Developing Indicators of Access to Care: The Case for HIV Disease
Vincent Mor 1
Since the first case was diagnosed in 1981, acquired immune deficiency syndrome, or AIDS, has become a leading cause of death among men age 25 to 44; more than 100,000 persons died of AIDS in 1981–1990 (CDC, 1990). The most recent Centers for Disease Control (CDC) projections are that as many as 153,000 persons were living with AIDS at the end of 1991 and that approximately 1 million persons are infected with the human immunodeficiency virus (HIV; CDC, 1991). Cumulatively, 275,000 cases of AIDS will have been reported by the end of 1991. Initially, the concentration of AIDS cases was in a few urban areas on the East and West coasts, but the incidence rate has been rising in communities outside the original epicenters, more rapidly in some risk groups than in the first-line cities.
The epidemic was first associated with homesexual contact, but HIV transmission through injection drug practice has been increasing in urban centers all along the East coast and even in the rural South. Linked to, but independent of, the rise in drug-related transmission is the incidence of AIDS among women and their children, which is expected to increase by about one-third per year. The incidence rate among bisexual and homosexual males in New York City, San Francisco, and Los Angeles who do not use drugs has actually been flat over the past several years. (Incidence is rising in this group in other areas of the country, however.) Given these
changing rates, the concentration of this epidemic will continue to increase in the nation's poor, drug-using, and minority populations.
The availability of effective curative and prophylactic treatments such as zidovudine (AZT) and pentamadine has provided a clinical rationale for early diagnosis (Fischl et al., 1987; Volberding et al., 1990). Life expectancy for those diagnosed with AIDS has been increasing over the first decade of the epidemic, even controlling for the changing pattern of diagnosed diseases that constitute the case definition of AIDS (CDC, 1990; Harris, 1990; Lemp et al., 1990; Piette et al., in press). Earlier diagnosis, longer survival, and extended years of medical treatment all imply that AIDS has become a chronic disease requiring a complex mix of health and social services that must be modulated as a patient's disease status and treatment affect both physiological and social functioning (Benjamin, 1988; Mechanic and Aiken, 1989).
Since the early days of the epidemic, the technology for treating and managing the multiplicity of opportunistic infections in AIDS has changed rapidly (Cohen et al., 1990). Variation in the length of hospital stays has also been repeatedly noted by region and provider (Andrulis et al., 1989; Kaplowitz et al., 1988; Seage et al., 1990). Recent studies have shown that experience with AIDS management has a positive impact on survival and that more experienced facilities appear to make more effective use of additional resources, compared with those with less experience (Bennett et al., 1990). Accompanying the change in clinical practice has been a reduction in the duration of hospital stays (Seage et al., 1990) and an increase in the use of outpatient and home treatment with ''high-tech" nursing care (e.g., home intravenous units, infusion and parenteral nutrition).
Because of both the rapid impoverishment of formerly employed persons who lose their private health insurance and the increasing prevalence of AIDS among the previously poor and uninsured, urban municipal and not-for-profit hospitals are disproportionately paying the price of the AIDS epidemic (Andrulis, 1989; Baily et al., 1990; Green and Arno, 1990). Since many of these institutions are facing other pressures, such as homelessness and the medical consequences of drug abuse, the unique pressures of AIDS only serve to complicate their ability to address an already complex mix of social problems.
Public reaction to the AIDS epidemic has been volatile and generally negative (Blendon and Donelan, 1988). Each new incident of transmission that breaks prior stereotypes is greeted with fear and hysteria in certain sectors. (The recent case of HIV transmission from a dentist to his patients provides a graphic example of this phenomenon.) At the same time, a substantial minority blame the victims for the behavioral transgressions that "caused" their condition (Blendon and Donelan, 1988). This prejudice isolates infected individuals and leads to job discrimination, denial of insurance benefits, and reduced access to personal health care.
With this background, HIV disease can serve as a sensitive indicator of the degree of access to personal health care services in the United States. However, relatively little is known about the reduced access experienced by this population, how to measure it, or what it means. Further conceptualization of these issues is required before specific indicators of reduced access can be suggested with any confidence.
Following the model proposed by the IOM Committee for Monitoring Access to Personal Health Services in Chapter 2, this paper reviews the issues and existing knowledge about the presence of barriers to access to health services that confront persons with HIV disease. The consequences for persons with HIV disease are also examined by measuring variations in utilization of health services. This review relies not only on the relatively sparse published literature but also on quantitative data and qualitative insights derived from an ongoing study examining the organization and delivery of health and social services to persons with AIDS in various cities around the country. This study is described briefly below, after which the relevant literature is reviewed and new data on these issues are presented. The last section of this paper proposes a series of indicators of barriers to access and the types of data systems necessary to monitor them.
ROBERT WOOD JOHNSON FOUNDATION AIDS HEALTH SERVICES PROGRAM AND EVALUATION
Between 1986 and 1990 the Robert Wood Johnson Foundation (RWJF) funded nine projects (AIDS Health Services Programs [AHSP]) in 11 cities to develop and coordinate specialized health and supportive community services for persons with disabling HIV disease (Mor et al., 1989). The emphasis in these projects was on developing networks of community-based care providers to offer new services and to coordinate the delivery of existing services. The central organizational features of the programs were a formal consortium of participating agencies and individualized case management. Program evaluation was conducted by Brown University's Center for Gerontology and Health Care Research, under the direction of this author. The evaluation integrates quantitative data obtained from computerized program "intake" records and two longitudinal cohorts of program clients. Clients were surveyed about their experience in obtaining health services and qualitative information was obtained from four rounds of site visits and a detailed review of program progress reports, correspondence, and budgets.
Several papers based on this evaluation have already appeared or are in press (Capilouto et al., 1991; Fleishman, 1990; Fleishman et al., 1989, 1990, 1991; Mor et al., 1989, 1992; Piette et al., 1990, in press; Stein et al., 1991). The major evaluation issues pertaining to the success of consortia building and case management as vehicles for service integration are still
being examined. The project information presented in succeeding sections represents both published and unpublished data. The more qualitative information is based largely on case studies; the quantitative information clearly is limited by its focus on program clients rather than on the experience of a broad cross section of persons with HIV disease in each city. Nonetheless, because this data source contains specific, detailed information relevant to the access barriers that confront those with HIV disease, it is useful as an illustrative device.
ISSUES IN ASSESSING ACCESS TO PERSONAL HEALTH CARE
Equitable access to health care services has been a major focus of health services research in the United States for four decades (Ginzberg, 1990). Research in the past two decades has emphasized factors associated with the rising costs of health care; now, as a result of growth in the number of uninsured over the past decade, access to care has once again emerged as a central topic in health services research. This section reviews the various barriers to use of services confronting persons with HIV disease. In keeping with the model proposed by the IOM committee, barriers have been classified as financial, structural, personal, and attitudinal. Each is defined in relation to the special issues facing persons with HIV disease.
Financial barriers to access to health care services include insurance coverage, provider reimbursement rates, and lack of investment in resources designated for the treatment of HIV disease. Other financial barriers can also be noted, such as out-of-pocket expenses for patients, the high cost of treatment, and the substantial indirect costs of the disease to the patient and society. These barriers are either covered under one of the topics mentioned above or are beyond the scope of this paper.
The absence of insurance coverage is a barrier to health care services regardless of a person's medical problems. In the case of HIV disease, however, the poverty and/or limited work history and savings of the population at risk, as well as the increasingly prolonged nature of the disease, mean that those infected are likely to endure limited access over a long period of time. Because the absence of private compared with public insurance may have a different, and potentially sequential, impact on the person with HIV disease, these two issues are treated separately.
Private insurance is almost always tied to employment, because the purchase of individual health insurance coverage is quite expensive and replete with coverage restrictions (Eby, 1989). Loss of employment traditionally has signaled loss of insurance. With passage of the Comprehensive Omnibus Budget Reconciliation Act of 1988, however, individuals have the right to continue purchasing health insurance from their previous employer, paying all the premiums themselves, for up to 18 months. Anecdotal evidence from many sources suggests that the loss of employment income, ongoing living expenses, and increased out-of-pocket medical expenses for copayments make paying these insurance premiums impractical, even at the group rate. The recent Ryan White bill (P.L. 101–381) allows states to pay these premiums. No data are available to estimate the proportion of formerly employed persons with HIV disease who are paying their own health insurance premiums.
There is also a dearth of data by which to quantify the rate at which loss of private insurance occurs following either onset of symptoms or formal AIDS diagnosis. Yelin and his colleagues (1991) report on the time between initial HIV symptom onset and work cessation in a cohort of 170 patients treated at San Francisco General Hospital. Three years after initial symptoms appeared, less than half of these patients were working; another 12 percent had reduced work loads.
Martin and colleagues surveyed 432 HIV-positive persons in Texas in 1988 (40 percent of distributed questionnaires were returned). The proportion of respondents with private insurance was only 41 percent (Martin et al., 1989). Kass (1989), comparing participants in the Multicenter AIDS Cohort Studies (MACS) with leukemia patients, reported that people with AIDS were more likely to be uninsured and more likely to have been turned down for insurance.
Based on intake records completed for all clients of the RWJF AHSPs, only 21.3 percent of some 14,000 clients were employed full-time at their entry into the program, and only half of these had private insurance. This result is consistent with the finding that many of the uninsured in the United States are employed. Half of those who were unemployed when they entered the program had no health insurance at all, either private or public.
Although it is likely that only a minority of all individuals with significant symptomatic HIV disease have private medical insurance, the impact of AIDS on specific private insurers is not insignificant. MetLife reviewed health claims for 1986–1989. In 1989, 6,450 people received group medical claim payments for AIDS-related diseases for a total of $111.2 million (Pickett et al., 1990). This figure constitutes almost a 300 percent increase over the AIDS-related claims incurred by the company in 1986.
