Aligning Medicaid and Medicare Advantage Managed Care Plans for Dual-Eligible Beneficiaries

Abstract

To coordinate Medicare and Medicaid benefits, multiple states are creating opportunities for dual-eligible beneficiaries to join Medicare Advantage Dual-Eligible Special Needs Plans (D-SNPs) and Medicaid plans operated by the same insurer. Tennessee implemented this approach by requiring insurers who offered Medicaid plans to also offer a D-SNP by 2015. Tennessee’s aligned D-SNP participation increased from 7% to 24% of dual-eligible beneficiaries aged 65 years and above between 2011 and 2017. Within a county, a 10-percentage-point increase in aligned D-SNP participation was associated with 0.3 fewer inpatient admissions (p = .048), 13.9 fewer prescription drugs per month (p = .048), and 0.3 fewer nursing home users (p = .06) per 100 dual-eligible beneficiaries aged 65 years and older. Increased aligned plan participation was associated with 0.2 more inpatient admissions (p = .004) per 100 dual-eligible beneficiaries younger than 65 years. For some dual-eligible beneficiaries, increasing Medicare and Medicaid managed plan alignment has the potential to promote more efficient service use.

Introduction

Providing high-quality services for dual-eligible beneficiaries is challenging because Medicare’s coverage of medical care is not coordinated with Medicaid’s coverage of longterm services and supports (LTSS) and behavioral health. Requiring beneficiaries to navigate two different programs is concerning, especially since the dual-eligible population already stands out among other Medicare beneficiaries for having more chronic conditions, greater functional and cognitive impairment, and higher exposure to social risk factors that can adversely affect health (Bynum et al., 2017; Coughlin et al., 2012; Kasper et al., 2010; Medicaid and CHIP Payment and Access Commission & Medicare Payment and Access Commission, 2018).

Poor Medicare and Medicaid coordination also raises questions about inefficiency. These programs have no incentive to avoid shifting costs across programs at the risk of diminished health care quality and greater overall spending (Grabowski, 2007). For example, reducing unnecessary hospitalizations among dual-eligible nursing home residents may reduce Medicare spending on inpatient and postacute services but increase Medicaid spending on nursing home services (Mor et al., 2010; Walsh et al., 2012).

Because of these conflicting incentives, state and federal policymakers have launched new efforts to align Medicare and Medicaid benefits. One approach focuses on Medicare Advantage Dual-Eligible Special Needs Plans (D-SNPs), which exclusively enroll dual-eligible beneficiaries. Although D-SNPs had no formal requirements to coordinate Medicaid benefits when they were launched in 2006 (Grabowski, 2009), more recent federal regulations have addressed this gap, including requiring D-SNPs to contract with state Medicaid programs by 2013 in order to continue operating in a given state. These contracts establish several aspects of how D-SNPs will operate, including which dual-eligible populations are eligible for coverage and how the D-SNP will provide or, in most cases, coordinate with the state to provide Medicaid benefits. Under more recent regulations effective in 2021, these contracts require D-SNPs to take further steps to support Medicaid programs, including covering key Medicaid benefits (behavioral support or LTSS). States are leveraging this Medicare contracting authority and their influence as contractors with Medicaid managed care plans to increase coordination between Medicaid managed care plans and Medicare D-SNPs (Medicare Payment Advisory Commission, 2018, 2019; Verdier et al., 2015). Eleven states have created opportunities for dual-eligible beneficiaries to simultaneously enroll in a Medicaid managed care plan and a Medicare Advantage D-SNP operated by the same insurer (Gifford et al., 2019; Integrated Care Resource Center, 2019). As of 2019, about 690,000 of the 1.6 million D-SNP members with full Medicaid have aligned enrollment with their Medicaid managed care plans (Medicare Payment Advisory Commission, 2019).

Despite the growing use of this alignment approach, little evidence exists on how this strategy affects dual-eligible beneficiaries and their use of health care services. This study addresses that gap by studying a period of increasing aligned D-SNP participation in Tennessee as the state instituted a requirement for all state Medicaid insurers to offer a Medicare D-SNP. We examined whether increased aligned plan participation between 2012 and 2017 is associated with changes in the use of acute medical services, prescription drugs, and LTSS.

New Contributions

To the best of our knowledge, this study is the first analysis to focus on the initial implementation of an aligned D-SNP strategy. Our analysis approach leverages an increase in aligned D-SNP participation as plans expanded D-SNP offerings. In contrast, other studies have focused on states with more established alignment models for dual-eligible populations that did not experience large changes in managed care participation during the study period (Anderson et al., 2020; Kim et al., 2019).

For states that are considering launching a similar alignment approach, these results provide important insight on participation and service use outcomes during the initial years of implementation.

Conceptual Framework

Aligned Medicare D-SNP and Medicaid plans may improve care coordination and promote more efficient service use because one insurer is responsible for both Medicare and Medicaid benefits, even if these benefits are administered through separate plans. The insurer has a financial incentive to figure out which combination of Medicare and Medicaid services will result in lower overall spending for dual-eligible beneficiaries. Furthermore, establishing communication channels between Medicare and Medicaid benefit representatives may be easier if both plans are operated by the same insurer. For example, a more coordinated care approach may be able to reduce hospitalizations and emergency room visits, encourage appropriate medication use, and promote the use of LTSS that are delivered at home instead of in a nursing home.

