Health Affairs, 22, no. 6 (2003): 143-155 doi: 10.1377/hlthaff.22.6.143 © 2003 by Project HOPE	New Online   House Health Reform Bill   Paying for Reform   Vetting AHIP's Report   HIV/AIDS Costs   Brief: Insurance Reform   HA Blog Top 10

This Article Abstract Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when Comments are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via Web of Science (2) Citing Articles via Google Scholar Google Scholar Articles by McLaughlin, C. G. Articles by Mortensen, K. Search for Related Content PubMed PubMed Citation Articles by McLaughlin, C. G. Articles by Mortensen, K. Related Collections Access To Care Hospitals Safety-Net Systems

Safety Net

Who Walks Through The Door? The Effect Of The Uninsured On Hospital Use

Abstract

 
Hospitals are concerned about the implications of an increase in the number of uninsured people. Using data from the 1999 Medical Expenditure Panel Survey (MEPS), we calculate what percentage of hospital inpatient, emergency department, and outpatient visits are accounted for by uninsured people and predict how those shares would change under three different scenarios. We find that although the burden of the uninsured would remain a severe problem for some hospitals, it would not likely increase much for most of them. This finding reflects the relatively low utilization rates among those most likely to lose coverage: nonelderly, nonpregnant, and nondisabled workers and their families.

The impact of an increase in the uninsured population is influenced by several factors: which currently insured people lose their coverage; the change in hospital utilization patterns as a result of that loss; and the differential between the revenue received for the care given to people without coverage and the marginal cost of providing that care.

Three scenarios. In this analysis we do not address the financial consequences of changes in the number of uninsured people, but we do predict the likely changes in the share of hospital visits accounted for by uninsured people under three possible scenarios: that the population groups most at risk of losing health insurance in the near future are (1) working adults and their dependents; (2) Medicaid eligibles who were part of the 1990s expansions, particularly nonelderly, nonpregnant, nondisabled adults and children with household incomes above the poverty threshold; and (3) Medicare eligibles ages 65–67.

Although the richness of benefits could change for those who retain coverage, we hypothesize that state and federal programs providing coverage for hospital services to pregnant women, nonelderly disabled adults and children, and the bulk of the elderly are not likely to be eliminated.

Changes already occurring. These hypotheses reflect in part the changes that are already occurring. A recent survey found that small businesses are the most likely to drop coverage.² As of January 2003 eleven states had implemented or scheduled to implement changes in Medicaid policies that would drop an estimated one million people from coverage.³ Although two states had also proposed cutbacks in Medicaid coverage for the elderly and disabled, the majority of the programmatic changes affected poor and near-poor working families. There have been no calls to make large cutbacks in the disability eligibility criteria for Medicare.

Study Data

Top Study Data Who Is Walking Through... Hospital Utilization Changes In... Results Of Three Simulations Discussion And Policy... Editor's Notes NOTES

 
We use two different data sets to look at who walks through the hospital door: the 2000 Nationwide Inpatient Sample (NIS), from the Agency for Healthcare Research and Quality’s (AHRQ’s) Healthcare Cost and Utilization Project (HCUP); and the 1999 Medical Expenditure Panel Survey (MEPS), also from AHRQ.

The NIS. The 2000 NIS provides information on 7.5 million discharges from 994 hospitals located in twenty-eight states that are matched with data from the American Hospital Association (AHA) 2000 survey. The data approximate a 20 percent stratified sample of U.S. community hospitals.

The NIS includes selected information about each hospital and the primary expected payer (private insurer, Medicare, Medicaid, patient, and no charge) for each discharge. For this analysis we assumed that if the primary payer was the patient, or if at intake the hospital anticipated no payment and therefore made no charge, the patient was uninsured. Using this information, we estimated the percentage of uninsured discharges for different broad categories of hospitals (by region, type, and size).

The NIS does not collect information about hospital emergency department (ED) and outpatient visits. In addition, while rich in details about each discharge, the NIS has very little nonmedical information about each patient and therefore is of limited value in simulating changes in use based on changes in coverage.

