Comparing HEDIS performance of Dual Eligible Special Needs Plans with other coverage types for dually eligible people

Abstract

People eligible for both Medicare and Medicaid coverage (“dually eligible individuals”) have lower levels of income and assets and often higher health care needs and costs than those eligible for Medicare but not Medicaid coverage. Their 3 most common Medicare coverage options are Medicare Advantage (MA) Dual Eligible Special Needs Plans (D-SNPs), non–D-SNP MA plans, and fee-for-service (FFS) Medicare with a stand-alone prescription drug plan. No prior study has examined clinical quality of care for dually eligible individuals across these 3 coverage types. To fill that void, we used logistic regression to compare these coverage types on 6 HEDIS measures of clinical quality of care that were available for both MA and FFS (constructed from claims files). D-SNPs and non–D-SNP MA plans significantly outperformed FFS for all 6 measures for dually eligible individuals, by approximately 5 percentage points for 2 measures and by 18–34 percentage points for the other 4 measures. For the 4 measures with the greatest advantage over FFS, performance was 3–8 percentage points higher in D-SNPs than in non–D-SNP MA plans.

Introduction

For those with both Medicare and Medicaid coverage (“dually eligible individuals”), 3 Medicare health care coverage options are most common: fee-for-service (FFS), a Dual Eligible Special Needs Plan (D-SNP), or a non–D-SNP Medicare Advantage (MA) plan. (Non-D-SNP MA plans include special needs plans catering to people with chronic conditions [C-SNP plans] and people who are institutionalized [I-SNP plans], as well as non–SNP MA plans. Among dually eligible non–D-SNP MA enrollees, fewer than 10% are in a C-SNP or I-SNP plan. This analysis excludes people with dual eligible coverage who are in I-SNPs. Because all people with dual-eligible coverage received Part D (PD) coverage, here both MA and FFS coverage always includes PD coverage; as such, “MA” refers to “MA-PD” and “FFS refers to “FFS + PD” in this paper.) D-SNPs are a type of MA plan intended to better meet the needs of dually eligible individuals and are restricted to such individuals. Through 2019, most dually eligible individuals selected FFS coverage, although this proportion has declined (from 80% in 2008 to 59% in 2019). Medicare Advantage enrollment for dually eligible individuals increased from 17% in 2008 to 41% in 2019¹; much of this can be attributed to D-SNP enrollment, which increased from 10% of all dually eligible individuals in 2008 to 22% in 2019 (12.3 million individuals in 2019).²

Compared with other people with Medicare, dually eligible individuals have higher health care needs and costs as well as lower levels of income and assets.³ Dually eligible individuals also face the challenge of navigating 2 complex public health insurance programs with different enrollment criteria, covered benefits, payment structures, and, in some cases, different provider networks.

To help mitigate these challenges, Section 2602 of the Patient Protection and Affordable Care Act (ACA; Pub L No. 111-148) created the Medicare-Medicaid Coordination Office to “make sure dually eligible individuals have full access to seamless, high-quality health care and to make the system as cost-effective as possible.”⁴ Medicaid covers services or cost-sharing not covered by Medicare, such as long-term care, premiums, and other out-of-pocket costs at the point of care.

D-SNPs were first authorized in the Medicare Prescription Drug, Improvement, and Modernization Act of 2003 (Pub L No. 108-173), and in 2018 were permanently re-authorized under the Bipartisan Budget Act of 2018 (Pub L No. 115-123). Insurance options for dually eligible people are continuing to evolve. Medicare-Medicaid Plans are an alternative to D-SNPs offered through demonstration projects but are being discontinued by 2025 (they enrolled 3% of people with dual coverage as of 2020).^5,6 There also are 3 types of D-SNPs with evolving criteria: Coordination Only, Highly-Integrated D-SNPs (HIDE-SNPs), and Fully Integrated D-SNPS (FIDE-SNPs). In 2020, 24% of all people with dual coverage were enrolled in coordination only D-SNPs, 3% were in HIDE- or FIDE-SNPs, and 51% had FFS coverage.