Method

Background: Tennessee’s Alignment Strategy

Tennessee’s Medicaid program, TennCare, requires all dual-eligible beneficiaries with full Medicaid coverage to enroll in Medicaid managed care plans. Plan benefits include coverage of LTSS. If beneficiaries do not choose a Medicaid plan, then the state automatically assigns them one. In 2013, TennCare announced that all Medicaid insurers must offer an aligned D-SNP in as many counties as possible by January 2015 and create a strategy for statewide D-SNP coverage by January 2016 (State of Tennessee Department of Finance and Administration, 2013; Appendix Table A1, in the online version of this journal, details the timeline). At the time of the announcement, one of the state’s three Medicaid insurers already offered statewide D-SNP coverage, one offered D-SNP coverage in a limited number of counties, and one had no D-SNP availability. By 2016, all three insurers had statewide D-SNP availability. TennCare prohibited new D-SNPs from non-Medicaid insurers but allowed D-SNPs in existence before 2014 to continue operating (Verdier et al., 2015).

In most cases, Tennessee’s dual-eligible beneficiaries gain aligned coverage by opting into D-SNPs that are aligned with their Medicaid managed care plan. Two exceptions late in our study period are notable. First, when Medicaid-only participants gain Medicare coverage and become dual eligible, they are automatically enrolled in the aligned D-SNP of their current Medicaid insurer. Beneficiaries can opt out for another source of Medicare coverage if desired. Each Medicaid insurer received federal permission to implement this “seamless conversion” approach in 2015 and 2016 (Centers for Medicare & Medicaid Services, 2021). Second, when two Medicaid insurers expanded from regional to statewide Medicaid coverage in 2015, the state reassigned some dual-eligible beneficiaries in the new coverage regions to these Medicaid plans unless the beneficiary opted out. If beneficiaries already belonged to a D-SNP offered by the newly available Medicaid insurer (but did not already have aligned benefits), then the reassignment process assigned them to the aligned Medicaid plan. Beneficiaries already in aligned plans were less likely to be reassigned to a new Medicaid insurer (Division of TennCare, 2019; personal email communication with Patti Killingsworth, TennCare Assistant Commissioner and Chief of Long-Term Services and Supports, October 10, 2019).

Data

Data sources span the years 2011-2017 for dual-eligible beneficiaries aged 65 years and over and the years 2011-2016 for dual-eligible beneficiaries under age 65. Two data sets identified dual-eligible beneficiaries’ insurance coverage: (1) TennCare eligibility files for Medicaid managed care plan coverage and (2) the Medicare Master Beneficiary Summary File for traditional Medicare or Medicare Advantage plan participation. We analyzed Medicaid plan encounter data, obtained from TennCare, to measure the use of nursing home services and home- and community-based services (HCBS). Tennessee hospital discharge data identified inpatient and emergency room use. Medicare Part D data identified prescription drug use. This latter data source was only available through 2015 and limited to dual-eligible beneficiaries who qualified for Medicaid with income less than 100% of the Federal Povery Level for dual-eligible beneficiaries younger than 65 years.

Population

The study population included dual-eligible beneficiaries with full Medicaid for at least 1 month during our study period (see the online Appendix Figure A1 for details). We excluded dual-eligible beneficiaries whose Medicare data could not be linked to Medicaid data (n = 10,906 [7.7%] beneficiaries under age 65 years; n = 5,036 [3.2%] beneficiaries aged 65 years and above). Among beneficiaries with linked data, we excluded records for the following reasons (n = 7,353 [1.4%] beneficiaries under age 65 years; n = 1,825 [4.9%] beneficiaries aged 65 years and above): belonging to multiple Medicaid plans in the same month, missing or invalid Medicaid plan identifiers, multiple Medicare beneficiary identifiers matched to one Medicaid participant identifier, out-of-state residence according to Medicare data, or discrepancies between Medicare and Medicaid records in the timing of Medicaid participation. We stratified analyses by age-group and, in some specifications, by LTSS use. The main model analyses included 130,677 beneficiaries age 65 years and over and 117,217 beneficiaries under age 65 years.

Measures

We categorized each beneficiary’s alignment status, which could change monthly, as follows: (a) an aligned D-SNP (same insurer for Medicaid and Medicare D-SNP plans), (b) a nonaligned D-SNP (different insurers for Medicaid and Medicare D-SNP plans), (c) traditional Medicare, and (d) Medicare Advantage plan other than a D-SNP.

Main study outcomes for Medicaid-covered services included number of nursing home users or HCBS users per month per 100 dual-eligible beneficiaries. HCBS services included in-home personal care services, adult day center services, assisted living residence, and other benefits like home-delivered meals. Main study outcomes for Medicare-covered services included number of hospital admissions, emergency room visits, and unique prescription drugs per month per 100 beneficiaries. This latter measure is the number of claims for a unique generic drug name, disregarding distinctions across claims in dosage or number of fills.