MEPS. To calculate which people are likely to change their insurance coverage and then simulate accompanying changes in use, we used data from the 1999 MEPS. A national probability survey, MEPS provides information about various individual characteristics, including health insurance coverage, as well as utilization data for hospital inpatient, ED, and outpatient visits during 1999. The MEPS Household Component (HC) sample is drawn from respondents to the National Health Interview Survey (NHIS), a nationally representative survey of the U.S. civilian noninstitutionalized population. MEPS uses an overlapping panel design in which data are collected in six rounds of interviews over two and a half years.

The 1999 file contains variables associated with 24,618 people. Of this sample, 23,565 were included in our analyses to generate nationally representative estimates. Information about hospital use for each person was then obtained from the hospital inpatient, ED, and outpatient files. These data allowed us to estimate the distribution of hospital visits according to insurance coverage and simulate how that distribution would change under various policy changes.

Who Is Walking Through The Door Now?

Top Study Data Who Is Walking Through... Hospital Utilization Changes In... Results Of Three Simulations Discussion And Policy... Editor's Notes NOTES

 
Estimates from the NIS. In 2000 the uninsured accounted for almost 5 percent of all inpatient discharges. There is variation around that average, with urban teaching hospitals and hospitals in the South carrying larger shares (Exhibit 1). Several analysts have looked at disproportionate shares of uncompensated care among hospitals, which is not quite the same as the share of visits accounted for by uninsured people, but the observed relationships are similar.⁴ While the average share of uninsured inpatient discharges is 4.8 percent, the average share for the top decile of hospitals—that is, those with the highest shares—was 14.6 percent, with individual hospitals ranging from 8.6 percent to 43.3 percent.⁵

We provide estimates of the lower and upper bounds for the share of visits accounted for by those without coverage.⁶ For example, the percentage of inpatient stays for patients without health insurance is at least 5 percent (full-year uninsured) and as much as 13 percent (full- and part-year uninsured combined) (Exhibit 2). Just as the overwhelming majority of the U.S. population has health insurance coverage, the overwhelming majority of inpatients have health insurance.

Use of EDs. Of more concern to many hospital administrators and policy analysts, however, is the use of EDs by those without insurance coverage. Given the relationship between coverage and access to private physicians’ offices, the ED is often viewed as the source of both nonurgent and urgent care for many without health insurance. We estimated that at least 10 percent of ED visits are made by those without coverage (Exhibit 2).

Interestingly, while uninsured people’s share of ED visits is much higher than their share of inpatient stays, it is proportional to the percentage of the population without coverage for the entire year. That is, in 1999, 11 percent of the population was without coverage the entire year, and 10 percent of ED visits were made by this population group. Another 11 percent were uninsured for some part of that year. Even if we assume that every ED visit made by those with part-year coverage was made while they were uninsured, clearly a conservative assumption, the resulting upper-bound estimate of the share of visits accounted for by uninsured people (24 percent) is still proportional to the upper bound of the percentage having no health insurance (22 percent).

While some may be surprised by this finding of proportional use, it is consistent with analyses of earlier MEPS data. In both 1997 and 1998 approximately 19 percent of adults ages 18–64 with insurance and 20 percent of those without insurance had one or more ED visits.⁷ Using 1997 and 1999 National Survey of American Families (NSAF) data, Stephen Zuckerman and Yu-Chu Shen found that, controlling for health status and other individual characteristics, the uninsured are significantly less likely to use the ED once or twice in a year than are the privately insured and are no different from the privately insured in their likelihood of being a frequent user.⁸

Use of outpatient services. Analysts would predict that because of access problems, including the barrier posed by high out-of-pocket prices, people without health insurance are less likely than people with insurance are to use the outpatient department for their medical care needs. The MEPS data support that prediction. We estimate that only 2 percent of outpatient visits are accounted for by the full-year uninsured and an additional 7 percent by the part-year uninsured, well below the proportion of the population lacking coverage all or part of the year.

