Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Data Needs for a Changing Health Care Delivery System To establish what improvements and innovations in national data sourc- es might best be pursued and with what priorities, policy makers must first appreciate the diverse features and trends in the health care arena today and the critical questions they pose. To identify current and future needs for health care data, they must grasp the changes that have led to the present configuration of the health care delivery system and the challenges it faces. These involve shifts and innovations in the organization of services, the relationships among different service settings, financing arrangements, and the coverage of the population and, ultimately, changes in the health of that population. This chapter briefly reviews key features of and developments in health care that provide the framework for considering reform of national health data sources both the proposed National Health Care Survey and broader data systems as well. The bedrock issues relate to costs, access, and quali- ty; cutting across them are trends relating to demographics and socioeco- nomic factors, health and disease, technology, human resources, and related social, legal, and ethical questions. Because space limitations prohibit a historical overview, the focus is on current factors and circumstances; read- ers interested in details of the evolution of the U.S. health care system are referred to Somers and Somers (1961), Torrens (1978), Starr (1982), and Aaron (199lb). A survey of users of national health care statistics under- taken by the panel is also described briefly; Appendix ~ has a more com- 19
20 TOWARD A NATIONAL HEALTH CARE SURVEY plete discussion. The last section returns to the subject of data sources and systems and their desirable characteristics. THREE CRITICAL ISSUES IN HEALTH CARE Health care pundits often characterize the goals of the U.S. health care system as being access for all to high- (or at least acceptable) quality care at affordable (or reasonable) cost. The nation's present system meets none of the goals of access, quality, and cost consistently, and certainly not simulta- neously, and some observers regard health care in this country as in crisis because of the extent to which these goals are not being accomplished (Aaron, 1991a). Furthermore, it is perhaps a commonplace to note the immense complexity of U.S. society and its health care system; nonetheless, many social, economic, and technological factors heavily influence what can, and cannot, be done to achieve progress toward those three goals. Health researchers and policy makers therefore have a compelling need for more and better data on a considerable number of issues if they are to make sense of current problems and the options available to address them. The issues and data presented below are intended to illustrate some of that complexity and thereby to underscore the difficulties facing those who must design and implement programs to make data available in a timely way. Costs and Expenditures A pressing issue confronting the health care system today is the contin- ued upward escalation in health care expenditures and costs of care. The figures are compelling (Office of National Cost Estimates, 1990; Levit et al., 1991; Aaron, 1991b). A quarter-century ago, national health care ex- penditures just exceeded $40 billion, or about 6 percent of the gross nation- al product (GNP). At the start of the 1990s, outlays stood at more than $600 billion, or 12 percent of GNP. To be precise, expenditures in 1991 were $670.9 billion (12.3 percent of GNP). Estimates for 1992, 1995, and 2000 have been put, respectively, at $809.0 billion (13.4 percent), $1,072.7 (14.7 percent) and $1,615.9 (16.4 percent) (Office of National Health Sta- tistics, 1991~. These levels of expenditures well outstrip those of other industrialized countries. For example, in 1989 the per capita expenditure on health in the United States was $2,354, a figure that exceeded per capita spending for Canada by 40 percent, for Germany by 91 percent, for the United Kingdom by 182 percent, and for Turkey by 1,245 percent (Schieber et al., 1991~. Rates of real growth in health care spending for Organisation for Economic Cooperation and Development (OECD) countries for the past three decades are not very different, however, suggesting that the United States has outs-
DATA NEEDS FOR A CHANGING HEALTH CARE SYSTEM 21 pent the rest of the world for almost the last half-century (J. Newhouse, personal communication, 1991~. These disparities in spending have not translated into better (or some- times even equivalent) coverage of the population or better health outcomes as reflected in standard indices (Schieber and Poullier, 1988, 1989; NCHS, 1990, 1991~. For example, in terms of infant and perinatal mortality rates, the United States ranks twentieth and sixteenth, respectively, in a compari- son among 24 OECD countries (Schieber et al., 1991~. The picture is not much better for life expectancy, in which the United States ranks sixteenth and seventeenth for female and male life expectancy at birth, respectively. An interesting exception is life expectancy at 80 years of age, in which the United States ranks second after Canada and Iceland (female) and Canada, Japan, and Switzerland (male). One analysis of 10 Western industrialized nations, which compared them in terms of extent of primary health services, levels of 12 standard health indices, and people's satisfaction with their health care systems, found that ratings for the United States on these vari- ables were generally low (as were those of West Germany) in contrast to Canada, Sweden, and the Netherlands, which had generally high ratings (Starfield, 1991~. Virtually every part of the health care sector has experienced increases in expenditures for care both the private and the public sectors (especially Medicare); both fee-for-service and prepaid capitated systems; and both inpatient and outpatient care. For example, spending for hospital care in- creased 9.3 percent between 1987 and 1988 (to just under $212 billion); spending for physician services increased 13.1 percent (to just over $105 billion) (Office of National Cost Estimates, 1990~. Furthermore, costs to third-party payers (i.e., insured costs) and to patients and families (i.e., out- of-pocket payments) have grown; for the latter, for instance, the increase between 1987 and 1988 was more than 10 percent (Office of National Cost Estimates, 1990~. Employer-based insurance is the primary way health care is financed in the United States. Increases in employers' group health premiums have been steep in recent years: 20 to 24 percent in 1989 alone by two different estimates (Cantor et al., 1991; Sullivan and Rice, 1991~. An average premi- um increase of only 14 percent for 1990 (Sullivan and Rice, 1991) may, however, partly reflect emerging changes in the health insurance picture: a shift toward health plans with utilization management requirements, more health maintenance organizations (HMOs) and preferred provider organiza- tion (PPO) arrangements, and growth in point-of-service plans, as well as greater cost-sharing by employees (in the form of higher proportions of total monthly premiums or higher deductibles or copayment rates). In short, the rise in what the nation spends on health in percentage terms well in excess of the rate of inflation- seems so inexorable that it
