decade of criticisms and calls for reform—what Kingdon (1984) called the "overwhelming power of an idea whose time has come."

The remarkably broad-based support for the OBRA '87 reforms represented a singular opportunity for comprehensive reform rather than the incremental efforts that had preceded OBRA. Not since the passage of Medicare and Medicaid had there been such a high level of agreement on nursing home reform issues among consumers and providers. Not only would Congress pass the legislation, but there was a good chance that the letter and spirit of the law, including the resident assessment provision, would be fully implemented.

Thus, in OBRA '87, Congress required that nursing facilities conduct a "comprehensive, accurate, standardized, reproducible assessment of each resident's functional capacities." Congress referred to this as the Minimum Data Set (MDS) and specified the basic functional areas that had to be included in the MDS. The CMS was charged with developing the instrument and regulations for implementing this requirement. This assessment became known as the Nursing Home Resident Assessment Instrument, or RAI, and incorporated the MDS.

The Constraints

The fundamental constraint facing policymakers and the research team charged with developing the RAI was the high likelihood that the consensus among providers and consumers would end. It is common for an industry that will be subject to a regulatory system to attempt to exert its influence on the agency charged with implementation (Thompson 1983; Wilson 1989). A breakup of the Coalition for Quality Care would have set the stage for the nursing home industry to weaken or delay the regulations implementing OBRA '87. Indeed, this has been a common pattern in other areas of health policy: recognizing that they are trapped in the converging streams of problem definition, policy solutions, and a changing political climate, interest groups often abandon the legislative battle and instead attempt to delay a new law's implementation or to chip away at the regulations promulgated to implement it (Weisskopf 1991).

Bureaucracies often play a significant role in formulating policy, but perhaps their most significant role is in implementing policy and administering programs over time (Morone 1993). Moreover, the "real," or operational, policy is the policy that is actually implemented rather than the policy enacted by the legislative body. In health care policy as in other areas, the bureaucracy charged with developing regulations and implementing legislation often modifies the final impact of a policy in order to address competing interests or goals, to respond to special interest groups, or to make a program work (Altman and Sapolsky 1976; Wilson 1989).

Certainly, CMS staff had ample evidence to support the notion that if they went "too far" in the regulations they developed to implement the OBRA provisions, they would arouse the ire of the industry. For example, in 1978 Congress had mandated uniform hospital reporting for those institutions participating in Medicare and Medicaid programs. In the view of associations representing the hospital industry, the agency overstepped its bounds in drafting the regulations and developing the manual that specified the contents and processes of the uniform reporting system. The hospitals complained to Congress, and the House of Representatives voted to eliminate the funds the agency needed to implement the program. Eventually, a compromise was reached, one that reduced the amount of data to be collected to a level the hospitals deemed was not too intrusive (Thompson 1983). Like all bureaucracies, CMS had to walk a delicate balance in implementing the OBRA '87 reforms while attempting to avoid alienating powerful external interest groups (Meier 1985; Wilson 1989). This was particularly true with respect to development and implementation of the RAI.

This dual possibility—that the Coalition for Quality Care would dissolve and that CMS might engender industry opposition—was particularly troubling, because earlier attempts to implement a uniform resident assessment had failed. During the mid-1970s, the Office of Nursing Home Affairs, then part of what is now the Department of Health and Human Services, sponsored development of a Patient Appraisal and Care Evaluation. This form was to be used for resident assessment, but the goal was that the resulting resident-level data would eventually be used in a more outcome-oriented survey process and as the basis for reimbursement (Lynch 1976; U.S. DHEW 1976). The nursing home industry argued that the system was too burdensome and complex for implementation as a regulatory requirement. As a result, the agency abandoned the proposed regulation and instead published the instrument as a tool nursing homes could use voluntarily (IOM 1986). Similarly, the Carter administration proposal on uniform assessment, abandoned by the Reagan administration, was partly the victim of renewed industry opposition. As previously discussed, the substantial difference between the administration's estimate of the cost impact of the proposed regulation and the one put forward by the nursing home industry doomed the proposal (Applied Management Sciences 1980; IOM 1986; U.S. DHHS 1980).

Finally, cost was a constraint. OBRA '87 required that the federal government and the states recognize any increased provider costs associated with meeting the new standards and adjust Medicare and Medicaid rates accordingly. Growing concern about federal budget deficits and the rising cost of nursing home care meant that any added cost would be a significant consideration in evaluating the utility and feasibility of regulations implementing the OBRA '87 mandates.

OUTREACH: CLINICAL AND TACTICAL DECISIONS IN DEVELOPING THE RAI

CMS and the research team that developed the nursing home resident assessment system were well aware of the constraints and potential pitfalls they faced. It was clear from the outset that the development of the RAI would involve both scientific and political decisions and actions.

CMS came to the RAI8 development process with a clear understanding of the constraints and potentially countervailing forces it faced. CMS staff recognized that the OBRA '87 reforms were so sweeping that CMS might increase OBRA's effectiveness by focusing special attention and effort on selected aspects of the legislation. As a result, CMS staff made a tactical decision to concentrate their initial efforts on some of OBRA '87's most basic provisions. These included implementing resident assessment, eliminating (or at least reducing) the use of physical restraints, reducing the inappropriate use of psychotropic medications, strengthening residents' rights and enhancing their quality of life, and enhancing the training and testing of nurse's aides. CMS viewed these as achievable goals, strongly related to resident well-being and amenable to influence by CMS interventions such as training and the survey process.

CMS also worked closely with the independent research team that had been selected to develop the RAI. It was clear from the outset that the research team was charged with developing an assessment system that was clinically relevant, feasible for the existing industry, and defensible in terms of the burden it would impose on providers. While the last two goals could be viewed as political, the project team also viewed them as valid from a quality of care perspective. In the research team's view, the RAI would be a failure if the existing provider community could not reliably use the assessment system. Similarly, achieving clinical excellence would be meaningless if, in the process, the RAI were to generate so much political opposition from the nursing home sector that the regulations would be abandoned. At the same time, the research team and CMS felt strongly that the chief aim of the RAI system was to improve clinical practice in nursing homes across the country. To achieve these multiple goals, the project team made several decisions.

Content of the RAI

One key decision concerned the RAI's scope. Initially, the research team focused on only one of the two basic substantive components of the RAI, the Minimum Data Set for Nursing Home Resident Assessment and Care Screening (MDS). The MDS is the core functional assessment instrument in the RAI and covers such congressionally mandated domains as physical functioning in the activities of daily living (ADLs), cognition, continence, mood, behavior, nutritional status, communication, vision, and psychosocial well-being. Development of this component was the project team's first task. The team also articulated the purpose of the MDS assessment, which was to identify a resident's strengths, preferences, and needs in key areas, providing a comprehensive picture of the resident's functional status in language that would be common across all disciplines. This assessment was intended to help the nursing facility staff develop a comprehensive and individualized plan of care for each resident.