Hospital Utilization Changes In Response To Changes Among The Uninsured

Top Study Data Who Is Walking Through... Hospital Utilization Changes In... Results Of Three Simulations Discussion And Policy... Editor's Notes NOTES

 
How changes among the uninsured in the next five to ten years will affect hospitals depends on how the size and the composition of the uninsured population change. Specifically, who is moving from being insured to being uninsured? What are the hospitalization patterns of these people, and how will their use of services change when they lose coverage? To begin to answer these questions, we need to look at who lacks health insurance now and what their patterns of hospital use are, and then make predictions about changes in public and private policies and the resulting changes in coverage and use.

In 1999, 215 million people had health insurance for the entire year. Of those, thirty-four million had only public coverage the entire year. For the remaining 181 million fully insured people, at least part of their coverage was private—that is, either they had private insurance coverage for at least part of the year, or they had both private and public coverage simultaneously (for example, a Medigap policy to supplement Medicare). Thirty million people had coverage of any kind for only part of the year, and thirty-one million were without coverage the entire year.

As noted above, most analysts predict that those numbers will increase, and there are early signs that this prediction is correct. The most recent estimates of the size of the uninsured population come from the Census Bureau’s 2002 Current Population Survey (CPS) (estimates for 2001). The number of people without coverage grew more rapidly then the number with coverage, yielding a net increase in the number without coverage: 14.6 percent lacked coverage during all of 2001, compared with 14.2 percent in 2000.⁹

Possible private-sector changes. Most of the decline in coverage reported in the CPS stemmed from a one-percentage-point decrease from 2000 to 2001 in the number of people with employer-sponsored insurance. Because of current increases in the number of unemployed workers and in the number of workers in part-time or temporary positions, we expect the number of people with employer coverage to decline further. In addition, health care costs continue to rise, leading to higher private-sector insurance premiums. In response, some small businesses, already less likely to offer insurance, are dropping coverage, and many large firms are raising their workers’ out-of-pocket premiums.¹⁰ We have evidence that employees will be sensitive to these increased premiums and that more employees at the margin will drop coverage.¹¹ As a result, we expect to see a decrease in the number of employees eligible for employer-sponsored coverage, a decrease in some employers’ offer rates, and a decrease in workers’ take-up rates.

It is reasonable to expect that members of the population groups more likely to gain insurance during the late 1990s, a period of low unemployment and relatively steady insurance premiums, would be those more likely to lose coverage now. Analysis of the 1996–2000 MEPS data revealed that although coverage rates differ among different age groups, the relative rates remained constant.¹² Similarly, although the percentage of part-time workers without coverage declined from 1996 to 1999, so did the percentage of uninsured full-time workers, yielding no change in the relative rate. However, some groups were more likely to gain coverage. While the percentage of uninsured workers in medium-size and large establishments remained constant from 1996 to 2000, the percentage of uninsured workers in small establishments (fewer than twenty-five employees) declined from 1996 to 1999, then went back in 2000 to the 1996 level.

Possible public-sector changes. At the same time that the private sector faces an economic downturn, state and federal budgets are being pinched, tax revenues are declining, and many public program costs are rising. Various state and federal programs are once again being reviewed, and health care program cuts are being considered in virtually every state.

During the 1990s most states expanded their Medicaid programs along various dimensions. In 1997 the State Children’s Health Insurance Program (SCHIP) was created, allowing states to further expand their coverage of children in low-income families.¹³ From 1990 to 2000 the number of Medicaid enrollees almost doubled, from approximately twenty-two million to more than forty-four million.¹⁴ Although much of that growth was attributable to expanding eligibility for children and pregnant women who were not receiving cash welfare, the program has grown to include other groups as well.¹⁵

However, as Alan Weil put it recently, "After a good run of five years, Medicaid is back in the crosshairs."¹⁶ Faced with tight budgets, historically high deficits, double-digit increases in Medicaid spending, and insufficient evidence that managed care can control Medicaid’s costs, many states may decide to roll back some of the expansions of the previous decade.¹⁷ As noted above, to date most of these contractions have affected low-income working adults and their families. The same analysis of 1996–2000 MEPS data indicated that, although not a statistically significant change, only workers earning less than $10 an hour experienced a decrease in coverage rates, in part as a result of Medicaid expansions.¹⁸

"It is not clear whether the uninsured’s lower inpatient use also reflects lack of access or less need."

