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.
4 MEASURING DISPARITIES IN ACCESS TO CARE Nicole Lurie Since the landmark Report of the Secretary's Task Force on Black and Minority Health, minority Americans have been consistently shown to have poorer health status and worse health outcomes than White Americans (DHHS, 1985~. These differences have remained so persistent that Healthy People 2010 specified the elimination of disparities in health as one of its two overarching goals (DHHS, 2002~. Included among the health disparities of most concern are those between different racial and ethnic groups and those associated with geography. There is now an emerging consensus that disparities, at least between different racial and ethnic groups, arise from several factors. These include differences in access to care and health insurance and in the amount and quality of care offered and received. They also appear to arise from factors not directly related to the health care system such as socioeconomic status, literacy, language, community factors affecting health, and differences in access to opportunity (Williams, 19991. The national goal of eliminating health disparities has led to a reexamination of the health care system and a call for intensive measures to move the nation well along toward meeting the goal (National Archives and Records Administration (NARA), ~ 9971. Learning from the quality movement, there has been heightened awareness that measurement is key to achieving success. Hence, in its reauthorization in 1999, the Agency for Healthcare Research and Quality (AHRQ) was directed to prepare a National Healthcare Disparities Report (NHDR), and it has formed an Office of Priority Populations (AHRQ, 1999~. These events coincide with a closely related effort: the development and publication of the National Healthcare Quality Report (TOM, 200 ~ c). An earlier report from the Institute of Medicine (TOM), Crossing the Quality Chasm, draws 99
1 00 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT attention to the intimate relationship between quality and disparities by its position that equity is one of the key aspects of quality (IOM, 200 Ib). In preparation for the NHDR, AHRQ has commissioned the TOM to create a committee to provide guidance on how to address various topics, including access to care. This paper is intended to provide background and fresh thinking for the Committee for Guidance in Designing a National Healthcare Disparities Report on the leading issues surrounding the measurement of disparities in access to care. 4-1. SETTING THE CONTEXT In a seminal article that provides a conceptual framework for access to care, Aday and Andersen identify a combination of policy variables such as the presence of health insurance, personal variables (predisposing, enabling, and need), and process of care elements (for example, use) that lead to an outcome (Aday and Andersen, 1974~. Since the article's publication over a quarter century ago, much has sadly remained the same such as the lack of health insurance and primary care. But much has also changed, and the challenge of measuring disparities in access to care in the early 21St century must account for these new realities. Several are highlighted here: . National demographics are shifting dramatically, and the U.S. is much more multi-ethnic than at any time in its history. Some states, such as California, are already "majority minority states," and the proportion of the population that is Hispanic is expected to grow dramatically in the next decade (U.S. Bureau of the Census, 1996~. Although African Americans as a group continue to have some of the worst health outcomes, discussion of health disparities for racial and ethnic minorities must move well beyond comparisons of Afiican Americans and Whites.
DISPARITIES IN ACCESS TO CARE 101 · Concurrently, there has been a shift of the population from rural to urban and suburban areas (U.S. Bureau of the Census, 2002a). Populations in rural areas have declined and aged as younger people have moved to more urban environments. The recent influx of immigrant populations, who often work on farms and in meat processing plants, has changed the ethnic composition of rural areas as well. . . The aging of the population and a longer life expectancy mean that chronic disease is now much more prevalent. Concomitantly, our understanding of how best to care for people with chronic disease is improving. Attention to end-of-life care has provided new opportunities for people to receive care at home, and the dying process has become more patient centered. At the other end of the age continuum, survival among severely impaired infants has increased, and these children often require continuous personalized medical care services. There have been major and continued shifts in the financing and organization of care. While most insured Americans are enrolled in some form of managed care, the heterogeneity among health plans means that it is harder to generalize about them. Nonetheless, some common elements have facilitated our ability to examine and address disparities. Explicit attention to responsibilities for enrollees (or identification of a denominator population) creates important opportunities to measure both access and quality within health plans. With this has come increased accountability, a key element of which is measuring and reporting. Managed care has also reintroduced the concept of population health. Once considered analogous to public health, it is now well recognized that caring for populations is effectively done both inside the personal health care system and outside of it. Managed care has also brought with it pressure to reduce health care expenditures and concerns that access and quality could be compromised. Uninsurance--and underinsurance--remain major problems for over 60 million Americans, and there is no foreseen decline on the horizon (Commonwealth Fund, 2000~. The amount that
1 02 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT individuals must pay for care is increasing, and half of all seniors report needing to cut back on essentials such as food and heat in order to pay for medications (Families USA, 2001~. The array of payment arrangements, deductibles, co-payments, and benefits has grown dramatically. Nonetheless, the uninsured, minorities, low- income populations, and rural residents disproportionately receive care in a separate, unmanaged system composed of community health centers, nonprofit clinics, various charity care arrangements, and emergency departments (IOM, 2000~. · Our understanding of what causes disease is becoming clearer. Specifically, it is now estimated that up to 50 percent of health status can be accounted for by health behaviors and only 15 to 20 percent by the health care delivery system (McGinnis et al., 2002~. In other words, health care system factors contribute proportionately less to health status when compared to other factors, although the benefits are greatest for those in poor health and without access to care. Importantly, advances in genetics have confirmed that race is not a biologic construct, but a social one. There is more genetic variation within racial groups than between them (President's Commission, 1983~. · Our understanding of health has broadened. Specifically, mental health and oral health have been more clearly defined as important components of health, and concepts like well-being or spirituality are increasingly thought to be integral to health. While the environment in which health services are provided has changed, the view of access to care has largely remained the same. 4-2. EVOLVING CONCEPTUAL FRAMEWORKS OF ACCESS TO CARE Our understanding of access to care has evolved over the last 25 years. The conceptual framework developed by Aday and Andersen identified the relationships among personal characteristics, policy variables, utilization, and outcomes. As the framework evolved, "potential access" became defined by characteristics of the delivery
DISPARITIES IN ACCESS TO CARE 103 system and of individuals in the area. These were further divided into the now familiar predisposing, enabling, and need variables, measured at both the individual and community levels. "Realized access" was represented by utilization. It was also represented by experiences with care and equity of access that required that the distribution of services be based on need (Andersen and Aday, 1978; Andersen et al., 1983~. The 1983 President's Commission for the Study of Ethical Problems in Medicine and Biomedicine and Behavioral Science Research further emphasized the issue of equity in access in declaring that "equitable access to health care requires that all citizens be able to secure an adequate level of care without excessive burdens"~ (President's Commission, 1983, p. 4~. While it did identify a standard of "equity," it did not define "adequate level of care" or how it might contribute to outcomes. The IOM revisited the issue of access to care in 1993, defining access as "the timely use of personal health services to achieve the best possible health outcomes"2 (IOM, 1993~. Of note, in choosing indicators that represented access-related outcome measures, it more explicitly linked access to quality and implied that everyone should have access to care to make these "best possible" outcomes achievable. The committee recommended a series of indicators that affect outcome measures through the processes of entering and staying in the health care system, the utilization of services, and quality of care. The report depicts a more linear relationship between access and outcomes starting with structural, financial, and personal barriers and moving through the use of services and "mediators" (which are largely measures of quality) to outcomes. It acknowledged the complexity of identifying access issues, but did not address the extent to which a poor performance on outcome measures is a direct function of poor quality of care or of barriers to access before or after an individual has entered the delivery system. The advent of managed care brought with it a set of additional factors that relate to access. This expanded framework is embodied in the work of Docteur et al., who identify a complex series of factors ~ Emphasis added. 2 Emphasis added.
1 04 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT related to access and quality in managed care (Docteur et al., 1996~. The authors include a number of contextual community variables such as available plan choices, active marketing and market characteristics, and stability of plan choices. They also recognize that access today may require navigating managed care plans, for example, to obtain referrals for care or to appeal denials. It is not clear how this challenge differs substantially from navigating the health care system as a whole. Finally, Envisioning the National Health Care Quality Report, issued by the TOM's Committee on the National Quality Report on Health Care Delivery, articulated a framework for consideration of quality, which is not unrelated to the frameworks described above (IOM, 200Ic). It takes the form of a matrix in which the rows describe consumer perspectives on health care needs (staying healthy, getting better, living with illness or disability, and coping with the end of life) and the columns describe components of health care quality (safety, effectiveness, patient centeredness, and timeliness). The relationships between access to care and this framework are important and are described in detail below. Neither the frameworks for describing access nor that for the National Healthcare Quality Report yet incorporate our knowledge of what creates health or evolution in the health care system. The Docteur et al. framework is something of an exception: it identifies community contextual variables (such as market competition) and available plan choices as access-related issues.3 However, as our understanding of factors that create health develops, it is clear that the nature of communities themselves and access to a well-functioning public health system are critical to achieving access to personal services. Increasingly, place appears to have an effect that is independent of socioeconomic status (Davey-Smith et al., 1998~. Consider, for example, factors that influence access to HIV care. A community's socioeconomic status and its physical and social environment affect the probability that one will become infected with HIV. Twenty-five percent of infected individuals in the U.S. do not even know they are infected and are thus completely unaware of the need for care (CDC, 3Andersen and Aday (Andersen and Aday, 1978) identify demographic characteristics of individuals in communities, such as the percent over age 65 or in poverty, as having potential impacts on access.
