Factors Affecting Costs in Medicaid Populations With Behavioral Health Disorders
Freeman, Elsie MD, MPH*,†; McGuire, Catherine A. BS‡; Thomas, John W. PhD*,§; Thayer, Deborah A. MBA*
*Muskie School of Public Service, University of Southern Maine, Portland, ME
†Integrated Care Projects, Maine DHHS, Office of Continuous Quality Improvement, Augusta
‡Health Data Resources, Muskie School of Public Service, University of Southern Maine, Portland, ME
§Emeritus of Health Policy and Management, University of Michigan, Ann Arbor, MI
Supplemental Digital Content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Website, www.lww-medicalcare.com.
Presented at the following conferences: Integration of Healthcare in Maine: Implications from the Multiple Chronic Conditions Project, Maine Multiple Chronic Condition Forum, Governor Hill Mansion, Augusta, ME. September 21, 2012; and The Hidden Costs of Behavioral Health Disorders: A Compelling Case for Improving Integrated Care, Quality Counts 2013 Conference: Aligning Forces to become the First State to Reach the Tripe Aim, Augusta Civic Center, Augusta, ME. April 3, 2013
The authors declare no conflict of interest.
Reprints: Catherine A. McGuire, BS, Health Data Resources, Muskie School of Public Service, University of Southern Maine, P.O. Box 9300, Portland, ME 04104. E-mail: firstname.lastname@example.org.
Persons with behavioral disorders incur higher healthcare costs. Although they utilize behavioral health (BH) services others do not, they also have higher utilization of medical services
To determine the degree to which higher costs for persons with BH disorders are attributable to utilization of BH services, multiple chronic medical conditions (CMCs) or other issues specific to populations with BH disorders.
Data base consisted of claims for 63,141 Medicaid beneficiaries, 49% of whom had one of 5 categories of BH disorder. Generalized linear models were used to identify relative impact of demographics, BH status, multiple CMCs and primary care access on total, behavioral, nonbehavioral, and medical/surgical costs.
Number of CMCs was associated with significant increases in all cost categories, including behavioral costs. Presence of any BH disorder significantly influenced these same costs, including those not associated with BH care. Effect size in each cost category varied by BH group.
BH status has a large impact on all healthcare costs, including costs of medical and other non-BH services. The number of CMCs affects BH costs independent of BH disorder. Results suggest that costs might be reduced through better integration of behavioral and medical health services.
Chronic medical conditions (CMCs) are recognized as a major driver of healthcare utilization, morbidity, and mortality in the United States, and their relationship to healthcare cost increases is well documented.1–5 The presence of multiple medical comorbidities is a further driver of increased morbidity, utilization, and cost, with an individual’s healthcare costs rising exponentially with the number of medical conditions present.6
Other studies provide convincing evidence that behavioral health (BH) disorders are associated with higher healthcare use and costs, both for general medical (non-BH) services and for BH services. This relationship has been shown for psychiatric disorders in general7–9 and for more specific conditions, including bipolar disorder,10 depression,11–13 and serious mental illness (SMI) diagnoses.14
This issue is especially important for Medicaid, the federal-state partnership providing publicly funded health insurance for eligible disabled and low-income Americans, as multiple CMCs and BH disorders are more prevalent in these populations.15 Medicaid provides one of 6 healthcare dollars spent in the United States, accounting for 20% of state and 7% of federal budgets. Medicaid members who are in the top 5% of the Medicaid population with respect to utilization account for 54% of overall Medicaid spending. The top 1% account for 25% of spending and among these 83% have 3 or more CMCs, including high rates of psychiatric disorders, associated with their increased utilization.16 Medicaid is the largest source of financing for BH services in the United States17 and its proportion of BH spending has increased from 17% in 1986 to 28% in 2005.18 Expansion of the number of persons covered by Medicaid under the federal Affordable Care Act (ACA) may lead to even greater expenditures in the future. To deal with the higher costs of caring for persons with multiple CMCs, providers and Medicaid administrators need information about the multiple factors that contribute to service use and costs for these populations.
