{"subscriber":false,"subscribedOffers":{}} Prevalence And Spending Associated With Patients Who Have A Behavioral Health Disorder And Other Conditions | Health Affairs

Cookies Notification

This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Find out more.
×

Research Article

Prevalence And Spending Associated With Patients Who Have A Behavioral Health Disorder And Other Conditions

Affiliations
  1. Ken Thorpe is the Robert W. Woodruff Professor and Chair of the Department of Health Policy and Management, Rollins School of Public Health, at Emory University, in Atlanta, Georgia.
  2. Sanjula Jain ( [email protected] ) is a doctoral student in health services research and health policy, Rollins School of Public Health, Emory University.
  3. Peter Joski is a senior associate in the Department of Health Policy and Management, Rollins School of Public Health, Emory University.
PUBLISHED:Free Accesshttps://doi.org/10.1377/hlthaff.2016.0875

Abstract

People with multiple medical conditions are a growing and increasingly costly segment of the U.S. population. Despite the co-occurrence of physical and behavioral health comorbidities, the US health care system tends to treat these conditions separately rather than holistically. To identify opportunities for population health improvement, we examined the treated prevalence of and health care spending on behavioral health disorders, by the number of coexisting physical disorders, among noninstitutionalized adults. The vast majority (85 percent) of spending was attributed to treatment of the physical comorbidities. Only 15 percent was attributed to treatments of the behavioral disorders; of these, a primary diagnosis of depression was most common, seen in 57 percent of the sample. These findings suggest the potential to improve outcomes and reduce spending by applying collaborative care models more broadly. Policies should promote payment and delivery reforms that advance the integration of behavioral health and primary care.

TOPICS

Patients with multiple medical conditions are driving high levels of growth in health care spending and constitute an increasingly costly segment of the US population. Treating people with multiple conditions can cost as much as seven times more than treating those with only one illness. 1 Approximately 86 percent of total health care expenditures are for patients with chronic diseases, and those with multiple conditions have largely contributed to the growth in total health spending. 2,3 Caring for patients becomes complicated when they seek treatment for a wide variety of co-occurring medical conditions. These conditions could range from behavioral disorders to cardiovascular-related diseases and acute care problems. Effective approaches to managing such complexity in primary care settings are needed.

While one in four Americans have multiple medical conditions, research suggests that comorbidities are common not only with physical conditions, but also with behavioral disorders. 48 The co-occurrence of the most commonly observed behavioral health condition—depression—with diseases such as diabetes is well established. 6,9,10 Over 15 percent of American adults have at least one behavioral condition; this inevitably creates overlaps between patient populations with behavioral health disorders and other physical diseases. 10,11 This overlap is to be expected given that behavioral health patients are less likely than patients without behavioral conditions to receive preventive services that often reduce the onset of physical ailments. 11

Background

Despite the large number of people diagnosed with a behavioral disorder, only 14.6 percent of adults used mental health services in 2013. 12 People who are dually eligible for Medicaid and Medicare report the highest rates of mental illness. 13 The high prevalence of co-occurring chronic physical and behavioral health conditions among older Medicare enrollees contributes to even greater spending. 14 Considering the high rates of multiple physical conditions already faced by the elderly, this population is less likely than younger and middle-aged adults to use behavioral health services. 15,16

Health care spending, is at minimum, two times greater for patients diagnosed with a behavioral disorder than for those without. 14,17 The majority of patients with behavioral disorders intentionally seek care in primary care settings, which suggests that a substantial number of these patients receive general medical care at the same time. 18,19 Even though people with behavioral disorders constitute a smaller population than those with physical diseases, the majority of spending for patients with behavioral health conditions is for care not related to behavioral health. 18 Moreover, patients with behavioral health disorders have significantly greater spending for general medical services compared to patients without a disorder. 10,19 Despite evidence that behavioral disorders are associated with comorbid conditions, most care delivery systems do not integrate behavioral with other medical care services. 20,21