Other measures included age (categorized in 5-year increments), sex, race or ethnicity (categorized as Black, Hispanic, or White), living in a rural county as defined by the Census Bureau, and whether beneficiaries originally qualified for Medicare due to disability or end-stage renal disease instead of qualifying due to age. We also identified whether beneficiaries qualified for Medicaid with income less than 100% of the federal poverty level. As a proxy for beneficiaries’ underlying health, we created variables to measure beneficiaries’ health care use over the past 12 months. These variables counted the total number of hospital admissions, emergency room visits, months with nursing home use, and months with HCBS use in the year prior to the month of observation.

Analysis Approach

For each year and alignment group, we reported the mean monthly number of nursing home users, HCBS users, inpatient admissions, emergency room visits, and prescription drugs per 100 beneficiaries, with adjustment for age and sex. Any differences in health care use observed in these descriptive trends could be partially due to favorable selection. If healthier beneficiaries were more likely to join aligned D-SNPs, then the health care use of beneficiaries in aligned D-SNPs would be lower, thus biasing estimates of whether aligned benefits are associated with changes in health care use. Instead of directly comparing dual-eligible beneficiaries in aligned plans and traditional Medicare, we estimated a fixed-effects model to understand the association between increased county-level participation in aligned benefits and the health care utilization outcomes of all dual-eligible beneficiaries, including those without aligned benefits. In this model, i indexes beneficiaries, c indexes counties, and t indexes months:

$${\text{Outcomes}}_{ict}={\alpha }_0+{\alpha }_1{\text{PercDSNPAlign}}_{c,t-6}+a_2{X}_{ict}+{\text{county}}_c+{\text{Month}}_t+{\text{ε}}_{ict}$$

The model’s unit of observation was a person-month for observations between 2012 and 2017 for older beneficiaries and between 2012 and 2016 for younger beneficiaries. The main explanatory variable (PercDSNPAlign_{c, t–6}) was the percentage of dual-eligible beneficiaries in a beneficiary’s county enrolled in an aligned plan. This percentage increased over time at variable rates across counties, allowing us to observe whether larger increases in aligned benefit participation were associated with larger changes in health care use. We lagged this variable by 6 months for two reasons: We assumed that the impact of aligned plans grew over time and lagging the variable provided assurance that beneficiaries’ health care use did not precede recent changes in aligned plan participation. For ease of interpretation, we scaled the average marginal effect of this variable to quantify the absolute monthly difference in outcome per 100 dual-eligible beneficiaries associated with a 10-percentage-point increase in the proportion of beneficiaries with aligned plan participation.

County fixed effects (County_c) controlled for time-invariant local characteristics that may influence health care use, such as provider availability. Monthly fixed effects (Month_t) accounted for statewide secular trends in health care use. This approach captured any direct results of aligned plans improving care for their own members, as well as any spill-over effects for traditional Medicare beneficiaries as more of a county’s population enrolled in managed care. The models also included the set of previously described covariates, X_ict (except for rural county status since the model included county fixed effects). Health care outcomes (Outcomes_ict,) were modeled using a generalized linear model with a Poisson and a binomial distribution used for count outcomes and binary outcomes, respectively. Standard errors were clustered at the county level; Tennessee has 95 counties.

Although the main variable of interest is the share of dual-eligible beneficiaries in aligned D-SNPs, changes in participation in other Medicare managed care options may have also influenced health care use among dual-eligible beneficiaries. To see if our results were robust to accounting for these other trends in managed care enrollment, we added two additional county-level variables to our model (both lagged by 6 months): (1) the percentage of beneficiaries enrolled in nonaligned D-SNPs and (2) the percentage of beneficiaries enrolled in other, non-D-SNP Medicare Advantage plans. Other sensitivity analyses included stratifying models by LTSS use, adding plan fixed effects with interaction terms for plan and region, omitting covariates that measure prior health care and LTSS use, and limiting the study years to 2012-2014 (prior to two Medicaid plans shifting to statewide coverage).

Results

From 2011 to 2017, aligned D-SNP participation increased from 7.3% to 23.7% of dual-eligible beneficiaries aged 65 years and above (Figure 1). There were also gains in how many dual-eligible beneficiaries had a Medicare D-SNP and Medicaid plan operated by two different insurers—participation in these nonaligned D-SNPs increasing from 9.3% to 16.0% of older beneficiaries. Participation in other Medicare Advantage plans was more stable, increasing slightly from 4.5% to 6.1% of older beneficiaries. Among dual-eligible beneficiaries under 65, participation over the years 2011-2016 increased from 10.7% to 31.4% in aligned D-SNPs, from 9.8% to 16.2% in nonaligned D-SNPs, and from 1.8% to 2.5% in other Medicare Advantage plans. By the end of 2017 and 2016, respectively, only 54.3% of older beneficiaries and 49.8% of beneficiaries under age 65 had traditional Medicare.

Figure 1.

Percentage of Dual-Eligible Beneficiaries by Alignment Status and Age-Group

Note. Authors’ analysis of Medicare, Medicaid enrollment data. Population includes dual-eligible beneficiaries with full Medicaid benefits. Other Medicare Advantage plans are all other Medicare Advantage plans that are not D-SNPs. Participation in traditional Medicare decreased from 77.6% in 2011 to 49.8% in 2016 among beneficiaries under age 65 years and from 78.9% to 54.3% in 2017 among beneficiaries aged 65 years and over (not shown). D-SNP = Dual-Eligible Special Needs Plan.