Why do populations with BH disorders have higher costs of care? Multiple factors may contribute to this observation. First, persons with these disorders utilize types of services—BH services—that other patients do not. Second, people with BH disorders have higher rates of CMCs than those without such disorders,10,19–22 perhaps in part because they are more likely to engage in health risk behaviors or to be prescribed psychopharmacological treatments that make them more susceptible to problems such as elevations in serum glucose and lipids.23 Thus, they may require more care to address their relatively high levels of morbidity. Third, as a consequence of alterations in mood, attention or cognition associated with their psychiatric conditions, persons with BH disorders may face greater challenges in creating and sustaining effective collaborations with medical providers. They are less likely to make and keep preventive care appointments, and are more likely to be “fired” from healthcare practices.9,10 In addition, they more often fail to adhere to recommended treatments for asthma, myocardial infarction, diabetes, and other chronic diseases.24–26 Perhaps as a result of their difficulties engaging effectively with health providers and systems, they rely more heavily on costly emergency room and hospital inpatient services.8,9
In this paper, data on a long-term adult Medicaid population in the state of Maine is used to investigate relationships between BH disorders and costs and to address the following questions:
- To what extent do populations with BH disorders have increased costs—(a) total; (b) BH; (c) medical/surgical; (d) and all non-BH costs?
- How do CMCs affect each of these types of costs?
- What types of BH conditions are associated with the largest increases in each type of cost?
- How does the contribution of BH disorders to increases in costs compare to that of other CMCs?
The population for this study consisted of Maine Medicaid members 18–60 years of age in 2007 with at least 1 service claim and with continuous coverage, defined as 22 to 24 months of enrollment in Medicaid, from January 1, 2007 to December 31, 2008. Medicaid recipients with any period of Medicare or other third party health insurance were excluded, as only a part of their claims data were available.
Consistent with the Centers for Medicare & Medicaid Services (CMS) regulations permitting evaluation of claims data without individual informed consent, the authors’ institutional review board waived the requirement of individual consent for this study.
Maine Medicaid (MaineCare) is a fee-for-service primary care case management payment system, providing a monthly capitation payment to primary care providers (PCP) to coordinate care and services. Thirty-one states use a Medicaid primary care case management payment system.27
MaineCare covers CMS mandated services required of all states, including inpatient and outpatient hospital, physician-related services, nursing home, home health, laboratory, x-ray, and has elected to cover a broad array of CMS optional benefits, including pharmacy, long-term home-based and community-based care services, rehabilitation services for mental illness/substance abuse (SA), occupational and physical therapy, chiropractic, and some selective dental services. Maine’s optional services include Medicaid waivers, whereby expanded services are reimbursed for certain populations (eg persons with physical, mental, or intellectual disabilities) to support community-based services to members as an alternative to institutional level of care.
In 2007–2008, MaineCare covered CMS mandated populations (eg, families qualifying for Temporary Assistance for Needy families, low-income aged, and disabled) and some optional populations (eg, Children in the State Children’s Health Insurance Program, their parents with incomes <150% of federal poverty level, nondisabled adults with incomes <100%). In 2010, Maine was ranked 28th among all the states in per-person Medicaid spending, with an average annual spend of $5478 per person, compared with the national average of $5592.28
Dependent variables in the analysis included 4 measures of per member per month (PMPM) cost: total, BH, non-BH health, and medical/surgical (med/surg). For this study, BH services were considered to include home and community care waivers restricted to persons with SMI or intellectual disability (ID), specialty BH residential care, BH case management, community support, psychiatric hospitalizations, outpatient psychiatric care, psychiatric medications, and substance abuse treatment. Non-BH PMPM consisted of all costs not comprised by BH PMPM and equaled total PMPM minus BH PMPM. Non-BH PMPM included costs of medical/surgical care plus expenditures for other services particular to MaineCare, e.g. non-emergency transportation. Medical/surgical PMPM represented 88% of non-BH costs and in this study included costs for inpatient and outpatient medical and surgical care, pharmacy, therapies, and ambulance.
