Abstract
Many older Americans do not receive needed care for mental health and substance use disorders (MHSUD) and there are substantial racial and ethnic disparities in receipt of MHSUD care across the lifespan. Medicare introduced cost-sharing parity for outpatient MHSUD care between 2010-2014, reducing beneficiaries’ out-of-pocket share of MHSUD spending from 50% to 20%. Among Traditional Medicare beneficiaries age 65+, we examined changes in MHSUD use and spending from 2008-2018 for low-income beneficiaries with the cost-sharing reduction vs. a control group of beneficiaries with free care throughout the study period among Black, Hispanic, Asian, and American Indian/Alaska Native vs. White beneficiaries. Among older Medicare beneficiaries, overall use of MHSUD services increased between 2008-2018. For White beneficiaries, MHSUD cost-sharing parity was associated with increased likelihood of having specialty MHSUD visits and medication use and reduced likelihood of having unmonitored MHSUD medication use and MHSUD ED visits and hospitalizations. However, cost-sharing parity was associated with smaller or no gains in MHSUD use for racial/ethnic minority beneficiaries compared with White beneficiaries, thus widening racial and ethnic disparities in MHSUD care.
Introduction
An estimated 14 to 20 percent of older Americans have a mental health condition or substance use disorder (MHSUD),(1) which are associated with increased likelihood of disability and nursing home admission, and higher health care spending compared to those who do not have such disorders.(2–4) Epidemiological data suggest that prevalence of anxiety and mood disorders are similar or more common among Black and Hispanic older adults as compared to White older adults, and that depressive symptoms are more severe for Black and Hispanic older adults.(5–7) However, Black, Hispanic, and Asian adults with mental illness use mental health services at less than two-thirds the frequency of White adults.(8) These disparities persist at older ages, even as most adults age 65+ qualify for Medicare.(7, 9)
In prior studies, poverty and patient out-of-pocket costs have been substantial contributors to underuse of MHSUD services and racial/ethnic disparities in access to specialty care.(10–12) These studies suggest that decreasing cost-sharing for MHSUD services could alleviate these differences. Recent efforts to address limited access to MHSUD treatment have focused on improving insurance coverage for MHSUD care, including requiring parity of coverage for MHSUD with other medical services.
Prior to 2014, Medicare included a mental health treatment reimbursement limitation that resulted in higher cost-sharing for MHSUD services versus medical and surgical care (50% vs. 20% coinsurance). The Medicare Improvements for Patients and Providers Act gradually introduced cost-sharing parity for MHSUD services, reducing coinsurance from 50% in 2009 to 45% in 2010-2011, 40% in 2012, 35% in 2013, and 20% in 2014 onward.
In prior studies, this policy change was not associated with increases in outpatient MHSUD visits overall among higher income beneficiaries or among lower-income beneficiaries with serious mental illness, most of whom qualified for Medicare before age 65 due to disability.(13, 14) There is no existing evidence on the effects of this policy on older, low-income adults, or on racial/ethnic disparities in use of MHSUD services. Prior studies on the effects of cost-sharing reductions for other types of services on racial/ethnic disparities are mixed, with some studies finding attenuation of disparities,(15) and others finding no change or even exacerbations of disparities.(16, 17)
We therefore investigated the association between implementation of MHSUD cost-sharing parity in Medicare and changes in MHSUD care use and spending among lower-income beneficiaries age 65+ years old by race/ethnicity to assess how this policy impacted mental health care disparities.
Methods
Study population and data sources
This study uses a 50% sample of Medicare claims data, 2008-2018, for non-institutionalized, traditional Medicare beneficiaries, age 65+ with an original reason for entitlement of age (i.e., beneficiaries qualifying due to disability or end-stage renal disease were excluded) who had Parts A, B, and D coverage and received Part B or D low-income subsidies in each year.
To identify the effects of cost-sharing parity, we compared changes in outcomes before and after the policy change for beneficiaries exposed to cost-sharing parity vs. beneficiaries with no cost-sharing (free care) because they received Part B cost-sharing subsidies (the federal income limit for these subsidies is 100% of the federal poverty level, FPL). To make the treatment group more closely resemble the control group, the treatment group was limited to low-income beneficiaries who received other subsidies (e.g., Part B premium or Part D subsidies only with income limits between 100-135% FPL; Appendix Exhibit A1).(18) For both groups, Medicare Part B covers outpatient mental health services, including psychiatric evaluations, psychotherapy, medication management, and partial hospitalizations.
Beneficiaries could enter the study cohort (e.g., if they newly received subsidies or Medicare) or exit the cohort (due to death, switching to Medicare Advantage, or no longer receiving subsidies) in each year. Because we measured annual outcomes, we required beneficiaries to have the same subsidy status for the entire year. We identified a coding inconsistency in October 2017 for many beneficiaries with full Part D low-income subsidies; these beneficiaries remained in the sample. Beneficiaries living in Florida were also excluded because of large decreases in MHSUD visits, 2008-2012, that were associated with fraud enforcement among Community Mental Health Centers and unrelated to the policy change.
The Mass General Brigham Institutional Review Board approved the study.
