By Benjamin D. Sommers, Katherine Swartz, and Arnold Epstein

Transcription

1 Web First doi: /hlthaff HEALTH AFFAIRS 30, NO. 11 (2011): Project HOPE The People-to-People Health Foundation, Inc. By Benjamin D. Sommers, Katherine Swartz, and Arnold Epstein Policy Makers Should Prepare For Major Uncertainties In Medicaid Enrollment, Costs, And Needs For Physicians Under Health Reform Benjamin D. Sommers is an assistant professor of health policy and economics at the Harvard School of Public Health and an assistant professor of medicine at Harvard Medical School and Brigham and Women s Hospital, in Boston, Massachusetts. Katherine Swartz is a professor of health economics and policy at the Harvard School of Public Health. Arnold Epstein is the John H. Foster Professor and chair of the Department of Health Policy at the Harvard School of Public Health. ABSTRACT The Affordable Care Act of 2010 will expand Medicaid to millions of Americans by How many enroll will greatly affect health care access, demand for clinicians, and the federal budget, yet the precision and validity of enrollment estimates made to date is unknown. We created a simulation model using two nationally representative data sets to determine the range of reasonable projections, estimating eligibility, participation, and population growth using prior research and our data. Our model predicted that the number of additional people enrolling in Medicaid under health reform may vary by more than 10 million, with a base-case estimate of 13.4 million and a possible range of 8.5 million to 22.4 million. Estimated federal spending for new Medicaid enrollees ranged from $34 billion to $98 billion annually, and we projected that 4,500 12,100 new physicians will be needed to care for new enrollees. In the end, Medicaid enrollment will be determined largely by the extent to which federal and state efforts encourage or discourage eligible people from enrolling. Yet our results indicate that policy makers should prepare to handle a broad range of contingencies and uncertainty in Medicaid expansion under health reform. The Affordable Care Act of 2010 will expand Medicaid eligibility to all US citizens and qualified residents with incomes below 138 percent of the federal poverty level, starting in The Congressional Budget Office (CBO) and the Centers for Medicare and Medicaid Services (CMS) estimate, respectively, that sixteen million and eighteen million new individuals will enroll in Medicaid in ,2 Understandably, both agencies are responsible for providing specific estimates that can be used for budget projections that do not allow for a range of possible values. As a result, current policy discussions have largely ignored the inherent uncertainty in projecting outcomes of the Medicaid expansion. Moreover, neither agency has released complete details on its methods, including some of the key parameters in the models, which makes it difficult to gauge their accuracy and precision. How many people end up enrolling in Medicaid will greatly affect health care access arguably the primary goal of reform. Moreover, whether new enrollees were previously uninsured or privately insured has important implications. Research indicates that extending Medicaid coverage to uninsured people increases their access to care. 3 However, because of low provider reimbursement rates, people who shift from private coverage to Medicaid may experience greater difficulty finding providers willing to treat them. 4 In addition, Medicaid enrollment estimates are critical for budget planning and ensuring that the provider workforce is adequate to care for newly insured people Health Affairs November :11

2 This article presents detailed projections under a variety of plausible economic and demographic assumptions to explore the reasonable range of projected new enrollment in Medicaid and to identify which factors most affect these results. Study Data And Methods We developed a simulation model to estimate how many people will enroll in Medicaid in 2019 as a result of the provisions in the Affordable Care Act. Using nationally representative data, we estimated how many people would be eligible for Medicaid and how many would enroll. We did this with two different surveys to assess the validity of our estimates: the Current Population Survey from 2009 and the Survey of Income and Program Participation from 2008, both conducted by the Census Bureau. The Current Population Survey is the primary source for government estimates regarding employment, annual income, and health insurance. The Survey of Income and Program Participation has a smaller sample but provides data on a monthly basis; it is the data set used by the CBO to estimate changes in insurance coverage. 