The team had 18 months in which to develop, test, and revise the RAI before the deadline by which the secretary of the Department of Health and Human Services had to formally designate the instrument called for in OBRA '87. Twelve months into the project, CMS directed the research team to include additional assessment guidelines in the RAI. The team had been working on 18 problem-focused Resident Assessment Protocols (RAPs) in specific problem areas or conditions that were common among nursing home residents or that represented a severe health risk if unaddressed. The RAPs represented an attempt to summarize the best available information on the nature, prevalence, and causes of problems and conditions afflicting nursing home residents; by helping staff to recognize these problems and conditions, the RAPs could assist them in making key care-planning decisions. Each RAP included a set of MDS items known as triggers—specified responses to these MDS items that would indicate that the resident being assessed had a problem, a risk for decline, or the potential for improved function. If such triggers were present, the RAP provided guidelines that specified additional, highly focused assessment to identify treatable causes of the condition and to focus decisions about care planning for the resident.

The research team initially intended to issue these RAPs, along with suggested care planning guidelines, as educational materials; their use by facilities would be voluntary rather than part of a regulatory mandate. One key CMS staff member, however, argued that the RAI's success would rest on its ability to inform the development of an individualized plan of care for each resident based on his or her needs, strengths, and preferences. This CMS policymaker envisioned the survey system as providing incentives for the actual provision of care and services specified in the care plan. The argument was convincing, and CMS moved to bring the mandated assessment system as close as possible to one in which there would be clear and specific guidelines regarding decisions about the content of the care plan. CMS asked the project team to include the RAPs along with the MDS in the resident assessment instrument. At the same time, this decision entailed stripping away the explicit material on what should be included in a care plan for each RAP area, since the congressional mandate on standardized assessment could not possibly be stretched to include explicit guidelines on the interventions that should be implemented in the care plan. Thus, the CMS bureaucracy's clinical and quality of care concerns dictated the expansion of the RAI to include the RAPs while its tactical concerns, driven by the desire to avoid generating provider ire, limited the content of the RAPs and explicit links to care plan specifications. The decision to include the specialized RAP assessment guidelines in the RAI significantly increased the clinical utility of the MDS. At the same time, the decision not to include the material that would have linked the assessment information to specific care plan interventions probably limited the system's eventual clinical impact.9

National Advisory Committee

CMS specified that the RAI project would have a national advisory committee that would include representatives of consumers, providers, and regulators.10 The initial goal was to ensure that all the key stakeholders, many of whom had participated in the Campaign for Quality Care, would be routinely informed about what was going on and would be consulted about key decisions. Research team members, particularly John Morris, the task leader for development of the RAI, felt that the advisory committee could be used even more creatively, and so its members were asked to volunteer as reviewers or to join substantive workgroups. The substantive workgroups were assigned the task of developing items, response categories, and definitions in 18 different areas, or domains, of the MDS (e.g., nutrition, activities, behavior, psychosocial well-being). They were also asked to work on the development of the second component of the RAI—that is, the 18 RAPs. The workgroups were comprised of project staff, members of an expert clinical panel, other consultants who were clinical experts, and volunteers from the advisory committee or the organizations they represented.

This use of the advisory committee had multiple advantages. It expanded the substantive expertise available to the research team, and it involved key stakeholders in critical decisions about the focus and content of the RAI. By involving them deeply in the development process, the project team virtually ensured that the key stakeholders would come to view the clinical utility of the system as their main objective. This buy-in by nearly all the stakeholders, and certainly by those who actively participated in the workgroups, helped maintain unity of purpose and vision. This decision also increased, or at least maintained, political support for the RAI among many of the groups that had participated in the original Campaign for Quality Care, including the provider associations. Finally, the involvement of key stakeholders from groups representing consumer advocates, nursing homes, ancillary providers, and nursing home staff helped ensure the feasibility of the system that was developed.

Burden

As has been noted, a major reason for the failure of CMS's earlier attempts to require comprehensive resident assessment was that the systems were viewed as too complex or unwieldy by the nursing home industry. In addition, the industry was able to portray the system proposed during the last years of the Carter administration as too costly. The RAI project team recognized the danger that its proposals could be cast in the same light, and it made three decisions designed to minimize this risk:

• First, the team concluded that it had to collect meaningful data from nursing homes on their existing assessment practices, including data on which staff members participated in assessment and care planning. Facilities were also asked to report the amount of time spent on assessment per resident. Similar data were collected from these facilities about their use of the RAI during reliability trials.
• Second, the research team decided that the additional "burden" the RAI placed on providers was acceptable. The data indicated that, on average, facilities would spend about 20 minutes more per resident on the MDS admission or annual assessment than they were currently spending. CMS and the project team concluded that this increase was defensible, especially since the difference was expected to drop slightly (by about 5 to 10 minutes) as staff became more familiar with the RAI and as they began to maintain medical records in all disciplines in a way that was consistent with the RAI. CMS, the team, and the advisory committee reviewed the scope and content of the RAI and were unable to identify any major sections or domains that could be eliminated without compromising the clinical utility of the assessment process. They therefore concluded that an admission assessment that took an hour, on average, and an annual reassessment that typically took 45 minutes did not present an unreasonable burden and were probably defensible, since it added key domains (e.g., mood and behaviors) to the usual assessment and given the overwhelmingly positive response to the RAI by facility staff participating in field trials (see below).
• Third, the team recognized the importance of being able to defend the RAI against the argument that it increased paperwork and took time away from resident care. The team concluded that one way to address this potential criticism would be to document the benefits of the assessment. Thus the project team asked detailed questions of the facility staff who participated in the various reliability and validity field trials, inviting them to identify items in the MDS that they felt were superfluous, MDS items that were not part of their standard assessment process, and any sources of information required by the MDS that they did not routinely use under current practices. The project team also asked participating facility staff whether they felt that the additional time associated with the RAI was worthwhile.

The results of this survey were striking and helped demonstrate the value of the additional effort. The facility nurses who tested the MDS reported that the MDS covered more domains or areas than the assessment instruments they were currently using. Those additional domains included such areas as mood, psychosocial well-being, oral status, and indicators of rehabilitation potential. Two-thirds of the facility nurses who participated in the trials felt that the additional time this took was "definitely worthwhile," and an additional 25 percent felt the additional time was "probably worthwhile" (Hawes et al. 1995; Morris et al. 1990). Politically speaking, however, the most useful finding was that the majority of facility staff reported that using the MDS required them to spend more time speaking with residents and nursing assistants. All the facility nurses reported that this was valuable and should not be abandoned (Morris et al. 1990). The research team felt that framing the issue as one of giving residents a greater voice in their care would defeat any possible opposition to the plan.