Race and ethnicity
We used race and ethnicity codes available in the Medicare Master Beneficiary Summary File, which include seven mutually exclusive categories: American Indian/Alaska Native, Asian/Pacific Islander, Black (or African-American), Hispanic, Non-Hispanic White (hereafter, White), Other, and Unknown. These data are largely based on Social Security Administration data and an imputation algorithm that uses beneficiaries’ first and last names to identify additional beneficiaries who are likely to be Hispanic or Asian.(19, 20)
Outcomes
We assessed four types of outcomes: (1) outpatient MHSUD visits (overall, with specialists, and with primary care providers [PCPs]), (2) fills for MHSUD prescription medications, (3) acute MHSUD events (ED visits and hospitalizations), and (4) MHSUD treatment spending. MHSUD visits included outpatient encounters with ICD-9-CM diagnosis codes of 291-319 (or corresponding ICD-10-CM codes), consistent with the application of the mental health treatment reimbursement limitation. We excluded claims with primary diagnoses of Alzheimer’s Disease or other related disorders (e.g., 290.X and 331.0) because the reimbursement limitation did not apply to medical management of these conditions. For visits for which multiple services were billed, only the services billed with MHSUD diagnoses were subject to the reimbursement limitation; thus, the implementation of cost-sharing parity could have been less apparent to beneficiaries for visits with PCPs versus specialists where services could be rendered for both MHSUD and non-MHSUD conditions.
MHSUD visits with specialists included psychiatrists, psychiatric nurse practitioners, psychologists, social workers, and counselors. MHSUD visits with PCPs included physicians with a specialty of general medicine, family medicine, and internal medicine, nurse practitioners, and physician assistants.
We examined whether beneficiaries filled any MHSUD medications in each year using Part D claims because receipt of medications could reflect access to outpatient treatment, particularly for primary care visits where MHSUD diagnoses are less consistently coded.(21) We assessed changes in acute MHSUD events, including emergency department (ED) visits and hospitalizations with a primary MHSUD diagnosis, as increased access to outpatient care could be associated with reductions in adverse clinical events.
Lastly, we assessed changes in outpatient, inpatient, pharmacy, and total MHSUD treatment spending based on total costs (Medicare spending plus patient out-of-pocket costs). Spending outcomes were converted to 2018 dollars using the Gross Domestic Product (GDP) implicit price deflator.(22)
Secondary outcomes included new MHSUD visits, defined as visits without an MHSUD visit in the last 12 months. We also assessed if beneficiaries receiving MHSUD medications had no MHSUD visits in the 12 months before or after these medication fills as an indicator of poor outpatient monitoring (“unmonitored medication use”).
Analysis
We calculated the unadjusted proportion of beneficiaries by race/ethnicity with any MHSUD visits, MHSUD medication fills, or acute care events in each year, and mean annual MHSUD spending.
We used an event study to plot difference-in-differences within each racial/ethnic group for those with the cost-sharing reduction vs. free care in each study year compared with the pre-policy year of 2009. These models included indicators for whether the beneficiary had the cost-sharing reduction (vs. free care), year (vs. 2009), interactions between policy exposure and year, beneficiary fixed effects to account for time invariant differences across the comparison groups, and annually updated covariates, including whether the beneficiary lived in a rural ZIP code (based on USDA rural-urban commuting codes 4-10), whether the beneficiary was attributed to an accountable care organization (ACO), state of residence, state/year (interaction) trends, and indicators for whether the beneficiary had 20 common physical health diagnoses as of the end of the prior year based on CMS Chronic Condition Warehouse criteria.
Lastly, we examined whether the associations between the policy change and outcomes differed for minority vs. White beneficiaries by including an interaction between race/ethnicity and a difference-in-difference estimator: i.e., interaction between parity exposure and the policy phase-in (2010-2013) and post-policy (2014-2018) periods vs. pre-policy (2008-2009). For all outcomes, we used linear models adjusted for the same covariates as above and adjusted for multiple comparisons across racial/ethnic groups and time periods within each outcome using the Benjamini-Hochberg False Discovery Rate correction.(23)
Sensitivity analyses
An increasing proportion of beneficiaries left Traditional Medicare for Medicare Advantage (MA) during the study (and thus exit the study cohort); thus, the sample size decreased over time. We observed higher levels of MA departures between 2013-2015 among those with free care vs. the cost-sharing reduction, particularly for Hispanic and Asian beneficiaries. To address concerns that differential selection into MA could bias our results (e.g., if relatively healthier beneficiaries choose MA), we conducted sensitivity analyses that excluded counties with high levels of MA transitions between 2013-2015 (>15% of beneficiaries moving to MA in 2013-2014 or 2014-2015 among 8 states). Patterns of MA enrollment were more comparable by race/ethnicity across groups after these exclusions (Appendix Exhibit A2).(18) We did not observe differential enrollment into MA for beneficiaries with MHSUD use vs. not for those with the cost-sharing reduction vs. free care.
Limitations
Because this study includes only two years of pre-policy data, we do not have power to assess whether pre-policy trends are parallel. The event study plots suggest that there were few differences in 2008 vs. 2009 for those with cost-sharing reduction vs. free care across outcomes and racial/ethnic groups. There were, however, declines in MHSUD visits with PCPs for beneficiaries with the cost-sharing reduction vs. free care pre-policy and it is possible that our findings of relative decreases in MHSUD visits with PCPs with the policy change reflect a continuation of these trends.
We cannot identify beneficiaries with a clinical need for MHSUD care, although underuse of MHSUD care among older adults is a persistent concern.(24) We also do not know if beneficiaries had supplemental Medigap or employer coverage that covers Medicare cost-sharing; however this is unlikely among the lower-income beneficiaries in this study. We also do not have data on care that was not billed to Medicare, such as care through the Veterans Administration or care from alternative healers.(25) The sample of AI/AN beneficiaries was the smallest among the groups in this study, which limits our ability to detect differences for AI/AN beneficiaries. This study focuses on low-income beneficiaries, which limits generalizability to higher-income beneficiaries. In addition, findings might not be generalizable to beneficiaries not enrolled in Part D, particularly those with no or less generous drug coverage (about 12% of Medicare beneficiaries in 2018).(26) Lastly, this is an observational study and there could be residual confounding.
