Health Care Use and Spending Among Need-Based Subgroups of Medicare Beneficiaries With Full Medicaid Benefits

Key Points Question For Medicare beneficiaries with full Medicaid benefits, how does the use of Medicare and Medicaid services and spending by payer differ across need-based subgroups? Findings This cross-sectional study including 333 240 Medicare and Medicaid beneficiaries found that substantial use of Medicare- and Medicaid-funded services across all subgroups and describes the services contributing the most to total spending differed across subgroups. Medicaid-funded services, including community-based services and nursing home care, were key contributors to total spending among subgroups needing long-term care. Meaning The diversity of health care use among dual-eligible beneficiaries requires integration strategies with comprehensive combined Medicare-Medicaid benefits to support whole-person, beneficiary-centered care for dual-eligible beneficiaries and their families.


Introduction
Medicare and Medicaid fund different benefits and services, and integration of the 2 programs is critical to better serve the 12 million people enrolled in both programs. 1Dual-eligible beneficiaries are, on average, sicker and frailer than other Medicare-or Medicaid-only beneficiaries because many are living with multiple chronic comorbidities and experience functional limitations, cognitive impairments, and mental health conditions. 2 The complex care needs of dual-eligible beneficiaries account for a disproportionate share of expenditures in both Medicare and Medicaid. 3Lack of coordination between programs creates misaligned incentives for payers and physicians, resulting in higher costs, fragmented care, and poor health outcomes. 4though evidence is limited, integrated models are intended to help align financial incentives with overall beneficiary experience and outcomes across the care continuum. 5,6Yet, relatively few dual-eligible beneficiaries are enrolled in integrated programs, such as the Program for All-Inclusive Care for the Elderly (PACE), or Medicare-Medicaid-managed care plans including the Financial Alignment Initiative (FAI), and Dual Special Needs Plans (D-SNPs).The PACE provides fully integrated financial, health care, and administrative processes; however, PACE models are challenging to scale.
Overall, D-SNPs are more widely used than PACE, with 3.8 million dual-eligible beneficiaries enrolled nationwide and varying degrees of coordination and integration between Medicare and Medicaid across states.Despite Centers for Medicare & Medicaid Services (CMS) and state efforts, only around 10% of dual-eligible beneficiaries nationally are enrolled in programs that integrate Medicare and Medicaid care models, payments, and administrative processes. 7pulation heterogeneity poses challenges in expanding access to, and enrollment in, integrated payment models.Beneficiaries enroll in Medicare based on age or long-term disability, and qualify for state Medicaid based on income or health care needs.Disabled adults comprise more than half of the dual-eligible population but only 15% of the general Medicare population. 8In addition to physical medical care, dually eligible beneficiaries often need behavioral health (BH) services, longterm services and supports (LTSS), and social supports. 3Long-term services and supports include both home and community-based services (HCBS) and facility-based care.Evaluations informing Medicare-Medicaid integration must consider average utilization and cost patterns of dual-enrollees across both programs, including among beneficiary subgroups with different health care needs.
0][11][12] The Medicare Payment Advisory Commission (MedPAC) has recently provided high-level data for the whole population, 8 but did not examine cost and use variations across key subgroups, such as BH users or HCBS waiver participants.4][15] However, services with the highest spending may vary for different subgroups due to diverse care needs and differing access to services through Medicaid waiver programs.
Evaluating Medicaid spending is challenging in states with high use of Medicaid-managed care organizations (MCO).Encounter data for MCOs are less accurate compared with fee-for-service (FFS) claims data, which are used for reimbursement. 16,17In North Carolina (NC), however, more than 95% of dual-eligible beneficiaries were served by FFS Medicaid in 2019.Thus, analysis of the NC dualeligible FFS population is more representative of the diversity of dual enrollees than in states where managed care is more prevalent.Compared with states with high managed care, evaluation of spending in NC may be more generalizable and applicable for policy recommendations due to availability of claims data on most of the Medicaid population.[20] Federal and state policy makers and administrators are currently developing strategies for Medicare-Medicaid integration. 21To inform the design of integrated programs that will meet the diverse needs of dual enrollees, we describe the Medicare and Medicaid health care use and spending for subgroups with different health care needs in NC.

Demographic Information
Patient demographics were extracted from the Medicaid member files, and Medicare values were used where Medicaid data were missing.Race and ethnicity data were self-reported in NC Medicaid and Medicare, and we retained missing values when unavailable in both data sources due to poor validity of imputation methods. 23Beneficiaries' county of residence was classified as rural or urban based on NC Department of Health and Human Services guidelines. 24The CMS Chronic Conditions Warehouse (CCW) indicators identified those with a condition date prior to or during the study period.