DISPARITIES rN ACCESS TO CARE 105 2002~. Late presentation of disease is a major determinant of bad outcomes in this disease, which disproportionately affects minority populations. While some of these individuals may have had missed opportunities for HIV detection in the personal health care system (a quality issue), the nature of the community and the failure of public health efforts to maximize knowledge of HIV serostatus are major barriers to access. Late stage presentation of other diseases, including certain cancers, diabetes, and heart disease, also provides examples of how access is tied to the effectiveness of the public health system. An example of the importance of a community environmental factor is oral health. It is well documented that low income and minority children have poorer oral health and less access to preventive or restorative dental services. Fluoridation is the single most effective intervention in preventing caries, but approximately one-third of the U.S. population does not have access to fluoridated water, particularly in rural areas (National Institute of Dental and Craniofacial Research, 2000~. Risk behaviors are irrefutable contributors to health outcomes. and many of these behaviors are modifiable. Changing health behavior is a joint function of the individual, the public health system (largely through increasing awareness and education), and the delivery system. Yet, the current conceptualization of access to care with regard to behavior change places the responsibility either solely with the patient or inside the health care system, often within the patient-provider encounter. A recent review of a campaign to promote smoking cessation during pregnancy found that the campaign did not focus on minority populations, which have the highest rates of smoking and SIDS (Lewis, 2001~. Women who are unaware of the risks of smoking during pregnancy are less likely to seek or use smoking cessation services. Finally, it is recognized that for most chronic disease, the major caregiver is the patient and/or family. Most care (for example, medication adherence, adherence to risk reduction behaviors, and self monitoring) actually occurs in the home, where the level that patients receive is largely influenced by their communities.
1 06 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT Hence, it is clear that the community, the public health system, the personal delivery system, and the individual all share responsibility in working to achieve optimal health outcomes. Figure 4-1 combines the frameworks discussed above to depict this relationship. FIGURE 4-1 Relationship between Population and Personal Delivery Systems Population Delivery System Community Characteristics Service Availability Community Demographics Normative Health Behaviors Physical and Social Environment Opportunity Structure (employment, education) Community Cultural Beliefs Market Characteristics (plan choice, structures) Sickness-creatin~ Environments Genetics_Nied *Includes entering and staying in system (use of services) Access to Care Structural System: Availability and Composition of Workforce Transportation Organization of Care System (health plan) Personal: Sociodemographic Characteristics Race Culture Language Attitudes Health Status Experience/Ability to Navigate System Need F. · ~ 1nancmg Issues Insurance Status Provider Reimbursement Paths through Care* Personal Delivery System Quality ~ e hi ~> Staying Healthy Getting Better . . Llvmg with Illness or Disability Coping oftLhifend _ v, cat .s _ .s _ ~ _ _ Outcomes Death Health Status Satisfaction Cost Effects on Individuals, Families, and Communities
DISPARITIES IN ACCESS TO CARE 4-3. ACCESS TO CARE AND THE QUALITY FRAMEWORK 107 The framework presented in Envisioning the National Health Care Quality Report highlights the fact that access to care can be a significant issue, even after someone has entered the health care delivery system (IOM, 2001c). Clearly, access throughout the lifespan to the full array of services identified in the rows of the framework is an important policy goal. Key aspects of access are related to each of the domains of safety, timeliness, effectiveness, and patient centeredness. Table 4-1 and Figure 4-2 identify examples of access indicators that can easily be placed in different cells of the quality framework. The following is a discussion of the relationship of access to the framework's four components of health care quality (timeliness, safety, effectiveness, and patient centeredness) and to the four consumer perspectives on health care needs (staying healthy, getting better, living with illness or disability, and coping with the end of life). Timeliness is unquestionably a component relevant to both for entering and staying in the system. With regard to safety, several studies have suggested that poor and minority patients are more likely to experience medical errors and are more likely to have cardiac procedures perfor~ned by providers with poorer outcomes. Whether this results from patients' uninformed choice of providers, the lack of availability of other providers to care for these individuals, or other factors is unclear. However, access to safe care is critical for optimal outcomes. Access to care that is not safe may actually be worse than no care.
'e · - c) ~ ~ - ~' c) - o o o it, ~ o ~ o Pro · - ~ ~ ~ ~ _4 r~- 1 ~ ~11' o Q~4 ~ i) o m ~0 i- q 00 ._ ~ o - - o Ct U. .~ ~ ~ Ct ¢ ~ ~0 Ct _ ~ Cot ~ :- %¢ o o · Ct _ Ct ~ ~ - o ¢ ~ at, · ~ - ° C - t ~ O Ct ~ - 0 En · - ~ ~ m
1 1 0 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT With regard to effectiveness, we are not ternbly concerned with ensuring access to care that is ineffective provided it is not harmful. However, data suggest that minority patients receive fewer effective, need-based treatments that improve health and quality of life such as knee replacements for severe arthritis or renal transplants for end stage renal disease. As is often the case, it may not be possible to disentangle access and quality because so many factors are involved. For example, insurance status is clearly related to arthritis care in general and to knee replacement rates in specific, which could lead one to conclude that knee replacement rates reflect financial access to care. However, barriers such as a lack of physician recommendation to undergo such a procedure or a poor explanation of patient options are quaTity-related access barriers to receipt of such a procedure. Even when the issue of insurance is removed, as for example in Medicare's program for end stage renal disease (ESRD), factors such as a lack of discussion or recommendations about options for treating renal failure appear to account for a substantial part of the difference in transplant rates between Whites and African Americans (Ayanian et al., 1999~. While an argument can be made that this is purely a quality of care issue, one cannot access treatments of which one is unaware. A comprehensive review of this issue can be found in the TOM study, Unequal Treatment (IOM, 2002b). That report suggests that provider-related factors such as bias, discrimination, and stereotyping- -as well as patient-related factors such as mistrust and variability in presentation of symptoms--explain much of the phenomenon of disparate access to effective care. While there are not yet good measures of many of these factors to more directly test this hypothesis, development and use of such measures should be a goal of subsequent editions of the NHDR. Regardless, it is safe to conclude that even once inside the health care system, there are numerous examples of disparate access to effective care. Access shares key relationships with another aspect of quality, patient centeredness, which includes such issues as workforce demographics and cultural competence. The degree to which the demographics of the health care workforce reflect the population cared
DISPARITIES IN ACCESS TO CARE 111 for is a significant access issue. First, minority physicians are more likely to provide care for minority patients (Komaromy et al., 1995~. The race and ethnicity of the health care provider are also factors in the choice of physician for up to 40 percent of minority patients (Commonwealth Fund, 2002; Saha et al., 2000~. Additional studies suggest that gender and race concordance is important in patient- doctor communication and receipt of preventive care (Cooper-Patrick et al., 1999; Docteur et al., 1996; Saha et al., 1999~. Access to a diverse workforce differs from cultural competence. There is no evidence to suggest that minority physicians are more culturally competent simply by virtue of their race or ethnicity. However, a diverse provider workforce improves cultural competence by exposing all providers to peers of different racial and ethnic backgrounds. Cultural and language barriers to access have long been recognized as impediments to patient-centered access. Beyond the provision of language access and provider knowledge of cultural beliefs and traditions, there has not been general agreement about what cultural competence is. However, core components have been proposed (Brach and Fraser, 2000; California Pan-Ethnic Health Network (CPEHN), 2001~. Each of the consumer perspectives on health care needs identified by the framework for the National Healthcare Quality Report should be examined for their implications for our understanding of health and what creates it. Access-related aspects of these domains are briefly discussed below: Staying healthy involves an array of access-related health behaviors and preventive services that occur across the lifespan, and many of these are well understood. There is, for example, a large body of work on access to breast and cervical cancer screening. In addition, appropriate follow-up after an abnormal Pap smear or mammogram is important to access, quality, and outcomes. This is an area in which large disparities are thought to exist. Getting better refers primarily to getting acute care, and measures are discussed below. Two aspects of the continuum of care have changed substantially since access to care was initially conceptualized: care for
1 12 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT chronic disease and end-of-life care. Living with chronic disease and disability has become more common, and it is clear that aggressive management can prevent complications and declines in function. This has led to the evolution of new models for chronic disease management (Wagner et al., 2001~. These often involve team-based care including, for example, nutritionists or physical therapists. If these models are demonstrated to be pathways for achieving the "best possible outcome," access to chronic disease management services will become even more important. Because minority patients are disproportionately affected by some of the conditions most amenable to disease management programs, including diabetes or asthma, access to these services should be a consideration in the NHDR. There are no large data systems that currently measure need-based access to, or use of, disease management services. However, some aspects of team- based care can be examined through medical claims. Coping with the end of life has also become more important as the population ages. Cultural differences associated with ethnicity and geography often affect the nature of end-of-life care and are associated with differing degrees of informal caregiving. The wide variation in preferences makes measures of service utilization somewhat difficult to interpret. Nonetheless, three common measures of use of end-of-life care services--use of nursing home care, use of home nursing, and use of hospice care--are available from Medicare administrative data. With regard to geography, only 11 counties in the U.S. do not have a hospice provider serving them. This does not necessarily mean that someone in need of hospice care can get it. Recent work indicates relatively small racial and ethnic differences in use of hospice services or skilled nursing facilities (Lynn and Shugarman, 2002~. Because Medicare covers much of the population needing both chronic care and end-of-life care, access to these services does not seem to be as problematic as it is for the uninsured. However, older rural residents are less likely than the rest of the population to be covered by Medicare. Further, Medicare alone does not provide sufficient financial access for many seniors for whom the co- payments, deductibles, and drug costs all pose financial barriers to care. The availability of Medicaid coverage for Tow-income senior
DISPARITIES IN ACCESS TO CARE 113 populations, which are disproportionately rural and minority, varies from state to state, again limiting access for many Tow-income Medicare beneficiaries. Some states with large minority and rural populations have some of the least generous Medicaid programs. Additional factors such as geographic access; the ability to understand the clinician or to participate in informed decision making; the ability to navigate the health care system; language access and availability of translation services; and access to understandable information before, during, and after the health care encounter all affect care throughout the continuum. This discussion has highlighted the relationships between access and quality in terms of the framework presented in Envisioning the National Health Care Quality Report. These relationships and their interface with community factors are depicted in Figure 4-2. 4-4. CHALLENGES TO EXAMINING DISPARITIES IN ACCESS A series of issues challenge the examination of disparities in access in the NHDR. They include issues related to data and measurement, socioeconomic position, cumulative effects, managed care, undercounted populations, and utilization as a measure of access. The following is a brief treatment of each. Data and Measurement A series of data and measurement issues will be relevant to consideration of the types of measures used and their presentation. A forthcoming report from the National Center for Health Statistics (NCHS) outlines some of the methodological issues in calculating disparities such as whether the differences are relative or absolute, whether they are framed in the positive or negative, and whether a summary disparity score for a given measure is useful (Keppel, 2002; National Center for Health Statistics, 2001~. Many of the decisions to be made regarding these issues will depend on the context in which they are being used.