22 TOWARD A NATIONAL HEALTH CARE SURVEY generates calls for rationing (Aaron and Schwartz, 1984, 1990; Callahan, 1987, 1990~. Equally firm counterarguments are advanced, however, to the effect that improving the efficiency of the health care system and reducing the provision of unnecessary and inappropriate care would help to keep the rate of expenditure growth under control and thereby forestall the need for rationing (Brook and Lohr, 1986; Wennberg, 1990a). Widely differing connotations of the term rationingas "social allocation," as "principled rationalization," and as "cost-effective retrenchment" (Brown, 1991, p. 30) tend to inhibit clear communication and complicate the debate. Many explanations for the upward spiral of expenditures have been advanced. Among them are: the effect of third-party insurance coverage in insulating people from the true costs of care; the ever-increasing, sophistica- tion of medical technologies; the changing demographics of the population; the ability of health care providers to generate demand for their services; lack of certain knowledge about the efficacy, effectiveness, and appropri- ateness of health care interventions; and rising expectations of patients and others about what health care can (or ought to be able to) do to extend life and enhance the quality of that life (or both). One of the challenges that must be met by data sources of the 1990s and beyond is to overcome the lacunae in information that keep researchers and decision makers from un- derstanding better which, if any, of these explanations are the most plausi- ble and amenable to change. The nation has seen many attempts to control rising expenditures. The shift from cost-reimbursement strategies for Medicare reimbursement of hospital care to the diagnosis-related group (DRG) prospective payment system for Medicare is one of the more dramatic examples. Direct controls on charges (price controls, fee schedules for physicians) have also been or will be tried. Efforts to constrain the use of services, as a means of control- ling costs, take several forms; higher levels of cost-sharing imposed by third-party payers and various approaches to utilization management and managed care are among the more prevalent. Incentives to use more ambu- latory (rather than inpatient) care also share the goal of controlling costs. Even the HMO movement can be seen in part as an effort to deal with the use and costs of care better than the traditional fee-for-service system has done (Luft, 1981, 1991~. More broadly, some observers believe that managed care itself an imprecisely defined and understood term variously embracing utilization management (IOM, 1989a), traditional and hybrid versions of HMOs and PPOs, and case management of high-risk patients~ffers a mechanism for constraining use and expenditures. The implications of these kinds of changes for data surveys and sources are many. They include the need for better systems for classifying and coding information on a large number of variables: sites and settings of care; personnel delivering care; types of care (with greater specificity than
DATA NEEDS FOR A CHANGING HEALTH CARE SYSTEM 23 is available today for, as examples, procedures, drugs, diagnostic tests, and other technologies); charges for care; and costs of care.) To have better data of these sorts, and to be able to track them over time and aggregate them across sources, those responsible for data collection must come to agreement on several issues. These matters include, in particular: (1) the minimum number of data elements necessary to convey specific and aggregate cost and expenditures data adequately and to link them to relevant populations and providers and (2) the operational definitions of those data elements. Access to Care Access to care refers essentially to the ability of persons needing health services to recognize that need and to seek and obtain the appropriate care in a timely way. Appropriate care can be thought of as necessary and effective care that can reasonably be expected to make a difference in the health and well-being of those receiving it; in other words, it should main- tain or improve the health status and quality of life of individuals and populations. The services of interest must be broadly conceived: . . primary care, including health promotion and preventive and screening services; specialized secondary and tertiary services, including inpatient care of the full range of acute and chronic illness, outpatient and inpa- tient care for mental and emotional disorders and for alcohol and drug abuse problems; dental care; rehabilitation therapy; and the like; various forms of sociomedical serviceshome health care, adult day care, and related social services that are aimed at maintaining ill or disabled individuals in their homes and communities and out of institutions; · long-term institution-based care; and . hospice and respite care and similar specialized services that may be provided at the end of life. Tracking the need for and use of this broad array of services presents a major challenge for the nation's data systems. Detecting the unmet need for tAs to the latter two points, economists correctly distinguish charges and costs on both theoretical and empirical grounds; policy makers and researchers often ignore the distinctions at some peril to appropriate analysis and interpretation of data.
24 TOWARD A NATIONAL HEALTH CARE SURVEY and lack of appropriate use of such services is an even more daunting task. Among the major difficulties are defining, classifying, and assessing the populations of interest in appropriate ways. Poor access to care for at least some groups in U.S. society has been an ever-present problem. It was recognized in the mid-twentieth century as sufficiently acute for elderly and poor people (especially families with de- pendent children) that the Medicare and Medicaid programs were enacted. In terms of providing insurance coverage to now more than 30 million elderly people, the Medicare program has been a remarkable success. Gen- erally, the elderly enjoy better access to care, in the forth of insurance coverage, than any other age group in the nation (IOM, 1990c; Aaron, l991b; NCHS, 1991~. Nevertheless, compared with employment-based pri- vate insurance, the Medicare program alone gives less protection against devastating out-of-pocket spending by older Americans because it does not cover prescription drugs and does not cap copayment costs for covered services (CBO, l991b). Only Medicare enrollees with employment-based retiree health benefits (about 30 percent), with so-called medigap supple- mental policies (31 percent), or Medicaid benefits (18 percent) are protect- ed to some degree from the risk of catastrophic out-of-pocket costs. By contrast, individuals and families with low incomes are not well covered by the federal-state Medicaid program. The Pepper Commission (the U.S. Bipartisan Commission on Comprehensive Health Care, 1990), for example, cites figures indicating that only 42 percent of people in pover- tyand only about 75 percent of those in extreme poverty (those with family incomes below 25 percent of the poverty line) were covered by Medicaid in 1987. Various categorical exclusions and remarkable varia- tions in eligibility rules across the states have kept a wide range of people out of the program, even if they are penniless. Some of these features are changing, however, as Congress has attempted to expand eligibility at least for pregnant women and young children and for those who experience an increase in income because of return to employment. Complicating the picture of access to health care for the poor is that Medicaid itself devotes increasing fractions of its resources to reimburse elderly people for long-term nursing home care. Of all Medicaid outlays in 1988, for example, 37 percent were for nursing home care, and these ac- counted for more than 44 percent of all expenditures on such care that year (Office of National Cost Estimates, 1990~. In 1990, as a fraction of the total spending on personal health care services in Medicaid, the outlays for nursing home care were just over 33 percent, or about 92 percent of the combined Medicare-Medicaid spending for this type of service (Office of National Health Statistics, 1991~. Still, today's most visible access problem lies with the 35 million or so people who lack any (or at least adequate) health insurance: 40 million in