Utilization and Spending Measures
Utilization and spending outcomes were ascertained for months in which beneficiaries met the criteria for a full dual enrollee and were enrolled in Medicare FFS parts A and B and not Medicare Advantage (eTable 1 in the Supplement 1).We measured utilization separately for each payer, and measures were not summed to avoid overcounting services paid by both payers.For example, Medicare is the primary payer for most acute and postacute services, and Medicaid typically For spending, we summed Medicaid-and Medicare-financed spending for a given person.
Spending was calculated as the total amount paid by Medicaid and Medicare, with a per-diem adjustment included in inpatient spending for Medicare.Medicaid spending measures included all Medicaid FFS payments including professional, facility, and dental claims.Medicare spending included all parts A and B claims.Payments were adjusted for inflation using the medical care component of the Consumer Price Index (CPI) to calculate all spending in 2017 dollars.

Statistical Analysis
Descriptive characteristics for the overall cohort and by subgroup are reported using frequencies and proportions for categorical variables and median and interquartile ranges (IQRs) for continuous variables.Yearly utilization and spending amounts were calculated as the sum of events, services, days, or spending divided by the total person-years eligible for the utilization outcomes, with accompanying 95% confidence intervals.No statistical testing was conducted for this descriptive analysis.We used SAS (version 9.4; SAS Institute, Inc) for all analyses.The analyses were conducted between 2021 and 2022.

Dual Enrollment
Our cohort included 333 240 NC Medicaid beneficiaries with full Medicaid benefits ever enrolled in Medicare during the study period.The median (IQR) age was 65 (IQR, 52-76) years, and 61.1% reported female identity.The most common racial identities included White (58.7%),Black (36.1%), and Asian (1.8%) (Table 1).Most beneficiaries maintained their full dual status throughout the entire study period or until death (71.4%); however, about 1 in 3 dual-eligible beneficiaries who were fullbenefit dual-eligible (FBDE) at the start of the period lost their full-dual eligibility status, primarily via loss of Medicaid benefits (eTable 3 and eFigure 2 in Supplement 1).

Need-Based Subgroup Profiles
Dual-eligible beneficiaries could be included in multiple need-based subgroups over the 4-year period; see eTable 4 in Supplement 1 for overlap between subgroups.Most of the cohort, 64.1% (n = 213 667), were included in the community well subgroup, which included people who did not meet criteria for the other need-based subgroups at any point in the period.The HCBS subgroup included 15.0% (n = 50 095), BH subgroup included 15.2% (n = 50 509), and the nursing home subgroup included 7.5% (n = 24 927).
The need-based subgroups differed in their demographics and health status (

Service Utilization by Need-Based Subgroup
Because Medicare is the primary payer and Medicaid covers beneficiary copays and non-Medicare covered days, we expected both Medicare and Medicaid claims would exist for acute care events.
Our linked data showed that fewer acute care events, such as ED visit, hospital admission, and inpatient days, were observed in Medicaid than Medicare (

Discussion
We found substantial use of both Medicare-and Medicaid-funded services across all need-based subgroups, demonstrating the need for both Medicare and Medicaid claims to build an accurate picture of dual-eligible beneficiaries' needs to support care coordination, program evaluation, and policy administration.We also found that the proportion of total spending paid by Medicaid varied by need-based subgroup.Medicaid contributed up to 70% of total spending for high-need populations including nursing home residents.Overall, however, the financial burden of care for dual-enrollees was split evenly between Medicare and Medicaid.Our estimates for total spending among the dualeligible population in NC was comparable to national estimates using similar data. 9These findings demonstrate the diverse health care needs within the dual populations, which has implications for state and federal policies considering strategies for Medicare-Medicaid integration.
States, supported by CMS, have been implementing innovative payment and delivery models to address the shortcomings of the current system and advance value-based care for dual-eligible beneficiaries.Three notable models include the PACE, FAI, and D-SNPs.The PACE model provides full financial integration in addition to integrated care and administrative processes for individuals aged 55 years and older who require nursing facility level of care.The FAI model mostly tested a capitated model whereby states, CMS, and plans entered into a contract, called a Medicare-Medicaid Plan (MMP), for a blended capitation rate.Although CMS has closed the window for new enrollment in the Financial Alignment Initiative demonstrations, D-SNPs have permanent authorization under the Bipartisan Budget Act of 2018.These D-SNPs are required to contract with the states (via State Medicaid Agency Contracts, SMACs) in which they operate, in addition to contracts with CMS, and must adhere to state and CMS requirements.Many states with D-SNPs have leveraged these SMACs

JAMA Health Forum | Original Investigation Health
The state of NC is currently implementing Medicaid transformation from FFS to Medicaid value-based care models.We examined FFS spending Care Use and Spending Among Need-Based Subgroups of Medicare Beneficiaries With Full Medicaid JAMA Health Forum.2023;4(5):e230973.doi:10.1001/jamahealthforum.2023.0973(Reprinted) May 12, 2023 2/12 Downloaded From: https://jamanetwork.com/ on 09/21/2023 both overall and for need-based subgroups defined from the Medicaid program perspective to support federal and state decision-making.