/ · - e :- 'e v 'e 'e .= o v v co lo lo lo 'e · - ~ A, o .~ 'e / In cO ~ -I .cn ~ . =, lo ~ ~ cO lo ~ - cO ~ a .= o A ~ Q o .' \ . ~ lo I ~ .' I cO En ~ ~ cat ~ o ~ o ._ ,, ~ o ~ ~ lo I ~ cut .~ Q En to En Cal cn o o I T C~ ~ O. _ cn ._ . ~ o , o .= L~
DISPARITIES IN ACCESS TO CARE 115 Another issue relates to heterogeneity of the populations studied. Each of the racial and ethnic groups and the groups of individuals considered "rural" varies tremendously by subgroup. African- and Caribbean-born Blacks clearly have different health status and health outcomes than American-born Blacks. Hispanics and Asians come from many countries, and they differ widely in the health behaviors and cultural patterns that affect access and use. There is tremendous variation in rural areas: culture, lifestyle, and health needs differ widely among, for example, rural North Dakota, Texas, and Appalachia. Aggregation by race or by urban/rural status masks many of these differences, just as examining the total population masks differences by major racial and ethnic groups. A more complicated set of issues concerns data availability and quality. The recent census includes the numbers of individuals who self report mixed race or ethnicity, a different kind of classification issue. Continued increases in those reporting more than one race or ethnicity will make comparisons from baseline data difficult. The Office of Management and Budget (OMB) is developing a crosswalk that will facilitate examination of changes over time. Racial misclassification continues to be an issue of concern in many data sources (IOM, 2002b). This is most likely to occur when an individual's race is not obtained by self report, but rather through observation by a third party such as a clerk assessing eligibility. While most of the large national survey efforts obtain self-reported information about respondent race and ethnicity, these data are not routinely collected and recorded in most health care databases, which are often sources of information regarding utilization of care. Race and ethnicity data are sometimes available by linking administrative and utilization data, and these vary in accuracy by data type (for example, Medicare and Social Security, Medicaid and eligibility files, and race and ethnicity). In addition, race and ethnicity data are available from hospital discharge data in states contributing to the Hospital Cost and Utilization Project (HCUP), although the amount of missing data is often substantial. Most health insurers and health plans do not routinely record this information, making it extremely difficult to measure utilization or quality for different racial
1 1 6 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT and ethnic subgroups. A recent review of all state laws regarding the collection of racial and ethnic data by health plans highlights the possibilities of collecting race and ethnicity data in all states, although laws in a few states preclude its collection prior to health plan enrollment (Youdelman, 2001~. Efforts to encourage collection and use of racial and ethnic data are underway in both the public and private sectors. This suggests that it should be possible to use this information for future measures of access and quality. Currently, Medicare data provide the greatest opportunity to measure disparities r In use OI SerV1CeS. Availability of data is not necessarily sufficient to answer important questions. In many epidemiological databases and national access and utilization surveys, minority subgroups are too small to make comparisons between groups possible. This is especially true for information at the state or health plan level. This problem becomes even more acute for measures drawn from sources such as the Health Plan Employer Data and Information Set (HEDIS) or the Consumer Assessment of Health Plans (CAH:PS) in individual health plans, particularly within given diagnoses. Recent attempts to examine such data for Medicare+Choice health plans found no plans with at least 30 enrollees in each of the major race and ethnicity classifications and less than a dozen plans in which there were sufficient numbers of Asian Americans or Hispanics to analyze. While there were many plans that had at least 30 African American and White enrollees, sample sizes that small make even simple comparisons very difficult Laurie et al., 2002~. Socioeconomic Position Some conceptual and philosophical challenges underlie interpretation of racial, ethnic, and geographic differences. Foremost among them is the consideration of socioeconomic status and how it relates to racial, ethnic, or geographic disparities. A long literature has attempted to disentangle these relationships, finding that if socioeconomic status is "controlled for," the magnitude of the disparity attributed to race or geography becomes smaller. However, such analyses are often used to dismiss the fact that serious disparities exist. Moreover, they do little to illustrate how race affects
DISPARITIES IN ACCESS TO CARE 117 socioeconomic position or opportunity or to disentangle the separate effects of education and income (Williams, 1999~. Furthermore, if we accept that community-level socioeconomic factors affect access or are etiologic in creating need, they need to be considered in measuring access, especially since addressing them may lead to improved outcomes. Some of these include conditions associated with poverty, such as poor access to quality schools, poor housing, undernutrition, and substance abuse. Each of these conditions contributes to poor health. They act to impede access in ways that go beyond the contributions of factors such as cost or transportation. In addition, they change the nature of perceived needs and the internal resources for addressing them. The immediate needs of caring for one's family often make prioritizing health needs nearly impossible.4 Ultimately, this is a report about the experience of different racial, ethnic, and geographic populations, and examining them means that their experiences cannot be disconnected from their environments. One way to address this problem, at least partially, is with need-based measures that pertain to populations whose needs for care are as similar as possible. Controlling for health status can move us in that direction. Certainly, adjustment for socioeconomic status will be important in some areas where presenting data by race and ethnicity within income or educational groupings may help in interpretation. Whether income or education is more relevant to access and whether one of these should be a preferred measure of socioeconomic status will, of necessity, depend on the consistency and the reliability of the data and findings of future research (Gornick, 2000~. Cumulative Effects The relationships between access to care and each of the key variables we have been discussing ethnicity, insurance status, and rural residence have been extensively examined. However, the 4 It is worth noting that both the stigma and the health consequences of being poor in an urban or a rural area often differ significantly. It is also important to keep in mind the wide variation in socioeconomic status within all racial and ethnic groups.