Counties varied in how quickly participation in aligned D-SNPs increased (see the online Appendix Table A2). For example, the median absolute increase from December 2011 to December 2012 in county-level aligned plan participation was 1.6 percentage points (interquartile range: 0.8-2.2 percentage points) for older dual-eligible beneficiaries. This median value ranged from 1.6 to 2.9 percentage points over the years 2012-2017 for older dual-eligible beneficiaries and from 2.3 to 4.9 percentage points over the years 2012-2016 for younger dual-eligible beneficiaries. By 2017, half the counties had more than 22.1% of older beneficiaries participating in aligned plans (interquartile range: 18.3% to 25.8%). Among younger dual-eligible beneficiaries, the median county-level aligned D-SNP participation was 30.6% (interquartile range: 25.7% to 33.3%) in 2016.

When comparing the characteristics of beneficiaries enrolled in aligned D-SNPs and traditional Medicare, the largest differences were observed in age and race (Table 1). Beneficiaries in aligned plans were more likely to be between the ages of 45 and 74 years rather than in the youngest or oldest age-groups. Aligned D-SNP members were more likely to be Black than traditional Medicare beneficiaries: 27.0% versus 20.0% among beneficiaries aged 65 and above and 28.6% versus 23.3% among beneficiaries under age 65, respectively. The proportion living in mostly urban counties was highest among aligned D-SNP members (68.8% of older beneficiaries; 71.3% of younger beneficiaries).

Table 1.

Characteristics of Dual-Eligible Beneficiaries by Alignment Status and Age-Group, January 2012.

	Under age 65 years				Age 65 years and above
Characteristic	Traditional Medicare	Aligned D-SNP	Nonaligned D-SNP	Other Medicare Advantage	Traditional Medicare	Aligned D-SNP	Nonaligned D-SNP	Other Medicare Advantage
Number	47,561	9,127	7,187	1,394	48,477	5,916	6,395	3,314
Age (years): mean (SD)	47.7 (11.0)	48.8 (10.2)	49.5 (9.8)	49.6 (10.0)	77.9 (8.9)	73.9 (7.1)	75.6 (7.5)	78.4 (8.8)
Under age 45 years (%)	35.7	31.1	28.7	28.3
Age 45-54 years (%)	31.5	34.6	34.9	33.5
Age 55-64 years (%)	32.7	34.3	36.4	38.2
Age 65-74 years (%)					41.9	61.4	51.2	38.6
Age 75-84 years (%)					32.7	28.8	35.2	34.0
Over age 85 years (%)					25.4	9.8	13.6	27.4
Female (%)	53.4	56.3	57.2	56.2	71.1	68.2	71.7	72.9
Race/ethnicity (%)
White	74.7	69.2	62.3	71.9	76.7	69.2	60.0	76.4
African American or Black	23.3	28.6	36.6	26.5	20.0	27.0	37.7	22.2
Hispanic or other	1.8	1.9	1.1	1.4	3.1	3.7	2.2	1.3
Rural (%)
Completely	5.9	3.9	3.5	3.4	7.3	5.5	4.4	3.6
Mostly	34.0	32.9	25.2	27.8	37.2	37.1	26.7	27.7
Mostly urban	60.1	63.2	71.3	68.9	55.5	57.4	68.8	68.8
Qualified Medicare beneficiaries	46.7	49.3	54.7	43.6	37.5	36.2	40.7	30.4
Originally qualified for Medicare due to disability	100.0	100.0	100.0	100.0	32.4	35.4	41.2	29.2
Service use prior year (per 100 beneficiaries)
Hospital admissions	45.2	31.2	31.0	46.4	62.6	39.1	43.4	76.4
Emergency room visits	175.8	168.2	171.5	168.7	128.4	111.6	117.3	150.8
Any nursing home use (%)	5.0	0.4	1.5	4.1	33.4	2.8	5.8	30.9
Any HCBS use (%)	8.5	5.6	7.2	8.4	15.9	11.5	8.4	15.9

Note. Authors’ analysis of Medicare, Medicaid enrollment data. Population includes dual-eligible beneficiaries with full Medicaid benefits. D-SNP = Dual-Eligible Special Needs Plan; HCBS = home- and community-based services.

Aligned D-SNP members were much less likely to use nursing home services than traditional Medicare beneficiaries (Figure 2). An average of 2.5% (95% confidence interval [CI]: 2.3%, 2.7%) of older aligned plan members per month in 2016 used nursing home services compared with 29.5% (95% CI: 29.1%, 29.9%) of traditional Medicare beneficiaries after adjustment for age and sex. Among beneficiaries under age 65, the average monthly percentage of beneficiaries with nursing home use was only 0.4% (95% CI: 0.3%, 0.5%) of aligned D-SNP members compared with 4.6% (95% CI: 4.4%, 4.8%) of traditional Medicare members (see the online Appendix Figure A2).

Figure 2.

