Costs of Caring for Uninsured People in Maine by Stephen Zuckerman, Randall R

costs of caring for uninsured people in maine By Stephen Zuckerman, Randall R. Bovbjerg, Jack Hadley, and Dawn Miller

Commissioned by the maine health access foundation may 2007

Strategic solutions for Maine’s health care needs MeHAF is Maine’s largest private health care foundation. Through its grant and program support, the Foundation advances strategic solutions to improve health and health care in Maine with a special emphasis on addressing the needs of people who are uninsured and medically underserved.

The mission of the Maine Health Access Foundation (MeHAF) is to promote affordable and timely access to comprehensive, high quality health care, and improve the health of every Maine resident. costs of caring for uninsured people IN MAINE Prepared by Stephen Zuckerman, Randall R. Bovbjerg, Jack Hadley, and Dawn Miller of The Urban Institute for the MAINE HEALTH ACCESS FOUNDATION

Strategic solutions for Maine’s health care needs about the authors stephen zuckerman, ph.d., is a Principal Research Associate in the Health Policy Center of The Urban Institute. He received his doctorate in economics from Columbia University in 1983 and has studied health economics for over 20 years. His current research interests are state health policy including coverage expansion for adults, racial and ethnic disparities, Medicare physician payment and Medicaid managed care. Dr. Zuckerman has also worked on research related to the health care safety net, hospital rate setting, health care price indices and health system reform. Prior to joining the Institute, he worked at the American Medical Association’s Center for Health Policy Research. randall r. bovbjerg, j.d., is a Principal Research Associ- ate at The Urban Institute. He has 30 years of experience in public the urban institute and private insurance; state and local health policy; medical injury, health policy center liability, and patient safety; safety net issues; and state regulation. The Urban Institute is a nonprofit policy research organization Current research includes a planning grant to help the District of created in 1968 to sharpen thinking about America’s problems and Columbia expand insurance coverage and case studies of state medi- efforts to solve them, improve government decisions and their cal boards’ performance in physician discipline. implementation, and increase citizens’ awareness about important jack hadley, ph.d., is a Principal Research Associate at The Ur- public choices. Its research agenda includes national issues that ban Institute and a Senior Fellow at the Center for Studying Health reflect, respond to, and at times anticipate society’s changing needs. System Change. His research has emphasized statistical analysis of In recent years, Institute researchers have also begun analyzing various issues in health care organization and financing, including similar issues in developing countries, Eastern Europe, and the determinants of health insurance coverage, the consequences of lack Russian Federation. Researchers identify and measure social of health insurance, access to care, physician and hospital payment, problems, assess their solutions, spot trends, evaluate social and and the effects of managed care on health care delivery systems. economic programs and policy options, and offer technical assistance in policy and program development. dawn miller, ba, is a Research Assistant with The Urban Institute's Health Policy Center. Her work focuses on analyzing en- Within The Urban Institute, the Health Policy Center analyzes rollment and expenditures for the Medicaid program utilizing ad- trends and underlying causes of changes in health insurance ministrative data from the Medicaid Statistical Information System coverage, access to care, and use of health care services by the and the CMS Form 64. Recently, she has investigated public revenue entire U.S. population. Researchers address issues that arise from sources to fund care for the uninsured, and has been providing sup- the inevitable trade-offs among health care costs, access, and port to senior health policy researchers using data from the Current quality. The center’s focus has been on Medicare and Medicaid, Population Survey and the Medicare Current Beneficiary Survey. public insurance programs that were created to serve the elderly, the disabled, and low-income households. Institute researchers also have studied proposals to control costs, incentives built into public and private provider reimbursement mechanisms, reform alternatives for the long-term care system, and malpractice tort law and insurance.

iv costs of caring for uninsured people in maine table of contents about the authors iv

Acknowledgements 2

executive summary 3

introduction 5

uncompensated care costs: provider and program data 6

uncompensated care costs: household survey data 12

sources of public revenues to support uncompensated care 17

conclusion 24

appendix a: costs of uncompensated care provided through the 27 ryan white comprehensive aids resources emergency (care) act

appendix b: methodology for estimating medical care costs of 28 uninsured people

appendix c: computing predicted expenditures, if fully insured 31

end notes 44

costs of caring for uninsured people in maine Acknowledgements

The authors gratefully acknowledge the comments and advice provided by John Holahan and the research assistance provided by Joel Ruhter and Saad Ahmad.

Thanks also for people who assisted with data collection: Katharine Addicott, Portland Community Free Clinic; Sarah Gagné Holmes, Maine Equal Justice Partners; Sophie Glidden, DHHS; Geoffrey Greene, DHHS; Patty Hamilton, City of Bangor; Nancy Kane, Harvard School of Public Health; Elizabeth Kilbreth, Muskie School of Public Service; Kala E. Ladenheim, National Conference of State Legislatures; Kevin Lewis, Maine Primary Care Association; Sue MacKenzie, Office of MaineCare services; Trish Riley, Govenor’s Office of Health Policy and Finance; Marianne Ringel, DHHS; Ellen Schneiter, Bureau of the Budget; Gordon Smith, Maine Medical How can Maine Association; Connie Warren, Office of MaineCare Services; David Wilson, DHHS; David Winslow, Maine Hospital Association; Charlotte Woodcock, Franklin Health Access; and Carol Zechman, cope with the costs of CarePartners. The Maine Health Access Foundation funded this study. Any views caring for uninsured expressed are those of the authors and do not represent those of the Foundation, The Urban Institute, its sponsors or Trustees. Mainers? One theme Additional copies of this report are available upon request at no charge. Please contact the Maine Health Access Foundation at 207-620-8266 or [email protected], or download the report emerged—“Maine through the Foundation’s website at www.mehaf.org. is small, so we all work together.” executive summary In this report, we provide estimates of the costs of uncompensated care that the 124,000 uninsured Maine residents received in 2005 and the revenues that may have been available to offset these costs. We use two alternative approaches to estimating uncompensated care costs of uninsured people. In the first approach, we draw on data reported by health care providers and public programs. 2005 costs of uncompensated care for The second approach uses household survey data on health care uninsured people in maine expenditures and, as such, develops estimates based on information reported directly by uninsured people. Prior to enumerating the sources of funds that offset the uncompensated costs of care $7.5 for uninsured patients, we provide some background on the recent $8.7 evolution of health policy and health care delivery in Maine. The information on the funding for uncompensated care is drawn from federal, state and local budget reports as well as from trade association materials and a series of interviews with key respondents. $33.0 $78.7 According to our analysis of provider/program data, hospitals are the largest provider of uncompensated care to uninsured Mainers, accounting for $78.7 million in 2005. Office-based physicians provided another $10.2 million in uncompensated care. $10.2 Among government programs and providers, the largest amount of uncompensated care to uninsured Mainers came through Veterans’ Affairs hospitals and clinics ($33.0 million). Other government providers included community health centers ($8.7 million) and the Indian Health Service (IHS) ($7.5 million). Across all providers total $138 million (all figures in millions) for which we had data, we estimate that uninsured people in Maine received about $138 million in uncompensated care in 2005.

According to estimates based on household survey data collected maine hospitals by the Medical Expenditure Panel Survey (MEPS), uninsured office-based physicians adults received an average of $1,277 in care and uninsured children received $1,382 in care. Personal out-of-pocket spending is the veterans’ affairs hospitals & clinics largest source of payment for care received by uninsured people (37%). Since some uninsured people have coverage for a portion community health centers of the year, private insurance and Medicaid pay for 9% of their care. indian health service Workers’ compensation, an insurance system for paying the medical care costs of injured workers, pays for about 3% of care for uninsured people. The remaining catagories of reported spending represent uncompensated care and account for about 51% of the care received by uninsured Mainers. The first category of uncompensated care for uninsured Mainers is “other public sources” and accounts for 23% of the care. This includes care provided by facilities run by Veterans’ Affairs, other federal programs, and other state and local programs. The second category—care financed by other private payments and payments from unknown sources—is

costs of caring for uninsured people in maine responsible for 9% of care to uninsured Mainers. Donated often smaller or more fragmented pieces of the delivery system whose care from private sources of payment is the final category of revenue streams were so varied they were difficult to quantify. We uncompensated care and represents 19% of care to uninsured believe that most public revenues are in the quantifiable section, but Mainers. These three categories of spending on uncompensated recognize that it is important to understand the roles played by the care for Maine’s uninsured people totaled $81 million dollars range of providers. in 2005. Our estimate of public sector revenue currently supporting uncom- The estimates of uncompensated care received by uninsured people pensated care for uninsured Mainers is approximately $110 million. derived from provider/program data and household survey data However, because of the more fragmented pieces of the system that are not identical. The MEPS data used for the household survey we identified but were not able to quantify, more than $110 mil- estimates are known to understate spending relative to the National lion is likely available to fund care to uninsured Mainers. About Health Expenditure Accounts (NHEA) and, although we made 93 percent of the quantifiable public support in Maine comes from adjustments to align the MEPS and NHEA aggregates, it is possible the federal government with 7 percent from states and localities. that the MEPS estimates could still understate costs. Moreover, the The picture in Maine is considerably different from the national MEPS data are from all surveyed households in the Northeast re- average, which shows that state and local funds typically constitute gion as opposed to only Maine. Although we adjusted the Northeast about one-third of public support. The reason that Maine appears data so that our estimates reflect the demographic characteristics to be so different, in part, is that the state has shifted funding of Maine’s population, it is also possible that there may be unique toward a coverage expansion rather than providing subsidies for features of Maine’s state policies and delivery system that are not uncompensated care. adequately captured in these adjustments. In aggregate, these estimates suggest that the 2005 public revenues The provider/program data are also imperfect. We are aware that in the system to support care to uninsured people are somewhat not all providers who care for uninsured patients could be included below the costs of the care they receive. Beyond the magnitude in the cost estimates based on provider data (e.g., clinics other than of the estimates, perhaps the major point to take away from this Federally Qualified Health Centers, or FQHCs), but interviews report is that the presence of large numbers of uninsured people suggest that we have captured the bulk of providers’ costs of caring and their inevitable need to receive health care has resulted in a for uninsured Mainers. In addition, there are several data elements complex mosaic of government programs and private initiatives to used in the calculations that are measured imprecisely and some defray the costs of that care. In the absence of large public hospitals assumptions that influence the various components of the overall or subsidies to offset the costs of care to uninsured Mainers (such estimate. On balance, we think the household survey estimates as those often provided as Medicaid DSH payments), understand- should be viewed as a lower bound on the costs of care to uninsured ing how providers in Maine serve uninsured patients will require people in Maine, while the provider/program estimates may poten- further study. tially overstate the actual costs of care.

We also reviewed the range of funding that is available to providers to offset the costs of providing uncompensated care to uninsured Mainers. Our goal is to provide a snapshot of funding that subsidizes uncompensated care in 2005 for uninsured Mainers, as opposed to assessing any broader policy approaches or strategies the state may be pursuing (e.g. Dirigo Health Reform). Even after Maine’s coverage expansions, there are many uninsured people in Maine. Care for these people is provided by a patchwork of organizations, such as hospitals, clinics, and direct public providers, with funding coming from Medicaid, Medicare, federal grants and other federal programs, the state and localities. We also describe the myriad important, yet

costs of caring for uninsured people in maine introduction Maine had about 124,000 uninsured non-elderly residents during the 2004/2005 period, according to estimates based on the Cur- rent Population Survey.1 Although many studies have shown that uninsured people receive less care than people with insurance, evidence suggests that uninsured people receive health care services and that costs on the health care system can be substantial. One national study based on data from 2001 showed that the 41 million Americans who were uninsured that year received about $35 billion in uncompensated care from hospitals, physician, clinics and a variety of publicly funded programs.2 More recent estimates for the state of Massachusetts showed that, in 2004, the uninsured received between $900 million and $1.3 billion in uncompensated care in that state alone.3

In this paper, we focus on people in Maine who are uninsured Household survey responses may understate the amount of care and provide estimates of the amount of uncompensated care they received as a result of recall problems and, to a far greater extent received in 2005 and the revenues available to offset those costs. than in the case of provider data, it is difficult to assign a cost to the Following previous studies, we use two alternative approaches to care reported on in surveys. The household survey data report pay- estimating uncompensated care costs for uninsured people. In the ments made to providers as opposed to costs. These types of data first approach, we draw on data reported by health care providers issues mean that we have to make a number of assumptions and and public programs. These data come from cost reports, surveys adjustments to produce estimates of uncompensated care. Having and budget documents and can be viewed as an accounting-based two fundamentally different approaches will allow us to compare estimate of uncompensated care. The second approach uses house- one to the other as a way of assessing the plausibility of each set of hold survey data on health care expenditures and, as such, develops findings. We also estimate how much care uninsured people might estimates based on information reported directly by uninsured use if they had insurance. These estimates are based on a compari- people. In addition, we derive estimates of the amount of care that son of spending by those who are currently uninsured to those who uninsured people would use if they were somehow insured for the have health insurance for a full year, controlling for differences in full year. their characteristics.

We use two approaches in this study because we recognize that In addition to providing estimates of the costs of uncompensated developing these estimates of uncompensated care may be sensitive care for uninsured people in Maine, this paper enumerates the to the methods used and assumptions made in working with the funding sources that may offset some of those costs. We also pro- available data. Providers may not keep track of the specific costs vide background on the recent evolution of health policy and health being incurred on behalf of individual patients and, therefore, link- care delivery in Maine. Our goal is to provide a snapshot of funding ing costs to any particular class of patients is not straightforward. potentially available in 2005, not to assess any broader policy ap- In the case of budget data, it can be difficult to know what portion proaches or strategies the state may be pursuing. The information of spending is actually used to provide uncompensated care to on potential funding for uncompensated care is drawn from federal, uninsured people. state and local budget reports as well as from trade association materials and a series of interviews with key respondents.

costs of caring for uninsured people in maine UNCOMPENSATED CARE COSTS: Provider and Program Data

hospitals insured patients in Maine is below 50 percent.9 In the calculations that follow, we assume that 35 percent of bad debt is associated with in- To estimate the cost of care provided to uninsured people in Maine sured patients and adjust hospitals’ reported bad debt costs accordingly. by hospitals in 2005, we rely on data collected by the Maine Health Data Organization (MHDO), an independent executive state agency Table 1 provides a summary of these calculations. In 2005, total free that collects a wide range of clinical and financial health care infor- care charges amounted to $60.6 million and bad debt charges to mation and makes this information accessible to the public. As part an additional $123.0 million. After adjusting by the cost-to-charge of this role, MHDO requires each Maine hospital to complete a fi- ratio for each hospital (direct expenses divided by gross patient nancial data template. These data are compared with each hospital’s service revenue), we arrive at allowable free care costs of $34.0 million audited financial statements for accuracy.