1 1 8 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT relationship between insurance status and access or outcome has usually been examined controlling for race and ethnicity or vice versa. Very few studies have examined the cumulative effects of these factors. A recent study has suggested that having a combination of variables may predict worse access or outcome than one alone. For example, being uninsured and Hispanic is associated with worse access than either alone (Burstin, 2001~. The addition of rural residence to this mix has not been examined. Managed Care As noted above, managed care provides unprecedented opportunities to define a denominator population and to monitor its utilization, quality, and outcomes. It is probably premature to measure its effects on access in rural areas. The challenges posed by racial and ethnic data in managed care plans have been discussed above. Nonetheless, two data sources can provide information on access for racial and ethnic minorities in managed care. The HEDTS measures for any use, mammography, and components of diabetes and mental health care are readily available from Medicare+Choice plans, and racial and ethnicity data can be obtained by linking to Social Security data. Recent analysis indicates that disparities remain quite substantial across all plans (Schneider et al., 2001~. CAMPS data are available for both Medicare+Choice enrollees and the commercially insured population through the National CAMPS Benchmarking Data Set. The instrument includes a self-reported measure of race and ethnicity, measures of any use, use of a specialist, and need to see a specialist. Data for both populations, even when adjusted for socioeconomic status, indicate significant disparities across all plans. They also indicate large plan-to-plan variation in the degree of disparities, highlighting the heterogeneity among health plans (Lurie et al., 2002~. A significant problem with CAMPS is the inability to examine response rates, particularly those for different racial, ethnic, or geographic populations. Managed care has the potential to reduce disparities by focusing on denominator populations. Alternatively, it has the potential to worsen them if cost control efforts create differential access barriers; if providers engage in differential advocacy; or if there
DISPARITIES IN ACCESS TO CARE 119 are problems navigating the system. These issues are well clescribec! elsewhere (TOM, 2002b). There is still debate in the literature about which aspects of manager! care are most important to measure, for example, capitation versus other payment mechanisms, gatekeeping, anc! restrictions on provider choice. However, each is relevant to . . . . . examining c .~spant~es. Undercounted Populations Two distinct populations of high need can significantly skew assessments of neec! for both care anc! access. Minority populations, particularly African Americans anc! Hispanics, are incarcerated at much higher rates than Whites, often clue to clifferential sentencing for drug-relatec! convictions (Iguchi et al., (in presser. Currently, most population-basec! measures of access, inclucling insurance status, clo not include those who are incarcerated. This has lee! to serious unclercounts of the uninsured in some areas, particularly African American anc! Hispanic men. Access to care in prisons is highly variable, as is access to insurance after release. Most parolees do not get jobs that provide health insurance. In some states, they are ineligible for MeclicaicI. Other populations that are seriously unclerrepresentec! when it comes to assessing access to care or insurance status are unclocumentec! immigrants and the homeless. Many clo not appear in the census or in surveys, yet they have substantial unmet health needs in both urban and rural areas. Utilization as a Measure of Access Some utilization measures that have long been considered indicators of access have been the subject of significant criticism, largely because they are felt to reflect preferences for care as much as unmet need. It has been argued that most people who are uninsured do not need or want insurance, although data consistently indicate this to be a small percentage of the uninsured. The issue is similar when examining the proportion with no visits, without a usual source of care, or without follow-up. When examined, only a small proportion of those without a usual source of care do not want one. The case of
1 20 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT children and vulnerable adults is a special one. The adult acting on their behalf may not bring them for care for reasons unrelated to need. In these cases, the access barrier is often intrinsic to their caregiver. Some differences in utilization are believed to represent overuse by the White population. As is the case with socioeconomic position, stratifying measures by health status permits comparison of populations with similar needs. Use of emergency departments for nonacute, first contact care is a slightly different issue. Care can, in fact, be obtained through emergency departments, and by law, emergency departments cannot refuse to see a patient. Increases in emergency department use are, in part, a function of not having an alternative place to go, not wanting or needing a regular provider, and convenience (Asplin, 2002~. This leads some to suggest that emergency department use is an expression of preferences rather than an access problem. From a policy perspective, this is probably not a desirable utilization pattern. Emergency department overcrowding has become a significant national concern, and as such, impedes access for those who truly need emergency care. Furthermore, when people seen in emergency departments are offered a primary care visit within a short time, most choose non-emergency department care (Baker et al., 1991; Bindman et al., 1995~. For chronic disease, it is likely that utilization patterns reflect a combination of access and quality as is the case for a set of quality indicators for follow-up care or receipt of tests such as eye exams for diabetes. This kind of utilization clearly depends in part on aspects of quality such as recommendations made by the provider or system for follow-up care, or the system's ability to prompt the physician to provide--and the patient to obtain--needed care. Patient follow-up is also likely to be a function of patient-centered issues such as the quality of communication and the ability to understand the care plan. However, we must be careful not to assume that access ceases to be a barrier once a patient has entered the system, particularly for chronic care. Co-payments frequently pose barriers to obtaining needed follow-up care, as do other barriers such as transportation. While it is impossible to disentangle access and quality, chronic disease related- utilization should be a part of measuring access, particularly when examined in relation to need.
DISPARITIES IN ACCESS TO CARE 121 A final challenge in considering utilization as a measure of access relates to cultural factors that influence care seeking. A person's cultural beliefs (whether from a different country or a rural community) might lead to different definitions of illness or need for care. in that case, many would argue that lack of care does not reflect barriers to access. However, factors such as the inability of the patient and provider to speak the same language, lack of trust, fears of discrimination or fears of disrespect of one's culture, a prohibition of care for a woman by her male partner, along with discrimination or a lack of cultural sensitivity on the part of the provider all constitute cultural barriers that are reflected in utilization patterns. There are no easy solutions to these problems. Utilization measures must be interpreted with these issues in mind. 4-5. PRINCIPLES GUIDING MEASUREMENT OF ACCESS IN THE NATIONAL HEALTHCARE DISPARITIES REPORT Outlined below are key principles that should guide the development of access measures for the NHDR: I. Every effort should be made not to marginalize the NHDR. This can be accomplished, in part, by assuring compatibility with the National Healthcare Quality Report, and by using, when possible, its conceptual framework for reporting on access-related quality measures. Because equity is such a fundamental component of quality, it is expected that many quality measures will be reported by racial and ethnic subgroups. Some overlap between disparities in access and quality measures should be expected, but it would be highly desirable to use similar reporting formats. 2. The NHDR should be anchored in well-accepted, extant measures of access for which data are regularly collected. This will avoid the need to create and validate measures, an expensive and time consuming process. It will also link the NHDR to other reports and sources that use these measures and support its continuity from year to year.
1 22 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT 3. As recommended by FiscelIa, measures should address access issues that are particularly relevant for minority or rural populations (Fiscella, 20021. These measures should be relevant for the general population, but they should focus on areas in which disparities are likely to exist. In addition to general population measures, the committee should identify a set of measures for which access to quality care is particularly (but not solely) relevant to minority populations. Language access may be such an example. 4. Prevalence and contribution to morbidity and mortality should guide selection of these conditions. The 1998 Federal Initiative to Eliminate Racial and Ethnic Disparities in Health has focused on six clinical areas, all of which meet these criteria (CDC, 2000~. They are infant mortality, immunization (child and adult), diabetes, cardiovascular disease, cancer screening and treatment, and HIV. 5. While the report should be anchored in a core of existing measures, new measures are sorely needed and should be developed for use in subsequent reports. These should account for the trends described above, including increase in chronic disease, a broader definition of health, and a recognition that factors outside of the immediate delivery system have major effects on access. 6. Data sources for this report will be an important consideration. Although reliance on federally collected data is most desirable, it is possible that reductions in funding for some kinds of data collection will seriously impair the ability to use necessary data. To assure that this report will provide important information about disparities in access, criteria for the use of data collected with nonfederal funding from such sources as foundations and professional societies should be considered. 4-6. CORE MEASURES The set of core measures traditionally considered to measure access should form the basis of this report. They are well accepted and relevant for all populations. All of them reveal disparities. Specific issues related to their use are discussed below.