Feasibility

Feasibility was also a critical issue. CMS and the project team felt that if the system were not feasible for use nationwide by a wide variety of facilities and in a wide variety of policy environments, it would lose political support. Enhancing the feasibility of the assessment process involved several tactical decisions aimed at maintaining political support for the RAI. These included the following:

• Recognizing that there were very different policy environments and practice patterns across the country, the team decided to conduct field tests in a wide range of facilities (differing in size, staffing, and ownership) and in states with differing regulatory climates and with different Medicaid payment rates. The states in which trials were conducted—Connecticut, Massachusetts, Minnesota, North Carolina, Ohio, and Tennessee—were specifically selected because of the variability of average R.N. staffing in their facilities and their average Medicaid rates above and below the national median (Hawes et al. 1995).
• In addition, the research team felt the RAI system had to be feasible for the average nursing home. This meant that the assessment and care planning decisions in the MDS and RAPs had to be amenable to use by the type of staff found in the average nursing home. At the time, most U.S. nursing homes did not have geriatric nurse practitioners on staff or ready access to geriatricians or geropsychiatrists. This affected the content of the RAPs to some degree, limiting the team's ability to specify what might be the most effective assessment strategies and interventions.
• Further, the project team did not require that facilities use multidisciplinary assessment teams, even though they viewed this as the most effective process. The RAI development team concluded that the critical task was to specify the result that was expected—a comprehensive, functional assessment—and that, within certain limits, it should allow facilities to determine the processes that would produce this result. 11 The requirement for interdisciplinary assessment and care planning teams had proved to be a serious impediment to the Carter administration's proposal, and the research team recognized that some facilities would clearly have a difficult time implementing any requirement for multidisciplinary teams. But the most compelling reason for not requiring interdisciplinary teams was the lack of conclusive empirical evidence at the time that multidisciplinary team assessment was essential to achieving accurate and comprehensive information. Indeed, the reliability trials indicated that one registered nurse who interviewed staff across all relevant disciplines could produce an accurate and comprehensive assessment of a resident's status. Thus, while the RAI development team preferred interdisciplinary teams, it did not have sufficient evidence to mandate their use, especially given the tactical considerations against adding such a requirement.
• Finally, the research team sought to achieve a high level of inter-rater reliability for the MDS in settings in which time constraints were a reality. Early field tests revealed that staff did not have the time to refer to a user's manual very often. The reality of low levels of training and continuing education for staff, combined with high staff turnover in nursing homes, presented serious challenges, and the research team felt that it might be overly optimistic to assume that all facility staff would attend adequate training sessions or consistently use the manual to clarify definitions or instructions. The project team therefore concluded that the best safeguard for effective operational use of the assessment system would be to include as much information as possible on the MDS form itself about the meaning of items, response codes, and key assessment instructions (e.g., the relevant assessment time frame).

Many of these decisions protected the RAI during the process of development and the early stages of implementation. Although the nursing home industry had participated in the Coalition for Quality Care and supported the OBRA '87 legislation, some segments of the industry lobbied CMS heavily during the RAI development phase.

For the first several months of the development, staff from the American Health Care Association (AHCA) and a group representing the major proprietary nursing home chains met at least monthly with CMS. In these meetings, industry representatives argued that the MDS went far beyond what Congress had envisioned when it enacted OBRA '87. They asserted that a "minimum" data set would consist of a few items with simple "yes" or "no" response categories rather than a five-page instrument with some 300 assessment items. These industry representatives continued to assert that CMS should cancel the project or at least circumscribe the latitude given to the RAI research team by limiting the size of the instrument to no more than two dozen or so items. CMS, however, held firm, mollifying the industry representatives by continuing to meet with them and assuring them that their concerns would be taken into account. Meanwhile, the development and testing process continued without interference.

Eventually, the industry representatives brought these meetings to a halt, largely because of the very positive reports they were receiving from their clinical staff who were members of the advisory committee or who were participating in the clinical workgroups. The field testing process, the generally positive responses of facility staff to the instrument, and the responsiveness of the research team to suggestions from facility staff, advisory committee members, and other clinical reviewers helped allay provider fears. Moreover, some of the large, multistate chains came to recognize the benefit of a uniform nationwide assessment system, which would eliminate many of the varying state requirements with which they had to contend. None of this would have been possible, however, without the support of many within CMS, both at the time of implementation and in subsequent years, who understood the clinical utility of the RAI as well as other potential uses for the MDS data. Such uses included generating quality indicators from data gathered in the survey process. In addition, other CMS bureaus were already working on proposals to use RAI data for determining residents' eligibility for Medicare-covered skilled nursing facility benefits and for determining Medicare and Medicaid payment rates based on resident acuity.

OUTCOME: EFFECTS OF THE RAI AND OTHER OBRA '87 PROVISIONS

The early effect of these reforms on nursing home residents' quality of care and quality of life was impressive. During the first three years of implementation of the RAI and other OBRA '87-mandated federal regulations, several studies found significant improvements in quality of care and resident outcomes. The most striking improvements were the substantial reductions in the inappropriate use of physical restraints and psychotropic drugs (see, e.g., Garrard, Chen, and Dowd 1995; Hawes et al. 1997; Kane et al. 1993). CMS had concentrated on these areas in its implementation activities, which included providing training to surveyors and a heightened focus on these issues in the survey process. In addition, the mandated resident assessment processes associated with the RAI focused on reducing the use of physical restraints and on identifying psychotropic medication use that had a significant negative effect on residents. There was widespread acceptance of these changes in practice patterns among the provider community.12

CMS also commissioned an evaluation of the effects of the RAI as part of the contract for its development.13 This quasi-experimental study used a complex, multistage, probability-based sample design, with cross-sectional and longitudinal data collected before (1990) and after (1993) implementation of the RAI for two independent cohorts (n > 2,000) of residents in a random sample of 254 nursing facilities in ten states (Fries et al. 1997; Hawes et al. 1997; Mor et al. 1997; Phillips et al. 1997).