Total costs for uncompensated care are calculated as follows. First, for each hospital we add the charges attributed to free care and The total hospital bad debt. Although accounting techniques vary, free care generally consists of services for which the hospital did not expect to receive payment; bad debt occurs when the hospital expects but does not uncompensated care receive payment. Bad debt may be incurred on uninsured as well as insured patients. We include bad debt associated with both emer- provided to uninsured gency and non-emergency services. The sum of charges for free care and bad debt is multiplied by each hospital’s cost-to-charge ratio, which is the ratio of total direct expense to gross patient service patients in 2005 is revenue.4 The resulting uncompensated care costs for each hospital are aggregated, resulting in the total cost of free care and bad debt for Maine in 2005.5 estimated to be

Finally, we adjust for the share of Maine’s uncompensated hospital care that is associated with individuals with insurance coverage. $78.7 million There is no data available on the share of free care provided to insured people in Maine. However since free care that is delivered in free care and $68.8 million in bad debt costs. The total cost of to insured patients is likely to account for only a small portion of uncompensated care in 2005 is $102.8 million. After netting out the costs, we omit this adjustment.6 Although most bad debt is thought share of bad debt attributable to insured patients, the cost of bad to be attributable to uninsured people, emerging evidence suggests debt drops by $24.1 million. We estimate the total costs of hospital that a portion of bad debt is owed by people with private health uncompensated care provided to uninsured patients in 2005 to be insurance.7 A 2004 survey of Maine hospitals produced estimates $78.7 million.10 that suggest about 50 percent of bad debt is generated by services provided to insured patients.8 However, only 17 Maine hospitals office-based physicians could breakout bad debt between privately insured and uninsured, Physicians practicing in their offices also provide a substantial self-pay patients. Given this low level of response and the fact that amount of uncompensated care to uninsured patients. Data that previous studies have credibly assumed that one-quarter (25%) of allows us to quantify the cost of this are drawn from unpublished bad debt is attributable to those with insurance in other studies of tabulations of the Community Tracking Study Physician Survey this type, we believe that the share of bad debt associated with fielded by the Center for Studying Health System Change, the Area table 1. estimate of costs of hospital care to the uninsured in maine, 2005 (in millions)

Charges for Charity Care $60.60

Charges for Emergency and Non-Emergency Bad Debt $123.00 Total Direct Expense $2,734.70 Total Gross Patient Service Revenue $4,914.80 Cost-to-Charge Ratio* 0.56 Charity Care Costs $34.00 Bad Debt Costs $68.80 Total—Charity Care and Bad Debt $102.80 Minus Bad Debt Attributable to Insured Patients (35%) <$24.10>

Estimated—Charity Care and Bad Debt (Uninsured) $78.70

*weighted average from GPSR n=36 hospitals

Resource File and Medical Economics. The primary source of infor- per week and weeks worked per year, and net income from medical mation is unpublished data from the Community Tracking Study practice in the prior year, practice expenses, the number of physi- Physician Survey fielded by the Center for Studying Health System cians in the state (by specialty) and out-of-pocket payments. Change. The survey was conducted by telephone in 2004/2005. We used information obtained from Medical Economics on practice The data used in this calculation are limited to 297 office-based expenses as a share of gross income by specialty to inflate physi- physicians in the New England Census Division. Net income data cians’ reported net income to an estimate of their gross billings.12 reported by survey respondents refers to 2003/2004 and was inflated Since the question about charity care includes care to reduced-fee to 2005 by multiplying by the average change in the Consumer Price patients, who pay some of the cost of the care, we also adjusted the Index from those years to 2005 (1.048). estimate to account for payments made by uninsured people. This The survey excluded physicians in specialties that typically do not estimate comes from Hadley and Holahan.13 have direct contact with patients (anesthesiology, radiology, and pathology), physicians who reported spending fewer than 20 hours per week in direct patient care, and physicians in residency train- Physicians provided ing. Responses from physicians who work in institutions (hospitals, medical schools, clinics) are excluded from the calculations because the charity care they provide will be captured in the estimates of $10.2 million in charity care provided by hospitals and clinics. We used information on the number of hours physicians report providing charity care in the previous month (defined as care provided to people for charity care in 2005 whom the physician received no payment or a reduced fee, excluding discounts from insurance plans)11, the number of hours worked

costs of caring for uninsured people in maine The details of the calculations are presented in Table 2. Summing across the totals for each specialty group, we estimate that physi- federally qualified cians provided $10.2 million in charity care in 2005. Medical and health centers surgical specialists account for over three-fourths of the charity The ambulatory care portion of the safety net in Maine also care provided by physicians. The estimate of physicians’ charity care includes Community Health Centers (CHCs) that provide health may double-count some charity care, since the charity care hours care for a reduced fee to medically underserved, uninsured, and reported do not indicate whether the physician provides the care in predominantly low-income populations. Complex and overlap- his/her own office or as a volunteer in a clinic or hospital. ping categories of health centers characterize the network of CHCs in Maine. Federally Qualified Health Centers FQHC( s) are health centers that receive some federal funding under section 330 of the Public Health Service Act. FQHCs must meet certain statutory requirements, such as location in a federally designated medically underserved area,

Table 2. Value of Charity Care Provided by Office-Based Physicians in Maine (by Specialty, 2005 $s)

GIM FP/GP PEDS MED SP SRG SP (No. of survey respondents) 63 46 44 80 64 1. Hours per month of charity care, from the CTS 3.95 5.82 3.46 8.57 7.79 Physician Survey (2004/2005) 2. Number of months worked per year, from the CTS 10.83 10.88 10.9 10.75 10.98 Physician Survey 3. Annual hours of charity care per year 42.8 63.3 37.7 92.1 85.5 (the product of row1 x row2) 4. Annual hours worked, from the CTS 2442 2400 2369 2294 2948 Physician Survey 5. Percent of annual effort devoted to charity care 1.75% 2.64% 1.59% 4.01% 2.90% (row3/row4) 6. Net income per physician in 2003/2004, from the CTS $125,026 $125,827 $141,265 $233,971 $250,464 Physician Survey 7. Adjustment factor for converting net income to gross 1.85 1.85 1.82 2.21 1.95 billings, from Medical Economics (Nov. 7, 2003) 8. Gross billings (product of row6 x row7) $231,298 $232,780 $257,102 $517,076 $488,405 9. Gross amount of charity care provided per physician $4,048 $6,145 $4,088 $20,735 $14,164 (product of row5 x row8) 10. Assume that uninsured pay 71% of charges out- $1,230 $1,867 $1,242 $6,302 $4,304 of-pocket or from other sources (calculated from re-weighted MEPS data for Maine) and inflate to 2005 $s by the change in the CPI (1.048) 11. Number of patient care, office-based 356 888 171 828 613 physicians in Maine (from the ARF) 12. Total uncompensated care provided by each specialty $0.44M $1.66M $0.21M $5.22M $2.64M

Source: Unpublished data from the 2004/2005 Community Tracking Study Physician Survey (rows 1, 2, 4, and 6).

costs of caring for uninsured people in maine provision of comprehensive primary health care services, and provi- costs for medical care and clinical services, including allocations for sion of services to patients regardless of ability to pay.14 facility and administration overhead costs. Costs for dental services and enabling services are not shown in Table 3 and are not included Statewide cost data on clinics in Maine are readily available for only in the estimates provided. those clinics that are FQHCs. We estimate the cost of care provided to uninsured patients in FQHCs, using the Uniform Data System The share of costs attributable to uninsured patients cannot be (UDS) maintained by the Bureau of Primary Health Care (BPHC). directly observed in the data available. However, information from The UDS contains medical cost and revenue data for each FQHC the UDS indicates that the share of total charges attributable to in Maine. In 2005, the FQHC network in Maine provided care to uninsured patients (uninsured patient’s charges divided by total approximately 125,000 patients through 16 federal grantees. In charges) is 20.4 percent, and we assume that this is equal to the combination with the state’s FQHC look-alike clinics, these grantees share of costs attributable to uninsured patients.17 Total costs for provided care in more than 60 distinct community based clinical medical care and clinical services for uninsured people, including sites in Maine.15 No data is available to estimate the cost of care allocated facility and administration costs, are then calculated by provided by non-FQHC clinics, although non-FQHC clinics may multiplying the share attributable to uninsured people by the total provide a significant volume of safety-net care in Maine. costs, resulting in $9.8 million in costs for uninsured Mainers. FQHC’s received $5.1 million in direct revenue from all patients. Table 3 shows cost figures derived from UDS data for Maine’s Assuming that uninsured patients accounted for 20.4 percent of FQHCs in 2005, the most recent data available. The calculation these payments, they would have paid FQHCs $1.0 million for their of cost estimates follows national estimates developed in a 2005 care, resulting in $8.7 million in total costs of uncompensated care study.16 In 2005, Maine’s FQHC’s reported a total of $48.0 million in to uninsured patients in 2005.

table 3. estimates of costs of uncompensated care to the uninsured at federally qualified health centers (fqhcs) in maine, 2005 ($millions)

Medical and Clinical Service Costs a. Medical Staff $32.2 67% b. Lab and X-ray $2.1 4% c. Medical and Other Direct Medical Care Services $7.7 16% d. Pharmacy and Pharmaceuticals $1.3 3% e. Mental Health, Substance Abuse and Other Professional $4.6 10% Total $48.0 100%

Share of Charges* (Uninsured) 20.4%

Medical and Clinical Service Costs (Uninsured) $9.8

Self-Pay Collections (Uninsured) $1.0

Estimated—Uncompensated Care Costs (Uninsured) $8.7

Notes: *Uninsured patients’ charges/all patients’ charges=$10.6M/$52.1M=20.4%. Source: Bureau of Primary Health Care, HRSA , Uniform Data System, Maine Rollup Report (2005).

costs of caring for uninsured people in maine Other Government Programs direct or acute medical care spending.19 The remaining 20 percent of expenditures funds long-term rehabilitative, psychiatric, and department of nursing home care and are excluded from the estimate of care to veterans affairs uninsured people. The Department of Veterans Affairs (VA) provides a range of Most veterans who use VA services have another source of health benefits to veterans, including health care, vocational training, care coverage, and most receive the bulk of their care from non-VA pensions, life insurance and indemnities, education, and disability sources.20 Because information on the services used specifically by compensation. Medical care accounts for over half of VA expendi- uninsured VA users is not available, we estimate their spending by tures in Maine each year.18 In FY 2005, VA expenditures for medical applying the share of uninsured VA users in Maine to the statewide services in Maine totaled $171.3 million (Table 4). We estimate that amount of direct medical care spending. In doing so, we assume $135.3 million, or approximately 80 percent of the total, represents that the VA-only, or otherwise uninsured, users use direct medical

table 4. department of veterans affairs (va) expenditures on care to the uninsured in maine, 2005 ($millions)

Total VA medical expenditures in ME, 2005a $171.3 Amount for direct medical care (79% of total)b $135.3 Percent of VA Users with Only VA Coveragec 24.4%

Estimated Direct Medical Care Expenditure on the Uninsured, 2005 $33.0

a US Dept. of Veterans Affairs expenditure data by locality. http://www1.va.gov/vetdata/page.cfm?pg=3 b 79% derived from the FY 2005 national VHA budget (in millions): acute hospital care services ($6,835) + outpatient care services ($14,589) + proportionate general operating expenses ($686)=total direct medical ($22,110)/total medical program budget ($28,093)=79%. See OMB, Annual Appropriations to the Department of Veterans Affairs, “Budget of the United States Government, Fiscal Year 2005—Appendix.” Data from Office of the Actuary, VA. c Two–year average of the 2004–2005 percentages in Maine using the March supplements of CPS.

10 costs of caring for uninsured people in maine services with the same frequency as the others. To the extent that they rely more or less heavily on VA care or consume disproportionately expensive or inexpensive services, our estimates may diverge from their actual medical spending.21 Table 4 shows that VA facilities 24.4 percent of those in Maine who reported being covered by VA had no other source of health coverage that year.22 Applying the CPS percentage of uninsured users of VA facilities to direct medical provided $33 million expenditures produces an estimate of $33 million in care to uninsured Mainers in 2005. of health care to indian health service The federal government attempts to meet its commitment to provide uninsured Mainers health care for American Indians and Alaska Natives through a system of hospitals and clinics on or near reservations, managed by the Indian Health Service (IHS) and, more recently, by Indian tribes. IHS facilities provide primary care services free of charge, and limited free specialty services are available through contracts with private summary of care to providers. However, services available through the IHS vary widely the uninsured (provider / across tribes, and IHS hospitals are not available in all service areas. Many communities have small clinics and they must contract out program data) for all specialty care, x-rays and other diagnostic tests and routine The largest provider of uncompensated care to uninsured people in preventive care such as mammograms. IHS has four operating units Maine is hospitals, accounting for $78.7 million in 2005. Office- in Maine, serving the Micmac, Maliseet, Passamoquoddy, and Pe- based physicians provided another $10.2 million in uncompensated nobscot tribes.23 Insurance and third party collections are estimated care. Among government programs and providers, the largest to pay for approximately 25 percent of total IHS clinic expenditures. amount of uncompensated care to uninsured people came through The remaining expenditures are funded by IHS appropriations. We Veterans’ Affairs hospitals and clinics ($33.0 million). Other exclude appropriations that cover non-patient care services such as government providers included community health centers ($8.7 health education and general public health services, and estimate that million) and the IHS ($7.5 million). Across all providers for which the remaining $7.5 million in federal appropriations as the costs of we had data, we estimate that uninsured people in Maine received caring for uninsured people at these facilities. about $138 million in uncompensated care in 2005.

costs of caring for uninsured people in maine 11 UNCOMPENSATED CARE COSTS: (Northeast, Midwest, South, and West) and a metropolitan statisti- Household Survey Data cal area indicator—spending in Maine cannot be determined using the MEPS alone. Therefore, we employ an approach that re-weights This section presents information on the cost of uninsured people MEPS observations for the Northeast region so that they correspond based on a methodology for constructing state-specific estimates of to the characteristics of observations from the CPS in Maine. the cost of medical care received by uninsured people, using data from We present details of this approach in Appendix B to this report. the Medical Expenditure Panel Survey (MEPS) in conjunction with the This Appendix also includes a discussion of other adjustments Current Population Survey (CPS). The MEPS, a nationally representa- that are required to produce estimates based on the MEPS data as tive survey of individuals and households conducted by the Agency for well as a detailed definition and description of how we estimated Health Care Policy and Research, is the most detailed source of health “donated” care, which is a component of the broader concept of care spending information available for this estimate. The CPS is the uncompensated care. most widely used data source for determining the size and characteristics of individual states’ uninsured populations. The upper portion of Table 5 reports the estimates of actual spending by uninsured people in Maine by source of payment. Uninsured Spending data for this analysis come from the MEPS Household adults received an average of $1,277 in care and uninsured children Component for the years 2002 through 2004. The MEPS collects in- received $1,382 in care. For all uninsured people, average spend- formation on health care use and expenditures, insurance coverage, ing per person is estimated to be $1,292. Applying these per capita health status, sources of payment, income, employment, and other spending figures to the estimated numbers of uninsured adults and sociodemographic characteristics for the U.S. civilian non-institu- children in Maine produces a total cost estimate of $161 million tionalized population. Respondents’ information is also adjusted and supplemented with data from medical providers, pharmacies, and insurance providers. Uninsured adults The analysis sample for this report is limited to persons who live in the Northeast census region. Newborns, people who die during received an average of the year, and those who are institutionalized for part of the year are included for the portion of the year that they satisfied the MEPS’ criteria for inclusion. The final MEPS sample includes 13,624 non- $1,277 in care and elderly people.24 CPS data are from the 2004 and 2005 March Supplement surveys uninsured children (years correspond to the year of insurance status and not year of interview). Conducted by the Bureau of the Census, the CPS uses information from over 50,000 households to provide estimates for the received $1,382 in care U.S. civilian non-institutionalized population. Although insurance coverage in the CPS is defined for the full calendar year prior to the for the medical care received by uninsured Mainers, with adults interview, benchmark analysis using other nationally representative accounting for about 86 percent of both the uninsured population surveys suggests that the CPS provides more of a point-in-time than and total costs. a full-year coverage estimate, and likely miscounts some part-year Using MEPS data on the sources of payment, we derive an esti- uninsured as full-year uninsured people.25 To produce comparable mate of the amount of uncompensated care that uninsured people definitions of coverage in the CPS and MEPS, we consider MEPS receive. Following previous research,27 we define uncompensated respondents to be “uninsured” if they have 7 or more months care as care received by uninsured patients that was not paid for of uninsurance.26 either out-of-pocket by the patient or by a traditional insurance Because MEPS is not designed to produce state-level spending esti- source (private, Medicaid, or Medicare). Clearly, donated care from mates the only geographic variables are the Census-defined regions private providers is one component of this estimate. Equally clearly,

12 costs of caring for uninsured people in maine table 5. medical care expenditures for the non-elderly uninsured in maine, by source estimated actual expenditures and predicted expenditures if fully insured, 2005 $s

non-elderly adults children all-non-elderly (n=104,671) (n=19,734) (n=124,405)

per capita total per capita total per capita total Percent (millions) (millions) (millions) Actual Spending, by Source Out-of-Pocket 510 53 328 6 484 60 37% Insurance Sourcesa Private 59 6 16 0 53 7 4% Medicaid 61 6 80 2 64 8 5% Worker’s Compensation 47 5 0 0 41 5 3% Uncompensated Careb Other Public 248 26 565 11 292 36 23% Other Sources 133 14 28 1 118 15 9% Donated Carec 219 23 365 7 239 30 19%

All Sources 1,277 134 1,382 27 1,292 161 100%

Predicted Spending if Fully Insured Out-of-Pocket 298 31 198 4 284 35 10% Total Expenditures 2,718 285 2,618 52 2,704 336 100% Notes:

a. Private and Medicaid are for people who are uninsured for less than a full year. b. Uncompensated care refers to payments made on behalf of the uninsured from sources other than health insurance, workers compensation, or out-of-pocket. c. Donated care is defined as the difference between actual payments from private sources (out-of-pocket and other) and the average amount of expected payment from a person with private insurance. Source: Urban Institute analysis of 2002–2004 MEPS Northeast sample reweighted to represent Maine’s uninsured population and the 2004–2005 CPS.

out-of-pocket spending, private insurance, Medicaid and workers’ accounts for 37 percent of their total care. Since some uninsured compensation do not represent uncompensated care. The two re- people have coverage for a portion of the year, private insurance and maining categories—“other public sources” and “other sources”— Medicaid pay for 9 percent of their care. Workers’ compensation, are not directly linked to any traditional type of health insurance. which is not insurance in the same sense as private health insur- These categories include, for example, philanthropic payments, ance but is nonetheless an insurance system for paying the medical payments from a wide range of publicly financed (non-insurance) care costs of injured workers, pays for about 3 percent of care for programs, and payments from other types of insurance, such as uninsured people. auto, homeowner, or personal liability. The remaining sources of payment fall into the category of uncom- Given these definitions, the largest source of payment for care re- pensated care. Other public sources account for 23 percent of care ceived by uninsured people is their own out-of-pocket spending, which delivered to uninsured patients and include care provided by the

costs of caring for uninsured people in maine 13 to an estimate of what the payment would have been if the person were covered by private insurance (See Appendix C), is the final category and represents 19 percent of care for uninsured patients.29 The three categories of spending that comprise uncompensated care totaled $81 million dollars in 2005, or 51 percent of the care received by Maine’s uninsured population.