5 CMS s estimates are based on yet another source, the Medical Expenditure Panel Survey, which we did not analyze. 2 Our samples included all people under age sixty-five, of whom there were 186,518 in the Current Population Survey and 91,849 in the Survey of Income and Program Participation. Medicaid Eligibility Our simulation estimated Medicaid eligibility for each individual under current law and under health reform, based on family income, age, disability, and parental status (pregnancy status is not available in these data sets). 6 8 The income eligibility cutoff under the Affordable Care Act is 138 percent of the federal poverty level (133 percent plus 5 percent that states are required to disregard in calculating eligibility). Combining eligibility with current insurance status produces the following categories: adults and children currently in Medicaid or the Children s Health Insurance Program; uninsured adults currently eligible for Medicaid but not enrolled; uninsured children currently eligible for Medicaid or the Children s Health Insurance Program but not enrolled; uninsured adults who will become eligible for Medicaid because of provisions of the Affordable Care Act; and adults with private coverage who will become eligible for Medicaid. We assumed that people who are already enrolled in Medicaid will remain in the program and that people with private coverage who are already eligible for Medicaid or the Children s Health Insurance Program will not change their enrollment status as a result of the Affordable Care Act. We excluded adults in Medicare from our analysis because the act prohibits them from acquiring Medicaid under the expanded eligibility criteria. Thus, uninsured children and adults currently eligible for Medicaid, but not enrolled, and adults who will become eligible for Medicaid constituted the potential new Medicaid enrollees under health reform in our study. We estimated the survey-weighted population in each category. Then, to compare our estimates with those of the CBO and CMS, we projected our estimates to Finally, we estimated the proportion of each of these groups that will enroll in Medicaid. Simulation Approach For each variable of interest, or parameter, in our simulation, we specified a base case, or best estimate.we also specified lower and upper estimates representing the plausible range based on prior research and our data.we used a Monte Carlo simulation to produce a range of outcomes with associated probabilities. Using this method, values for each numeric parameter were randomly drawn from a normal distribution with our base case defined as our mean, and high and low values set equal to two standard deviations from the mean. The parameters were independent of one another, except for participation rates among uninsured adults and children, which we calculated from the same random drawing. For the categorical parameters of data source and method of calculating family income, we used the basecase, low, and high estimates each for one-third of our simulations. We conducted 1,000 simulations and calculated a 95 percent confidence interval for each outcome.we identified the parameters most responsible for variation in our projections by changing them one at a time from the base-case to the upper or lower estimate and measuring the resulting enrollment change. The parameters of our study are shown in Exhibit 1. We discuss in detail in the online Appendix 9 some of the issues surrounding these parameters and the analytic choices we made in selecting the base-case, high, and low estimates. Outcomes Our primary outcome was the number of new enrollees in Medicaid in 2019 as a result of health reform. Secondary outcomes were the number of new physicians needed to care for previously uninsured Medicaid enrollees; the number of existing physicians experiencing a change in payer mix from private coverage to Medicaid; and federal costs of the expansion (under the Affordable Care Act, the federal government pays for more than 90 percent of it). 2 We calculated workforce outcomes using November :11 Health Affairs 2187