Monthly Number of Nursing Home Users and HCBS Users per 100 Dual-Eligible Beneficiaries Age 65 Years and Above by Alignment Status

Note. Authors’ analysis of Medicare, Medicaid enrollment data and Medicaid encounter data. Population includes dual-eligible beneficiaries with full Medicaid benefits. Adjusted for age and sex. Trad. Medicare = traditional Medicare; D-SNP: Dual-Eligible Special Needs Plan; Non D-SNP MA = Medicare Advantage plan other than D-SNPs; HCBS = home- and community-based services.

Among older dual-eligible beneficiaries, those in aligned plans and traditional Medicare had similar HCBS use trends, with universal increases in HCBS use from 2011 to 2013. By 2017, an average of 11.6% (95% CI: 11.1%, 12.1%) and 12.0% (95% CI: 11.7%, 12.3%) beneficiaries aged 65 and above used HCBS in aligned plans and traditional Medicare per month, respectively. Among dual-eligible beneficiaries under 65, aligned plan members were less likely to use HCBS services than traditional Medicare beneficiaries (see the online Appendix Figure A2). Per month, an average of 5.2% (95% CI: 5.0%, 5.5%) of younger aligned plan members used HCBS in 2016 compared with 9.1% (95% CI: 8.9%, 9.4%) of younger traditional Medicare beneficiaries.

Differences between traditional Medicare and aligned D-SNP members varied across Medicare-covered services depending on the age-group and beneficiaries’ use of LTSS (see the online Appendix Figures A3 and A4). For example, among beneficiaries without LTSS use, aligned D-SNP members had lower rates of hospitalization after adjustment for age and sex than beneficiaries with traditional Medicare in both age-groups. In contrast, among beneficiaries aged 65 and above with LTSS use, traditional Medicare beneficiaries had lower inpatient admission rates than aligned D-SNP members. Some differences were also observed for emergency room and prescription drug use, such as higher rates of prescription drug use and lower rates of emergency room visits for traditional Medicare LTSS users compared with aligned D-SNP LTSS users in both age-groups.

Because these differences in health care use by alignment status may reflect selection into aligned D-SNPs instead of the effect of aligned benefits, we examined how county-level increases in aligned D-SNP participation were associated with health care use for all dual-eligible beneficiaries (Table 2). Among older beneficiaries, a 10-percentage-point increase in the share of dual-eligible beneficiaries with aligned D-SNP participation in a county was associated with 0.3 fewer inpatient admissions (95% CI: −0.6, 0.0; p = .048) and 13.9 fewer prescription drugs (95% CI: −27.8, −0.1; p = .048) per month per 100 dual-eligible beneficiaries. Although not statistically significant at the 5% level, there were marginal associations with shifts in LTSS use: A 10% increase in aligned plan participation was associated with 0.3 fewer nursing home users (95% CI: −0.6, 0.0; p = .06) and 0.2 more HCBS users (95% CI: −0.1, 0.5; p = .12) per 100 dual-eligible beneficiaries. Relative to the monthly rate of service use among all older dual-eligible beneficiaries throughout the study period, these shifts range from a 1.4% decline in nursing home use (mean 22.2 nursing home visits per month per 100 beneficiaries) to a 5.9% relative decline in inpatient admissions (mean 5.1 inpatient admissions per month per 100 beneficiaries). No significant differences were observed for emergency room use among all older dual-eligible beneficiaries, although stratified results for LTSS users show a marginally significant association for fewer emergency room visits in that population (0.8 [95% CI: −1.7, 0.0] fewer emergency room visits per 100 beneficiaries, p = .06).

Table 2.

Change in Health Care Use Among All Dual-Eligible Beneficiaries With Increased County-Level Aligned D-SNP Participation by Age-Group and Use of LTSS.

	Under age 65 yearsa			Age 65 years and above
Characteristic	All (N = 4,163,849)	Non-LTSS users (N = 3,714,911)	LTSS users (N = 448,938)	All (N = 4,533,829)	Non-LTSS users (N = 2,990,463)	LTSS users (N = 1,543,366)
Change in number per month per 100 beneficiaries with 10-percentage-point increase in aligned plan participation
Nursing home users	—			−0.3 (−0.6, 0.0)
HCBS users	0.0 (−0.1, 0.2)			0.2 (−0.1, 0.5)
Hospital admissions	0.2** (0.1, 0.4)	0.2* (0.0, 0.4)	0.2 (−0.2, 0.7)	−0.3* (−0.6, −0.0)	−0.2 (−0.5, 0.1)	−0.3 (−0.7, 0.1)
Emergency room visits	−0.2 (−0.6, 0.3)	−0.1 (−0.7, 0.4)	0.2 (−0.7, 1.0)	−0.4 (−1.2, 0.3)	−0.2 (−1.2, 0.7)	−0.8 (−1.7, 0.0)
Prescription drugs	−0.1 (−9.6, 9.4)	−0.6 (−9.8, 8.5)	3.1 (−18.2, 24.4)	−13.9* (−27.8, −0.1)	−6.6 (−17.8, 4.7)	−19.7 (−47.5, 8.1)
Mean number of users or events per month per 100 beneficiaries
Nursing home users	3.1	—	—	22.2	—	65.2
HCBS users	7.7	—	71.6	12.0	—	35.3
Hospital admissions	3.5	3.2	5.8	5.1	4.4	6.5
Emergency room visits	15.5	15.6	14.8	12.1	11.5	13.1
Prescription drugs	501.1	470.7	694.9	554.4	453.7	749.5