We note that the estimate of per person spending for uninsured adults in Maine is somewhat lower than those made for three other Northeastern states using a similar methodology to re-weight regional MEPS data.30 Estimated spending for uninsured adults was $1,689 in New York in 2005, $1,754 in Connecticut in 2005 and $2,318 in Massachusetts in 2004, compared to $1,277 in Maine. Veterans’ Affairs, other federal programs, and other state and local However, these comparisons are not adjusted for some major dif- programs. Other sources, which include other private payments ferences in the characteristics of Maine’s uninsured population and payments from unknown sources, are responsible for 9 percent compared to those in other states. In particular, a much higher of care for uninsured patients.28 Donated care from private sources proportion of Maine’s uninsured adults live outside of metropoli- MSA of payment, which is derived from MEPS information about actual tan statistical areas ( s) and appear to be in better health than payments received from private, non-insurance sources compared the insured in Maine, unlike uninsured persons in many other comparisons. As shown in Appendix Table B4, about 60 percent of Maine residents, both insured and uninsured, live outside of MSAs, The largest payment compared to only 10 percent of Northeast residents generally. That same table also shows that Maine residents appear to be in poorer health than Northeast residents generally, with 10 percent report- source of care received ing fair or poor health compared to 8 percent for the Northeast as a whole. However, comparing the health status of the re-weighted MEPS data for insured and uninsured Maine residents indicates that by uninsured people uninsured adults in Maine are in significantly better health than insured adults (Appendix Table B5). In particular, only 1.3 percent of uninsured adults in Maine report that they are in poor health, is their own out-of- compared to 5.5 percent of insured adults. Similarly, much smaller proportions of uninsured Maine adults report any Activities of pocket spending, Daily Living (ADL) or Independent Activities of Daily Living (IADL) limitations, social or cognitive limitations, diabetes, hypertension, heart conditions, cancer, or musculoskeletal condi- 37 percent of their tions. As a result, a much higher proportion of uninsured adults in Maine report no medical spending at all, 36 percent compared to 11 percent of insured adults, and a much smaller proportion have very total care large annual medical spending exceeding $20,000, only 0.8 percent of uninsured adults compared to 6.8 percent of insured adults.

14 costs of caring for uninsured people in maine comparison of cost of on balance, we think the household survey estimates should be viewed as a lower bound on the costs of care for uninsured Mainers, uncompensated care while the provider/program estimates may potentially overstate the estimates between actual costs of care. provider/ program data potential effects of and household survey data providing insurance to The estimates of uncompensated care received by uninsured people derived from provider/program data and household survey data are people who are uninsured not identical. The provider/program data suggest that the unin- The lower portion of Table 5 reports predicted out-of-pocket and sured received about $138 million in care in 2005, while estimates total spending by uninsured people under the assumption of full- from household surveys suggest about $81 million. The MEPS data year insurance coverage. Our basic approach is to use multivariate used for the household survey estimates are known to understate models to estimate the relationship between health care spending spending relative to the National Health Expenditure Accounts and various personal characteristics including several measures (NHEA) and, although we made adjustments to align the MEPS and of health status as well as the share of the year the individual had NHEA aggregates, it is possible that the MEPS estimates could still insurance. To predict spending for the uninsured population if they understate costs.31 Moreover, the MEPS data are from all surveyed had insurance coverage, we use their characteristics and assume households in the Northeast region as opposed to only Maine. that they had health insurance for the entire year, i.e., we use the While differences in the characteristics of Maine’s uninsured popu- models to estimate the effect of having insurance on the result- lation undoubtedly explain some of the difference in per person ing increase in medical spending. The details of these methods are costs compared to other states, it is also possible that there may be contained in Appendix C. If fully insured, uninsured adults would unique features of Maine’s state policies and delivery system that receive just over twice as much medical care, $2,718, as they do cur- are not adequately captured in the re-weighted MEPS data for the rently, while spending by uninsured children would increase slight- Northeast census region. Since the re-weighting process is only able ly less than twofold, to $2,618 per child. Since adults make up 86 to adjust for differences in personal characteristics, these estimates percent of Maine’s uninsured population, total spending also more may not fully capture differences in costs due to specific policies than doubles, increasing to a new total of $336 million. In spite of and/or organizational arrangements that are distinctly different in the increase in total spending if insured, out-of-pocket spending by Maine compared to other Northeastern states.

The provider/program data are not perfect either. First, the estimate of the share of VA users who are uninsured is based on a relatively Hospitals or their small sample of CPS respondents and is quite imprecise. Similarly, we do not have hard data to show that all IHS revenues not offset by third-party or self-pay payments necessarily reflect costs of care affiliates employ received by uninsured patients. Third, to the extent that a greater share of hospital bad debt than we assumed is associated with insured patients, we would be over-counting the bad debt component more than one-third of the costs of caring for uninsured persons. Finally, not all providers who care for uninsured patients could be included in the cost estimates based on provider data (e.g., clinics other than FQHCs of primary care and clinics providing services not funded through the federal programs for which data were available), but interviews regarding the range of these other providers suggest that we have captured the physicians in Maine bulk of providers’ costs of caring for uninsured persons. However,

costs of caring for uninsured people in maine 15 uninsured people actually decreases, from $484 per person to $284 per person. This represents 10 percent of projected total spending. Medicaid is said to The incremental resource cost of extending insurance coverage to all Maine residents is equal to the difference between current spending pay 20 or 30 percent by the uninsured population and what projected spending would be if uninsured Mainers were insured. This amount is estimated to be roughly $175 million. The estimate is not based on any particular more for essentially benefit package, but rather assumes an “average” of the range of benefits held by low- and lower-middle-income people with insurance coverage. The $175 million estimate is simply the additional costs the same service if that would be spent by those who are currently uninsured if they had insurance. This estimate should not be treated as an estimate of the costs of providing universal coverage in Maine. A program with that provided by a objective would be considerably more expensive because it would likely displace some current private coverage and, as a result, absorb hospital-employed the costs of care currently paid for with private insurance. physician as opposed to an independent physician

16 costs of caring for uninsured people in maine sources of public revenues to support uncompensated care

In this section of the report, we review a range of funding that is available to providers to offset the costs of providing uncompensated care to uninsured people in Maine. As is the case in any state, care for uninsured Mainers is provided by a patchwork of organizations, such as hospitals, clinics, and direct public providers. Funds for this purpose come from Medicaid, Medicare, federal grants and other federal programs, the state and localities. Funding mechanisms include grants, public insurance program add-ons, and in- kind donations. We start with a brief overview of state polices and expansion of Medicaid coverage to previously ineligible childless describe health care delivery in Maine. We then address the larger, adults, up to incomes of 100 percent of the federal poverty level.36 and therefore more often easily quantifiable, sources of revenue re- This was accomplished through a Section 1115 HIFA waiver.37 DSH- lated to care for uninsured Maine people. We also include a section based funding for this “non-categorical” eligibility expansion is with descriptive information on the myriad important, yet often some $100 million a year (combined state and federal).38 Enrollment smaller or more fragmented pieces of the delivery system whose began in late 2002 and soon exceeded projections, as did the cost revenue streams were so varied they were difficult to quantify. We per enrollee, so that the program had to be scaled back.39 believe that most public revenues to care for uninsured residents In 2003, Maine launched an additional initiative, Dirigo Health in Maine are captured in the first section, but recognize that it is Reform, subsidizing a form of private coverage and slightly important for a reader to understand the importance of the actors expanding Medicaid. It also seeks to hold down health care described in the second section. costs and improve health care quality.40 The increases in state- supported coverage during the early 2000s have been accompanied state policy on uninsured by a decrease in Maine’s rate of uninsurance, even as the national people and health care rate has climbed.41 Policy makers believe that these expansions have also reduced the amount of bad debt and charity care incurred by delivery in maine hospitals.42 Indeed, the Dirigo expansion was to be partly funded In the five years prior to our study there was a major shift in Maine by a Savings Offset Payment assessed on health payers and meant health policy on uninsured persons. Through the late 1990s, Maine to recapture bad debt and charity care savings.43 This report does provided support for providers that serve uninsured patients largely not address the extent of change in bad debt and charity care, which through disproportionate share hospital (DSH) Medicaid funding. is disputed.44 Instead we focus is on what sources of public revenue DSH payments supplement basic Medicaid payments to hospitals in continue to be available to offset the costs of uninsured care as recognition of the extra burdens of serving Medicaid and uninsured of 2005. populations.32 As of 1998, Maine used hospital tax revenues to draw Several features of Maine’s health care delivery system deserve high- down more than twice as much federal DSH support as the national lighting here because of their relevance to how uninsured services average, whether measured per Medicaid enrollee or per uninsured are provided and financed. First, among the state’s 39 hospitals, person.33 Maine repealed its hospital tax and did not make DSH pay- there is no large public general hospital, although there are two ments for inpatient care from 1999 through 2002.34 small public district-related hospitals.45 There are large, state-owned There was lobbying to reinstate DSH support for uncompensated mental health facilities, and interviewees suggested that the state care, some two-thirds of which would have been federal.35 However, is heavily reliant on Medicaid to help fund mental health services. the state opted instead to dedicate its DSH allotment to fund an However, these facilities are not the focus of this report.

costs of caring for uninsured people in maine 17 Second, Maine is heavily rural, with very low population density46 Moreover, when the facility and professional payments are com- and widely dispersed, small hospitals.47 The state also does not bined, interviewees indicated that Medicaid pays 20 or 30 percent have the kind of large urban hospitals that in many states provide more for essentially the same physician service if provided by a hos- large shares of care to uninsured patients.48 Moreover, the Uni- pital-employed physician as opposed to an independent physician. versity of New England’s College of Osteopathic Medicine is the On the cost side, it appears that malpractice premiums decline, but state’s only medical school; there is no conventional medical school costs of regulatory compliance may rise. In general, hospitals are extremely important to health care delivery. In one rural county, sources explained that the local hospital not only employed physi- All of Maine’s health cians but also ran the nursing home and ambulance service. facilities are seen as safety nets—a role Public Revenue to Care for Uninsured Maine People enshrined in state law state funding for hospital care Medicaid provides by far the largest amount of state funding for as well as residents’ health care, matched by an even larger federal share. In 2005 the federal share of Medicaid spending was 64.89 percent.55 Medicaid in expectations 2005 provided a modest DSH allotment for care delivered to uninsured patients of about $11.6 million (Table 6).56 Maine’s 2005 payments went to two hospitals that qualified for assistance based with associated teaching facilities that often serve as safety nets.49 on earlier years’ rules on DSH allotments. As already noted, the Instead, all of Maine’s health facilities are seen as safety nets—a role state has reallocated the bulk of its inpatient DSH funding to the enshrined in state law as well as residents’ expectations.50 Unusu- already approved childless-adults expansion. Pending federal ally, Maine law requires that hospitals provide needed free care to approval, proposed state rules will reduce Maine’s DSH support for all low-income applicants; this obligation goes well beyond federal conventional inpatient hospital care to $200,000 a year.57 Emergency Medical Treatment and Active Labor Act (EMTALA) requirements that emergency rooms treat emergency patients. Dur- Beyond DSH, Maine evidently targets no other state program ing 2005, free care was required for those with incomes below 100 dollars directly to helping providers offset the costs of caring for percent of the federal poverty level; interviewees explained that the uninsured patients. We heard this description repeatedly from standard is rising to 150 percent. Many hospitals’ own guidelines go interviewees, both inside and outside of state government. This to 200 percent, often with a sliding scale of reduced fees.51 The same seems to be consistent with the state’s broader strategy of expanding free-care obligation applies to physician practices owned by hospi- access to coverage as opposed to subsidizing direct services. tals,52 and to surgicenters and some other institutions as well.53 The state does target disproportionate payment of another kind to Third, many physicians in Maine have become employees of hos- Critical Access Hospitals (CAHs), 13 facilities among the state’s 39 pitals or federally qualified health centers (FQHCs). It is estimated general hospitals that have qualified under Medicare standards as that hospitals or their affiliates employ more than one-third of crucial to their areas. Many interviewees stressed that Maine’s rural primary care physicians in the state.54 One motivation appears to be nature and low population density make CAHs especially important financial: FQHCs receive essentially full cost reimbursement under in the state, and the state has opted to pay its CAHs not merely the Medicare and Medicaid when they bill for employee physicians. normal cost-based rate for CAHs but 117 percent of costs. This

18 costs of caring for uninsured people in maine table 6: public funding available to support care for the uninsured in maine, by source, 2005 ($ millions)

federal state/local total Medicaid Inpatient dsh* $7.53 65% $4.07 35% $11.60 cah payments $0.65 65% $0.35 35% $1.00 Medicare dsh/ime $45.76 100% $0.00 0% $45.76 Veterans system $33.00 100% $0.00 0% $33.00 Indian Health Service $7.50 100% $0.00 0% $7.50 fqhcs $7.70 89% $1.00 11% $8.70 mch Title v Block Grant $0.22 17% $1.08 83% $1.30 Ryan White care Act $0.84 95% $0.04 5% $0.88 National Health Service Corps $0.00 0% $0.48 100% $0.48

Total $103.20 94% $7.02 6% $110.23

*Note: Medicaid inpatient dsh estimate excludes dsh payments for Institutions of Mental Disease, or imds. Additional sources not listed in Table 6 also played a role in assisting uninsured people in Maine. We have not quantified such support either because it is very small, reliable data are not available, or both. (See page 22) additional increment was recently implemented, and staff report it has an annual cost of about $7 million—only about 2 percent of ob- ligations to all hospitals at the time, but important to the small CAHs. The state opted

