Providing better care for dually insured Medicare/Medicaid beneficiaries with advanced chronic illness

Kim et al.¹ have reported one of the few published large-scale, multi-state evaluations of the impact of enrolling patients in Fully-Integrated Dual Eligible (FIDE) Special Needs Plan (SNP) insurance models available to Medicare and Medicaid beneficiaries, compared with enrolling in standard SNP duals plans (D-SNPs). The authors found no overall impact of increased financial integration at the plan level on a range of outcomes that integration of care at the patient-provider level have impacted.^{2, 3} However, for two higher-risk dual subgroups, either those using home- and community-based services or those meeting modified Independence at Home high needs qualifying criteria,⁴ Kim et al. found FIDE beneficiaries were likely to have lower hospitalization rates, compared to patients in dually eligible insurance models with less required integration.

It is dispiriting to see essentially no overall difference in service use patterns between individuals enrolled in FIDE plans, where Medicaid funds that support Long Term Services and Supports (LTSS) are managed by the same legal entity as Medicare funds, and other D-SNPs which have variable levels of coordination between Medicare and Medicaid services. This finding follows a decade when D-SNP growth reached 40% of eligible Duals, and FIDE SNP growth 8% of D-SNP enrollees. This finding is in line with other work by several of the study's authors, for example, noting a modest increase in Home and Community Based Services (HCBS) services but little impact on outcomes in Pennsylvania's effort to integrate care in its managed LTSS program, Community Health Choices.⁵ The finding of limited impacts suggests that a more robust, better targeted, intervention is needed than simply merging the pools of dollars and encouraging health plans to find ways to integrate care. Still, the strategy of combining Medicare and Medicaid funding under a single private sector insurer, hopefully lessening the artificial separation of medical care and community-based supports, continues to be favored by many subject experts and legislators as a means of reducing care fragmentation and blunting the cost curve for the most vulnerable and costly US citizens.^6-9

While funding may be integrated at the plan level in these new models, the Medicare and Medicaid benefits are often structurally separate, or are operationally separate due to organizational culture. Due to data limitations, the authors were unable to test the effects of different organizational cultures or structures on outcomes, nor the strength of the integration in some obvious ways, such as testing whether FIDE plans were better able to substitute less expensive Medicaid HCBS services for more expensive Medicare services and institutional long-term care. Ultimately, we still lack insight into how well the top-level financial integration translates into operational structures and care processes to drive change.

In Table 1, we list and contrast models where care integration is centered on provider teams, and models where integration relies more on plan-level financial administration and care coordination, ordered by decreasing levels of integration. In contrast to payer-managed care relying on a centralized administrative care coordination approach, approaches to provider-integrated care have proved effective at reducing combined costs for high-cost dual-eligible beneficiaries, with the pooled funds linked to an effective local, personal-scale clinical model that is specific to the needs of the targeted population and is motivated to provide parsimonious care through value-based funding.

One successful model is the Program for All-inclusive Care of the Elderly (PACE), a full-risk contract focused on vulnerable patients, almost all dually insured by Medicare and Medicaid, who are chronically ill and are nursing home-eligible at entry. PACE organizations have shifted care away from nursing homes and hospitals, with half as many hospital admissions for PACE patients than a well-matched comparison group in one analysis,¹⁰ while delaying long-term institutionalization of nearly 20 months.¹¹ In PACE, Medicaid and Medicare resources are managed by an interdisciplinary team (IDT) that directly cares for participants, with the PACE regulations holding the team responsible for providing any services needed to promote or maintain health. This differs from the brokered arrangement more typical of SNPs where care managers may be disconnected from care delivery. The broad PACE mandate encourages and empowers IDTs to creatively deploy resources where needed rather than to follow traditional rules. The impact PACE has on the lives of participants, and healthcare utilization, comes at the cost of a major capital investment in a PACE center, which functions as both social center and full service clinic, plus higher overhead due to mandated IDT personnel and administrative costs.¹² PACE is complex to operate, requiring coordination and delivery of the range of services provided by large insurers, delivered to a population typically under 750 beneficiaries. While the small size fosters trust between the IDT and participants, PACE growth has been hindered by the challenge of necessary capital investment, and by being both provider and health plan, offering broad insurance benefits to small populations. A recent ASPE analysis by RTI found PACE, among integrated care plan types, to consistently stand out as a high performer to reduce unnecessary institutional care.²

