Comparison of Health Care Utilization by Medicare Advantage and Traditional Medicare Beneficiaries With Complex Care Needs

Key Points

Question  Do rates of health care utilization for beneficiaries with complex care needs differ between traditional Medicare and Medicare Advantage (MA)?

Findings  In this cross-sectional study of 1 844 326 Medicare beneficiaries, those enrolled in MA had lower rates of hospital stays, emergency department visits, and 30-day readmissions. The largest relative differences were observed for hospital stays, which ranged from −9.3% to −11.9% across different cohorts of beneficiaries with complex care needs.

Meaning  Among Medicare beneficiaries with complex care needs, those enrolled in MA had lower rates of acute care utilization, suggesting that managed care activities in MA may influence the nature and quality of care provided to these beneficiaries.

Importance  Medicare beneficiaries with co-occurring chronic conditions and complex care needs experience high rates of acute care utilization and poor outcomes. These patterns are well described among traditional Medicare (TM) beneficiaries, but less is known about outcomes among Medicare Advantage (MA) beneficiaries. Compared with TM, MA plans have additional levers to potentially address beneficiary needs, such as network design, care management, supplemental benefits, and value-based contracting.

Objective  To compare health care utilization for MA and TM beneficiaries with complex care needs.

Design, Setting, and Participants  This cross-sectional study analyzed beneficiaries enrolled in MA and TM using claims data from a large, national MA insurer and a random 5% sample of TM beneficiaries. Beneficiaries were segmented into the following cohorts: frail elderly, major complex chronic, and minor complex chronic. Regression models estimated the association between MA enrollment and health care utilization in 2018, using inverse probability of treatment weighting to balance the MA and TM cohorts on observable characteristics. The study period was January 1, 2017, through December 31, 2018. All analyses were conducted from December 2020 to August 2022.

Exposures  Enrollment in MA vs TM.

Main Outcomes and Measures  Hospital stays (inpatient admissions and observation stays), emergency department (ED) visits, and 30-day readmissions.

Results  Among a study population of 1 844 326 Medicare beneficiaries (mean [SD] age, 75.6 [7.1] years; 1 021 479 [55.4%] women; 1 524 458 [82.7%] White; 223 377 [12.1%] with Medicare-Medicaid dual eligibility), 1 177 896 (63.9%) were enrolled in MA and 666 430 (36.1%) in TM. Beneficiary distribution across cohorts was as follows: frail elderly, 116 047 with MA (10.0% of the MA sample) and 104 036 with TM (15.6% of the TM sample); major complex chronic, 320 954 (27.2%) and 158 811 (23.8%), respectively; and minor complex chronic, 740 895 (62.9%) and 403 583 (60.6%), respectively. Beneficiaries enrolled in MA had lower rates of hospital stays, ED visits, and 30-day readmissions. The largest relative differences were observed for hospital stays, which ranged from −9.3% (95% CI, −10.9% to −7.7%) for the frail elderly cohort to −11.9% (95% CI, −13.2% to −10.7%) for the major complex chronic cohort.

Conclusions and Relevance  In this cross-sectional study of Medicare beneficiaries with complex care needs, those enrolled in MA had lower rates of hospital stays, ED visits, and 30-day readmissions than similar beneficiaries enrolled in TM, suggesting that managed care activities in MA may influence the nature and quality of care provided to these beneficiaries.

Medicare beneficiaries with complex, co-occurring, chronic conditions experience high rates of acute care utilization and poor outcomes.^1-3 These patterns have been well described among beneficiaries enrolled in traditional Medicare (TM) and attributed, in part, to a fragmented delivery and financing system.^1-4 To date, efforts at payment and delivery system reform in TM have not been consistently linked to improvements in quality and outcomes for these populations.^5-8

Less is known about outcomes for beneficiaries with complex care needs enrolled in Medicare Advantage (MA). Medicare Advantage plans now manage the financing and delivery of care for approximately one-half of Medicare beneficiaries,⁹ an increasing share of whom have complex needs.¹⁰ Compared with TM, MA plans have additional levers to potentially address beneficiary needs, such as network design, care management, supplemental benefits, and value-based contracting. As enrollment in the MA program grows and Medicare beneficiaries develop increasingly complex care needs, it is important to understand whether outcomes for these beneficiaries differ between MA and TM. We used data from a large, national MA insurer and a national sample of TM beneficiaries to compare health care utilization for Medicare beneficiaries with complex care needs.

