Mental Health By Richard G. Frank and Arnold M. Epstein 10.1377/hlthaff.2013.0769 HEALTH AFFAIRS 33, NO. 6 (2014): 1006–1013 ©2014 Project HOPE— The People-to-People Health Foundation, Inc.
doi:
Richard G. Frank (frank@ hcp.med.harvard.edu) is the Margaret T. Morris Professor of Health Economics in the Department of Health Care Policy, Harvard Medical School, in Boston, Massachusetts. Arnold M. Epstein is a professor of health policy and management, Harvard School of Public Health, in Boston.
Factors Associated With High Levels Of Spending For Younger Dually Eligible Beneficiaries With Mental Disorders ABSTRACT The Affordable Care Act focused attention on how conflicting rules and payment arrangements in Medicare and Medicaid can produce high costs and fragmented care for people who are dually eligible for the two programs. Nearly half of such dual eligibles have severe and persistent mental disorders. Using Medicare data for the period 2006–09, we examined factors that were associated with high levels of spending for dual eligibles younger than sixty-five with a mental disorder. We found that these beneficiaries were nearly twice (1.86 times) as expensive as young dual eligibles who did not have a mental disorder. We identified functional limitations, multiple chronic conditions, and substance use disorders as being associated with high levels of spending in this subpopulation. We conclude that case management that coordinated medical, mental health, and substance use treatment along with psychosocial rehabilitation services could yield savings, primarily to the Medicare program. Because only Medicaid pays for case management and psychosocial rehabilitation services, Medicaid spending may need to rise if overall savings are to be realized.
P
eople who are eligible for both Medicare and Medicaid, commonly called “dual eligibles,” are among the most vulnerable members of the US population. Medicare beneficiaries qualify for full or partial Medicaid coverage if their incomes and assets fall below certain thresholds or if they have high health care bills. Each state sets its own income thresholds and scopes of benefits for the dual eligibles, within federal guidelines. The nine million people who were dual eligibles in 2008 accounted for 39 percent of Medicaid’s spending and 29 percent of Medicare’s in that year.1 Dual eligibles often receive substandard care.1 Roughly one-third of them are under age sixty-five and are sometimes called “young duals”: They receive Medicare benefits because of a disability and account for 37 percent of spending for all dual eligibles.2 People with se1006
Health Affairs
June 2014
33:6
vere and persistent mental illnesses account for nearly half of young dual eligibles.2 Young dual eligibles with severe mental illnesses typically have complex needs.2 To remain in the community, they require medical care, specialty psychiatric care, prescription drugs, long-term services and supports (such as rehabilitation services and case management), and human services (for example, services related to housing and employment). It has long been realized that “the organization of services for adults with severe mental disorders is the linchpin of effective treatment.”3 Thus, the integration and coordination of care to meet their complex needs is important. About 28 percent of young dual eligibles have depression, and about 12 percent have schizophrenia or bipolar disorder.Young dual eligibles with mental disorders tend to have higher chronic disease burdens and to account for high levels
of spending on emergency department (ED) services and inpatient care, compared to young dual eligibles who have other disabling conditions.2 Dual eligibles with schizophrenia have one of the highest rates of hospital readmissions of any group of people with a particular disease. Improved care coordination has been shown to improve care and reduce spending on ED visits and hospitalizations.3–6 For the dual-eligible population, Medicare pays for traditional medical services such as physician visits, hospital stays, and prescription drugs. For people with disabling mental disorders, Medicaid pays for long-term services and supports such as case management, residential or nursing home care, and psychosocial rehabilitation services. As part of its efforts to reform the payment and care delivery systems, the Affordable Care Act tried to align rules and payment arrangements between Medicare and Medicaid. Current conflicts between the two programs’ regulations can result in incentives that produce high costs and fragmented care for dual eligibles. For example, Medicaid bed-hold policies—that is, paying nursing homes a fee to hold a bed open for a patient who has been hospitalized—in conjunction with Medicare’s high payment rates for postacute stays at skilled nursing facilities create incentives for the rehospitalization of nursing home patients. To address this problem, in 2011 the federal government, under authority provided by the Affordable Care Act, established a demonstration program that allows states to align the two programs’ incentives, improve care coordination, and share in any savings that accrue.1 As of April 2014 seventeen states had launched demonstrations in this program or were planning to launch them by the end of the year.7 Ten of those states have signed memoranda of understanding with the Centers for Medicare and Medicaid Services (CMS). Two states initiated their programs ahead of others: Washington State began its demonstration for dual eligibles in October 2012 and expanded it in May 2013, and Massachusetts launched its demonstration in October 2013.7 The demonstrations can include arranging housing, medical care, behavioral health services, long-term care, and other support services for dual eligibles who have functional limitations. Seven of the demonstrations (those in Illinois, Massachusetts, Michigan, New York, Ohio, South Carolina, and Washington) emphasize care for people with severe mental illnesses. The success of these interventions may depend on directing efforts toward those segments of the population that are the most costly to the Medi-
care and Medicaid programs and whose quality of care is most in need of improvement. We analyzed patterns of spending among dual eligibles younger than sixty-five who suffer from mental disorders. It has been known for some time that people with severe mental disorders are overrepresented in the Medicare population in this age group. However, little attention has been devoted to dual eligibles with these conditions.8 We identified factors that were associated with higher spending levels and examined the distribution of spending (and, as a result, potential savings) between Medicare and Medicaid.
Study Data And Methods Data We used data from the Medicare Current Beneficiary Survey (MCBS) for the period 2006– 09. By using those four years and oversampling dually eligible beneficiaries, we were able to obtain sufficient numbers of cases to conduct our analysis. The MCBS is a representative survey of the Medicare population. It contains a rich array of information, including beneficiaries’ demographic characteristics, health care use, living arrangements, functional impairments, health status, and program participation. It also includes Medicare claims data. Each year the survey produces a sample of about 14,500 beneficiaries. Response rates for surveys in the study period ranged from 85 percent to 88 percent. Survey respondents are typically kept in the survey for three years. Thus, in each year the sample includes both new and continuing respondents. The MCBS oversamples people ages eighty-five and older and those younger than age sixty-five. About 93 percent of the sample resides in the community; the remaining 7 percent live in long-term care facilities. CMS sponsors the survey and provides public-use files on cost and use of, and access to, care. Our analyses relied primarily on the cost and use files. The data in these files include linkages between survey responses and Medicare claims, as well as use of and spending on all health care not covered by Medicare. A beneficiary is reported as having a mental disorder if he or she had a claim with a primary diagnosis that called for services associated with care of a mental illness (International Classification of Diseases, Ninth Revision, codes 290–319), use of a psychotherapeutic drug (MCBS variable THERCC or therapeutic class ¼ 80), or treatment under a specialty mental health procedure code during the survey year. Our sample of dual eligibles younger than sixty-five ranged from 1,163 to 1,306 per year. Between 788 and 883 of J u n e 201 4
3 3: 6
Health Affairs
1007
Mental Health them each year were reported as having a mental disorder. We studied individuals in some analyses (such as our description of average costs) and total person-years in other analyses (for example, our regression models). For analyses of individuals, we chose the second year if the beneficiary was surveyed twice and the middle year if he or she was surveyed three times. This approach resulted in 4,999 person-year observations. Twenty-eight percent of dual eligibles reported to have mental disorders had depression, 18 percent had bipolar disorder, and 21 percent had schizophrenia. The remainder had other classes of diagnoses, including anxiety, substance use disorders, and obsessive-compulsive disorder. These figures accord well with data in the Medicare Payment Advisory Commission’s (MedPAC’s) 2010 report to Congress.2 Analysis We examined four patterns of spending for dual eligibles. First, we assessed total spending on those with mental disorders compared to spending on those with other forms of disability. Second, we described spending on young dual eligibles by type of service (inpatient care, outpatient care, drugs, or other services) and source of payment (Medicare or Medicaid). Third, we examined total spending by demographic characteristics, health, and functional status of beneficiaries and by the source of payment. And fourth, we examined the shares of spending that different population segments accounted for, starting with the group whose costs were the highest. All spending estimates are in 2009 dollars, adjusted by the Consumer Price Index for All Urban Consumers.9 We used multivariate logistical regression models to examine the explanatory power of key demographic, health, and function characteristics on high levels of spending overall and by Medicare and Medicaid separately. This approach enabled us to identify high spenders
Exhibit 1 Average Annual Spending For People Dually Eligible For Medicare And Medicaid Who Were Younger Than 65 And Had A Mental Disorder, By Type Of Spending, 2006–09 Type of spending
Patients
Total spending ($)
Spending by Medicare (%)
Spending by Medicaid (%)
Inpatient Outpatient Drug Othera Total
827 2,705 2,875 3,278 3,318
4,862 2,344 6,246 10,705 24,155
83 75 87 31 60
6 12 5 44 23
SOURCE Authors’ analysis of data from the Medicare Current Beneficiary Survey, 2006–09. NOTES Spending was adjusted to 2009 dollars using the Consumer Price Index. N ¼ 3; 318. aResidential, case management, and partial hospital spending.