As an example of the ongoing evolution, Section 3205(b) of the ACA authorized an additional frailty payment by CMS for certain individuals enrolled in FIDE-SNPs. To implement this provision, CMS adopted a regulatory definition of FIDE-SNP, which applied beginning in 2012. FIDE-SNPs directly cover Medicaid benefits for enrollees. Hence, for FIDE-SNPs, the same insurer is at risk for both Medicare and Medicaid spending, which is intended to incentivize the most cost-effective mix of services. In addition, starting in 2013, all D-SNPs subtypes had to contract with a state Medicaid agency. These changes are intended to better integrate Medicaid and Medicare coverage for enrollees in D-SNPs. Research has shown that D-SNPs offer more supplemental benefits than non-SNP MA plans and are more profitable for plan sponsors.^7,8

Five studies have considered D-SNP performance for people with dual-eligible coverage relative to other coverage options. Haviland et al⁹ compared 2014–2019 experiences of care for people with dual-eligible coverage across the 3 different coverage types and compared performance in this period with performance in earlier years. Using data from 671 913 respondents to the Medicare Consumer Assessment of Healthcare Providers & Systems (MCAHPS) Survey (https://www.ma-pdpcahps.org/), they found that 2015–2019 immunization and overall ratings of care were higher for those in D-SNPs than in the other 2 coverage options. However, D-SNP enrollees did not report better experiences with coordination of care, doctors, or receiving care quickly than similar people in the other 2 coverage options. This study found that experiences with care reported by D-SNP enrollees improved from earlier to later in the study period, making care in D-SNPs more similar to care provided by other coverage types, except regarding immunizations.

Roberts and Mellor¹⁰ used 2015–2019 Medicare Current Beneficiary Survey data to compare experiences of people with dual-eligible coverage across the 3 coverage types on measures of access to care, use of care, and satisfaction with care. They restricted their analysis to those eligible for full Medicaid benefits (including Medicaid services not covered by Medicare as well as coverage of some Medicare cost sharing; see Appendix A). The study used 9885 respondent-year observations and was limited by modest sample sizes. The D-SNP enrollees reported better access to dental care and greater satisfaction with out-of-pocket care expenses and access to specialists than people with dual-eligible coverage in non–D-SNP MA plans. Compared with people with dual-eligible coverage in FFS, D-SNP enrollees also reported better access to a primary care provider, higher rates of 2 of 3 preventive care measures (including an immunization), and greater ease of getting to a doctor from home. Full-benefit D-SNP enrollees did not report better performance on the measures the authors considered related to coordination of care.

A third study compared the use of services for people with dual-eligible coverage in 2 of the 3 coverage types: D-SNPs and non–D-SNP MA (but not the most common option, FFS).¹¹ Compared with people with dual-eligible coverage in non–D-SNP MA plans, D-SNP enrollees had lower hospital and nursing facility admission rates and higher rates of home- and community-based services.

The fourth study is 1 of only 2 studies to consider Healthcare Effectiveness Data and Information Set (HEDIS) measures of clinical quality and was conducted by the Medicare Payment Advisory Commission.⁸ This study compared 2016 clinical quality for people with dual-eligible coverage and enrolled in D-SNPs or non–D-SNP MA (not FFS) but only among a subset of those with dual coverage, those with partial Medicaid benefits (see Appendix A). Those with partial Medicaid benefits comprise approximately 26% of D-SNP enrollees. Medicaid covers any MA premium for those with partial Medicaid coverage but only pays cost-sharing for other services covered by Medicare in some partial Medicaid coverage levels (see Appendix A). For this subgroup, D-SNP enrollees received recommended clinical care at a rate similar to that of non–D-SNP MA plan enrollees on 35 of 39 HEDIS measures for those younger than 65 and 36 of 42 HEDIS measures for those aged 65 or older. For those younger than 65, rates of receiving recommended care were higher in D-SNPs for 2 of the other 4 measures and lower in D-SNPs for the other 2. For those aged 65 or older, rates of receiving recommended care were higher in D-SNPs for 4 of the other 6 measures and lower in D-SNPs for the other 2.

The fifth study, also conducted by the Medicare Payment Advisory Commission,¹² compared clinical performance on 33 HEDIS measures between 5 different types of MA plans that are available to people with dual-eligible coverage who select MA coverage. They compared 3 groupings of D-SNPs (Coordination Only, Unaligned HIDE-SNPs and FIDE-SNPs,¹³ and Aligned HIDE-SNPs and FIDE-SNPs) with Medicare-Medicaid plans and other MA plans with dually eligible enrollees. Across the 5 types of plans and 33 measures, results did not indicate that any of the plan types had consistently better or worse performance than the others. No comparison was made to clinical performance for dually eligible people with FFS coverage.