In this cross-sectional study, MA beneficiaries were identified from individuals enrolled in plans offered by a large, national insurer. Traditional Medicare beneficiaries were identified from a random 5% national sample. By using a previously developed taxonomy,^1,2 we identified beneficiaries with complex care needs on the basis of claims from January 1, 2017, to December 31, 2017, and segmented these beneficiaries into the following mutually exclusive cohorts: frail elderly, major complex chronic, and minor complex chronic. We limited our sample to beneficiaries aged 65 years or older as of January 1, 2018, who were alive and continuously enrolled in either MA alone or TM alone from January 1, 2017, through December 31, 2018. We excluded from both populations beneficiaries with end-stage kidney disease and those who enrolled in hospice at any time from January 1, 2017, through December 31, 2018. From the MA population, we excluded beneficiaries whose providers delegated claims processing to a third party and those in plans contractually excluded from research. From the TM population, we excluded beneficiaries without both Part A and Part B coverage. Additional detail on the taxonomy and cohort construction is provided in the eMethods in the Supplement.

This study was reviewed by the Humana Healthcare Research Human Subject Protection Office, which determined it not to be human participant research. The study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines.

Utilization measures included inpatient admissions, observation stays, emergency department (ED) visits, and all-cause 30-day readmissions. We grouped inpatient admissions and observation stays together as hospital stays. Our measure of 30-day readmissions identified individuals experiencing a hospital stay within 30 days of discharge from another hospital stay. In addition to all-cause hospital stays and ED visits, we identified potentially avoidable events using the Agency for Healthcare Research and Quality Prevention Quality Indicators definition¹¹ for potentially avoidable hospital stays and the New York University ED visit algorithm¹² for potentially avoidable ED visits (additional detail provided in the eMethods in the Supplement). All utilization measures were assessed using medical claims from January 1, 2018, through December 31, 2018.

For each beneficiary, we extracted age, sex, race, Medicare-Medicaid dual eligibility status, disability status, and county of residence as of January 1, 2018, from enrollment files. Race was assessed according to the Centers for Medicare & Medicaid Services (CMS) beneficiary race code and categorized as Black, White, and other (Asian, Hispanic, North American Native, other, and unknown). County of residence was used to assign beneficiaries to a US Census region and a core-based statistical area and to classify the population density (urban, suburban, or rural) of their residence. We assessed the following 2 measures of comorbidity for each beneficiary: the CMS Hierarchical Conditions Category (CMS-HCC) score and a claims-based frailty index.¹³ For MA beneficiaries, monthly CMS-HCC scores from January 2018 to December 2018 were obtained directly from the CMS and averaged to construct an annual value for 2018. These 2018 CMS-HCC scores reflect comorbidities documented in 2017 and demographic characteristics in 2018. For TM beneficiaries, we calculated 2018 CMS-HCC scores using medical claims from 2017 and demographic characteristics from 2018 according to the logic used by the CMS.¹⁴ Frailty was assessed using claims from July 1, 2017, to December 31, 2017, for all beneficiaries.

For MA beneficiaries, we used enrollment data from the MA insurer to determine whether the beneficiary was enrolled in a health maintenance organization (HMO) or preferred provider organization (PPO) plan. For MA beneficiaries in HMO plans, we used contract data to identify the payment arrangement for their attributed primary care organization, classified according to the following taxonomy: fee for service, shared savings with upside-only financial risk (upside-only risk), and shared savings with upside and downside financial risk (2-sided risk). Additional detail is provided in the eMethods in the Supplement.

We used inverse probability of treatment weighting (IPTW)¹⁵ to balance the MA and TM populations on observable differences. To implement the IPTW approach, we began by estimating the propensity to enroll in MA using logistic regression models with the following covariates: age category (65-69, 70-74, 75-79, 80-84, or 85-97 years), sex, race, dual eligibility, disability, US Census region, county-level population density, CMS-HCC score, and frailty index. Before IPTW, CMS-HCC scores for MA beneficiaries were deflated by 5.91%, consistent with the coding intensity adjustment applied in 2018 for MA plan payments.¹⁶ A separate propensity score model was estimated for each of the 3 cohorts (frail elderly, major complex chronic, and minor complex chronic) and within each cohort, for the subset of beneficiaries experiencing a hospital stay. Using these propensity score models, each beneficiary was assigned a stabilized weight calculated as the inverse of their propensity score multiplied by the proportion of MA beneficiaries (for MA beneficiaries) and the inverse of 1 minus their propensity score multiplied by the proportion of TM beneficiaries (for TM beneficiaries). We compared the demographic and clinical characteristics of the MA and TM populations before and after IPTW adjustment using standardized mean differences. We considered standardized mean differences greater than 0.10 to reflect meaningful differences between groups.¹⁵