1008
H e a lt h A f fai r s
J u n e 201 4
3 3: 6
and potential targets of care coordination efforts. To obtain the appropriate standard errors for our coefficient estimates, our estimation approach took account of the repeated measures with non-Gaussian data.We accomplished this by using the generalized estimating equations estimator and taking account of the complex sample design.10 Limitations There are several limitations to this study. Our survey data were based on interviews with respondents living in the community and thus are self-reported. The information on health status and functioning was not clinically validated. The Medicare spending data were based on claims, while Medicaid and other spending data were based on self-report and CMS’s imputations of costs. The sample reflected primarily fee-for-service spending, largely because relatively few dual eligibles had enrolled in managed care arrangements. Because of the limited number of people younger than sixty-five in the MCBS, our analyses had somewhat limited power. This was especially true when we stratified them according to dual eligibles’ different levels of function and comorbidity. Our national sample did not allow us to describe regional variation. It is tempting to conclude that integrated care programs would be most cost-effective if they were targeted at patients associated with higher costs. However, we cannot be sure of that without an intervention study. Our analyses of spending in relation to mental health diagnosis may be overestimates if high-cost users with mental health conditions were more likely than people with other conditions to be diagnosed in our data.
Study Results Dual eligibles younger than sixty-five incurred average annual costs for all health and long-term services and supports of $20,383, based on the data for the period 2006–09 and expressed in 2009 dollars. For those identified as having a mental disorder, average annual spending was estimated to be $24,155 (Exhibit 1), in contrast to $12,948 for young dual eligibles with other disabilities (a ratio of 1.86). Thus, spending on young dual eligibles with a mental disorder was slightly less than twice as much as spending on young dual eligibles with other types of disabilities. Spending By Type Of Service And Payer Sixty percent of all spending for young dual eligibles with a mental disorder was paid for by Medicare, and 23 percent was paid for by Medicaid (Exhibit 1). Roughly 25 percent was spent on
drugs, and nearly 40 percent was spent on other services—including institutional providers such as nursing homes, residential treatment, and nonmedical services such as case management. Other services were more often paid for by Medicaid than by Medicare. The remainder went to inpatient and outpatient care that was largely paid for by Medicare. In a given year, 25 percent of dual eligibles with a mental disorder were hospitalized, and 11.5 percent were hospitalized twice or more. In contrast, 14 percent of dual eligibles without a mental disorder were hospitalized, and 4.9 percent had two or more inpatient stays. Exhibit 2 plots total health care spending during the study period for all young dual eligibles with mental disorders according to source of payment: Medicare or Medicaid. The exhibit highlights three important phenomena. First, there was considerable variation in spending levels in both programs for young dual eligibles. Second, despite that variation, the proportion of relatively low spenders was much greater in Medicaid than in Medicare. This is illustrated in Exhibit 2 by the density of points along the vertical axis clustered near the origin of the horizontal axis. As a result, median spending was about $667 per year in Medicaid, versus $6,665 in Medicare. The respective numbers for average annual spending were $5,460 for Medicaid and $14,178 for Medicare. Third, there were few cases of high levels of spending by both Medicare and Medicaid. In other words, young dual eligibles who needed high-intensity services appeared to need chiefly either high-cost medical care (such as inpatient and ED care) that was primarily paid for by Medicare or services that were mostly paid for by Medicaid—expensive institutional placements (for example, in a nursing home) or intensive community-based support services. Spending By Diagnosis, Functional Status, And Substance Abuse Average annual spending in 2009 dollars for young dual eligibles without limitations in activities (or instrumental activities) of daily living ranged from a low of $10,021 for those without a mental disorder to $23,471 for those with substance use disorders (Exhibit 3). For young dual eligibles with limitations, the range in spending was also large. The results underscore that it is combinations of mental disorders with a functional impairment or comorbid substance use disorders that generate high levels of spending. For a beneficiary with depression, adding either a functional impairment or a substance use disorder approximately doubled average annual spending (Exhibit 3). Interestingly, for a beneficiary with a
diagnosis not related to mental health, the presence of a substance use disorder resulted in a 2.5fold increase in average annual spending. Predictors Of High Spending By Dual Eligibles Disease burdens appeared greater for young dual eligibles with a mental disorder than for those without such disorders. However, even the subpopulation with a mental disorder had variable spending levels (Exhibit 4). The top 10 percent of spenders accounted for 43 percent of spending overall. The highest-spending half of the population accounted for nearly 90 percent of all spending. Of course, the corollary is also true: The lowest-spending half of the population accounted for about 11 percent of spending. We used a logistic regression model to determine the likelihood that a young dual eligible would be in the top 10 percent of either Medicare or Medicaid spending. The model results revealed both similarities and some notable differences between the two programs in the factors that explained high levels of spending (Exhibit 5). For example, having functional limitations in activities of daily living (see Exhibit 3 notes) or
Exhibit 2 The Relationship Between Medicare And Medicaid Spending For People Eligible For Both Programs Who Were Younger Than 65 And Had A Mental Disorder, 2006–09
SOURCE Authors’ analysis of data from the Medicare Current Beneficiary Survey. NOTE Spending is total health care spending during the study period for all young dual eligibles with mental disorders.