Factors hypothesized as influencing cost differences are listed in Table 1. Age was grouped into 2 categories with the younger group corresponding to women’s childbearing years. As females are known to consume more health services than males, the model adjusted for sex. Because of the difference in access to healthcare between rural and nonrural patients, the model adjusted for place of residence.29,30 Two indicators on the nature of the PCP relationship, lack of PCP, and fragmentation of PCP care were also included. Fragmentation of primary care, defined for this study as receipt of primary care services from multiple PCP practices, is associated with increased costs.31–33 BH disorders of each Medicaid member was identified using ICD-9 diagnosis codes submitted on claims data and each member was assigned to one of the mutually exclusive BH groups shown in Table 1. The disease typology proposed by Hwang and colleagues was used to classify the number of CMCs, excluding BH disorders from the count of CMCs (see Variable Descriptions, Supplemental Digital Content 1, http://links.lww.com/MLR/A651 for a complete definition and coding for all independent variables).
To determine how explanatory variables were associated with each of the 4 cost measures, Stata’s implementation of generalized linear models (GLM) was employed, using Medicaid member as the unit of analysis. Unlike linear regression, GLM analyses are able to analyze relationships with dependent variables—such as healthcare costs—that do not meet Gaussian distribution criteria. Stata’s implementation of the GLM procedure further reduces potential bias by providing for inclusion of cases with $0 cost in the population analyzed. In our data, 2.7% of cases had $0 total PMPM, 37.1% had $0 BH PMPM, 2.9% had $0 non-BH PMPM, and 3.6% had $0 medical-surgical PMPM. Four alternative GLM models (identity link, Gaussian family; log-link, Gaussian family; log-link, gamma family and identity link, gamma family) were compared using Bayesian information criterion, Akaike information criterion, and deviance. On the basis of this analysis, the first 3 models were rejected in favor of the linear link, gamma family. Statistics are reported based on that model.
The total number of individuals in the study cohort was 63,141, of whom 65.3% were female (Table 1). The racial distribution of the cohort reflected that of Maine as a whole, with 93.6% white. Among nonwhites, African Americans (3% overall) and Native Americans (2.1% overall) were the most prominent.
The long-term cohort in this study represented 47.2% of adult Maine Medicaid members aged 18–60 and accounted for 75.5% of the total expenditures in 2007–2008.
Burden of CMCs
As shown in Table 2, compared with those without a BH disorder, Maine Medicaid members with BH disorders generally had higher rates of CMCs. The prevalence of CMCs differed among BH groups, with the highest rates occurring among those with intellectual disability/autism/traumatic brain injury (ID/Au/TBI) and those with SMI. All BH groups, except for those with SA-only, had higher numbers of CMCs than those without a BH disorder.
Higher PMPM for all groups with BH disorders were observed as compared with those with no BH disorder (Table 2). PMPM costs varied by the type of BH disorder. BH service costs comprised 40.6% of total expenditures for the whole cohort, and 49% of total PMPM for those with BH disorders, with significant variation among the BH groups. Although the BH costs of those with ID/Au/TBI were almost 70% of their total costs, BH costs for the other groups were less than half of their total PMPM, ranging from 13% for those with non-SMI mental illness to 48% for those with SMI.
Table 3 shows factors related to differences in PMPM costs for each measure of PMPM. For total PMPM costs, all independent variables were significant at P≤0.0001 with only 1 exception, age. Although total PMPM costs increased monotonically with the number of CMCs, the presence of any BH disorder increased total costs more than did the presence of only 1 CMC. Having ID/Au/TBI was the single most costly condition. PMPM cost increases associated with this category of BH disorder dramatically exceeded those attributable to having 4 or more CMCs. Costs related to dual diagnosis mental illness/substance abuse (MI/SA), SMI, or to SA-only were larger than those associated with having 3 or more CMCs. Males had lower costs than females, as did patients without a PCP relationship. Members who received fragmented PCP care incurred higher costs than those who experienced care from only 1 PCP practice.