JAMA Health Forum | Original Investigation
Health Care Use and Spending Among Need-Based Subgroups of Medicare Beneficiaries With Full Medicaid JAMA Health Forum.2023;4(5):e230973.doi:10.1001/jamahealthforum.2023.0973(Reprinted) May 12, 2023 3/12 Downloaded From: https://jamanetwork.com/ on 09/21/2023 contributes to acute and postacute spending by reimbursing clinicians for Medicare copays and coinsurance, or additional days beyond Medicare limits.We were unable to distinguish short-term from long-term postacute services in the Medicaid claims because the Medicaid billing codes do not differ between the 2 types of services.

Table 1
lower prevalence of beneficiaries older than 65 years (17.5%), higher prevalence of Black (37.2%) beneficiaries, and fewer chronic conditions.Compared with the community well, the nursing home subgroup had higher prevalence of beneficiaries older than 65 years (91.1%) and lower prevalence of Black beneficiaries (23.9%).The nursing home subgroup was the most likely to have 6 or more chronic conditions (88.7%), with a median of 11 chronic conditions, and experienced the highest mortality (55.4%) of all need-based subgroups.
Advantage enrollment was more common among nursing home residents (44%) than other groups.Compared with community well, HCBS users had higher prevalence of Black and rural beneficiaries and poorer health status, with a higher median number of chronic conditions(10), and higher mortality during the study period (30.7%).Compared with community well, the BH users had JAMA Health Forum | Original Investigation Health Care Use and Spending Among Need-Based Subgroups of Medicare Beneficiaries With Full Medicaid JAMA Health Forum.2023;4(5):e230973.doi:10.1001/jamahealthforum.2023.0973(Reprinted) May 12, 2023 4/12 Downloaded From: https://jamanetwork.com/ on 09/21/2023

Table 3 )
. The difference between acute events observed in Medicare and events observed in Medicaid claims was highest among the nursing home resident subgroup.For nursing facility use, the Medicare and Medicaid days often reflect different types of care.Medicaid nursing facility days are primarily long-term care, but also include Medicaid payments for post-acute care days that overlap with Medicare post-acute days.Medicare-covered skilled nursing facility days are most common among nursing home residents.Other utilization measures included home health days, hospice days, and total BH service visits.More home health visits and hospice days were observed in Medicare than Medicaid claims.More behavioral health service visits were observed in Medicaid than Medicare claims.

Table 1 .
Characteristics of Need-Based Subgroups Among North Carolina (NC) Full-Benefit Dual-Eligible Beneficiaries, 2014 to 2017 a,bHealth Care Use and Spending Among Need-Based Subgroups of Medicare Beneficiaries With Full Medicaid b 1915 waiver populations results presented in the eTable 2 in Supplement 1. c Mortality among beneficiaries who were full benefit dual-eligible in January 2014 (n = 206 874).JAMA Health Forum | Original Investigation

Table 2 .
Medicare Program Enrollment by Need-Based Subgroups, North Carolina (NC), 2014 to 2017 a,b Unless otherwised noted, data are reported as number (percentage) of beneficiaries.Includes NC Medicaid beneficiaries with full Medicaid benefits ever enrolled in Medicare during the study period.Beneficiaries may belong to 1 or more need-based subgroups within the study period.The values represent beneficiaries with any enrollment (Ն1 month) in Health Care Use and Spending Among Need-Based Subgroups of Medicare Beneficiaries With Full Medicaid a each benefit plan during the study period, out of all beneficiaries in that subgroup.Beneficiaries may switch plans over time, resulting in Parts A/B and Part C not summing to 100%.b 1915 Waiver populations results presented in eTables 2 and 3 Supplement 1. c Parts A and B were counted in the months where they enrolled in each program, respectively, but not enrolled in Plan C (ie, Parts A and B while FFS).JAMA Health Forum | Original Investigation JAMA Health Forum.2023;4(5):e230973.doi:10.1001/jamahealthforum.2023.0973(Reprinted) May 12, 2023 6/12 Downloaded From: https://jamanetwork.com/ on 09/21/2023