June 2014
33:6
H ea lt h A f fai r s
1 009
Mental Health Exhibit 3 Average Annual Spending For People Dually Eligible For Medicare And Medicaid Who Were Younger Than 65, By Diagnosis, 2006–09
Diagnosis Depression Substance use disorder Bipolar disorder Schizophrenia Other, mental healthb Other, not mental health
No ADL or IADL limitations
ADL limitations
IADL limitations
Substance use disorder
No. 337 156 228 192 505 604
No. 449 158 173 133 772 529
No. 711 273 323 348 1,289 1,009
No. 225 —a 138 98 318 45
Spending ($) 20,078 23,471 19,150 21,079 19,222 10,021
Spending ($) 41,829 40,243 39,364 41,129 41,276 19,922
Spending ($) 33,675 32,628 29,882 27,757 33,154 14,694
Spending ($) 38,083 —a 33,120 36,059 33,456 26,846
SOURCE Authors’ analysis of data from the Medicare Current Beneficiary Survey, 2006–09. NOTES Spending was adjusted to 2009 dollars using the Consumer Price Index. ADL is activities of daily living (bathing, dressing, eating, using the toilet, and transferring into or out of a bed or chair). IADL is instrumental activities of daily living (managing money, meal preparation, shopping, doing housework, using the telephone, and taking medications). “Limitations” is functional limitations. aNot applicable. b Includes anxiety disorders and personality disorders.
multiple chronic conditions increased the likelihood that a beneficiary would have a high level of spending in both Medicare and Medicaid. Yet having multiple chronic conditions had a more powerful impact on the probability of incurring high costs in Medicare than in Medicaid (odds
Exhibit 4 Combined Medicare And Medicaid Spending For People Eligible For Both Programs Who Were Younger Than 65 And Had A Mental Health Disorder, By Level Of Spending, 2006–09 100
80
ratio for Medicare: 16.70; for Medicaid: 3.44). In contrast, having functional limitations had a similar impact on the odds of being in the top 10 percent of spending in Medicare (odds ratio: 2.10) and Medicaid (odds ratio: 1.82). The differences in explanatory factors for being a high-cost beneficiary were linked to the different services that each program funds. For example, limitations in instrumental activities of daily living (see Exhibit 3 notes) and a diagnosis of schizophrenia predicted high levels of Medicaid spending but not Medicare spending. People with these conditions frequently need long-term services and supports that are covered by Medicaid—including case management, residential services, and personal assistance—in addition to medical care.
Percent of population
Discussion 60
40
20
0 0
20
40
60 Percent of spending
80
100
SOURCE Authors’ analysis of data from the Medicare Current Beneficiary Survey. NOTE “Spending” is total spending from all payment sources during the study period.
1010
Health Affairs
J u n e 201 4
3 3: 6
It has been well established that beneficiaries who are eligible for both Medicare and Medicaid represent a high-cost group whose care could be improved.11 Sections 2601 and 2602 of the Affordable Care Act focus on measures that aim to improve both the alignment between Medicare and Medicaid in serving dually eligible beneficiaries and the coordination of clinical services for a population with complex needs. The National Commission on Fiscal Responsibility and Reform recommended placing dual eligibles in Medicaid managed care plans, reflecting the view that health plans that managed all of the care for a population with complex needs within a budget would likely provide better and more efficient care than that provided under unmanaged fee-for-service arrangements. The commission estimated that if its recommendation were followed, $13 billion would be saved in the period 2015–20.12
Exhibit 5 Impact Of Demographic, Clinical, And Functional Characteristics Of People Dually Eligible For Medicare And Medicaid Who Were Younger Than 65 On The Likelihood Of Being In The Top 10 Percent Of Medicare Or Medicaid Spending, 2006–09 Medicare
Medicaid
Characteristic
Likelihood of being in top 10%
SE
Likelihood of being in top 10%
Age Female
−0.0006 −0.064
(0.0059) (0.131)
Black Asian Latino
0.579*** −0.740 −0.392
(0.148) (0.797) (0.286)
1.82
0.744*** 0.061 2.815***
(0.129) (0.141) (0.156)
2.10
Schizophrenia Bipolar disorder Depression Substance use disorder
−0.095 0.051 0.100 0.506***
(0.171) (0.168) (0.149) (0.157)
Other mental health diagnosis, any year In 2007 In 2008 In 2009
−0.125 −0.092 −0.030 −0.167
Constant
−4.350***
ADL limitations IADL limitations 2+ chronic conditions
OR
SE
0.011 0.028
(0.006) (0.134)
−0.109 −0.215 0.057
(0.154) (0.664) (0.134)
OR
16.70
0.597*** 0.470*** 1.236***
(0.133) (0.150) (0.125)
1.82 1.60 3.44
(0.177) (0.198) (0.161) (0.197)
1.91
1.66
0.646*** −0.009 0.226 −0.329
(0.156) (0.138) (0.142) (0.151)
0.337*** 0.063 0.009 0.063
(0.150) (0.112) (0.129) (0.141)
1.03
(0.327)
−4.034***
(0.310)
SOURCE Authors’ analysis of data from the Medicare Current Beneficiary Survey, 2006–09. NOTES “Top 10%” of spending is the top 10 percent of Medicare or Medicare spending for dual eligibles with a mental health problem. SE is standard error. ADL is activities of daily living. IADL is instrumental activities of daily living. “Limitations” is functional limitations. QIC for Medicare is 2,441; for Medicaid, 2,954. OR is odds ratio. ***p < 0:01
UnitedHealth Group also proposed using managed care for all dual eligibles and projected that savings over a ten-year period would total $87 billion.13 Still more optimistic estimates of savings from coordinating care for dual eligibles have reached $20 billion per year.14 However, other analyses, citing the scarcity of existing organizations that currently have the capacity to realize significant savings, are less optimistic about the potential for savings. Specifically, there are few organizations that currently provide coordinated care that encompasses medical care, mental health services, substance use disorder care, and long-term services and supports that form the basis for the projected savings.15 Efforts to coordinate care are expensive and complex to implement. They typically target specific types of costly services that are thought to be overused, such as inpatient hospital care, nursing home stays, or ED use. The presumption is that efforts to make care more efficient will be more likely to succeed if they target high-cost beneficiaries who incur preventable costs.16 We focused on people who were younger than sixty-five and eligible for both Medicare and Medicaid. They likely present markedly different
care coordination challenges than do older adults in nursing homes. For instance, nursing home residents are confined to the facility, and the services that address the range of residents’ needs can be assembled under a single roof. In contrast, organizing the services that dual eligibles living in the community need involves dealing with numerous service delivery organizations (including those that provide medical and behavioral health services; public programs such as the Supplemental Nutrition Assistant Program, formerly known as food stamps; and employment services) that are typically in separate locations. We were especially interested in young dual eligibles with mental disorders because they are numerous (about 40 percent of all young dual eligibles), incur above-average spending, and are viewed as potentially benefiting from coordinated care arrangements. Our findings offer some important observations about the development of policy related to young dual eligibles. First, young dual eligibles who had a mental disorder were on average considerably more expensive than those who did not have such a disorder. Moreover, in the case of the beneficiaJ un e 2 0 1 4
33 : 6
Health A ffairs
1011
Mental Health ries with a mental disorder, the bulk of the spending—some of which resulted from relatively high use of inpatient services—was paid for by Medicare. Second, relatively few people incurred high expenditures in both programs. This is because there is a clear division of responsibility between Medicare and Medicaid for funding different types of services that meet different sets of needs. These patterns have important implications when considered together with recent findings from programs that aim to coordinate care for people with severe mental disorders. Recent results from programs in Pennsylvania17 and Washington State suggest that multidisciplinary care management teams targeting their services to patients at elevated risk of needing high-cost care can reduce the use of inpatient and emergency services. Thus, the increased use of services such as case management, care coordination teams, and psychosocial rehabilitation services that are typically paid for by Medicaid may lead to reductions in Medicare—and potentially in overall—expenditures. Of course, the mix of state and federal funding for Medicaid and Medicare is very different in each program. Thus, in constructing shared savings programs, it is important to think carefully about the timing of when savings in each program will occur and what investments are needed in each program to realize any benefits overall. Our analyses also showed that within the highcost group of young dual eligibles with a mental disorder, there is considerable heterogeneity in spending. Discussions of initiatives aimed at dual eligibles often assume that all such beneficiaries are the target of policies intended to yield better care and cost savings. The results depicted in Exhibit 4 show that even within the population of young dual eligibles with mental disorders, 25 percent of the population accounted for 70 percent of the spending. Thus, targeting initiatives within the population of dual eligibles is likely to be important. Our regression models identified a number of characteristics of beneficiaries who incur high levels of spending in each program. These can be used to target care coordination efforts. For example, our results showed that having substance use disorders, limitations in activities of daily living, and multiple chronic medical conditions beyond mental disorders substantially increased the probability that beneficiaries would be among the highest spenders in both Medicaid and Medicare.18,19 Thus, it may be beneficial for care management teams to focus on identifying people with these conditions and ensuring that they get evidence-based treatment for 1012
Health Affairs
J u n e 201 4
3 3: 6
Our findings underscore the importance of contracting with organizations that can deploy care coordination resources in nuanced and targeted ways.