Note. Authors’ analysis of Medicare, Medicaid enrollment data; Medicaid encounter data; Medicare Part D data; and Tennessee hospital discharge records. Population includes dual-eligible beneficiaries with full Medicaid benefits. N indicates number of person-months. Analysis included control variables for county fixed effects, month fixed effects, age, race, sex, original reason for Medicare eligibility, qualifying for Medicaid through a Qualified Medicare Beneficiary pathway, and inpatient, emergency room, nursing home, and HCBS use over the past 12 months. Study period varied by population. Under age 65 years: 2012-2016 for inpatient admissions, emergency room visits, and LTSS use; 2012-2015 for prescription drug use. Age 65 years and above: 2012-2017 for all outcomes. LTSS = long-term services and supports; D-SNP = Dual-Eligible Special Needs Plan; HCBS = home- and community-based services.

^aPrescription drug analyses for the under-65-years population had the following number of person-months due to there being one less year of Part D data and exclusion of individuals who qualified for Medicaid with income greater than 100% Federal Poverty Level): all (N = 1,709,938); non-LTSS users (N = 1,477,927); LTSS users (N = 232,011).

^*p < .05.

^**p < .01.

Among younger beneficiaries, a 10-percentage-point increase in the share of dual-eligible beneficiaries with aligned D-SNP participation in a county was associated with 0.2 more inpatient admissions (95% CI: 0.1, 0.4; p = .004) per 100 dual-eligible beneficiaries. Relative to average inpatient use among younger beneficiaries, this estimate represents a 5.7% increase in inpatient use. In stratified results by LTSS use, this significant association was present for non-LTSS users: 0.2 more inpatient admissions (95% CI: 0.0, 0.4; p = .01) per 100 dual-eligible non-LTSS users. No other outcomes were significantly different with increased aligned D-SNP participation among younger beneficiaries.

Although the direction of our estimates remained consistent in the model that included additional covariates for the county-level percentage of dual-eligible beneficiaries in non-aligned D-SNPs and other Medicare Advantage plans, the statistical significance of some estimates did change (Table 3). In this model, increased aligned D-SNP participation was significantly associated with 0.3 more HCBS users per 100 beneficiaries aged 65 and above (95% CI: 0.0, 0.05, p = .04). Estimates for changes in inpatient admissions and prescription drug use among older adults were still negatively associated with increased aligned D-SNP participation but were no longer statistically significant. Among younger beneficiaries, increased aligned participation was still associated with more inpatient admissions in the extended model. For most outcomes in both age-groups, neither nonaligned D-SNP participation nor other Medicare Advantage plan participation was associated with changes in service use. The notable exception was that participation in other Medicare Advantage plans was associated with greater use of HCBS in both age-groups: 0.5 (95% CI: 0.0, 1.1; p = .05) and 0.6 (95% CI: 0.2, 0.9; p = .001) additional HCBS users per 100 beneficiaries among older and younger beneficiaries, respectively. Nonaligned D-SNP participation was associated with fewer hospital admissions among younger beneficiaries (−0.3 [95% CI: −0.5, −0.1] inpatient admissions per 100 dual-eligible beneficiaries, p = .02).

Table 3.

Change in Health Care Use Among All Dual-Eligible Beneficiaries With Increased County-Level Managed Care Participation by Age-Group.

	Under age 65 years			Age 65 years and above
Characteristic	Aligned D-SNP	Nonaligned D-SNP	Other MA	Aligned D-SNP	Nonaligned D-SNP	Other MA
Change in number per month per 100 beneficiaries with 10-percentage-point increase in aligned plan participation
Nursing home users	—	—	—	−0.2 (−0.5, 0.1)	0.2 (−0.1, 0.5)	0.4 (−0.2, 0.9)
HCBS users	0.1 (−0.0, 0.2)	0.0 (−0.2, 0.2)	0.6** (0.2, 0.9)	0.3* (0.0, 0.5)	−0.2 (−0.5, 0.2)	0.5* (0.0, 1.1)
Hospital admissions	0.2* (0.0, 0.4)	−0.3* (−0.5, −0.1)	0.4 (−0.1, 0.8)	−0.2 (−0.5, 0.1)	−0.2 (−0.6, 0.2)	0.4 (−0.0, 0.8)
Emergency room visits	−0.2 (−0.8, 0.4)	−0.4 (−1.5, 0.7)	0.3 (−1.9, 2.6)	−0.4 (−1.2, 0.3)	0.3 (−0.8, 1.5)	0.0 (−1.2, 1.2)
Prescription drugs	1.4 (−8.4, 11.1)	10.1 (−2.2, 22.3)	−1.3 (−15.4, 12.9)	−13.2 (−27.3, 1.0)	2.4 (−13.3, 18.1)	7.1 (−17.1, 31.3)