The Social Security eligibility age is already going up, from age sixty-five to age sixty-seven, with those first affected turning sixty-five this year. For those born in 1938 who decide to retire when they are eligible for regular Social Security benefits, the wait will be two months longer than before—until two months after their sixty-fifth birthday. Although some policymakers have proposed raising the eligibility age for Medicare, it is not affected by the current Social Security changes and will remain at sixty-five.¹⁹ However, given dual pressures on revenues and expenditures, eligibility for Medicare could be pushed back as well.

Affected population groups’ hospital use. We divided the population into four groups for our analysis: elderly (age sixty-five or older); nonelderly and disabled; pregnant; and nonelderly, nonpregnant, and nondisabled.²⁰ We considered a nonelderly person to be disabled if he or she received Medicare or Supplemental Security Income (SSI) because of disability during the year. Although MEPS does not have a convenient variable for pregnancy, we considered a woman pregnant during 1999 if she used any inpatient, ED, or outpatient services related to pregnancy.²¹ Finally, for one of our scenarios, we also divided children and nonelderly adults according to whether they came from a household at 125 percent or more of the federal poverty level.

The hospital utilization rates of these population groups differ greatly. The elderly have the highest rate of outpatient department use, and the nonelderly disabled have the highest rate of inpatient and ED use (Exhibit 3). Both groups were more likely than the nonelderly, nonpregnant, and nondisabled group was to have multiple hospital visits in all three categories (data not shown). Approximately 10 percent of the nonelderly disabled population had multiple visits in all three categories, while 5 percent or less of the elderly had multiple inpatient and ED visits and 15 percent had multiple outpatient visits. In contrast, only 4 percent of the nonelderly, nonpregnant, and nondisabled group had multiple outpatient visits, 2 percent had multiple ED visits, and less than 1 percent had multiple inpatient visits during the year. The overwhelming majority of the nonelderly, nonpregnant, and nondisabled group are healthy working adults and their dependents.

Results Of Three Simulations

Top Study Data Who Is Walking Through... Hospital Utilization Changes In... Results Of Three Simulations Discussion And Policy... Editor's Notes NOTES

 
Using these base data, we performed three simulations, looking at changes in utilization by the three population groups more likely to experience a decrease in health insurance coverage because of private- and public-sector changes.

Private-sector changes. Extrapolating from the one-percentage-point drop in employer-sponsored coverage between 2000 and 2001, we tested a scenario in which 5 percent of the nonelderly, nonpregnant, nondisabled group with employer-sponsored coverage lose their full-year coverage over the next five to ten years. We took a random draw of eight million of the 164 million in this group (5 percent of the total) and randomly assigned four million into the part-year coverage group and the other four million into the full-year uninsured group, reflecting the current fifty-fifty allocation of those having at least some time without coverage.

Although not everyone faces the same probability of losing coverage, the 1996–2000 MEPS data suggest that the probability of losing employer coverage does not vary by age group or full/part-time work status. Rather, small-firm workers are more likely to lose coverage. Predictions according to hourly pay are less clear.²²

Taking a random draw of those currently covered by employer-sponsored insurance will introduce bias in our estimates if hospital use varies across the same groups experiencing differential loss of coverage. We do not know the size of the establishment for individual workers, although we do know income. We tested and found no significant difference across income groups in hospital discharge rates or ED use. Lower-income workers are more likely to use the hospital outpatient department, perhaps reflecting a higher tendency of high-income insured workers to use private physicians’ offices for ambulatory care.

As a result of our simulation, there was a 13 percent increase in the number of part-year uninsured people (from thirty to thirty-four months) and an equal increase in the number of full-year uninsured people. Although this seems large, the percentage of the total population in each of these categories increased only one percentage point, from 13 percent to 14 percent.

There are two ways to simulate the change in hospital use for this group. One is to assume that utilization reflects only need and that people’s behavior will not change when their insurance coverage status changes. That is, that demand is perfectly inelastic, impervious to the out-of-pocket price. In this case, we can simply add current utilization figures to those of the existing group of uninsured people. The other is to assume that utilization reflects a response to out-of-pocket price as well as need (that is, that the price elasticity of demand is not equal to zero) and therefore that utilization will change when insurance coverage is lost. In this case, we simply inflate the utilization figures of the existing group to reflect an increase in the size of the group. In this way, we provide upper and lower bounds of what will actually happen.

