Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 95
5 The Cost of the Additional Care That the Uninsured Would Use If They Had Insurance Coverage Providing health insurance would increase the life expectancy and reduce the morbidity of those who now lack it. In Care Without Coverage, the Committee concluded that those without coverage have higher morbidity and decreased life expectancy as a result of being uninsured. Insurance improves the health and wellbeing of the insured by increasing access to preventive services, timely care, and medical treatment. This chapter presents benchmark estimates of the additional monetary cost of the health care that those without coverage would be expected to use if they had coverage. The more timely care, better prevention, and better access to health services are the mechanisms by which insurance improves the health and well-being of the insured. Here the Committee considers estimates of the costs of the additional care that would be provided if everyone in the United States had continuous health insurance coverage. The costs of this “additional care,” the difference between the amount of health care services that the uninsured actually use and what they might be expected to use if they had coverage, are new economic costs of expanding coverage to the entire U.S. population. The Committee reviews and compares the results from one earlier and two recent statistical analyses of differences in utilization and expenditures among those who are uninsured, have private health insurance, and have public coverage, adjusting for observable differences among these populations. These analyses help predict the new costs (at the societal level) of the additional services that those who are now uninsured would use if they gained health insurance. The cost projections do not speak to the distribution of new costs among taxpayers, enrollees, and other payers nor do they make any assumptions about redistribution of present cost burdens for care provided to the uninsured among payers and providers of that care. Each set of projections of the
OCR for page 96
increase in costs is predicated on there being no major structural changes or reforms in the way health care is financed and delivered. The forecasts also assume no major differences in the mix of plans that would cover the uninsured from those which now cover the privately or publicly insured populations, in terms of scope of benefits, extent of coverage (levels of deductibles, copayments, stop losses, and other cost-sharing elements), or utilization review and other care management practices. In addition, the estimates that follow assume that the current capacity of the health care system is sufficient to meet the increased demand without generating additional inflation in health care prices. This assumption may not hold in those localities with particularly large uninsured populations and limited health care delivery resources. In any case, estimates of inflation in health care prices would depend on the particular features of a proposed health care reform, including any explicit cost control elements. The possible alternatives are too numerous to be considered in this report. In what follows, the Committee presents results from the studies by Long and Marquis (1994), by Miller, Banthin, and Moeller (2003), and by Hadley and Holahan (2003b). We do not review the older literature on comparisons of alternative health plans because the estimates depend on very specific features and are usually targeted at the cost of the entire reform proposal, not just the increase in cost for covering those currently uninsured. Typically, the latter is much smaller than the former. We also do not review studies of the effects of cost sharing on demand for health services because there are a number of reviews in the literature that evaluate the response to out-of-pocket costs; see Zweifel and Manning (2000) for one of the most recent. LONG AND MARQUIS ESTIMATES, 1994 Stephen Long and Susan Marquis (1994) predicted the additional ambulatory and hospital care that the population without health insurance could be expected to use if they had health insurance. Their work was based on data from the middle to late 1980s from three nationally representative surveys: the National Medical Expenditure Survey (NMES), the Survey of Income and Program Participation (SIPP), and the Health Interview Survey (HIS). Based on their estimates of differences in use, they predicted the utilization experience that uninsured people would be expected to have if they were in a typical employment-based plan. The 1987 NMES provided the uninsured population sample, updated by the Census Bureau’s Current Population Survey (CPS) to account for changes in the size and mix in the uninsured population between 1987 and 1992. The authors termed the difference in utilization for this standard population in the uninsured and insured states as the “access gap” attributable to lacking coverage. The model assumed that other aspects of health care financing and delivery, such as the mix of prepaid and fee-for-service health plans, cost-sharing provisions, and scope of benefits, re-
OCR for page 97