their substance use disorders and the long-term services and supports they need to address their functional limitations. Facilitating appropriate linkages and communication between primary care and specialty behavior health providers also may be beneficial. Data on all of these predictive factors are not typically available in existing Medicaid or Medicare claims. However, efforts such as those in Massachusetts and Washington State to appropriately manage care for dual eligibles are resulting in new beneficiary assessment methods and data collection. This opens up new opportunities to make use of analyses such as those presented here. Our findings underscore the importance of contracting with organizations that can deploy care coordination resources in nuanced and targeted ways. It seems natural to target interventions first at beneficiaries with high levels of hospitalization and other expensive medical services, since if the interventions are successful, they may yield considerable savings. Further research should identify the expenditures that are most likely to be preventable.
Conclusion Our data show that people younger than sixtyfive who are dually eligible for Medicare and Medicaid have high costs of care, especially when they suffer from a severe mental disorder. The cost of care for such people is highly variable and well associated with clinical comorbidity and functional problems. Programs seeking to improve the quality of care and reduce costs for this group of patients through care coordination will probably be more cost-effective if they are tar-
geted at the people within the overall group who are most at risk of generating the highest spending levels, such as those with comorbid substance use disorders and functional impairments. Policy makers should design gain-sharing arrangements carefully. It seems likely that Medicare savings (which would be realized solely by the federal government) could be achieved as a
result of well-managed increases in spending on services covered by Medicaid (which are paid for by a mix of state and federal funds). This means that it might be wise for the federal government to share with states some portion of any savings achieved by Medicare, as a way to encourage the investments in Medicaid services that are needed to produce the gains. ▪
The authors are grateful to the Commonwealth Fund for support of this research and to Yanmei Liu for expert programming assistance.
NOTES 1 House of Representatives, Committee on Energy and Commerce, Democrats. Hearing on “dualeligibles: understanding this vulnerable population and how to improve their care,” Subcommittee on Health (June 21, 2011) [Internet]. Washington (DC): House of Representatives; 2011 Jun 21 [cited 2014 Apr 7]. Available from: http:// democrats.energycommerce.house .gov/index.php?q=hearing/hearingon-dual-eligibles-understandingthis-vulnerable-population-andhow-to-improve-their2 Medicare Payment Advisory Commission. Report to the Congress: aligning incentives in Medicare [Internet]. Washington (DC): MedPAC; 2010 Jun [cited 2014 Apr 7]. Available from: http://www.medpac.gov/ documents/jun10_entirereport.pdf 3 Department of Health and Human Services. Mental health: a report of the surgeon general [Internet]. Washington (DC): HHS; 1999 [cited 2014 Apr 7]. Available from: http:// profiles.nlm.nih.gov/ps/access/ NNBBHS.pdf 4 Gilmer T, Hamblin A. Hospital readmissions among Medicaid beneficiaries with disabilities: identifying targets of opportunities [Internet]. Hamilton (NJ): Center for Health Care Strategies Inc.; 2010 Dec [cited 2014 Apr 7]. (Faces of Medicaid Data Brief). Available from: http://www .chcs.org/usr_doc/CHCS_read mission_101215b.pdf 5 Mechanic D, Aiken LH. Improving the care of patients with chronic mental illness. N Engl J Med. 1987;317(26):1634–8. 6 Stein LI, Test MA, Marx AJ. Alternatives to the hospital: a controlled study. Am J Psychiatry. 1975; 132(5):517–22. 7 Health Management Associates. Dual eligible financial alignment demonstration calendar. HMA Weekly Roundup: Trends in State Health Policy [serial on the Internet]. Lansing (MI): HMA; 2014
8
9
10
11
12
13
Apr 9 [cited 2014 Apr 23]. Available from: http://www.health management.com/assets/WeeklyRoundup/040914-HMARoundup.pdf Foote SM, Hogan C. Disability profile and health care costs of Medicare beneficiaries under age sixty-five. Health Aff (Millwood). 2001;20(6): 242–53. We adjusted by the Consumer Price Index for All Urban Consumers (CPIU) because we were attempting to adjust for economywide price increases that affected purchasing power, not just medical care price increases. It is also important to point out that for the most of the past twenty years, prices for behavioral health services and long-term care have grown more slowly than those for general medical care. Thus, using the CPI-U allowed us to make a reasonable adjustment to people’s ability to purchase behavioral health services over time. Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized estimating equation approach. Biometrics. 1988;44(4): 1049–60. Erratum in Biometrics. 1989;45(1):347. Medicaid and CHIP Payment and Access Commission. Report to the Congress on Medicaid and CHIP [Internet]. Washington (DC): MACPAC; 2013 Mar [cited 2014 Apr 8]. Available from: http:// cnsnews.com/sites/default/files/ documents/MACPAC%20REPORT2013.pdf National Commission on Fiscal Responsibility and Reform. The moment of truth [Internet].Washington (DC): White House; 2010 Dec [cited 2014 Apr 8]. Available from: http:// www.fiscalcommission.gov/sites/ fiscalcommission.gov/files/ documents/TheMomentof Truth12_1_2010.pdf UnitedHealth Center for Health Reform and Modernization. US deficit reduction: the Medicare and Medic-
14
15
16
17
18
19
aid modernization opportunity [Internet]. Minnetonka (MN): UnitedHealth Group; 2010 Oct [cited 2014 Apr 8]. (Working Paper No. 4). Available from: http://www.united healthgroup.com/hrm/UNH_ WorkingPaper4.pdf Thorpe KE. Estimated federal savings associated with care coordination models for Medicare-Medicaid dual eligibles [Internet].Washington (DC): America’s Health Insurance Plans; 2011 Sep [cited 2014 Apr 8]. Available from: http://www.ahip coverage.com/wp-content/uploads/ 2011/09/Dual-Eligible-StudySeptember-2011.pdf Gold MR, Jacobson GA, Garfield RL. There is little experience and limited data to support policy making on integrated care for dual eligibles. Health Aff (Millwood). 2012; 31(6):1176–85. Brown R, Mann DR. Best bets for reducing Medicare costs for dual eligible beneficiaries: assessing the evidence [Internet]. Menlo Park (CA): Kaiser Family Foundation; 2012 Oct [cited 2014 Apr 8]. (Issue Brief). Available from: http:// kaiserfamilyfoundation.files .wordpress.com/2013/01/8353.pdf Kim JY, Higgins TC, Esposito D, Gerolamo AM, Flick M. SMI Innovations Project in Pennsylvania: final evaluation report [Internet]. Princeton (NJ): Mathematica Policy Research; 2012 Oct 1 [cited 2014 Apr 8]. Available from: http:// www.mathematica-mpr.com/ publications/PDFs/health/SMI_ Innovations_PA_final.pdf Our results for dual eligibles differ somewhat from findings for the Medicaid-only population (see Notes 4 and 19). Buck JA, Teich JL, Bae J, Dilonardo J. Mental health and substance abuse services in ten state Medicaid programs. Adm Policy Ment Health. 2001;28(3):181–92.