Results
The study included 286,276 beneficiaries with the cost-sharing reduction and 734,280 with free care in 2008 (Exhibit 1). In 2008, White beneficiaries with the cost-sharing reduction vs. free care had slightly older age and were more likely to be female; Black, Hispanic, and Asian beneficiaries with the cost-sharing reduction were younger, on average, and less likely to be female compared with their counterparts with free care. As expected, those with free care were more likely to have diagnoses for 20 common physical conditions (Appendix Exhibit A3).(18) There was a greater proportion of White beneficiaries among those with the cost-sharing reduction vs. free care (73% vs. 46% in 2008, Exhibit 1). The cost-sharing reduction group included 15% Black, 9% Hispanic, 2% Asian, and 1% AI/AN beneficiaries compared with 16% Black, 20% Hispanic, 15% Asian, and 1% AI/AN beneficiaries with free care in 2008. Beneficiaries with the cost-sharing reduction vs. free care had lower mean CMS-Hierarchical Condition Category (HCC) risk scores, were less likely to live in a low SES ZIP code and were more likely to live in a rural ZIP code (except AIAN beneficiaries) and to be attributed to an ACO in 2018 (except Hispanic and AIAN beneficiaries) (p<.05).
Exhibit 1.
Characteristics of elderly low-income Medicare beneficiaries exposed and unexposed to the MHSUD cost-sharing reduction by race/ethnicity in 2008 and 2018
| 2008 | Total | White | Black | Hispanic | Asian | AI/AN |
|---|---|---|---|---|---|---|
| Cost-sharing Reduction | ||||||
| Number | 286,276 | 208,462 | 43,419 | 24,641 | 5,662 | 2,440 |
| Mean Age | 77.1** | 77.7** | 76.4** | 74.3** | 73.5** | 75.1** |
| % Female | 71%** | 74%** | 71%** | 58%** | 53%** | 67% |
| % in Low SES ZIP | 26%** | 19%** | 48%** | 53%** | 16%** | 47%** |
| % in Rural ZIP | 35%** | 40%** | 23%** | 18%** | 6%** | 71% |
| Mean CMS-HCC score | 1.02** | 1.06** | 0.98** | 0.83** | 0.71** | 0.97** |
| Free Care | ||||||
| Number | 734,280 | 337,366 | 119,224 | 148,782 | 108,791 | 8,436 |
| Mean Age | 77.2 | 77.5 | 77.6 | 76.3 | 77 | 76.1 |
| % Female | 70% | 73% | 77% | 66% | 62% | 68% |
| % in Low SES ZIP | 37% | 22% | 51% | 62% | 32% | 58% |
| % in Rural ZIP | 24% | 35% | 24% | 12% | 2% | 71% |
| Mean CMS-HCC score | 1.24 | 1.32 | 1.28 | 1.16 | 1.05 | 1.14 |
| 2018 | Total | White | Black | Hispanic | Asian | AI/AN |
|
| ||||||
| Cost-sharing Reduction | ||||||
| Number | 235,870 | 158,725 | 35,484 | 26,579 | 8,779 | 2,694 |
| Mean Age | 76.0** | 76.4** | 76.2** | 74.5** | 73.7** | 75.2 |
| % Female | 66%** | 68%** | 68%** | 56%** | 53%** | 66% |
| % in Low SES ZIP | 19%** | 12%** | 37%** | 39%** | 11%** | 46%** |
| % in Rural ZIP | 33%** | 39%** | 22%** | 17%** | 7%** | 67% |
| % in ACO | 28%** | 29%** | 30%** | 22% | 23%** | 7%* |
| Mean CMS-HCC score | 1.14** | 1.2** | 1.15** | 0.93** | 0.82** | 1.13** |
| Free Care | ||||||
| Number | 671,100 | 297,721 | 91,867 | 136,648 | 114,173 | 8,532 |
| Mean Age | 76.3 | 76 | 75.9 | 76.4 | 77.7 | 75.1 |
| % Female | 65% | 66% | 67% | 65% | 63% | 65% |
| % in Low SES ZIP | 27% | 14% | 41% | 48% | 22% | 53% |
| % in Rural ZIP | 21% | 33% | 21% | 14% | 2% | 69% |
| % in ACO | 24% | 25% | 27% | 22% | 22% | 6% |
| Mean CMS-HCC score | 1.34 | 1.42 | 1.42 | 1.26 | 1.2 | 1.29 |
Source: Authors’ analysis of Medicare claims data
Notes: MHSUD stands for Mental Mealth and Substance Use Disorders. CMS-HCC stands for the Centers for Medicare and Medicaid Services Hierarchical Condition Category. ACO stands for Accountable Care Organization. P-value levels indicate characteristics for which the cost-sharing reduction group differs significantly from the free care group within each racial/ethnic group and year.
p<.05
Consistent with the assumptions underlying our approach, we find that although there were differences in observable characteristics across the racial/ethnic groups at baseline, directional changes within each racial/ethnic group over time (e.g., 2018 vs. 2008) were similar for those with and without the cost-sharing reduction.