mained unchanged. The model assumed no changes in unit prices for services as a result of the increased coverage or any cost-containment efforts. Long and Marquis reported that uninsured adults have approximately 60 percent of the outpatient visits and 70 percent of the hospital days annually as they would if they had been insured. Uninsured children were estimated to use ambulatory services at about 70 percent of the rate that they would have if insured, and to use hospital days at 80 percent of the insured level. Adults and children in fair or poor health had a greater “access gap” than those with good or excellent health (i.e., with health insurance coverage they could be expected to increase their use of services relatively more than those in better baseline health). Long and Marquis estimated that the incremental cost of ambulatory and hospital inpatient services to the population gaining health insurance would amount to $20 billion in 1993 dollars (about $ 28.6 billion in 2001 dollars), nearly a 50 percent increase in the spending for this population (the uninsured) for these services. This amount represented 2.2 percent of total national spending for health care in 1993. Adjusting the estimates for the difference in the number of uninsured in 2001 and in 1993, the estimate grows to $35 billion, assuming no differences in the mix of demographic, health, and socioeconomic characteristics of the uninsured between the late 1980s and 2001. MILLER, BANTHIN, AND MOELLER ESTIMATES Edward Miller, Jessica Banthin, and John Moeller (2003), economists at the Agency for Healthcare Research and Quality (AHRQ), used more recent data to develop estimates of the annual cost of expanding coverage to all of the currently uninsured updated to 2002. Their model estimates reflect the health status, utilization, and expenditures of respondents in the 1996, 1997, and 1998 Medical Expenditure Panel Survey (MEPS). In order to produce current cost estimates, the sample population weights from the 3 years were adjusted to those of the March 2002 CPS, and 1996, 1997, and 1998 expenditures were calibrated by type of service and source of payment to the 2002 National Health Accounts to reflect the growth in expenditures between the sample year and 2002. This pooled MEPS sample produced estimates of roughly 34 million persons under age 65 uninsured for the entire year and 31 million additional persons uninsured for some part of the year. Because the uninsured, privately insured, and publicly insured differ in many characteristics beyond their insurance status, the authors adjusted differences in the estimated costs of services for differences in their age, race, sex, proportion of women of child-bearing age, family income and size, education, region, urbanicity, disability, health and decedent status, body mass index, and number of chronic conditions. They used a standard set of models that are employed to deal with the very skewed nature of expenditure data in order to make their estimates robust to extreme cases (Duan et al., 1983). The authors developed alternative models to produce upper- and lower-
OCR for page 98
bound estimates of the incremental service costs of expanding coverage to the uninsured. The lower-bound model was based on the experience of those who lacked health insurance for a full year and were not eligible for either private or public insurance. To study this case, they limited the “uninsured” sample to respondents who were uninsured for the full year, who had no offer of workplace coverage, and who were not eligible for Medicaid. The authors used the experience of this subgroup that did not have an opportunity to obtain coverage to minimize the correlation of unobserved differences in the propensity to use health care services with actual health insurance status. The authors reasoned that, because the population of uninsured individuals who had an offer of or eligibility for coverage includes individuals who are least likely to use health care (and thus forgo an opportunity to take up coverage), averaging in their utilization experience with that of the “insured” group reduces the difference between the utilization of the uninsured group and the insured group. This makes the estimated incremental costs of insuring the uninsured less than they would be otherwise. Those uninsured without an offer of or eligibility for coverage accounted for just one-third (32 percent) of all persons who were uninsured for all or part of the year. The remaining uninsured were included in either a private insurance (group or individual) or Medicaid category if they had that coverage for at least one month or if they had an offer of workplace coverage or qualified for but were not enrolled in Medicaid. In order to predict the incremental expenditures under universal private or public health insurance, the authors “scaled up” the projected expenditures for the subgroup of uninsured without an offer of or eligibility for coverage to account for the entire population of individuals uninsured for part or all of a year. The second model providing an upper-bound estimate of the increased utilization under universal coverage used as the “uninsured” group all of those individuals who were uninsured for any time during the year regardless of their eligibility or ineligibility for private or Medicaid coverage. Their predicted utilization was based on the experience of only those individuals covered for the full year by either private group insurance or Medicaid, thus maximizing the differences in service use and costs between the model’s