The CMS-funded evaluation and other independent studies found significant improvements in a variety of other process quality measures. These included both decreases in bad practices, such as use of indwelling urinary catheters and physical restraints, and increases in good practices, such as accurate medical record-keeping, comprehensive care plans, advance directives, toileting programs, and activities for residents (Hawes et al. 1997; Marek et al. 1996; Teno et al. 1997). One study also found improved resident outcomes after implementation of the OBRA '87 regulations, with reductions in the rate of decline among elderly residents in such areas as physical and cognitive functioning and continence (Phillips et al. 1997). Consistent with this finding, there was also a significant reduction in hospital admissions of nursing home residents. (Mor et al. 1997).

These empirical findings of improved quality were supported by the reports of facility nursing staff and administrators who were surveyed about the RAI's effects on quality. For example, significant majorities of administrators and directors of nursing reported that the MDS, RAPs, and training manual had improved the quality of assessment, problem identification, and care planning in their facilities (Hines et al. 1994; Marek et al. 1996). The positive effects of the RAI and other aspects of the OBRA '87 nursing home reforms were also presented in testimony before members of the U.S. Senate by state and federal surveyors, ombudsmen, facility owners, medical directors, residents' family members, researchers, and the CMS administrator (U.S. Senate 1995; Vladeck 1995).

Although the evaluations of various aspects of OBRA '87 all pointed to significant improvements in quality, whether measured as changes in the process of care or in resident outcomes, the improvements were not universal (Fries et al. 1997; Hawes et al. 1997; Phillips et al. 1997). In addition, many observers have argued that quality of care in the nation's nursing homes started to decline again after the mid-1990s or have noted significant remaining quality problems (IOM 2000; U.S. GAO 1998, 1999a, 1999b, 1999c). These realities illustrate the difficulty of sustaining reforms and achieving permanent and widespread improvements in quality of care and quality of life in nursing homes.

SUSTAINABILITY: WHAT ARE THE PROSPECTS FOR THE INTERVENTIONS AND OUTCOMES DISCUSSED ABOVE?

Whether or not the RAI will survive and the positive effects of OBRA's implementation persist depends on forces at least partially exogenous to the nursing home industry. In this section, we identify the factors that suggest that the RAI, other OBRA '87 reforms, and their positive effects are likely to survive despite their continued susceptibility to reversal. The incident that best exemplifies the difficulty of sustaining regulatory reforms is the 1995 proposal to block-grant Medicaid. This episode demonstrated the political vulnerability of the RAI and other OBRA '87 reforms and the underlying factors that make nursing home reforms difficult to achieve and even harder to maintain over time, but it also illustrates how research can be used to influence public policy.

A Changed Political Environment

In 1994, the Republican Party won a majority of state governorships and took control of the U.S. Congress from the Democrats. Republicans embraced new House Speaker Newt Gingrich's (R., GA) "Contract with America," which, among other things, emphasized deregulation and shifting control from the federal government to the states.

The first shot at the OBRA reforms came from the National Governors Association (NGA) in January 1995. The NGA, now comprising 32 Republicans, 17 Democrats, and 1 Independent, adopted a Medicaid policy position that criticized OBRA '87 for placing the federal government in the position of "micromanaging the nursing home program" (Pear 1995a). Congress followed suit in March, with committee chairs floating the idea of giving responsibility for nursing home regulation to the states. This was an attractive idea in the context of recreating Medicaid as a program of block grants to states and, in the process, enabling the balancing of the federal budget by 2002 while still paying for the tax cuts being proposed by House Republicans (National Journal's Congress Daily 1995a). Under the block-grant proposal, states would receive fewer federal Medicaid dollars but would enjoy greater authority over how those dollars were spent. State governors would likewise have greater autonomy to set and enforce quality and staffing standards for nursing homes. This notion appealed to the many governors who viewed Medicaid expenditures on nursing homes as the "black hole" of state Medicaid budgets.

A successor to the Campaign for Quality, known as the Long-Term Care (LTC) Campaign, was formed to fight these proposals (National Journal's Congress Daily 1995b). Of the 141 groups represented by the campaign, most had a general interest in Medicaid policy but only a secondary or tertiary interest in nursing home quality (these included the AARP, the Alzheimer's Association, the American Public Health Association, and the AFL-CIO). The NCCNHR was one of the few member-groups for which the OBRA '87 regulations were a primary concern.

The campaign initially focused on getting one or more of the key congressional committees to back away from Medicaid block grants and from eliminating federal regulation of nursing homes. This effort met with little success. Proposals to give the states authority and control over Medicaid continued to be made, without any public hearing at which resident advocates could present their views about the likely impact of this policy on nursing home quality. Even opposition from the White House, voiced at the 1995 White House Conference on Aging, did not affect the Congress.

In the fall of 1995, the House passed the Medicaid Transformation Act, which would have repealed Medicaid and recreated it as a block-grant program. It replaced the OBRA '87 federal regulations with meager requirements that essentially eliminated federal regulation and allowed states to establish and maintain standards for nursing homes. This would have meant the end of the RAI and would have effectively rolled back standards to a situation with greater variability and even fewer constraints than the pre-1987 situation so roundly criticized by the decades of state and federal studies.

Consumer advocates were stymied. The House held no hearings on the bill's nursing home provisions, and the LTC Campaign seemed unable to find a meaningful public forum at which to present its views. The campaign's hopes rested on Senate reversal of the House's dismantling of the federal regulations governing nursing homes, but there was no comprehensive strategy for achieving this outcome.

At this point, the research community became a key player in the policy debate. Members of the research community entered into discussions with the public policy staff of the Alzheimer's Association and other members of the LTC Campaign, particularly AARP and NCCNHR. These discussions focused on the important role that the positive evidence concerning the effects of OBRA '87 might play in any debate over abandoning the nursing home reforms.

These discussions made it clear that research findings about the effects of the RAI and other OBRA '87 provisions could be useful adjuncts to testimonials from family members and resident advocacy groups. The policy experts from these organizations noted that the research findings would lend both scientific legitimacy and the weight of numbers to anecdotal stories from family members and providers. Members of the LTC Campaign worked with Senate staff to set up three hearings at which research findings were presented along with testimony from family members, consumer advocates, providers, ombudsmen, state regulatory agency staff, and the CMS administrator (U.S. Senate 1995; Vladeck 1995).