The strongest influence on BH PMPM costs was BH diagnosis (Table 3). Although CMCs had a smaller impact on BH PMPM costs than on total PMPM, it is noteworthy that BH costs also increased with the number of CMCs. Demographic characteristics such as age, sex, place of residence, or members’ PCP relationships had little effect on BH PMPM costs.
Non-BH and Med/Surg Costs
The non-BH PMPM cost change observed for each of our independent variables was approximately equal to each variable’s total PMPM cost change minus its BH PMPM cost change (Table 3). As with total PMPM, non-BH costs increased with the number of CMCs, and were higher among those with BH disorders. The cost increase associated with SA-only was higher than that for 1 CMC. Costs related to MI/SA were higher than those associated with 2 CMCs, and ID/Au/TBI-related costs were higher than those incurred for members with 3 CMCs.
Medical/surgical costs represented (88%) of non-BH costs and overall, the patterns for these 2 cost measures were alike for most of the independent variables. Med/surg PMPM costs increased most dramatically with the number of CMCs but were also higher for those with BH disorders. Med/surg PMPM cost increases for SA-only, MI/SA and ID/Au/TBI were the same as for those with 1–2 CMC’s. As with non-BH PMPM, med/surg PMPM costs were lower for persons 45 years and older, males, and those not having a PCP relationship. Both non-BH and med/surg PMPM costs were higher for persons residing in nonrural areas and for those experiencing fragmented PCP care. Table 3 also shows that the difference between med/surg costs and non-BH costs was magnified among members with SMI, SA, MI/SA, and ID/Au/TBI. The increased discrepancy between non-BH and med/surg PMPM costs for the subgroups was likely due to their heavier use of services, such as long-term care services, nonemergent transportation, special therapies, and dental care, which were included in non-BH cost, but not in med/surg cost.
Compared with those with no BH disorders, all BH groups had higher rates of utilization of physician services, emergency room, hospitalization, 30-day hospital readmissions, and potentially avoidable hospitalizations (Table 4). Emergency room utilization for medical conditions was 1.7–3 times higher for persons with BH disorders compared with those without. Thirty-day hospital readmission rates were 2- to 10-fold higher and rates of potentially avoidable hospitalization 2–14 times greater in those with BH diagnoses.
Utilization of BH services accounted for 40.6% of total costs for long-term Maine Medicaid members; populations with BH disorders had BH PMPM costs ($597.62) that were, however, less than all their non-BH PMPM costs ($620.73). This finding suggests that initiatives designed to contain costs for Medicaid members with BH disorders will likely be ineffective if they focus primarily on the BH service sector.
Costs in all cost categories, including BH costs, increased with increasing numbers of CMCs, and on average, those with BH disorders had more CMCs than their peers without BH disorders. Thus, the elevated costs associated with BH disorders are in part because people with BH disorders are sicker and need more care. However, after controlling for the number of CMCs, total, med/surg, and non-BH costs remain significantly higher for those with BH disorders than for those without. This observation suggests that to understand why people with BH disorders generate relatively high costs, factors besides their physical health status must be considered.