To estimate the association between MA enrollment and health care utilization, we constructed a series of beneficiary-level regression models. Hospital stays and ED visits were modeled using ordinary least-squares regression. Readmissions were modeled using logistic regression. For all models, the primary independent variable was an indicator variable for MA enrollment (vs TM enrollment), with observations weighted by the inverse probabilities of treatment. All regressions also included core-based statistical area fixed effects to account for differences in the geographic distribution of MA and TM beneficiaries in the study population.

For hospital stays and ED visits, we calculated adjusted rates of utilization per 1000 beneficiaries per year, as well as absolute and relative adjusted differences between MA and TM beneficiaries. These adjusted rates were estimated using least-squares means, with adjusted differences representing the associated marginal effects. For readmissions, we calculated the odds ratios (ORs) associated with MA enrollment.

Finally, we conducted a set of exploratory analyses to evaluate whether plan design and value-based payment moderated the association between MA enrollment and health care utilization. To explore differences by plan design, we stratified the MA population by plan type, changing our primary independent variable to one with 3 values: TM, MA HMO, and MA PPO. To explore the impact of value-based payment, we stratified the MA population by primary care payment arrangement, changing our primary independent variable to one with 4 values: TM, MA fee for service, MA upside-only risk, and MA 2-sided risk. For this analysis, the MA population was restricted to beneficiaries in HMO plans because the insurer principally implemented advanced value-based payment arrangements within HMO plans.

All analyses were performed between December 2020 and January 2022 using SAS Enterprise Guide, version 8.3 (SAS Institute). All hypothesis tests were 2-sided, with P < .05 indicating significance. We calculated 95% CIs using robust SEs.

Cross-sectional comparisons of MA and TM beneficiaries are subject to potential bias from differences in documentation patterns across the 2 programs. In our primary analysis, we deflated CMS-HCC scores for MA beneficiaries by 5.91%, consistent with the coding intensity adjustment applied to MA plan payments in 2018.¹⁶ Because some estimates of coding intensity exceed the adjustment applied by the CMS, we conducted a sensitivity analysis in which we deflated CMS-HCC scores by 10%, a value exceeding estimates of coding intensity in comparable, non–provider-owned MA plans.^16-18 Because we used a claims-based taxonomy to assign beneficiaries to different cohorts, differences in documentation patterns also had the potential to bias our findings upstream of comorbidity adjustment. To address this concern, we examined outcomes for another cohort of beneficiaries with complex care needs^1,2—those younger than 65 years with disabilities. Because identification of this cohort is based on demographic characteristics alone, it should not be subject to bias from differential documentation of comorbidities. Additional detail on these sensitivity analyses is provided in the eMethods in the Supplement.

Among the study population of 1 844 326 Medicare beneficiaries (mean [SD] age, 75.6 [7.1] years; men, 822 847 [44.6%]; women, 1 021 479 [55.4%]; Black, 221 419 [12.0%]; White, 1 524 458 [82.7%]; other race, 98 449 [5.3%]; dual eligibility, 223 377 [12.1%]), 1 177 896 (63.9%) were MA beneficiaries and 666 430 (36.1%) were TM beneficiaries. The study population was distributed across the cohorts as follows: frail elderly, 116 047 beneficiaries with MA (10.0% of the MA sample) and 104 036 beneficiaries with TM (15.6% of TM sample); major complex chronic, 320 954 with MA (27.2% of the MA sample) and 158 811 with TM (23.8% of the TM sample); and minor complex chronic, 740 895 with MA (62.9% of the MA sample) and 403 583 with TM (60.6% of the TM sample). Participant flow through the study is shown in the eFigure in the Supplement. Compared with TM beneficiaries, MA beneficiaries were generally younger, more likely to be Black, more likely to have a disability, more likely to be less frail, and more likely to live in the South (eTable 1 in the Supplement). After IPTW adjustment, the MA and TM populations had similar demographic and clinical characteristics across all cohorts (Table 1).