Trends in MHSUD Use and Spending
The proportion of beneficiaries with one or more annual MHSUD outpatient visit increased between 2008 and 2018 across all groups (Exhibit 2; Appendix Exhibit A4 includes all years) (18), with higher use among those with free care vs. the cost-sharing reduction. In both groups, White beneficiaries were more likely to have an MHSUD visit vs. minority beneficiaries before and after the policy change. Patterns were similar for MHSUD visits with specialists and PCPs; among those with the cost-sharing reduction, about twice as many beneficiaries had MHSUD visits with PCPs vs. specialists in 2008. Trends were also similar for new MHSUD visits, which comprised about two-thirds of all MHSUD visits in 2009 (Appendix Exhibit A5).(18)
Exhibit 2.
Percentage of beneficiaries with annual MHSUD Visits and Medication Fills by cost-sharing status in 2008 and 2018
| Year | White | Black | Hispanic | Asian | AI/AN | |
|---|---|---|---|---|---|---|
| Any MHSUD Outpatient Visit | ||||||
|
| ||||||
| Cost-sharing Reduction | 2008 | 6% | 4% | 4% | 3% | 6% |
| 2018 | 11% | 6% | 6% | 4% | 8% | |
| Free Care | 2008 | 11% | 7% | 7% | 5% | 8% |
| 2018 | 17% | 11% | 10% | 8% | 10% | |
|
| ||||||
| MHSUD Visit with a Specialist | ||||||
|
| ||||||
| Cost-sharing Reduction | 2008 | 2% | 1% | 1% | 1% | 2% |
| 2018 | 3% | 2% | 2% | 1% | 2% | |
| Free Care | 2008 | 5% | 3% | 3% | 2% | 2% |
| 2018 | 7% | 4% | 3% | 2% | 3% | |
|
| ||||||
| MHSUD Visit with a PCP | ||||||
|
| ||||||
| Cost-sharing Reduction | 2008 | 4% | 2% | 3% | 2% | 4% |
| 2018 | 8% | 4% | 4% | 3% | 5% | |
| Free Care | 2008 | 6% | 4% | 4% | 3% | 6% |
| 2018 | 11% | 7% | 7% | 5% | 8% | |
|
| ||||||
| Any MHSUD Medication Fill | ||||||
|
| ||||||
| Cost-sharing Reduction | 2008 | 30% | 17% | 18% | 13% | 23% |
| 2018 | 37% | 19% | 20% | 14% | 23% | |
| Free Care | 2008 | 41% | 25% | 29% | 27% | 26% |
| 2018 | 45% | 27% | 30% | 25% | 27% | |
Source: Author’s analysis of Medicare claims data
Notes: MHSUD stands for Mental Mealth and Substance Use Disorders. AI/AN stands for American Indian/Alaska Native. Annual trends from 2008-2018 are available in Appendix Exhibit A4.
Thirty percent of White beneficiaries with the cost-sharing reduction had any MHSUD medication use in 2008 vs. 17% of Black, 18% of Hispanic, 13% of Asian, and 23% of AI/AN beneficiaries (Exhibit 2). Among those filling MHSUD medications, a large proportion had unmonitored use: range from 80% among AI/AN beneficiaries to 87% for Asian beneficiaries in 2008 among those with the cost-sharing reduction; this rate decreased over time (Appendix Exhibit A5).(18)
Among all racial/ethnic groups, White beneficiaries with the cost-sharing reduction were most likely to have an ED visit (0.7%) or hospitalization (0.4%) with a primary MHSUD diagnosis in 2008 (Appendix Exhibit A6) (18). Rates were lowest among Asian beneficiaries.
Among beneficiaries with the cost-sharing reduction, White beneficiaries had the highest levels of mean MHSUD spending ($284) compared with a low of $75 among Asian beneficiaries (Appendix Exhibit A7).(18) Total MHSUD treatment spending trended downward, largely due to reductions in prescription drug spending.
Annual changes in outcomes associated with MHSUD parity by race/ethnicity
Exhibit 3 presents the event study plot for MHSUD visits with specialists, i.e., the adjusted relative change in likelihood of having a visit in each year vs. 2009 for beneficiaries with the cost-sharing reduction vs. free care by racial/ethnic group. For White and Black beneficiaries, the cost-sharing reduction was associated with relative increases in the likelihood of having a specialist visit, with larger and earlier increases for White vs. Black beneficiaries. There were no significant changes, however, for Hispanic, Asian, and AI/AN beneficiaries. For MHSUD visits with PCPs, the cost-sharing reduction was associated with relative decreases in visit rates for most racial/ethnic groups (Appendix Exhibit A8) (18). In sum, the cost-sharing reduction was associated with relative increases in the percentage of White beneficiaries with any MHSUD visit in 2016-2018 vs. 2009, but relative decreases for Black, Hispanic, and Asian beneficiaries (Appendix Exhibit A8) (18).
EXHIBIT 3. Difference-in-differences results: cost-sharing reduction vs. free care compared with the pre-policy year of 2009, by race/ethnicity, through 2018.
Source: Authors’ analysis of Medicare claims data
Notes: MHSUD stands for Mental Health and Substance Use Disorders. The shaded area indicates the policy implementation period when cost-sharing was gradually reduced from 50% to 20%. The event study plots difference-in-differences in outcomes for those with the cost-sharing reduction vs. free care in each year vs. 2009 stratified by racial/ethnic group. The model included indicators for whether the beneficiary was exposed to the cost-sharing reduction (vs. free care), year (vs. 2009), interactions between policy exposure and year, beneficiary fixed effects, and annually updated covariates (i.e., rural ZIP code, whether the beneficiary was attributed to an accountable care organization, state of residence, state by year trends, and indicators for 20 physical health diagnoses). See Appendix Exhibit A8 for event study plots with confidence intervals for all outcomes and Appendix Exhibit A13 for full model results.