As noted above, patient factors linked to BH disorders, such as difficulties in self-care and engaging productively with providers, may play an important role in hindering individuals with BH disorders from obtaining timely, cost-effective interventions that would prevent the progression of disease and help contain costs. Aspects of delivery system design may exacerbate the impact of these patient factors. Mental illness, SA, intellectual disabilities, and physical health problems are currently addressed within separate care delivery systems, with none being accountable for utilization, quality, or outcomes across both physical and BH for their priority populations, nor having structures in place that identify and address interactions among multiple medical and psychiatric comorbidities in complex populations.34
Failure of coordination between the behavioral and physical healthcare systems creates additional barriers for patients with BH disorders who must coordinate services in several domains, and who likely face special challenges in navigating even 1 care delivery system. Assuming these inferences regarding the role of patient and system factors are correct, then overall healthcare costs, as well as BH and all non-BH costs, may be reduced through improved integration of behavioral and physical health services. Accountable Care Organizations (ACOs), which are promoted in the ACA of 2010, offer the possibility of improved integration across multiple providers caring for specific populations. Fisher et al35 notes that through integration of services, “ACOs can help reduce unnecessary medical care and improve health outcomes, leading to a decrease in utilization of acute care services.” Other system reforms, for example, CMS Health Homes for Medicaid members with multiple CMCs and Patient Centered Medical Homes also offer the promise of a more holistic, person-centered approach, which involves greater integration between behavioral and physical healthcare services.36
Another key finding of this study was that BH disorders alone can affect medical costs, and that the presence of some BH disorders can have as large an effect on total cost as the presence of multiple CMCs. This observation suggests BH disorders cannot be regarded as equivalent to other types of CMCs for purposes of risk adjusting healthcare costs. Our results are consistent with evaluations of commercially available and commonly utilized risk adjustment methodologies (eg, Adjusted Clinical Groups, Diagnostic Cost Groups), which have found that the systems tended to underestimate costs for persons with BH disorders and overestimate costs for persons without those conditions.37–39
The findings and policy implications reported here for Maine are most applicable to other fee-for-service Medicaid programs that cover optional services, particularly home-based and community-based care. They are also relevant for those states that under the ACA elect to provide Medicaid benefits for all adults <133% of federal poverty levels and to those Medicaid programs that have subpopulations of high-cost users with increased prevalence of multiple CMCs and BH disorders.15 The high rates of potentially avoidable hospitalization, emergency service use, and 30-day readmissions among the BH populations in this study suggest that there are opportunities for Medicaid programs for improving the cost effectiveness of health service delivery for these populations. Whether high utilization results in better outcomes has been a matter of debate, with some research showing better outcomes in certain settings40 and others suggesting that more care may be associated with worse outcomes.41 In addition, those with BH disorders may require specialized interventions to achieve the same disease-specific outcomes as their less complex peers. Research is sparse as to what is required to achieve equivalent clinical outcomes in multimorbid populations, but studies designed to increase physical activity or reduce weight in obese populations with SMI required extensive one-to-one support from trained staff over 12 months or longer to be effective.42,43 Thus, specialized interventions that effectively address the medical issues of those with BH disorders may require adaptations tailored to their complex needs, for example, specialized staff over longer time, likely resulting in additional expenditures.
It is not clear whether the magnitude of cost differences observed in this study would be found in commercially insured or Medicare populations. Although persons in these insured groups may have fewer socioeconomic stressors than Medicaid members, these other insurers have members with chronic disease and comorbid depression, who have been shown to have increased utilization, elevated costs, and worse clinical outcomes.44,45 Thus, the overall pattern of findings observed in this study may have relevance beyond the Medicaid program.
One limitation of this study is that the demographics of Maine did not permit evaluation of the impact of racial or ethnic differences. The finding of higher costs independent of the number of CMCs in the BH cohorts also does not take into account the possibility that at any level of CMC, or for a specific CMC, BH populations may have more severe and complex trajectories, requiring more care to achieve equivalent outcomes. The administrative data used in this study contained only limited information on the personal characteristics of those included in the study population. Although we know that all of our subjects had incomes that were low enough to qualify for Medicaid, no information was available on clinical characteristics (eg, blood pressure, smoking status), capacity for self-care, or the many other factors that might relate to outcomes. Thus, there exists the possibility of endogeneity in relationships identified in our analyses.
In summary, BH disorders have a large impact on all healthcare costs, including costs of medical and other non-BH health services. Although persons with BH disorders have higher numbers of CMCs, their chronic disease burden alone does not entirely explain their higher utilization and costs. Utilization patterns for emergency services, avoidable hospitalizations, and 30-day readmissions suggest opportunities for improving efficiency and cost effectiveness of care for populations with BH disorders. Results also suggest costs might be reduced through better integration of behavioral and medical health services.