Inpatient stays. The resulting increase in the share of inpatient stays accounted for by uninsured people is quite small (Exhibit 5). The share for full-year uninsured patients increases by less than one percentage point, whether we assume that they change their behavior or don’t change it. Including part-year uninsured people yields increases of less than two percentage points under both assumptions.

Outpatient visits. The change in outpatient shares accounted for by the uninsured is particularly sensitive to the assumptions being made. If newly full-year uninsured people do not change their behavior in response to losing coverage, there would be a nearly 40 percent increase in their share of outpatient visits, from 2.1 percent to 2.9 percent. We would not expect this to occur. Rather, we would expect demand to be sensitive to out-of-pocket price and therefore that those who lose coverage would reduce outpatient use. For those who now lack coverage all year, a more reasonable expected response would be a smaller increase in outpatient visits, from 2.1 percent to 2.4 percent.

Changes in Medicaid coverage. It is difficult to know precisely which people covered by Medicaid in 1999 were newly eligible because of the Medicaid expansions of the 1990s. Approximately twenty-one million children, eight and a half million adults in families, four million elderly people, and seven million blind and disabled people were covered in 1999.²⁵ We hypothesized that neither the pregnant nor the aged, blind, and disabled were going to lose their Medicaid coverage. It is more likely that states seeking to roll back Medicaid expansions would target the nonelderly, nonpregnant, nondisabled adults and children in families with incomes 125 percent or more above poverty.²⁶

There were approximately 6.7 million children under age twenty-one and 2.9 million adults who fit this category in 1999; 42 percent of those were insured all year with public coverage only, 20 percent were insured all year with some private coverage, and 38 percent were covered only part of the year. We converted those with public coverage only or part-year coverage into full-year uninsured, and those with some private coverage into part-year uninsured, and simulated what this effect would have on hospital use.

Some of these people might be able to obtain employer-sponsored insurance. If so, these conversions provide a worst-case scenario. Given the kinds of jobs such people can obtain and the simultaneous decline in small firms’ offer rates, it is not obvious that the crowding-out rates estimated by some during the 1990s would lead to a symmetric "crowding in" in response to Medicaid rollbacks. In addition, a study that looked specifically at the parental Medicaid expansions found no major crowding-out effects on private coverage.²⁷

The result was a 26 percent increase in the number of people who were uninsured all year and a 6 percent increase in the number of those with part-year coverage. Those moving into the part-year covered group previously had some private coverage and, mirroring the behavior of that group, were less likely to use the ED and more likely to use the outpatient department. Those moving into the full-year uninsured group were more likely to use all three categories of hospital services. Therefore, if these people kept their same hospital utilization patterns even though they were losing Medicaid coverage, the share of all three types of visits would go up by three to four percentage points (Exhibit 5). As discussed before, we would expect these people to reduce their use of the outpatient department, which suggests that a more modest increase (one percentage point) is more likely.

Changes in Medicare coverage. Following the changes already put into place for Social Security’s retirement age, we simulated what would happen if the Medicare eligibility age were to rise from sixty-five to sixty-eight. There were 5.6 million people ages 65–67 covered by Medicare in 1999. Two-thirds also had some private coverage. It is reasonable to expect some of these people to postpone retirement and retain their employer coverage, others to become dependents on their spouse’s employer-sponsored policy, and still others to qualify for Medicaid.

For this simulation, we took those 5.6 million people and changed the distribution of coverage to mimic the distribution of those who were ages 62–64 in 1999, putting 74 percent into full-year coverage with some private coverage, 9 percent into full-year public-only coverage, 5 percent into part-year coverage, and 12 percent with no coverage the entire year. The result would add more than a half-million people ages 65–67 to the uninsured population. Although the effect on the utilization patterns, and potentially the health status, of these people could be quite serious, the numbers are just too small to have any measurable effect on the share of hospital visits by uninsured people, no matter what we assume about their behavior upon losing coverage or having it only part of the year.