uninsured and insured groups. From these models, the authors estimated the cost of expanding coverage to each of four subgroups of the uninsured: full-year uninsured, who either had or did not have an offer or eligibility for coverage, and part-year uninsured, who either had coverage for less than 6 months or for 6 through 11 months. Expenditures for those uninsured for part of the year only, who started off with much higher annual health care spending, increased much less over the baseline, ranging from a 3 percent decrease in spending for those uninsured for less than 6 months assuming Medicaid coverage, to a 58 percent increase in baseline spending for those uninsured 6 months or more, assuming private coverage. Providing private coverage to those who currently are not eligible for either public or private coverage would increase their per capita annual expenditures by 205 percent (by $1,911 in 2002 dollars) of their baseline (uninsured) expenditures of $934. Alternatively, expanding Medicaid coverage to include these uninsured
OCR for page 99
TABLE 5.1 Simulated Medical Spending per Uninsured Person: Two Alternative Benchmarks by Length of Time Uninsured and Existence of an Offer or Eligibility for Coverage (2002 dollars) Simulated Spending Baseline Insurance Status Baseline “Average” Private Group Coverage “Average” Medicaid All Uninsured $1,217 $2,302 (89%)a $1,962 (61%) Part-Year Uninsured < 6 months $1,779 $2,028 (14%) $1,732 (-3%) = 6 months $1,350 $2,128 (58%) $1,815 (35%) Full-Year Uninsured With offer or eligibility $778 $1,978 (154%) $1,685 (117%) No offer or eligibility $934 $2,844 (205%) $2,421 (159%) aPercentage increase from baseline spending level. SOURCE: Miller et al., 2003. Data from 1996, 1997, and 1998 MEPS. individuals would increase per capita expenditures somewhat less (by about 160 percent or $1,487 in 2002 dollars) over baseline expenditures. See Table 5.1 for the results for each of the “uninsured” subgroups: ineligible, eligible but not covered, covered for less than 6 months, and covered for 6 months or more. Table 5.2 shows the projected expenditure increases in the aggregate for the four subgroups of the uninsured and overall. The estimated incremental health expenditures for all groups of the uninsured, assuming a pattern of utilization and spending that matched the average of all group private insurance, range from $44.9 to $57.4 billion for 2002 and, assuming utilization and spending comparable to the national average per capita Medicaid experience, range from $35.1 to $38.1 billion. These estimates deduct the estimated value of in-kind uncompensated care (that is, uncompensated care for which no revenue support stream can be identified) provided to the uninsured of $13.5 billion under the private plan alternative and of $10.6 billion under the Medicaid alternative. Miller and colleagues also estimated a range of values for the new costs that would be associated with administering the newly insured health services. They base their estimates on national health account estimates of the percentages of public and private health expenditures spent on insurance administration (4.5 and 8.3 percent, respectively). Administrative costs for the incremental services that the uninsured would use with health insurance range from $3 billion based on a Medicaid expansion to $9–$10 billion with an expansion based on private coverage.
OCR for page 100
TABLE 5.2 Simulated Incremental Spending for All Uninsured: Two Alternative Benchmarks (2002 dollars, in billions) Simulated Spending “Average” Private Group Coverage “Average” Medicaid All Uninsured Lower bound $44.9a $35.1b Upper bound $57.4a $38.1b aIncremental medical spending includes a reduction of an estimated $13.5 billion in in-kind uncompensated care. bIncremental medical spending includes a reduction of an estimated $10.6 billion in in-kind uncompensated care. SOURCE: Miller et al., 2003. Data from 1996, 1997, and 1998 MEPS. HADLEY AND HOLAHAN ESTIMATES Jack Hadley and John Holahan (2003b), economists at The Urban Institute, conducted a similar analysis using somewhat different modeling and working assumptions. These authors used the same 3-year pooled sample of MEPS respondents for 1996–1998 to simulate the costs of coverage for the uninsured as they had used to estimate actual health expenditures for the uninsured (as described in Chapter 3). The larger sample from pooling 3 years of data allowed them to examine the effects for children and adults separately; some of the literature suggests that patterns of utilization for children and adults respond differently to cost sharing (Valdez et al., 1986; Newhouse et al., 1993). Hadley and Holahan made adjustments to the MEPS data comparable to those the AHRQ economists had made. They inflated the expenditures to 2001 levels and adjusted for undercounting of uncompensated care in MEPS using National Health Accounts statistics. Unlike the analysis by Miller and colleagues, Hadley and Holahan limited their sample of privately insured people to those with family incomes less than 400 percent of the federal poverty line. Thus, they assumed that the uninsured’s response to insurance will be the same as that for other low- to moderate-income individuals. They used a similar set of covariates to adjust for observable differences between the insured and uninsured. Their statistical model was based on the generalized linear model version of the two-part model (introduced by Blough et al., 1999). They treated an individual as uninsured if he or she was uninsured for any part of the year and they explicitly controlled for the fraction of the year without insurance. Thus, they made no distinctions based on being eligible for but not taking up public or private coverage. Hadley and Holahan considered two different scenarios for health insurance