The authors would like to acknowledge, AHRQ for supporting this research through our Grant award R21 HS019522-01 and Federal Project Officer, Richard Ricciardi. They would also like to acknowledge the support they were provided by the Maine Department of Health and Human Services, Office of Quality Improvement (OQI), and Office of MaineCare Services for allowing the use of the data for this study and Dr Jay Yoe of OQI for his support and technical assistance with the identification of behavioral health using MaineCare claims data. At the Muskie School, they would like to acknowledge several project staff who participated in this study including lead programmer, Tina Gressani, Project Managers, Anne Conners, and Mark Rubin, Project Assistants, Jenny Mackenzie and Victoria Abbott and Dr John Devlin, Medical Director of the Mattina R. Proctor Diabetes Center at Mercy Hospital, Portland Maine for consultation on diabetes identification in claims data and treatment. They would like to recognize Ayse Akincigil, PhD from Rutgers University for consultation on pharmacological measures used in the larger study. In addition, they would like to thank Andrew Coburn, PhD and Jean Talbot, PhD from the Muskie School for reviewing and critiquing the manuscript.
1. Charlson M, Charlson RE, Briggs W, et al .Can disease management target patients most likely to generate high costs? The impact of comorbidity.J Gen Intern Med. 2007; 22:464–469.
2. Hwang W, Weller W, Ireys H, et al .Out-of-pocket medical spending for care of chronic conditions.Health Aff (Millwood). 2001; 20:267–278.
3. Naessens JM, Stroebel RJ, Finnie DM, et al .Effect of multiple chronic conditions among working-age adults.Am J Manag Care. 2011; 17:118–122.
4. Schneider KM, O’Donnell BE, Dean D .Prevalence of multiple chronic conditions in the United States’ Medicare population.Health Qual Life Outcomes. 2009; 7:82–93.
5. Friedman B, Jiang HJ, Elixhauser A, et al .Hospital inpatient costs for adults with multiple chronic conditions.Med Care Res Rev. 2006; 63:327–346.
6. Lehnert T, Heider D, Leicht H, et al .Review: health care utilization and costs of elderly persons with multiple chronic conditions.Med Care Res Rev. 2011; 68:387–420.
7. Cawthorpe D, Wilkes TC, Guyn L, et al .Association of mental health with health care use and cost: a population study.Can J Psychiatry. 2011; 56:490–494.
8. Sayers S, Hanrahan N, Kutney A, et al .Psychiatric comorbidity and greater hospitalization risk, longer length of stay, and higher hospitalization costs in older adults with heart failure.J Am Geriatr Soc. 2007; 55:1585–1591.
9. Hutter N, Schnurr A, Baumeister H .Healthcare costs in patients with diabetes mellitus and comorbid mental disorders—a systematic review.Diabetologia. 2010; 52:2470–2479.
10. Centorrino F, Mark T, Talamo A, et al .Health and economic burden of metabolic comorbidity among individuals with bipolar disorder.J Clin Psychopharmacol. 2009; 29:595–600.
11. Pan Y, Knapp M, Yeh L, et al .Treatment costs for depression with pain and cardiovascular comorbidities.J Psychiatr Res. 2013; 47:329–336.
12. Arnow B, Blasey C, Lee J, et al .Relationships among depression, chronic pain, chronic disabling pain, and medical costs.Psychiatr Ser. 2009; 60:344–350.
13. Unützer J, Schoenbaum M, Katon W, et al .Healthcare costs associated with depression in medically ill fee-for-service Medicare participants.J Am Geriatr Soc. 2009; 57:506–510.
14. Sokal J, Messias E, Dickerson FB, et al .Comorbidity of medical illnesses among adults with serious mental illness who are receiving community psychiatric services.J Nerv Ment Dis. 2004; 192:421–427.
16. Boyd C, Leff B, Weiss C, et al .Clarifying multmorbidity patterns to improve targeting and delivery of clinical services for Medicaid populations.CHCS Data Brief. 2010; 1–14.
18. Mark TL, Levit KR, Vandivort-Warren R, et al .Changes in US spending on mental health and substance abuse treatment, 1986-2005, and implications for policy.Health Aff (Millwood). 2011; 30:284–292.
20. Druss B, Liping Z, Von Esenwein S, et al .Understanding excess mortality in persons with mental illness.Med Care. 2011; 49:599–604.