DATA NEEDS FOR A CHANGING HEALTH CARE SYSTEM 25 1996 by one estimate (Aaron, 1991a).2 This group tends to be dispropor- tionately poor. Of the approximately 32 million uninsured in 1987, for example, 30 percent came from families with incomes between 0 and 100 percent of the federal poverty level (and 32 percent from families between 101 and 200 percent), compared with 21 percent from families with in- comes greater than 300 percent of poverty (U.S. Bipartisan Commission on Comprehensive Health Care, 1990~. As a percentage of all people in vari- ous income groups, 32 percent of those with annual family incomes be- tween O and 100 percent of poverty were uninsured, compared with 6 per- cent of those with incomes greater than 300 percent of poverty (U.S. Bipartisan Commission on Comprehensive Health Care, 1990~. In short, about 15 percent of the population faces severe constraints on their ability to receive appropriate and timely care of the sorts outlined above. Although eventual- ly they may be able to acquire services for acute or chronic illnesses, those services may be sporadic, fragmented across providers, unnecessarily de- layed, and of questionable quality. The issue is especially serious for children, particularly those in unin- sured families not sufficiently poor to be eligible for Medicaid. That is, the uninsured are disproportionately young; more than a quarter are children under 18. More concretely, about 83 percent of all children through age 17 were covered by some health insurance plan or Medicaid in 1988, but only 72 percent of children in families with annual incomes under $10,000 were so covered, compared with 92 percent of children in families with incomes over $40,000 (Bloom, 19901. Lack of or inadequate insurance coverage and attendant high or insur- mountable out-of-pocket costsare not the only hindrances to access to arid appropriate use of care. Various other problems associated with residence (e.g., rural areas, inner cities) can be cited (although they differ by site). For example, some residents of rural areas not adjacent to metropolitan areas may face long-distance travel to sites of care, small or inadequately staffed and equipped facilities, and a dearth of physicians (especially spe- cialists).3 People in the inner cities may have to cope with overcrowded 2Estimates of the number of uninsured people in this nation vary considerably, depending partly on definitions and partly on time frames. The Pepper Commission cites about 31.5 million (15 percent of the nonelderly) who had no health insurance in 1987; Ries (1991) cites about 33.9 million (14 percent) in 1989. Short (1990) puts the number of people lacking coverage for all or part of 1987 at 48 million (about 20 percent of the population); more timebound estimates suggest that 36 million people were completely uninsured in the first quarter of 1987, 34 million in last quarter. 3The issue of whether rural areas are underserved by physicians, or by physicians in special- ty practice compared with primary practice, is a tangled one. It is discussed in more detail in the section on human resources below.
26 TOWARD A NATIONAL HEALTH CARE SURVEY facilities and clinics, fraudulent "Medicaid mills," and the lack of physi- cians able to communicate in languages other than English. In addition to these problems, access to care may be further limited for specific population groups. These include ethnic and racial minorities, non- English-speakers, people with various impairments and disabilities (physi- cal handicaps, blindness, deafness), people with various stigmatizing condi- tions, such as the acquired immune deficiency syndrome (AIDS), and homeless people. Finally, certain other features of the current health care landscape im- pinge on people's access to care. Very critical may be the current medical malpractice and liability climate, in which fear of being sued or the need to pay high malpractice premiums, or both, drives physicians out of certain hi~h-risk specialties. This is especially true for obstetrics or at least the delivery of obstetric care (IOM, 1989b). Low rates of reimbursement for physicians' services are also declared to be a barrier to access. This is a significant issue at least for Medicaid, where low fees are one tactic for (presumably) controlling expenditures, insofar as physicians will refuse to take any (or at least new) Medicaid patients. Some observers are also concerned that physician payment reform in Medicare (national fee schedules, control on volume of services through volume performance standards, and limits on balance billing,) will, first, drive some payments for services below those typical even for Medicaid and, second, negatively affect access for the elderly to key health services (AARP, 1991b). Quality of Care A recent Institute of Medicine report has defined quality of care as "the degree to which health services for individuals and populations increase the likelihood of desired health outcomes and are consistent with current pro- fessional knowledge" (IOM, 1990c: 21, emphasis added). The report char- acterized problems with quality of care as stemming from three sources: overuse of unnecessary and inappropriate care; underuse of needed and effective care; and poor technical and interpersonal performance on the part of health care providers. The emphasis on health services, individuals and populations, and outcomes plus the breadth of questions that flow from overuse, underuse, and poor care highlight the extraordinary set of topics that standard surveys must consider in measuring and monitoring quality of care. Attention to quality of care and to the evaluation and improvement of health care has waxed and waned over the years, often in response to pro- grams intended to deal with expenditures on and access to care. There has been some resurgence in interest in quality of care, quality assurance, and
DATA NEEDS FOR A CHANGING HEALTH CARE SYSTEM 27 continuous quality improvement in recent years (Health Affairs, 1988; In- quiry, 1988; Lohr et al., 1988; Goldfield and Nash, 1989; Palmer et al., 1991~. Strenuous efforts to stem the use of health care services and control expenditures prompt various interest groups (e.g., health care providers, patients) to predict harmful consequences for the quality of health care. A case in point is the predictions of harm to quality of care for Medicare beneficiaries secondary to the implementation of a DRG-based prospective payment system predictions not borne out to date (Kahn et al., 1990a, 1990b). Nevertheless, many participants in health care delivery, as well as many patients, express concern about the quality of the care they deliver or receive now or will in the future. Quality of care is often seen as having three dimensions: (1) the so- called structural characteristics of providers and practitioners (various orga- nizational and professional factors); (2) the process of care (or what is done to and for patients with respect to diagnosis, treatment, rehabilitation, and palliation); and (3) the outcomes of care. These three dimensions are im- portant in traditional approaches to quality assessment and assurance (Donabedian, 1966, 1980, 1982, 1985) as well as in newer approaches involving continu- ous quality improvement and total quality management (Batalden and Buchanan, 1989; Berwick, 1989; Berwick et al., 19901. The last-named dimension- variously termed outcomes, health status, health-related quality of life, pa- tient well-being, and the like has gained considerable attention and utility in recent years (Brook et al., 1976; Gilford, 1988; Ellwood, 1988; Lohr, 1988; Tarlov et al., 1989~. Most experts in this area agree that understanding the relationship be- tween the process of care and outcomes of that care is essential, and they lament the striking lack of information that might demonstrate those linkag- es. Generally, the process of carethat is, the services renderedhas been and is easier to define, document, and evaluate than are outcomes of care. As a consequence, existing data systems do not include adequate indicators of outcomes that could be used to evaluate quality. What is needed to measure this dimension of quality of care i.e., the outcomes of care, or the individual's health status adequately? Health status is a complex, multidimensional construct reflecting significant as- pects of an individual's or a population's life circumstances. It usually incorporates five major areas of interest physical health, mental health, social functioning, role functioning, and general health perceptions. Some experts add pain as a key domain; some add cognitive functioning, or ener- gy and vitality, or both. The literature may differ somewhat in the typology of these (or related) domains, but the basic classifications are becoming relatively standard (Katz, 1987; Ware, 1987; Lohr and Ware, 1987; Patrick and Erickson, 1988; Lohr, 1989; Mosteller and Falotico-Taylor, 1989; Patrick and Bergner, 1990~.