Table 3 .
Health Care Utilization Rates per Person-Year a by Need-Based Subgroups 21e primary vehicle for providing integrated services for full-benefit dual-eligible beneficiaries.For example, states can require D-SNP contracts in their state to meet CMS criteria for fully-integrated D-SNPs (FIDE-SNP).The implications of this research for state and federal policies are particularly timely given the potential legislation pending at the national level to expand access to Medicare-Medicaid integrated programs.To increase access to integrated care, policy makers should consider expansion of PACE to new regions and populations.Multiple Senate and House bills (H.R.6770, S.1162, S.3626)21propose expanding eligibility and geographic expansion for PACE, and other bills support integration planning at the state level (S.4264, S.4273, S.3630).Expansion of PACE can play an important role in future integration efforts in 2 ways.First, it allows more individuals who are eligible for PACE to enroll and receive integrated care.Second, PACE expansion increases opportunities for collaborative and innovative partnerships between non-PACE and PACE organizations serving beneficiaries.For example, in Massachusetts, managed care plans have partnered with PACE organizations to help deliver a PACE-like model in the community.In the absence of new legislation, states can use a combination of PACE and FIDE-SNPs to expand integrated options for dual-eligible beneficiaries.Given the overlap and churn between need-based groups, a single plan and program offering the broad range of services will minimize disruption to care during transitions where integration and continuity are most needed.Integrated plans should include comprehensive Medicare and Medicaid services, including LTSS, BH, social services and other supplemental benefits to support wholeperson care.Medicare benefit packages have historically been restricted; however, more flexible benefits became available with the introduction of Medicare Advantage special needs plans in 2006, North Carolina (NC) dual-eligible Medicare and Medicaid beneficiaries, 2014 to 2017, were identified from the Medicare 100% sample inpatient, skilled nursing facility, home health, outpatient, carrier, durable medical equipment, and hospice claims linked to NC Medicaid equivalent claims at the beneficiary level.The sample includes 283 322 patients with at least 1 month of dual-eligibility for full NC Medicaid benefits and Medicare FFS included in the analysis.Rates shown as spending per person-year.Skilled nursing facilities, nursing homes, and intermediate care facilities (SNF/NH/ICFs) represent the institutional spending for these locations.Medicaid inpatient spending is depicted in yellow but has low values compared with the other categories (range, $17.73-$169.11per person-year).BH indicates behavioral health; HCBS, home and community-based services.SUPPLEMENT 1. eFigure 1. Medicare-Medicaid linkage and cohort inclusion criteria eTable 1. Utilization and cost variable algorithms eTable 2. Characteristics of need-based subgroups among NC full-benefit dual-eligible beneficiaries, 2014-2017 eTable 3. Dual status in January 2014 by need-based subgroup eFigure 2. Changes in Dual Status Throughout the Study Period by Need-Based Subgroup, among those who were full-benefit dual eligible (FBDE) beneficiaries in January 2014 eTable 4. Overlap in Beneficiaries Between Need-Based Subgroups eTable 5. Proportion of total fee-for-service healthcare spending per person-year funded by Medicaid Programs among North Carolina full benefit dual eligible, overall eTable 6. Proportion of total fee-for-service healthcare spending per person-year funded by Medicaid Programs among Community Well beneficiaries eTable 7. Proportion of total fee-for-service healthcare spending per person-year funded by Medicaid Programs among 1915(c) Waiver beneficiaries eTable 8. Proportion of total fee-for-service healthcare spending per person-year funded by Medicaid Programs among Home-and Community-Based Service (HCBS) users eTable 9. Proportion of total fee-for-service healthcare spending per person-year funded by Medicaid Programs among Intensive Behavioral Health service users eTable 10.Proportion of total fee-for-service healthcare spending per person-year funded by Medicaid Programs among Nursing Home Residents b Services covered by Medicare only.c as and have expanded over time to include LTSS, BH, palliative care, and other support services.New regulations require FIDE-SNPs to provide comprehensive Medicare and Medicaid services, including either long-stay nursing home care or behavioral health services.States contracting with D-SNPs should ensure options for integrated D-SNPs with comprehensive benefits are available in all regions, including rural areas.Similar to PACE, integrated programs should conduct integrated assessments of dual-eligible beneficiaries' health risks, needs, and preferences to provide a comprehensive understanding of the whole person.A standard functional assessment tool would ensure development of an individualized person-centered care plan that is designed to meet the unique needs of high-risk beneficiaries and should be updated as needed to address beneficiaries' needs as they change over time and across care settings.Use of combined Medicare-Medicaid data can support implementation of tailored approaches for integration and care coordination to best serve each subgroup.Not only do Medicare and Medicaid fund different services, creating blind-spots in care, there are gaps in claims for services