Note. Authors’ analysis of Medicare, Medicaid enrollment data; Medicaid encounter data; Medicare Part D data; and Tennessee hospital discharge records. Population includes dual-eligible beneficiaries with full Medicaid benefits. Analysis included control variables for county fixed effects, month fixed effects, age, race, sex, original reason for Medicare eligibility, qualifying for Medicaid through a Qualified Medicare Beneficiary pathway, and inpatient, emergency room, nursing home, and HCBS use over the past 12 months. Study period varied by population. Under age 65 years: 2012-2016 for inpatient admissions, emergency room visits, and LTSS use; 2012-2015 (excluding individuals who qualified for Medicaid with income greater than 100% FPL) for prescription drug use. Age 65 years and above: 2012-2017 for inpatient admissions, emergency room visits, LTSS use, and prescription drug use. D-SNP = Dual-Eligible Special Needs Plan; HCBS = home- and community-based services; LTSS = long-term services and supports.

^*p < .05.

^**p < .01.

Except for some changes in statistical significance, other sensitivity analyses that explored adding additional fixed effects or omitting controls for prior health care use were largely consistent with the findings from our main specification regardless of the covariates included in the model (see the online Appendix Table A3). Among younger beneficiaries in these early study years, there were significant increases in inpatient and emergency room use associated with increased aligned plan participation.

Discussion

After Tennessee required Medicaid insurers to offer a Medicare Advantage D-SNP, the proportion of dual-eligible beneficiaries in aligned D-SNPs more than tripled. Compared with beneficiaries remaining in traditional Medicare, aligned D-SNP members were less likely to be in the oldest or youngest age-groups and more likely to be Black. In analyses adjusted for age and sex, aligned D-SNP members had much lower nursing home use. Aligned D-SNP members under age 65 were less likely to use HCBS than traditional Medicare beneficiaries, but HCBS use was similar among older adults.

Because these health care use differences may reflect favorable selection into aligned D-SNPs, we examined health care use among all dual-eligible beneficiaries as the percentage of aligned plan members increased within a county. Among beneficiaries aged 65 and over, increases in aligned D-SNP participation were significantly associated with several modest differences—lower inpatient and prescription drug use and marginal associations with reduced nursing home use. In contrast, increased aligned D-SNP participation among beneficiaries under age 65 was associated with increases in inpatient use. Changes in other types of managed care enrollment were largely not associated with changes in health care use, but increased participation in other, non-DSNP Medicare Advantage plans was associated with increased HCBS use in both age-groups.

This evidence about Tennessee’s D-SNP integration approach joins mixed findings from other alignment initiatives (Medicare Payment Advisory Commission, 2016). Tennessee achieved substantial growth in aligned D-SNP participation largely through opt-in enrollment and increasing aligned plan availability. In contrast, some states in the Financial Alignment Initiative have struggled to maintain participation in Medicare–Medicaid managed care plans even with passive enrollment (Grabowski et al., 2017). Similar to trends observed in Oregon, Tennessee dual-eligible beneficiaries in aligned plans were less likely to be nursing home residents (Kim et al., 2019). The conflicting results for older and younger dual-eligible beneficiaries from the Tennessee approach echo the disparate results from other alignment approaches. Evidence from states with high participation in aligned approaches, such as Minnesota’s well-established integrated plans and Ohio’s new Medicare-Medicaid plans, suggest that these strategies were associated with lower inpatient use, increased HCBS use, and lower nursing home use (Anderson et al., 2020; Bayer et al., 2018). Fewer data are available about younger dual-eligible beneficiaries. Massachusetts, which limited Medicare–Medicaid plan membership to beneficiaries under age 65, has experienced low participation rates, increased inpatient use, and decreased nursing home use (Gattine et al., 2019). Findings from Oregon suggest that beneficiaries in aligned Medicare and Medicaid managed care plans had larger improvements in quality-of-care measures over a 4-year period than those in unaligned managed care plans (Kim et al., 2019).

Several factors specific to Tennessee are important to consider when interpreting these results. First, mandatory participation in Medicaid managed care plans could facilitate voluntary Medicare managed care participation. Second, the extra benefits offered by many D-SNPs, such as dental or hearing coverage, may be more appealing since Tennessee’s Medicaid program does not cover these services for dual-eligible beneficiaries. Third, in 2012, Tennessee’s Medicaid program ranked in the bottom half of states in terms of share of LTSS in HCBS versus institutional services (Eiken et al., 2014), suggesting that Tennessee had greater potential to increase HCBS use. Other states with more established HCBS use may have less room to expand HCBS use with greater aligned plan participation. Fourth, Tennessee is one of the country’s most experienced states in overseeing Medicaid managed care plans and retains considerable control over Medicaid benefit design.

Finally, Tennessee implemented mandatory managed care participation for long-term services and supports (MLTSS) in 2010, shortly before our study period began. Some trends observed in our study may reflect insurers simultaneously gaining experience with MLTSS and aligned D-SNPs. In an evaluation that compared Tennessee’s dual-eligible LTSS users with a matched control group from Alabama and Georgia, this shift to MLTSS was associated with less use of any HCBS and more inpatient use over the years 2010-2014 (Libersky et al., 2018). This analysis was not stratified by age-group but is consistent with our finding of more inpatient admissions among younger dual-eligible beneficiaries.