A growing body of research has assessed the treated prevalence and costs of co-occurring behavioral and physical health comorbidities. However, these studies have been limited to examining selected conditions within specific populations. 2224 To identify opportunities to improve population health and reduce spending, it is necessary to examine broader populations diagnosed with a behavioral health disorder and other comorbid physical conditions. 10,23 In this study we present new evidence quantifying the national disease burden by examining the treated prevalence and health care spending of the US adult noninstitutionalized population diagnosed with, and treated for a behavioral health disorder—inclusive of mental health and substance-related disorders—and comorbid physical conditions. We were interested primarily in determining how much of total spending among adult behavioral health patients is attributable to treatment for nonbehavioral comorbidities. We also conducted subset analyses for patients with depression to assess treated prevalence rates for diagnosed individuals and spending associated with the most common behavioral disorder among adults. 6,9 Our analyses focused on the total noninstitutionalized adult population, with subset analyses for those enrolled in Medicare, Medicaid, or both (dual eligibles).

Study Data And Methods

Data

Data for our analysis came from the Medical Expenditure Panel Survey (MEPS), pooled from the years 2010–13. MEPS is a nationally representative survey of the US civilian noninstitutionalized population. The survey collects data on self-reported medical conditions, insurance coverage, patient demographics, health services use, and health care spending. The sample was limited to adults ages eighteen and older with at least one behavioral health disorder diagnosis. Behavioral health patients were defined as having at least one Clinical Classifications Software (CCS) code between 650 and 663. The all-adult baseline sample included all payers: Medicaid, Medicare, dual Medicaid-Medicare, Tricare, employer sponsored, other forms of private coverage, and uninsured.

Analysis

The baseline data were divided by insurance category to create several subsets of enrollees. This analysis focuses on Medicaid, Medicare, and dual enrollees because of the disproportionate spending attributed to behavioral health among beneficiaries in these programs. We considered Medicaid beneficiaries as those with Medicaid-sponsored health insurance for twelve months during the year with no simultaneous Medicare coverage during any months of the same year. Medicare beneficiaries were defined as those with Medicare-sponsored health insurance for twelve months during the year with no simultaneous Medicaid coverage during any months during the year. Dual eligibles had twelve months of both Medicaid and Medicare coverage. The four groups of interest—all adults, Medicaid, Medicare, and dual Medicaid-Medicare—were further stratified to create a subset representative of adults with a primary diagnosis of depression when considering all behavioral health conditions, denoted by CCS code 657.

Additional sensitivity analyses (presented in online Appendix B, Exhibits B1–4) 25 were performed using at least one month of coverage data for each of the three insurance programs. “Medicaid 1+ month” consisted of individuals with at least one month of Medicaid coverage during the year and no Medicare coverage during any other months of the same year. “Medicare 1+ month” consisted of individuals with Medicare coverage for at least one month during the year and no Medicaid coverage during any other months. “Dual Medicaid-Medicare 1+ month” consisted of individuals with both Medicaid and Medicare coverage during the same month at any point during the year.

Medical condition data were coded using the International Classification of Diseases , Ninth Revision (ICD-9). The ICD-9 codes were collapsed to three-digit codes and subsequently coded into 259 clinically relevant medical conditions using Clinical Classifications Software codes. 26 These 259 medical conditions were collapsed to 60 medical condition categories.

The baseline data set included records of patients with medical condition codes denoting a behavioral health disorder. We considered only individuals who had evidence of treatment related to a mental or substance-related disorder. Treatment was determined by at least one medical event—office-based visit, emergency department visit, prescribed medication, home health service, or inpatient or outpatient care—associated with a self-reported behavioral health condition. Comorbidity calculations were based on the number of co-occurring physical conditions per person treated for a behavioral disorder in the baseline data set. This procedure was then repeated for the depression subset, calculating treated prevalence and comorbidity rates only for behavioral health patients with a primary CCS code and evidence of treatment for depression.