Overall, D-SNPs have shown similar performance to non–D-SNP MA and FFS on patient experience (including coordination of care), except for higher flu immunization levels. D-SNPs have shown similar performance on clinical measures (HEDIS) for different subgroups of people with dual coverage relative to different types of MA coverage, but no comparisons to FFS (the most common coverage type for people with dual coverage) have been made to date.

In what follows, we compare the performance of D-SNPs relative to both of the other 2 most common coverage types, FFS and non–D-SNP MA, for all people with dual-eligible coverage on 6 HEDIS measures of clinical quality. This study goes beyond the 5 studies discussed above by considering quality of clinical care for all those eligible for both Medicare and Medicaid and by comparing clinical quality across all 3 coverage types, including FFS, the most common coverage type for people with dual coverage.

Data and methods

Analysis

HEDIS data from reporting years 2018 and 2019 were combined for people dually eligible for Medicare and Medicaid (including those with both partial and full Medicaid) with MA and FFS Medicare coverage to examine differences across the 3 main coverage types (D-SNPs, non–D-SNP MA, and FFS) for the 6 clinical quality-care measures described above. Enrollee-level HEDIS data were submitted by MA organizations. Analogs to HEDIS measure scores for people with FFS coverage were constructed from 100% Medicare FFS claims: 2015–2019 FFS inpatient, outpatient, carrier, and hospice claims file and Medicare enrollment files using HEDIS 2018 specifications¹⁴. The following additional inclusion criteria were applied to the lookback period for each measure: (1) no months of enrollment in MA, (2) no hospice utilization, and (3) no carrier claims where Medicare was not the primary payer.¹⁵

The primary comparison of the 3 coverage types excludes US territories, as their dual-eligible coverage options differ from those in the 50 states and Washington, D.C.¹⁶ As the main focus is on D-SNP quality of care, we also consider 2 versions of D-SNP coverage, both excluding and including US territories, adding a fourth group to our analysis: D-SNPs including US territories.

To assess differences in performance by coverage type, we estimated the proportion of dually eligible individuals who received the recommended clinical care for each of the 6 HEDIS measures by coverage type. We tested for statistically (P < .05) and practically (at least 3 percentage points) significant mean differences of the D-SNP and non–D-SNP MA estimates compared with FFS, as the majority of 2019 dually eligible people had FFS coverage. Estimates and statistical tests are from logistic regressions predicting each HEDIS measure from D-SNP and non–D-SNP MA indicators. Models were run twice, with and without territories. Sampling weights were used for the 3 measures for which plans may choose a census or a sample: Colorectal Cancer Screening; Diabetes Care: Nephropathy; and Diabetes Care: Retinal Eye Exam. As the data for each HEDIS measure were either a census or a stratified random sample from all plans (for D-SNPs and non-SNP MA) and states (for FFS), the sampling procedure creates no dependencies necessitating clustering of standard errors.¹⁷

Finally, pooled 2018–2019 MCAHPS data¹⁸ were used to describe the demographic and health characteristics of all people with dual-eligible coverage across coverage types, since eligibility varies by HEDIS measures and most of the measures pertained to specific subpopulations, such as females (eg, Breast Cancer Screening) or restricted age groupings. We used chi-square procedures to test for significant differences (P < .05) across coverage groups.

Similar demographic tables were produced for the population eligible for each HEDIS measure Appendixes B–G), although only a subset of the characteristics listed above were available within the HEDIS data: sex, age, disability status, rurality of residence, residential census division, and linked race-and-ethnicity probabilities.¹⁹

Data sources

Data come from HEDIS reporting years 2018 and 2019. HEDIS is a comprehensive set of standardized performance measures across 6 domains of care: Effectiveness of Care, Access/Availability of Care, Experience of Care, Utilization and Risk Adjusted Utilization, Health Plan Descriptive Information, and Measures Reported Using Electronic Clinical Data Systems.

MCAHPS 2018–2019 survey data¹⁸ were used to describe the demographic and health characteristics of dually eligible people within each of the 3 coverage types: D-SNPs, non–D-SNP MA, and FFS. Sampling weights were created to account for sample design and nonresponse.²⁰

Outcome measures

We used the only 6 HEDIS measures that are available for both MA and FFS (in analogous forms); they were all included in the 2019 MA and Part D Star Ratings²¹ and are tied to quality bonus payments for MA plans (see Appendix H).

Dual eligibility status

Anyone with Medicare who was additionally eligible for Medicaid or enrolled in a D-SNP in CMS's Integrated Data Repository was counted as a person with dual-eligible coverage.