Little evidence exists about how the insurance coverage of persons with HIV changes with disease progression. Kaufman and his colleagues
(1990) linked hospital discharge records in the state of New York between 1984 and 1986 and found that 17 percent of the patients whose status on first admission was as a private payer had been changed to government reimbursement by their last indicated admission. Unfortunately, the authors do not note the proportion of patients who had no insurance coverage at all. The Texas sample surveyed by Martin and colleagues reported that 60 percent were privately insured at initial HIV diagnosis and 41 percent were insured at the time of the survey (Martin et al., 1989). Interviews with a nonrandomized sample of 1,386 program clients in 9 of the 11 cities funded by the RWJF revealed that 44 percent reported having had private insurance at the time that they learned they had HIV disease. Of those, 37 percent had lost their insurance within three to six months.
Access to personal health care services appears to be strongly linked to the availability of private insurance coverage. Zucconi and her colleagues surveyed HIV-positive men and found that the number of reported physician visits was strongly associated with being insured (Zucconi et al., 1989). those RWJF survey sample clients with private insurance reported significantly more physician and clinic visits than those without insurance and were significantly less likely to use an emergency room (Mor et al., 1992). However, in neither that study nor one by Seage and colleagues (1990) was there a relationship between having private insurance and the probability of being hospitalized.
Although loss of insurance may be the longitudinal experience of individuals, changes in the insurance mix of the HIV population will occur as the composition of the population shifts and as early identification and treatment of HIV disease become more commonplace. The growth of public insurance coverage for medical treatment of HIV disease undoubtedly will increase with the changing population (Green and Arno, 1990); in addition, the liability of private insurers may grow to cover the early treatment expenses of those with HIV disease who are still employed. Given the promise of early intervention efforts among those with asymptomatic HIV disease, this is an issue of importance for the near future (Arno et al., 1990). Considerable concern has been voiced about the private insurance industry's willingness to continue to provide coverage to persons with high-cost conditions such as HIV (Lipson, 1988; Parmet, 1987); however, the empirical documentation of instances of coverage rejection, withdrawals, and limitations is difficult. Kass (1989) reported that 11 percent of AIDS cases had been turned down for health insurance. As one part of the RWJF program evaluation, 355 clients in three cities were asked about their experience with private insurance. Of those who had health insurance at the time of HIV diagnosis, 40 percent subsequently lost it. For most (23 out of 54 persons), this occurred because they were no longer employed; the rest
lost their coverage because their policy was terminated or they were no longer able to pay the premiums.
Denial of coverage on the basis of retrospective interpretations of pre-HIV diagnosis claims and the widespread adoption of HIV testing as a precondition for health and life insurance applications are strategies being used by the insurance industry to limit liability (IOM, 1986). Indeed, a 1987 survey of state laws revealed that only eight states forbid HIV testing by insurers (Faden and Kass, 1988). Moreover, the growing cost of health insurance premiums for employers may make companies increasingly willing to adopt "carve-out" and prior-condition exemption practices to limit the perceived catastrophic effect of high-cost AIDS patients in their group.
Public insurance is playing an ever-increasing role as a payer for health and social services for persons with AIDS (Green and Arno, 1990). Between 1983 and 1987 in New York City, Los Angeles, and San Francisco, the percentage of hospitalizations of AIDS patients financed by Medicaid increased between 50 percent and 100 percent, whereas the percentage financed by private insurance dropped between 25 percent and 50 percent. When compared with non-AIDS cases, costs for AIDS admissions were 50 percent to 275 percent more likely to have been paid by Medicaid than by private insurance. According to the U.S. AIDS Hospital Survey (Andrulis et al., 1989), there is considerable regional variation in the payer source mix for hospitalizations. In 1987, among public hospitals in the East, 71 percent of admissions were paid by Medicaid; it paid for only 18 percent of admissions to public hospitals in the South. Private hospitals are also less likely to serve Medicaid patients in all regions but the South. In contrast, nearly half of all outpatient visits by AIDS patients to private hospitals were covered by Medicaid in virtually all regions. Public hospitals, particularly in the South and West, were much more likely than private hospitals to incur "bad debt" or to use local financing to offer free outpatient clinic care to AIDS patients.
Green and Arno (1990) found that in Los Angeles, almost all (98.9 percent) privately insured AIDS admissions occurred in private hospitals, whereas only 42 percent of Medicaid admissions went to private hospitals. Similar patterns were observed in the other cities studied. The authors remark that "the differences are so large … as to suggest nearly distinct systems of care depending upon type of insurance." Site visit experience in the 11 RWJF AHSP demonstration communities reinforced this notion of a bifurcated system.
Regional variation in reliance on Medicaid is due to the enormous variation among the states in eligibility policies. Unless a state has a categorical eligibility program (e.g., Aid to Families with Dependent Children, or AFDC) or a program for the medically needy, individuals must meet state
income and asset tests before they are eligible to receive Medicaid. And even the presence of a medically needy program does not guarantee continuous coverage under Medicaid in most states. The income test in some is below 50 percent of the poverty line, meaning that individuals who have become presumptively eligible for Social Security Disability Income (SSDI) may receive just enough money to make them ineligible for Medicaid. States like Texas and Louisiana have very low-income limits for Medicaid.
The services covered under Medicaid also vary considerably by state. Some, like Louisiana, arbitrarily limit the number of inpatient care days as well as the number of physician and home health visits covered by Medicaid. Almost all Medicaid programs have updated their pharmacy formulary to include treatment drugs such as AZT and pentamadine (Buchanan, 1988), although in some states there is a perception of limited public supply.
Some states have chosen to expand the range of services that can be provided to Medicaid recipients (e.g., home care, case management, homemaker, transportation, and home medical visits) by requesting "waivers" from the Health Care Financing Administration (HCFA). HCFA approval of a waiver depends on a state's demonstration of how the waiver will produce savings. Expanding home services as a substitute for more expensive inpatient care is almost always the rationale. This means that publicly insured persons with AIDS in waivered states will have access to a broader, more diverse array of health care services than will be the case for all but the most fortunate patients with private insurance. New Jersey has had waivers in place since 1987; other states, however, have such limited Medicaid coverage that it is virtually impossible to demonstrate how the waivers would be cost neutral, much less produce savings. Even when a case can be made for cost savings, when properly evaluated these programs almost always result in higher costs.
The time lapse between loss of private coverage and eligibility for public insurance may be prolonged. The devastating personal effects of this lapse may result in impoverishment, since home ownership (the one asset spared by Medicaid spend-down rules) is not the norm in this population. (Only 10.8 percent of respondents to the RWJF survey owned their own homes.) In examining respondent satisfaction with the health care they received, the RWJF survey found that those who previously had private insurance but were no longer covered reported significantly lower satisfaction with their access to health care. During the interim, from the time people lose their private insurance to the time they become eligible for public coverage, there is some evidence that outpatient and physician visits are less frequent (Mor et al., 1989; Zucconi et al., 1989). If these individuals require hospitalization, the hospitals that accept them (presumably public facilities) incur added bad debt.
Provider Reimbursement Rates
Provider reimbursement levels also may affect access to the personal health care system. It is well documented that patients with Medicaid as their primary source of payment for health care have a more difficult time finding a private physician to treat them (Holahan, 1984; Perloff et al., 1987). A major reason is that Medicaid reimbursement rates tend to be substantially lower than customary charges and reimbursements from other payers. If a provider receives lower-than-average levels of reimbursement for treating Medicaid patients and AIDS patients cost more to treat than the average patient, the provider may be even more reluctant to treat AIDS patients. The implications of inadequate reimbursement for the AIDS patient may differ for hospitals, clinics, physicians, home health providers, and nursing homes. Consequently, they are discussed separately below.
The cost-to-charge ratios of hospitals reflect whether, for various types of services, their accounting cost for producing a service is higher or lower than what they charge for it. This cost in turn might differ from their reimbursement for the service by third-party payers. Andrulis and his colleagues (1989) surveyed U.S. hospitals and asked them to estimate their costs per inpatient day and per outpatient visit for treating AIDS patients. Across all regions, both public and private hospitals reported losing from $4 per patient day among private hospitals in the South to $386 per day among southern public hospitals. Reporting hospitals also noted that they were experiencing losses with non-AIDS medical/surgical patients, but these losses were substantially lower. Similar findings were reported for outpatient visits.
Hospitals that note discrepancies between costs incurred and reimbursements received for AIDS hospital admissions may attempt to minimize their exposure to financial risk by limiting the number of AIDS admissions, particularly those insured by Medicaid. Green and Arno (1990), as well as Andrulis and coworkers (1987a), have shown that Medicaid patients are underrepresented among AIDS admissions to private hospitals, compared with public hospitals.
The hospital outpatient clinic has become a central point of treatment for persons with HIV disease. Public hospitals have a long tradition of providing care to the poor and uninsured, and these clinics are training sites for medical schools and affiliated academic medical centers. Andrulis and colleagues (1987a) found an average of 161 HIV outpatients (median = 36) with an average of 1,460 visits per year in the 80 public hospitals they surveyed. On average, the 196 private hospitals surveyed served far fewer outpatients. Whether this differential is due to the diversion of high-cost, low-reimbursement patients to the public sector or to the fact that few private hospitals specialize in this care is not known at this time.
Access to physician services can also be compromised by the rate at which physicians are reimbursed relative to their "cost" for providing a particular service. Green and Arno (1990) revealed that in both New York City and San Francisco, Medicaid pays between 10 and 50 percent of what private insurance companies reimburse for the same service. For example, in New York, Medicaid pays only $11 for an office visit with a new patient, whereas a private insurer pays $84. In San Francisco, chemotherapy with infusion is reimbursed at $12 by Medicaid and at $28 by a private insurance carrier.