CAHs’ rural areas have higher rates of uninsurance than more urbanized locations, but CAHs have no special obligation to serve to pay Critical Access uninsured persons (beyond that of all Maine hospitals). The extra payment thus helps maintain access to care for all area residents, among whom uninsured residents constitute somewhat over 10 Hospitals not merely percent. By this reasoning, one might attribute slightly less than $1 million a year to care for uninsured patients.58 In Table 6, we allot the normal cost-based $1 million of this payment to maintaining uninsured access at CAHs. medicare payments to rate but 117% of costs hospitals Medicare provides hospitals with payments that can be used cy programs. Although most of IME payments reimburse hospitals to offset some of the costs of uncompensated care. Under the for actual costs they incur for intensity of care that is not fully re- Prospective Payment System (PPS), Medicare adjusts basic pay- flected in base PPS payments, a portion of the payment is intended ment rates for hospitals that serve a large share of low-income to compensate for the teaching hospitals’ social mission, including beneficiaries. These payments are also called DSH payments and, in the provision of uncompensated care to uninsured patients. In total, Maine hospitals received about $37 million in 2005.59 These 2004, six hospitals in Maine received IME funds, the majority of payments vary greatly by hospital, according to the most recent data which (nearly $13.9 million) went to the Maine Medical Center.61 for 2004. In all, 25 of Maine’s hospitals received these payments Research suggests that about one-third of IME payments should be in amounts ranging from under $100,000 to nearly $12 million.60 viewed as subsidies for teaching hospitals’ social mission.62 In 2005, Medicare also pays hospitals an allotment for the costs of Maine teaching hospitals received a total of $25 million in IME indirect medical education (IME) to hospitals that sponsor residen- payments, of which we estimate $8.3 million to have been available

costs of caring for uninsured people in maine 19 to subsidize uncompensated care. Therefore, through DSH and IME federal lawyers, and any payouts come from the federal government. payments, Medicare provided about $45.8 million in 2005 available In addition, FQHCs are also entitled to preferred access for recruit- to offset costs of care to uninsured patients (Table 6). ing National Health Service Corps practitioners, participation in the section 340b drug program that provides access to low, VA-level support for the va system, prices, and certain loan guarantees. The value of these cost-reduc- indian health service, ing benefits is difficult to quantify, but substantial. In addition, by federal rule, the state Medicaid program must assure and fqhcs that FQHCs are paid on a cost-realted basis for services provided Only federal revenues support the Veterans Affairs Health system to Medicaid enrollees. This provision applies even to centers only and the Indian Health Service, and mainly federal revenues support qualified for but not actually receiving section 330 grant support— the federally designated FQHCs. Spending on uninsured patients Maine has one such “look-alike” center. If services are provided for these types of providers is estimated to be exactly the same as under managed care contracts at a lower rate of payment, the state the provider cost estimates presented earlier in this report, which must make a supplemental “wraparound” payment to bring total netted out contributions from other sources. The VA constitutes the reimbursement to the full rate. FQHCs also benefit from special largest source of funding, of which an estimated $33.0 million (all Medicare payment rules. It is often said that giving FQHCs higher federal) is spent on uninsured patients. The IHS is likewise entirely reimbursements than are available to others is a kind of subsidy in federal, and we estimate $7.5 million is spent on uninsured patients recognition of their safety net role. We do not assign a dollar value through this program. For FQHCs we estimate the costs of serving to any of these types of additional help to FQHCs, as they do not di- uninsured patients to be $8.7 million, and as a whole FQHCs appear rectly provide resources designated to meet costs of uncompensated to have sufficient federal and state grant revenues to cover these care, but their value probably exceeds that of direct grants alone. costs. In Table 6, we apportion the share of federal/state revenues covering these costs based on the share of total grant revenues that stem from each of these sources. support for uninsured Several additional forms of federal support for FQHCs go beyond care at other providers the ryan white care act. The Ryan White Comprehensive As in all states, AIDS Resources Emergency (Ryan White CARE) Act is adminis- tered by the federal Health Resources and Services Administration (HRSA). It is the largest source of federal discretionary funding care for uninsured in for care to persons living with HIV/AIDS. States and cities use the grants to fund outpatient and inpatient services; medications; and support services to low-income, uninsured, and underinsured per- Maine is provided sons living with HIV/AIDS. Most direct medical care delivered via the Ryan White CARE Act originates from its Titles I and II.63 by a patchwork Title I provides emergency assistance to the metropolitan areas most affected by the HIV/AIDS epidemic. Maine does not receive these funds. Title II grants support ambulatory health care, insur- of organizations ance coverage continuation, home-based care, medications, and support services for people living with HIV/AIDS across the state. section 330 grants; one is free FQHC malpractice coverage for all The majority of Title II funds are earmarked for AIDS Drug patients, regardless of health insurance status. Claims are handled Assistance Program (ADAP), which provides medication to under the Federal Tort Claims Act in federal court, defended by individuals without insurance coverage or who cannot get all of

20 costs of caring for uninsured people in maine their medication needs met through their insurance payer. In FY Title V patients were enrolled in Medicaid, but the remaining 62 2005, Maine’s total ADAP and Title II Ryan White CARE budget was percent had unknown insurance coverage. We therefore applied approximately $1.6 million for the estimated 477 persons living with the national shares of uninsured patients, that is, 8 percent among AIDS in Maine in 2005.64 Of this total, we estimate that $900,000 children with special health needs and 9.5 percent among all others is attributable to care for uninsured persons, with $44,000 coming served. After subtracting infrastructure expenditures, the resulting from state revenues. (Because of its complexity, the derivation of estimate is $1.3 million in support to provide care for uninsured this estimate is presented in Appendix A.) patients. MCH grant programs other than the MCH Block Grant, e.g., the Abstinence Education Program and Healthy Start, are not

included here. maternal and child health block grants. The Title V

table 7. maternal and child health (mch) block grant spending on care for uninsured in maine, 2005 ($millions)

children with othersa all users special health needs Total mch Block Grant expendituresb $3.3 $14.7 $18.0 Total, subtracting infrastructure expenditures $2.5 $11.2 $13.7 Percent of users uninsuredc 8.0% 9.5%

Estimated mch Block Grant spending on uninsured $0.2 $1.1 $1.3

Notes:

a. Includes pregnant women, infants less than 1 year, children 1 to 22 years, and all others. b. Includes Federal allocation, Maine match and overmatch, and program income. Expenditures on administration are redistributed proportionately. c. Because Maine reports a very high rate of unknown insurance status (62%), these percentages are based on national figures. 9.5% for the “others” category is the weighted average of 6.2% (pregnant women). 8.9% (infants < 1 year old). Source: Maternal and Child Health Bureau, hrsa, Title V Information System (tvis), FY 2005, https://performance.hrsa.gov/mchb/mchreports.

Maternal and Child Health (MCH) Block Grant is a federal program clinics staffed with national health service corps of assistance to states that targets health improvement for all moth- professionals. The National Health Service Corps (NHSC) is ers and children, but with particular emphasis on low-income, a US Department of Health and Human Services program that uninsured, and underinsured persons. States allocate much or most seeks to improve the health of medically underserved persons by MCH funds to local service providers. Title V programs in Maine recruiting health professionals to serve in communities with the serve nearly 120,000 pregnant women and children each year.65 greatest need. In return for commitments to serve in primary States must spend $3 for every $4 in federal funding received. Most health care shortage areas, NHSC assists physicians with their states, including Maine, contribute more than the required match. education loans and training. NHSC professionals staffed 24 sites In 2005 Maine contributed 82.9 percent of total MCH funds.66 in Maine in 2005.67

Total MCH block grant funding in Maine (state plus federal) was The NHSC Uniform Data System (UDS) reports detailed cost and $18 million in 2005 (Table 7), but estimating the share related to charge information for the non-FQHC clinics in Maine with a uninsurance was complex because data on patients’ insurance NHSC assignee.68 In calendar year 2005, $11.4 million in care was coverage is incomplete for Maine. Approximately 38 percent of delivered at these sites. Uninsured patients’ unpaid charges represented

costs of caring for uninsured people in maine 21 9 percent of charges for all patients. After applying this percentage free clinics. A second type of non-FQHC clinic in Maine is the to total expenses at NHSC sites, we estimate that approximately $1 free clinic. Free clinics are private, non-profit organizations that million was for uninsured patients. About half of this amount was provide free care to uninsured people, largely through a network of collected from self-pay patients. NHSC sites provided the remainder, volunteer providers.73 One source identified nine free clinics state- an estimated $500,000 in uncompensated care, to uninsured Maine wide.74 These clinics served an estimated 3,000 uninsured patients residents in 2005. State, local and other non-federal sources paid for in Maine in 2004.75 There is little government funding for these or- this care. Federal grants do not finance direct care in these clinics.69 ganizations, and they function primarily through volunteer service provision, donated goods, and support from affiliated hospitals. Other Sources Providing Care school-based health clinics. Twenty-nine of Maine’s schools were served by 27 school-based health centers (SBHC) in To Uninsured Mainers the 2004–2005 school year. Uninsured or self-pay students com- prised 11 percent of SBHC users and 9 percent of visits.76 SBHCs free-standing clinics provide basic primary care services regardless of ability to pay, of- other than fqhcs ten through a staff of nurse practitioners, physician assistants, and registered nurses with backup from physicians.77 SBHCs in Maine Like other states, Maine has a number of non-FQHC clinics of vari- receive funds from numerous sources including federal, state, and ous types. These include rural, school-based and free clinics. Some local sources, and the funding mix at each site is diverse. Most sites are hospital-affiliated and others are free-standing. According to SBHCs in Maine receive significant financial and operational sup- interviewees, non-FQHCs generally are smaller than FQHCs, offer port from the Maine Center for Disease Control and Prevention’s less comprehensive services, and lack the same obligation to serve Teen and Young Adult Health (TYAH) Program. In most cases, all presenting patients (unless they are hospital-based). the SBHC sponsor is a local health system, although local health rural health clinics (rhcs). In 2005, there were 39 rural departments and school districts also run SBHCs. Others SBHCs are 78 health clinics in Maine.70 By federal law, RHCs must be located in FQHCs or sponsored by physician group practices. rural, underserved areas, although some qualify for “automatic” designation by meeting different criteria. They benefit from special local health departments Medicare and Medicaid payment rules that require cost-based reim- Local health departments (LHDs) play only a small role in provid- bursement, much like FQHCs, but this support shores up a different ing medical services to uninsured persons. Interviewees suggested kind of safety net. The federal goal in assisting RHCs is to keep basic that only two cities, Portland and Bangor, had significant efforts in health care accessible in rural areas to all patients, including Medi- this area.79 care and Medicaid enrollees among other insured clients. Nonethe- less, RHCs care for many uninsured, self-pay patients who receive free portland public health department. The Portland care or pay reduced fees. They do not, however, receive reimburse- Public Health Division, a division of the city’s Health and Human ment targeted toward care for this population.71 Dollar figures for Services Department, is organized into five service areas: epidemiol- RHCs’ public revenues and uncompensated care costs are not readily ogy, family health, health promotion, indigent care and infectious available, and we are unable to assign a dollar value to them. disease. Many of the Division’s activities, such as maternal and child health home visits, immunization clinics, and disease surveillance It is notable that in 2005, the HRSA Office of Rural Health Policy are best classified as population-based public health; however, it also Grants provided grant funding of $1.3 million to various grantees plays a role in the provision of clinical services to uninsured persons. in Maine.72 However, these grants do not directly support clinical services; they include an obesity intervention program, support for A venue in which uninsured people might receive care is at one of the purchasing automated external defibrillators, and assistance for gen- City’s six school-based health centers. At these centers, which are a eral rural healthcare network development. Accordingly, we do not partnership between the Portland Public Health Division, Portland consider them to be funds related to caring for uninsured Mainers. Public Schools and Maine Medical Center, patients receive clinical

22 costs of caring for uninsured people in maine services regardless of their ability to pay. The City of Portland also type of funding include loan guarantees, access to favorable bond has a Positive Health Care team funded by the Ryan White CARE Act. rating categories, and grants to upgrade information technology. This team provides comprehensive primary care services to individuals living with HIV/AIDS regardless of their ability to pay.80 private support for Two of the City’s programs are most directly related to indigent uninsured people health care, including care for uninsured patients. First is the Maine has developed cooperative local mechanisms for coping with Health Care for the Homeless Program, which runs an FQHC that the burdens of uninsurance. Two particularly organized approaches provides primary care check-ups, dental services, immunization, operate in the Portland area (CarePartners) and in Franklin County and testing to people who are homeless. The dental clinic also in western Maine (Franklin Health Access). Community Health serves uninsured people who are not homeless. Second is the Port- Connections in the Greater York Area and the Maine Health Alli- land Community Free Clinic, which is a public-private partnership ance in Northern Maine are other examples of health care access between the City, Mercy Hospital, and private philanthropy/volun- programs in the state.83 teers. The clinic provides primary and specialty health care to low- income, uninsured adults.81 Although the Free Clinic is part of the carepartners. Closely affiliated with MaineHealth system, Portland Public Health Division, it does not regulary receive public CarePartners serves residents of Kennebec, Lincoln and Cumberland funds. It is funded largely through its relationship with Mercy Counties. CarePartners is a “managed uncompensated care program” Hospital (which provides both direct funding to the clinic and free that helps low-income individuals sign up for public insurance. If services to its patients), private donations, and the donated labor of individuals do not qualify for public programs (or if private premi- its volunteers. ums would exceed 5 percent of their gross income), they are enrolled in CarePartners own program so they can obtain free care donated bangor public health department. The Bangor Public by participating physicians/service providers. The program not only Health Department emphasizes health education, monitoring, and facilitates donated care (with a small co-pay), but also provides en- other public health services, in which a key effort is to improve rollees with care management services, a match with a primary care coordination. The city conducts public health home nursing and provider, and prescription drug services (which it offers largely by runs several clinics including for STDs, vaccinations, pediatric dental aggressively accessing free medication programs of pharmaceutical care, and HIV/AIDS screening, counseling, and care, but not a gener- companies). Approximately 1,000 Mainers are enrolled in CarePart- al free clinic or even well child services. In all, the city supports only ners each year. In 2004, the program estimated that annual donated a few full time equivalent staff; most funding comes from federal provider and hospital services totaled $4 million.84 monies passed through the state. Referrals for clinical services are made to the Penobscot Community Health Center. That center is the franklin health access. This “managed uncompensated principal source of care for uninsured patients, along with a medical care program” accepts applications from needy people, seeks to residency clinic run out of the Eastern Maine Medical Center. place those eligible into public coverage, and refers others for free or reduced-fee care within a network of collaborating service provid- state programs indirectly ers.85 Franklin Community Health Network was begun by the local hospital in 1991. Participating providers accept the Health Access’s supporting care for assessment of eligibility for free care or reduced fees. About 800 pa- uninsured people tients a year receive help, some paying their share through “barter” of in-kind services valued at an hourly rate favorable to the patient We also do not include dollar values for a number of other state or family. The program began with grant funding from the Robert programs that help medical providers. Such programs are often Wood Johnson Foundation and others. A key donor of ongoing small, not tied to direct patient care, and sometimes only margin- revenue is the area’s hospital, which is central to provision of care in ally related to uninsured populations. For example, Maine offers the area. educational assistance for some physicians, intended to improve access for all, especially in underserved areas.82 Other examples of this

costs of caring for uninsured people in maine 23 of uninsured patients.

The two estimates of costs of care to uninsured patients are far- ther apart than has been observed in prior studies of the nation or of individual states86—$81 million based on the MEPS household surveys to $138 million based on provider and program data. As we discussed, there are reasons to believe that the household survey estimate may be understating the costs of care to uninsured Main- ers, while the provider/program estimates may overstate these costs. First, the MEPS data were collected from 2002–2004, while the provider data are almost all from 2005. Discrepancies could arise if the inflation adjustment applied to the MEPS is not consistent with the cost experience in Maine, or if the actual experience in 2005 differed from the average between 2002 and 2004.