The cost-sharing reduction was associated with increases in the percentage of White beneficiaries that filled any MHSUD medication in each year vs. 2009. Changes for other racial/ethnic groups were largely not significant. For Black, Hispanic, and White beneficiaries, the cost-sharing reduction was associated with relative decreases in percentage with MHSUD ED visits and hospitalizations in most post-policy years vs. 2009 (Appendix Exhibit A8) (18).
Differences in MHSUD Parity effects by race/ethnicity
MHSUD service use
For White beneficiaries, the cost-sharing reduction was not associated with mean changes in overall MHSUD visits in the policy phase-in (2010-2013) or post-policy (2014-2018) periods (Exhibit 4). However, the percentage with MHSUD specialist visits increased during policy phase-in and post-policy vs. pre-policy (e.g., +0.46pp post-policy, p<0.001; a 21.9% increase) vs. beneficiaries with free care. Conversely, there were relative decreases in MHSUD visits with PCPs for White beneficiaries with the cost-sharing reduction (e.g., −0.31pp post-policy, p=0.001; a 7.8% decline, Exhibit 4).
Exhibit 4.
Change in percentage of Medicare beneficiaries with annual MHSUD service use for racial/ethnic minority vs. White beneficiaries with the MHSUD cost-sharing reduction vs. free care during policy phase-in (2010-2013) and post-policy (2014-2018) vs. pre-policy (2008-2009)
| White | Black | Hispanic | Asian | AI/AN | |
|---|---|---|---|---|---|
| Overall MHSUD Visits | Estimate | Estimate | Estimate | Estimate | Estimate |
| Pre-policy | 6.3% | 4.0% | 4.5% | 2.6% | 5.9% |
| 2010-2013 | −0.06 | −0.30* | −0.48*** | 0.13 | 0.67 |
| 2014-2018 | 0.16 | −0.71**** | −1.09** | −0.94*** | −0.01 |
| MHSUD Visits with Specialists | |||||
| Pre-policy | 2.1% | 1.2% | 1.3% | 0.9% | 1.7% |
| 2010-2013 | 0.15**** | −0.09 | −0.25*** | −0.28** | −0.3 |
| 2014-2018 | 0.46**** | −0.29*** | −0.53**** | −0.61**** | −0.83** |
| MHSUD Visits with PCPs | |||||
| Pre-policy | 4.0% | 2.5% | 2.9% | 1.6% | 4.1% |
| 2010-2013 | −0.2** | −0.12 | −0.16 | 0.22 | 1.04* |
| 2014-2018 | −0.31*** | −0.32* | −0.51** | −0.39 | 0.66 |
| MHSUD Medication Fill | |||||
| Pre-policy | 29.8% | 16.7% | 18.2% | 12.9% | 22.7% |
| 2010-2013 | 0.24** | −0.33 | −0.58** | 0.19 | −0.95 |
| 2014-2018 | 0.95**** | −0.77** | −1.54**** | 0.07 | −1.21 |
| MHSUD ED Visit | |||||
| Pre-policy | 0.7% | 0.6% | 0.4% | 0.1% | 0.7% |
| 2010-2013 | −0.08** | −0.02 | 0.06 | 0.21** | −0.02 |
| 2014-2018 | −0.08** | −0.09 | 0.00 | 0.17* | 0.06 |
| MHSUD Hospitalization | |||||
| Pre-policy | 0.4% | 0.3% | 0.1% | 0.1% | 0.2% |
| 2010-2013 | −0.10**** | 0.01 | 0.05 | 0.1* | 0.10 |
| 2014-2018 | −0.12**** | 0.02 | 0.00 | 0.04 | 0.08 |
Source: Authors’ analysis of Medicare claims data
Notes: MHSUD stands for Mental Mealth and Substance Use Disorders. The pre-policy row presents the percentage of beneficiaries with use in 2008. Estimates for White show results for parity vs. free care by time period vs. 2009 (difference-in-differences). Estimates for races other than White show difference-in-differences estimate for each racial/ethnic group vs. White. P-value levels indicate when a result is significantly different from 0 after applying the Benjamini—Hochberg False Discovery Rate correction. Models also include beneficiary fixed effects and adjust for annually updated covariates (i.e., rural ZIP code, whether the beneficiary was attributed to an accountable care organization, state of residence, state*year trends, and indicators for 20 physical health diagnoses). See Appendix Exhibit A14 and A15 for full model results.
p<0.1,
p<0.05,
p< 0.01,
p<0.001
Compared with White beneficiaries, changes for the cost-sharing reduction vs. free care groups in the likelihood of having any MHSUD visit in the post- vs. pre-policy period were significantly lower for Black, Hispanic, and Asian beneficiaries. Findings were similar for MHSUD visits with specialists. Black and Hispanic vs. White beneficiaries with the cost-sharing reduction vs. free care also experienced greater declines in MHSUD visits with PCPs in the post- vs. pre-policy period (Exhibit 4). In secondary analyses, the policy change was largely not associated with changes in the likelihood of having new MHSUD visits (Appendix Exhibit A9).(18)
The cost-sharing reduction was associated with increases in the relative percentage of beneficiaries filling any MHSUD medication for White beneficiaries (e.g., +0.95pp post-policy, p<0.001; a 3.2% increase, Exhibit 4), and relative reductions in the percentage of beneficiaries with unmonitored MHSUD medication use (Appendix Exhibit A9) (18). For Black and Hispanic beneficiaries, changes in the percentage filling MHSUD medications with the cost-sharing reduction vs. free care in the post-vs. pre-policy periods significantly lagged behind that of White beneficiaries (Exhibit 4).