Conversations with medical personnel held during visits to the RWJF cities to educate the medical community about HIV treatment techniques revealed that, in many localities, only a small cadre of physicians care for persons with HIV disease. Rizzo and his colleagues analyzed responses of the 1988 American Medical Association (AMA) physician survey and found that about one-half of all physicians reported having treated at least one HIV-positive person (Rizzo et al., 1990). Although there is no hard evidence, many physicians who are willing to treat patients with HIV disease transfer them to the care of the public hospital clinic once a patient's private insurance lapses. They indicate that they do not treat patients without insurance and, given the low reimbursement rates, will retain ongoing continuity of care responsibility for only a small number of patients who become covered by Medicaid.
Nursing homes have not played a prominent role in meeting the health and social service needs of AIDS patients in most communities in the United States. Only 4 of 47 skilled nursing facilities surveyed in Oregon in 1989 had served an AIDS patient (White and Berger, 1991). Marder and Linsk (1990) documented that none of 42 hospitalized persons with AIDS who required discharge arrangements were placed in a nursing home. Similar experiences have been reported by hospital discharge planners and case managers in the cities funded by the RWJF. With certain exceptions (e.g., publicly operated county facilities), the nursing home industry has not served persons with HIV disease.
Although there are many reasons for this phenomenon, one that is frequently mentioned is inadequate reimbursement. Recently, several states have instituted "enriched" Medicaid reimbursement for AIDS patients in nursing facilities. For example, Florida providers receive a supplemental fee of nearly $75 per diem for serving AIDS patients—almost double what the state pays for a patient without AIDS. Nonetheless, as recently as the winter of 1991, only one nongovernmental facility in south Florida admitted persons with AIDS. Providers in locations as disparate as Minnesota, Florida, and California all maintain that it costs $200 a day to care for an AIDS patient in a nursing home, and in California no supplement is available. Swan and his colleagues, reporting the results of a time and motion study of
93 persons with AIDS in California nursing homes, found that they required 6.5 hours of nursing time per day, which is a much higher rate than that for traditional nursing home patients (Swan et al., 1990). Providers say that the specialized reimbursement is still inadequate.
New York is developing AIDS residential health care facilities (RHCF) to provide subacute care, rehabilitative care, active nursing treatment, palliative care, and case management services. The state has approved more than 200 beds, and some facilities have been established in New York City and Rochester. These facilities, as well as specialized AIDS health-related facilities (HRF) or nursing homes, are to be paid a substantially higher rate than facilities serving other nursing home patients, both because they entail a new class of licensed facility and to encourage providers to offer this service for AIDS patients. Presumably, this effort will reduce the access barriers facing those with advanced HIV disease in New York.
In view of the major emphasis on community-based care for persons with HIV disease, the role of home health agencies (HHA) is critical. The recent study in Oregon by White and Berger (1991) found that 13 of 16 agencies located in areas with a relatively high prevalence of HIV had served persons with AIDS. Fleishman and Masterson-Allen (1991)surveyed 263 HHAs in 11 cities with high rates of HIV infection and found that 76 percent of these agencies had served patients with HIV, although only 57 percent had served an AIDS patient in the past three months. Smaller, proprietary agencies and those without Medicaid certification were less likely than other agencies to have ever served an HIV-infected patient. However, in areas with many HHAs, it appears that a higher proportion of agencies able to provide intensive home nursing are willing and anxious to do so.
HHAs noted that reimbursement was a major barrier to serving AIDS patients (Fleishman and Masterson-Allen, 1991). A study by the Home Care Association of New York determined that AIDS patients "need 13 percent more nursing time than terminal cancer patients and 29 percent more than the average home care patient" (DeHovitz, 1990). Based on this study, the average nursing visit rate for HIV-infected patients in New York was increased by 30 percent in 1988.
An obvious barrier to receiving home care or home hospice benefits is the lack of housing. Numerous studies have documented the difficulty of discharging AIDS patients from the hospital because of a lack of housing alternatives. Fleishman and Masterson-Allen (1991) also found that home health agencies reported that the greatest difficulty associated with serving AIDS patients was the absence of a stable home environment. According to their respondents, this problem was compounded by the fact that in many communities those with HIV disease lived in the least safe neighborhoods, thereby requiring personal guards or other costly strategies to ensure the safety of nurses and aides.
It is precisely this higher cost of caring for persons with AIDS that has made hospices anxious about assuming the liability for all medical care of persons with AIDS who elect to be served by a hospice. Since 1989, coverage of hospice services by Medicaid has been mandated in all states: hospices can now receive standard, prospectively set reimbursement for home and inpatient care. The hospice is responsible, however, for all medical services consumed including hospitalization, drugs, and infusion therapies. Because AIDS treatment standards are evolving rapidly, optimal strategies for symptom relief might require hospitalization or continuous infusion at home, or both; these services are extraordinarily costly even if persons with AIDS elect palliation and no aggressive care. It is little wonder that hospices are ambivalent about enrolling patients with AIDS as hospice beneficiaries and, depending on state reimbursement levels, may choose to provide similar care as an HHA rather than a hospice (personal communications to the author during site visits in Florida and Louisiana).
Investment in Resources for Treating HIV Disease
Investment in treatment resources by the state, county, or local municipality has clearly had an effect on access to personal health care for persons with HIV disease across the country. Rowe and Ryan (1988) catalogued state expenditures for AIDS between 1986 and 1988. Florida, California, New York, and New Jersey were among the highest spenders. Direct expenditures for patient care climbed from 4 percent to 19 percent of the total between 1986 and 1988, largely because of the added emphasis on direct care in New York. In general, however, most of the state expenditures were directed toward education, testing, and counseling.
The most obvious examples of local investment are the public-sector hospitals. Some have their own tax base in the county or metropolitan area property tax. The budgets of others may derive from direct line-item allocations by the state legislature. In Louisiana, for example, Charity Hospital in New Orleans was designated a statewide referral center for AIDS treatment, education, and training. In this particular instance, given Louisiana's limited Medicaid program, the concentrated state investment in the Charity Hospital AIDS program has enhanced access to high-quality outpatient care far beyond what would have been possible if the hospital were relying only on Medicaid reimbursement.
Another example is the transformation in West Palm Beach County of hospital districts with their own taxing authorities into a countywide ''health district" with spending authority to cover not only hospital care but home and community care as well. Although the referendum covered all aspects of health care for the poor and uninsured (there are no public hospitals in the county), the first beneficiaries were the many uninsured patients with
HIV disease and the public clinics and programs serving that population. Providers benefit from guaranteed funding of AIDS patients care, and patients benefit from increased access and choice.
Structural barriers to access to the personal health care system have traditionally included lack of a primary source of care, lack of appropriate service providers in one's area, distance from providers, and extended waiting time for providers. In each case, these structural barriers theoretically detract from the patient's ability and willingness to obtain, and then adhere to, appropriate medical care.
Lack of a Primary Source of Care
Having a regular source of ambulatory care has traditionally been viewed as a sine qua non for access to medical care. Without a regular source of care, it is argued, individuals will not have the benefits of preventive care and periodic checkups. When they do become ill, there will be no one who is sufficiently knowledgeable to make informed clinical decisions about their treatment. National surveys have consistently shown that the 15 percent of the population without a regular source of care are less likely to be insured and to have had recommended screening exams (Hayward et al., 1991).
Although in the general population the proportion of people without a regular source of care is only 15 percent, this figure may be higher among those with HIV disease. In the general population, the groups most likely not to have a regular source of care, men between 18 and 45 and those without insurance, are precisely the groups with HIV disease. An added difficulty in trying to estimate the proportion of HIV-infected people without regular care is the lack of comparability between populations. Because HIV disproportionately strikes those who are underrepresented in national telephone surveys (the method used to ascertain those with a regular source of care), it is improper to extrapolate these national estimates to the HIV population.
Several major issues complicate an understanding of what is and is not a "usual source of care" for persons with HIV disease: (1) the nature of the relationship between nonmedical treatment systems and medical care providers; (2) the role of the hospital clinic and whether, with its rotating staff, it provides the continuity generally associated with a primary care physician; (3) the role of disease progression in stimulating a shift in primary care responsibility from the generalist to the specialist; and (4) the implications of using multiple physicians in the ongoing care of a patient.
Persons with HIV disease who are active intravenous drug abusers form the group that is least likely to have a usual source of care (Mor et al., 1992). This finding raises concerns about how the drug treatment and medical care systems interact. The impact of drug use on the health care system has been well documented and appears to be growing, particularly in New York (Myers et al., 1990). The number of drug- and alcohol-related hospital discharges throughout the state increased 1 percent per year. In New York City, discharges, patient days, and length of stay for those with substance abuse diagnoses have been increasing at rates of more than 10 percent a year. In 1987, in New York City, "one out of every seven patient days for AIDS care was devoted to patients with both AIDS and substance abuse diagnoses" (Myers et al., 1990) and it is this group of individuals with dual diagnoses that is increasing most rapidly.
Little information is available to indicate where these individuals obtain ambulatory medical care, although many observers believe that they rely on hospital emergency rooms. Drug treatment programs, whether residential or outpatient (and with or without methadone maintenance) have tended not to be connected to medical treatment systems, despite the high risk of medical complications associated with illicit drug use. The state of New York has begun to recognize this failing, given the growing HIV seroprevalence rate among intravenous drug users, and has instituted various demonstration projects to open "multi-modality treatment centers which would include outpatient, inpatient, enhanced methadone services, AIDS services, central intake," and other services, all provided in one setting. In addition, the existing substance abuse network is being linked to primary health and mental health services. Whether this approach will reduce the rate of hospitalizations that occur through the emergency room among Medicaid-funded, HIV-infected persons in New York City remains to be seen.