Second, another possible reason is that the household survey cost estimates are based on a sample of people in the Northeast census The estimates of the region that has been adjusted to have similar characteristics to residents of Maine. Since the population in the Northeast is substantially more likely to reside in a metropolitan area than the popula- cost of uncompensated tion in Maine, the extent of the adjustment required with respect to this attribute was large. Although the adjusted sample providing data care for the 124,000 on health care costs reflects the characteristics of Maine residents, the size of the adjustment could raise a potential question about the uninsured Mainers applicability of this estimate to Maine. Third, none of the adjustments to the MEPS that account for demographic differences between Maine and the Northeast can cap- range from $81 to ture institutional differences in how care is provided, especially in non-metropolitan areas, or policy changes such as the expansion of Medicaid and the introduction of the Dirigo Health Reform. $138 million For example, it may be that the Medicaid expansions resulted in disproportionate enrollment of uninsured adults with significant health care needs, so that the remaining uninsured people have CONCLUSION well below average costs. This paper presents estimates of the costs of uncompensated care for In contrast, almost all of the provider and program data used to the about 124,000 uninsured Mainers from two perspectives: estimate the costs of care provided to uninsured patients are derived (1) the providers and programs designed to support this care; from providers located in Maine. The only exception is the physician and (2) the households with uninsured individuals who are us- component and that represents less than 10 percent of the aggre- ing health care services. These two sets of estimates indicate that gate estimate. However, estimates based on provider/program data uninsured residents receive large amounts of health care in Maine, required assumptions about the share of hospital bad debt associ- although the amount of care is much lower than it would be if they ated with the insured population and the share of VA and IHS users had insurance. Because there is no one agreed-upon method for who are uninsured. Alternative assumptions that were still plausible estimating the costs of care provided to uninsured Mainers, we draw could have produced lower estimates. Nevertheless, we feel the esti- on two substantially different sources of data to provide a range of mate based on provider and program data is probably closer to the likely costs. We then compare these two estimates to the public sec- truth, although it still may be on the high side. tor revenues that are available to offset the uncompensated care costs

24 costs of caring for uninsured people in maine Our estimate of public sector revenue available to support uncom- Beyond the magnitude of the estimates, perhaps the major point pensated care to uninsured patients is approximately $110 million. to take away from this paper is that the presence of large numbers This estimate may be high to the extent that available revenues are uninsured persons and their inevitable need to receive health care used for other purposes than to help the uninsured, including to have resulted in a complex mosaic of government programs and offset payment shortfalls on patients with coverage. Moreover, private initiatives to defray the costs of that care. The Veterans some $11 million in 2005 consisted of Medicaid disproportionate- Affairs system, for example, plays a large role in caring for share-hospital payments that, according to current plans, are likely uninsured residents. In general, better organization and less to be cut to $200 thousand a year. The estimate may also be low to fragmentation could lead to more efficient care delivery and the extent that it excludes local sources of funding for uninsured more effective treatment of uninsured Mainer’s health needs. people that we identified but were not able to quantify. Moreover, As it became evident that there was a potential gap between our the uninsured could be disproportionately more or less expensive estimates of the costs of caring for uninsured Mainers and the than average—and that difference would make our per-head public subsidies available to offset those costs, we asked several estimates of revenues inaccurate. About 94 percent of the interviewees to consider how Maine communities may cope with quantifiable public support comes from the federal government. this situation. One theme that emerged was that “Maine is small, so we all work together.” This collaborative effort includes National estimates suggest that public sources of revenue avail- providers supporting each other, lobbying hard for additional able to offset the cost of uncompensated care, in the aggregate, are Medicaid spending and resisting cuts in private or public payments 80 to 85 percent as large as costs.87 Our estimate for Maine lies in whenever possible. However, in the absence of large public hospitals this range as well, based upon costs of $138 million. The balance of or subsidies to offset the costs of care for uninsured people (such revenue versus cost varies substantially by sector and by provider. as those often provided as Medicaid DSH payments), understand- Revenues flow much more heavily to federal providers and hospitals ing how providers in Maine serve uninsured patients will require than to private physicians or non-FQHC clinics that do not qualify further study. for federal-state assistance. Of the public share of revenue, federal sources contribute about two-thirds nationally, while state and local revenues constitute about one third.88 In Maine, however, the picture is considerably different. Quantifiable direct state and local support for uncompensated care in Maine is low, with this category Large numbers of of funds combining for only about 6 percent of the public total. The reason for this appears to be, in part, that the state has used Medicaid DSH to fund a coverage expansion rather than for uninsured persons provider subsidies that can offset some of the costs of uncompensated care. and their need In aggregate, these estimates suggest that the 2005 public revenues in the system to support care for uninsured patients are somewhat below our provider/program estimates of costs for the care they re- for care has resulted ceive, but above our household survey estimates. However, there are several cautionary notes that warrant consideration. First, for some providers, it is likely that revenues in excess of costs for privately in a complex mosaic of insured patients (or, even some publicly insured ones) may be used to cross-subsidize care for uninsured patients, but the extent of this government programs cross-subsidy is hard to measure. Second, not every provider who cares for an uninsured patient gets paid or is made whole. For example, there are generally no programs that compensate and private initiatives private physicians for the cost of treating uninsured patients in their offices. costs of caring for uninsured people in maine 25 APPENDIces

26 costs of caring for uninsured people in maine APPENDIX A: COSTS OF UNCOMPENSATED CARE PROVIDED THROUGH THE RYAN WHITE COMPREHENSIVE AIDS RESOURCES EMERGENCY (CARE) ACT

Most direct medical care delivered via the Ryan White CARE Act patients who are uninsured to get a total ADAP spending on the originates from its Titles I and II.89 Title I provides emergency as- uninsured of approximately $775 thousand. The state contributed sistance to the metropolitan areas most affected by the HIV/AIDS $60,000 to ADAP, and the rest of funding for Title II (ADAP and epidemic. Maine does not receive these funds. Title II grants non-ADAP) was federal. support ambulatory health care, insurance coverage continuation, About half a million dollars of Title II Ryan White funding was for home-based care, medications, and support services for people liv- non-ADAP programs. We estimate that about two-thirds of this ing with HIV/AIDS across the state. The majority of Title II funds funding was for direct medical care. The share of patients who are are earmarked for AIDS Drug Assistance Program (ADAP), which uninsured and who receive direct medical services from Title II provides medication to individuals without insurance coverage grantees is lower than for patients receiving medications through or who cannot get all of their medication needs met through their ADAP. In 2004, providers reported that the share of clients nation- insurance payer. Maine received $1.6 million in Title II Ryan White wide served by Title II who were uninsured was 31 percent.91 Apply- funds in 2005 (Table A1). Just over $1 million of this was used for ing this percentage to the cost of direct medical care results in total ADAP, and the rest was for other Ryan White programs. In general, Ryan White spending in Maine of $106 thousand through Title II. the Ryan White program is intended to be the payer of last resort and to only fill in current gaps in the funding of HIV/AIDS care. We estimate that the sum of spending on the uninsured from ADAP ($775 thousand) and non-ADAP Title II ($106 thousand) was $880 To estimate the cost of medical care delivered to the uninsured thousand in 2005. Approximately $40,000 of this funding was pro- through the Ryan White program, spending is divided into two cat- vided by the state, and the rest was from the federal government. egories: ADAP, which represents the largest piece of the budget, and Title II, non-ADAP direct care spending. “Direct Care spending” excludes the cost of home health services, enabling and support services, and program evaluation. Unlike some other states, Maine does not report using ADAP funds to purchase/maintain health insurance coverage.

Data on the insurance status of individuals served by the broader Ryan White CARE act are not systematically collected. However, the National ADAP Monitoring Project consistently reports the percentage of ADAP patients who are covered by Medicaid, Medicare, or private insurance. The remaining share of patients, 73 percent, is counted as the share of uninsured for ADAP (Table A1).90 We multiply the $1 million in total ADAP budget by the share of ADAP

costs of caring for uninsured people in maine 27 APPENDIX b: methodology for estimating medical care costs of the uninsured

Using data from the Medical Expenditure Panel Survey (MEPS) and and systematically lower in the MEPS–HC than those reported in the the Current Population Survey (CPS) we estimate the cost of the NHEA. For example, one of the largest sources of difference is the fact medical care for the uninsured in Maine in three stages. that the MEPS–HC universe is limited to the civilian, non-institutionalized population. Consequently, the MEPS–HC excludes all spending • First, we adjust the MEPS data for inflation and to align them with (both acute and long-term) for nursing home residents and people in the National Health Expenditure Accounts (NHEA), which acute-care or specialty hospitals (e.g., psychiatric or rehabilitation) incorporates several fundamental differences in definitions and for more than 30 days.93 Since the NHEA is the standard benchmark methods relative to the MEPS. for measuring health spending, we apply a set of adjustment factors • Second, the Northeast census region population in the MEPS by major source of payment in the MEPS–HC to align the MEPS–HC is “re-weighted” to mirror the characteristics of Maine residents total spending estimates with those from the NHEA.94 represented by the CPS. We use CPS data from the 2004 and 2005 March Supplement sur- • Third, using the Maine-reweighted MEPS sample, we calculate veys (years correspond to the year of insurance status and not year per capita spending on medical care by age for the uninsured, of interview) to adjust the MEPS–HC person weights so that the including a separate estimate of donated care based on the differ- MEPS–HC Northeast sample reflects the size and characteristics of ence between amounts charged to the uninsured and the amounts Maine’s population. Conducted by the Bureau of the Census, the paid from private, non-insurance sources of payment. CPS uses information from over 50,000 households to provide estimates for the U.S. civilian noninstitutionalized population. aligning the meps and Although insurance coverage in the CPS is defined for the full cal- the nhea endar year prior to interview, benchmark analysis using other nationally representative surveys suggests that the CPS provides more Before constructing the estimates of medical spending in Maine, of a point-in-time than a full-year coverage estimate, or at least we adjust the MEPS–HC expenditure data to account for medical miscounts some part-year (un)insured as full-year (un)insured.95 care price inflation over time and fundamental differences between To match coverage definitions in the CPS and MEPS–HC, we define the MEPS–HC and the National Health Expenditure Accounts92 “full-year” Medicare and Medicaid coverage in MEPS–HC to be 12 (NHEA) produced by the Centers for Medicare and Medicaid months of Medicare, and 12 months of Medicaid, SCHIP, or other Services (CMS). First, medical care payments and charges from comprehensive state coverage. For other types of coverage, we take 2002–2004 are inflated to 2005 dollar values using the medical care advantage of the fact that the MEPS provides more detailed insur- component of the Consumer Price Index. Second, we add prescrip- ance coverage information that allows us to assign public or private tion drug expenditures to total charges, because total charges in the coverage based on which type of coverage was more prevalent in the MEPS–HC do not include charges for prescription drugs. Third, we year for those who were insured. We selected prevalence definitions impute an estimate of donated care from private sources of payment (number of months with a particular type of coverage) to yield a that the MEPS–HC is likely to underestimate. coverage distribution in the MEPS–HC Northeast sample that is very Lastly, fundamental differences in definitions and methods of similar to the distribution in the CPS Northeast sample, using the 96, 97 measuring expenditures in the MEPS–HC and the NHEA produce standard full-year definitions in CPS. People with multiple types estimates of national health expenditures that are significantly of coverage are assigned to a single category using the following

28 costs of caring for uninsured people in maine hierarchy that gives priority to the type of coverage highest on We use the predicted probabilities from the probit models to define the list: (1) Medicare, (2) ESI (own or dependent coverage), (3) the initial Maine-adjusted person weight in the MEPS–HC as: Medicaid/SCHIP/other state coverage, (4) non-group, and (5) other ME = MEPS * [ (1–P ) / P ] * [ P / (1–P ) ], where insurance. A person is classified as uninsured if they do not report perwt perwt cps cps meps meps any type of insurance coverage. MEPSperwt = the unadjusted person weight in MEPS-HC Northeast re-weighting the meps sample, Pcps = the actual probability of living in Maine among persons in the northeast sample to CPS Northeast sample, resemble maine P = the predicted probability of living in Maine among persons meps statistical re-weighting. Because the MEPS-HC is not de- in the MEPS–HC Northeast sample. signed for state-level spending estimates—the only geographic variables are the Census-defined regions (Northeast, Midwest, South, and West) and a metropolitan statistical area indicator—spending cell-based re-weighting. After the initial statistical-based in Maine cannot be determined using the MEPS–HC alone. Conse- adjustment, the MEPS–HC person weights are adjusted a second quently, we estimate a probit model of the probability of residence time in order to match the size and distribution of the re-weighted in Maine among CPS respondents in the Northeast, and apply the MEPS–HC Northeast sample to the size and distribution by key parameters from this model to identically defined variables for characteristics of the Maine CPS sample. In this step, we parti- MEPS–HC respondents in the Northeast to predict the probability tion the MEPS–HC and CPS samples into cells defined by age, race, of living in Maine for each MEPS–HC observation.98 (The dependent insurance status, and adults’ labor force status. We then construct variable in the probit model is a dichotomous indicator that equals the ratio of the sum of CPS weights to the sum of MEPS–HC weights 1 if the person lives in Maine.) and apply it to each individual’s MEPS–HC weight in the cell. (Since we are pooling data from multiple years, we first adjust the sums of 0 18 We estimate separate probit models for children (ages – ) and the weights so that they represent a single year’s population.99) The non-elderly adults (ages 19–64). The models control for age, gender, final adjusted MEPS–HC person-weights reflect the CPS estimates of race and ethnicity, marital status, household size, education, urban the number of uninsured, privately insured, and publicly insured residence, employment status, industry, and family income relative people in Maine by age, race/ethnicity, and non-elderly adults’ to poverty. (Appendix Tables B1 and B2 report the parameters of the labor force status. (Appendix Table B3 shows the distributions of probit models for Maine residence.) MEPS–HC and CPS samples across the age, insurance status, and labor force status cells, and the adjustment factors calculated from the ratios of the weighted populations in each cell.)

Appendix Table B4 illustrates the result of the re-weighting process by comparing the average sociodemographic characteristics of the MEPS–HC Northeast sample, before and after the re-weighting process, to the CPS Maine population. Before re-weighting, the MEPS–HC

costs of caring for uninsured people in maine 29 400% of the federal poverty level; a larger percentage unemployed computing donated care. Following prior methods used to and a smaller proportion not in the labor force; smaller propor- estimate the cost of the uninsured nationally and in several other tions in fair or poor health; and different distributions of workers states, we adjust the uninsured’s total expenditures in MEPS–HC by industry and occupation. The overall effect of the re-weighting to include the cost of care donated from private sources of pay- is summarized by the sum of the absolute differences between the ment.100 (Donated care from public hospitals and clinics is already MEPS–HC and the CPS values (shown in the bottom row of the imputed by the MEPS–HC and included in total expenditures.) We table). Before the re-weighting process, this measure has a value of define “donated” care as the difference between payments actu- 196.7 percentage points; after the re-weighting the sum of absolute ally received from uninsured patients and payments providers values shrinks by almost three-fourths to 54.2 percentage points. would expect to receive for the same services from privately insured people. The expected payment is the ratio of payments to charges Adjusting the weights in this way enables us to use the MEPS–HC for the full-year privately insured (the discount rate, represented data for individual respondents from the Northeast to estimate by prv_discnt in the formula below) applied to total charges for medical spending and sources of payment that reflect both the size care received by the uninsured, excluding care paid for by private and the characteristics of Maine’s population as represented in the insurance, public insurance, or other public sources.101 The private CPS. The revised weights indicate the number of number of Maine discount rate used is 71.0% for adults and 76.3% for children.102 residents with similar characteristics represented by each sample individual in the MEPS–HC. Table B5 illustrates the use of the re- We use the following formula to calculate donated care for each weighted data by comparing selected characteristics of non-elderly uninsured person.103 Maine residents, as represented in the re-weighted MEPS data, by age and insurance status. doncare = payments expected if privately insured—actual payments received;

= [prv_discnt * totchgs * {(slf+opr+osr)/totexp}]–(slf+opr+osr),

where

prv_discnt = total payments / total charges, for the full-year privately insured,104 slf = total out-of-pocket payments, opr = total payments from other private sources, osr = total payments from other unclassified sources.

30 costs of caring for uninsured people in maine APPENDIX c: Computing predicted expenditures, if fully insured We estimate a two-part statistical model of medical care spending Separate models are fit for total expenditures and out-of-pocket to calculate how much more medical care the uninsured would use expenditures. Total expenditures include donated care if the person if they had full-year insurance coverage. The two-part model con- is uninsured. The models are estimated separately for non-elderly sists of a logistic regression model of the probability of having any adults and children. Appendix Tables C1 and C2 report the model medical spending, and a model of the amount of medical spending coefficients for adults and children. for people with positive expenditures.105 This approach is often used Insurance status is measured using a continuous variable for the when estimating a statistical model on spending data that are not percentage of the year the person is insured (with either private or normally distributed because a large share of people have no spend- public insurance). To simulate the impact of full-year insurance ing and a small share have very high spending. coverage on the uninsured’s spending, the insurance measure is set The sample is restricted to the non-elderly (ages 0 to 64), and ex- to 1 and the models’ coefficients are used to predict the probability cludes persons with any Medicare coverage and persons with private of having any expenditures and the amount of expenditures for coverage who have family incomes greater than 400 percent of the people with positive spending, assuming full-year coverage. The federal poverty level. Non-elderly Medicare beneficiaries are either per capita predictions are multiplied by the number of non-elderly disabled or have end-stage renal disease, and therefore their medical uninsured adults and children in Maine to determine total state- care is not likely to be typical of either the uninsured or those with wide expenditures for the uninsured, if they were fully insured. The private coverage. The high-income privately insured are excluded amount of this spending over and above total current spending for on the assumption that their spending behavior is not as comparable the uninsured in Maine represents the estimated incremental medi- to the uninsured population’s as the privately insured with low and cal care cost of complete insurance cove. lower-middle incomes. The models control for age, gender, urban/ rural residence, race and ethnicity, marital status, education, family income relative to poverty, self-reported health and mental health status, ADL/IADL help received, a range of functional and activity limitations, and the presence of many chronic health conditions.106

costs of caring for uninsured people in maine 31 APPENDIX table a1

ryan white care act spending on medical care to the uninsured in maine, 2005

aids Drug Assistance Program (adap) Total adap Budget, Federal and State Sourcesa $ 1,062,831.0 Percent of adap Patients Uninsured 73% ADAP Spending on Uninsured, 2005 $ 775,866.6

Title II Federal Grants to Maine (excluding adap) $ 501,672.2 Estimated Share for Direct Medical Careb 68% Percent of CARE Act Patients Uninsuredc 31% Title II Spending on Uninsured, 2005 $ 105,752.5

Total, Ryan White Care to Uninsured, 2005d $ 881,619.1

Sources: National adap Monitoring Project 2006 Annual Report, National Alliance of State and Territorial aids Directors, Kaiser Family Foundation, March 2006; Kaiser State Health Facts Online, Ryan White care Act Budget, www.statehealthfacts.org; 2004 care Act Data Report, http://hab.hrsa.gov/reports/2004_Data_Summary/page1.htm.