3 Web First Exhibit 1 Simulation Parameters For Projections Of New Medicaid Enrollment, 2019 Parameter Low estimate Base-case estimate High estimate Parameters affecting how many people are eligible Data set CPS CPS SIPP Classification of income for determining Medicaid eligibility a Survey s definition of family income Health insurance unit $2,047 $2,456 Proportion of noncitizens eligible for 15% 30% 45% Medicaid b 10-year growth rate ( ) in low-income uninsured adults c,d 5% 20% 30% Parameters affecting how many people enroll Percent of newly eligible adults who switch 15% 30% 60% from private insurance to Medicaid ( crowdout ) Percent of previously eligible uninsured adults 20% 33% 50% who enroll in Medicaid f Percent of previously eligible g uninsured 20% 33% 50% children who enroll in Medicaid c Percent of newly eligible uninsured adults who enroll in Medicaid f 45% 62% 80% Parameters affecting federal Medicaid spending Annual federal Medicaid costs per newly $1,638 enrolled child h Annual federal Medicaid costs per newly enrolled adult h $3,684 $4,606 $5,527 Health insurance unit, continuous eligibility for children, income disregards SOURCES See exhibit notes. Full bibliographic information is available in the online Appendix (see Note 9 in text). NOTES CPS is Current Population Survey. SIPP is Survey of Income and Program Participation. a (1) Kenney et al. (2009); (2) Kaiser Commission on Medicaid and the Uninsured (2011). See the online Appendix for full explanation of these parameters. b Sommers (2010). c Authors analysis of data in our sample. d (1) Census Bureau (2009); (2) Census Bureau (2010). e (1) Blumberg et al. (2000); (2) Cutler and Gruber (1996); (3) Thorpe and Florence (1998); (4) Gruber and Simon (2008). f Sommers and Epstein (2010). g Eligible for Medicaid or Children s Health Insurance Program. h (1) Note 19 in text; (2) Buettgens et al. (2010); (3) Holahan et al. (2010); (4) Congressional Budget Office (2010). multivariate-adjusted national averages for outpatient visits by people who had no insurance, private insurance, or Medicaid, as well as number of visits per US physician. Prior research suggests that uninsured adults acquiring coverage will increase their use of services. To a lesser extent, people shifting from private coverage to Medicaid will also increase their use of services, because of lower cost sharing in Medicaid. Our annual estimates were 2.7 outpatient visits for uninsured people, 3.5 visits for people with private insurance, and 3.8 visits for Medicaid beneficiaries. 10 The average number of visits per US physician, across all specialties, was 1, ,12 Although one recent analysis of health reform considered only primary care physician needs (assuming roughly 3,500 annual visits per primary care physician), 13 we followed the approach of prior work in considering the need for all physician services and specialties, which resulted in a lower estimated number of visits per physician. 14 Survey data analyses were conducted using the statistical analysis software Stata, version We used Microsoft Excel for the Monte Carlo simulation. Limitations Extrapolating from past data may be unreliable in the context of dramatic changes that are coming in the health care system. We assumed that the parameters followed a normal distribution, with the high and low estimates plus or minus two standard deviations from the mean. However, this approach may be conservative. Our workforce projections assumed that previous utilization patterns based on insurance coverage and current physician productivity will remain stable under the Affordable Care Act. Our estimates of Medicaid eligibility were subject to some imprecision because of differences in how the Census Bureau s surveys measure income and disability, compared to Medicaid, as well as a lack of data on financial assets. These issues affect any survey-based simulation approaches, 2188 Health Affairs November :11

4 such as those used by the CBO and CMS. Finally, our projections do not include nonphysician providers such as nurse practitioners and physician assistants. They will, of course, also play an important role in caring for new Medicaid enrollees; in fact, they may take on more patient care overall in the context of a reformed health care system. Our projections do not take into account the possibility that the work of seeing and treating patients may be allocated differently across new teams of health care providers, particularly those in which allied health professionals play a greater role. Study Results Calibration Check We compared our model s estimates to the projections provided by the CBO and CMS. Using our base-case assumptions and data from the Survey of Income and Program Participation, to follow the CBO s approach, we projected that 17.3 million people would be newly enrolled in Medicaid in Using the same data and assuming that 55 