The hospital clinic is often the only option available for persons with AIDS. (However, few data are available concerning the relative proportion of persons with HIV disease served by hospital clinics, compared with private physicians' offices.) Among 408 respondents to the RWJF AHSP evaluation who were clients of a community-based social service agency, only 38 percent reported that their usual source of care was a private physician or health maintenance organizations, whereas 44 percent named the clinic at the local public hospital. Among those identified in the RWJF public hospital clinic sample, almost all (95 percent) named the clinic as their usual source of care.
The public hospital clinics in RWJF communities have evolved from being merely infectious disease specialty clinics into continuity of care clinics. In some, a concerted effort is made to establish primary care teams that follow a panel of patients. Often, these clinics are staffed by the most experienced physicians treating HIV in that community, a strong incentive
to receive care there even among those individuals who have private health insurance.
Based on site visits to and telephone contact with clinics in non-RWJF cities, it appears that the transformation of traditional public hospital medical clinics into primary care programs has occurred in other cities as well. The clinic environment remains complex, since many patients are followed by special research clinics that may or may not involve the same medical staff as in the regular clinic. Yet this problem of multiplicity of providers complicated by participation in research protocols is not unique to the public clinic patient population. Indeed, being treated in multiple care settings may be more burdensome for the private patient, particularly as this inevitably involves travel from one location to another and not just shifting from clinic to clinic in the hospital. Using a public hospital clinic rather than a private physician as a regular source of care does not necessarily signify poorer quality or less continuity.
These growing "islands of excellence" in the public hospital system will, however, come under increasing pressure with early identification of patients for prophylactic therapy. This swelling group, many of whom are uninsured, cannot reasonably be accommodated by the existing capacity of public hospital-based clinics. Staff in the RWJF study found that time spent by newly diagnosed, asymptomatic individuals on a waiting list in a community often is six months or longer. In some locations, health planners are making a concerted effort to arrange a system of linkages between publicly supported community health centers and the public hospital HIV specialty centers. In some areas, these locations are linked to existing sexually transmitted disease clinics; in other areas, HIV-infected individuals are integrated into community health centers.
Lack of Appropriate Service Providers
It is still unclear what pattern of patient transfer from generalist to specialist is most appropriate in the case of HIV disease. (This level of uncertainty is not surprising; there are few standards for this process for any chronic disease.) The process of referral and transfer is complicated by insurance coverage, negative attitudes toward HIV-infected patients, and the varied expertise of the physicians from many different backgrounds who specialize in AIDS care. It is possible that HIV disease may become a subspecialty within medicine. Indeed, there are already examples of divisions of HIV, separate from divisions of infectious disease, emerging within departments of medicine at major medical schools.
A related issue pertains to the implications for patients of receiving care from many different physicians. Theoretically, the rationale for a single, regular source of care is that this provider will be most knowledgeable
about the patient's condition, values, and preferences. In the event that specialty care is required, the primary source of care should be able to coordinate it and to integrate the results of multiple consultations. Unfortunately, there is little empirical evidence to substantiate the success of this model for any chronic disease, much less HIV disease. Consequently, no data suggest that seeing several physicians is more or less appropriate than seeing one primary physician.
Data from the survey of RWJF AHSP participants whose regular source of care was the public hospital clinic revealed that 11 percent also saw a private physician and 9 percent reported a clinic visit to a different hospital. Among those with a private physician as their usual source of care, 16 percent reported also using the public hospital clinic. Across both groups of respondents, nearly 15 percent reported having used three or more different sources of care in the past four months, and only 43 percent reported using only one source. This pattern was no more likely for patients whose regular source of care was a public hospital clinic than it was for those whose usual source of care was a private physician.
Whether seeing multiple physicians in different locations actually leads to more diffuse care still needs to be investigated. Factors such as these will provide an interesting indicator of continuity of care, once the practice and its consequences are better understood.
Distance from care Providers
The distance patients with HIV disease must travel to obtain medical care may influence compliance with recommended medical regimens. As long as the epidemic remained within the confines of urban centers and their suburban environs, distance or travel time was not a major impediment unless the patient was seriously impaired or homebound. For example, the survey of RWJF program clients found that 70 percent of respondents traveled less than 30 minutes to reach the clinic or their physician, and only 14 percent traveled more than 45 minutes. Nonetheless, even this relatively short travel time can present insurmountable problems for some patients. Although only 32 percent of RWJF survey clients needed assistance with transportation, for most of that group (70 percent) this need was unmet. In fact, the lack of assistance with transportation was greatest among the most physically impaired. Thus, even though transportation problems affect only a minority of patients, when present, they constitute a major barrier.
Extended Waiting Times for Providers
Waiting time in public hospital clinics is traditionally long and a principal complaint among patients. Perception of waiting time is strongly correlated
with other measures of access to medical care such as satisfaction with access. Theoretically, perceptions about waiting time may affect compliance because patients may be more likely to miss scheduled appointments if they anticipate a long wait. The survey of RWJF program clients found that 35 percent routinely waited more than an hour for an appointment; only 36 percent reported routinely waiting less than half an hour. In response to the question, How would you rate your satisfaction with the length of time spent waiting?, more than 30 percent said "fair" or "poor," and another 30 percent said merely ''good." Generally, responses to such questions on satisfaction are heavily weighted toward the positive. The strongly negative ratings on this aspect of access clearly indicate that waiting time is a problem.
Public clinics have long waiting lists, which means that individuals seeking medical care will be triaged unless their condition is serious. Site visit discussions with public HIV clinics in RWJF demonstration and nondemonstration communities revealed that waits of six months or longer were standard for asymptomatic HIV-infected individuals seeking treatment and a source of primary care knowledgeable about HIV disease. Such extended waits can be seen as a real deterrent to care and presumably are far more characteristic of the public than of the private health care system.
The availability of providers who care for persons with HIV disease is not well quantified. Anecdotal evidence from conversations with staff from public hospitals and community-based organizations suggests that in most communities, only a handful of physicians are willing to treat HIV patients.
In 1985, one-half of the physicians not specializing in radiology or pathology who responded to the AMA socioeconomic monitoring system survey reported that they had treated at least one HIV-positive patient. This finding, however, provides little basis for estimating the proportion of primary care physicians who retain clinical management responsibility for these patients, as has been recommended (Northfelt et al., 1988; Rizzo et al., 1990). Indeed, the call has been sounded from various quarters for general internists to assume primary care responsibility for persons with HIV disease by engaging in clinical tasks as diverse as sexual counseling to managing the administration of retroviral agents (American College of Physicians, 1988; DeHovitz, 1990; Northfelt et al., 1988). These physicians may agree that there is an obligation to treat those with HIV, but the evidence in most communities suggests that this perceived obligation may not extend to retaining primary care responsibility for them. It is not known whether private primary care physicians cease to see their HIV patients because the patients have no insurance, because the physicians are afraid of the risk of contagion or of losing their other patients, or because they are uncertain about their clinical skills. However, the comment made repeatedly in each of the 15 communities studied in the RWJF evaluation is that only a handful
of private physicians are seeing the vast majority of persons with HIV disease. The public sector is similarly constrained.
A related issue is the availability of dental care for persons with HIV disease in many communities across the country. The topic of access to dental care has been hotly debated for several years. Thus, it is difficult to separate access barriers due to fear of spreading the disease from the low levels of access to dental care in this population (Capilouto et al., 1991; Neidle, 1989; Vercusio et al., 1989).
Despite these concerns for the future, the gap in availability of medical and nursing care was not given the highest priority among health and social service providers in RWJF-funded communities. More important, they felt, was the absence of housing and inadequate funding for home nursing, attendant services, and subacute long-term care facilities. Future demand for health services, and its potential for compromising access by outstripping the existing supply of physicians and clinic slots, will have to be carefully monitored and probably will vary considerably from community to community and by risk group.
This category of barriers includes those factors traditionally associated with differential access, regardless of disease or age—for example, education, ethnicity, or income. Historically, these factors have been shown to influence the behavior of providers in treating patients and the ability of patients to adhere to a prescribed treatment regimen. These personal characteristics tend to be correlated with socioeconomic status, making it difficult to disentangle the effects of personal characteristics from the effects of living in a particular neighborhood, the lack of insurance coverage, and reliance on public medical resources for medical care.
One clear indicator of access to care is whether infected individuals who should be receiving a given therapy are actually receiving it. Holmberg and colleagues (1990) found that as many as half of all AIDS patients were not receiving aerosolized pentamidine when it was the therapy of choice. In the RWJF AHSP evaluation sample, among those eligible to receive aerosolized pentamidine, men were four times as likely as women to have received it. This finding held true after controlling for disease duration, drug use, and insurance status, all of which were also significantly related to treatment (Piette et al., in press). Hidalgo and colleagues (1990) also found a surprisingly small number in the Maryland Medical or Pharmacy Assistance program on pentamidine. Findings on who received the drug were similar to those in the RWJF study: those more likely to have received the drug were gay, white men.
Another drug known to be of benefit to persons with HIV disease is
AZT (zidovudine). Since 1987, when it was shown to be effective in prolonging survival in patients with advanced disease, the federal and state governments have invested heavily in making this drug available to the infected population (Buchanan, 1988). But there is evidence that certain subgroups of the AIDS population are less likely than others to receive this treatment. Moore and colleagues (1991) studied the files of 714 adults in Maryland diagnosed with AIDS and found that women, minorities, and intravenous drug users were less likely to have received AZT than were gay, white men, even after controlling for ability to pay and access to a regular source of care at a clinic. Study data showed that 63 percent of white patients had received AZT, compared with only 43 percent of nonwhite patients. A similar pattern was observed for receipt of AZT among clinically eligible clients of the RWJF evaluation survey (Stein et al., 1991). After controlling for disease stage and past history of Pneumocystis carinii pneumonia, intravenous drug users, women, and minorities were significantly less likely to have received AZT than whites, males, and those with insurance. These effects remained as strong when the analyses were restricted to patients who receive their medical care from public hospital clinics.
Other utilization-based indicators of access differentials related to personal background have not been examined carefully for persons with HIV disease. The RWJF evaluation survey found that among persons served by the public hospital clinic, whites, those with more education, and nondrug users reported more frequent clinic visits in the past three months than the average patient, even after controlling for disease advance, symptom severity, functional status, and living arrangements (Mor et al., 1992). The opposite set of relationships was observed when the dependent variable was the likelihood of using an emergency room; use was more likely among minorities, women, and drug users.
Whether these differences signify reduced access to care must be carefully considered, because a greater number of visits does not necessarily translate into higher quality of care. Indeed, observed utilization differences do not imply provider bias; intravenous drug users may not have been socialized to, or have adopted, the norms of being compliant patients. Nevertheless, most public hospital clinics in the RWJF program reported high rates of missed appointments for clinic patients. If missed appointments were found disproportionately among minority and drug-using patients, it could explain their lower rate of use.
In conceptualizing utilization per se as an indicator of access, the value implications of "more is better" must be considered. It is not helpful to blame patients for noncompliance if they do not understand its importance. Yet even if public hospital systems could assume responsibility for ensuring patients' compliance, this orientation might engender unnecessary dependence. Some condition-specific approach that would leave less room for
debate about values is probably necessary to determine whether more or less physician utilization is necessarily good or bad. At the same time, it must be recognized that not even this more restricted approach will reveal whether the observed differences in utilization arise from provider biases or from patients' insistence or noncompliance.
The attitudes of providers can influence patient access to the personal health care system. This statement applies to primary care medicine, specialty medical or surgical care, dental care, hospital care, nursing home, and home health aide care. Because all of these health care providers are part of society in general, no discussion of access problems attributable to the attitudes of health care workers would be complete without considering broader societal messages and influences.
Societal reactions to the emergence of an infectious chronic disease that is fatal have varied considerably over the past decade, ranging from serious proposals for quarantining persons with HIV infection to massive voluntary efforts, both financial and otherwise, directed at providing support to those suffering from the disease (Blendon and Donelan, 1988; Brandt, 1988; Musto, 1986). These attitudes are reflected in the health care professions and influence patient access to the personal health care system. In addition, attitudes of non-HIV patients may play a role: the views of such patients about being treated by a health care provider who treats persons with HIV may affect provider attitudes and behavior. A recent random sample survey of the U.S. population found that 25 percent of the public would stop seeing their doctor if they knew that their physician was also treating an HIV-infected person (Gerbert et al., 1989). Perhaps attitudinal barriers, as much as economic, structural, and personal barriers, can be assumed to influence the behavior of health care professionals, and to therefore play a subtle role in reducing access.
A substantial amount of research has been devoted to examining the attitudes of health care professionals and their willingness to treat persons with HIV infection. These topics were addressed in a disproportionately large number of social science abstracts at the fifth and sixth international AIDS conferences. A wide array of attitude surveys have been devised to measure fears, attitudes, and thoughts about working with persons who have HIV disease (Bernstein et al., 1990; Damrosch et al., 1990; Emanuel, 1988; Gordin et al., 1987; Kelly et al, 1987; MacDowell, 1989; McGrory et al., 1990; Merrill et al., 1989; Richardson et al., 1987; Shultz et al., 1988). The major categories of providers who have served as subjects for this research are physicians, residents, and medical students—as distinct from nurses and nursing students. Finally, a limited degree of research has focused on the
policies and practices of institutions such as nursing homes and home health agencies.
Merrill and colleagues (1989) classified the concerns of physicians and students about working with persons with AIDS into three groups: fear of contagion, homophobia, and discomfort with dying patients. McGrory and coworkers (1990) surveyed medical students at Columbia University and found low levels of homophobia but high rates of prejudice against treating intravenous drug users. Kelly and others (1987) surveyed physicians in three major cities asking about prejudice and willingness to interact socially with similar patients but under two disease scenarios: AIDS versus leukemia. In contrast to their view of leukemia patients, respondents saw AIDS patients as responsible for their illness and dangerous to others; they also indicated that they would experience greater discomfort in socializing with AIDS patients than with leukemia patients.
Negative attitudes about working with HIV-infected individuals also reflect broader social biases. The role of religious and moral beliefs as influences on the attitudes of health care providers' toward people with HIV has also been investigated. Francis (1989) assessed religiously based moral beliefs about AIDS and compared black and white clergy, medical students, and physicians as well as a sample of the public in the South. The disquieting results suggested that a majority of rural whites outside the health professions believe that HIV is a divine retribution and that 1 in 10 responding physicians feel this way.
Fear of contagion is another critical factor that may influence provider behavior. The survey by Kelly and coworkers (1987) suggested that fear of contagion was a central theme underlying physicians' negative attitudes toward AIDS patients. Educational programs targeting health professionals appear to be effective, both in increasing knowledge of the practices necessary to avoid infection and in changing behavior (Muskin and Stevens, 1990). However, they appear to have a differential impact on medical and dental students, with medical students more likely than dental students to manifest positive changes in attitude (Bernstein et al., 1990). Yet even after exposure to a training program, one-third of medical students and two-thirds of dental students said that they did not want to select a specialty that would bring them into contact with a high percentage of AIDS patients during their training. The authors suggest that these anxieties may influence the career choices of such students. Indeed, the study by Merrill and colleagues (1989) found that fear was the biggest factor in students' not wanting to work with HIV-infected patients. Several studies have noted that practicing physicians, house staff, students, and nursing staff all appear to have considerable skepticism about experts' assurances that they have a low probability of becoming infected (Francis, 1989; Wallack, 1989). Wallack found that this lack of trust was greatest among minority staff.
Several surveys have investigated the attitudes of nurses toward caring for HIV-infected patients. These studies generally reveal that the attitudes of nurses and nursing students are far more favorable toward AIDS patients than those of the general public. Variations in nurses' and nursing students' attitudes and fears about working with HIV-infected patients appear to be related to their knowledge of HIV transmission and their prior contact with patients (Gordin et al., 1987). Damrosch and colleagues (1990) surveyed critical care nurses in a teaching hospital and a denominational community hospital and found that teaching hospital nurses had more favorable attitudes than other nurses. Nonetheless, these authors reported that, given a choice, many of these nurses (45 percent in the teaching and 65 percent in the community hospital) would refuse to care for persons with AIDS.
Interestingly, although 26 percent of staff at a New York City teaching hospital felt that they would become infected with HIV and develop AIDS as a result of occupational exposure, 97 percent were nevertheless committed to continuing to care for these patients. This apparently contradictory finding means that interpreting attitudinal information without also investigating how health care professionals actually behave could result in incorrect, excessively pessimistic interpretations. Judgments of the impact of attitudes on access to health care for HIV-infected individuals must rely on more than surveys of knowledge and attitudes. The literature suggests that attitudes and a sense of comfort in caring for AIDS patients improve with educational exposure and time. Yet health care workers seem to remain anxious about infection and might elect not to run the risk of exposure, were they not committed to their jobs and professions. The behavior of interest, however, is whether health care workers refuse to treat HIV-infected patients if asked to do so. To date, there have been only limited instances of this phenomenon in the health care field, particularly once institutional and professional leaders set the pace.
Compared with the large number of attitudinal studies, relatively little research addresses institutional policies regarding the treatment of AIDS patients. The major medical societies have explicit policy statements on the responsibility of physicians to treat patients with HIV disease (AMA Council on Ethical and Judicial Affairs, 1988; Emanuel, 1988). There is apparently no consensus among the membership, however, and some studies suggest that between one-quarter and one-half of all physicians feel they have the right to refuse to treat some patients (Merrill et al., 1989; Rizzo et al., 1990).
Early in the epidemic, most hospitals established policies that prohibited staff from refusing to care for patients with HIV disease (McCarthy, 1988). In major metropolitan areas, most hospitals have served HIV-infected patients. With the increasing dehospitalization of HIV care, the
establishment of specialty units for the tertiary care of complex and advanced cases of HIV disease, such as already occurs for cancer, may be seriously entertained. Evidence that patient outcomes are improved and that resources are more effectively used when care is provided by experienced hospitals suggests that such proposals may have merit (Bennett et al., 1990). In New York, there have been calls to establish specialty medical facilities, in both subacute and acute care settings, for the exclusive treatment of patients with HIV disease (Mayor's Task Force on AIDS, 1989; Weinberg, 1990). Rothman and Tynan (1990) reviewed many of the advantages of separate facilities as well as the historical evidence for not using segregation as a way of improving quality, and they strongly advocate an integrated approach to mainstreaming patients with HIV disease. An argument for this position is that if specialty facilities become the strategy to ensure rational allocation of responsibility for care, it would be impossible to differentiate the desirable practice of specialization from the undesirable practice of discriminating against patients with HIV disease by deflecting them to other facilities.
In a recent study by Fleishman and Masterson-Allen (1991) of home health agencies in 11 cities with high AIDS prevalence, executives of agencies that have served HIV patients were asked to respond to an open-ended telephone interview regarding their experiences. Staff stress was a problem associated with treating persons with AIDS, but the stresses tended to relate to the complexity of patients' nursing needs and the severity of their illness rather than to fear of contagion or to prejudices. In almost all cases, agencies developed specific policies about caring for patients with HIV disease and implemented management policies that prohibited staff from refusing to work with these patients.
In contrast, nursing homes have not encouraged the acceptance of HIV-infected patients. A survey of nursing homes in Ohio found that none had actually treated an HIV-infected patient and that only 25 percent would even consider admitting such a patient (MacDowell, 1989). Most felt that specialized facilities, such as those being fostered in New York, were the most appropriate setting for these patients. Administrators expressed concerns about possibly losing both staff and current residents if AIDS patients were admitted. These results were consistent with the reaction of the nursing home industry in general.
VIABILITY OF DEVELOPING ACCESS INDICATORS FROM EXISTING DATA SETS
Ideally, social indicators should be derived easily from routinely gathered statistics. Age- and sex-adjusted mortality ratios of blacks and whites are an example of such indicators, as are hospital discharge rates per 100,000
persons in certain age-sex groups for given diagnostic conditions. Indicators of access to the personal health care system are not as easily available, however, particularly for the relatively rare condition of HIV disease.
This section proposes indicators of barriers to health care access for persons with HIV disease consistent with the three goals outlined by the IOM committee: (1) avoiding premature mortality from diseases amenable to early case-finding; (2) avoiding premature mortality from life-threatening conditions for which effective medical management exists; and (3) providing services that reduce morbidity or improve functioning. Indicators might be derived from existing, ongoing data systems that are already in place, special-purpose merges of normally unlinked data systems, and surveys of the population that should contain samples of persons with HIV disease.
Decreased Mortality Using Case-Finding
Before the advent of effective antiviral treatments to increase survival, any discussion of HIV testing was largely influenced by concerns about confidentiality. It may therefore seem strange, given the debate over the past decade about "anonymous" versus "voluntary" testing, to use HIV testing as an indicator of access to care. But antiviral therapy is now advocated early in the course of disease. In view of available life-prolonging therapeutics, those who are at risk and who are not tested are at a disadvantage. Early case-finding is as important for HIV disease as it is for breast and cervical cancer. Just as use of mammography is a "utilization"-based indicator of access to the personal health care system that is relevant for breast cancer, it is now reasonable to use the rate of HIV testing in the population as a utilization indicator of access.
Measures of the testing rate per 1,000 persons within age-sex-race profiles, using aggregated data from anonymous test sites, should provide an indication of the level of "access" to early identification for each population subgroup. In states such as Colorado, where HIV infection is a reportable condition, the rate of testing could be calculated after removing from the denominator the number of people in that subpopulation who are already infected.
The measurement task is not a simple one, however, because a myriad of conceptual and technical complications are associated with using such data as the basis for examining differential access. First, in many states, testing is available outside of state-operated anonymous test sites in settings that still protect anonymity. In addition, anecdotal evidence points to considerable out-of-state travel, particularly from states that have aggressive partner notification programs, to secure even greater anonymity in testing. Moreover, estimates of the at-risk population, whether gay men or intravenous drug users, are notoriously inaccurate; this means that the denominators
for these rates would have to be calculated separately for various age, sex, and racial groups. It also implies a recognition that the numerators (the number of people tested in each group) might reflect many persons from outside the geographic population base of interest.
Screening programs for treatable conditions are considered effective if there is evidence that they are used by a large unbiased proportion of the population or if there is evidence of a reduction in the rate at which the disease is identified in a later, untreatable stage. As applied to HIV disease, there are several potential indicators of the success of an early identification program. One example relies on the existing CDC-maintained national AIDS "registry" of all cases reported by each state's department of health. In addition to data on risk group and presenting diseases, the registry includes data on mortality. One class of deaths includes those that occur within the same calendar quarter in which the case was reported. Early identification programs should reduce the prevalence of such cases in registries.
Another example of an outcome indicator of early identification programs is the proportion of first admissions for HIV-related conditions that occur in late stages. Turner and her colleagues (1989) have developed a staging system for AIDS, based on hospital diagnoses, that predicts in hospital mortality. By using linked hospital discharge abstract record systems such as exist in New York, it is possible to identify an individual's first HIV-related hospitalization. By applying a disease staging system, the rate of presentation with advanced HIV disease at first hospitalization can be determined.
Several conceptual and methodological limitations and constraints must be considered in evaluating the validity of these outcome indicators. With respect to the AIDS registry, a host of concerns arise about reporting biases and the completeness of case ascertainment and mortality follow-up. Furthermore, case reporting for AIDS is based on an arbitrary set of clinical symptoms identified early in the epidemic. There has already been one change in the definition of AIDS that complicated use of the registry for epidemiological purposes. The designation of AIDS implies presumptive eligibility for total disability under the SSDI (Social Security Disability Insurance) program, meaning that an AIDS diagnosis represents an entitlement for financial and other benefits. Given the underrepresentation of opportunistic infections concentrated among women in the AIDS case definition criteria, there is now considerable pressure to change the definition again. Obviously, any such change will alter the validity of indicators based on the registry and will limit longitudinal comparisons.
Certain states (e.g., New York, California, Massachusetts, Maryland) have statewide, uniform hospital discharge abstract reporting systems that include information on the charges or costs incurred per discharge, in addition
to information on length of stay, discharge diagnoses, and payer source. New York has assembled the Statewide Planning and Research Cooperative System (SPARCS), which links the discharges of individuals, thus facilitating historical analyses of changes in an individual's payment sources and diagnoses (Kaufman et al., 1990). This diagnostic data set can be used to determine whether a patient's first HIV-related hospitalization occurs during an early or late stage of the disease. This estimate can be aggregated to the county or catchment area level to characterize the effectiveness of early identification programs in given locales.
The validity of measures derived from systems like the SPARCS as indicators of late diagnosis can be undermined by rapidly changing patterns of care; for example, increasing emphasis on outpatient treatment, both at home and in clinics, may lead to the avoidance of hospitalization for some patients. Thus, relying only on hospital-based statistical indicators under this scenario will confuse newly diagnosed with end-stage, drug-using patients.
Reduced Mortality Using Medical Management
Among the indicators of appropriate utilization of services that are assumed to address the overall goal of mortality reduction is the receipt of therapeutics that are known to affect survival. Consensus has been reached on the soundness of the evidence showing the effectiveness of retroviral drugs and antimicrobial agents, both in response to illness and for prophylaxis. Utilization-based indicators of access should be based on the rate of use of these drugs in various subgroups of the population. These rates can be derived from special statewide merged data sets such as the HIV Information System in Maryland, which links health department AIDS reporting information with Medicaid and Blue Cross/Blue Shield claims (including pharmaceuticals) and with hospital discharge abstracts (Hidalgo, 1990).
Another approach, which was used in the evaluation of the RWJF multicity project, is to survey individuals about their receipt of these treatments. The Agency for Health Care Policy and Research (AHCPR) has funded the AIDS Cost and Services Utilization Survey (ACSUS), which is currently in the field. The survey asks respondents about their use of AIDS-related therapeutics and will be abstracting comparable data from physician and hospital records.
Reliance on specialized data bases and surveys requires knowing well in advance the information that needs to be collected. Changes in the types of drugs being used and in how they are reimbursed may undermine the accuracy and validity of these types of indicators unless a mechanism can be developed for continuous updating of treatment-related information. For example, a new drug may be introduced and rapidly disseminated among
treating physicians while a survey is in the field; without a mechanism to capture such a change in treatment, the effects of the drug may be missed and the validity of the earlier target treatments (which are replaced by the new one) as access indicators will be undermined. Special merged data sets may be similarly compromised, particularly if a lag occurs between acceptance in the field of the new treatment's effectiveness and the treatment's acceptance by insurers, particularly public insurers, as a reimbursable claim.
The outcome of access to medical management, including antiviral treatment, should be to increase survival. In spite of the limitations of the national CDC registry, it is an appropriate vehicle by which to examine this issue. However, an adjustment must be made for the effect of late-stage diagnosis, since prior to diagnosis the disease could not have been managed. Thus, survival differentials among those who live at least three to six months postdiagnosis are a potential indicator of access.
Although these comparisons are broadly applicable, there is reason to believe that survival differentials are related not merely to differences in medical management but also to the health status of the host and the efficiency of the mode of transmission of the virus. Intravenous drug users, for instance, may have compromised their health before being exposed to HIV. In addition, transmission by contaminated needles, rather than sexually, may be much more efficient at spreading the virus within an organism. Consequently, the mere comparison of survival rates among risk groups, after adjusting for late-diagnosed cases, could overstate the effects of medical management. Use of a special-purpose merged data system, such as the one in Maryland, makes it possible to statistically control for differences in risk groups and to assess the effects of treatment across all risk groups through stratified analyses (Moore et al., 1991).
Reduced Morbidity and Improved Function
Providing of services that reduce morbidity and improve functioning is often considered synonymous with out-of-hospital care. Evidence from a number of studies of persons with AIDS suggests a strong preference for care at home and for maintaining control over medical decisionmaking (Teno et al., 1990, 1992). Thus, it is safe to say that this population values time spent at home and away from the hospital.
Examples of utilization indicators of access to services that can help reduce morbidity include the rate of hospital admissions through the emergency room, the proportion of people with AIDS who use home health agency services, and the number of additional days spent in the hospital due to administrative discharge delays. Data on home health use must be derived from periodic surveys of the population such as the ACSUS. Hospital admissions through the emergency room and the number of administrative,
or ''outlier," days can be obtained from the National Center for Health Statistics (NCHS), which maintains ongoing data on a large number of hospital discharge abstracts used to estimate population-level hospital use rates by discharge diagnosis.
Considerable care must be taken in interpreting the meaning of admission through the emergency room and extended hospital stays, since the medical care system is not necessarily the most efficient structure for redressing other societal inequities (e.g., limited social support and financial resources for intravenous drug users, lack of drug treatment program openings that could facilitate their discharge). Moreover, in many parts of the country, physicians instruct patients to enter the hospital through the emergency room, reflecting a hospital operating strategy. Because it is likely that hospitals in different areas use their emergency rooms in very different fashions, differences in rates of emergency room use by subgroups may reflect administrative styles and not merely access differences.
Outcome indicators of access associated with the goal of reducing morbidity can be conceived of as avoiding unnecessary hospitalization and receiving care at home. The first of these can be measured by using a population-based uniform hospital discharge abstract system matched with a statewide AIDS registry at the zip code, or census tract, level. The registry data provide the denominator of AIDS cases, and the hospital discharge abstracts provide the numerator of admissions for specific conditions, such as Pneumocystis carinii pneumonia (PCP). Rates of PCP-related hospitalization for minorities and nonminorities can be compared on the assumption that effective prophylactic treatment would minimize hospitalization.
A methodological limitation of the validity of this measure involves the transience of patients: many will not necessarily continue receiving care in a single hospital. Patients' addresses on discharge abstracts are often coded geographically by zip code or census tract; matching these codes with comparably coded registry information is problematic because either address may be incorrect, particularly among certain groups. Naturally, this bias would undermine the validity of any comparison.
Since 1987, when uniform coding of AIDS with the International Classification of Diseases-9 framework began for hospital discharge and for vital and health statistics, death certificates have offered another source of data from which selected indicators can be derived. Among HIV disease-related deaths, the distribution of the location of death can be examined. As of 1989, the nationally standardized death certificate coding scheme has included a data element for deaths occurring at home as opposed to in the hospital, in a nursing home or at another site. Those AIDS patients who die at home could be compared on the basis of gender, age, race, and geographic area as an indicator of the availability and use of home health and out-of-hospital services.
Much health and social policymaking over the past several years has been devoted to buttressing community support services to reduce reliance on hospital care. Recently, McMillan and colleagues (1990) demonstrated that policies on hospices have affected the site of death of Medicare beneficiaries dying of cancer; the study used Medicare claims merged with death certificates. Some evidence suggests that AIDS patients also would prefer to die at home (Teno et al., 1992). Consequently, the proportion of deaths that occur at home can be a useful barometer of access to services that is particularly applicable when looking at age- and sex-adjusted rates within certain metropolitan areas known to have strong home support services.
Despite the promise of the various data systems that have been suggested as the basis for deriving access indicators, the rapidly changing circumstances of the epidemic, the variability in state programs offering subsidized or free care to the poor, and the regional variation in Medicaid coverage make it difficult to imagine a data system that would have complete disease, health utilization, and outcome data on a representative sample of persons with HIV. Consequently, even given a better understanding of the meaning of the suggested indicators, the result is likely to be a series of incomplete snapshots. Each will have its own limitations and biases of which users, particularly policymakers, must be aware. Yet despite these limitations, many of the merged data system files proposed above hold considerable promise and should be explored further. By "cross walking" findings from the more detailed surveys, such as ACSUS, with the population-based utilization estimates derived from merged data systems, a more informed opinion could be obtained about the validity of the resulting access indicators. Ideally, when indicators such as those suggested here are assembled into a montage, they will form a more coherent, more consistent picture.
AMA (American Medical Association) Council on Ethical and Judicial Affairs. 1988. Ethical issues involved in the growing AIDS crisis. Journal of the American Medical Association 259:1360–1361.
American College of Physicians, Health and Public Policy Committee, and the Infectious Diseases Society of America. 1988. The acquired immunodeficiency syndrome (AIDS) and infection with the human immunodeficiency virus (HIV). Annals of Internal Medicine 258:2714–2717.
Andrulis, D. P. 1989. Crisis at the Frontline. New York, NY: The Twentieth Century Fund, p. 50.
Andrulis, D. P., Beers, V. S., Bentley, J. D., et al. 1987a. The provision and financing of medical care for AIDS patients in U.S. public and private teaching hospitals. Journal of the American Medical Association 258:1343–1346.
Andrulis, D. P., Beers, V. S., Bentley, J. D., and Gage, L. S. 1987b. State Medicaid policies and hospital care for AIDS patients. Health Affairs (Winter)6:110–118.
Andrulis, D. P., Weslowski, V. B., and Gage, L. S. 1989. The 1987 U.S. Hospital AIDS survey. Journal of the American Medical Association 262:784–794.
Arno, P. S., Shenson, D., Siegel, N. F., Franks, P., and Lee, P. R. 1990. Economic and policy implications of early intervention in HIV disease. Journal of the American Medical Association 262:1493–1499.
Baily, M. A., Bilheimer, L., Wooldridge, J., Langwell, K., and Greenberg, W. 1990. Economic consequences for Medicaid of human immunodeficiency virus infection. Health Care Financing Review 12:97–108.
Benjamin, A. E. 1988. Long term care and AIDS: Perspectives from experience with the elderly. Milbank Quarterly 66:415–443.
Bennett, C. L., Gertler, P., Guze, P. A., Garfinkle, J. B., Kanouse, D. E., and Greenfield, S. 1990. The relation between resource use and in-hospital mortality for patients with acquired immunodeficiency syndrome related Pneumocystis carinii pneumonia. Archives of Internal Medicine 150:1447–1452.
Bernstein, C. A., Tabkin, J. G., and Wolland, H. J. 1990. Medical and dental students' attitudes about AIDS issues. Academic Medicine 65:458–460.
Blendon, R. J. and Donelan, K. 1988. Discrimination against people with AIDS: The public's perspective. New England Journal of Medicine 319:1022–1026.
Brandt, A. M. 1988. AIDS in historical perspective: Four lessons from the history of sexually transmitted diseases. American Journal of Public Health 78:367–371.
Buchanan, R. J. 1988. State Medicaid coverage of AZT and AIDS-related policies. American Journal of Public Health 78:432–436.
Capilouto, E. I., Piette, J., White, B. A., and Fleishman, J. 1991. Perceived need for dental care among persons living with acquired immunodeficiency syndrome . Medical Care 29:745–754.
CDC (Centers for Disease Control). 1990. HIV prevalence estimates and AIDS case projections for the United States: Report based upon a workshop. Morbidity and Mortality Weekly Report 39(RR-16):1–31.
CDC (Centers for Disease Control). 1991. Mortality attributable to HIV infection/AIDS— United States, 1981–1990. Morbidity and Mortality Weekly Report 40:41–44.
Cohen, P. T., Sande, M. A., and Volberding, P. 1990. The AIDS Knowledge Base. Waltham, Mass.: Medical Publishing Group.
Damrosch, S., Abbey, S., Warner, A., and Guy, S. 1990. Critical care nurses' attitudes toward, concerns about, and knowledge of the acquired immunodeficiency syndrome. Heart-Lung 19:395–400.
DeHovitz, J. A. 1990. The increasing role of primary care in the management of HIV-infected patients. New York State Journal of Medicine 90:119–120.
Eby, C. 1989. The share of AIDS medical care costs paid by private insurance in the United States. P. 1047 in Abstracts from the Fifth International Conference on AIDS, Montreal, Quebec, June 4–9. Abstract T.H.P. 24. Ottawa: International Development Research Centre.
Emanuel, E. 1988. Do physicians have an obligation to treat patients with AIDS? New England Journal of Medicine 315:1686–1690.
Faden, R. R. and Kass, N. E. 1988. Health insurance and MDs: The status of state regulatory activity. American Journal of Public Health Statistics 78:437–438.
Fischl, M. A., Richman, D. D., Grieco, M. H., et al. 1987. The efficacy of azidothymidine (AZT) in the treatment of patients with AIDS and AIDS-related complex: A double-blind, placebo-controlled trial. New England Journal of Medicine 317:185–191.
Fleishman, J. 1990. Research issues in service integration and coordination. Pp. 167–187 in Community-Based Care for Persons with AIDS: Developing a Research Agenda, AHCPR
Conference Proceedings. DHHS Pub. No. (PHS)90-3456. Rockville, Md.: U.S. Government Printing Office.
Fleishman, J. A., Cwi, J., and Mor, V. 1989. Sampling and accessing people with AIDS: A study of program clients in nine locations. Pp. 181–186 in Proceedings of the Fifth Conference on Health Survey Research Methods, Floyd J. Fowler (ed.). DHHS Pub. No. (PHS)89-3447. Washington, D.C. : U.S. Government Printing Office.
Fleishman, J. A., and Masterson-Allen, S. 1991. Delivery of home care to people with AIDS. Final report prepared for the Agency for Health Care Policy and Research.
Fleishman, J. A., Mor, V., and Piette, J. 1991. AIDS case management: The client's perspective. Health Services Research 26(Oct.):447–470.
Fleishman, J. A., Piette, J., and Mor, V. 1990. Organizational response to AIDS. Evaluation and Program Planning 13:31–38.
Francis, R. A. 1989. Moral beliefs of physicians, medical students, clergy, and lay public concerning AIDS. Journal of the National Medical Association 81:1141–1147.
Gerbert, B., Maquire, B. T., Hulley, S. B., and Coates, T. J. 1989. Physicians and acquired immunodeficiency syndrome: What patients think about human immunodeficiency virus in medical practice. Journal of the American Medical Association 262:1969–1972.
Ginzberg, J. 1990. Health Services Research: Key to Health Policy. Cambridge, Mass.: Harvard University Press.
Gordin, F., et al. 1987. Knowledge of AIDS among hospital workers: Behavioral correlates and consequences . Journal of Acquired Immune Deficiency Syndromes 1:183–188.
Green, J. and Arno, P. S. 1990. The Medicaidization of AIDS: Trends in the financing of HIV-related medical care. Journal of the American Medical Association 264:1261–1266.
Harris, J. E. 1990. Improved short term survival of AIDS patients initially diagnosed with Pneumocystic carinii pneumonia. Journal of the American Medical Association 263:397–401.
Hayward, R. A., Bernard, A. M., Freeman, H. E., and Corey, C. R. 1991. Regular source of ambulatory care and access to health services. American Journal of Public Health 81:434–438.
Hidalgo, J., Sugland, B., Bareta, J. C., Allen, N., Moore, R. and Chaisson, R. E. 1990. Access, Equity, and Survival: Use of Zidovudine (ZVD) and Pentamidine by Persons with AIDS. Paper presented at the Sixth International Conference on AIDS, San Francisco, California, June 23.
Hidalgo, J. 1990. Development and application of statewide acquired immunodeficiency syndrome (AIDS) information systems in health services planning and evaluation. Evaluation and Program Planning 13:39–46.
Holahan, J. 1984. Paying for physician services in state Medicaid programs. Health Care Financing Review 5:99–110.
Holmberg, S. D., Harrison, J. S., Buckbinder, et al. 1990. Therapeutic and Prophylactic Drug Use by Homosexual and Bisexual Men in Three U.S. Cities. Paper presented at the Sixth Conference on AIDS, San Francisco, California, June 23.
IOM (Institute of Medicine). 1986. Confronting AIDS. Washington, D.C.: National Academy Press.
Kaplowitz, L. G., Turshen, J., Myers, P. S., Staloch, L. A., Berry, A. J., and Settle, J. T. 1988. Medical care costs of patients with acquired immunodeficiency syndrome in Richmond, VA: A quantitative analysis. Archives of Internal Medicine 148:1793–1797.
Kass, N. 1989. Access to insurance and perceived discrimination by homosexual men. P. 1047 in Abstracts from the Fifth International Conference on AIDS, Montreal, Quebec, June 4–9. Abstract T.H.P. 20. Ottawa: International Development Research Centre.
Kaufman, G. I., Grabau, J. C., Schmidt, E. M., and Han, Y. 1990. HIV infected hospital patients in New York city state. The development of longitudinal information from a hospital discharge data system. New York State Journal of Medicine 90:238–242.
Kelly, J. A., St. Lawrence, J. S., Smith, S., Hood, H. V., and Cook, D. J. 1987. Medical students' attitudes toward AIDS and homosexual patients. Journal of Medical Education 62:549–556.
Lemp, G. F., Payne, S. F., Neal, D. Temelso, T., and Rutherford, G. W. 1990. Survival trends for patients with AIDS. Journal of the American Medical Association 263:402–406.
Lipson, B. 1988. A crisis in insurance. New England Journal of Public Policy 285–305, Winter/Spring.
MacDowell, N. M. 1989. Willingness to provide care to AIDS patients in Ohio nursing homes. Journal of Community Health 14:205–213.
Marder, R. E. and Linsk, N. L. 1990. Assessment of discharge planning referral to nursing homes for people with AIDS and HIV. P. 287 in Abstracts from the Fifth International Conference on AIDS, Montreal, Quebec, June 4–9. Vol. 3, Abstract No. S.D. 791. Ottawa: International Development Research Centre.
Martin, C., Tortolero, S., Hines, A., Jardin, L., and Johnson, C. 1989. How do persons with symptomatic HIV disease finance their care in Texas? P. 1047 in Abstracts from the Fifth International Conference on AIDS, Montreal, Quebec, June 4–9. Abstract No. T.H.P. 22. Ottawa: International Development Research Centre.
Mayor's Task Force on AIDS. 1989. Assuring care for New York City's AIDS population. New York, N.Y.: Mayor's Task Force on AIDS, March.
McCarthy, C. 1988. The role of the American Hospital Association in combating AIDS. Public Health Reports 103:273–277.
McGrory, B. J., McDowell, D. M., and Muskin, P. R. 1990. Medical students' attitudes toward AIDS, homosexual and intravenous drug-abusing patients: A re-evaluation in New York City. Psychosomatics 31:426–433.
McMillan, A., Mentnech, R. M., Lubitz, J., McBean, A. M. and Russell, D. 1990. Trends and patterns in place of death for Medicare enrollees. Health Care Financing Review 12:1–7.
Mechanic, D. and Aiken, L. H. 1989. Lessons from the past: Responding to the AIDS crisis. Health Affairs 8(Fal):16–32.
Merrill, J. M., Laux, L., and Thornby, J. I. 1989. AIDS and student attitudes. Southern Medical Journal 82:426–432.
Moore, R. D., Hidalgo, J., Sugland, B. W., and Chaisson, R. E. 1991. Zidovudine and the natural history of the acquired immunodeficiency syndrome. New England Journal of Medicine 324:1412–1416.
Mor, V., Piette, J., and Fleishman, J. A. 1989. Community-based case management for persons with AIDS. Health Affairs 8(Winter):139–153.
Mor, V., Fleishman, J. A., Dresser, M., and Piette, J. 1992. Variation in health service use among HIV-infected patients. Medical Care 30:17–29.
Muskin, P. R. and Stevens, L. A. 1990. An AIDS educational program for third year medical students. General Hospital Psychiatry 12:390–395.
Musto, D. F. 1986. Quarantine and the problem of AIDS. Milbank Memorial Fund Quarterly 64(suppl. 1):97–117.
Myers, L. P., and Vladeck, B. C. 1990. New York City's Hospital Occupancy Crisis: Caring for a Changing Population. A report from the Bigel Institute for Health Policy, United Hospital Fund of New York, 1988, and the New York State Governor's Statewide Anti-Drug Abuse Council. Anti-Drug Abuse Strategy Report.
Neidle, E. A. 1989. AIDS-related changes in dental practice. Journal of Dental Education 9:525–528.
Northfelt, D. W., Hayward, R. A., Shapiro, M. F. 1988. The acquired immune deficiency syndrome is a primary care disease. Annals of Internal Medicine 109:773–775.
Parmet, W. E. 1987. AIDS and the limits of discrimination law. Law, Medicine, and Health Care 15:61–72.
Perloff, J. D., Kletke, P. R., and Neckeman, K. M. 1987. Physicians decisions to limit Medicaid participation: Determinants and policy implications. Journal of Health Politics, Policy and Law 12:221–235.
Pickett, N. A., Drewry, S. J., and Comer, E. L. 1990. AIDS: MetLife's experience. Statistical Bulletin—Metropolitan Life Insurance Company 71:2–9.
Piette, J., Fleishman, J. A., Dill, A., and Mor, V. 1990. A comparison of hospital and community case management programs for persons with AIDS. Medical Care 28:746–755.
Piette, J., Stein, M., Mor, V., Fleishman, J. A., Mayer, K., Wachtel, T., and Carpenter, C. In press. Patterns of secondary prophylaxis with aerosol pentamidine among persons with AIDS (letter). Journal of Acquired Immune Deficiency Syndrome.
Richardson, J. L., Lochner, T., McGuigan, K., and Levine, A. M. 1987. Physicians' attitudes and experience regarding the care of patients with acquired immunodeficiency syndrome (AIDS) and related disorders (ARC). Medical Care 25:675–685.
Rizzo, J. A., Marder, W. D., and Willke, R. J. 1990. Physician contact with and attitudes toward HIV-seropositive patients: Results from a national survey. Medical Care 28:251–260.
Rothman, D. J. and Tynan, E. A. 1990. Advantages and disadvantages of special hospitals for patients with HIV infection. New England Journal of Medicine 323:764–768.
Rowe, M. J. and Ryan, C. C. 1988. Comparing state-only expenditures for AIDS. American Journal of Public Health 78:424–429.
Seage, G. R., Landers, S., Lamb, G. A., and Epstein, A. M. 1990. Effect of changing patterns of care and duration of survival on the cost of treating the acquired immunodeficiency syndrome (AIDS). American Journal of Public Health 80:835–839.
Shultz, J. M., MacDonald, K. L., Heckert, K. A., and Osterholm, M. T. 1988. The Minnesota AIDS Physician Survey. Minnesota Medicine 71:277–283.
Stein, M. D., Piette, J., Mor, V., Wachtel, T. J., Fleishman, J. A., Mayer, K. H., and Carpenter, C. 1991. Differences in access to zidovudine (AZT) among symptomatic HIV-infected persons. Journal of General Internal Medicine 6:35–40.
Swan, J. H., Benjamin, E. A., and Lee, P. R. 1990. Resource use for AIDS care in nursing homes. P. 286 in Abstracts from the Fifth International Conference on AIDS, Vol. 3. Abstract F.D. 816.
Teno, J., Fleishman, J. A., Brock, D. W., and Mor, V. 1990. The use of formal prior directives among patients with HIV-related diseases. Journal of General Internal Medicine 5:490–494.
Teno, J. Fleishman, J. A., and Mor, V. 1992. AIDS patients' preferences for community-based care. Unpublished paper.
Turner, B. J., Kelly, J. V., and Ball, J. K. 1989. A severity classification system for AIDS hospitals. Medical Care 27:423–437.
Vercusio, A. C., Neidle, E. A., Nash, K. D., Silverman, S. Horowitz, A. M., and Wagner, K. S. 1989. The dentist and infectious diseases: A national survey of attitudes and behavior. Journal of the American Dental Association 118:553–562.
Volberding, P. A., Lagakos, S. W., and Koch, M. A. et al. 1990. Zidovudine in asymptomatic human immunodeficiency virus infection: A controlled trial in persons with fewer than 500 CD4-positive cells per cubic millimeter. New England Journal of Medicine 332:941–949.
Wallack, J. J. 1989. AIDS anxiety among health care professionals. Hospital and Community Psychiatry 40:507–510.
Weinberg, L. 1990. Bill would require states to raise Medicaid AIDS rates. Modern Healthcare, p. 17.
White, C. M. and Berger, M. C. 1991. Response of hospitals, skilled nursing facilities and home health agencies in Oregon to AIDS: Reports of nursing executives. American Journal of Public Health 81:495–496.
Yelin, E. H., Greenblatt, R. M., Hollander, H., and McMaster, J. R. 1991. The impact of HIV-related illness on employment. American Journal of Public Health 81:79–84.
Zucconi, S., Kingsley, L., Lave, J., and Ishii, E. 1989. Impact of insurance coverage on health care utilization for HIV infected persons: Results of a survey. P. 289 in Abstracts from the Fifth International Conference on AIDS, Vol. 3, Abstract S.D. 797.