Notes:

a The ADAP budget is spent almost entirely on medications. Unlike some other states, Maine does not report using ADAP funds were used to purchase/maintain health insurance coverage. b Based on share of these funds used for direct medical care in Missouri. c Providers reported that 31% of care Act clients in 2004 were uninsured (national estimate). (2004 care Act Data Report, Section 2, Items 32). d Maine does not receive Title I funds because these funds provide emergency assistance only to eligible metropolitan areas (emas) most severely affected by the hiv/aids epidemic. In order to avoid double-counting, this table does not include the Title III funds Maine receives. Although some of these allocations pay for primary care, a share supports fqhcs and community health centers, which are counted in separate sections.

32 costs of caring for uninsured people in maine APPENDIX table b1 probit regression model for maine residence, among adults (19–64) in the northeast (unweighted n=50,843) variable coefficient variable coefficient Private insurance 0.00 industry

Public insurance 0.34 Mining -0.79

Female -0.03 Construction 0.14*

Metropolitan residence -1.02* Manufacturing 0.18 race ethnicity Retail 0.33

African-American -1.02* Transportation 0.24

Asian American -0.34 Information 0.25*

Native American 0.22* Financial 0.28

White Hispanic -0.86* Professional 0.17 family structure Education/Health 0.29

Married, spouse not present -0.10* Hospitality 0.13

Widowed, divorced, seperated 0.02* Other 0.16

Never married -0.13 Government 0.20 Number of adults in houshold -0.07 general health status

Number of veterans in household 0.20* Excellent -0.07 education Very good -0.07

High school diploma 0.05* Good -0.14

College degree -0.02* Fair -0.02 Post college education -0.08 age family income 19–24 0.13*

100–200% fpl 0.03 25–29 0.06

200–400 % fpl 0.04 30–34 0.06

400%+ fpl -0.10 40–44 0.02 employment status 45–49 0.02

Unemployed 0.02 50–54 0.00

Not in labor force -0.03* 55–59 -0.10*

Self-employed 0.08 60–64 -0.14* occupation interactions

Management -0.16 Female*Married, spouse not present 0.16

Professional -0.16* Female*Widowed, divorced, seperated -0.05

Service -0.09 Female*Never married 0.02

Sales -0.24 Female*High school diploma 0.02

Support -0.12 Female*college degree 0.12

Construction -0.09 Female*post college education 0.01

Production -0.18 Female*unemployed -0.13

Military 0.23 Female*not in labor force -0.05

Female*self-employed 0.07

Female*firm size 1–9 -0.06

Female*firm size 10–24 -0.07

* Statistically significant at the 5% level Constant -0.98* Source: Urban Institute analysis of 2004–2005 cps

costs of caring for uninsured people in maine 33 APPENDIX table b2 probit regression model for maine residence, among children (0–18) in the northeast

(unweighted n=27,165) variable coefficient Private insurance 0.13* Public insurance 0.37* Female 0.00 Metropolitan residence -1.03* race/ethnicity African-American -0.91* Asian American -0.43* Native American 0.17 White Hispanic -0.72* family structure Married, spouse not present 0.02 Widowed, divorced, seperated -0.06 Never married -0.08 Number of adults in houshold -0.07* Number of veterans in household 0.22* education High school diploma 0.14* College degree 0.09 Post college education -0.01 family income 100–200% FPL -0.06 200–400 % FPL -0.01 400%+ FPL -0.15* parent's employment status Unemployed -0.18* Not in labor force -0.01 general health status Excellent -0.14 Very good -0.13 Good -0.18 Fair -0.10 age 2–4 -0.06 5–9 -0.03 10–12 -0.02 13–18 -0.01 constant -0.89*

Source: Urban Institute analysis of 2004–2005 CPS * Statistically significant at the 5% level

34 costs of caring for uninsured people in maine APPENDIX table b3 cell-based reweighting adjustment, by age, race/ethnicity, and insurance status

meps northeasta cps maine b annual pop annual pop cell-based weight race/ethnicity insurance work status unwtd obs wtd % (sum of wtsc) unwtd obs wtd % (sum of wts) adjustmentd Adults White, Non-Hispanic Private Employed 3,582 42.8 468,161 2,311 41.8 457,719 0.978 White, Non-Hispanic Private Unemployed 230 2.2 24,113 70 1.3 14,379 0.596 White, Non-Hispanic Private Not in Labor Force 581 5.2 56,635 409 7.9 86,457 1.527 White, Non-Hispanic Public 515 10.9 119,041 525 9.1 99,856 0.839 White, Non-Hispanic Uninsured Employed 399 5.5 60,166 352 6.8 74,343 1.236 White, Non-Hispanic Uninsured Unemployed 67 0.8 8,399 47 1.1 11,518 1.371 White, Non-Hispanic Uninsured Not in Labor Force 122 1.4 15,072 102 2.0 21,654 1.437 African American Private Employed 544 0.1 1,452 10 0.1 1,475 1.015 African American Private Unemployed 39 0.0 168 0 0.0 0 1.000 African American Private Not in Labor Force 94 0.0 178 2 0.0 479 2.684 African American Public 437 0.2 2,037 8 0.1 1,419 0.697 African American Uninsured Employed 160 0.1 610 2 0.0 411 0.674 African American Uninsured Unemployed 34 0.0 51 0 0.0 0 1.000 African American Uninsured Not in Labor Force 63 0.0 103 3 0.1 494 4.774 Asian & Amer. Indian Private Employed 290 0.8 8,975 41 0.6 6,993 0.779 Asian & Amer. Indian Private Unemployed 11 0.0 118 1 0.0 148 1.255 Asian & Amer. Indian Private Not in Labor Force 61 0.1 1,037 14 0.2 2,368 2.283 Asian & Amer. Indian Public 64 0.2 2,419 27 0.4 4,787 1.979 Asian & Amer. Indian Uninsured Employed 65 0.2 1,723 16 0.3 2,840 1.649 Asian & Amer. Indian Uninsured Unemployed 5 0.0 183 2 0.0 307 1.678 Asian & Amer. Indian Uninsured Not in Labor Force 26 0.0 259 4 0.1 536 2.068 Hispanic Private Employed 502 0.2 1,772 20 0.2 2,536 1.431 Hispanic Private Unemployed 30 0.0 116 0 0.0 0 1.000 Hispanic Private Not in Labor Force 82 0.0 248 2 0.0 372 1.497 Hispanic Public 407 0.2 1,613 12 0.2 1,902 1.179 Hispanic Uninsured Employed 256 0.1 638 5 0.1 628 0.984 Hispanic Uninsured Unemployed 33 0.0 262 0 0.0 0 1.000 Hispanic Uninsured Not in Labor Force 138 0.0 141 1 0.0 110 0.782 Children White, Non-Hispanic Private 1,763 18.0 196,579 1,329 18.1 198,759 1.011 White, Non-Hispanic Public 553 8.7 95,309 455 6.4 70,106 0.736 White, Non-Hispanic Uninsured 153 1.1 11,707 119 1.6 17,226 1.472 African American Private 346 0.1 1,513 10 0.2 1,822 1.205 African American Public 585 0.2 2,408 13 0.2 2,292 0.952 African American Uninsured 82 0.0 217 0 0.0 0 1.000 Asian & Amer. Indian Private 152 0.3 3,757 16 0.2 2,280 0.607 Asian & Amer. Indian Public 52 0.1 1,245 24 0.3 3,570 2.867 Asian & Amer. Indian Uninsured 16 0.0 249 3 0.1 605 2.430 Hispanic Private 281 0.2 1,842 26 0.3 3,497 1.898 Hispanic Public 703 0.4 4,440 9 0.1 1,147 0.258 Hispanic Uninsured 84 0.0 263 3 0.1 553 2.101 All Nonelderly 13,607 100.0 1,095,217 5,993 100.0 1,095,584 1.000

a. 2001–2003 MEPS Northeast sample with 13,607 unweighted observations. b. 2003–2004 CPS Maine sample with 5,993 unweighted observations. c. Sum of weights after statistical reweighting. d. Ratio of CPS Maine weighted population to MEPS Northeast weighted population. Source: Urban Institute analysis of 2002–2004 MEPS data reweighted to represent Maine population costs of caring for uninsured people in maine 35 APPENDIX table b4

percentage distrubutions of population characteristics before and after reweighting meps northeast (2002–2004) to cps maine (2004–2005)

meps before difference meps after difference cps for characteristic reweighting meps–cps reweighting meps–cps maine gender, race, ethnicity (%) Female 50.4 -0.2 48.8 -1.8 50.6 White, non-Hispanic 71.2 -24.9 96.0 0.0 96.0 African-American 13.2 12.4 0.8 0.0 0.8 Asian American 5.3 3.1 2.2 0.0 2.2 White Hispanic 10.4 9.4 1.0 0.0 1.0 metropolitan residence (%) 89.6 48.9 39.7 -1.0 40.7 marital status Married, spouse not present 1.1 -0.1 0.8 -0.4 1.2 Married, spouse present 57.9 -2.0 58.4 -1.5 59.8 Widowed, divorced, seperated 11.3 -6.0 16.0 -1.3 17.3 Never married 22.7 1.2 18.8 -2.6 21.4 adult/parent education (%) Less than high school 10.3 1.8 9.2 0.8 8.4 High School diploma 59.7 -7.5 68.9 1.7 67.3 College degree 19.4 2.6 15.4 -1.4 16.8 Post college education 10.6 3.4 6.5 -0.7 7.2 family income relative to poverty (%) Less than 100% FPL 11.6 -0.7 12.0 -0.3 12.3 100–200% FPL 14.1 -2.3 17.2 0.8 16.4 200–400% FPL 29.3 -6.3 34.8 -0.9 35.7 400%+ FPL 44.9 9.3 36.0 0.4 35.7 employment status (%) Employed 76.5 1.8 76.9 2.2 74.7 Unemployed 5.9 1.9 4.0 0.0 4.0 Not in labor force 16.9 -4.3 18.9 -2.3 21.2 Self-employed 7.0 -0.3 9.6 2.3 7.3 general health status (%) Excellent 37.4 1.3 34.6 -1.6 36.2 Very good 32.0 -1.5 34.0 0.5 33.5 Good 22.4 2.5 19.1 -0.9 19.9 Fair 6.1 -1.3 8.7 1.2 7.5 Poor 2.0 -0.9 3.6 0.7 2.9 insurance coverage, non-elderly (%) Private 73.6 2.5 71.1 0.0 71.1 Public 14.9 -2.0 16.9 0.0 16.9 Uninsured 11.5 -0.5 12.0 0.0 12.0

36 costs of caring for uninsured people in maine APPENDIX table b4 (continued)

percentage distrubutions of population characteristics before and after reweighting meps northeast (2002–2004) to cps maine (2004–2005)

meps before difference meps after difference cps for characteristic reweighting meps–cps reweighting meps–cps maine age (%) 0–1 2.6 0.0 2.7 0.1 2.6 2–4 4.5 0.9 3.4 -0.2 3.6 5–9 7.3 0.6 7.1 0.4 6.7 10–12 4.8 0.3 4.3 -0.2 4.5 13–18 10.0 -0.4 10.1 -0.3 10.3 19–24 8.6 -0.4 7.6 -1.4 9.0 25–29 7.6 1.2 6.2 -0.1 6.3 30–34 8.0 0.2 9.0 1.2 7.8 35–39 8.1 0.5 8.2 0.6 7.5 40–44 9.8 0.5 9.4 0.0 9.3 45–49 9.0 -1.2 9.8 -0.3 10.2 50–54 7.8 -1.4 9.3 0.1 9.2 55–59 6.8 -0.9 7.9 0.2 7.7 60–64 5.2 -0.1 5.0 -0.2 5.2 industry (%) Natural resources 1.0 -1.0 2.1 0.2 1.9 Mining 0.1 0.0 0.2 0.2 0.1 Construction 6.8 0.2 9.1 2.5 6.6 Manufacturing 9.8 -0.1 8.8 -1.0 9.9 Retail 11.4 -2.2 13.9 0.3 13.6 Transportation 4.4 0.6 4.4 0.5 3.8 Information 2.3 0.5 1.7 -0.1 1.8 Financial 6.3 0.8 4.3 -1.1 5.4 Professional 9.3 3.0 5.8 -0.4 6.3 Education/health 17.3 -3.5 18.2 -2.6 20.8 Hospitality 6.5 0.5 6.5 0.5 6.0 Other services 3.5 -0.6 3.4 -0.7 4.1 Public 4.5 1.0 6.0 2.5 3.5 Military 0.2 0.2 0.1 0.1 0.0 occupation (%) Management 14.0 3.1 10.7 -0.1 10.8 Professional 17.8 1.6 14.8 -1.4 16.3 Service 12.5 -2.2 15.7 1.1 14.7 Sales 8.7 0.1 9.5 0.8 8.6 Support 9.6 -2.2 8.9 -2.8 11.7 Farming 0.7 -0.4 1.3 0.2 1.1 Construction 9.3 -0.4 11.7 2.0 9.7 Production 10.3 -0.5 11.2 0.4 10.8 Military 0.2 0.2 0.1 0.1 0.0

Source: Urban Institute analysis of the MEPS (2002–2004) CPS (2004–2005).

costs of caring for uninsured people in maine 37 APPENDIX table b5 selected characteristics by insurance status

all uninsured most insured majority observations the year of the year (unweighted n) (13,372) (1,693) (11,679) Percent of year insured 85.8 7.4 97.4 Metropolitan residence 39.7 37.7 40.0 Female 48.8 35.3 50.8 age 0–9 13.2 5.3 14.4 10–19 14.3 9.0 15.1 19–34 22.8 38.8 20.5 35–64 49.6 46.9 50.0 race/ethnicity White, non Hispanic 96.0 95.1 96.2 White, Hispanic 1.0 1.0 1.0 African American 0.8 0.7 0.8 Asian/Native American 2.2 3.2 2.1 education Less than high school 9.2 21.9 7.3 High school diploma 68.9 61.7 70.0 College degree 15.4 12.7 15.8 More than college 6.5 3.7 6.9 family income LT 100% FPL 12.0 20.5 10.7 100–200% FPL 17.2 24.0 16.2 200–400% FPL 34.8 32.9 35.1 400%+ FPL 36.0 22.6 38.0 marital status Married 58.4 30.0 62.5 Married no spouse present 0.8 1.1 0.8 Widowed, divorced, seperated 16.0 25.7 14.6 Never married 18.8 37.5 16.1 general health status Excellent 34.6 27.5 35.7 Very good 34.0 36.3 33.7 Good 19.1 25.6 18.1 Fair 8.7 9.2 8.6 Poor 3.6 1.3 3.9

38 costs of caring for uninsured people in maine APPENDIX table b5 (continued) selected characteristics by insurance status

all uninsured most insured majority observations the year of the year (unweighted n) (13,372) (1,693) (11,679) activity limitation Unable to perform usual activities 3.3 1.3 3.6 Social or cognitive limitation 7.4 5.5 7.6 Activity limitation 4.4 6.3 4.2 Deceased or institutionalized 0.5 0.9 0.4 Required aid help 2.5 2.7 2.4 conditions Diabetes 4.5 1.7 4.9 Hypertension 9.9 4.5 10.7 Asthma 6.2 5.0 6.4 Back disorders 10.8 9.6 11.0 Infectious disease 20.0 14.9 20.7 Endocrine conditions 13.2 6.5 14.2 Blood disorders 1.2 0.3 1.3 Cardiovascular conditions 5.7 2.1 6.2 Bronchitis 4.5 3.3 4.7 Digestive conditions 15.3 8.6 16.3 Genito urinary conditions 11.7 4.5 12.8 Skin condition 9.3 5.5 9.8 Musculoskeletal condition 16.4 8.6 17.6 Fracture 3.5 3.8 3.4 Completely unable to work 4.5 1.5 5.0 Inner ear Infection 5.0 2.0 5.4 Cancers 2.3 0.3 2.6

Source: Urban Institute analysis of the meps (2002–2004) cps (2004–2005).