percent of uninsured adults newly eligible for Medicaid would enroll in the program (instead of our baseline rate of 62 percent), we matched the CBO s estimate of 16.0 million enrollees. However, we cannot verify that 55 percent is indeed the CBO s assumed take-up rate, because that parameter has not been released. Using our base-case assumptions and Current Population Survey data, but substituting the participation rate assumed by CMS for uninsured adults newly eligible for Medicaid (97 percent), 15 we obtained a total of 17.9 million, essentially matching CMS s projection of 18.0 million. These calibration checks demonstrate that our overall simulation approach is consistent with the CBO and CMS models. However, our goal was not to replicate their results, but instead to offer an independent estimate using our preferred parameters. These results are presented in the next section. Projections And Confidence Intervals Our base-case estimate of new Medicaid enrollment was 13.4 million people (95% confidence interval: 8.5 million, 22.4 million). Exhibit 2 presents eligibility and enrollment projections by group, and Exhibit 3 presents the enrollment estimates in graphical form. In each scenario, roughly half of the enrollees were newly eligible uninsured adults. Exhibit 4 presents clinician workforce and cost estimates. Assuming that current patterns of utilization and physician productivity persist, we estimated that 7,400 additional physicians would be needed to care for people who are newly enrolled in Medicaid, compared to the current physician supply (95% confidence interval: 4,500, 12,100). We also estimated that 5,300 full-time-equivalent physicians would shift from caring for privately insured to Medicaid-covered patients. This latter effect does not represent a shortfall in physicians but simply a substitution in payment sources for existing physicians. And we projected that the federal government would spend an additional $58 billion on Medicaid annually (95% confidence interval: $34 billion, $98 billion). The largest determinants of variation were the choice of data sets (depending on which of the two we used, the estimated enrollment varied by 3.9 million people), classification of family income (depending on whether we used the income of the survey family or that of the health Exhibit 2 Alternative Projections For New Medicaid Enrollment Under Health Reform, 2019 Population group Base case Number eligible (millions) Number enrolled (millions) 95% confidence interval Number eligible (millions) Number enrolled (millions) Newly eligible adults, switched from private insurance , , 6.1 Newly eligible uninsured adults , , 11.3 Previously eligible uninsured adults , , 3.3 Previously eligible uninsured children a , , 3.4 Total new Medicaid enrollment b 13.4 b 8.5, 22.4 SOURCE Authors analysis. NOTES Numbers might not sum to totals because of rounding. Subgroup enrollment figures in the 95 percent confidence intervals do not sum to total enrollment because each 95 percent confidence interval was calculated independently. a Eligible for Medicaid or Children s Health Insurance Program. b No totals are provided because the exhibit does not include all groups, and thus the sum of the rows would not accurately indicate the total number of individuals eligible for Medicaid. See Study Data and Methods section in text for more detail. November :11 Health Affairs 2189

5 Web First Exhibit 3 Alternative Projections For New Medicaid Enrollment Under Health Reform, By Eligibility And Insurance Status, 2019 Millions of new Medicaid enrollees Base case Newly eligible adults, switched from private insurance Previously eligible uninsured children Previously eligible uninsured adults Newly eligible uninsured adults SOURCE Authors analysis. NOTES CI is confidence interval. The figure maintains the proportions for each subgroup from the 95 percent CI in Exhibit 2, scaled to the total enrollment for the upper and lower bounds. Exhibit 4 insurance unit, enrollment varied by 3.1 million), and the participation rate among newly eligible uninsured adults (depending on whether we used the baseline take-up rate of 62 percent or the high case rate of 80 percent, enrollment varied by 2.3 million). Discussion Using a range of plausible assumptions, our model predicted that the number of additional people enrolling in Medicaid under health reform may vary by more than 10 million, with a base-case estimate of 13.4 million and a 95 percent confidence interval of 8.5 million to 22.4 million. Government estimates were higher than our base case. The