June 2014
33:6
Health A ffairs
1013
doi:
Richard G. Frank (frank@ hcp.med.harvard.edu) is the Margaret T. Morris Professor of Health Economics in the Department of Health Care Policy, Harvard Medical School, in Boston, Massachusetts. Arnold M. Epstein is a professor of health policy and management, Harvard School of Public Health, in Boston.
Factors Associated With High Levels Of Spending For Younger Dually Eligible Beneficiaries With Mental Disorders ABSTRACT The Affordable Care Act focused attention on how conflicting rules and payment arrangements in Medicare and Medicaid can produce high costs and fragmented care for people who are dually eligible for the two programs. Nearly half of such dual eligibles have severe and persistent mental disorders. Using Medicare data for the period 2006–09, we examined factors that were associated with high levels of spending for dual eligibles younger than sixty-five with a mental disorder. We found that these beneficiaries were nearly twice (1.86 times) as expensive as young dual eligibles who did not have a mental disorder. We identified functional limitations, multiple chronic conditions, and substance use disorders as being associated with high levels of spending in this subpopulation. We conclude that case management that coordinated medical, mental health, and substance use treatment along with psychosocial rehabilitation services could yield savings, primarily to the Medicare program. Because only Medicaid pays for case management and psychosocial rehabilitation services, Medicaid spending may need to rise if overall savings are to be realized.
P
eople who are eligible for both Medicare and Medicaid, commonly called “dual eligibles,” are among the most vulnerable members of the US population. Medicare beneficiaries qualify for full or partial Medicaid coverage if their incomes and assets fall below certain thresholds or if they have high health care bills. Each state sets its own income thresholds and scopes of benefits for the dual eligibles, within federal guidelines. The nine million people who were dual eligibles in 2008 accounted for 39 percent of Medicaid’s spending and 29 percent of Medicare’s in that year.1 Dual eligibles often receive substandard care.1 Roughly one-third of them are under age sixty-five and are sometimes called “young duals”: They receive Medicare benefits because of a disability and account for 37 percent of spending for all dual eligibles.2 People with se1006
Health Affairs
June 2014
33:6
vere and persistent mental illnesses account for nearly half of young dual eligibles.2 Young dual eligibles with severe mental illnesses typically have complex needs.2 To remain in the community, they require medical care, specialty psychiatric care, prescription drugs, long-term services and supports (such as rehabilitation services and case management), and human services (for example, services related to housing and employment). It has long been realized that “the organization of services for adults with severe mental disorders is the linchpin of effective treatment.”3 Thus, the integration and coordination of care to meet their complex needs is important. About 28 percent of young dual eligibles have depression, and about 12 percent have schizophrenia or bipolar disorder.Young dual eligibles with mental disorders tend to have higher chronic disease burdens and to account for high levels
of spending on emergency department (ED) services and inpatient care, compared to young dual eligibles who have other disabling conditions.2 Dual eligibles with schizophrenia have one of the highest rates of hospital readmissions of any group of people with a particular disease. Improved care coordination has been shown to improve care and reduce spending on ED visits and hospitalizations.3–6 For the dual-eligible population, Medicare pays for traditional medical services such as physician visits, hospital stays, and prescription drugs. For people with disabling mental disorders, Medicaid pays for long-term services and supports such as case management, residential or nursing home care, and psychosocial rehabilitation services. As part of its efforts to reform the payment and care delivery systems, the Affordable Care Act tried to align rules and payment arrangements between Medicare and Medicaid. Current conflicts between the two programs’ regulations can result in incentives that produce high costs and fragmented care for dual eligibles. For example, Medicaid bed-hold policies—that is, paying nursing homes a fee to hold a bed open for a patient who has been hospitalized—in conjunction with Medicare’s high payment rates for postacute stays at skilled nursing facilities create incentives for the rehospitalization of nursing home patients. To address this problem, in 2011 the federal government, under authority provided by the Affordable Care Act, established a demonstration program that allows states to align the two programs’ incentives, improve care coordination, and share in any savings that accrue.1 As of April 2014 seventeen states had launched demonstrations in this program or were planning to launch them by the end of the year.7 Ten of those states have signed memoranda of understanding with the Centers for Medicare and Medicaid Services (CMS). Two states initiated their programs ahead of others: Washington State began its demonstration for dual eligibles in October 2012 and expanded it in May 2013, and Massachusetts launched its demonstration in October 2013.7 The demonstrations can include arranging housing, medical care, behavioral health services, long-term care, and other support services for dual eligibles who have functional limitations. Seven of the demonstrations (those in Illinois, Massachusetts, Michigan, New York, Ohio, South Carolina, and Washington) emphasize care for people with severe mental illnesses. The success of these interventions may depend on directing efforts toward those segments of the population that are the most costly to the Medi-
care and Medicaid programs and whose quality of care is most in need of improvement. We analyzed patterns of spending among dual eligibles younger than sixty-five who suffer from mental disorders. It has been known for some time that people with severe mental disorders are overrepresented in the Medicare population in this age group. However, little attention has been devoted to dual eligibles with these conditions.8 We identified factors that were associated with higher spending levels and examined the distribution of spending (and, as a result, potential savings) between Medicare and Medicaid.
Study Data And Methods Data We used data from the Medicare Current Beneficiary Survey (MCBS) for the period 2006– 09. By using those four years and oversampling dually eligible beneficiaries, we were able to obtain sufficient numbers of cases to conduct our analysis. The MCBS is a representative survey of the Medicare population. It contains a rich array of information, including beneficiaries’ demographic characteristics, health care use, living arrangements, functional impairments, health status, and program participation. It also includes Medicare claims data. Each year the survey produces a sample of about 14,500 beneficiaries. Response rates for surveys in the study period ranged from 85 percent to 88 percent. Survey respondents are typically kept in the survey for three years. Thus, in each year the sample includes both new and continuing respondents. The MCBS oversamples people ages eighty-five and older and those younger than age sixty-five. About 93 percent of the sample resides in the community; the remaining 7 percent live in long-term care facilities. CMS sponsors the survey and provides public-use files on cost and use of, and access to, care. Our analyses relied primarily on the cost and use files. The data in these files include linkages between survey responses and Medicare claims, as well as use of and spending on all health care not covered by Medicare. A beneficiary is reported as having a mental disorder if he or she had a claim with a primary diagnosis that called for services associated with care of a mental illness (International Classification of Diseases, Ninth Revision, codes 290–319), use of a psychotherapeutic drug (MCBS variable THERCC or therapeutic class ¼ 80), or treatment under a specialty mental health procedure code during the survey year. Our sample of dual eligibles younger than sixty-five ranged from 1,163 to 1,306 per year. Between 788 and 883 of J u n e 201 4
3 3: 6
Health Affairs
1007
Mental Health them each year were reported as having a mental disorder. We studied individuals in some analyses (such as our description of average costs) and total person-years in other analyses (for example, our regression models). For analyses of individuals, we chose the second year if the beneficiary was surveyed twice and the middle year if he or she was surveyed three times. This approach resulted in 4,999 person-year observations. Twenty-eight percent of dual eligibles reported to have mental disorders had depression, 18 percent had bipolar disorder, and 21 percent had schizophrenia. The remainder had other classes of diagnoses, including anxiety, substance use disorders, and obsessive-compulsive disorder. These figures accord well with data in the Medicare Payment Advisory Commission’s (MedPAC’s) 2010 report to Congress.2 Analysis We examined four patterns of spending for dual eligibles. First, we assessed total spending on those with mental disorders compared to spending on those with other forms of disability. Second, we described spending on young dual eligibles by type of service (inpatient care, outpatient care, drugs, or other services) and source of payment (Medicare or Medicaid). Third, we examined total spending by demographic characteristics, health, and functional status of beneficiaries and by the source of payment. And fourth, we examined the shares of spending that different population segments accounted for, starting with the group whose costs were the highest. All spending estimates are in 2009 dollars, adjusted by the Consumer Price Index for All Urban Consumers.9 We used multivariate logistical regression models to examine the explanatory power of key demographic, health, and function characteristics on high levels of spending overall and by Medicare and Medicaid separately. This approach enabled us to identify high spenders
Exhibit 1 Average Annual Spending For People Dually Eligible For Medicare And Medicaid Who Were Younger Than 65 And Had A Mental Disorder, By Type Of Spending, 2006–09 Type of spending
Patients
Total spending ($)
Spending by Medicare (%)
Spending by Medicaid (%)
Inpatient Outpatient Drug Othera Total
827 2,705 2,875 3,278 3,318
4,862 2,344 6,246 10,705 24,155
83 75 87 31 60
6 12 5 44 23
SOURCE Authors’ analysis of data from the Medicare Current Beneficiary Survey, 2006–09. NOTES Spending was adjusted to 2009 dollars using the Consumer Price Index. N ¼ 3; 318. aResidential, case management, and partial hospital spending.