Total health care spending was determined from the treatment-related event files for office-based visit, emergency department, prescribed medication, home health, and inpatient and outpatient encounters across the continuum of medical care for behavioral health patients. Health care spending included costs related to general medical and behavioral health services for those diagnosed with a mental or substance-related disorder. We then calculated behavioral health–specific spending by summing health service costs exclusive to behavioral health treatment events specified by behavioral disorder CCS codes. For people treated for depression, patient events may include depression and other diagnoses. To examine depression spending among patients, we developed two measures to provide a range of depression-related spending. For average annual total depression-specific health care spending among all adults treated for depression, we determined a lower spending bound ($45.2 billion) using the Bureau of Economic Analysis (BEA) Health Care Satellite Account (HCSA) methodology and an upper spending bound ($46.3 billion) via the Altarum approach. 27 These higher- and lower-range estimates differed by 2.4 percent. Given the relatively small percentage difference across the two methods, our analyses for depression spending were based upon depression identified as the primary diagnosis using the upper-bound Altarum methodology.

Survey weights were adjusted to account for the pooling of four years of data. STATA version 14.0 was used for data analysis. Survey estimation commands were used to adjust for complex survey design of MEPS. Total spending was inflated to National Health Expenditures Personal Health Care values for the respective years. 28 All spending amounts were calculated in 2013 dollars using the gross domestic product deflator.

Limitations

We note a few limitations of our study. Since our sample includes only noninstitutionalized adults, our findings are not representative of the US adult population as a whole. The increase in behavioral health disorder prevalence and receipt of treatment since 2013 suggests that our results may be underestimates of current spending. 29 The practice of medicine suggests that there may be some misallocation, where separate behavioral health spending may be difficult to tease apart from spending related to concurrent behavioral health and physical conditions. We did not account for variation in Medicaid estimates due to differences between state programs, nor did we explicitly delineate private insurance trends. Despite these limitations, our study provides a foundation for further spending analyses.

Study Results

On average, 15.5 percent of US adults were treated for at least one behavioral health disorder during 2010–13. Exhibit 1 highlights the average annual treated prevalence of behavioral health disorders among Medicaid beneficiaries (25.3 percent), Medicare beneficiaries (20.9 percent) and dual Medicaid-Medicare eligibles (38.7 percent). A high degree of comorbidity among people with behavioral disorders was consistent across all groups. Across all three insurance types, more than half of adults treated for a behavioral disorder had four or more comorbid physical conditions. Moreover, nearly 80 percent of Medicare beneficiaries and dual eligibles diagnosed with a behavioral disorder had at least four additional physical comorbidities.

Exhibit 1 Treated prevalence of behavioral health disorders and comorbid physical conditions, 2010–13

All adults age 18+MedicaidMedicareDual Medicaid-Medicare
Overall treated prevalence for behavioral health disorders
Percent of population treated15.5%25.3%20.9%38.7%
Number of individuals treated36,867,3232,834,4598,102,9491,792,808
Distribution of behavioral disorder treated prevalence by number of comorbid conditions
Behavioral health disorder only8.0%9.2%0.9%3.6%
+1 condition11.711.73.53.8
+2 conditions12.210.94.86.8
+3 conditions12.410.47.96.5
+4 or more conditions55.857.882.979.2

SOURCE Authors’ analysis of pooled 2010–13 data from the Medical Expenditure Panel Survey. NOTES Percentages by number of comorbid conditions correspond to noninstitutionalized adult treated prevalence rates. Data in the “All adults” category do not differentiate by payer type.

Exhibit 2 presents treated prevalence trends among people treated for a primary behavioral health diagnosis of depression This subset population of patients treated for depression accounted for 57 percent of those treated for all behavioral health disorders. We calculated that, on average, 8.8 percent of all noninstitutionalized adults were treated annually for depression between 2010 and 2013. Relative to Medicare and Medicaid, dual eligibles had the greatest proportion of adults treated for depression (24.4 percent). Similar to those treated for any behavioral health disorder, we found that a substantial majority of people treated for depression had multiple physical comorbidities. Over 60 percent of adults treated for depression had four or more comorbid physical conditions, with an even greater proportion among Medicare and dual-eligible enrollee populations.