DISPARITIES IN ACCESS TO CARE 123 Insurance Status That insurance is often a prerequisite to receiving care has been thoroughly documented. Disparities in the presence and type of insurance are also well known, as are the associations among insurance status and utilization, quality, and health outcomes. The presence of any insurance coverage should remain a core measure. Measures of underinsurance should be developed for future use and are discussed later. Issues related to the uninsured are the subjects of a separate, ongoing TOM project (IOM, 2002a). Methodological issues in the reporting of insurance status must be considered. These are relevant regardless of whether the focus is on racial and ethnic disparities or on the general population. First, multiple federal and private data collection efforts assess insurance status. Each asks the questions Mightily differently, and the wording of some of the measures has changed over time, leading to different estimates and frustration for policy makers (Berk and Schur, 1998~. The Current Population Survey (CPS) has added a validation question to its measurement of insurance status, and this has lowered estimates of the uninsured in many states and made time trends difficult to interpret (U.S. Bureau of the Census, 2002b). Measures of insurance status are well summarized in Table B! of Coverage Matters (TOM, 2001a). The NHDR would be best served by choosing a measure that is likely to remain stable over time. While the absence of insurance has repeatedly been demonstrated to inhibit access to care, relationships between other aspects of insurance coverage and access are more complicated. Even in the presence of insurance, most Americans face co-payments, deductibles, and other degrees of cost sharing. Many have policies that do not cover primary and preventive care or medications. The latter is a particular feature of Medicare. Based on a definition proposed by Bashshur et al. (Bashshur et al., 1993), underinsurance can be characterized as "a situation in which the consequences of having less than full coverage are so burdensome that they inhibit realized access to needed care likely to result in the best possible outcome." Bashshur et al. distinguish structural elements of underinsurance (for example, benefit packages), experiential elements (for example, degree of out-
1 24 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT of-pocket coverage or part-time coverage), and perceptual elements (for example, the assessment of the insured individual about adequacy of coverage). Thus, underinsurance must be defined relative to a standard and does not necessarily mean having to pay for all or some portion of care. Because primary and preventive care are necessary for staying healthy, underinsurance can be said to exist when coverage for these services are not included in a benefit package. Similarly, underinsurance exists when essential medications are not covered although this is the current standard for many plans, including Medicare. While some authors have defined specific levels of out-of- pocket payment relative to income that constitute underinsurance, that level likely varies with income, making a uniform definition difficult. Use of a uniform standard could thereby be misleading, particularly when considering underinsurance for groups who, on average, have different incomes. It may be most useful to present information about premiums, covered benefits, any co-payments, and any deductibles for different races and ethnicities by income groupings, thus allowing the reader to further interpret the data and avoid the pitfalls discussed above. Proportion of Adults and Children without a Visit and Their Health Status These measures of utilization are often considered to measure access to care. General problems with utilization measures have been described above. Most guidelines call for young children to have preventive visits at least annually (or at least biannually for adolescents). However, apart from certain screening services, standards are not so clear for healthy adults. While insurance status is one predictor of foregoing a visit, minorities are less likely than Whites to have any visits, even within insured populations. In these cases, insurance is not a major determinant of use. Lack of visits may also reflect lack of need (real or perceived) or desire for care, or other predisposing and enabling barriers such as language, culture, or geography. The interpretation of"no visits" is less likely to be problematic for those with chronic health conditions or those in worse health. Nonetheless, data on whether or not a visit was obtained are easily available and reliable to the extent that racial and ethnic data are
DISPARITIES IN ACCESS TO CARE 125 accurate. Stratifying by health status will help with interpretation of the data. Regular Site of Care and Regular Provider Measures of access have traditionally included measures of continuity of care and primary care, usually embodied in having a regular site of care or regular provider. The controversies around these measures relate primarily to the small number of individuals who do not want a primary provider, and the debate about whether a regular provider is as important as a regular system. In several studies, lack of a regular site and/or provider were the primary reasons for not seeking care when needed (Ettner, 1999~. Additional studies show that both measures contribute independently. As managed care continues to evolve, attributes of the system may become more important than having a regular provider. These measures are readily available from national surveys, and the bulk of the evidence suggests that they continue to be useful, particularly if conditioned on health status. fOM Indicators The 1993 TOM report identified a series of access indicators important to quality or outcome-related objectives. These are discussed briefly. Measures Related to Birth Outcomes Although recent studies have questioned the relationships between prenatal care (particularly the amount) and birth outcomes, they may reflect that we are victims of our own success. As it is evermore possible to save babies born at increasing levels of prematurity, the number of Tow birth weight babies has continued to rise. Other poor outcomes, such as maternal transmission of HIV or congenital syphilis, remain strongly linked to prenatal care. Fortunately, rates of both have decreased dramatically. Hence, measures of adequacy of prenatal care continue to be relevant.
1 26 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT Immunization High levels of immunization for vaccine-preventable illness remain important public health goals. However, several factors should be considered in the selection of immunization measures. First, as the vaccination schedule becomes more complex, we are likely to see greater disparities in the use of some of the newer vaccines. This raises the question of whether being up-to-date on diphtheria, pertussis, and tetanus (DPT)+polio+measles, mumps, and rubella (MMR) is the right measure, or whether uptake of newer vaccines (hepatitis B series, Hib, and conjugate pneumococcal vaccines) are measures that are more likely to reflect ongoing disparity challenges. The quality movement teaches us that measurement is important in improving performance. An unintended consequence of switching measures may be less vigilance in the public health and provider community for achieving high rates of coverage. Second, an increasing number of parents are choosing not to immunize children for reasons unrelated to access. Whether these numbers are large enough to affect population-wide rates and whether this phenomenon varies by race and ethnicity is unclear. Third, we now recognize that immunization is as essential for adults as for children, and that adult immunization rates are lower than those for children. Both child and adult immunization remain important access indicators. Early Detection and Diagnosis of Treatable Disease Despite the recent controversy, receipt of mammography and receipt of Pap smears continue to be important measures of access to care. Data indicate that preferences and culture play relatively minor roles in not receiving these services. Additional potential measures in this category are suggested in Section 4-7. Reducing Effects of Chronic Disease: Preventable Hospitalizations for Ambulatory Care Sensitive Contlitions Preventable hospitalizations for ambulatory care sensitive (ACS) conditions are most useful as measures of access when used alongside hospitalization for conditions not associated with access to
DISPARITIES IN ACCESS TO CARE 127 care. ACS admissions are a function of access to care (they occur more often among Tow income and uninsured populations), quality of care, and factors that are less related to the health care system. These data are usually examined with regard to geography, and ACS admissions are more concentrated in low-income areas. The challenge is to have enough knowledge about the denominator to be able to interpret the numerator. To the extent that the denominator (by race, ethnicity, and socioeconomic status) can be derived from census data, ACS admissions are probably useful measures of access. They do not, however, address other sources of regional variation such as practice patterns. Several methods to address this issue, such as creating ratios of ACS to non-AC S admissions, appear promising, but they need additional validation prior to use in a national report (Billings, 2002~. Reducing Morbidity and Pain through Timely Treatment As suggested above, annual dental care visits remain an appropriate measure of access to oral health care, and there is broad agreement on the need for annual visits at the very least. Furthermore, oral health care is known to be associated with profound disparities by race, ethnicity, and geography. Examining reports of delayed care when needed continues to remain an important way to measure access. A need-based measure (such as care for serious symptoms) would be ideal, but data are not consistently available. As an alternative, it would be useful to stratify this measure by those in fair or poor health to allow further inferences about timely treatment for exacerbations of chronic disease. While access for those usually in good health is also a concern, this information is harder to interpret absent a need-based measure (TOM, 1993~. 4-7. SUGGESTIONS This section makes suggestions in four categories: 1. existing measures; 2. measures that can be readily constructed with currently available data; 3. measures that can be constructed with currently available data, but that need to be pilot tested and/or validated; and 4. measures that need to be developed.
128 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT :~. Existing measures The following core measures should continue to serve as a foundation for the NHDR:s 1. Presence of any insurance. Usual source of care and provider, stratified by health status. Proportion of children and adults with no visits, stratified by health status. 4. Proportion needing care and not getting it, stratified by health status. Preventable hospitalizations for ambulatory care sensitive conditions for both children and adults, provided denominator data are suitable. 6. Adequacy and timeliness of prenatal care. 7. Rates of congenital syphilis. 8. Childhood immunization. 9. Proportion of children and adults without a dental visit. 10. Proportion of women over age 50 with/without a mammogram or proportion of adult women with/without a Pap test (adjusted for hysterectomy status). s The need for stratification of the second, third, and fourth measures by health status has been discussed above. There is much less disagreement about the need for care for those in worse health. In the case of childhood visits or immunizations, one could still consider utilization patterns a matter of parental preferences. However, as discussed earlier, children who do not get care that is generally felt to be needed may still be experiencing access barriers, even if their caregivers account for their lack of access. Rates of congenital syphilis have been declining, but they are higher than rates of perinatal HIV. Measurement should continue until rates have decreased by another 50 percent or until a new measure is identified.
DISPARITIES IN ACCESS TO CARE 2. Readily Constructed Measures 129 The following describes some new measures that can be readily constructed from existing data. In addition, the literature supports their use as indicators of access or access-related quality. New measures can also be readily constructed from existing data on consumer perspectives on care (staying healthy, getting better, living with illness or disability, and coping with the end of life) that are part of the NHDR's framework. These are described below. Insurance In addition to the presence of any insurance, the following measures are important in understanding disparities in insurance status: presence of part-time coverage, any coverage for primary/preventive care or medications, any co-payment, and any deductible. As discussed above, the cost-sharing measures should be stratified by income within racial and ethnic groups. Staying Healthy ~ . Rates of neonatal transmission of HIV. 2. Proportion of children with screening provided by Medicaid's Early Prevention, Screening, Diagnosis, and Treatment (EPSDT) Program. 3. Distribution of housing with lead paint that has not been rehabilitated. 4. Proportion of adolescents with no visit and their health status. 5. Proportion of adolescents with up-to-date vaccinations. 6. Proportion of adults with pneumococcal or influenza vaccines. 7. Deaths from pneumococcal pneumonia. 8. Deaths from complications of influenza. 9. Rates of colon cancer screening by flexible sigmoidoscopy or colonoscopy. 10. Proportion without a blood pressure check in past tWQ years.