Cost-sharing parity was associated with decreases in MHSUD ED visits and hospitalizations for White beneficiaries (e.g., for ED visits: −0.08pp post-policy, p=0.03; an 11.4% reduction, Exhibit 4). For Asian vs. White beneficiaries, the cost-sharing reduction was associated with relative increases in MHSUD ED visits during policy phase-in vs. pre-policy period (+0.21pp, p=0.01, Exhibit 4).
MHSUD Spending
The cost-sharing reduction was not associated with changes in annual MHSUD outpatient spending (Appendix Exhibit A10) (18). For White beneficiaries, the cost-sharing reduction was associated with relative increases in MHSUD pharmacy spending, and relative decreases in MHSUD inpatient and total spending.
Changes in MHSUD pharmacy spending for racial/ethnic minority beneficiaries associated with the cost-sharing reduction were smaller compared with White beneficiaries. For Hispanic and Asian beneficiaries, changes in MHSUD inpatient spending associated with the cost-sharing reduction increased relative to White beneficiaries, as did changes in total MHSUD spending for Asian beneficiaries in the policy implementation period.
Findings were robust in sensitivity analyses that excluded counties with higher transitions from Traditional Medicare to Medicare Advantage (Appendix Exhibits A11 and A12).(18)
Discussion
Medicare implemented parity for outpatient MHSUD services between 2010-2014 and reduced beneficiaries’ cost-sharing for MHSUD care to the same level as for medical care for the first time since the program’s inception. Among low-income and elderly White beneficiaries, these reductions in MHSUD cost-sharing were associated with increases in specialty visits, greater MHSUD medication use, fewer acute care events (ED visits and hospitalizations), and lower annual total MHSUD spending.
For White beneficiaries, relative rates of MHSUD specialty visits increased and primary care visits decreased with the policy change. MHSUD diagnoses could be inconsistently coded by primary care providers, which could have mitigated the effect of the policy change on primary care vs. specialty visits. However, the policy was also associated with increases in the receipt of MHSUD medications and reductions in unmonitored MHSUD medication fills for White beneficiaries, in contrast to prior studies that have found increased psychiatric prescribing without a corresponding specialty visit or diagnosis.(21)
Racial and ethnic minority beneficiaries, however, were less likely to receive MHSUD outpatient treatment compared with White beneficiaries prior to the policy change, and cost-sharing parity was associated with smaller or no gains in MHSUD use compared with White beneficiaries, thus widening disparities. For Asian beneficiaries, the cost-sharing reduction was associated with some relative increases in acute MHSUD events and spending compared with White beneficiaries.
These findings raise critical questions about why the benefits of this policy accrued more to White vs. minority beneficiaries. In contrast to evidence from prior parity policies in private health plans which has found a limited impact on MHSUD care use and expenditures,(27, 28) parity in Traditional Medicare was associated with some gains in MHSUD use and potential improvements in quality among elderly White beneficiaries. Traditional Medicare does not employ utilization management techniques to constrain use and spending, which could have blunted prior parity efforts.(29) However, while the cost-sharing reduction improved the affordability of MHSUD outpatient treatment for all beneficiaries, it did not address other potential barriers to treatment that could disproportionately impact racial and ethnic minority beneficiaries.
For example, historical experiences of disrimination could continue to limit use of services for minority beneficiaries.(30, 31) Among individuals with high levels of depressive symptoms, 37% of Black and 21% of Latino individuals reported experiencing discrimination from their providers because of their race/ethnicity, or language or accent.(32) Having limited English proficiency has also been associated with lower detection of mental health need and service use.(33–35) About three-quarters of Medicare beneficiaries with limited English proficiency are Hispanic or Asian.(36) Our findings for Asian beneficairies are consistent with prior work that has found that disparities in MHSUD use have been wider for Asian vs. White adults compared with other racial/ethnic groups, and that Asians may be more likely to delay reporting psychiatric symptoms until they are severe.(37)
Prior studies have found worse health and health insurance literacy among Black and Hispanic vs. White individuals, which could also limit responses to the parity policy.(38, 39) Minority beneficiaries might have also been more likely to face cost-related barriers or perceived barriers to MHSUD care even after policy implementation. Although levels of wealth vary widely for White vs. non-White households,(40) beneficiaries in this study met income and asset tests for Medicare subsidies, thus reducing these differences. Moreover, racial/ethnic differences in use were also persistent among the group with free care, suggesting that out-of-pocket costs may not be the primary driver of these differences.
Prior studies have found that Black and Hispanic individuals are more likely to live in areas with a greater supply of mental health providers,(41) but that living in such areas is associated with lower rates of initiation of mental health care.(42) In prior analyses, we did not find variation in the overall effects of the cost-sharing reduction for elderly beneficiaries living in areas with greater vs. lower supply of psychiatrists.(43) However, we have not assessed whether increases in use by White beneficiaries crowded-out the availability of specialty MHSUD providers for minority beneficiaries or investigated the availability of culturally competent providers.(44)
Conclusion
Use of MHSUD services increased over time in this cohort of low-income and elderly Medicare beneficiaries. While MHSUD cost-sharing parity in Medicare was associated with increased receipt of outpatient MHSUD treatment among White beneficiaries, use increased less among minority beneficiaries in the five years after full parity implementation, exacerbating existing disparities. Although parity reduced beneficiaries’ MHSUD out-of-pocket costs, attention to other potential structural barriers to MHSUD care for older adults from racial and ethnic minority communities is critically needed.