It also became clear that research findings had to be presented to policymakers in a less technical way than that demanded by peer-reviewed journals and that, in this era of fiscal constraint, researchers had to address the cost implications of study findings. The RAI development and evaluation team, for example, had found that, against all expectations, hospital use by nursing home residents declined after the implementation of the RAI and other OBRA '87 provisions, with no increase in mortality. To make this finding more salient to legislators, the research team extrapolated the study's result to the nursing home population as a whole. They used data from 1993 (the year the reduction was observed) to estimate the overall fiscal impact of the decreased hospital use. (Researchers assumed that such reductions, observed in the 254 selected facilities in 10 states studied in 1993, would be repeated nationwide.) A 25 percent reduction in hospital use by nursing home residents would, they estimated, yield a $2 billion savings on Medicare-reimbursed inpatient stays alone, not counting any of the expenditures associated with patient copayments, ambulance costs for transportation, or any subsequent post-acute care. This estimate was particularly important because it addressed assertions by the NGA and House leaders that the OBRA '87 regulations were expensive and counterproductive (National Journal's Congress Daily 1995b; Pear 1995a).

It also became apparent that science could affect policy if research findings were used to "redefine" the problem. Initially, both the NGA and the House Republicans defined the problem within the context of their perceived electoral mandate for the "Contract with America," one part of which emphasized deregulation, reliance on market forces, and shifting as much control as possible from the federal government to the states. As has been indicated, the effort to end the OBRA '87 reforms and federal regulation of nursing homes was caught up in a larger effort to reduce the budget and block-grant the Medicaid program—and thus in the battle between the Republican and Democratic parties. The Democrats tended to portray the House Republicans' "Contract with America" and their plan to block-grant Medicaid as an attempt to provide tax cuts for the wealthy at the expense of the most vulnerable members of society. Some moderate Republicans in the Senate joined the Democrats in expressing opposition to ending federal regulation of nursing homes, but these opponents did not really redefine the issue so much as state their opposition to moving responsibility from the federal government to the states. As a result, the nursing home issue did not initially generate much public attention or political heat.

The research community was able to help redefine the issue in several ways. First, most of the IOM members sent a letter to members of Congress. The letter noted that overturning OBRA '87 and giving regulatory responsibility back to the states would amount to returning to the pre-1987 situation that the IOM committee and scores of state reports had repudiated. The letter emphasized that the variability in state regulation would bring poor quality and inadequate regulation back to many nursing homes. And the letter briefly summarized research findings about the positive effects of the OBRA '87 nursing home reforms on quality of care and quality of life for residents.

Second, in their testimony and in speaking with the press, researchers translated what they believed to be the most significant findings about OBRA '87 into human terms. For example, one researcher took the quantitative finding about the magnitude of restraint reduction and converted that into the number of elderly nursing home residents who, as a result of OBRA, could be untied or never restrained at all—a total of 250,000 people each year (Hawes 1995).

Third, researchers helped the LTC Campaign and Senate supporters redefine the issue of nursing home deregulation from that of ending "federal micromanagement" and of "freeing states to make relevant decisions" to one of diminishing nursing home quality. Key researchers and former IOM committee members spoke to the press as part of this attempt to refocus the debate by redefining the issue. In both press interviews and in testimony before Congress, the researchers specifically countered the arguments about "over-regulation" and shifted the focus to quality of care and the well-being of residents.

For example, in an interview with a reporter for the New York Times, Rep. Thomas J. Bliley, Jr. (R., VA), chair of the House Commerce Committee, asserted that Congress intended to "end an eight-year experiment with federalization of nursing home standards. . . . By all accounts . . . the Washington-run nursing home regulation has been a bureaucratic nightmare: confusing, expensive and counterproductive" (Pear 1995a). But members of the LTC Campaign gave the reporter the names of three researchers whose empirical findings contradicted Bliley's assertion. An IOM committee member told the Times that ending federal regulation would cause a significant reduction in quality and be a significant step backward. Another IOM committee member called the proposal "a return to the bad old days." A researcher told the reporter that the OBRA reforms were "paying off," slowing rates of decline among residents and reducing their hospital use, saving an estimated $2 billion annually on acute care. These themes were picked up in a piece by syndicated columnist Ellen Goodman, who wrote a scathing column entitled "Remaking the Nursing Home Horror Tale" (Goodman 1995).

Democratic and moderate Republican Senators echoed this focus on the positive effects of OBRA '87 and the RAI in subsequent floor debates and newspaper interviews (Carlson 1995; Kennedy 1995; National Journal's Congress Daily 1995b). Thus the debate in the press and among supporters shifted away from arguments about "federal micromanagement" and freeing the states to protecting the elderly and preserving improvements in nursing home quality.

The success the advocates were having in redefining the issue clearly had an effect. In floor debates in both the House and Senate during the fall of 1995, Democrats and moderate Senate Republicans increasingly cited nursing home quality issues and ending OBRA '87 federal regulations as serious stumbling blocks to passage of the House budget bill. Many Democrats made the issue a major argument for rejecting the entire Republican budget proposal. Indeed, President Clinton subsequently cited the need to retain federal nursing home standards as part of his rationale for vetoing the initial budget bill.

The issue of abandoning the OBRA '87 federal standards and the adverse impact this would have on quality became a lightning rod, attracting increased attention to the split between Republicans and Democrats. Moreover, the negative press the issue generated spilled over to Republican proposals on Medicaid block grants, the budget bill, and tax cuts. Columns such as Goodman's "Remaking the Nursing Home Horror Tale" and Margaret Carlson's Time magazine piece "Back to the Dark Ages" (Carlson 1995) were matched by articles on the nursing home issue in most major national newspapers, following the lead of the New York Times and Washington Post. Sen. Robert Dole (R., KS), then the Republican majority leader, complained on the Senate floor about the attention the nursing home reform issue was getting: "I know there is a lot of politics about nursing homes. I know the liberal media bought into the spin put on by Democrats" (Dole 1995).14

As a result of the redefinition of the issue and the attention it was attracting, the Senate reversed its original concurrence with the House budget bill and retained the nursing home standards. This was done in part because moderate Senate Republicans joined the Democrats on the issue of nursing home regulation. They threatened to vote against the entire budget bill unless the federal OBRA '87 standards were retained. Some observers also felt that Dole, who at the time was running for the presidency, agreed to retain the OBRA '87 reforms in order to "deflect criticism from Democrats and deprive them of political issues that could be used in [the] elections" of 1996 (Pear 1995b).

In any such web of political machinations, it is always difficult to determine the exact, or even approximate, weight to give to individual events or factors. Some might argue that the Democrats and moderate Senate Republicans would have carried the day without the aid of policy-relevant research. Arguably, they could simply have raised the specter of reductions in quality, paraded a long line of consumer horror stories before reporters and their congressional colleagues, and carried the day in that fashion. The point here, however, is not whether the availability of policy-relevant research was the deciding factor. It is simply the recognition that those involved in a policy struggle and those reporting on that struggle saw science-based evidence as a very relevant element in the policy debate and used it vigorously to what seems to have been good effect.