costs of caring for uninsured people in maine 39 APPENDIX table c1 two-part spending models, non-elderly adults

coefficents any expenditure total expenditure variable (logit) (glm) (unweighted n) (5,502) (4,435)

insurance Percent of year insured 1.16* 0.61* Metropolitan residence 0.23 0.07 Female 1.35* 0.18* age 19–24 -0.28 0.03 25–29 -0.29 -0.22 30–34 -0.57 0.18 35–39 -0.07 0.01 40–44 -0.26 0.00 45–49 -0.13 0.03 50–54 -0.04 -0.05 55–59 -0.17 0.02 race/ethnicity White, Hispanic -0.28 -0.09 African American -0.70* 0.04 Asian/Native American -0.08 -0.42* education Less than high school -0.44 -0.42* High school diploma 0.04 0.03 College degree 0.05 0.18 family income

100–200% FPL -0.49 -0.23

200–400% FPL -0.34 -0.50*

400%+ FPL -0.27 -0.58* marital status Married no spouse -0.31 -0.44* Widowed, divorced, seperated -0.91* 0.02 Never married 0.16 0.07 general health status Very good -0.06 0.19 Good 0.26 0.27* Fair -0.64 0.44* Poor -2.79* 0.57*

40 costs of caring for uninsured people in maine APPENDIX table c1 (continued) two-part spending models, non-elderly adults

coefficents any expenditure total expenditure variable (logit) (glm) (unweighted n) (5,502) (4,435)

activity limitation adl/iadl 1.02 0.34 Unable to perform usual activities 2.97 0.48 Social or cognitive limitation 0.00 0.15 Activity limitation -0.69 0.16 Deceased or institutionalized 0.00 0.79* Required aid help 4.88* 0.44 conditions Diabetes 5.16* 0.54* Hypertension 3.85* 0.10 Asthma 3.88* 0.15 Back disorders 1.27* 0.24* Infectious disease 0.42 0.24* Endocrine conditions 0.49 0.16 Blood disorders 4.48* 0.51 Cardiovascular conditions 4.72* 0.41* Bronchitis 3.18* 0.16 Digestive conditions 1.68* 0.28* Genito uninary conditions 4.73* 0.31* Skin condition 5.05* 0.37* Musculoskeletal condition 2.12* 0.34* Fracture 1.07 0.95* Inner ear Infection 3.55* 0.43 Cancers 5.58* 1.12* Pregnancya 1.31 Constant -0.01 7.05*

Source: Urban Institute analysis of 2002–2004 meps data reweighted to represent Maine population Note: * Statistically significant at the 5% level a. Omitted from logit model because variable perfectly predicts having expenditures

costs of caring for uninsured people in maine 41 APPENDIX table c2 two-part spending models, children

coefficents any expenditure total expenditure variable (logit) (glm) (unweighted n) (3,609) (3,091)

insurance Percent of year insured 1.53* 0.17 Metropolitan residence -0.68* 0.14 Female 0.19 -0.36* age 0–1 0.72 -0.09 2–4 -0.62 -0.72* 5–9 -0.05 -0.66* 10–12 0.04 -0.20 race/ethnicity Minority -0.12 0.24 education Less than high school -2.34* 0.52* High school diploma -1.90* 0.49* College degree -1.19 0.31 family income

100–200% fpl -0.17 -0.38* 200–400% fpl -0.29 -0.32* 400%+ fpl 0.00 -0.10 marital status Widowed, divorced, seperated -0.21 0.59 Never married 0.32 -0.35 general health status Very good 0.10 0.29* Good 0.89* 0.18 Fair -0.12 1.28* Poor 6.98* 0.64 activity limitation ADL/IADL 0.19 1.96 Unable to perform usual activitiesa 1.08 Social or cognitive limitation 0.21 -0.52 Activity limitation -2.17* 0.80 Deceased or institutionalized -1.58 0.68 Required aid helpa 0.88

42 costs of caring for uninsured people in maine APPENDIX table c2 (continued) two-part spending models, children

coefficents any expenditure total expenditure variable (logit) (glm) (unweighted n) (3,609) (3,091)

conditions Diabetesa 0.11* Hypertensiona -0.43 Asthma -0.23 0.81* Back disorders 2.40* 1.63* Infectious disease 1.95* 0.17 Endocrine conditionsa -0.08 Blood disordersa -0.39 Cardiovascular conditions 4.57* -1.26* Bronchitis 4.86* 0.15 Digestive conditions 2.11* 0.46* Genito urinary conditions 2.17* 0.69* Skin condition 3.30* 0.19 Musculoskeletal conditiona 0.27 Fracture 4.99* 0.90* Inner ear infection 2.70* 0.49* Cancersa 3.32* Pregnancya 0.39 Constant 2.89* 6.56*

Source: Urban Institute analysis of 2002–2004 MEPS data reweighted to represent Maine population Note: * Statistically significant at the 5% level a. Omitted from logit model because variable perfectly predicts having expenditures.

costs of caring for uninsured people in maine 43 end notes

44 costs of caring for uninsured people in maine 1 The estimate of the number of uninsured in Maine is based on the 10 If we had assumed that only 25 percent of bad debt was due to recently revised CPS that corrects for a nationwide data processing insured patients, our estimate of total hospital uncompensated problem that understate the numbers of people who were assigned care costs provided to the uninsured in 2005 would have been dependent coverage. $85.6 million. Alternatively, had we assumed that 50 percent of See http://www.census.gov/Press-Release/www/releases/. bad debt was due to insured patients, our estimate of total hospital uncompensated care costs provided to the uninsured in 2005 2 J. Hadley and J. Holahan, “How Much Medical Care Do the would have been $68.4 million. Uninsured Use and Who Pays for It?” Health Affairs Web Exclusive (12 Feb 2003): w366–w381. 11 Some of this care could have been provided to insured patients, and this could tend to overstate our estimate of the cost of un- 3 J. Holahan, R. Bovbjerg, and J. Hadley, “Caring for the Uninsured compensated care provided by physicians to the uninsured. in Massachusetts: What Does It Cost, Who Pays and What Would Full Coverage Add to Medical Spending?” Report for the Blue 12 Weiss, G. “Exclusive Survey: Practice Expenses,” Medical Cross Blue Shield of Massachusetts Foundation, November 2004. Economics 80(31), Nov. 7, 2003.

4 Multiplying charges by a cost-to-charge ratio is a standard way of 13 J. Hadley and J. Holahan, “The Cost of Care for the Uninsured,” deriving the cost of resources used to provide care. Kaiser Family Foundation publication, May 10, 2004.

5 This figure may exclude some costs borne by hospitals for uncom- 14 J. Taylor, “Fundamentals of Community Health Centers,” pensated care. For example, hospital-based clinics and hospital- National Health Policy Forum, George Washington University, affiliated faculty practice plans provide a significant amount of August 31, 2004. uncompensated care. However, due to differences in accounting 15 John Snow, Inc. State of Maine Primary Care Safety Net methodologies between hospital systems, the total cost of uncom- Environmental Scan. Feb. 2006. Submitted to the Maine pensated care provided in these settings may not be captured by Health Access Foundation. the data provided through the AHA’s Annual Survey. 16 J. Hadley, M. Cravens, T. Coughlin, and J. Holahan, “Federal 6 Holahan, Bovbjerg, and Hadley, 2004. Spending on the Health Care Safety Net: 2001–2004,” Kaiser 7 Kane N. House Subcommittee on Oversight, Committee on Ways Commission on Medicaid and the Uninsured, November 2005. and Means, US House of Representatives, “Medical Bad Debt, A 17 This approach is analogous to the approach used to analyze Growing Public Health Crisis,” June 22, 2004. uncompensated care costs in hospitals. 8 Kane N. “Bad Debt and Free Care Baseline Analysis and Recom- 18 Between 55.0% and 55.6% of VA expenditures in Maine were mendations for Purposes of Determining Savings Offset Pay- devoted to medical care in FY 2002, 2003, and 2004. See ments” Governor’s Office of Health Policy and Finance, Department of Veterans Affairs, “Geographic Distribution of December 9, 2004. VA Expenditures,” 2002–2004, www.va.gov/vetdata/Geograph- 9 Zuckerman, Stephen, Bovbjerg, Randall R., Hadley, Jack, icInformation/index.htm (6 April 2005). Cravens, Matthew and Clemans-Cope, Lisa, “The Cost of Care 19 80 % represents the share of national VA appropriations budgeted for Missouri’s Uninsured,” Missouri Foundation for Health Report, for acute hospital and outpatient care services, along with a October 2006. http://mffh.org/CoverMoDataBook2.pdf Holahan, proportionate amount of general operating expenses (adminis- John, Bovbjerg, Randall R., and Hadley, Jack “Caring for the Unin- tration and oversight). See Office of Management and Budget, sured in Massachusetts: What Does it Cost, Who Pays, and What “Budget of the United States Government, Fiscal Year 2005— Would Full Coverage Add to Medical Spending?” Blue Cross and Appendix,” http://www.whitehouse.gov/omb. Blue Shield Foundation of Massachusetts Report, November 2004. http://roadmaptocoverage.org/pdfs/roadmapReport.pdf 20 Congressional Budget Office, “The Potential Cost of Meeting Demand for Veterans’ Health Care,” March, 2005.

costs of caring for uninsured people in maine 45 21 Nationally, VA enrollees with more serious service-connected 28 These other sources might include payments from private disabilities—especially those in Priority Groups 1, 4, and 5— philanthropies and other types of insurance (e.g., automobile tend to rely more heavily on VA care and are sicker, on average. or liability). This latter category would not literally represent uncompensated care, but we include it in the definition because 22 Calculated as a two-year average of the 2004 and 2005 percent- it is combined with other sources in the MEPS data. ages. “Other sources” include CHAMPUS/Tricare, CHAMPVA, Medicare, Medicaid, SCHIP, and all private plans. The 24.4 29 The costs of donated care are absorbed as a loss by providers or percent estimated of the share of uninsured VA users who are may be cross-subsidized from payments for other payers that uninsured is roughly comparable to national averages. Although exceed the costs of care or from payments through government VA is the main payer for CHAMPVA, it is a form of insurance programs (e.g., DSH, IME or GME payments). rather than direct care; services may be received at non-VA 30 John Holahan, Randall Bovbjerg and Jack Hadley, “Caring for facilities. As such, budget appropriations for CHAMPVA are the Uninsured in Massachusetts: What Does It Cost, Who Pays excluded in estimating care provided to the uninsured. and What Would Full Coverage Add to Medical Spending?” Blue 23 “2001 FEHP Disparity Index and IHCIF Calculations for Cross and Blue Shield Foundation of Massachusetts, November Operating Units: Nashville Area.” Indian Health Service, Indian 2004; Jack Meyer and Jack Hadley, “Mapping Health Spending Health Care Improvement Fund. http://www.ihs.gov/nonmedi- and Insurance Coverage in Connecticut,” report prepared for the calprograms/lnf/ihcif2002/nashville.pdf. Last accessed 4.6.2007. Universal Health Care Foundation of Connecticut, November 2005; Randall Bovbjerg, Stan Dorn, Jack Hadley, John Holahan 24 Some people appear in the sample twice because interviews are and Dawn Miller, “Caring for the Uninsured in New York: What conducted over multiple years. The person-level weight is differ- Does It Cost, Who Pays and What Would Full Coverage Add to ent for each year’s record. Because of MEPS’ sample design, and Health Care Spending?” Final report prepared for the New York to increase the number of observations, we leave in all persons Community Trust. (Washington: The Urban Institute Press, who appear in multiple years’ data files. October 2006). 25 Congressional Budget Office, “How Many People Lack Health 31 Jack Hadley and John Holahan, “Who Pays and How Much? Insurance and For How Long?”, Economic and Budget Issue The Cost of Caring for the Uninsured,” Kaiser Commission on Brief, May 2003, http://www.cbo.gov/showdoc.cfm?index=4211 Medicaid and the Uninsured, February 2003; Sing et al., 2006, &sequence=0; J. Holahan, G. Kenney, and L. Nichols, “Towards “Reconciling Medical Expenditure Estimates from the MEPS and a Federal Survey of Health Insurance Coverage and Access,” the NHEA, 2002,” Health Care Financing Review, Fall 28(1): 25–40. Urban Institute Working Paper, May 2004; National Institute for The MEPS-HC values by source of payment are adjusted using Health Care Management Research and Educational Foundation, the following multipliers derived from Sing et al. (2006): 1.125 “Health Insurance Coverage in the U.S.: The New Census Bureau for Medicare, 1.53 for Medicaid, 1.262 for private, and 0.977 for all Numbers for 2000 and The Trend into 2001,” September 2001. other sources. However, we do not adjust the MEPS-HC estimate 26 The 7-month threshold is applied as a percentage of available of out-of-pocket spending because it is obtained directly from months of insurance data, or 58.3%, for those who enter or leave household respondents and is more likely to be accurate than the the survey part way through a year. NHEA estimate, which is constructed essentially as a residual.

27 Jack Hadley and John Holahan, “Who Pays and How Much? The 32 David Rousseau and Andy Schneider, Current Issues in Medicaid Cost of Caring for the Uninsured,” Kaiser Commission on Financing—An Overview of IGTs, UPLs, and DSH, Issue Brief for Medicaid and the Uninsured, February 2003; Jack Hadley, the Kaiser Commission on Medicaid and the Uninsured, pub. Matthew Cravens, Terri Coughlin and John Holahan, “Federal no. 7071, April 2004, accessible from http://www.kff.org/medic- Spending on the Health Care Safety Net from 2001–2004: Has aid/7071.cfm. Spending Kept Pace with the Growth of the Uninsured,” Kaiser Family Foundation, Washington DC, November 2005.

46 costs of caring for uninsured people in maine 33 The amount of federal funding was $491 per uninsured person in 40 Trish Riley and Elizabeth Kilbreth, “Health Coverage in the Maine (the state share added about half that amount again), as States—Maine’s Plan for Universal Access,” New England compared to $195 per uninsured person for the U.S. as a whole. Journal of Medicine 350(4):330–332, 2004. The initiative’s See Table 5-2, pp.160–161, Teresa A. Coughlin and Stephen Zuck- webpage is http://www.dirigohealth.maine.gov/ erman, “States’ Use of Medicaid Maximization Strategies to Tap 41 The following Census Bureau estimates for uninsured rates differ Federal Revenues: Program Implications and Consequences,” from those provided in this report’s accompanying chart book 145–178 in Federalism and Health Policy, edited by John Holahan, because the chart book excludes the elderly (who have higher Alan Weil, and Joshua M. Weiner/ Washington, DC: The Urban rates of insurance) from its estimates: According to the U.S. Cen- Institute Press, July 2003. DSH funding supported the state’s sus Bureau, for 1999–2000 (two year average), 11.4% of Mainers former Hospital Uncompensated Care and Governmental were uninsured vs. 14.2% of all Americans. For 2004–05 the cor- Payment Shortfall Fund, Maine Rev. Stat. Ann., tit. 22, § 395-B responding figures were 10.7% for Maine and 15.9% for the U.S. (Charity care), repealed. Compare Robert J. Mills, U.S. Census Bureau, Current Popula- 34 DSH funding has continued throughout for “institutions tion Reports P60–215, Health Insurance Coverage: 2000, September for mental disease,” or IMDs, which are outside the scope of 2001 (Table D) http://www.census.gov/prod/2001pubs/p60- this report. 215.pdf with Carmen DeNavas-Walt, Bernadette D. Proctor, and Cheryl Hill Lee, U.S. Census Bureau, Current Population 35 Maine’s Medicaid Payment Shortfall: An Analysis of Medicaid Reports, p60–231, Income, Poverty, and Health Insurance Reimbursement to Maine Hospitals, Prepared by Baker, Newman Coverage in the United States: 2005, U.S. Government Printing & Noyes, LLC for the Maine Hospital Association, July 2002 Office, Washington, DC, August 2006 (Table 10) http://www.themha.org/pubs/Maine_s%20Medicaid%20Payme http://www.census.gov/prod/2006pubs/p60-231.pdf nt%20Shortfall.pdf 42 See discussion at pp. 46 and following in Maine Continuation 36 Paul Saucier, MaineCare and Its Role in Maine’s Healthcare System, State Planning Grant Final Report, Submitted to USDHHS/HRSA report to Kaiser Commission on Medicaid and the Uninsured by The Governor’s Office of Health Policy and Finance Maine from Muskie School of Public Service, January 2005. http:// State Government And The Health Policy Institute, The Muskie muskie.usm.maine.edu/Publications/ihp/MaineCare2005Kaiser. School of Public Service, University of Southern Maine, pdf; Alshadye Yemane and Ian Hill. Model Waiver Evaluation— February, 2006 (2P09OA00042-02-00) The Health Insurance Flexibility and Accountability Initiatives: http://statecoverage.net/pdf/mainefinalreport.pdf Case Study Report for Maine. Washington, DC: The Urban Institute, Final Report to the Office of Research and Demonstra- 43 Frequently Asked Questions About the DirigoChoice SOP tions, Centers for Medicare & Medicaid Services, July 2005. Payment State of Maine, Dirigo Health Plan website. http://www.dirigohealth.maine.gov/sop_pay_faq.html 37 CMS Medicaid Waiver and Demonstration List. http://www.cms.hhs.gov/MedicaidStWaivProgDemoPGI/ 44 Jerry Harkavy, “Maine Court Sets May Deadline for Dirigo MWDL/itemdetail.asp?filterType=none&filterByDID=-99&sort Savings Report,” Insurance Journal, April 16, 2006 http://www. ByDID=2&sortOrder=ascending&itemID=CMS042935&intNu insurancejournal.com/news/east/2006/04/16/67330.htm mPerPage=10 45 Mayo Regional Hospital is a hospital-district owned hospital 38 $100 million is the 2004 figure, according to Saucier op cit. with 25 beds. Cary Medical Center, with 65 beds, is a private Federal rules have allowed slow increase in DSH, but if Maine not-for profit, but its building is owned by the Caribou Hospital exceeds any year’s ceiling it will lose federal support. The state is District. Bed counts from Maine Hospitals Association authorized to cap enrollment under the waiver, which it did in 2005. http://www.themha.org/members/hospitalsbysize.htm