28 TOWARD A NATIONAL HEALTH CARE SURVEY A partial list of variables pertinent to domains of health-related quality of life (which some regard as synonymous with health status and others take to be broader) includes: survival and life expectancy; various symptoms, such as pain; numerous physiologic states, such as blood pressure or serum glucose levels; physical function states of many sorts, for instance, mobility and ambulation, sensory functioning (such as seeing or hearing), sexual func- tioning, a range of capacities relating to impairment, disability, and handicap, and specific measures of activities of daily living; emotional and cognitive function status, such as anxiety and depression or positive well-being; per- ceptions about present and future health; and satisfaction with health care. Aspects of quality of care other than outcomes must also be considered. One is the continuity of care, which implies that one must evaluate the flow of illness and wellness, and the services sought and rendered, through an entire episode of care (rather than in a single encounter) and thus across time and settings of care. A second, related factor is coordination of care. Yet another is the satisfaction of both the recipient and the provider with the care rendered, a facet of health care that some place directly in any listing of outcomes; satisfaction can in turn relate to a broad set of ques- tions concerning accessibility, availability, and costs of care in addition to ratings of the technical and interpersonal performance of clinicians and institutions. Adequately addressing this inventory of information in data systems of the future poses an immense challenge to policy makers and researchers alike. HEALTH CARE REFORM The calls for health care reform and the proposals emanating from sev- eral blue-ribbon panels in recent years have centered on finding ways to solve particular problems of access to care (especially those of the unin- sured and underinsured). The aim is to overcome the access barriers with- out exacerbating the problems of costs or undermining the quality of care received. A peculiarity of the problem of being uninsured in the United States is that most people in this situation are in families with at least one employed worker a seeming anomaly, insofar as the great bulk of health insurance in this country is employer-based. Perhaps not surprisingly, therefore, the main ideas being advanced generally call for the enactment of various kinds of mandrels for employers to offer at least basic insurance plans to their employe-~-; or to contribute to a public-sector pool that will cover such individuals (ideas generally referred to as "play or pay"~. Among the better known proposals, for instance, are those emanating from the Pepper Com- mission. These would, among other things, require all individuals to obtain health care coverage from their employers or from a public program, the
DATA NEEDS FOR A CHANGING HEALTH CARE SYSTEM 29 latter a new version of Medicaid to cover the unemployed and the poor (U.S. Bipartisan Commission on Comprehensive Health Care, 1990~. These kinds of proposals including ones being advanced by the Amer- ican Federation of Labor and Congress of Industrial Orgar~izatior~s, by the American Medical Association, by various health policy and health eco- nomics experts, by business coalitions, and politiciansare relatively mid- dle-of-the-road, politically speaking. As noted, they would either require compulsory private insurance through employers or direct that employers either provide private insurance to employees or pay an equivalent tax, with government insuring nonworkers and the poor (Blendon and Edwards, 1991; Haglund, 1991; JAMA, 1991~. More radical or reform-minded proposals appear to favor, on one hand, a Canadian-style national health program (MarTnor and Mashaw, 1990; Woolhandler and Himmelstein, 1989, 1991; Himmelstein and Woolhandler, 1989) and, on the other hand, competition- oriented approaches or programs to require individuals to acquire their own health insurance and receive tax credits for doing so (Butler, 1991~. Debates over the attempt to ration Medicaid care in Oregon (the Oregon Basic Health Services Act) (Fox and Leichter, 1991; Grannemann, 1991; Sipes-Metzler, 1991) underscore the immense diversity of proposals now being aimed at health care reform and the appreciable emotional overlay that accompanies them (Brown, 1991; Callahan, 1991; Etzioni, 1991~. Widening interest in what can be learned for the United States from other countries may help to break what some perceive to be a health care policy gridlock (Health Affairs, 1991; Peet, 1991; Reinhardt, 1991; Starfield, 1991~. The directions that health care reform is likely to take, as reflected in these kinds of ideas, are not at all clear. The implications for data survey and collection efforts are also murky. Those designing and implementing such survey programs will surely need to give strict attention to (1) key elements of health care financing and reimbursement proposals or legislated strategies for reform; (2) measures of expenditures, access, outcomes, and quality of care; (3) providers of health care, broadly defined; and (4) the populations affected by such strategies, particularly children, ethnic minori- ties, and the elderly (and the very old). Cutting across these points is the requirement that planning and implementation of reform require respect- able, comprehensive data that still permit a focus on special groups. OTHER FACTORS INFLUENCING HEALTH CARE POLICY AND DATA NEEDS Demographics and Socioeconomic Factors The evolving nature of U.S. society poses notable challenges to health care delivery. Most significant, perhaps, is what is called the "graying of