Implications for the Future Sustainability of the RAI

The near-death experience suffered by the RAI and other OBRA '87 reforms in 1995 demonstrated that regulations that impose on the latitude of providers are always vulnerable. In this case, consumer advocates and their allies in Congress successfully resisted a major change.

A struggle over the CMS regulations implementing OBRA '87 mandates on enforcement was going on at roughly the same time as the battle over block-granting Medicaid and nursing home regulation. The nursing home industry, particularly AHCA, strongly opposed full implementation of the enforcement provisions, and their implementation had been delayed for nearly five years after the other major OBRA '87 regulations had gone into effect. Consumer advocacy groups tended to view this as a betrayal, and AHCA's executive director classified both the delay of the enforcement regulations and their promulgation in 1995 as major victories (Edelman 1997–98; Pear 1997; Weissenstein 1997).

In fact, AHCA had extraordinary access to the Clinton White House. As was noted earlier, a prominent AHCA member chaired the Democratic Party's Business Council and became the party's finance chairman. At the same time, nursing home interests contributed more than $1.1 million to the 1996 Clinton campaign (Weisskopf 1997). Given this reality, many within the CMS bureaucracy felt that the delay in implementation and other actions consumers viewed as "watering down OBRA" were in fact necessary to save both other provisions of OBRA and the key elements of the enforcement provisions.

What these episodes illustrate is that regulation remains a politically charged area of public life in the United States. Thus it remains an open question whether all the provisions of OBRA '87 are sustainable. Indeed, they seem particularly vulnerable when reductions in government spending and constraints on Medicaid are central issues in Washington and in the states. Further, the staffing shortages currently experienced by many nursing homes nationwide not only endanger the ability of facilities to deliver high quality care, but they also provide fodder for the debate over whether regulations are excessive.

The regulations that appear to be most sustainable are those that represent what consumers and providers identify as directly clinically relevant and helpful. For example, provider and consumer support for reducing the use of physical restraints remains widespread and strong, and few would advocate a repeal of the requirement on restraint use. Similarly, consumer advocates tend to have a very positive view of the RAI, and providers argue that the MDS has been one of the most effectively implemented of the OBRA '87 regulations. Even most administrators and directors of nursing who complained about the paperwork burden associated with the MDS acknowledged that the RAI had improved their ability to plan care and monitor the effects of interventions (Hines et al. 1994; Marek et al. 1996).

Two other factors add to the long-term sustainability of the RAI. First, many homes are using MDS data in quality improvement initiatives, to understand their resident case mix, and for management (e.g., in projecting staffing needs for different units), and many nursing home chains find MDS data useful for management purposes and prefer the uniformity of requirements across states associated with a federally mandated instrument. Second, many states and the Medicare system itself have incorporated the use of MDS data into their payment and regulatory systems, making the MDS data even more central to facility operations.15

As a result of these factors, the RAI is likely to survive. It will, however, be subject to continued attacks as "an excessive paperwork burden." If advocates of the RAI do not recognize this threat and neglect to redefine the issue in terms of the RAI's effect on the quality of care and life experienced by residents, the RAI will be vulnerable to attack.

The RAI faces three other potential impediments to its sustainability. One arises from a new federal requirement for uniform data reporting elements across all post-acute care programs. Currently, different instruments are used for reporting patient-level data in home health (Outcome and Assessment Information Set, or OASIS) and nursing homes (RAI/MDS). The issue of what will be required for rehabilitation facilities remains undecided. The initial split between OASIS and the RAI occurred, in part, because different branches of the CMS bureaucracy were charged with the development and testing of the home health and nursing home instruments. Unlike the RAI, the OASIS instrument is not intended as a comprehensive assessment that can be used to generate a care plan; it is only intended to produce indicators of the quality of an agency's performance. There has been some agitation about the CMS mandate for two different instruments by states that want uniformity of instruments and data elements across all long-term care settings and prefer the RAI/MDS to be the core of any assessment program. CMS has not resolved the issue by moving to one basic instrument, however, and such a move toward real uniformity seems unlikely. The development of a "crosswalk" formula that allows comparisons between OASIS and MDS data items seems a more likely outcome.

Another potential threat to the sustainability of the RAI involves the lack of attention it has generated among physicians, as well as some physicians' opposition to some OBRA '87 provisions. The American Medical Directors Association (AMDA), which represents nursing home medical directors, has been supportive of both the RAI and other OBRA '87 requirements (Russell 1995), but other groups, notably the American Medical Association (AMA), have largely ignored the RAI. Even during the RAI development process, the official AMA representative was only an occasional attendee at the national advisory committee meetings and did not participate in the clinical workgroups. In part, this may reflect the traditionally low salience of issues related to nursing home care for most physicians, but it may also be a reflection of physicians' general antipathy toward any system that attempts to improve quality through the use of uniform guidelines, which some physicians regard as "cookbook medicine" (McGlynn and Brook 2001). Some physicians, for example, have been particularly critical of OBRA '87's attempt to reduce inappropriate use of psychotropic medications, arguing that this regulatory initiative infringes on their ability to prescribe medications that best address the needs of individual patients. Finally, some geriatricians have criticized the RAI because it does not contain some of the items they need for particular quality measures for specific medical conditions. Despite these factors, the objections of academic geriatricians have little practical impact on the RAI's sustainability. Just as they were not involved in the development process, so most physicians (other than nursing home medical directors) have remained aloof from the politics surrounding OBRA '87, and the medical directors have been generally supportive of both the RAI and OBRA '87. The real potential harm to the RAI comes from most physicians' lack of knowledge about how to use the MDS data effectively to evaluate changes in their patients and what they can do to ensure greater accuracy of assessment by facility staff.

Finally, yet another threat to the RAI emerges from the argument advanced by some observers that the MDS data are not reliable. Granted, the field trials demonstrated that facility staff members were able to produce assessments that were reliable across assessors (Hawes et al. 1995; Morris et al. 1990), and the evaluation found that there had been an overall improvement in the accuracy and comprehensiveness of information in residents' medical records after implementation of the RAI (Hawes et al. 1997). And in a similar vein, some state Medicaid agency evaluations found that facility assessment data were, on average, accurate (Boulter 1995; Jobe 1995). Despite this, concerns about shortages of registered nurses, high staff turnover, limited training, and facility commitment have translated into disquiet about the accuracy of the MDS data provided by facilities (Harrington et al. 1997). Whatever the reality of facility performance, it is difficult to attribute the problem to the RAI. If facilities do not adequately train staff or put needed assessment practices in place, and if state regulatory agencies do not insist on accuracy, the problem will continue to exist, regardless of the nature of the instrument.