Demographic and health characteristics

Data on race and ethnicity, education, and general and mental health status (each poor to excellent) were self-reported on the MCAHPS survey. Data on age, sex, disability status, rurality, and residential census division came from CMS administrative records.

People who had disability insurance benefits as their original reason for Medicare entitlement in CMS’s Medicare Beneficiary Summary File were classified as having a disability.

People were classified as living in a rural or urban area based on the zip code of their mailing address and the corresponding US Census Bureau core-based statistical area (CBSA). We classified anyone in a metropolitan statistical area (anywhere with a core urban area with a population at least 50 000) as an urban resident; anyone living in a micropolitan statistical area or outside of a CBSA was classified as a rural resident.

Limitations

Relative to MA plans, there is limited information on clinical quality, as assessed by HEDIS measures, for those with FFS coverage. Medicare Advantage contracts are required to report 45 HEDIS measures and D-SNPs are required to separately report at the benefit package level on 23 of these 45 measures. For people with FFS coverage, only 6 HEDIS measures have been constructed for the purpose of comparing quality on measures used in MA Star Ratings.¹⁵ To compare with the FFS coverage option, which was selected by a majority of dual-eligible individuals during the study period, we restricted our analysis to only these 6 measures. This is a limitation, and we note that performance on these 6 measures may not reflect performance on a wider set of clinical measures. Further, for HEDIS measures reported using the hybrid method, FFS performance may be underestimated because medical record reviews were not feasible. In addition, people with dual-eligible coverage who select different coverage types may differ in ways that are related to performance on these clinical measures. While we follow current practices of not applying case-mix adjustment to HEDIS measures, differences in performance across coverage types may be due not only to the coverage type but also to other confounding factors.

Results

Table 1 displays the proportion of people with dual-eligible coverage among all people with Medicare who are eligible for each of the 6 HEDIS measures, within each coverage type. The D-SNPs serve only people with dual-eligible coverage. In non–D-SNP MA plans, 8%–15% of those eligible for the 6 HEDIS measures were people with dual-eligible coverage. In FFS, 19%–40% of those meeting the denominator definitions for each of the 6 HEDIS measures were people with dual-eligible coverage.

Table 2 shows the distribution of coverage type among people with dual-eligible coverage who were eligible for each HEDIS measure. Disease-Modifying Anti-Rheumatic Drug Therapy for Rheumatoid Arthritis and Osteoporosis Management in Women Who Had a Fracture had the smallest denominators of all 6 measures. Among people with dual-eligible coverage, the majority of enrollees eligible for each HEDIS measure had FFS coverage (64%–95%), 3%–18% were enrolled in D-SNPs, and 2%–18% were enrolled in non–D-SNP MA plans.

Figure 1 shows the estimated proportion of people with dual-eligible coverage getting recommended care on each HEDIS measure, by coverage type. For all measures, proportions of people with dual-eligible coverage getting recommended care were significantly higher (both statistically significant at the 5% level and at least 3 percentage points in magnitude) for those with either MA coverage type relative to FFS. Several of these differences are larger than 20 percentage points. D-SNPs and non–D-SNP MA plans had similar performance on some measures and, where there were differences, estimates of performance were higher in D-SNPs relative to non–D-SNP MA plans by 3 to 8 percentage points. The largest differences in performance between the 3 coverage types were observed for Breast Cancer Screening (FFS: 53%; non–D-SNP MA: 71%; D-SNP: 75%); Colorectal Cancer Screening (FFS: 41%; non–D-SNP MA: 72%; D-SNP: 75%); Diabetes Care: Retinal Eye Exam (FFS: 49%; non–D-SNP MA: 75%; D-SNP: 79%); and Osteoporosis Management in Women Who Had a Fracture (FFS: 25%; non–D-SNP MA: 44%; D-SNP: 51%).

When enrollees residing in a US territory were included in the D-SNP population, overall performance for D-SNPs increased slightly (by 0.2%–1.0%) across all measures, except for Colorectal Cancer Screening and Diabetes Care: Nephropathy. The largest increases were by 1 percentage point and were observed for Breast Cancer Screening and Osteoporosis Management in Women Who Had a Fracture.