The need to restructure beneficiaries' insurance and provider networks, a hindrance to growth for staff models like PACE, was a major barrier to managed care-based financial alignment models operated in the first decade of the ACA by the Medicare-Medicaid Coordination Office.¹³ Of the 13 models CMS tested, all but two were managed care integration models, with inconsistent findings on reduced nursing home stays, and generally with increased Medicare and Medicaid spending. The relatively successful model operated by Washington state, one of the two managed fee-for-service models, used well-organized primary care health homes anchored by Area Agencies on Aging, essentially setting a global budget covering both LTSS and medical providers in a shared savings arrangement. Through Performance Year 8, overall Medicare savings were 8.6%, with significant declines in both long-term and skilled nursing facility use. The model effectively links provider teams and patients to resources, the goal of integrated care.¹⁴ The Washington Health Home FAI is an integration model that offers a targeted approach to high-needs beneficiaries built around care teams that are interdisciplinary, albeit interorganizational.

As highlighted in Table 2, another targeted approach uses home-based primary care (HBPC), demonstrated to be a successful clinical model for selected patients inside the Department of Veterans Affairs health system.¹⁵ HBPC has also been tested at individual sites in Medicare fee-for-service, and then in the multisite Independence at Home (IAH) demonstration, a Medicare shared savings program financial design. Here, finances are linked with a clinical model of care organized by a small interprofessional mobile primary care team, focused on managing the patient at home, saving money by reducing hospital and nursing home use. Most individuals nearing life's end prefer to remain at home, and to eventually die at home; they can do so only when appropriately supported. Clinicians who visit the home have unique and more accurate perspectives on patients' needs and earn patient and family trust in the provider team. In the IAH demonstration, using the HBPC clinical team model, targeted for high-needs cases, hospitalizations were reduced 19%.¹⁶ An analysis of a subset of IAH sites suggests that substantial reductions can be made in long-term institutional care, which is primarily funded by Medicaid,¹⁷ through close integration with local LTSS providers.

Under IAH, total Medicare costs were lower by 9% on average through demonstration Year 5, and up to 18% by Year 8.¹⁶ Using the prospectively modeled cost targets generated by CMS with a calibrated risk-adjustment model, savings exceeded $5100 per beneficiary per year by 2019, the year before the pandemic. A successor CMMI demonstration focused on similar high-need individuals, delivered by mobile interdisciplinary HBPC teams, found similar savings in 2021–2022.¹⁸

These successful examples highlight the importance of integrating services at the patient level by an IDT, rather than at the contractual level by management through an external care manager. Patient-level integration may be essential to impact the clinical trajectories of complex dual-eligible beneficiaries. The use of pooled funds managed by the interprofessional team is both to support the care plan for patients, including both clinical and social supports, as well as to support the interprofessional team itself, such that the team can attract and retain the needed workforce. Like HBPC and PACE, the Washington FAI provides a structure to facilitate integration at the patient level, without requiring a mediating insurer. There is nothing preventing FIDE SNPs from adopting clinical models of care that have been shown to help vulnerable patients remain in the community and that are vested in local clinical care teams that are familiar with the local resources and familiar to the patient. The difficulty and cost of such focused, interdisciplinary care requires focusing efforts on a higher-needs subset of dual beneficiaries. What is clear from the Kim study is that simply combining funds at an insurer level, while simple from a governmental contracting perspective, is no guarantee of integrating care nor of effectiveness at the patient-provider interface. Future integration studies should broaden their vision beyond payer managed care, and for FIDE-type plans should focus on measures of integration at the “tip of the spear,” to identify more effective practices.

All the authors contributed to the concept, design, data acquisition, analysis, and preparation of the manuscript.

Dr. Kinosian has participated in PACE since 1998; Drs. Boling, Taler, and Kinosian were participants in the Independence at Home demonstration, and currently participate in the High Needs ACO-REACH demonstration.

None.