CBO estimated 16 million new Workforce And Cost Estimates For New Medicaid Enrollment, 2019 Estimate Base case 95% confidence interval Number of new physicians needed 7,400 4,500, 12,100 Number of physicians switching from privately insured to Medicaid-covered patients 5,300 2,200, 12,500 Additional annual federal spending on Medicaid $58 billion $34 billion, $98 billion SOURCE Authors analysis. enrollees, but the difference from our base case is partially attributable to the CBO s use of data from the Survey of Income and Program Participation. When we used our base-case parameters and that survey instead of the Current Population Survey, we estimated 17.3 million enrollees, which is much closer to the CBO s figure than our base case. However, we used the Survey of Income and Program Participation s monthly income to more closely parallel state eligibility rules, 16 while the CBO used annual income data from that survey, 5 so the comparison is not perfect. It is likely that our model differs in other ways from that the CBO s, but not all of the details of that agency s simulation have been made public. Furthermore, the CBO assumed that enrollment in the Children s Health Insurance Program would fall after 2015 because the program although reauthorized through 2019 has funds allocated to it only through In contrast, we assumed that funding for the program and enrollment would remain stable. However, this has little effect on estimates of the net impact of the Affordable Care Act because both the pre- and postreform projections from the CBO made the same assumption of decreased funding for the Children s Health Insurance Program by ,18 CMS projected eighteen million new Medicaid enrollees, more than our estimate because of CMS s assumption of extremely high Medicaid participation (97 percent) among currently uninsured adults. 15 This rate is far higher than any Medicaid take-up rates reported in the research literature or state-specific analyses. Our analysis revealed much uncertainty in both the number of people who will become eligible for Medicaid and how many of them will enroll in the program. Putting our results in other terms, we found only a 25 percent chance that the number of new enrollees will be between fifteen and nineteen million the range that appears most likely based on government projections. The sources of uncertainty can be divided into two categories: methodological factors and policy factors. The first category includes which data set one uses and how one estimates eligibility from survey data. In particular, the effect of the data set on projections may be underappreciated by policy makers because different government agencies employ different data sets for similar analyses. 2,5 Policy factors are the second source of uncertainty especially the Medicaid participation rate among newly eligible people who are uninsured. If states can encourage greater participation, then our higher estimates may be attain Health Affairs November :11

6 able. However, if health reform implementation funding is reduced a real possibility, given ongoing fiscal pressures enrollment efforts will be compromised. Our results indicate that 4,500 12,100 additional physicians may be needed to care for new Medicaid patients (Exhibit 4), compared to the current physician supply. Many current physicians will also experience a shift from private to public insurance as a source of payment, although this represents less than 1 percent of total physician visits nationally. 12 These changes may pose major challenges in health care access because in recent years an increasing share of physicians have stopped accepting Medicaid patients. 4 The Affordable Care Act does provide enhanced Medicaid reimbursement to primary care clinicians for , 2 but this may not be enough to ensure an adequate supply of providers for new Medicaid patients. We estimated that additional federal spending on Medicaid could vary from $34 billion to $98 billion in 2019 (Exhibit 4). Although health reform will dramatically increase the number of people covered by Medicaid, the bulk of Medicaid costs will remain primarily attributable to the care of the disabled and elderly people already in the program. Even our high-end estimate of $98 billion is less than a quarter of the $477 billion that the CBO expects the federal government to spend on Medicaid in To our knowledge, our analysis is the first to use repeated simulations to estimate new Medicaid enrollment under health reform and to indicate the precision of these estimates with a range of likely values. Our results were largely consistent with an Urban Institute analysis that considered two scenarios: a new Medicaid enrollment of 15.9 million adults under standard outreach, and a new enrollment of 22.8 million under enhanced outreach. 