The aligned D-SNP strategy used by Tennessee and other states addresses some, but not all, benefit integration issues. Aligned D-SNPs (and nonaligned D-SNPs) are responsible for coordinating with Medicaid plans about beneficiaries’ care needs and transitions (Barth et al., 2019), a basic communication role lacking for beneficiaries in traditional Medicare. Another advantage is that an insurer can benefit financially at an organizational level if aligned plans improve overall efficiency, which provides some alignment of financial incentives. Still, it is not clear how much alignment alone affects individual plan operations or whether plans are formally evaluated by their joint performance.

Important integration issues remain in aligned D-SNP models. First, there is no formal mechanism for states to share in any Medicare savings from aligned D-SNPs. Tennessee only financially benefits from reduced inpatient stays if these decreases delay LTSS use or reduce the use of other services primarily funded by Medicaid. Tennessee’s Medicaid program may indirectly benefit if supporting aligned D-SNPs enhances the state’s ability to retain and attract competitive Medicaid plan bids, which is possible given insurers’ interest in states where they can offer multiple Medicare and Medicaid product lines (Archibald et al., 2019). Second, providing Medicare and Medicaid benefits through two separate plans leaves some questions open about which plan is the primary payer when Medicare and Medicaid benefits overlap, especially since Medicare Advantage plans can offer services traditionally covered by Medicaid, like nonemergency medical transportation (Willink & DuGoff, 2018). Notably, participation in other Medicare Advantage plans was a stronger predictor of increased HCBS use than participation in aligned D-SNPs, possibly due to universal incentives across all Medicare Advantage plans to supplement Medicare-financed home health benefits with Medicaid coverage of home-based services.

Although LTSS users may benefit most from better care coordination, Tennessee’s aligned D-SNPs reached very few nursing home users and, among younger beneficiaries, fewer HCBS users. Because Tennessee has mandatory managed care enrollment for LTSS, nursing homes must contract with Medicaid plans to provide Medicaid services for dual-eligible beneficiaries, which means that the state has established LTSS managed care plan networks, a process that was challenging in other alignment models (Lipson et al., 2018). However, when comparing traditional Medicare and D-SNP skilled nursing facility (SNF) benefits, nursing homes may find it easier to be paid under traditional Medicare coverage policies. Nursing homes have reported challenges navigating SNF coverage in Medicare Advantage plans, including issues with network contracting, prior authorization, and burdensome appeal processes when payments are denied (Gadbois et al., 2018). Dual-eligible beneficiaries who use both SNF and long-term residential coverage may be more likely to join D-SNPs if SNF benefits are easily accessible in their nursing home of choice. Given Tennessee’s enrollment patterns among LTSS users, aligned D-SNPs currently hold the most promise when it comes to deterring or delaying nursing home use among community residents and coordinating services for HCBS users.

Finally, offering unbiased assistance to dual-eligible beneficiaries as they navigate Medicare enrollment decisions is important given how many complex factors must be considered. Having the same insurer provide Medicare and Medicaid benefits may increase confusion regarding which Medicaid benefits, such as HCBS, are available to dual-eligible beneficiaries regardless of Medicare coverage and may lead to unfounded concerns about disenrolling from D-SNPs. Many Tennessee dual-eligible beneficiaries joined a D-SNP that was not operated by their Medicaid insurer, suggesting that other coverage factors, like Medicare provider networks, may be more important to beneficiaries than benefit alignment. Increased participation in these nonaligned D-SNPs was not associated with any changes in health care use other than reduced inpatient use among younger dual-eligible beneficiaries, raising concerns about the value of this enrollment option.

This study has several limitations. First, our study approach analyzes whether the shift to aligned D-SNP participation is associated with changes in health care and long-term care use, but we cannot establish whether greater aligned D-SNP participation caused these changes. Aligned plan participation may have been more likely to increase in counties that had greater potential, for example, for reduced nursing home use among older adults. Second, we cannot assess whether changes in health care or LTSS use are associated with changes in health outcomes or quality of care. Due to data limitations, the study period for beneficiaries under age 65 is 1 year shorter than the study period for older beneficiaries. Thus, any differences in results for these two populations may reflect the fact that we could observe more long-term effects of aligned D-SNP participation for older beneficiaries. Finally, some concurrent changes to the Tennessee Medicaid program, such as suspending Medicaid eligibility reverification processes in 2014 (Artiga & Pham, 2019) and the statewide expansion of two Medicaid plans in 2015, may influence patterns of health care use. In our analytic models, the monthly fixed effects are included to account for these secular trends that may influence health care patterns for all of Tennessee’s dual-eligible beneficiaries outside of increases in aligned D-SNP participation.

Tennessee’s experience with aligned D-SNPs suggests that this strategy may be effective at reaching dual-eligible beneficiaries and have modest success in reducing inpatient and institutional care among older beneficiaries. However, results to date have shown less evidence of success among younger beneficiaries and among nursing home users of all ages. Federal and state policymakers should consider these advantages and limitations of the alignment approach when considering initiatives to expand aligned D-SNP availability.