30 TOWARD A NATIONAL HEALTH CARE SURVEY America": the population is becoming older, on average, and the elderly population itself is aging. Data from the mid-1980s suggests, for instance, an increase in the population age 65 and older of 75 percent between 1960 and 1986, compared with a 30-percent increase among those under age 65 (NCHS, 1989~. Between 1987 and 2030, those under age 20 are projected to drop as a percentage of the total population from 29 to 24 percent and those age 20 to 64 to drop from 59 to 56 percent; by contrast, those age 65 and older will increase from 12 to 21 percent (Waldo et al., 1989~. The oldest elderly (those age 85 and older) typically the sickestconstitute the fastest-growing portion of the elderly population (Gilford, 1988~. Other factors and trends are significant as well. With respect to the elderly, the numerical predominance of women over men is striking. Fur- thermore, the vast majority of the elderly (and hence elderly women) live in the community, and of those about one-third live alone. Of those residing in nursing homes, most are very old (age 85 and older) and female. Poverty rates among the elderly have been declining in recent years, but even so about one in three elderly people was below 150 percent of the poverty line in the mid-1980s (Special Committee on Aging, 1987-1988~. The proportion of the total population accounted for by immigrants and nonimmigrant minorities is growing. Toward the end of the 1980s, the racial composition of the U.S. population was about 84 percent white, 12 percent black, 8 percent Hispanic (of any race), 3 percent Asian and Pacific Islander, and 1 percent American Indian and Alaskan native (NCHS, 19911. The Hispanic and Asian populations have both grown considerably since 1980 (in both absolute and percentage terms), the former about equally from immigration and natural increase and the latter almost entirely from immigration (NCHS, 1991). The contrasts between minority populations, particularly blacks and Hispanic groups, and the majority (white) population with respect to socioeconomic and health factors are dramatic (U.S. DHHS, 1991a). Virtually every health status and health utilization measure shows those subgroups as disadvan- taged relative to whites; in some cases (e.g., life expectancy and infant mortality for blacks), the disparities are widening. The problems are so acute that the DHHS secretary has made minority health issues a top priori- ty for the 1990s. Apart from factors relating to age and ethnicity, the health care sector may be affected by trends in family income and makeup. One consideration is the growth in the number of families in poverty. The rise in the number of children in poverty is especially stark (Johnson et al., 1991~: between 1979 and 1989, child poverty rates increased by 21 percent; one in five children currently lives in poverty; children have almost twice the likeli- hood of being poor as any other age group; and the younger the child, the greater the chance of being poor.
DATA NEEDS FOR A CHANGING HEALTH CARE SYSTEM 31 Likewise, both the lack of formation of and the breakup of stable fami- lies through divorce, separation, and out-of-wedlock births, for instance contribute to the growth of single-parent families and families in which the single parent may herself be barely out of childhood. Homeless people, an extremely heterogeneous group that can range from functioning families tem- porarily down on their luck to mentally ill individuals who have been dis- charged from institutions to the community, is another population for which the current health care system has little to offer (IOM, 1988; Bruckner et al., 1990~. Various types of nontraditional households, such as homosexual couples for whom obtaining conventional health insurance may be a prob- lem, are yet another element of society that poses challenges to health care delivery and policy. (The predicament is exacerbated by the loss of private health insurance among homosexual men with AIDS, although this may be more a consequence of loss of employment owing to sickness than to exclu- sionary insurance practices per se Kass et al., 19911. Nevertheless, the difficulties posed to survey and surveillance systems of the 1990s and be- yond by these extremely differing views of '~fam~lies"4 are profound. Health and Disease The causes of death, illness, and disability in this country, as in all industrialized countries and many industrializing nations, have changed sig- nificantly in the twentieth century. In broad terms across the U.S. popula- tion, a number of infectious diseases have come under control in the past 50 years or can now be cured. Because these were often diseases of the young, life expectancy has risen. Chronic illnesses have taken the place of infectious disease as the lead- ing causes of death and disability. They are increasingly prevalent because of the aging of the population and increasingly complex as greater numbers of people develop multiple chronic problems. Furthermore, the numbers of people suffering various disabilities~wing to genetic disorders, chronic 4A recent compilation of definitions offamily, for documenting how nontraditional families may be recognized for purposes of bereavement and sick leave, extension of health benefits, and similar matters underscores the variability in how this term is understood and used (legal- ly, socially, and emotionally). One definition from California is "a unit of intimate transacting and interdependent persons who share the same values and goals, responsibility for decisions and resources, and a commitment to one another over time"; another has it that a family is "a unit of interdependent and interactin, persons, related together over time by strong social and emotional bonds and/or by ties of marriage, birth, and adoption, whose central purpose is to create, maintain, and promote the social, mental, physical and emotional development and well-being of each of its members." Similarly, a definition from New York holds that "a more realistic . . . view [is that] a family includes two adult lifetime partners whose relationship is long-term and characterized by an emotional and financial commitment and interdependence" (all cited in National Gay and Lesbian Task Force Policy Institute, 1990-1991).