Discussion And Policy Implications

Top Study Data Who Is Walking Through... Hospital Utilization Changes In... Results Of Three Simulations Discussion And Policy... Editor's Notes NOTES

 
Using 1999 MEPS data, we estimated that the increase in the burden to hospitals resulting from a decrease in insurance coverage, on average, would be relatively small for most hospitals. The actual effect of this burden on any one hospital would depend on its current share of uninsured patients, its reliance on private insurance and Medicaid payments, its mix of inpatient and outpatient revenues, and the marginal cost of treating those additional uninsured patients. Just as the burden of uncompensated care is not evenly distributed among hospitals, so would any increase in that burden not be evenly distributed. Also, because hospitals with disproportionately high shares of uncompensated care are more likely to have high shares of Medicaid patients, any cutbacks in Medicaid coverage would have larger effects on these hospitals.

Our results suggest that even large increases in the numbers of uninsured people will not cause overwhelming numbers of uninsured patients to walk through the doors of most hospitals. However, these numbers mask considerable, and important, variation in the share of patients who are uninsured among hospitals.

Our data also indicate that the uninsured as a group are less likely to use hospital outpatient services than nonelderly, nonpregnant, and nondisabled adults and children with coverage. It is unclear whether those who lose coverage would continue to use hospital outpatient services, switch to the ED or freestanding health centers for nonemergency care, or stop seeking health care on an outpatient basis.

Editor's Notes

Top Study Data Who Is Walking Through... Hospital Utilization Changes In... Results Of Three Simulations Discussion And Policy... Editor's Notes NOTES

 
An earlier version of this paper was presented at the conference, "The American Hospital: What Does the Future Hold?" in Washington, D.C., 21 April 2003.

Catherine McLaughlin is a professor in the Department of Health Management and Policy at the University of Michigan in Ann Arbor and director of its Economic Research Initiative on the Uninsured. Karoline Mortensen is a research assistant with the initiative.

NOTES

Top Study Data Who Is Walking Through... Hospital Utilization Changes In... Results Of Three Simulations Discussion And Policy... Editor's Notes NOTES

 