Exhibit 2 Treated prevalence for depression and comorbid physical conditions, 2010–13

All adults age 18+MedicaidMedicareDual Medicaid-Medicare
Overall treated prevalence for depression
Percent of population treated8.8%16.8%11.8%24.4%
Number of individuals treated21,040,8211,877,4634,551,1111,130,319
Distribution of depression treated prevalence by number of comorbid conditions
Depression only6.6%5.7%0.6%3.4%
+1 condition10.210.72.34.1
+2 conditions10.29.54.54.4
+3 conditions12.411.18.05.7
+4 or more conditions60.763.184.782.4

SOURCE Authors’ analysis of pooled data from the 2010–13 Medical Expenditure Panel Survey. NOTES Percentages by number of comorbid conditions correspond to noninstitutionalized adult treated prevalence rates. Data for the “All adults” category do not differentiate by payer type.

For patients treated for any type of behavioral health disorder, the total amount (behavioral and nonbehavioral health expenditures) spent on health care during 2010–13 averaged $672.4 billion ( Exhibit 3 ). Among treated patients, health spending was primarily related to inpatient care (28.4 percent), prescriptions (28 percent), and office-based medical provider visits (21.9 percent). For people treated for depression, the total amount spent on health care averaged $426.5 billion.

Exhibit 3 Average annual health care spending among people treated for depression only and all behavioral health disorders, 2010–13

Average annual spending attributed to treatment for behavioral health conditionsAverage annual spending among people treated for a behavioral health condition
Insurance categoryAnnual total (billions)Per capitaAnnual total (billions)Per capita
Spending for subset of adults treated for depression only
All adults age 18+$46.3$2,202$426.5$20,268
Medicaid5.93,12137.620,027
Medicare8.61,885121.226,631
Dual Medicaid-Medicare4.13,64939.534,923
Spending for all adults treated for a behavioral health disorder
All adults age 18+$101.2$2,745$672.4$18,238
Medicaid12.94,56851.218,073
Medicare21.22,621198.124,450
Dual Medicaid-Medicare11.46,34464.035,686

SOURCE Authors’ analysis of pooled 2010–13 data from the Medical Expenditure Panel Survey. NOTES Total spending includes costs for all health services and medical care received for behavioral and non-behavioral health–related purposes among noninstitutionalized adults treated for a behavioral health disorder. Of the $101.2 billion attributed to treatment for behavioral health conditions, most spending was due to prescription drugs (48.4 percent) and office-based medical provider visits (23.1 percent), whereas of the total $672.4 billion spent for medical care among people diagnosed with a behavioral health condition, spending was primarily related to inpatient care (28.4 percent), prescription drugs (28 percent), and office-based medical provider visits (21.9 percent). Data for the “All adults” category do not differentiate by payer type.

Exhibit 3 outlines behavioral health–related and total health care spending by insurance type. Only 15 percent of total health care spending among adults treated for behavioral disorders ($101.2 billion) was attributed to behavioral health–related care. Of direct behavioral health spending, the majority of expenditures were attributed to prescription drugs (48.4 percent) and office-based visits (23.1 percent). Medicaid spent the greatest proportion of total health care spending on care for behavioral health, yet the least amount ($18,073) per capita. Comparatively, spending for dual Medicaid-Medicare eligibles was nearly twice that amount: $35,686 per capita.

Across the three insurance categories, the highest spending occurred among patients with four or more comorbidities ( Exhibit 4 ). We found that 80.4 percent of average total health spending for all adults treated for a behavioral health disorder, regardless of insurance, was attributed to patients with at least four comorbid conditions. Total spending increased by the number of additional conditions. For patients with one to three comorbidities, incremental spending was relatively small compared to those with four or more conditions. Similar trends were observed for patients treated for depression.