32 TOWARD A NATIONAL HEALTH CARE SURVEY illnesses and the aging process, improved treatment of previously life-threatening wounds and injuries suffered in wartime, and various types of other injuries such as automobile accidents are also growing. The conditions them- selves are more complex than the infectious diseases they have replaced, and people live longer with them, often with a poor quality of life (IOM, l991d). These and related developments all pose likely increasing demands for home health, rehabilitative, and related types of care and for better data on incidence, prevalence, measures of severity and comorbidity, im- pairment and disability, and quality of life (Stoto, 19921. Superimposed on this basic pattern of the incidence and prevalence of acute and chronic conditions is a set of sociomedical conditions that have had or threaten to have great impact on the need for health care. The AIDS epidemic has perhaps been perceived as the most menacing problem in recent years, for several reasons its essentially 100-percent fatality rate, the rapid growth in the numbers and diversity of infected persons, the real and perceived threat to the nation's blood supply, and the biological diver- sity of the human immunosuppressive virus (HIV) itself. Other significant health problems stem from the of use and abuse of illegal drugs, alcohol, and tobacco (U.S. DHlIS, 1991a). Among these are (1) the use of crack cocaine and sequelae such as crack-addicted and devel- opmentally affected babies and children of school age and (2) disability and premature death from automobile accidents involving alcohol abuse. The contribution of smoking to disease and premature death is well established; rates of smoking among certain populations (teenagers, women, and minor- ities) have not come down even in the face of that evidence. Interpersonal violence in the form of homicide and child and spouse abuse arise partly from some of the causes noted above and partly from the generally increas- ing levels of stress in U.S. society secondary to racial, economic, and other tensions. Finally, intrapersonal violence, that is, suicide and attempted suicide, is also on the rise, at least among teenagers and the elderly (Rosen- berg et al., 1987; Meehan et al., 1991~. Healthy People 2000, the 1990 report of the secretary of DHHS outlin- ing national health promotion and disease prevention objectives, is a partial response to these changing mortality and morbidity patterns and trends (U.S. DHHS, 1991a). It proffers three national health goals: increase the span of healthy life, lower health disparities among Americans, and provide access to preventive services for all Americans. These translate into perhaps 400 different objectives, and thus statistical series that will need to be moni- tored (Stoto, 1992) an imposing task indeed. Changing Values and Expectations People's attitudes toward health and health care services have altered considerably in past years. Patients now expect to have a greater say in the
DATA NEEDS FOR A CHANGING HEALTH CARE SYSTEM 33 decisions about how they shall be cared for. That is, they expect to be informed about all (reasonable) options open to them and about the expect- ed benefits and risks of those options, so that they may express and exercise their own preferences for or against certain health states or the sequelae of different health care services. These choices may be about care for essentially symptomatic (but not life-threatening) problems, such as the alternatives of surgery and "watchful waiting" for benign prostatic hypertrophy (Wennberg et al., 1987; Barry et al., 1988; Fowler et al., 1988~. They may be about care for possibly life- threatening diseases, such as the alternatives of lumpectomy (with or with- out chemotherapy, hormonal therapy, or radiotherapy) or mastectomy for breast cancer (IOM, 1990a). Or they may be about care at the end of life, when the decision may be not to start, or not to continue, various extraordi- nary technologies that might prolong life at the significant expense of the quality of that life. Regardless of the situation, however, patients more and more can be expected to want to exercise some autonomy about what is, or is not, done for them. For policy makers to be able to make sense of utilization (and hence expenditures) data as well as outcomes and quality-of-care data, those responsible for amassing that information will need to be increasingly sen- sitive to these issues. They will also need to design their data collection strategies to take patient preferences and autonomous actions stemming from those preferences adequately into account. Technology, Innovation, and Biomedical Advance One public policy goal immediately following World War II was to increase investment in biomedical research. Rapid increases in funding expanded substantially the scope and size of the National Institutes of Health and related funding of biomedical and clinical investigators. This funding and the simultaneous federal investment in the education of scientists re- sulted in tremendous advances in fundamental biology (molecular biology, _ neurobiology, genetics) and in the application of basic science to the pre- vention, diagnosis, and treatment of disease. Although some have criticized the emphasis on the development of halfway technologies such as dialysis and transplantation (Thomas, 1972), there is little doubt that these technolo- gies have radically changed medical care (Altman and Blendon, 1979; Ban- ta et al., 1981; Gelijus, 1991), as well as dramatically influenced the health and biomedical policy scene (IOM, l991c). Moreover, issues concerning the conduct of and support for research and development of new technolo- gies (the early innovation stages, for example) are being more carefully studied and debated (IOM, 1990-1992~; more broadly, the field of technolo- gy assessment has received much attention in the last decade or two (OTA, 1982; IOM, 1985~.
34 TOWARD A I\lATIONAL HEALTH CARE SURVEY Developments during the past four decades cover virtually every seg- ment of health care that is, prevention, diagnosis, and treatment- and vir- tually every aspect of health and illness. Among the more dramatic achievements have been those related to life-threatening or fatal illness:5 open heart sur- gery and bypass grafting; cardiac assist devices, including total artificial hearts; heart, lung, kidney, and liver transplants, and a vast array of phar- maceutical agents for medical and mental illnesses. Less visible but per- haps no less dramatic have been diagnostic steps: automated laboratories; computer-aided tomography (CAT) scans, magnetic resonance imaging (MRI), and positron emission tomography. In the prevention area, vaccines against a number of diseases (for example, hepatitis B and poliomyelitis) are bring- ing these diseases under greater control. Many impressive technologies come together in units such as those devoted to coronary care and neonatal intensive care. Finally, major breakthroughs can be expected in molecular biology, genetics, and bioengineering the prospects of mapping and se- quencing the human genome, for example, are monumental and daunting (CLS/NRC, 1991). Initially, because of the cost, the reimbursement system, and the inten- sity of treatment, the hospital was the focus of placement for most of the new technology, and many of the nation's hospitals sought to acquire in as timely as way as possible the relevant equipment and professional staffs. This pattern is now changing in two ways. First, in the last two decades, automation, advances in anesthesia and surgical techniques and other tech- nologies, and reimbursement and cost containment policies have stimulated the movement of certain types of technology out of the hospital setting. New facilities have emerged and grown to address this health care market. Second, assuming the evidence continues to lend credence to information that certain technologies (especially complex surgical procedures) require high volume and specific skills to reach acceptable levels of mortality, morbidity, and quality of care (Luft et al., 1987; OTA, 1988), greater re- gionalization of these technologies (typically in hospital settings) may well occur. Reshaping the hospital industry and developing vertical systems of care will also accelerate regionalization. Many observers expect that the availability, coupled with the costs, of these existing and anticipated technologies will outstrip society's ability to provide their benefits to everyone. As noted in the discussions above con- cerning costs and access, demands are increasing to develop more rational approaches to distributing the benefits of technology. Apart from the many proposals for health care reform that have emerged in 1990-1991, other key steps include the expansion of outcomes and effectiveness research (Brook 5Two recent IOM studies have explored many issues relating to two expensive, life-saving technologies. One concerns end-stage renal disease (IOM, 1991a; Levinsky and Rettig, 1991), the other the total artificial heart (IOM, l991b).