21. Muskin P .Major depressive disorder and other medical illness: a two-way street.Curr Psychiatr. 2010; 9:S15–S20.
22. Li C, Barker L, Ford E, et al .Diabetes and anxiety in US adults: findings from the 2006 Behavioral Risk Factors Surveillance Survey.Diabet Med. 2008; 25:878–881.
23. McEvoy J, Meyer J, Goff D, et al .Prevalence of the metabolic syndrome in patients with schizophrenia: baseline results from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) schizophrenia trial and comparison with national estimates from NHANES III.Schizophr Res. 2005; 80:19–32.
24. Cluey S, Cochrane G .Psychological disorder in asthma with poor control and poor adherence to inhaled steroids.Respir Med. 2001; 95:37–39.
25. Ziegelstein R, Fauerback J, Stevens S, et al .Patients with depression are less likely to follow recommendations to reduce cardiac risk during recovery from a myocardial infarction.Arch Int Med. 2000; 160:1818–1823.
26. Lustmann P, Anderson R, Freedland K, et al .Depression and poor glycemic control: a meta-analytic review of the literature.Diabetes Care. 2000; 23:934–942.
29. Ziller EC, Anderson NJ, Coburn AF .Access to rural mental health services: service use and out-of-pocket costs.J Rural Health. 2010; 26:214–224.
30. Koller D, Eisele M, Kaduszkiewicz H, et al .Ambulatory health services utilization in patients with dementia—is there an urban-rural difference? Int J Health Geogr. 2010; 9:59–68.
31. DeVoe JE, Saultz JW, Krois L, et al .A medical home versus temporary housing: the importance of a stable usual source of care.Pediatrics. 2009; 124:1363–1371.
32. Liu CW, Einstadter D, Cebul RD .Care fragmentation and emergency department use among complex patients with diabetes.Am J Manag Care. 2010; 16:413–420.
33. Fleury MJ, Grenier G, Bamvita JM, et al .Mental health service utilization among patients with severe mental disorders.Community Ment Health J. 2011; 47:365–377.
34. Starfield B .Is patient-centered care the same as person-focused care? Perm J. 2011; 15:63–69.
35. Fisher E, McClellan M, Bertko J, et al .Fostering accountable health care: moving forward in Medicare.Health Aff (Millwood). 2009; 28:219–231.
37. Hermann R, Rollins C, Chan J .Risk-adjusting outcomes of mental health and substance-related care: a review of the literature.Harv Rev Psychiatry. 2007; 15:52–69.
38. Ettner S, Frank R, McGuire T, et al .Risk adjustment of mental health and substance abuse payments.Inquiry. 1998; 35:223–240.
39. Rosen A, Loveland S, Anderson J .Diagnostic cost groups (DCGs) and concurrent utilization among patients with substance abuse disorders.Health Serv Res. 2002; 37:1079–1103.
40. Stukel T, Fisher E, Alter D, et al .Association of hospital spending intensity with mortality and readmission rates in Ontario hospitals.JAMA. 2012; 307:1037–1045.
41. Fisher E, Wennberg D, Stukel T, et al .The implications of regional variations in Medicare spending. Part 2: health outcomes and satisfaction with care.Ann Intern Med. 2003; 138:288–298.
42. Bartels S, Pratt S, Aschbrenner K, et al .Clinically significant improved fitness and weight loss among overweight persons with serious mental illness.Psychiatr Serv. 2013; 64:729–736.
43. Daumit GL, Dickerson FB, Wang N-Y, et al .A behavioral weight-loss intervention in persons with serious mental illness.N Engl J Med. 2013; 368:1594–1602.
44. Sherwood A, Blumenthal J, Hinderliter A, et al .Worsening depressive symptoms are associated with adverse clinical outcomes in patients with heart failure.J Am Coll Cardiol. 2011; 57:418–423.
45. Moraska A, Chamberlain A, Shah N, et al .Depression healthcare utilization, and death in heart failure: a community study.Circ Heart Fail. 2013; 6:387–394.
behavioral health disorders; healthcare costs; chronic medical comorbidities; Medicaid
Copyright © 2014 by Lippincott Williams & Wilkins
Highlight selected keywords in the article text.