1 30 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT 11. Proportion without cholesterol screening. 12. Proportion of individuals with ESRD whose first presentation is for dialysis. 13. Proportion of adults who are edentulous. It should be noted that rates of neonatal HIV transmission could easily be examined and are related to the goal of promoting good birth outcomes. However, those rates have fallen rapidly and probably do not provide additional information about comparatively higher rates of congenital syphilis. For the consumer perspective of staying healthy, a goal of promoting healthy development should accompany the goal of promoting good birth outcomes. This would include measures of developmental screening and screening for lead (see below). Measures of lead prevalence and screening have the advantage of addressing a community health issue, which ultimately is access to lead-free housing. Immunization is an issue for adults as well as for children. Currently, racial and ethnic disparities in adult immunization are significantly greater than those for children. Preventing complications of disease through early detection and treatment continues to be a major goal of care. Currently, cancer screening for women is the most widely accepted measure, and monitoring cancer screening rates has done much to increase them. Colon cancer screening should be added to this list because it affects men as well as women and is a costly procedure in the absence of health insurance. Some would argue that nonreceipt of these tests is largely a reflection of personal preference. However, it remains the case that the most common reason they are not done is lack of physician recommendation, which has been shown to be an area in which disparities occur. Cancer screening is not the only way to prevent complications. Avoiding complications of chronic diseases such as diabetes can be achieved with access to high quality care. However, this requires that individuals know they have the condition and that they need care. This
DISPARITIES IN ACCESS TO CARE 131 is the case for chronic diseases such as diabetes and heart disease. While additional measures need to be developed, data on hypertension screening and cholesterol awareness are currently available from the Behavioral Risk Factor Surveillance Survey (BRFSS). These measures would reflect the role of community/public health in identifying the need for care that triggers attempts to access the system as well as quality for those in care. Because chronic renal failure is such a prevalent condition among some minority populations, examining incident cases of renal failure whose first presentation is dialysis would be a similar indicator. Finally, the proposed measure for oral health care (percent of adults who are edentuTous) serves to assess a community-level factor and to reflect long term access to quality dental services. Getting Better 1. Rates of knee or hip replacement for arthritis among Medicare beneficiaries. 2. Receipt of thrombolytic therapy for an acute myocardial infarction (AMT). 3. Rates of admission or readmission for serious mental illness after first diagnoses. 4. Proportion of admissions for mental health care without a follow-up visit in 30 days. 5. Rate of suicide. 6. Proportion of those needing mental health or substance abuse treatment who did not receive it. 7. Breast cancer survival, adjusted for stage at presentation. The first two measures examine access to effective but costly care, and both are known to be associated with disparities. Timeliness of thrombolytic therapy is also important. A private data source collects this information (National Registry of Myocardial Infarction, 2002), and it is not available in recurrently collected federal datasets. The mental health measures examine the supply of mental health providers in a community (there are fewer in minority and rural
1 32 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT communities) ant! unmet neeci. Admission for serious mental illness, like admission for asthma or heart failure, is largely avoidable through access to high quality care. Finally, rates of suicide serve to reflect community levels of depression awareness as well as access to mental health services. Suicide rates among Native Americans are quite high, ant! they are increasing rapidity in African American men. It will be important to adjust for regional differences in suicide rates when making these comparisons. In the long run, rates of successful ant! unsuccessful suicide attempts wouic! serve as a better measure, but such a measure needs further clevelopment. One exception wouic! be for aclolescents, where data on suicide attempts are reporter! through the Youth Risk Behavior Survey (YRBS). The proposer! measures regarding substance abuse reflect the importance of this problem for the population as a whole, the reporter! disparities in access to substance abuse treatment, ant! the particular role that substance abuse plays in the HIV epidemic. The proposer! breast cancer measure reflects the fact that the gap in screening between African Americans and Whites has been largely closed. Adjusting for stage at diagnosis also permits examination of populations with similar needs. Because equal treatment is felt to leac] to equal outcomes, this measure likely reflects access to high quality care. Living with IlIness or Disability Proportion of HIV-infectec! inclivicluals who know their status. 2. Proportion of HIV-infectec] inclivicluals who know their status . . ant are recelvmg care. 3. Proportion of ESRD patients referrer! for transplant evaluation or receiving renal transplant. 4. Proportion of Medicare patients acimittec! for heart failure or acute MI who saw a specialist. Proportion of diabetics who receiver! an eye or foot exam from a health professional. 6. Proportion of diabetics who saw a nutritionist for counseling.
DISPARITIES IN ACCESS TO CARE 133 7. Availability of Meclicaic] coverage for oilier aclults below 200 percent of the Fecleral Poverty Limit (FPL). The HIV epidemic has a disproportionate impact on minority communities. These proposer! measures reflect access that is clepenclent on the functioning of the public health system as well as entering ant! remaining in the health care system itself. Disparity in access to specialty services for people with chronic disease is a well- clocumentec! problem. The issue of renal transplantation was previously cliscusseci. Use of this specialty measure for heart disease obviates the need] to risk adjust to examine clifferential use of procedures. The diabetes measures reflect the evolving needs for disease management ant] team care. Diabetes is the recommenclec] disease entity here because disease management has been best stucliec! for this conclition. In aciclition, there is a high prevalence of diabetes in Hispanic, Native American, ant! Afiican American populations. As cliscusseci, insurance coverage for low-income elclerly with chronic . . . c disease remams a serious access issue. Coping with the End of Life I. Proportion of Medicare beneficiaries with cleath from a cancer diagnosis who receiver! hospice services or home care in the last six months of life. 2. Proportion of those coverer! by Medicare or Meclicaic! who cliec! from HIV ant! who receiver] hospice services in the last year of life. 3. Proportion of Medicare beneficiaries receiving care in a skillet! nursing facility in the last year of life. Availability of Meclicaic] coverage for oIcler aclults below 200 percent of the FPL. Unfortunately, all of these potential measures are baser] on kinds of utilization that are likely to reflect personal ant] cultural differences in preference for ens! of life care, ant! they will Beet! to be interpreted in that light. Contrasting patterns of ens! of life care for cancer ant! HIV will enable examination of disparities in care for conditions with different degrees of social acceptance.
1 34 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT Indicators Relevant to the Entire Continuum of Care . Racial and ethnic distribution of matriculating medical, dental, and nursing students. 2. Proportions of racial, ethnic, urban, and rural populations with Internet access. 3. Literacy, especially health literacy. Issues related to a diverse workforce are discussed above. Information regarding Internet access is available through the Department of Commerce and can serve as an indicator of access to information, especially since this will be a major vehicle for obtaining information in the future. The Department of Education is scheduled to begin data collection in the summer of 2002 on the National Assessment of Adult Literacy Survey (NAALS). A question regarding access to care is currently planned. This will facilitate further understanding of the relationships between literacy and access, as well as provide a direct assessment of literacy. 3. Other New Measures Existing data could also support measures on important aspects of access, but these will require pilot testing and/or validation. The following lists those measures and briefly describes ways that they could be tested and/or validated. Proportion of children on Medicaid who are screened for elevated lead levels. Proportion of people with diabetes who first present with end organ damage. 3. Accessibility of a mental health provider. 4. Proportion of the population who needed mental health care or substance abuse treatment, but who did not receive it. Lead screening is now a Medicaid requirement, and this information is, in theory, available from state Medicaid programs. Work will be required to examine the reliability of this information prior to use. The diabetes measure can presumably be obtained for the
DISPARITIES IN ACCESS TO CARE 4-2 through 4-4. 135 Medicare population by examining new diabetes diagnoses and secondary diagnoses reflecting end organ damage within a specified time frame after diagnosis. Validation of the measure could be accomplished through chart review or potentially through linking with data from the National Health and Nutrition Examination Survey (NHANES) and Medicare. Geographic accessibility of mental health providers can theoretically be determined by mapping community demographics, community mental health centers, and information from professional societies. However, this cannot provide information on wait time to appointment, whether providers are taking new patients, the number of patients who have Medicaid, or the number of patients who are uninsured. Nonetheless, the presence of providers in a community is a prerequisite to actually getting an appointment. Information about language capability of mental health professionals is available from health plans and professional societies. Proposed measures and probable data sources appear in Tables 4. Measures That Need to Be Developed Measures most in need of development are those that reflect patient-centered aspects of care, culture, and the community role in access to care. These are described below. Stereotyping/bias/discrimination. These factors may account for racial and ethnic differences in care. While this has been easier to measure in other settings such as housing or banking in which testers can be used, it is more difficult in the health care setting. This issue was the subject of a recent conference at the National Institutes of Health (NTH) on measuring bias and discrimination in health, but no consensus was reached regarding which measures to use. Recently, the Commonwealth Fund released a study reporting on the frequency of perceived discrimination in the health care setting. Analysis of the relationships between those items and measures of access to care is underway. A related issue is trust, which is relevant for both the individual provider and the system overall. Here, some measures have been developed, and these show quite divergent levels of trust, at least
136 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT between Whites and African Americans. However, these are not widely used, and further development is necessary. TABLE 4-2 Staying Healthy MEASURE PROPOSED SOURCE Rates of Neonatal Centers for Disease Control and Transmission of HIV Prevention (CDC) Proportion of Children with Centers for Medicare and Medicaid EPSDT Screening Services (CMS) _ Proportion of Children CMS Screened for Elevated Lead Levels Distribution of Unrehabilitated Nepal lment of Housing and Urban Housing with Lead Paint Development (HUD) _ . Proportion of Adolescents with Youth Risk Behavior No Visit and Their Health Status Proportion of Adolescents with CDC Up-to-date Vaccination Proportion of Adults with Behavioral Risk Factor Pneumococcal or Influenza Surveillance System (BRFSS) Vaccine Deaths from Pneumococcal CMS, Health Cost and Utilization Pneumonia Project (HCUP) Deaths from Complications of CMS, HCUP Influenza .. Rates of Colon Cancer CDC, BRFSS Screening by Flexible Sigmoidoscopy or Colonoscopy Proportion without a Blood BRFSS Pressure Check in Past Two Years Proportion without Cholesterol BRFSS Screening _ Proportion of Diabetics Who CMS, National Health and First Present with End Organ Nutrition Examination Survey Damage (NHANES) Proportion of Individuals with ESRD End Presentation of Dialvsis , Proportion of Adults Who Are National Health Interview Survey Edentulous (NHIS)
DISPARITIES IN ACCESS TO CARE TABLE 4-3 Getting Better 137 MEASURE PROPOSED SOURCE Rates of Knee or Hip Replacement Centers for Medicare and Medicaid for Arthritis among Medicare Services (CMS) Beneficiaries Receipt of Thrombolytic Therapy for CMS an Acute Myocardial Infarction (AMI) . . Mental Health Provider Accessibility Center for Mental Health Services (CMHS) Proportion of Population Who CMHS Needed Mental Health Care But Did Not Receive It Rates of Admission for Serious CMS Mental Illness Proportion of Medicare Admissions CMS for Mental Health Care without a Follow-un Visit in 30 Dave . , Rate of Suicide National Center for Health Statistics (NCHS) ~ , Proportion of Those Needing Substance Abuse and Mental Health Substance Abuse Treatment Who Services Administration (SAMHSA), Did Not Receive It Organization of American States (OAS) ~ , Substance Abuse Provider National Household Survey on Drug Accessibility, SAMHSA OAS , Breast Cancer Survival, Adjusted for Surveillance, Epidemiology and End Stage at Presentation Results (SEER), CMS Cultural competency. While measures that examine the cultural proficiency of the individual provider and of a health care institution are currently under development through funding from Health Resources and Services Administration (HRSA) and the California Endowment, a generally accepted, consistent, and operational definition of cultural competency is still needed. To be most useful, measures should be validated and have a clear and demonstrable relationship to access and outcomes.