1008
H e a lt h A f fai r s
J u n e 201 4
3 3: 6
and potential targets of care coordination efforts. To obtain the appropriate standard errors for our coefficient estimates, our estimation approach took account of the repeated measures with non-Gaussian data.We accomplished this by using the generalized estimating equations estimator and taking account of the complex sample design.10 Limitations There are several limitations to this study. Our survey data were based on interviews with respondents living in the community and thus are self-reported. The information on health status and functioning was not clinically validated. The Medicare spending data were based on claims, while Medicaid and other spending data were based on self-report and CMS’s imputations of costs. The sample reflected primarily fee-for-service spending, largely because relatively few dual eligibles had enrolled in managed care arrangements. Because of the limited number of people younger than sixty-five in the MCBS, our analyses had somewhat limited power. This was especially true when we stratified them according to dual eligibles’ different levels of function and comorbidity. Our national sample did not allow us to describe regional variation. It is tempting to conclude that integrated care programs would be most cost-effective if they were targeted at patients associated with higher costs. However, we cannot be sure of that without an intervention study. Our analyses of spending in relation to mental health diagnosis may be overestimates if high-cost users with mental health conditions were more likely than people with other conditions to be diagnosed in our data.
Study Results Dual eligibles younger than sixty-five incurred average annual costs for all health and long-term services and supports of $20,383, based on the data for the period 2006–09 and expressed in 2009 dollars. For those identified as having a mental disorder, average annual spending was estimated to be $24,155 (Exhibit 1), in contrast to $12,948 for young dual eligibles with other disabilities (a ratio of 1.86). Thus, spending on young dual eligibles with a mental disorder was slightly less than twice as much as spending on young dual eligibles with other types of disabilities. Spending By Type Of Service And Payer Sixty percent of all spending for young dual eligibles with a mental disorder was paid for by Medicare, and 23 percent was paid for by Medicaid (Exhibit 1). Roughly 25 percent was spent on
drugs, and nearly 40 percent was spent on other services—including institutional providers such as nursing homes, residential treatment, and nonmedical services such as case management. Other services were more often paid for by Medicaid than by Medicare. The remainder went to inpatient and outpatient care that was largely paid for by Medicare. In a given year, 25 percent of dual eligibles with a mental disorder were hospitalized, and 11.5 percent were hospitalized twice or more. In contrast, 14 percent of dual eligibles without a mental disorder were hospitalized, and 4.9 percent had two or more inpatient stays. Exhibit 2 plots total health care spending during the study period for all young dual eligibles with mental disorders according to source of payment: Medicare or Medicaid. The exhibit highlights three important phenomena. First, there was considerable variation in spending levels in both programs for young dual eligibles. Second, despite that variation, the proportion of relatively low spenders was much greater in Medicaid than in Medicare. This is illustrated in Exhibit 2 by the density of points along the vertical axis clustered near the origin of the horizontal axis. As a result, median spending was about $667 per year in Medicaid, versus $6,665 in Medicare. The respective numbers for average annual spending were $5,460 for Medicaid and $14,178 for Medicare. Third, there were few cases of high levels of spending by both Medicare and Medicaid. In other words, young dual eligibles who needed high-intensity services appeared to need chiefly either high-cost medical care (such as inpatient and ED care) that was primarily paid for by Medicare or services that were mostly paid for by Medicaid—expensive institutional placements (for example, in a nursing home) or intensive community-based support services. Spending By Diagnosis, Functional Status, And Substance Abuse Average annual spending in 2009 dollars for young dual eligibles without limitations in activities (or instrumental activities) of daily living ranged from a low of $10,021 for those without a mental disorder to $23,471 for those with substance use disorders (Exhibit 3). For young dual eligibles with limitations, the range in spending was also large. The results underscore that it is combinations of mental disorders with a functional impairment or comorbid substance use disorders that generate high levels of spending. For a beneficiary with depression, adding either a functional impairment or a substance use disorder approximately doubled average annual spending (Exhibit 3). Interestingly, for a beneficiary with a
diagnosis not related to mental health, the presence of a substance use disorder resulted in a 2.5fold increase in average annual spending. Predictors Of High Spending By Dual Eligibles Disease burdens appeared greater for young dual eligibles with a mental disorder than for those without such disorders. However, even the subpopulation with a mental disorder had variable spending levels (Exhibit 4). The top 10 percent of spenders accounted for 43 percent of spending overall. The highest-spending half of the population accounted for nearly 90 percent of all spending. Of course, the corollary is also true: The lowest-spending half of the population accounted for about 11 percent of spending. We used a logistic regression model to determine the likelihood that a young dual eligible would be in the top 10 percent of either Medicare or Medicaid spending. The model results revealed both similarities and some notable differences between the two programs in the factors that explained high levels of spending (Exhibit 5). For example, having functional limitations in activities of daily living (see Exhibit 3 notes) or
Exhibit 2 The Relationship Between Medicare And Medicaid Spending For People Eligible For Both Programs Who Were Younger Than 65 And Had A Mental Disorder, 2006–09
SOURCE Authors’ analysis of data from the Medicare Current Beneficiary Survey. NOTE Spending is total health care spending during the study period for all young dual eligibles with mental disorders.
June 2014
33:6
H ea lt h A f fai r s
1 009
Mental Health Exhibit 3 Average Annual Spending For People Dually Eligible For Medicare And Medicaid Who Were Younger Than 65, By Diagnosis, 2006–09
Diagnosis Depression Substance use disorder Bipolar disorder Schizophrenia Other, mental healthb Other, not mental health
No ADL or IADL limitations
ADL limitations
IADL limitations
Substance use disorder
No. 337 156 228 192 505 604
No. 449 158 173 133 772 529
No. 711 273 323 348 1,289 1,009
No. 225 —a 138 98 318 45
Spending ($) 20,078 23,471 19,150 21,079 19,222 10,021
Spending ($) 41,829 40,243 39,364 41,129 41,276 19,922
Spending ($) 33,675 32,628 29,882 27,757 33,154 14,694
Spending ($) 38,083 —a 33,120 36,059 33,456 26,846
SOURCE Authors’ analysis of data from the Medicare Current Beneficiary Survey, 2006–09. NOTES Spending was adjusted to 2009 dollars using the Consumer Price Index. ADL is activities of daily living (bathing, dressing, eating, using the toilet, and transferring into or out of a bed or chair). IADL is instrumental activities of daily living (managing money, meal preparation, shopping, doing housework, using the telephone, and taking medications). “Limitations” is functional limitations. aNot applicable. b Includes anxiety disorders and personality disorders.