Exhibit 4 Average annual health care spending, by payer and number of comorbid conditions, among people treated for depression only and for all behavioral health disorders, 2010–13

All adults age 18+ ( N = 36,867,323) Medicaid ( n = 2,834,459) Medicare ( n = 8,102,949) Dual Medicaid-Medicare ( n = 1,792,808)
Number of comorbid conditionsDepression onlyAll BH disordersDepression onlyAll BH disordersDepression onlyAll BH disordersDepression onlyAll BH disorders
Spending in billions of US dollars
No additional conditions$ 5.49$ 12.13$ 0.51$ 1.92$ 0.15$ 0.67$ 0.35$ 0.95
+1 condition13.7927.441.202.150.962.660.460.80
+2 conditions18.8539.561.422.571.963.370.442.18
+3 conditions34.3152.612.263.085.858.321.162.62
+4 or more conditions354.02540.6232.2141.51112.27183.137.0757.43
Per capita spending in US dollars
No additional conditions$ 3,967$ 4,121$ 4,740$ 7,383$ 6,188$ 9,075$ 9,144$ 14,735
+1 condition6,4386,3896,0086,4659,2949,3489,82511,764
+2 conditions8,8088,7807,9488,3549,6688,7178,84317,762
+3 conditions13,12911,54110,91510,41916,04013,04318,17622,296
+4 or more conditions27,74326,28627,18325,32729,12727,24739,77740,441

SOURCE Authors’ analysis of pooled 2010–2013 Medical Expenditure Panel Survey Data. NOTES Depression spending values represent a subset of all behavioral health disorder–related health care spending. Data for the “All adults” category do not differentiate by payer type. Sample sizes are numbers of treated individuals. BH is behavioral health.

Unabridged versions of Exhibits 14 are presented in Appendix C, Exhibits C1–4, with confidence intervals for our estimates. 25

We next examined the most common comorbid disease combinations. Pooling data from the period 2010–13, we found that the most common comorbid conditions were hypertension, hyperlipidemia, arthritis, endocrine disorders, and pulmonary disease (Appendix A). 25 Hypertension and hyperlipidemia occurred in greater proportions among treated Medicare beneficiaries and dual eligibles. However, no significant differences in the types of common comorbidities were found between patient populations by insurance type.

Sensitivity analyses were performed to estimate treated prevalence and spending for people with at least one month during the year of Medicaid-only, Medicare-only, or dual Medicaid-Medicare coverage (Appendix B). 25 Despite the increased sample size, 1+ month estimates and observed trends for adults diagnosed with behavioral disorders and comorbid conditions were consistent with our 12-month results.

Discussion

The results of our study underscore the association between increased numbers of physical comorbidities and increased behavioral health spending among US adults treated for one or more behavioral health disorders. The spending attributed to behavioral disorder–related care is small relative to the costs of the additional comorbid physical conditions. Medicaid, Medicare, and dual eligibility account for more than a third of adults treated for a behavioral disorder. Medicare and dual Medicare-Medicaid patients have the greatest proportion of beneficiaries treated for a behavioral disorder and four or more co-occurring physical conditions. Medicare and dual-eligible populations were also found to have a higher concentration of people treated for depression. Our findings confirm that Medicaid spends proportionally more for behavioral health services than Medicare spends. Yet the drivers of high overall spending among people treated for a behavioral disorder are primarily Medicare and dual-eligible patients.

Given that only 15 percent of total health care spending for people diagnosed with a behavioral disorder was attributable to behavioral health–specific care, our data suggest that the remaining 85 percent of spending represents costs related to medical care for physical comorbidities. The most common comorbidities identified among treated behavioral health patients were primarily chronic conditions. Observed comorbidities reiterate the strong correlation between behavioral health disorders and physical disease. This phenomenon will likely present situations in which certain prescribed medical treatments may simultaneously address symptoms related to co-occurring behavioral and general medical conditions for an individual patient.

Our estimates suggest that 63 percent of total health care spending among people treated for a behavioral disorder is for people treated for depression. Unlike with most behavioral health conditions, people are typically diagnosed with depression by a primary care physician, not a behavioral health specialist. 30 While our findings demonstrate the high treated prevalence of physical comorbidities alongside behavioral conditions, other research has conversely found that at least 40 percent of high primary care utilizers have a mental disorder—notably depression and anxiety. 31 Primary care has become the de facto mental health system for depression and other mental health conditions; this fact emphasizes the need for collaborative care models. 32