LESSONS

Two different kinds of lessons have emerged from the RAI's history: lessons related to the development and implementation of reforms and lessons about regulatory reform and the politics of the policy process.

Implementation Lessons

Many of the lessons about developing and implementing legislative reforms emerge from the decisions made by CMS and the research team during the development and testing of the RAI. These important decisions have contributed to the clinical utility and political viability of the instrument and the assessment system. Some lessons about what might have been done differently have also emerged over time. Among the implementation-related lessons are these:

• It is probably impossible to achieve comprehensive quality reform without adequate staffing in facilities. The IOM recommendations and the resulting OBRA '87 nursing home reforms failed to adequately address issues related to nursing home staffing. They gave no recommendation on staffing levels, other than minimums for licensed staff. For example, the issue of staff-to-resident ratios for certified nursing assistants, who perform an estimated 80 percent of the hands-on care in nursing homes, was largely ignored. As a result, most nursing homes implemented the OBRA '87 reforms without increasing their staffing levels for direct-care staff. These levels were already too low in the early 1990s, and the problem has intensified as many facilities experience difficulty in hiring and retaining sufficient staff even for inadequate staff-to-resident ratios.
• More attention needs to be paid to training. The research team wrongly assumed that the national and state associations, in conjunction with the state regulatory agencies, would offer high quality, ongoing training in assessment and care planning to facility staff. Unfortunately, this did not occur uniformly across the United States. Indeed, too much of the training that was provided concentrated on the minutiae of "filling out the MDS form." What was needed was more training in how to link the MDS data to the RAP guidelines, a process designed to identify treatable causes of functional deficits and risk factors for decline that could be addressed in a well-specified and executed care plan. Sadly, many of the staff working in long-term care lack the time or knowledge to use the assessment data to develop and implement an effective care plan. (Italy provides a different lesson—one that other countries might follow. In at least one region, Italian nursing homes have used the RAI and RAPs as part of a comprehensive effort to educate clinical staff working in long-term care about geriatrics and about linking assessment information to effective care-plan interventions for nursing home residents.)
• More attention should have been paid to ensuring that administrative staff, including directors of nursing, understand more about the potential multiple uses of the MDS data. Once administrators understand that they can rely on the data for a variety of purposes—from evaluating care interventions, to allocating staff resources, to benchmarking their own facilities' performance—they should have a greater commitment to ensuring that the data are accurate. Some training efforts should therefore have focused more directly on how facility administrative staff can use the MDS data to plan, manage, and evaluate more effectively.

Policy-Reform Lessons

The passage of the OBRA '87 nursing home reforms and the subsequent implementation of the RAI, as well as their near-repeal, reveal some important lessons about attempts to improve quality in American nursing homes:

• Legislation, rather than administrative action, may be needed for comprehensive reforms. Because of the highly politicized nature of regulation in American politics, all regulations, even those aimed at improving quality, are subject to the vagaries of changes in administrations and their priorities. Legislative mandates are more difficult to abandon. Even when originally enacted with bipartisan backing and strong support from consumers and providers, however, legislated changes in federal rules governing nursing homes remain politically vulnerable, particularly if they are caught up in another, broader political agenda.
• Reform is more likely when the bureaucracy charged with promulgating and implementing the regulations views the changes as consistent with its mission. Equally important, the implementation should be carried out in a way that does not endanger the bureaucracy by arousing the ire of the nursing home industry. Efforts to create consumer and provider "buy-in" to changes in regulation—and to maintain that enthusiasm over time—are essential and well worth the effort.
• Research can play a critical role in policy formation and implementation. For the last three decades, efforts to reform nursing home care have had a history similar to that of efforts to address a host of other social problems. Such problems are only episodically the focus of significant attention from the public, the media, and legislative bodies, and they affect only limited numbers of people. Most people have not suffered from poor-quality nursing home care, and even groups that represent older people, such as AARP, do not place this topic high on their policy agenda—or at least not on a regular basis. Moreover, the cost of addressing the causes of poor quality is likely to be substantial. In such a situation, researchers can help maintain at least a low level of attention to the problem by consistently casting their results in terms of policy issues. They can also work to translate and communicate those findings in a meaningful way to consumer advocates and policymakers. Strategic thinking and planning can create "alarmed discovery" of the depth and seriousness of a problem—placing the issue higher up on the policy agenda and opening a window for incremental or comprehensive reform. And research can also lead to more effective regulations, particularly if it focuses on issues of clinical utility, feasibility of implementation, and cost impact.
• When "alarmed discovery" of a quality problem occurs, consumer advocates must recognize that their window of opportunity is limited. Using such an opportunity effectively demands rapid action to craft a coalition of interest groups and achieve consensus about the needed legislative or regulatory change. Consumer advocates should focus on research findings that identify practices and policies that promote quality and then use those findings to help redefine the issue as one of quality of care, not governmental regulation.
• Finally, policies that both consumer advocates and providers view as clinically relevant and achievable are more sustainable over time. It is therefore critically important to involve providers, along with consumers and clinicians, as partners in the development and implementation of quality initiatives.

NOTES

¹Public policy governing nursing homes is largely a function of the federal standards set for participation in the two health care programs that are the primary payers for nursing home care, that is, Medicare and Medicaid. These programs set standards, specify inspection processes, and mandate enforcement remedies that apply to all facilities that participate in the programs and receive payment for the care of covered program beneficiaries. More than 95 percent of nursing homes participate in Medicare or Medicaid (Strahan 1997). Medicare is a federally funded program that covers acute, ambulatory, and rehabilitative care for the elderly, as well as for those nonelderly persons who are permanently disabled and meet program eligibility criteria. Medicaid is a combined federal and state program that covers specified health care services, including nursing home care for persons who are poor or, in many states, defined as meeting financial criteria for being "medically needy." Because Medicare coverage is restricted to skilled, rehabilitative care, in 1998 it covered only 12 percent of expenditures on nursing homes. Medicaid accounted for 46 percent of the expenditures and helped pay for nearly 68 percent of the residents. The remainder of dollars spent on nursing home care came from out-of-pocket spending by individuals (33%) and other payers, such as private insurance and the Veterans Administration (9%). (The 68 percent of residents who are Medicaid-eligible contribute most of their monthly income—from Social Security and/or pensions—and Medicaid pays the remaining charges after their income is exhausted. This is why Medicaid is a payer for 68 percent of the residents but accounts for only 46 percent of the expenditures on nursing home care.)

²For example, after Ohio led the nation in nursing home fires resulting in multiple deaths, the state legislature enacted mandatory sprinkler requirements (Ohio 1978).