DATA NEEDS FOR A CHANGING HEALTH CARE SYSTEM 35 and Lohr, 1985; Ellwood, 1988; NCHSR, 1988; Roper et al., 1988; IOM, 1990d; Wennberg, 1990b). Development of clinical practice guidelines, for instance by the Agency for Health Care Policy and Research in the U.S. Public Health Service, by medical specialty societies, and by other groups is another indicator of public, payer, professional, and policy maker re- sponse to concerns about overuse or misuse of services, spiraling health care spending, and quality of care (Chassin, 1988; Brook, 1989, 1990, 1991; USPSTF, 1989; AMA, 1990a, l990b; Audet et al., 1990; Fletcher and Fletcher, 1990; IMCARE, 1990; IOM, 1990b, 1992; Leape, 1990; Eddy, 1991, forth- coming; GAO, 19911. Interest has grown in the last two decades regarding the distribution, cost, and use of the accelerating number of new as well as established technologies. Some data exist on these matters. For example, the Ameri- can Hospital Association surveys in-hospital availability of certain devices such as CAT scans, MRIs, and lithotripsy; manufacturers have proprietary information on sales of drugs and devices; and information on procedures can be found in insurance claims systems. Furthermore, special studies and surveys (like those now conducted by and being planned for NCHS) can provide at least some cross-sectional information. The fact remains, however, that there is a paucity of information on the quantity and distribution of specific technologies according to settings, types of practitioners, and similar variables, as well as on their use according to demographic and health characteristics of patients; no comprehensive na- tional data are collected on technology use. For example, as many types of technologies shift from the hospital sector to the outpatient sector, informa- tion from insurance claims is less and less helpful in tracking their use, because of the less comprehensive and less reliable coding of inflation on ambulatory care claimswith the possible exception of coding for visits and procedures, when it is done with CPT-4 (Current Procedural Terminol- ogy) codes rather than ICD-9-CM (International Classification of Diseases, ninth version, clinical modification) codes. As another illustration, although the Medicare Part A and Part B files collectively provide the best single source of information on use of health care (in this case, of course, just for the elderly), they provide no information on outpatient use of prescribed medications- a source of considerable concern about quality of care and out-of-pocket spending for the elderly (for example, 16 percent of annual out-of-pocket spending between $1,000 and $2,000 is for prescription drugs- (AARP, 199 la). Data that link diagnoses with use of technology are not generally avail- able except from selected clinical data banks. Commonality of procedures coding is essential to permit full utilization of different data banks on the use of technology. Outcome and quality assessment require data on diagno- sis, use of drugs, devices, and procedures.
36 TOWARD A NATIONAL HEALTH CARE SURVEY Human Resources Certain trends in the emergence and use of different types of health professionals have significant implications for data sources. Among the more traditional health professionsfor example, medicine and nursing a move away from general practice is clear, although if the distinction is made between primary care and specialty practice, the trend is not so dis- tinct. For example, in 1970, about 15 percent of all physicians were in general and family practice (27 percent of those in office-based practice); in 1987, the figure was 9 percent (16 percent for office-based practice) (NCHS, 1991~. However, categorizing all nonfederal allopathic physicians as pri- mary care or specialty care physicians, with the former being all general and family practice, internal medicine, obstetrics/gynecology, and pediat- rics, changes the picture somewhat. In 1975-1976, 45 percent of this group . . ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ r. was in primary care, 5u percent in specialty care; In 196-/-lYbb, the figures were 44 percent and 51 percent (Frenzen, 1991) hardly a meaningful shift over time. One lingering debate is the extent to which physicians (both primary care and specialists) have diffused into nonmetropolitan or rural areas. Evi- dence amassed by researchers at the RAND Corporation in the early 1980s suggested, for instance, that most rural residents were within a half-hour drive of most types of specialists (Williams et al., 1983) and that even in physician-poor states, half the towns of 30,000 to 50,000 residents had neurosurgeons (Newhouse et al., 1982; J. Newhouse, personal communica- tion, 1991~. The expectation was that the diffusion of specialists into rural areas would increase as the total physician supply grew. Other investiga- tors (cited in Frenzen, 1991) have presented different data and arrived at different conclusions or predictions~hiefly that few physicians would (or did) move into the most rural areas of the country. Examining physician manpower data for 1976 to 1987 for 10 categories of urban-rural locations, Frenzen (1991) concludes that the supply of physicians increased every- where in the country, but that the increase was more rapid in metropolitan counties and nonmetropolitan urbanized areas not adjacent to metropolitan statistical areas, leading to a widening over time between urban and rural areas in the availability of physicians. (The exception to this comment concerns osteopaths, who were relatively more likely to settle in rural areas than allopathic physicians, especially in those states traditionally hospitable to that profession). Superimposed on these patterns is the growth in the numbers of, or demand for the services of, various types of health professionals, such as dentists, psychologists, speech therapists, physical therapists, nurse practi- tioners, nurse midwives, and physician assistants. Within this picture, how- ever, is a marked shortage (perceived or real) of certain kinds of personnel,
DATA NEEDS FOR A CHANGING HEALTH CARE SYSTEM 37 particularly nurses and technicians (IOM, 1989c; NCHS, 1991~. Aiken and Mullinex (1987) and the Commission on Nursing (of the secretary of the DEPTHS) conclude that cyclical imbalances between the supply of and de- mand for nurses result from excess demand, not an insufficient supply of nurses (L. Aiken, personal communication, 1991~. Notwithstanding this evidence, many people in the health care sector, as well as much of the public, believe that a nursing shortage exists (Harris-Wehling, 1990; Walk- er, 1990~. The emergence of new types of providers- nurse practitioners, nurse midwives, and physician assistants, for example is another notable trend. These types of practitioners present a cost-effective alternative to physi- cians, with no discernable decrement in quality of care (OTA, 1986~. As public policy shifts to permit the public to choose such providers, and as reimbursement to such health professionals from third-party payers expands, demand for and use of their services may well grow as well. USERS OF HEALTH CARE DATA The panel determined early in its deliberations that it needed to learn more about (1) the kinds of health care data that actual users of the National Center for Health Statistics provider surveys desired now or might want in the future, (2) their satisfaction (or lack of it) with the inflation currently in those surveys, and (3) their views on the reliability, validity, costs, and accessibility of those surveys. To that end, the panel acquired information from users through presentations at panel meetings and focus group inter- views. The methods and findings of those data-gathering efforts are given in greater detail in Appendix B. This section briefly summarizes the re- sults. Data needs most frequently identified concerned, not surprisingly, the major health policy issues discussed earlier. Among the specific issues related to costs were expenditures for prescription drugs and long-term care. With respect to access, users were concerned about populations lacking insurance, the elderly (especially as regards long-term and home health care), children and adolescents, people in rural areas, and people with HIV infections. Other issues concerned the impact of changes in financing and reimbursement generally, the effects of the shift from inpatient hospital care to ambulatory settings, and the effects of prevention programs on health habits, lifestyles, and outcomes. The users interviewed raised other points as well. One topic concerned the need for person-based data, not just provider-based data, so that analy- ses can be done on episodes of care and across multiple providers and settings. A second involved the large number of different types of provid- ers that NCHS does not now survey. Third, some respondents criticized the