138 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT TABLE 4-4 Living with Illness or Disability MEASURE PROPOSED SOURCE Proportion of HIV-~nfected Centers for Disease Control and Individuals Who Know Their Prevention (CDC) Status Proportion of HIV-~nfected CDC Individuals Who Know Their Status and Are Receiving Care Proportion of ESRD Patients United States Renal Data System Referred for Transplant System (USRDS) Evaluation or Receiving Renal Transplant Proportion of Medicare Patients Centers for Medicare and Admitted for Heart Failure or Medicaid Services (CMS) Acute Myocardial Infarction (AMI) Who Saw a Specialist Proportion of Diabetics Who CMS Received an Eye or Foot Exam from a Health Professional Proportion of Diabetics Who Saw CMS a Nutritionist for Counseling Availability of Medicaid Current Population Survey (CPS), Coverage for Older Adults below CMS 200% of Federal Poverty Level (FPL) Language access. The Office of Civil Rights in the Department of Health and Human Services (DHHS) has promulgated standards for language access, but there are not yet regularly accepted ways to assess this. Both institutional and patient-reported measures are critical since this is one of the more actionable domains of access. New measures should be based on considerations of whether language access is available and, if so, the extent of its quality. Measures of community/public health contribution to access. The need for these measures to assess performance of the public health system and the role of communities has been discussed above. Topical areas might include comparisons of estimated incidence of disease in a community or population to actual rates of detection/reporting, measures of awareness and health information in communities, and measures of community beliefs and values. Many would argue that
DISPARITIES IN ACCESS TO CARE 139 access at a community level to resources associated with underlying causes of disease ought to be included in a report about disparities in access. These might include levels of educational attainment, employment, housing quality, or access to supermarkets. Measures of access to specialty care and to disease management services. As chronic diseases become more prevalent and complicated to treat, access to specialty care may be necessary to achieve the best possible outcomes. Utilization measures are fraught with the problems described above. Self-reported measures can be constructed from CAMPS measures on the need to see a specialist and whether one was actually seen. However, these measures rely on the patient's understanding of whether specialist care is needed and therefore may not be ideal. As in the HIV and renal transplant examples, not knowing that care is needed is a major barrier to access. With regard to disease management, utilization of specific services common in disease management programs (for example, podiatry or physical therapy) can be examined with claims data. However, comprehensive disease management cannot. Better need-basec! measures of access to mental health and oral health care. While these measures should be applied across the lifespan, the current Youth Risk Behavior Survey offers an opportunity to apply them to adolescents. Currently, measures of depression and suicidal ideation are collected, but mental health use is not. A measure combining rates of successful and unsuccessful suicide attempts would also be useful. Better need-based measures that are both general and disease specific. The importance of such measures has been discussed above. Priority should be given to the development of need-based measures for conditions that are prevalent in minority populations and that contribute significantly to the burden of morbidity and mortality. 4-8. CHOOSING AMONG POTENTIAL MEASURES Given the need to develop new measures, measures should be selected in three groupings. First, the traditional core measures should
140 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT continue to be used. Ultimately, new measures should fulfill the following criteria: They should represent issues that affect all populations, but that affect minority populations in an important way. For disease-specific measures, priority should be given to those conditions that were the focus of the 1998 Federal Initiative to Eliminate Racial and Ethnic Disparities in Health. They should cover the lifespan. They should capture disparities that are known to exist. They should add important information beyond core measures. There is a strong likelihood that the health of minority populations would improve if the focus of measurement were addressed. It is also possible that addressing some foci would improve health for all populations without decreasing disparities. Because the primary aim is improved health, measures should not be discarded for this reason. They are particularly important for specific populations, even if they are less salient to Whites. 7. They fill gaps in the quality framework, including the continuum of care, attributes of quality, or care over the lifespan. 8. They reflect patient-centered or community-centered aspects of access. 9. They incorporate an expanded definition of health. This is particularly important for mental health since it is an important co-morbidity for chronic diseases such as diabetes and etiologic in much care-seeking behavior. Second, new measures should be selected from priority measures that do not involve substantial development. The following identifies the measure and explains the rationale: 1. Proportion of adolescents with no visit, and their health status. Adolescents would constitute a new age group for this measure.
DISPARITIES rN ACCESS TO CARE 141 2. Proportion of adolescents with up-to-date immunization status. Adolescents would constitute a new age group for this measure. 3. Proportion of adults with pneumococcal or influenza vaccine. Adults would constitute a new age group for this measure. In addition, these vaccinations are public health priorities. 4. Rates of colon cancer screening. This measure would extend the measurement of cancer screening to men. 5. Proportion without a blood pressure check. This measure would capture the contributions of the community as well as of the delivery system. Proportion of individuals with ESRD who require dialysis at their first presentation. This measure would capture the contributions of the community as well as of the delivery system. 7. Proportion of HIV-infected individuals who know their serostatus. This measure would reflect the contribution of the public health system. 8. Proportion of children on Medicaid who are screened for elevated lead levels. This measure would reflect the interface between the community and the health system. 9. Rates of admission for serious mental illness after first diagnosis. This measure would expand the concept of health to include mental as well as physical illness. 10. Proportion needing mental health or substance abuse treatment who did not receive it. This measure would expand the concept of health to include mental as well as physical health. 11. Proportion of diabetics who received an eye or foot exam from a health professional or who saw a nutritionist for counseling. This measure addresses health care for a major chronic disease. 12. Insurance issues such as the availability of Medicaid coverage for low income seniors, the proportion of people with insurance that covers primary/preventive care and medicines, and the proportion of people whose insurance includes co-
142 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT payments and/or deductibles. This measure captures different dimensions of insurance coverage. 13. Racial and ethnic composition of the health care workforce. This measure captures a new dimension of access, the importance of which is explained above. discussed. 14. Literacy. This measure captures a new dimension of access, the importance of which is explained above. 15. Proportion of Medicare beneficiaries with death from a cancer diagnosis or HIV who received hospice services or home care in the last six months of life. This measure addresses end-of- life care, which is relatively neglected as an area of measurement (TOM, 200 Ic). Lastly, work should begin on new measures in all of the areas 4-9. CONCLUSION A combination of existing measures, measures that can be developed from existing data, and new measures should be included in the NHDR. When utilization measures are used as access measures, comparisons of groups with similar health needs will facilitate interpretation. Additional measures should be developed, particularly those representing the contribution of community and public health measures to access as well as those focusing on prevalent health conditions in minority populations.