Table 3 describes the demographic characteristics of people with dual-eligible coverage within each coverage type. Non–D-SNP MA enrollees were generally more similar to D-SNP enrollees than people with FFS coverage, with a few exceptions. D-SNP enrollees were more likely to be 64 years or younger, have a disability, reside in urban areas, reside in the mid-Atlantic census division, have lower educational attainment, and be Hispanic or Black than non–D-SNP MA enrollees. Both D-SNP and non–D-SNP MA enrollees were statistically different than people with FFS coverage for almost all levels of all demographic characteristics. D-SNP enrollees were more likely to be 70 to 79 years old, be female, reside in urban areas, reside in the mid-Atlantic and South Atlantic census divisions, be in better general and mental health, and be Hispanic and Black than people with FFS coverage. They also were less likely to have a disability and had lower mean educational attainment than people with FFS coverage. Non–D-SNP MA enrollees more often were 65 years or older, female, resided in urban areas and the Pacific census division, were in worse general health, and were Hispanic compared with people with FFS coverage. They also less often had a disability and had lower mean educational attainment than people with FFS coverage.

For each HEDIS measure, Appendixes B–G describe a more limited set of demographic characteristics of the measure-eligible enrollee population among people with dual-eligible coverage within each coverage type.

Discussion

This article summarizes differences in clinical quality-of-care performance among people with dual-eligible coverage between the 3 most common coverage types: D-SNP, non–D-SNP MA, and FFS. D-SNPs and non-D-SNP MA plans significantly outperformed FFS coverage for all 6 HEDIS measures included in this analysis. Several of the improvements in rates of receiving recommended care were substantial. For 2 measures—Diabetes Care: Nephropathy and Disease-Modifying Anti-Rheumatic Drug Therapy for Rheumatoid Arthritis—the improvements for people with dual-eligible coverage in either MA plan type were approximately 5 percentage points; for the other 4 measures—Breast Cancer Screening, Colorectal Cancer Screening, Diabetes Care: Retinal Eye Exam, and Osteoporosis Management in Women Who Had a Fracture—the improvements were between 18 and 34 percentage points. For the 2 measures with modest improvements (5 percentage points) in MA relative to FFS, D-SNPs had similar performance to non–D-SNP MA plans. For the other 4 measures, estimates of performance were 3 to 8 percentage points higher in D-SNPs than in non–D-SNP MA plans. Additional work is needed to understand how D-SNPs are attaining higher performance on measures such as Osteoporosis Management in Women Who Had a Fracture, where rates were 7.8 percentage points higher in D-SNPs than in non–SNP MA plans.

This study has several limitations. First, we cannot infer from this cross-sectional analysis that the reported associations are causal. Despite our efforts to control statistically for data and sampling differences, it is possible that quality differences are partly or wholly due to active selection by individual beneficiaries (or brokers) of coverage types, the availability of plans in geographic areas, or other unmeasured confounders. Second, our conclusion that D-SNPs provide better quality care for people with dual-eligible coverage than FFS is based on a limited number of quality measure (although these were the only ones available to support this analysis). Finally, for measures reported using the hybrid method, FFS performance may be underreported using the FFS claims construction method. We note, however, that 1 of the 2 measures showing small improvements and 2 of the 4 measures showing large improvements in MA relative to FFS were hybrid measures. This suggests that the possible underreporting in FFS does not account for the differences observed. This issue does not impact the comparison between D-SNPs and non–D-SNP MA plans.

Conclusion

Although less than one-third of people with dual-eligible coverage were enrolled in D-SNPs in 2019, our analysis suggests that these plans may be superior to other coverage options for meeting the clinical care needs of this subset of the Medicare population. This superior performance may be the result of the substantial oversight required by the Medicare Modernization Act of 2003 (Pub L No. 108-173) and the Bipartisan Budget Act of 2018 (Pub L No. 115-123),²² integration of Medicare and Medicaid coverage under FIDE-SNPs, or investment of pay-for-performance subsidies by D-SNPs into additional benefits and support services for people with dual-eligible coverage. Additional research is needed to investigate these and other mechanisms that may underlie this superior performance. Lessons learned from such analyses could be used to improve clinical care for those with dual-eligible coverage enrolled in Medicare FFS, which was found to consistently underperform both D-SNPs and non–D-SNP MA plans in this regard.

Acknowledgments

The authors thank Biayna Darabidian, MPP, and Katherine Osby, BA, for help with preparation of the manuscript.

Supplementary material

Supplementary material is available at Health Affairs Scholar online.

Funding

This study was funded by the Centers for Medicare & Medicaid Services (contract/task order no. GS-10F-0275P/75FCMC20F0101).

Notes

Author notes