1. M.Chernew, D. Cutler, and P.S. Keenan, "Rising Health Care Costs and the Decline in Insurance Coverage," Economic Research Initiative on the Uninsured Working Paper no. 8, September 2002, www.umich.edu/~eriu/pdf/wp8.pdf (21 August 2003); J.Gabel et al., "Job-Based Health Benefits in 2002: Some Important Trends," Health Affairs (Sep/Oct 2002): 143–151; T.Gilmer and R. Kronick, "Calm before the Storm: Expected Increase in the Number of Uninsured Americans," Health Affairs (Nov/Dec 2001): 207–210; J.Holahan and M. Pohl, "Changes in Insurance Coverage: 1994–2000 and Beyond," 3 April 2002, www.healthaffairs.org/WebExclusives/Holahan_Web_Excl_040302.htm (18 August 2003); and L.Ku et al., "Proposed State Medicaid Cuts Would Jeopardize Health Insurance Coverage for One Million People" (Washington: Center on Budget and Policy Priorities, 6 January 2003).
2. Henry J.Kaiser Family Foundation/Health Research and Education Trust, Employee Health Benefits: 2002 Annual Survey (Washington: Kaiser/HRET, 4 September 2002).
3. Ku et al., "Proposed State Medicaid Cuts."
4. J.Hadley and J. Holahan, "How Much Medical Care Do the Uninsured Use, and Who Pays for It?" 12 February 2003, www.healthaffairs.org/WebExclusives/Hadley_Web_Excl_021203.htm (18 August 2003); P.J.Cunningham and H.T. Tu, "A Changing Picture of Uncompensated Care," Health Affairs (July/Aug 1997): 167–175; L.E.Fishman, "What Types of Hospitals Form the Safety Net?" Health Affairs (July/Aug 1997): 215–222; and K.E.Thorpe, C.S. Florence, and E.E. Seiber, "Hospital Conversions, Margins, and the Provision of Uncompensated Care," Health Affairs (Nov/Dec 2000): 187–194.
5. The top decile equates to the top 100 hospitals. We exclude five outliers with noticeably higher shares of uninsured visits than the other ninety-five (up to 77 percent) and significantly lower Medicare and, in two cases, Medicaid, shares.
6. While it is possible to develop a month-to-month link for each person between coverage and use of each type of hospital service and then denote whether the hospital visit occurred while uninsured for those with part-year coverage or while covered with private or public insurance for those with mixed coverage during the year, we have not done so.
7. U.S. Department of Health and Human Services, Health, United States, 2001 (Washington: DHHS, 2001), 274–275, Table 79.
8. S.Zuckerman and Y. Shen, "Characteristics of Occasional and Frequent Emergency Department Users: Do Insurance Coverage and Access to Care Matter?" Medical Care (forthcoming).
9. U.S. Census Bureau, "Health Insurance Coverage: 2001," 30 September 2002, www.census.gov/hhes/hlthins/hlthin01/hlth01asc.html (18 August 2003).
10. Kaiser/HRET, Employee Health Benefits: 2002 Annual Survey.
11. Chernew et al., "Rising Health Care Costs."
12. J.Rhoades, J. Vistnes, and J. Cohen, The Uninsured in America: 1996–2000, MEPS Chartbook no. 9, Pub. no. 02-0027 (Rockville, Md.: Agency for Healthcare Research and Quality, 2002).
13. L.Dubay, I. Hill, and G. Kenney, "Five Things Everyone Should Know about SCHIP," New Federalism Issues and Options for States, Series A, no. A-55 (Washington: Urban Institute, October 2002).
14. Centers for Medicare and Medicaid Services, A Profile of Medicaid: Chartbook 2000 (Baltimore: CMS, [pub. date?]), Figure 1.7; and CMS, "MSIS Statistical Reports for Federal Fiscal Years 1999 and 2000," 29 May 2003, www.cms.hhs.gov/Medicaid/msis/msis99sr.asp (18 August 2003).
15. A.Weil, "There’s Something about Medicaid," Health Affairs (Jan/Feb 2003): 13–30.
16. Ibid., 14.
17. Ku et al., "Proposed State Medicaid Cuts."
18. Rhoades et al., "The Uninsured in America."
19. A.J.Davidoff and R.W. Johnson, "Raising the Medicare Eligibility Age: Effects on the Young Elderly," Health Affairs (July/Aug 2003): 198–209.
20. We were not able to identify which workers (and their dependents) work for firms more likely to stop offering insurance. Nor were we able to identify which current Medicaid recipients were newly eligible because of recent expansions.
21. MEPS gives the reason for each visit. We recognize that we may have missed some pregnant women who used no hospital services either because they were newly pregnant at the end of 1999, or because they received no prenatal care during the year and did not have a hospital delivery. Given that our interest is in identifying the share of hospital services going to those without insurance and predicting changes in that share, this measurement error is benign.
22. Rhoades et al., "The Uninsured in America."
23. HHS, Health, United States, 2001, 274–275, Table 79. The near-poor are those with incomes of 100–199 percent of poverty.
24. Ibid.
25. Kaiser Commission on Medicaid and the Uninsured, "The Medicaid Program at a Glance," January 2001, http://www.kff.org/content/2001/2004b/2004b.pdf (21 August 2003).
26. Although several states specifically targeted low-income parents, studies have shown that children whose parents are not eligible for coverage are less likely to be enrolled. See, for example, A.Aizer and J. Grogger, "Parental Medicaid Expansions and Child Health Insurance Coverage," 8 May 2003, www.jcpr.org/conferences/health_policy/aizer_grogger.pdf (18 August 2003).
27. Ibid.

                      What's this?

This article has been cited by other articles:

				M. F. Newton, C. C. Keirns, R. Cunningham, R. A. Hayward, and R. Stanley Uninsured Adults Presenting to US Emergency Departments: Assumptions vs Data JAMA, October 22, 2008; 300(16): 1914 - 1924. [Abstract] [Full Text] [PDF]