OCR for page 101
coverage for the currently uninsured. In the first, they assumed that the uninsured would be covered by the same mix of private health insurance plans as those of low- to moderate-income people with private insurance coverage for the full year. In the second, they assumed that the uninsured would have the same experience as people with public insurance coverage (e.g., Medicaid or the State Children’s Health Insurance Program) for the full year. In both cases, the estimation and the predictions assumed that observable differences, other than insurance status, did not change. For each alternative, they pooled the data from the uninsured with either the publicly insured or the privately insured to estimate the parameters of the model, but they did not pool all three groups simultaneously. This analysis accounted for the length of time those who were uninsured during the year had any public or private insurance coverage, demographic characteristics (including age, race, sex), education, family income relative to poverty, marital status, and census region, and health-related characteristics including health status, mental health status, functional status, acute and chronic conditions, and whether the respondent died or was institutionalized during the year. Table 5.3 presents the per capita spending projected for the uninsured in 2001 assuming that they had either the same mix of private insurance policies or the same mix of public insurance coverage as their insured counterparts. Based on their estimates, Hadley and Holahan predict that per capita expenditures for adults uninsured for a full year would increase to 265 percent of their baseline amount (from $1,158 annually to $3,069) modeled on private group coverage utilization and spending experience, and would increase to 209 percent of baseline (to TABLE 5.3 Simulated Medical Spending per Uninsured Person: Two Alternative Benchmarks, by Age and Baseline Insurance Status (2001 dollars) Simulated Spending Age and Baseline Insurance Status Baselinea (Actual) “Average” Private Coverage “Average” Public Coverage All Uninsured $1,383 $2,676 $2,121 Full-year uninsured 989 2,650 2,068 Part-year uninsured 1,813 2,705 2,178 Uninsured Adults $1,644 $3,187 $2,568 Full-year uninsured 1,158 3,069 2,419 Part-year uninsured 2,240 3,331 2,751 Uninsured Children (<18) $733 $1,408 $1,008 Full-year uninsured 475 1,374 996 Part-year uninsured 943 1,434 1,016 aDoes not include value of uncompensated care. SOURCE: Hadley and Holahan, 2003b. Data from 1996, 1997, and 1998 MEPS.
OCR for page 102
TABLE 5.4 Simulated Total Spending: Two Alternative Benchmarks by Baseline Insurance Status (2001 dollars, billions) Simulated Spending Baseline Insurance Status Baseline Spendinga “Average” Private Coverage “Average” Public Coverage Total Change in Spending, $ (%) Total Change in Spending, $ (%) All Uninsured $98.9 $167.6 $68.7 (69.5) $132.8 $33.9 (34.3) Full-year uninsured 40.6 86.7 46.1 (113.5) 67.6 27.0 (66.5) Part-year uninsured 58.3 80.9 22.6 (38.8) 65.2 6.9 (11.8) aIncludes value of uncompensated care. SOURCE: Hadley and Holahan, 2003b. Data from 1996, 1997, and 1998 MEPS. $2,419) modeled on average national Medicaid experience. Likewise, for children who are uninsured for the full year, spending would more than double from $475 per capita to between $996 and $1,374 per capita, depending on whether utilization and spending is modeled on Medicaid or private group insurance. Although the baseline estimates of per capita spending are not adjusted to include the value of uncompensated care (which is undercounted in MEPS), the authors adjusted the total baseline cost estimate for medical care used by the uninsured to take into account uncompensated care. The projected increases in spending for children and adults who are uninsured only part of the year are smaller. They range from an 8 percent increase for part-year uninsured children based on Medicaid experience to a 49 percent increase for adults uninsured for part of the year if they had the average private group experience. Overall, health care expenditures for individuals uninsured for any amount of time during the year are projected to increase to $132.8 billion, 134 percent of the baseline of $98.9 billion in 2001, if their utilization and spending followed the experience of those with Medicaid coverage, and to $167.6 billion, 170 percent of baseline, if at average private group insurance levels. If we consider the increased (i.e., new economic) costs over the uninsured alternative rather than the total cost of additional benefits to the uninsured under universal coverage, then the increase in health care spending would be $33.9 billion for the public alternative and $68.7 billion if the private plan alternative is used (see Table 5.4). Hadley and Holahan note that their estimates can vary by as much as 10 to 20 percent based on issues of sample and model specification.