20 By considering a greater number of parameters than prior work, we found an even wider range of possible outcomes. Furthermore, unlike existing projections, we used a probabilistic simulation model that allowed for significant ranges in all of the key parameters underlying the projections, as opposed to single-point estimates. 1,2,20 We believe that this approach offers a more realistic assessment of possible outcomes under health reform. In spite of the limitations in our approach outlined above, we believe that our approach was conservative. This suggests that there may be an even wider range in eventual outcomes than our findings indicate. Conclusion Our results indicate that policy makers should prepare to handle a broad range of contingencies in the Medicaid expansion under health reform. Some of this uncertainty in enrollment projections is methodological, but some is amenable to public policy. Medicaid participation will be determined largely by the extent to which federal and state efforts encourage or discourage eligible people from enrolling. The results of these efforts will have major implications for health care access, Medicaid budgeting, and provider workforce needs for decades to come. This work was supported in part by funding from the Agency for Healthcare Research and Quality (Grant No. R03 HS to Benjamin D. Sommers as principal investigator). The authors are grateful to John Shatto, Office of the Actuary at the Centers for Medicare and Medicaid Services, for his insights into Medicaid enrollment projections; to David Auerbach, RAND, and two anonymous reviewers for helpful feedback on prior drafts of the manuscript; and to Meredith Roberts Tomasi, Harvard School of Public Health, for research assistance. This work was completed while Sommers was working at the Harvard School of Public Health. [Published online October 26, 2011.] November :11 Health Affairs 2191

7 Web First NOTES 1 Elmendorf DW. Letter from Douglas W. Elmendorf, director, Congressional Budget Office, to US House of Representatives Speaker Honorable Nancy Pelosi regarding HR 4872, the Reconciliation Act of 2010 [Internet]. Washington (DC): Congressional Budget Office; 2010 Mar 20 [cited 2011 Oct 6]. Available from: 113xx/doc11379/AmendReconProp.pdf 2 Foster RS. Estimated financial effects of the Patient Protection and Affordable Care Act, as amended [Internet]. Baltimore (MD): Centers for Medicare and Medicaid Services; 2010 Apr 22 [cited 2011 Oct 6]. Available from: PPACA_ pdf 3 Finkelstein A, Taubman S, Wright B, Bernstein M, Gruber J, Newhouse JP, et al. The Oregon health insurance experiment: evidence from the first year. Cambridge (MA): National Bureau of Economic Research; 2011 Jul. (NBER Working Paper No ). 4 Cunningham P, May J. Medicaid patients increasingly concentrated among physicians. Washington (DC): Center for Studying Health System Change; Congressional Budget Office. CBO s health insurance simulation model: a technical description. Washington (DC): CBO; 2007 Oct. 6 Kaiser Commission on Medicaid and the Uninsured. Where are states today? Medicaid and CHIP eligibility levels for children and non-disabled adults [Internet]. Washington (DC): The Commission; 2011 Feb [cited 2011 Oct 6]. (Fact Sheet). Available from: upload/ pdf 7 Kaiser Commission on Medicaid and the Uninsured. Medicaid financial eligibility: primary pathways for the elderly and people with disabilities [Internet]. Washington (DC): The Commission; 2010 Feb [cited 2011 Oct 6]. Available from: pdf 8 Ross DC, Marks C. Challenges of providing health coverage for children and parents in a recession: a 50 state update on eligibility rules, enrollment and renewal procedures, and cost-sharing practices in Medicaid and SCHIP in 2009 [Internet]. Washington (DC): Kaiser Commission on Medicaid and the Uninsured; 2009 Jan [cited 2011 Oct 6]. Available from: pdf 9 To access the Appendix, click on the Appendix link in the box to the right of the article online. 10 Hadley J, Holahan J. Covering the uninsured: how much would it cost? Health Aff (Millwood). 