DISPARITIES IN ACCESS TO CARE 143 Reference I,ist Aday, L. A. and R. Andersen. 1974. A framework for the study of access to medical care. Health Serv Res 9 (3~:208-220. AHRQ. 1999. Reauthorization Fact Sheet, AHRQ Publication No.OO-P002. Rockville, MD: Agency for Healthcare Research and Quality. Andersen, R. and L. A. Aday. 1978. Access to medical care in the US: realized and potential. Med Care 16 (7~:533-46. Andersen, R. M., A. McCutcheon, L. A. Aday, G. Y. Chin, and R. Bell. 1983. Exploring dimensions of access to medical care. Health Serv Res 18 (1~:49-74. Asplin, B. 2002. (Health Partners). 29 April 2002. Personal communication to Nicole Lurie. Ayanian, J. Z., P. D. Cleary, J. S. Weissman, and A. M. Epstein. 1999. The effect of patients' preferences on racial differences in access to renal transplantation. N Engl J. Med 341 (22~: 1 66 1 -69. Baker, D. W., C. D. Stevens, and R. H. Brook. 1991. Patients who leave a public hospital emergency department without being seen by a physician: causes and consequences. JAMA 266 (8~: 1085-90. Bashshur, R., D. G. Smith, and R. A. Stiles. 1993. Defining underinsurance: a conceptual framework for policy and empirical analysis. Med Care Rev 50 (2~:199-218. Berk, M. and C. L. Schur. 1998. Measuring access to care: improved information for policymakers. Health Aff 17 ( 1~: 1 80-86. Billings, J. (New York University). 11 April 2002. Personal communication to Nicole Lurie. Bindman, A. B., K. Grumbach, D. Osmond, M. Komaromy, K. Vranizan, N. Lurie, J. Billings, and A. Stewart. 1995. Preventable hospitalizations and access to health care. JAMA 274 (4~:305-11. Brach, C. and I. Fraser. 2000. Can cultural competency reduce racial and ethnic health disparities? A review and conceptual model. Med Care Res Rev 57 (1~:181-217.
144 GUIDANCE FOR THE NATIONAL HEALTH CARE DISPARITIES REPORT Burstin, H. 2001. Double jeopardy: the impact of race/ethnicity and health insurance. Paper presented at the Society for General Internal Medicine, 24th Annual Meeting. California Pan-Ethnic Health Network (CPEHN). 2001. "Diverse patients, disparate experience: the use of standardized patient satisfaction surveys in assessing the cultural competence of health care organizations." Online. Available at http://www.cpehn.org/pdfs/diverse_patients.pdf Accessed June 13, 20023. CDC. 2000. "CDC FY2000 Performance Plan - XV. Eliminating Racial and Ethnic Disparities." Online. Available at http://www.cdc.gov/od/perfplan/2000xv.htm [accessed June 10, 2002~. . 2002. "Centers for Disease Control and Prevention; Division of HIV/AIDS Prevention." Online. Available at www.cdc.gov/hiv/pubs/facts.htm [accessed Apr. 25, 20023. Commonwealth Fund. 2000. "Health Insurance, the Uninsured." Online. Available at http://www.cmwf.org/uninsured°/OSFoverview.html [accessed Mar. 1, 2002~. . 2002. Diverse communities, common concerns: accessing health care quality for minority Americans, Report #523. Cooper-Patrick, L., J. J. Gallo, J. J. Gonzales, H. T. Vu, N. R. Powe, C. Nelson, and D. E. Ford. 1999. Race, gender, and partnership in the patient-physician relationship. JAMA 282 (6~:583-89. Davey-Smith, G., C. Hart, G. Watt, D. Hole, and V. Hawthorne. 1998. Individual social class, area-based deprivation, cardiovascular disease risk factors, and mortality: the Renfrew and Paisley study. Journal of Epidemiology Community Health 52:399-405. DHHS. 1985. Report of the Secretary's Task Force on Black and Minority Health. Washington DC: DHHS. . 2002. Healthy People 2010. Washington DC: DHHS. Docteur, E. R., D. C. Colby, and M. Gold. 1996. Shifting the paradigm: monitoring access in Medicare managed care. Health Care Financ Rev 17 (4~:5-21.
DISPARITIES IN ACCESS TO CARE Ettner, S. 1999. The relationship between continuity of care and the health behaviors of patients. Med Care 37~6~:547-55. 145 Families USA. 2001. Getting Less Care: The Uninsured with Chronic Health Conditions. Washington, DC: Families USA. Fiscella, K. 2002. Using existing measures to monitor minority healthcare quality. In Improving Healthcare Quality for Minority Patients. National Quality Forum. Washington DC: National Forum for Health Care Quality Measurement and Reporting. Gornick, M. 2000. Vulnerable Populations and Medicare Services; Why Do Disparities Exist. New York: The Century Foundation Press. Iguchi, M. Y., J. A. London, N. G. Forge, L. Hickman, and K. Reihman. (in press). Elements of well-being affected by criminalizing the drug user: an overview. Public Health Rep. IOM. 1993. Access to Health Care in America. Washington DC: National Academy Press. . 2000. America 's Health Care Safety Net. Washington DC: National Academy Press. . 2001 a. Coverage Matters: Insurance and Health Care. Washington DC: National Academy Press. . 2001b. Crossing the Quality Chasm. Washington DC: National Academy Press. . 2001c. Envisioning the National Health Care Quality Report. Ed. Hurtado, M, E. Swift, and J. Corrigan. Washington DC: National Academy Press. . 2002a. Care Without Coverage. Too Little, Too Late. Washington DC: National Academy Press. . 2002b. Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care. Ed. Smedley, B., A. Stith, A. Nelson. Washington DC: National Academy Press. Keppel, K. (National Center for Health Statistics). May 2002. Personal communication to Nicole Lurie.
146 GUIDANCE FOR THE NATIONAE HEALTH CARE DISPARITIES REPORT Komaromy, M., N. Lurie, and A. Bindman. 1995. California physicians' willingness to care for the poor. West JMed 162 (2~:127-32. Lewis, G. (Office of Minority Health). August 2001. Personal communication to Nicole Lurie. Lurie, N., C. L. Zhan, J. Sangl, and A. Bierman. 2002. Racial and Ethnic Differences in Consumer Assessments of Health Care: Evidence from Commercial and Medicare Populations. Unpublished paper. Lynn, J. and L. Shugarman. 2002. Closing the Gap: The Attenuation of Disparities in End of Life Care. Data presented for the Medicare Claims End of Life Work Group, New York City. McGinnis, J. M., P. Williams-Russo, and J. R. Knickman. 2002. The case for more active policy attention to health promotion. Health AM (2~:78-93. National Archives and Records Administration (NARA). 1997. "President's Initiative on Race: Building One America for the 21st Century." Online. Available at ht~q?://clinton3.nara.gov/Initiatives/OneAmerica/america.htinl [accessed June 2, 20024. National Center for Health Statistics. 2001. "NCHS-publications and information products - health, United States, 2001." Online. Available at ht~q?://www.cdc.gov/nchs/hus.htm Accessed Apr. 10, 20023. National Institute of Dental and Craniofacial Research. 2000. Oral Health in America: A Report of the Surgeon General. Washington. Washington DC: National Institute of Dental and Craniofacial Research, National Institutes of Health. National Registry of Myocardial Infarction. 2002. Online. Available at www.nrmi.org [accessed Apr. 25, 2002~. President's Commission. 1983. Securing Access to Health Care: The Ethical Implications of Differences in the Availability of Health Services. Vol. 1. Washington DC: President's Commission for the Study of Ethical Problems in Medicine and Biomedicine and Behavioral Science Research.
DISPARITIES IN ACCESS TO CARE 147 Saha, S., M. Komaromy, T. D. Koepsell, and A. B. Bindman. 1999. Patient- physician racial concordance and the perceived quality and use of health care. Arch Intern Med 159 (9~:997-1004. Saha, S., S. H. Taggart, M. Komaromy, and A. B. Bindman. 2000. Do patients choose physicians of their own race? Health A~19 (4~:76-83. Schneider, E. C., A. M. Zaslavsky, and A. M. Epstein. 2001. Racial disparities in the quality of care for enrollees in Medicare managed care. JAMA 287 (10~: 1506-1509. U.S. Bureau of the Census. 1996. Population Projections of the United States by Age, Sex, Race and Hispanic Origin: 1995 to 2050. Washington, DC: U.S. Government Printing Office. . 2002a. "Census 2000 Gateway." Online. Available at htip://www.census.gov/main/www/cen2000.htrnl [accessed Apr. 25, 2002]. U.S. Bureau of the Census. 2002b. "Current Population Survey." Online. Available at www.bls.census.gov/cps/cpsrnain.htm Accessed Apr. 26, 2002~. Wagner, E. H., B. T. Austin, C. Davis, M. Hindmarsh, J. Schaefer, and A. Bonomi. 2001. Improving chronic illness care: translating evidence into action. Health AM 20 (6~:64-78. Williams, R. 1999. Race, SES and health: the added effects of racism and discrimination. Ann N YAcad Sci 896:173-88. Youdelman, M. 2001. Racial and ethnic data collection: a review of state laws. National Health Law Project. Interim Report.