multiple chronic conditions increased the likelihood that a beneficiary would have a high level of spending in both Medicare and Medicaid. Yet having multiple chronic conditions had a more powerful impact on the probability of incurring high costs in Medicare than in Medicaid (odds
Exhibit 4 Combined Medicare And Medicaid Spending For People Eligible For Both Programs Who Were Younger Than 65 And Had A Mental Health Disorder, By Level Of Spending, 2006–09 100
80
ratio for Medicare: 16.70; for Medicaid: 3.44). In contrast, having functional limitations had a similar impact on the odds of being in the top 10 percent of spending in Medicare (odds ratio: 2.10) and Medicaid (odds ratio: 1.82). The differences in explanatory factors for being a high-cost beneficiary were linked to the different services that each program funds. For example, limitations in instrumental activities of daily living (see Exhibit 3 notes) and a diagnosis of schizophrenia predicted high levels of Medicaid spending but not Medicare spending. People with these conditions frequently need long-term services and supports that are covered by Medicaid—including case management, residential services, and personal assistance—in addition to medical care.
Percent of population
Discussion 60
40
20
0 0
20
40
60 Percent of spending
80
100
SOURCE Authors’ analysis of data from the Medicare Current Beneficiary Survey. NOTE “Spending” is total spending from all payment sources during the study period.
1010
Health Affairs
J u n e 201 4
3 3: 6
It has been well established that beneficiaries who are eligible for both Medicare and Medicaid represent a high-cost group whose care could be improved.11 Sections 2601 and 2602 of the Affordable Care Act focus on measures that aim to improve both the alignment between Medicare and Medicaid in serving dually eligible beneficiaries and the coordination of clinical services for a population with complex needs. The National Commission on Fiscal Responsibility and Reform recommended placing dual eligibles in Medicaid managed care plans, reflecting the view that health plans that managed all of the care for a population with complex needs within a budget would likely provide better and more efficient care than that provided under unmanaged fee-for-service arrangements. The commission estimated that if its recommendation were followed, $13 billion would be saved in the period 2015–20.12
Exhibit 5 Impact Of Demographic, Clinical, And Functional Characteristics Of People Dually Eligible For Medicare And Medicaid Who Were Younger Than 65 On The Likelihood Of Being In The Top 10 Percent Of Medicare Or Medicaid Spending, 2006–09 Medicare
Medicaid
Characteristic
Likelihood of being in top 10%
SE
Likelihood of being in top 10%
Age Female
−0.0006 −0.064
(0.0059) (0.131)
Black Asian Latino
0.579*** −0.740 −0.392
(0.148) (0.797) (0.286)
1.82
0.744*** 0.061 2.815***
(0.129) (0.141) (0.156)
2.10
Schizophrenia Bipolar disorder Depression Substance use disorder
−0.095 0.051 0.100 0.506***
(0.171) (0.168) (0.149) (0.157)
Other mental health diagnosis, any year In 2007 In 2008 In 2009
−0.125 −0.092 −0.030 −0.167
Constant
−4.350***
ADL limitations IADL limitations 2+ chronic conditions
OR
SE
0.011 0.028
(0.006) (0.134)
−0.109 −0.215 0.057
(0.154) (0.664) (0.134)
OR
16.70
0.597*** 0.470*** 1.236***
(0.133) (0.150) (0.125)
1.82 1.60 3.44
(0.177) (0.198) (0.161) (0.197)
1.91
1.66
0.646*** −0.009 0.226 −0.329
(0.156) (0.138) (0.142) (0.151)
0.337*** 0.063 0.009 0.063
(0.150) (0.112) (0.129) (0.141)
1.03
(0.327)
−4.034***
(0.310)
SOURCE Authors’ analysis of data from the Medicare Current Beneficiary Survey, 2006–09. NOTES “Top 10%” of spending is the top 10 percent of Medicare or Medicare spending for dual eligibles with a mental health problem. SE is standard error. ADL is activities of daily living. IADL is instrumental activities of daily living. “Limitations” is functional limitations. QIC for Medicare is 2,441; for Medicaid, 2,954. OR is odds ratio. ***p < 0:01
UnitedHealth Group also proposed using managed care for all dual eligibles and projected that savings over a ten-year period would total $87 billion.13 Still more optimistic estimates of savings from coordinating care for dual eligibles have reached $20 billion per year.14 However, other analyses, citing the scarcity of existing organizations that currently have the capacity to realize significant savings, are less optimistic about the potential for savings. Specifically, there are few organizations that currently provide coordinated care that encompasses medical care, mental health services, substance use disorder care, and long-term services and supports that form the basis for the projected savings.15 Efforts to coordinate care are expensive and complex to implement. They typically target specific types of costly services that are thought to be overused, such as inpatient hospital care, nursing home stays, or ED use. The presumption is that efforts to make care more efficient will be more likely to succeed if they target high-cost beneficiaries who incur preventable costs.16 We focused on people who were younger than sixty-five and eligible for both Medicare and Medicaid. They likely present markedly different
care coordination challenges than do older adults in nursing homes. For instance, nursing home residents are confined to the facility, and the services that address the range of residents’ needs can be assembled under a single roof. In contrast, organizing the services that dual eligibles living in the community need involves dealing with numerous service delivery organizations (including those that provide medical and behavioral health services; public programs such as the Supplemental Nutrition Assistant Program, formerly known as food stamps; and employment services) that are typically in separate locations. We were especially interested in young dual eligibles with mental disorders because they are numerous (about 40 percent of all young dual eligibles), incur above-average spending, and are viewed as potentially benefiting from coordinated care arrangements. Our findings offer some important observations about the development of policy related to young dual eligibles. First, young dual eligibles who had a mental disorder were on average considerably more expensive than those who did not have such a disorder. Moreover, in the case of the beneficiaJ un e 2 0 1 4
33 : 6
Health A ffairs
1011
Mental Health ries with a mental disorder, the bulk of the spending—some of which resulted from relatively high use of inpatient services—was paid for by Medicare. Second, relatively few people incurred high expenditures in both programs. This is because there is a clear division of responsibility between Medicare and Medicaid for funding different types of services that meet different sets of needs. These patterns have important implications when considered together with recent findings from programs that aim to coordinate care for people with severe mental disorders. Recent results from programs in Pennsylvania17 and Washington State suggest that multidisciplinary care management teams targeting their services to patients at elevated risk of needing high-cost care can reduce the use of inpatient and emergency services. Thus, the increased use of services such as case management, care coordination teams, and psychosocial rehabilitation services that are typically paid for by Medicaid may lead to reductions in Medicare—and potentially in overall—expenditures. Of course, the mix of state and federal funding for Medicaid and Medicare is very different in each program. Thus, in constructing shared savings programs, it is important to think carefully about the timing of when savings in each program will occur and what investments are needed in each program to realize any benefits overall. Our analyses also showed that within the highcost group of young dual eligibles with a mental disorder, there is considerable heterogeneity in spending. Discussions of initiatives aimed at dual eligibles often assume that all such beneficiaries are the target of policies intended to yield better care and cost savings. The results depicted in Exhibit 4 show that even within the population of young dual eligibles with mental disorders, 25 percent of the population accounted for 70 percent of the spending. Thus, targeting initiatives within the population of dual eligibles is likely to be important. Our regression models identified a number of characteristics of beneficiaries who incur high levels of spending in each program. These can be used to target care coordination efforts. For example, our results showed that having substance use disorders, limitations in activities of daily living, and multiple chronic medical conditions beyond mental disorders substantially increased the probability that beneficiaries would be among the highest spenders in both Medicaid and Medicare.18,19 Thus, it may be beneficial for care management teams to focus on identifying people with these conditions and ensuring that they get evidence-based treatment for 1012
Health Affairs
J u n e 201 4
3 3: 6
Our findings underscore the importance of contracting with organizations that can deploy care coordination resources in nuanced and targeted ways.