38 TOWARD A NATIONAL HEALTH CARE SURVEY high level of geographic aggregation of current survey data, indicating a desire for data for (or the ability to extrapolate to) smaller geographic re- gions (e.g., states, rural areas, and perhaps even counties or communities). A fourth issue involved specific subpopulations (minorities defined by eth- nicity and race as well as groups classified according to age, health status, socioeconomic level, and similar variables). A final matter of great concern to the respondents interviewed was the desire to be able to link the various NCHS data bases with each other as well as with Health Care Financing Administration (HCFA) data, social security records, and other databases. IMPLICATIONS FOR DATA SOURCES AND SYSTEMS NCHS surveys have been, for decades, models of how to conduct com- plex national surveys in the health care field. Vast numbers of epidemio- logic, health policy, and even effectiveness studies have relied, and must continue to rely, on information from the Vital and Health Statistics series and other NCHS publications. In recent years, agency staff have pioneered methods of applying composite health status measures to information such as from the National Health Interview Survey (Erickson et al., 1989~. Yet the political and economic constraints under which the agency was forced to operate in the 1980s, which perhaps have led to many of the difficulties cited below and in other chapters, should not be underestimated. Despite this admirable record, existing data sources that might be ex- pected to provide at least some of the information needed to monitor and evaluate changes in the availability, financing, and quality of health care are rapidly becoming outdated and less comprehensive than is desirable. Often they do not cover the universe of providers and sites of health care; neither do they cover the universe of patients or potential users of health care. They lack sufficient information on exactly what services are provid- ed and what the outcomes of those services are; they are inexact with re- spect to financial data (charges, costs, and expenditures); they are not time- ly; and at times they are inaccurate, incomplete, and unreliable. In short, current national data systems, when taken together, do not provide the in- formation needed to allow researchers and policy makers to assess the ef- fect of changes in financing, organization, and delivery of health care, or the impact of other social and economic trends, on the appropriateness, quality, costs, and outcomes of care. Among the directions in which database systems will need to move are the following: . Better insurance claims data from the fee-for-service system and analogs from the prepaid, capitated systems (and from hybrid sys- tems), especially among people younger than age 65;
DATA NEEDS FOR A CHANGING HEALTH CARE SYSTEM 39 · More information on clinical services and physiologic outcomes from medical records; . . More information from patients (and their surrogates or proxies) on health-related quality of life, perceptions of health status, and satis- faction with care; Better information on how much is spent for treating particular types of patients with particular types of illnesses (both direct aIld indirect costs, including out-of-pocket costs); and · More specific information on the services and technologies being given and used, and on the types of personnel providing them. A recent IOM meeting convened to begin planning activities in the area of database development for clinical evaluation also brought to the fore several significant "logjam" issues (Lohr and Durch, 19911.6 These are relevant for the planning of future NCHS surveys in general. Among the more salient were various technical aspects of data collection, data analy- ses, and data exchange; a myriad ot legal Issues (privacy and confidential~- ty, malpractice); use of data (e.g., clinical applications, feedback in quality- assurance programs, administrative and managerial applications); and the impact of reimbursement and financing on data collection and database development. No single survey, information source, or data collection mechanism system is likely ever to meet all these criteria, address all these technical problems, or meet all users' needs. Collectively, however, they should be flexible enough to adapt to health care system changes as they occur and to meet special needs on an ad hoc basis. They must be able to record shifts in the health care delivery system and related industries, to register changes in policies at the national level and perhaps subnational level, and to document the impact of these shifts and policy changes on providers and patients alike. More technically speaking, they also need to be flexible enough to 6Participants seemingly agreed that database is a singularly elastic tells,. It was (and is) used to refer to such dissimilar sources of information as the following (examples in parenthe- ses): computerized administrative files (Social Security and Internal Revenue Service files); insurance claims files in both the public sector (Medicare, Medicaid; Department of Veterans Affairs) and the private sector (the "big five" health insurance companies); provider records of all sorts (hospitals; managed care organizations and networks; pharmacies; laboratories); regis- tries (for cancer [tumors], trauma, bone marrow transplant, end-stage renal disease); re.;,ulatory systems (state systems for reporting hospital adverse events and incidents; credentialing agen- cies; the National Practitioner Data Bank for malpractice reporting); secondary research data- bases (the Rand Corporation's Health Insurance Experiment); surveys (of the National Center for Health Statistics; of provider associations); and vital and health statistics (e.g., birth and death certificate files).
40 TOWARD A NATIONAL HEALTH CARE SURVEY facilitate integration and linkage of data generated in special studies with those compiled in ongoing data systems. The panel concludes that, for the NCHS surveys to be able to meet future demands for information related to all the health policy and environ- mental issues summarized in this chapter, they must be developed in a coordinated manner. That is, the continued uncoordinated development of new, discrete data systems in NCHS and throughout the department with different definitions, documentation, coding protocols, and little ability to be linked for analysis would be a serious mistake. In the panel's judgment, the overriding focus of the integrated National Health Care Survey should be to provide statistics on a continuous basis that reflect the state of health care in the United States with respect to the key dimensions of costs, ac- cess, availability, quality, and effectiveness both now and through time. These and similar issues, especially as they relate to the National Health Care Survey, are taken up in succeeding chapters of this report.