³California's "Little Hoover Commission" is a state oversight agency credited with issuing reports, such as a 1983 study of skilled nursing facilities (California 1983) that led to state reforms, including the creation of a system of citations and fines that put teeth in the regulatory system. New York's Moreland Act Commission was created partly in response to scandals in facilities owned by Bernard Bergman and to concerns that the abysmal care and regulatory failures found in Bergman's facilities were widespread. The series of reports issued by the Moreland Act Commission (New York 1975, 1976) led to some reforms in New York. Similarly, the Ohio legislature's Nursing Home Commission (Ohio 1978) found and documented significant and widespread quality problems and followed with a blueprint for needed policy changes (Ohio 1979), a residents' bill of rights, and two omnibus reform bills. The bill of rights and one reimbursement reform bill were enacted before the two-year commission expired, but the other omnibus quality improvement reform bill, which came to be known as "little OBRA," was not enacted until 11 years later, following the passage of OBRA '87 at the national level.

⁴As of 1995, 54 percent of U.S. nursing homes were owned by chains, a proportion that had increased from 41 percent in 1985 (Strahan 1997). This percentage is apparently holding fairly steady, with chain ownership at 55 percent in 1999. The ten largest chains controlled 19 percent of the facilities and 21 percent of the beds nationwide in 1999 (Vickery 2000).

⁵Many of these findings were echoed in a class action lawsuit, Smith v. O'Halloran, brought by nursing home residents in Colorado against facilities in that state as well as what is now the U.S. Department of Health and Human Services. The case, which went all the way to the U.S. Supreme Court, was decided in favor of the defendants. Among other things, the 10th Circuit Court ruled that the federal government had to substantially change the way it surveyed nursing homes. (Smith v. O'Halloran, 557 F. Supp. 289 [D. Colo. 1983], rev'd sub nom. Smith versus Heckler, 747 F.2d 583 [10th Cir. 1984].)

⁶Similarly, in Ohio, segments of the nursing home industry had defeated several proposals for a residents' bill of rights. When a fifth bill was introduced, the for-profit trade association estimated that implementing a residents' bill of rights would cost more than $1.50 per resident per day. The staff of the Ohio legislature's Nursing Home Commission estimated the cost at only a few pennies per resident day. To demonstrate their confidence in the estimate, commission staff asserted that they would perform all the notification, resident and family contacts, and hearing components of the bill for $0.42 per resident day, and use the proceeds to fund the operation of the commission, including its fraud hearings, in perpetuity. The industry dropped its cost argument, and the bill was enacted by the Ohio legislature.

⁷For the sake of clarity for present-day readers, we use CMS—the agency's current designation—throughout this paper rather than Health Care Financing Administration (HCFA), as the agency was known during the period under discussion.

⁸Contract DHHS 500-88-0055 was from HCFA's Health Standards and Quality Bureau (HSQB) for a project called Develop the Resident Assessment System and Database for Nursing Homes. The project team was led by North Carolina's Research Triangle Institute and included among its subcontractors the Social Gerontological Research Center of the Hebrew Rehabilitation Center for Aged (Boston), the Center for Gerontology and Health Care Research at Brown University (Providence), and the Institute of Gerontology at University of Michigan at Ann Arbor. The key investigators were Catherine Hawes, Brant E. Fries, Vincent Mor, John N. Morris, and Charles D. Phillips. Katharine Murphy was the lead nurse clinician, and Sidney Katz, M.D., headed the clinical panel. Key players at HCFA/CMS included Helene Fredeking, head of HSQB's Office of Long-Term Care; Wayne Smith, head of HSQB during the development of the RAI; Bruce Vladeck, the HCFA administrator; and (sequentially) project officers Allan Friedlob, Elliana Steele-Friedlob, and Sue Nonemaker.

⁹The RAI system also includes a training manual, mandated quarterly review forms, and a form on which staff summarize decisions about care planning and note the location of any planned intervention. The RAI must be completed when the resident is admitted to the facility and annually thereafter, as well as any time there is a significant change in the resident's status. The quarterly review contains a few key MDS items and is completed every three months, in the intervals between annual assessments.

¹⁰The advisory committee included representatives from nursing home trade associations, consumer advocacy groups, ancillary providers (e.g., mental health professionals, physical therapists), medical directors, nurses, social workers, and others viewed as key stakeholders. It also included representatives from state and federal agencies, including Medicaid agencies and state survey agencies.

¹¹The RAI did specify that the facility's assessment had to draw information from multiple sources, including interviews with residents and family members, direct observation of residents, interviews with direct-care staff across all shifts, and medical records.

¹²Some among the provider community had moved to reduce use of physical restraints prior to the enactment of OBRA '87. The most notable example of this trend was the "Untie the Elderly" initiative sponsored by Kendall-Crossland, a group of nonprofit facilities in Pennsylvania. Although the idea was gaining acceptance among providers and staff, few facilities had actually implemented programs or achieved significant reductions before October 1990, when reductions were required by the new regulations (Hawes et al. 1997; Kane et al. 1993). The change in use of psychotropics was nearly as striking, and there is some evidence suggesting facilities began reducing the use of psychotropics between the passage of the OBRA '87 legislation and the effective date for implementation of the new regulations in 1990 (e.g., Hawes et al. 1997; Kane et al. 1993).

¹³One issue that has been of concern to some commentators is the use of the development team for the evaluation. The obvious concern here is that the development team might be so invested in the success of the instrument that it might unwittingly fail to provide an unbiased evaluation. A number of factors make this theoretical concern one that is very unlikely to have operated in reality. First, the development team was committed to improving the quality of care and quality of life experienced by nursing home residents. Thus, like others, the team wanted to know whether the system worked as intended. More important, a number of external safeguards were in place. The evaluation design and analysis plan were reviewed and approved by a group of senior external researchers brought in by CMS to ensure that the evaluation would be scientifically sound. In addition, all the evaluation results were published in peer-reviewed journals, and thus faced careful scrutiny by anonymous reviewers. Finally, "objectivity" in research is not something achieved merely by separating development from evaluation. Developers can be scrupulous in their adherence to scientific norms and as objective as any other evaluators in their analyses of results, and any evaluator may bring a set of preconceived notions to the task. Similarly, the demands of clients can drive evaluations performed by anyone. Individual researchers—developers or evaluators—can be so wedded to a hypothesis that they unconsciously skew their observations. There are no easy structural remedies for this problem. In the end, good science demands that researchers adhere to the fundamental rules of scientific inquiry and that they let others review their design and results. Both these requirements were met by the RAI evaluation.

¹⁴Dole's comment about the media "buying the spin" Democrats put on the nur