2003; 24:w DOI: /hlthaff.w Association of American Medical Colleges. Physician shortages to worsen without increases in residency training. Washington (DC): AAMC; National Center for Health Statistics. Ambulatory medical care utilization estimates for 2007 [Internet]. Hyattsville (MD): NCHS; 2011 Apr [cited 2011 Oct 6]. (Vital and Health Statistics, Series 13, No. 169). Available from: nchs/data/series/sr_13/sr13_169.pdf 13 Hofer AN, Abraham JM, Moscovice I. Expansion of coverage under the Patient Protection and Affordable Care Act and primary care utilization. Milbank Q. 2011; 89(1): Anderson GF, Hurst J, Hussey PS, Jee-Hughes M. Health spending and outcomes: trends in OECD countries, Health Aff (Millwood). 2000;19(3): Centers for Medicare and Medicaid Services, Office of the Actuary actuarial report on the financial outlook for Medicaid. Baltimore (MD): CMS; Marron DB. Statement of Donald B. Marron, acting director: Medicaid spending growth and options for controlling costs. Washington (DC): Congressional Budget Office; 2006 Jul Congressional Budget Office. Spending and enrollment detail for CBO s August 2010 baseline: Children s Health Insurance Program (CHIP) [Internet]. Washington (DC): CBO; 2010 [cited 2011 Oct 18]. Available from: CHIPAugust2010FactSheet.pdf 18 Congressional Budget Office. Spending and enrollment detail for CBO s March 2010 baseline: Children s Health Insurance Program (CHIP) [Internet]. Washington (DC): CBO; 2010 [cited 2011 Oct 18]. Available from: 19 Congressional Budget Office. Spending and enrollment detail for CBO s August 2010 baseline: Medicaid [Internet]. Washington (DC): CBO; 2010 [cited 2011 Oct 18]. Available from: MedicaidAugust2010FactSheet.pdf 20 Holahan J, Headen I (Urban Institute, Washington, DC). Medicaid coverage and spending in health reform: national and state-by-state results for adults at or below 133 percent FPL. Washington (DC): Kaiser Commission on Medicaid and the Uninsured; 2010 May Health Affairs November :11

8 ABOUT THE AUTHORS: BENJAMIN D. SOMMERS, KATHERINE SWARTZ & ARNOLD EPSTEIN Benjamin D. Sommers is an assistant professor of health policy and economics at the Harvard School of Public Health. In this month s Health Affairs, Benjamin Sommers and coauthors describe how they used a simulation model to estimate the number of new enrollees who will be eligible for Medicaid in 2014 under the Affordable Care Act. They also estimated federal spending for new Medicaid enrollees and the number of new physicians who will be needed to care for them, and conclude that thereisawiderangeofuncertainty in all these estimates. For example, their base-case estimate of the number of additional Medicaid enrollees is 13.4 million, but a reasonable range could be between 8.5 million and 22.4 million. Sommers says he and his colleagues were all surprised about how much uncertainty there actually is. Sommers, an assistant professor of health policy and economics at the Harvard School of Public Health and an assistant professor of medicine at Harvard Medical SchoolandBrighamandWomen s Hospital, is currently on leave from Harvard while he serves as a senior adviser in health policy in the Office of the Assistant Secretary for Planning and Evaluation at the Department of Health and Human Services. His health policy research has focused on public health insurance, health care financing, and medical decision making. He is also a practicing primary care physician. He received a doctorate in health policy with a concentrationinhealtheconomics from Harvard University and a medical degree from Harvard Medical School. Katherine Swartz is a professor of health economics and policy at the Harvard School of Public Health. Katherine Swartz is a professor of health economics and policy at the Harvard School of Public Health. Her research interests for the past thirty years have focused on people without health insurance, efforts to increase access to health care coverage, how to pay for expanded health insurance coverage, and, most recently, on policy issues related to the elderly. Swartz is a member of the Institute of Medicine and the National Academy of Social Insurance. She received her doctorate in economics from the University of Wisconsin. Arnold Epstein is thejohnh.foster Professor and chair of the Department of Health Policy at the Harvard School of Public Health. Arnold Epstein is the John H. Foster Professor and chair of the Department of Health Policy and Management at the Harvard School of Public Health and a member of the Division of General Medicine at Brigham and Women's Hospital. He is a practicing physician and health services researcher focusing on access and quality of care, especially for disadvantaged populations. He received his medical degree from Duke University and his master s degree in government from Harvard University. November :11 Health Affairs 2193