Supplementary Material
Contributor Information
Vicki Fung, Massachusetts General Hospital and Harvard University, Boston, Massachusetts.
Mary Price, Massachusetts General Hospital and Harvard University.
Alex McDowell, Massachusetts General Hospital and Harvard University.
Andrew A. Nierenberg, Massachusetts General Hospital and Harvard University.
John Hsu, Massachusetts General Hospital and Harvard University.
Joseph P. Newhouse, Harvard University.
Benjamin Lê Cook, Cambridge Health Alliance, Cambridge, Massachusetts, and Harvard University.
References
- 1.Institute of Medicine. The Mental Health and Substance Use Workforce for Older Adults: In Whose Hands? . Washington, DC; 2012. [PubMed] [Google Scholar]
- 2.Andrews AO, Bartels SJ, Xie H, Peacock WJ. Increased risk of nursing home admission among middle aged and older adults with schizophrenia. Am J Geriatr Psychiatry. 2009;17(8):697–705. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Barry LC, Murphy TE, Gill TM. Depressive symptoms and functional transitions over time in older persons. Am J Geriatr Psychiatry. 2011;19(9):783–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bartels SJ, Clark RE, Peacock WJ, Dums AR, Pratt SI. Medicare and medicaid costs for schizophrenia patients by age cohort compared with costs for depression, dementia, and medically ill patients. Am J Geriatr Psychiatry. 2003;11(6):648–57. [DOI] [PubMed] [Google Scholar]
- 5.Pickett YR, Bazelais KN, Bruce ML. Late-life depression in older African Americans: a comprehensive review of epidemiological and clinical data. Int J Geriatr Psychiatry. 2013;28(9):903–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Reynolds K, Pietrzak RH, El-Gabalawy R, Mackenzie CS, Sareen J. Prevalence of psychiatric disorders in U.S. older adults: findings from a nationally representative survey. World Psychiatry. 2015;14(1):74–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Vyas CM, Donneyong M, Mischoulon D, Chang G, Gibson H, Cook NR, et al. Association of Race and Ethnicity With Late-Life Depression Severity, Symptom Burden, and Care. JAMA Netw Open. 2020;3(3):e201606. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Substance Abuse and Mental Health Services Administration. Racial/Ethnic Differences in Mental Health Service Use among Adults. Rockville, MD: Substance Abuse and Mental Health Services Administration; 2015. [Google Scholar]
- 9.Jimenez DE, Cook B, Bartels SJ, Alegría M. Disparities in mental health service use of racial and ethnic minority elderly adults. J Am Geriatr Soc. 2013;61(1):18–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Alegria M, Canino G, Rios R, Vera M, Calderon J, Rusch D, et al. Inequalities in use of specialty mental health services among Latinos, African Americans, and non-Latino whites. Psychiatric Services. 2002;53(12):1547–55. [DOI] [PubMed] [Google Scholar]
- 11.Chow JC, Jaffee K, Snowden L. Racial/ethnic disparities in the use of mental health services in poverty areas. American Journal of Public Health. 2003;93(5):792–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Weinick R, Zuvekas S, Cohen J. Racial and ethnic differences in access to and use of health care services, 1977 to 1996. Medical Care Research and Review. 2000;57(Supplement 1):36–54. [DOI] [PubMed] [Google Scholar]
- 13.Cook B, Flores M, Zuvekas S, Newhouse JP, Hsu J, Sonik R, et al. The Impact Of Medicare’s Mental Health Cost-Sharing Parity On Use Of Mental Health Care Services. Health Affairs. 2020;39(5):819–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Fung V, Price M, Nierenberg AA, Hsu J, Newhouse JP, Cook BL. Assessment of Behavioral Health Services Use Among Low-Income Medicare Beneficiaries After Reductions in Coinsurance Fees. JAMA Netw Open. 2020;3(10):e2019854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Choudhry NK, Bykov K, Shrank WH, Toscano M, Rawlins WS, Reisman L, et al. Eliminating medication copayments reduces disparities in cardiovascular care. Health affairs. 2014;33(5):863–70. [DOI] [PubMed] [Google Scholar]
- 16.Sabatino SA, Thompson TD, Guy GP Jr., de Moor JS, Tangka FK. Mammography Use Among Medicare Beneficiaries After Elimination of Cost Sharing. Med Care. 2016;54(4):394–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Trivedi AN, Leyva B, Lee Y, Panagiotou OA, Dahabreh IJ. Elimination of Cost Sharing for Screening Mammography in Medicare Advantage Plans. N Engl J Med. 2018;378(3):262–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.To access the Appendix, click on the Details tab in the article online.