39 Yemane and Hill 2005.

costs of caring for uninsured people in maine 47 46 “Even when compared to the United States as a whole, with its 55 USDHHS, ASPE, Federal Medical Assistance Percentages vast rural areas, Maine’s population density is only slightly more http://aspe.hhs.gov/health/fmap.htm than 50% of average US population density.” p. 18, 2007 State 56 CMS Form 64 2005 inpatient DSH. Health Plan, issued April 2006 http://www.dirigohealth.maine.gov/dhlp09.htm 57 Final Rule: MaineCare Benefits Manual, Chapters II & III, Section 45, Hospital Services. Maine Department of Health and 47 Two thirds of hospitals have 100 or fewer beds, according to the Human Services. July 1, 2006. http://www.maine.gov/dhhs/bms/ hospital association, http://www.themha.org/members/hospi- rules/downloads/c_ii_iii_s_45_f_complete.pdf talsbysize.htm. For local maps of various providers see Patricia Fairchild, Jonathan Stewart, Alec McKinney, and Ashley Marks, 58 On the other hand, we were told that the CAH above-cost State of Maine Primary Care Safety Net Environmental Scan, Bos- payment add-on can also be conceptualized as an offset to CAHs’ ton, MA: John Snow, Inc. Final Report to Maine Health Access payments of a hospital revenue tax. The revenue levy of 2.23% Foundation, February 2006, Appendix 3 of net patient revenues (i.e revenue collected) passed in 2003 http://www.mehaf.org/pictures/jsi_b.pdf to support Medicaid spending growth and draw down federal match; hospitals benefited along with all other providers from 48 There are no Maine members of the National Association of more generous Medicaid spending, but most were net payers (to Public Hospitals and Health Systems, for example. the extent that they could not immediately pass on the increased 49 However, Maine Medical Center in Portland is a large tertiary costs to their payers). Maine Hospital Association, MaineCare referral center with many training programs and numerous Hospital Reimbursement (undated, internal evidence suggests resident physicians. 2005) www.themha.org/pubs/Medicaid%20Flyer%20FINAL.pdf. For this analysis, we assume that the CAH add-on helps offset the 50 Free Care Guidelines, 10-144 Code of Maine Rules (CMR) cost of care for the uninsured to some extent. chapt 150, Department of Health and Human Services, Office of Mainecare Services http://www.maine.gov/sos/ 59 Data on Medicare payments were provided by MedPAC, cec/rules/10/144/144c150.doc. February 13, 2007.

51 Maine Hospital Association, “Survey of Free Care Policies as 60 2004 CMS Data on DSH, IME, and GME by hospital. 2005 data of March 20, 2007” were incomplete. http://www.themha.org/pubs/charitypolicy.pdf 61 2004 CMS Data on DSH, IME, and GME by hospital. 2005 data 52 “Maine’s Hospitals: A Caring Mission.” Maine Hospital were incomplete. Association. April 2005. http://www.themha.org/pubs/ 62 Prospective Payment Assessment Commission, Medicare and the A%20Caring%20Mission.pdf American Health Care System (Washington, DC: June 1997) and 53 Maine Statute. Title 22, Chapter 401, General Provision. Prospective Payment Assessment Commission, Report and §1715. Access requirements applicable to certain health care Recommendations to Congress (Washington, DC: March 1997). providers. http://janus.state.me.us/legis/statutes/22/ 63 Some Title III and IV allocations pay for primary care (early title22ch401.rtf (page 17) intervention services with Title III and services to women 54 Patricia Fairchild, Jonathan Stewart, Alec McKinney, and Ashley and children with Title IV). However, to avoid funds double- Marks, State of Maine Primary Care Safety Net Environmental counting in our estimate of care to the uninsured, we exclude Scan, Boston, MA: John Snow, Inc. Final Report to Maine Health these Titles. A share of Title III allocations support Federally Access Foundation, February 2006, http://www.mehaf.org/pic- Qualified Health Centers, which are counted in the section on tures/jsi_b.pdf. An Urban Institute interview references a recent community health centers; and Title IV funds some maternal survey findings that 41% of physicians work for a hospital or and child health services. FQHC, higher for primary care physicians.

48 costs of caring for uninsured people in maine 64 Kaiser State Health Facts Online, “Maine: Estimated Number Sanford Kiwanis Free Clinics, Bates Street Clinic, Penobscot of Persons Living with AIDS, All Ages, At the End of 2005.” Community Health Center, Knox County Health Center, and The Henry J. Kaiser Family Foundation, available at Leavitt’s Mill Health Center. The list does not claim to be www.statehealthfacts.org. comprehensive. Source: Consumers for Affordable Health Care, Guide to Health Care: Navigating Maine’s Health Care System: 65 MCH Bureau, Title V Information System (TVIS), “Program 2007 Edition, section 10, Resource Directory, Last updated on Data: Number of Individuals Served by Title V, by Class of 3/29/06 http://www.mainecahc.org/healthcare/ Individuals: Maine,” available at https://perfdata.hrsa.gov/ otherprogramsinmaine.htm#regional mchb/mchreports/search/program/prgsch04_result.asp 75 Patricia Fairchild, Jonathan Stewart, Alec McKinney, and 66 MCH Bureau, Title V Information System (TVIS), “Federal-State Ashley Marks, State of Maine Primary Care Safety Net Title V Block Grant Partnership Expenditures by Source of Environmental Scan, Boston, MA: John Snow, Inc. Final Report Funding: Maine,” available at https://perfdata.hrsa.gov/mchb/ to Maine Health Access Foundation, February 2006, http://www. mchreports/Search/special/finsch06_history_result.asp mehaf.org/pictures/jsi_b.pdf 67 Patricia Fairchild, Jonathan Stewart, Alec McKinney, and Ashley 76 Maine Assembly on School-Based Health Care. “Maine Marks, State of Maine Primary Care Safety Net Environmental School-Based Health Center Data Summary 04-05.” http:// Scan, Boston, MA: John Snow, Inc. Final Report to Maine Health measbhc.org/mainesbhcdatasummary0405.shtml. Last accessed Access Foundation, February 2006, http://www.mehaf.org/pic- 4.3.2007; Maine Assembly on School-Based Health Care. “Maine tures/jsi_b.pdf School-Based Health Centers: Sound Investment in Health and 68 Although many NHSC clinicians serve in FQHCs, the NHSC data Education,” http://measbhc.org/uploads/maine_sbhc_facts-1- system only reports on sites that are not FQHCs. This avoids 06.doc. Last accessed 4.11.2007. double-counting uncompensated care from FQHCs. 77 Patricia Fairchild, Jonathan Stewart, Alec McKinney, and 69 NHSC Uniform Data System (UDS) report, 2005. Ashley Marks, State of Maine Primary Care Safety Net Environmental Scan, Boston, MA: John Snow, Inc. Final Report 70 Lenardson, Jennifer D. and Hartley, David. “Issue Brief: to Maine Health Access Foundation, February 2006, Appendices Maine’s Rural Health Challenges.” Prepared for the Legislative . Policy Forum on Health Care by the Muskie School of Public Service at the University of Southern Maine and the Margaret 78 Patricia Fairchild, Jonathan Stewart, Alec McKinney, and Ashley Chase Smith Policy Center at the University of Maine. Funded Marks, State of Maine Primary Care Safety Net Environmental by Maine Health Access Foundation. January 26, 2007. Scan, Boston, MA: John Snow, Inc. Final Report to Maine Health Available at www.mdf.org. Access Foundation, February 2006, http://www.mehaf.org/pictures/jsi_b.pdf 71 Lenardson and Hartley, 2007. 79 Fairchild et al., 2006 also classifies the city of Auburn as part of 72 HRSA Office of Rural Health Policy Grants-Maine. the primary care safety net, but lists only population-oriented or http://ruralhealth.hrsa.gov/map/maine.htm other public health services.

73 Patricia Fairchild, Jonathan Stewart, Alec McKinney, and 80 Portland Health and Human Services Department web site. Ashley Marks, State of Maine Primary Care Safety Net http://www.portlandmaine.gov/hhs/hhs.asp. Environmental Scan, Boston, MA: John Snow, Inc. Final Report Last accessed 4.2.2007. to Maine Health Access Foundation, February 2006, http://www. mehaf.org/pictures/jsi_b.pdf 81 Portland Health and Human Services Department web site. http://www.portlandmaine.gov/hhs/hhs.asp. 74 This includes the Oasis Free Clinic, Portland Community Last accessed 4.2.2007. Free Clinic, Ellsworth Free Clinic, Biddeford Free Clinic,

costs of caring for uninsured people in maine 49 82 Hospitals and the federal government also provide assistance. 89 Some Title III and IV allocations pay for primary care (early Maine Primary Care Association, Choosing Primary Care in intervention services with Title III and services to women Maine with a $200K Debt Load: Still a Feasible Choice, and children with Title IV). However, to avoid funds double- White Paper, March 2006 http://69.80.208.229/mepcaorg/ counting in our estimate of care to the uninsured, we exclude pictures/pdfs/Loan%20Repayment%20White%20Pape%2003- these Titles. A share of Title III allocations support Federally 12-06.pdf Qualified Health Centers, which are counted in the section on community health centers; and Title IV funds some maternal 83 Patricia Fairchild, Jonathan Stewart, Alec McKinney, and and child health services. Ashley Marks, State of Maine Primary Care Safety Net Environmental Scan, Boston, MA: John Snow, Inc. Final Report 90 Because these data are not available for Maine specifically, we use to Maine Health Access Foundation, February 2006, http://www. the national uninsured rate among ADAP patients as an estimate mehaf.org/pictures/jsi_b.pdf for this rate in Maine. J. Kates et al, National ADAP Monitoring Project: 2005 Annual Report, (Washington, DC: 84 Catherine Ormond and Sarah Gerrish, “Opportunities and The Henry J. Kaiser Family Foundation and the National Challenges: Improving Access and Health Outcomes through Alliance of state and Territorial AIDS Directors, 2005). the CarePartners Program,” February 2006, Supported by a Grant from the Maine Health Access Foundation. 91 2004 CARE Act Data Report, Section 2, Items 32, available at http://hab.hrsa.gov/reports/TII2003DB/default.htm#9TOC 85 See Franklin Health Access http://www.fchn.org/healthaccess. 92 The National Health Expenditure Accounts are the standard 86 Jack Hadley and John Holahan, “How Much Medical Care Do benchmark for measuring medical care spending in the the Uninsured Use, and Who Pays for It?” Health Affairs (Web United States. See http://www.cms.hhs.gov/NationalHealth Exclusive) 10.1377/hlthaff.w3.66 February 12, 2003. Holahan, ExpendData/ for an overview and further details. John, Bovbjerg, Randall R., and Hadley, Jack. Caring for the Uninsured in Massachusetts: What Does it Cost, Who Pays, and 93 Long-term medical care services for people who reside in the What Would Full Coverage Add to Medical Spending? Blue Cross community are generally included. and Blue Shield Foundation of Massachusetts Report, November 94 Note 6, op. cit. 2004. http://roadmaptocoverage.org/pdfs/roadmapReport.pdf Zuckerman, Stephen, Bovbjerg, Randall R., Hadley, Jack, 95 Note 2, op. cit. Cravens, Matthew and Clemans-Cope, Lisa, “The Cost of Care for Missouri’s Uninsured,” Missouri Foundation for Health 96 For persons who enter or leave the survey part way through a Report, October 2006. http://mffh.org/CoverMoDataBook2.pdf year, the threshold is applied as a percentage of available months of insurance data—58.3% instead of 7 months, for instance. 87 Jack Hadley and John Holahan, “How Much Medical Care Do the Uninsured Use, and Who Pays for It?” Health Affairs (Web 97 The MEPS-HC threshold decisions are also guided by the Exclusive) 10.1377/hlthaff.w3.66 February 12, 2003. assumptions that (a) CPS understates insurance coverage (overstates uninsurance) relative to other national surveys, so 88 Jack Hadley and John Holahan, “How Much Medical Care Do the MEPS-HC uninsurance definition should be relaxed to match the Uninsured Use, and Who Pays for It?” Health Affairs (Web CPS; (b) MEPS-HC picks up more public coverage than the CPS, Exclusive) 10.1377/hlthaff.w3.66 February 12, 2003. particularly Medicaid, so MEPS-HC public definitions should be tightened; (c) CPS overstates non-group coverage relative to MEPS-HC because the CPS non-group questions come earlier in the survey, and fewer logical edits are performed for this category. Using a consistent threshold of 6, 7, 8, 9, or 10 months, or a December point-in-time measurement does not change the MEPS-HC coverage rate for any insurance type more than 0.5%.

50 costs of caring for uninsured people in maine 98 We follow the re-weighting procedure described by Barsky, et al. (“Accounting for the Black-White Wealth Gap: A Nonparametric Approach,” Journal of the American Statistical Association, September 2002), who applied the methodology developed by Rosenbaum and Rubin (“The Central Role of the Propensity Score in Observational Studies for Causal Effects,” Biometrica, 1983; “Reducing Bias in Observational Studies Using Subclassification on the Propensity Score,” Journal of the American Statistical Association, 1984).

99 Because we pool two years of CPS data, and three years of MEPS data, this requires dividing the sum of the CPS weights by two and the sum of the MEPS weights by three.

100 Hadley and Holahan, 2003; John Holahan, Randall Bovbjerg, and Jack Hadley, 2004. John Holahan, Randall Bovbjerg and Jack Hadley, “Caring for the Uninsured in Massachusetts: What Does It Cost, Who Pays and What Would Full Coverage Add to Medical Spending?” Blue Cross and Blue Shield Foundation of Massachu- setts, November 2004; Jack Hadley and John Holahan, “Who Pays and How Much? The Cost of Caring for the Uninsured,” Kaiser Commission on Medicaid and the Uninsured, February 2003.

101 We assume that the difference between payments and charges for these sources of payment represents contractual allowances, which are conceptually distinct from discounts from charges due to an inability to pay.

102 The expected payment is set to (prv_discnt * totchgs) for uninsured people with charges but no payments.

103 We set donated care to zero if the calculation produces a negative number.

104 Excludes people with any payments from Medicare, workers’ comp, other state and local, other public, and VA.

105 The second model is estimated as a generalized linear model using Newton-Raphson (maximum likelihood) optimization. For specifics see StataCorp LP, “Stata help for glm,” 2005, http://www.stata.com/help.cgi?glm

106 The 17 chronic conditions (8 in the children’s models) are aggregations of the 3-digit ICD-9-CM codes into similar and clinically meaningful categories. The ICD-9-CM codes and accompanying data are stored in each year’s MEPS-HC Medical Conditions file.

costs of caring for uninsured people in maine 51 Strategic solutions for Maine’s health care needs

150 Capitol Street, Suite 4 | Augusta, Maine 04330 www.mehaf.org | 207.620.8266