Access to Health Care:

Originally published by the Center for Studying Health System Change (HSC). HSC was a nonpartisan policy research organization funded principally by the Robert Wood Johnson Foundation.

Access to Health Care: Bridging the Gap Between Policy and Research

Issue Brief No. 08 | April 1997 | Heidi H. Whitmore

Researchers have long studied access to health care, but the focus of that work has evolved over time. Where earlier efforts concentrated on documenting whether the poor and uninsured faced worse access than others, the emphasis more recently shifted toward monitoring how access was changing from year to year. However, only a small number of measures had been applied consistently enough to track these shifts over time, and the measures that were available suggested that access for the uninsured was declining. Other measures with greater potential value existed, but the data needed to apply them consistently had not been collected. This Issue Brief, drawn from a seminar held by the Center for Studying Health System Change, examined the research efforts underway to measure access to care and the challenges they faced.

Defining and Measuring Access to Care

Tracking access to care was far from straightforward. Access itself was a broad and frequently vague concept that took in multiple dimensions of health care, from the supply and availability of providers to insurance coverage and the specific barriers individuals encountered when trying to get care. No gold standard existed for measuring it, and researchers were constantly developing new approaches to reflect changes in how services were delivered along with growing interest in outcome-based measures.

This was not just an academic problem. Policy makers had a strong interest in health care system research, particularly in understanding how policy changes affected the way health care was delivered. But longitudinal data on access to care were scarce, and because much of the available research appeared to produce conflicting results, policy makers had been reluctant to rely on it.

Researchers had relied primarily on population-based surveys to study access. A separate line of research used the volume of charity or uncompensated care delivered by hospitals and other providers as a proxy for health services reaching the uninsured. But this approach had important limitations. Changes in the amount of uncompensated care might not accurately reflect changes in access, because those changes could be driven by the supply of resources available to providers rather than the actual demand from people lacking insurance. For instance, a decline in uncompensated care might signal that hospitals had fewer resources to provide it, rather than any reduction in the community's need for free care. In some communities, uncompensated care trends would mirror access trends, but in others they would reflect supply-side constraints instead.

There were additional drawbacks to the uncompensated care approach. Government subsidies and private grants to hospitals for uncompensated care were sometimes excluded from the total, which meant the measure could understate the actual amount of free care provided. At the same time, uncompensated care figures could overstate the true volume because they traditionally combined charity care and bad debt -- and some bad debt had nothing to do with caring for uninsured patients. A hospital might record bad debt because an insurance company failed to pay for services delivered to one of its covered members. Furthermore, most uncompensated care databases did not distinguish between inpatient and outpatient care, so shifts from one setting to the other went uncaptured.

Population-based surveys offered a more direct and widely used approach to measuring access, with several advantages over uncompensated care data. These surveys employed a broader set of measures capturing multiple dimensions of access: aspects of primary care, the process of seeking care (such as ease and convenience of reaching a doctor), barriers to care (such as language and transportation problems), and unmet health needs. By comparing vulnerable populations -- the uninsured, the poor, those in poor health -- to the rest of the population, researchers could draw inferences about who was at greatest risk of lacking access.

Population-based surveys were not ideal instruments for capturing health outcomes related to access, however. They could not tell researchers whether patients were receiving good-quality care in the most appropriate setting at the lowest possible cost. To examine those questions, researchers turned to hospital discharge data to measure rates of avoidable hospitalizations -- conditions like asthma and diabetes that could be treated in ambulatory settings, and for which hospitalization pointed to a lack of access to high-quality primary care. Interest in these outcome-based measures was expected to grow.

Trends in Access

Peter Cunningham of the Center for Studying Health System Change illustrated the difficulty of drawing conclusions about access for the uninsured based on uncompensated care trends. Using physician surveys conducted by the American Medical Association (AMA) alongside reports of physician visits from the National Health Interview Survey (NHIS), he found that the volume of charity care provided by physicians increased substantially between 1990 and 1994. Yet the average number of physician visits for uninsured people did not change during the same period and consistently remained well below the level for insured people. These findings raised serious doubts about whether changes in uncompensated care could serve as a reliable indicator of changes in access.

Making inferences about access from hospital uncompensated care data was even more problematic. While hospital uncompensated care grew somewhat during the 1980s, the number of uninsured people also increased during the same period. Moreover, it was not clear whether increases in hospital use indicated better or worse access overall. The percentage of hospital discharges for ambulatory-sensitive conditions -- avoidable hospitalizations -- rose between 1980 and 1994 for the general population, but the rate of increase was twice as high for the uninsured. This suggested that access to primary care for uninsured individuals had deteriorated. In an era that increasingly emphasized primary care and cost-effectiveness, whether rising hospital uncompensated care represented improved or diminished access for the uninsured was far from clear.

Lack of Consistency

Despite the relatively large number of surveys conducted by both government agencies and private organizations, the ability to observe trends in access to care over the prior 10 to 15 years had been limited. This was largely because of inconsistency in the measures used and methodological differences among surveys that could influence results even when the measures were identical.

Valid comparisons over time generally required longitudinal surveys that maintained the same basic design, methodology, and virtually identical questions. Until recently, the only annual health care survey was the NHIS, conducted by the National Center for Health Statistics (NCHS). Other surveys had been done infrequently: the National Medical Expenditure Survey (NMES) was conducted by the Agency for Health Care Policy and Research (AHCPR) in 1977, 1987, and 1996, and the Access to Care Survey was conducted by The Robert Wood Johnson Foundation (RWJF) in 1978, 1982, 1986, and 1994.

Even with these surveys, tracking certain access measures remained difficult. The NHIS did not include measures of access beyond service use until the late 1980s, and many researchers considered service use alone to be of limited value as an access measure. Questions also changed from one survey to the next, further complicating comparisons.

P. Ellen Parsons of NCHS noted that constructing measures to capture the concept of access was challenging, and the task grew even harder given the rapid pace of health system change. Some researchers questioned whether tracking usual source of care -- a traditional and widely used measure -- was even relevant for people enrolled in managed care plans. Traditional survey methods also did not collect data on the severity and stage of disease at diagnosis, information that was critical for developing outcome-based measures.

A Case in Point: Unmet Health Needs

Unmet health needs was one of the most important access measures included in surveys, yet it was virtually impossible to make accurate inferences about changes in unmet needs over time because there was no standardization across surveys.

A recent study by Karen Donelan of the Harvard School of Public Health and colleagues examined access problems and their consequences among both insured and uninsured populations. The study used several measures to assess access problems, including difficulty obtaining needed medical care, difficulties paying medical bills, and the severity of health problems for which individuals had trouble getting care. The survey asked respondents whether there was a time when they needed care but did not get it, and -- unlike previous surveys -- also asked whether respondents had difficulty getting care even if they eventually received it.

The Donelan study found that a substantially higher percentage of the population had unmet health needs than previous surveys had reported. Among uninsured respondents, 45 percent said they had unmet needs, compared with a low of 6 percent in the 1987 NMES, 13 percent in the 1993 NHIS, 20 percent in the 1986 RWJF Access to Care survey, and 34 percent in the 1994 RWJF Access to Care survey.

Marc Berk of Project HOPE identified several factors that could explain these wide disparities. Many prior surveys, including the NMES, did not include the additional question about difficulty obtaining care even when eventually received, and including it likely pushed the estimate upward. Longer surveys that placed access questions toward the end tended to show lower rates of unmet need, while the Donelan study used a relatively brief interview with access questions asked at the beginning. Surveys with higher response rates tended to produce lower rates of unmet need, and telephone surveys generally showed higher rates than in-person interviews; the Donelan study had a lower response rate and used telephone interviewing.

These inconsistencies made it virtually impossible to draw firm conclusions about trends in unmet health needs from the various surveys. Researchers could reach sharply different conclusions about changes in access depending on which studies they used for comparison. For example, comparing the 1986 RWJF survey with the 1993 NHIS showed a roughly 50 percent decrease in unmet health needs, but comparing the 1987 NMES with the 1996 Donelan study showed an increase of approximately 700 percent.

Addressing the Problem of Consistency

These problems highlighted the need for greater standardization in how access to care was measured and for continuous monitoring over time. Recent activities by the federal government and private organizations suggested that these needs were beginning to be recognized.

On the federal side, the annual NHIS had undergone a major restructuring beginning in the early 1990s and now addressed barriers to care, unmet needs, and other access-related issues on a regular basis. Periodic supplements to the NHIS examined the duration of respondents' relationship with their usual source of care, satisfaction with care, reasons for delays in receiving care, efforts to obtain needed care, whether care was eventually obtained, and the health impact of not getting needed care.

AHCPR had created the Medical Expenditure Panel Survey (MEPS) to replace the infrequently conducted NMES. The MEPS would be conducted annually, allowing continuous monitoring of access alongside health care utilization, insurance coverage, and expenditures. It would approach access issues similarly to the NHIS and would be able to track the relative burden of out-of-pocket health care costs.

The Medicare Current Beneficiary Survey (MCBS), conducted by the Health Care Financing Administration (HCFA), was another recent government effort to improve access monitoring. Although the MCBS was primarily designed to track health care expenditures for the Medicare population, it included special supplements on access that were administered on an ongoing basis.

The Center for Studying Health System Change had also made tracking access a priority through its household and physician surveys as part of the Community Tracking Study. The Center planned to track changes in access using measures such as difficulty obtaining needed care, ease and convenience of getting care, usual source of care, charity care, and physicians' ability to obtain specialist referrals. These surveys would be conducted every two years to allow longitudinal tracking. In addition, the Center would monitor other aspects of access -- such as the organization and viability of the health care safety net -- through site visits to 12 communities across the country.

With so many changes in the health care system affecting access for vulnerable populations -- declining ability of providers to shift costs, cutbacks in public funding for the safety net, erosion of employment-based insurance coverage -- the need for researchers to get their findings about access trends into the hands of policy makers had never been greater. Understanding these trends was critical to assessing how well the health care system was performing.

Presenters and Participants

Peter Cunningham, Center for Studying Health System Change; Karen Donelan, Harvard School of Public Health; P. Ellen Parsons, National Center for Health Statistics. Commentator: Chip Kahn, Ways and Means, Subcommittee on Health, U.S. House of Representatives. Moderator: Paul B. Ginsburg, Center for Studying Health System Change.

This Issue Brief is based on a health policy research seminar sponsored by the Center for Studying Health System Change, which took place in Washington, D.C., on March 7, 1997.

Sources and Further Reading

Center for Studying Health System Change, Issue Brief No. 08, April 1997.