Evidence-based, whole person–oriented collaborative care delivery models such as accountable care organizations and patient-centered medical homes have the greatest potential to manage the care of patients with mild-to-moderate behavioral comorbidities. 3335 The IMPACT Collaborative Care Program, exclusive to costly Medicare and Medicaid beneficiaries with mental disorders and multiple chronic comorbidities, found that integration of collaborative care–based interventions within primary care settings was effective in significantly improving health status and reducing overall spending for enrolled patients. 36,37 Systematic implementation of evidence-based collaborative care is particularly important for the future of public-sector health care in an era of constrained public-sector budgets, given that Medicare and Medicaid beneficiaries have a higher prevalence of physical and behavioral health comorbidities. 11

More than seventy randomized controlled trials have demonstrated that the collaborative care model applied to patients with common mental disorders such as depression is more effective and cost-effective than usual care across a variety of patient populations. 32,37 Randall Brown and colleagues concluded that consistent face-to-face contact between care coordinators and patients, evidence-based patient education, comprehensive medication management, timely transition plans, and communication between physicians and care coordinators were critical to the success of collaborative care. 38 Further deployment of these collaborative care features to integrate behavioral health services with primary care has the potential to save, at a minimum, $15 billion within Medicaid programs alone. 39 The key to delivery system reform will be to instill a sense of shared accountability between behavioral health and general medical providers across the continuum of care. 31,40,41 The continuation of incentives and payment models that facilitate accountability for clinical outcomes at the point of care will be critical to establishing integrated care across the continuum.

Policy Implications

The Affordable Care Act includes initiatives that, if capitalized upon, would have the potential to transform care coordination and delivery across the continuum of care. 42 Many of these provisions, such as Medicaid expansion, incentives for health homes, and increased access to preventive services, have begun to augment access to behavioral health services. Most recently, the launch of the Center for Medicare and Medicaid Innovation (CMMI) and State Innovation Models have provided states and provider organizations with the opportunity to facilitate multipayer payment and delivery system reforms to promote behavioral health demonstrations. The preliminary success of these demonstrations suggests that future efforts should continue to leverage CMMI models to accelerate the integration of behavioral health services and primary care.

Despite policy efforts, the segmentation of the health care system creates obstacles to care access and quality stemming from factors such as benefit differences and variation in clinical documentation systems for physical and behavioral health care. 43 To address gaps in care, future policies will require greater integration of value-based payment and delivery models to support shared responsibility between primary care and behavioral health care providers across all payers. In addition to collaborative care models, policy makers and provider organizations need to develop specific behavioral health care management processes for primary care. 44 Additional complementary strategies such as telemedicine, Healthcare Employer Data and Information Set (HEDIS) reporting incentives, and all-payer data sources that can support the effective use of behavioral health data in physical health settings, and vice versa, will be critical to the integration of physical and behavioral health care.

Given the magnitude of comorbidity conveyed by our findings, the next step is to evaluate the specifics of corresponding health services utilization. It is important to further analyze the types of prescriptions and health services (for example, physician or emergency department visits) used by the population of patients with co-occurring behavioral health and physical conditions. Another area of interest may be the investigation of “double comorbidities,” to determine how many people are treated for multiple behavioral disorders in addition to multiple physical comorbidities. Patterns in types of commonly observed medical conditions can inform future clinical practice adaptations in the holistic treatment of disease as opposed to individual diagnoses. Future efforts should build upon these findings to determine how different delivery models address the needs of different patient populations. 33 Thus, future studies that assess health care spending and utilization patterns after 2013 will be invaluable for future policy analysis and delivery system innovation.

Conclusion

A substantial majority of spending among patients diagnosed with a behavioral health disorder is attributed to the treatment of nonbehavioral comorbid conditions. Despite evidence of the high spending amounts and comorbidity risk associated with behavioral health conditions, many, if not most, current models are ineffective in providing the necessary population health services. Greater integration of primary and behavioral health services through collaborative care models is critical for meeting the complex health needs of patients treated for behavioral disorders and physical comorbid conditions. Value-based payment and delivery system reform policies that facilitate the adoption of care coordination processes have the potential to both improve health outcomes and greatly reduce associated health care spending.

NOTES

   
Loading Comments...