their substance use disorders and the long-term services and supports they need to address their functional limitations. Facilitating appropriate linkages and communication between primary care and specialty behavior health providers also may be beneficial. Data on all of these predictive factors are not typically available in existing Medicaid or Medicare claims. However, efforts such as those in Massachusetts and Washington State to appropriately manage care for dual eligibles are resulting in new beneficiary assessment methods and data collection. This opens up new opportunities to make use of analyses such as those presented here. Our findings underscore the importance of contracting with organizations that can deploy care coordination resources in nuanced and targeted ways. It seems natural to target interventions first at beneficiaries with high levels of hospitalization and other expensive medical services, since if the interventions are successful, they may yield considerable savings. Further research should identify the expenditures that are most likely to be preventable.
Conclusion Our data show that people younger than sixtyfive who are dually eligible for Medicare and Medicaid have high costs of care, especially when they suffer from a severe mental disorder. The cost of care for such people is highly variable and well associated with clinical comorbidity and functional problems. Programs seeking to improve the quality of care and reduce costs for this group of patients through care coordination will probably be more cost-effective if they are tar-
geted at the people within the overall group who are most at risk of generating the highest spending levels, such as those with comorbid substance use disorders and functional impairments. Policy makers should design gain-sharing arrangements carefully. It seems likely that Medicare savings (which would be realized solely by the federal government) could be achieved as a
result of well-managed increases in spending on services covered by Medicaid (which are paid for by a mix of state and federal funds). This means that it might be wise for the federal government to share with states some portion of any savings achieved by Medicare, as a way to encourage the investments in Medicaid services that are needed to produce the gains. ▪
The authors are grateful to the Commonwealth Fund for support of this research and to Yanmei Liu for expert programming assistance.
NOTES 1 House of Representatives, Committee on Energy and Commerce, Democrats. Hearing on “dualeligibles: understanding this vulnerable population and how to improve their care,” Subcommittee on Health (June 21, 2011) [Internet]. Washington (DC): House of Representatives; 2011 Jun 21 [cited 2014 Apr 7]. Available from: http:// democrats.energycommerce.house .gov/index.php?q=hearing/hearingon-dual-eligibles-understandingthis-vulnerable-population-andhow-to-improve-their2 Medicare Payment Advisory Commission. Report to the Congress: aligning incentives in Medicare [Internet]. Washington (DC): MedPAC; 2010 Jun [cited 2014 Apr 7]. Available from: http://www.medpac.gov/ documents/jun10_entirereport.pdf 3 Department of Health and Human Services. Mental health: a report of the surgeon general [Internet]. Washington (DC): HHS; 1999 [cited 2014 Apr 7]. Available from: http:// profiles.nlm.nih.gov/ps/access/ NNBBHS.pdf 4 Gilmer T, Hamblin A. Hospital readmissions among Medicaid beneficiaries with disabilities: identifying targets of opportunities [Internet]. Hamilton (NJ): Center for Health Care Strategies Inc.; 2010 Dec [cited 2014 Apr 7]. (Faces of Medicaid Data Brief). Available from: http://www .chcs.org/usr_doc/CHCS_read mission_101215b.pdf 5 Mechanic D, Aiken LH. Improving the care of patients with chronic mental illness. N Engl J Med. 1987;317(26):1634–8. 6 Stein LI, Test MA, Marx AJ. Alternatives to the hospital: a controlled study. Am J Psychiatry. 1975; 132(5):517–22. 7 Health Management Associates. Dual eligible financial alignment demonstration calendar. HMA Weekly Roundup: Trends in State Health Policy [serial on the Internet]. Lansing (MI): HMA; 2014
8
9
10
11
12
13
Apr 9 [cited 2014 Apr 23]. Available from: http://www.health management.com/assets/WeeklyRoundup/040914-HMARoundup.pdf Foote SM, Hogan C. Disability profile and health care costs of Medicare beneficiaries under age sixty-five. Health Aff (Millwood). 2001;20(6): 242–53. We adjusted by the Consumer Price Index for All Urban Consumers (CPIU) because we were attempting to adjust for economywide price increases that affected purchasing power, not just medical care price increases. It is also important to point out that for the most of the past twenty years, prices for behavioral health services and long-term care have grown more slowly than those for general medical care. Thus, using the CPI-U allowed us to make a reasonable adjustment to people’s ability to purchase behavioral health services over time. Zeger SL, Liang KY, Albert PS. Models for longitudinal data: a generalized estimating equation approach. Biometrics. 1988;44(4): 1049–60. Erratum in Biometrics. 1989;45(1):347. Medicaid and CHIP Payment and Access Commission. Report to the Congress on Medicaid and CHIP [Internet]. Washington (DC): MACPAC; 2013 Mar [cited 2014 Apr 8]. Available from: http:// cnsnews.com/sites/default/files/ documents/MACPAC%20REPORT2013.pdf National Commission on Fiscal Responsibility and Reform. The moment of truth [Internet].Washington (DC): White House; 2010 Dec [cited 2014 Apr 8]. Available from: http:// www.fiscalcommission.gov/sites/ fiscalcommission.gov/files/ documents/TheMomentof Truth12_1_2010.pdf UnitedHealth Center for Health Reform and Modernization. US deficit reduction: the Medicare and Medic-
14
15
16
17
18
19
aid modernization opportunity [Internet]. Minnetonka (MN): UnitedHealth Group; 2010 Oct [cited 2014 Apr 8]. (Working Paper No. 4). Available from: http://www.united healthgroup.com/hrm/UNH_ WorkingPaper4.pdf Thorpe KE. Estimated federal savings associated with care coordination models for Medicare-Medicaid dual eligibles [Internet].Washington (DC): America’s Health Insurance Plans; 2011 Sep [cited 2014 Apr 8]. Available from: http://www.ahip coverage.com/wp-content/uploads/ 2011/09/Dual-Eligible-StudySeptember-2011.pdf Gold MR, Jacobson GA, Garfield RL. There is little experience and limited data to support policy making on integrated care for dual eligibles. Health Aff (Millwood). 2012; 31(6):1176–85. Brown R, Mann DR. Best bets for reducing Medicare costs for dual eligible beneficiaries: assessing the evidence [Internet]. Menlo Park (CA): Kaiser Family Foundation; 2012 Oct [cited 2014 Apr 8]. (Issue Brief). Available from: http:// kaiserfamilyfoundation.files .wordpress.com/2013/01/8353.pdf Kim JY, Higgins TC, Esposito D, Gerolamo AM, Flick M. SMI Innovations Project in Pennsylvania: final evaluation report [Internet]. Princeton (NJ): Mathematica Policy Research; 2012 Oct 1 [cited 2014 Apr 8]. Available from: http:// www.mathematica-mpr.com/ publications/PDFs/health/SMI_ Innovations_PA_final.pdf Our results for dual eligibles differ somewhat from findings for the Medicaid-only population (see Notes 4 and 19). Buck JA, Teich JL, Bae J, Dilonardo J. Mental health and substance abuse services in ten state Medicaid programs. Adm Policy Ment Health. 2001;28(3):181–92.
June 2014
33:6
Health A ffairs
1013