OCR for page 103
COMPARING RESULTS Finding: Estimates of the cost of the additional health care that would be provided to the uninsured once they became insured range from $34 to $69 billion per year, assuming no structural changes in the systems of health care financing or delivery, average scope of benefits, or provider payment. This incremental cost of services amounts to 2.8 to 5.6 percent of national spending for personal health care services in 2001. These three studies provide a range of estimates of what additional health care services for the uninsured will cost when they become insured. Adjusting the Long and Marquis estimates for differences in the cost of medical care between 1993 and 2001, and prorating the expenses for the larger number of uninsured in 2001, the Long and Marquis estimate amounts to about $35 billion per year. The Long and Marquis estimates are based on differences in visit rates and hospital days per capita, assuming that there are no differences in the cost per visit or the cost per inpatient day between the insured and uninsured. The two studies that used much more current data from the MEPS differed from the earlier analysis by Marquis and Long in that they looked directly at health care expenditures (including uncompensated care) rather than estimating costs from differences in outpatient and inpatient use rates. Thus the two MEPS-based studies allow for the possibility of differences in intensity of treatment per encounter, a possibility ruled out by the Long and Marquis approach. Although the details of the two MEPS studies differed in important aspects, the range of estimates from the two both suggest that the cost of providing those now uninsured with the extra health services used by those with coverage could range from about $34 billion to about $69 billion (Hadley and Holahan, 2003b; Miller et al., 2003). Locating a more specific estimate within that range would depend on whether the new coverage was like current public insurance plans or current private insurance plans. The range of estimates largely reflects differences between the expanded coverage through private and public plans, which in turn reflects significant differences in provider payment rates between private and public health plans, on average (Hadley and Holahan, 2003b; Miller et al., 2003). The differences among the sets of estimates are quite small and may be due to differences in statistical modeling. Long and Marquis used separate two-part models for inpatient and outpatient care to allow for differences in responses to insurance status. Hadley and Holahan used a similar two-part model and the AHRQ economists used a four-part model. Although no attempt has been made to find how much of the difference is due to differences in model specification, Hadley and Holahan note that their own results could change by as much as 10–20 percent based on alternative specifications.
OCR for page 104
CONCLUSION This chapter has addressed the cost implications of expanding coverage to those without insurance. This is the correlate to the costs of uninsurance: What might it cost society to provide health insurance coverage to the roughly 41 million Americans who lack it? A precise answer to this question would require specifying many features of the proposed scheme for universal health coverage, including the administrative framework, the scope of benefits covered, the level of cost sharing, the level of provider payments for covered services, and the mechanism that would finance the insurance coverage (e.g., general revenues, payroll tax, employer mandate). Different choices in each of these areas could dramatically change the estimates of the costs of providing universal coverage. The Committee’s objective here is not, however, to provide a precise estimate of the net new costs of covering the uninsured population. Rather, it is to provide a range of the likely new real resource costs, based on the best evidence to date, of extending to the uninsured coverage similar to that of individuals who now have public or private health insurance. These estimates should be treated as benchmarks. The Committee presents a range of estimates of the projected costs of covering those now uninsured that are plausible, generic, and illustrative. The incremental service costs are based on the costs of the coverage that accounts for the differences in morbidity and mortality between otherwise similar insured and uninsured Americans (IOM, 2002a,b). If these benchmarks are appropriate, then the evidence suggests that the costs of the additional health care that would be provided to the uninsured once they become insured will be on the order of $34 to $69 billion a year, assuming no other major changes in scope of benefits, provider payment, or the structure of the health care financing and delivery systems. This amounts to a 2.8 to 5.6 percent increase in spending for personal health care services for 2001. It is equivalent to between one-third and two-thirds of the 8.7 percent growth in national expenditures for personal health care services between 2000 and 2001 (Levit et al., 2003).
Representative terms from entire chapter: