Date of Award

January 2026

Document Type

Thesis

Degree Name

Medical Doctor (MD)

Department

Medicine

First Advisor

Sarwat I. Chaudhry

Second Advisor

Jose F. Figueroa

Abstract

Abstract

Background: Cardiovascular disease (CVD) is the leading cause of death in the US for people who are 65 years and older. Healthcare coverage has rapidly changed in recent years as over half of the Medicare population is now enrolled in Medicare Advantage (MA), the private alternative to traditional Medicare (TM). Within MA, private insurance plan design varies substantially, and the impact of a healthcare insurance plan is important especially for socially vulnerable adults (e.g., Medicaid-eligible). In particular, as beneficiaries with CVD are often high utilizers of healthcare resources, complex plan availability (e.g., number of plans, supplemental benefits, network design) may be detrimental as beneficiaries become overwhelmed and select plans poorly fit to their needs for choice complexity. Rather, beneficiaries with CVD may be better served by chronic condition special needs plans (C-SNPs) which are specialized MA plans exclusively focused on a particular disease with tailored benefits like care coordination, curated formularies, and transportation. These plans undergo review for their ability to provide care coordination, maintain sufficient specialized provider networks, and ensure beneficiary satisfaction. In contrast, there is an expanding market of nonspecialized conventional MA plans that may offer similar benefits but do not have disease eligibility criteria and do not undergo similar federal government scrutiny. To date, we have a limited understanding of the evolving healthcare coverage landscape for Medicare beneficiaries with CVD, how plan benefit design and plan quality in CVD C-SNPs compare with conventional MA plans, and the impact of CVD C-SNP enrollment on utilization and clinical outcomes for these beneficiaries.

Research Aims: The objectives of this research are to 1) describe enrollment patterns of beneficiaries with CVD in MA and plan choices available to them from 2016 to 2022, 2) understand how plan benefit design and plan quality in specialized CVD C-SNPs compare with nonspecialized conventional HMO/PPO MA plans (supplemental benefits as our primary outcome, cost-sharing and star ratings as secondary outcome), and importantly, 3) the impact of plan choice on clinical outcomes (mortality as our primary outcome) and utilization (readmissions, composite re-visit rates, and composite any outcome as a secondary outcome) for beneficiaries hospitalized with heart failure (HF) between specialized HF C-SNPs and nonspecialized conventional HMO/PPO plans.

Hypothesis or Question: In Aim 1, we hypothesize that more patients with CVD will enroll into MA as well as have more plan choices available to them over time. For Aim 2, we hypothesize that specialized CVD C-SNPs have more favorable plan designs (e.g., low to no-cost sharing, offering of meal and transportation benefits) and greater plan quality than non-specialized conventional HMO/PPO MA plans. For Aim 3, we hypothesize that beneficiaries with CVD in specialized plans will have increased utilization and better outcomes relative to beneficiaries enrolled in non-specialized plans.

Methods/approach: We used national Medicare data from 2016 to 2022. Medicare beneficiary-level information was obtained from the Medicare Master Beneficiary Summary File and plan-level information was obtained from the Plan Benefits Package data and the Contract Year 2025 Landscape file. To be eligible for the study, patients had to have diagnoses of arrythmias (atrial fibrillation or flutter), heart failure (HF), ischemic heart disease, or history of acute myocardial infarction derived from the Chronic Condition Data Warehouse. Medicare claims data for Traditional Medicare beneficiaries and Medicare Advantage encounter data was used to characterize utilization and health outcomes. Data on hospitalizations relied on the Medicare Provider Analysis and Review file. For Aim 1, traditional descriptive statistics were used to characterize and compare the specific types of insurance coverage and enrollment patterns of Medicare beneficiaries with CVD with a linear regression model adjusted for beneficiary sociodemographic attributes, qualifying reason for Medicare, dual-eligibility for Medicaid with fixed effects for year and county to assess changes in plan choice complexity within a given county over time (2016 - 2022). For Aim 2, Fisher’s exact test was used to determine differences in supplemental benefit offerings. Welch’s two-sample t-test was used to determine differences in Part C and Part D premiums. Wilcoxon rank-sum tests were used to determine differences in Part C, D, and Overall star ratings. For Aim 3, in a sample of MA beneficiaries hospitalized with HF in 2021-2022, inverse probability weighting of beneficiary sociodemographic attributes, clinical comorbidity profile, qualifying reason for Medicare, and dual-eligibility for Medicaid was conducted. Then, a series of multivariable logistic regression models adjusted for the aforementioned with state fixed effects and hospital random effects was used to determine rate differences in 30-day and 90-day outcomes of: mortality, readmission, composite hospital re-visit rates (inpatient stays, observation stays, emergency department visits), and any outcomes rates (composite of mortality, readmissions, and hospital re-visit rates).

Results: In our first analysis, we found that the number of beneficiaries with CVD increased as did the proportion of beneficiaries with CVD relative to the total number of beneficiaries from 2016 to 2022. By 2022, we observed a greater proportion of beneficiaries enrolled in MA versus TM (52.5% versus 47.5) in addition to a significant increase in the average potential plan availability (21.4 (95% CI: 21.0, 22.7) to 44.7 (95% CI: 44.2, 45.2)), especially among socially vulnerable full-benefit dual-eligible beneficiaries with CVD (30.4 (95% CI: 29.2, 31.6) to 58.3 (95% CI: 56.7, 59.8)). For our second analysis, examining plans available in 2025, the study sample included 3,560 plans representative of 21,943,772 beneficiaries (1,278,224 or 5.8% in C-SNPs; 20,665,548 or 94.2% in conventional MA plans). C-SNPs were more likely to offer meal and transportation benefits (6.2% vs 2.9%, p < 0.01; 33.3% vs 2.8%, P < 0.0001, respectively) with lower premiums (Part C: $0.59 vs $9.36, P < 0.0001; Part D: $5.76 vs $10.13, P < 0.0001). Conventional MA plans had greater average star ratings (Part C: 3.66 vs 3.74, P < 0.0001; Part D: 3.74 vs 3.76, P < 0.0001; Overall: 3.74 vs 3.79 P < 0.0001), indicating greater plan quality than CVD C-SNPs. For the third analysis, the study population was 198,210 MA beneficiaries hospitalized with HF (6.8% in C-SNPs for HF and 93.2% in conventional MA plans). Upon inverse probability weighting, beneficiary characteristics were well balanced. With adjustment, mortality was similar at 30-days (-0.23 percentage points (pp), (95% CI: -1.15, 0.70)) and 90-days (0.4 pp (95% CI: -0.85, 1.66) with no differences in readmission rates or revisit rates at 30-days or 90-days (30-day readmissions: -0.71 pp (95% CI: -1.89, 0.48); 90-day readmissions: 0.94 pp (95% CI: -0.49, 2.37); 30-day revisit rates: -0.83 pp (95% CI: -2.18, 0.52); 90-day revisit rates: 0.94 pp (95% CI: -0.58, 2.47).

Statement of scientific impact and relevance for communities of interest: In this study, we found that Medicare beneficiaries with CVD are increasingly enrolling into MA with greater plan choice available to them. Our study demonstrates that specialized CVD plans did not outperform nonspecialized conventional MA plans in plan quality, utilization, and outcomes despite the lower cost-sharing profile and favorable supplemental benefit offerings. Our findings suggest that the federal government may not be realizing the full potential for these specialized CVD plans, warranting scrutiny by policymakers interested in reforming and regulating the MA market that is increasingly enrolling beneficiaries with CVD into specialized CVD plans.

Comments

This thesis is restricted to Yale network users only. This thesis is permanently embargoed from public release.

Download

Share

COinS