- 19.Eicheldinger C, Bonito A. More accurate racial and ethnic codes for Medicare administrative data. Health care financing review. 2008;29(3):27–42. [PMC free article] [PubMed] [Google Scholar]
- 20.Ayanian JZ, Landon BE, Newhouse JP, Zaslavsky AM. Racial and ethnic disparities among enrollees in Medicare Advantage plans. The New England journal of medicine. 2014;371(24):2288–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wiechers IR LD, Rosenheck RA. Prescribing of psychotropic medications to patients without a psychiatric diagnosis. Psychiatric Services. 2013;64(12):1243–8. [DOI] [PubMed] [Google Scholar]
- 22.Dunn A, Grosse SD, Zuvekas SH. Adjusting Health Expenditures for Inflation: A Review of Measures for Health Services Research in the United States. Health Serv Res. 2018;53(1):175–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Pracitcal and Powerful Approach to Mulitple Testing. Journal of the Royal Statistical Society: Series B (Methodological). 1995;57(1):289–300. [Google Scholar]
- 24.Byers AL, Arean PA, Yaffe K. Low Use of Mental Health Services Among Older Amercians with Mood and Anxiety Disorders. Psych Serv. 2012;63(1):66–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Beals J, Novins DK, Whitesell NR, Spicer P, Mitchell CM, Manson SM. Prevalence of mental disorders and utilization of mental health services in two American Indian reservation populations: mental health disparities in a national context. Am J Psychiatry. 2005;162(9):1723–32. [DOI] [PubMed] [Google Scholar]
- 26.Medicare Payment Advisory Commision (MedPAC). Report to the Congress: Medicare Payment Policy. Washington, D.C.; Mar 2019. [Google Scholar]
- 27.Goldman HH, Frank RG, Burnam MA, Huskamp HA, Ridgely MS, Normand SL, et al. Behavioral health insurance parity for federal employees. N Engl J Med. 2006;354(13):1378–86. [DOI] [PubMed] [Google Scholar]
- 28.McConnell KJ, Gast SH, Ridgely MS, Wallace N, Jacuzzi N, Rieckmann T, et al. Behavioral health insurance parity: does Oregon’s experience presage the national experience with the Mental Health Parity and Addiction Equity Act? Am J Psychiatry. 2012;169(1):31–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Barry CL, Ridgely MS. Mental health and substance abuse insurance parity for federal employees: how did health plans respond? J Policy Anal Manage. 2008;27(1):155–70. [DOI] [PubMed] [Google Scholar]
- 30.Alang SM. Mental health care among blacks in America: Confronting racism and constructing solutions. Health Serv Res. 2019;54(2):346–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Saldana AM, Saldana AM, Mohayed MO, Bailey RK. Psychiatry’s Dark Secrets: Black Lives Don’t Matter. J Health Care Poor Underserved. 2021;32(3):1225–35. [DOI] [PubMed] [Google Scholar]
- 32.Sonik RA, Creedon TB, Progovac AM, Carson N, Delman J, Delman D, et al. Depression treatment preferences by race/ethnicity and gender and associations between past healthcare discrimination experiences and present preferences in a nationally representative sample. Soc Sci Med. 2020;253:112939. [DOI] [PubMed] [Google Scholar]
- 33.Bauer AM, Chen CN, Alegría M. English language proficiency and mental health service use among Latino and Asian Americans with mental disorders. Med Care. 2010;48(12):1097–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Garcia ME, Hinton L, Gregorich SE, Livaudais-Toman J, Kaplan C, Karliner L. Unmet Mental Health Need Among Chinese and Latino Primary Care Patients: Intersection of Ethnicity, Gender, and English Proficiency. J Gen Intern Med. 2020;35(4):1245–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Bauer AM, Alegría M. Impact of patient language proficiency and interpreter service use on the quality of psychiatric care: a systematic review. Psychiatr Serv. 2010;61(8):765–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Proctor K, Wilson-Frederick SM, Haffer SC. The Limited English Proficient Population: Describing Medicare, Medicaid, and Dual Beneficiaries. Health Equity. 2018;2(1):82–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Cook BL, Trinh NH, Li Z, Hou SS, Progovac AM. Trends in Racial-Ethnic Disparities in Access to Mental Health Care, 2004–2012. Psychiatr Serv. 2017;68(1):9–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Villagra VG, Bhuva B, Coman E, Smith DO, Fifield J. Health insurance literacy: disparities by race, ethnicity, and language preference. Am J Manag Care. 2019;25(3):e71–e5. [PubMed] [Google Scholar]
- 39.Berard LDH, Mackenzie CS, Reynolds KA, Thompson G, Koven L, Beatie B. Choice, coercion, and/or muddling through: Older adults’ experiences in seeking psychological treatment. Soc Sci Med. 2020;255:113011. [DOI] [PubMed] [Google Scholar]
- 40.Bhutta N, Chang A, Dettling LJ, Hsu JW. Disparities in Wealth by Race and Ethnicity in the 2019 Survey of Consumer FInances. Washington: Board of Governors of the Federal Reserve SYstem; 2020. Sept 28. [Google Scholar]
- 41.Cook BL, Doksum T, Chen CN, Carle A, Alegría M. The role of provider supply and organization in reducing racial/ethnic disparities in mental health care in the U.S. Soc Sci Med. 2013;84:102–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Cook BL, Zuvekas SH, Chen J, Progovac A, Lincoln AK. Assessing the Individual, Neighborhood, and Policy Predictors of Disparities in Mental Health Care. Med Care Res Rev. 2017;74(4):404–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Fung V, Hsu J, Newhouse JP, Cook BL. The Unevent Effects of Implementation of Mental Health Coinsurance Parity in Medicare on Psychiatry Visits. 10th Annual Conference of the American Society of Health Economists; Jun 22; Virtual Conference2021. [Google Scholar]
- 44.Eken HN, Dee EC, Powers AR 3rd, Jordan A. Racial and ethnic differences in perception of provider cultural competence among patients with depression and anxiety symptoms: a retrospective, population-based, cross-sectional analysis. Lancet Psychiatry. 2021;8(11):957–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.