Beneficiaries enrolled in MA had lower rates of hospital stays, ED visits, and 30-day readmissions. The largest relative differences were observed for hospital stays, which ranged from −9.3% (95% CI, −10.9% to −7.7%) for the frail elderly cohort to −11.9% (95% CI, −13.2% to −10.7%) for the major complex chronic cohort.

Compared with TM beneficiaries, MA beneficiaries had significantly lower rates of hospital stays across all cohorts, with adjusted differences per 1000 beneficiaries of −65.46 (95% CI, −76.63 to −54.30; P < .001) for the frail elderly cohort, −64.27 (95% CI, −70.88 to −57.65; P < .001) for the major complex chronic cohort, and −29.28 (95% CI, −31.93 to −26.64; P < .001) for the minor complex chronic cohort (Table 2). Among beneficiaries experiencing a hospital stay, 30-day readmissions were significantly less frequent for MA beneficiaries in the major complex chronic and minor complex chronic cohorts, with adjusted ORs of 0.922; (95% CI, 0.899-0.947; P < .001) and 0.917; (95% CI, 0.892-0.943; P < .001), respectively. There was no significant difference in readmissions within the frail elderly cohort (adjusted OR, 0.979; 95% CI, 0.948-1.010; P = .17). Adjusted differences in ED visits per 1000 beneficiaries were −29.43 (95% CI, −41.71 to −17.15; P < .001) for the frail elderly cohort, −39.12 (95% CI, −46.59 to −31.65; P < .001) for the major complex chronic cohort, and −28.71 (95% CI, −31.94 to −25.48; P < .001) for the minor complex chronic cohort (Table 2). Unadjusted results and results for measures of potentially avoidable hospital stays and ED visits are provided in eTables 2 and 3 in the Supplement.

Medicare Advantage beneficiaries enrolled in HMO and PPO plans both had lower rates of hospitalizations and ED visits than TM beneficiaries across all cohorts, but the magnitude of the difference was larger for those enrolled in HMO plans (Table 3). Across primary care payment arrangements, MA beneficiaries had lower rates of hospitalizations and ED visits across most cohorts (Table 4). The largest difference was observed for MA beneficiaries attributed to primary care organizations reimbursed under 2-sided risk arrangements.

After deflating CMS-HCC scores for MA beneficiaries by 10%, we still observed significantly fewer hospital stays, ED visits, and 30-day readmissions across cohorts, with the continued exception of 30-day readmissions in the frail elderly cohort. For beneficiaries younger than 65 years with disabilities, those enrolled in MA experienced significantly fewer hospital stays, ED visits, and readmissions than those enrolled in TM. Full results from these sensitivity analyses are provided in eTables 4, 5, and 6 in the Supplement.

In this cross-sectional study of Medicare beneficiaries with complex care needs, we found that those enrolled in MA had lower rates of acute care utilization than similar beneficiaries enrolled in TM. The largest relative differences were observed for hospital stays, which ranged from −9.3% in the frail elderly cohort to −11.9% in the major complex chronic cohort.

Our findings are consistent with previous comparisons of MA and TM populations, which have generally revealed lower rates of hospitalizations and ED visits among MA beneficiaries.^19-23 The current study extends these observations, which were based on broad MA and TM populations, into specific cohorts of beneficiaries with complex care needs. Previous comparisons of MA and TM beneficiaries have been mixed with respect to the association between MA enrollment and hospital readmissions.^22,23 Our findings of decreased rates of readmissions among most beneficiaries with complex care needs suggests that the association between MA enrollment and readmissions may be influenced by the underlying complexity and care needs of the patient population.

Although the association between MA enrollment and lower rates of acute care utilization was observed across cohorts of beneficiaries in this study, the magnitude of this association varied. Beneficiaries in the major and minor complex chronic cohorts had comparatively lower rates of acute care utilization than those in the frail elderly cohort. Heterogeneity in the association between MA enrollment and care utilization is consistent with previous research findings that the association between MA enrollment and outcomes varies across disease states.^22-26

In exploratory analyses, we observed that elements of insurance design in MA moderated the association between MA enrollment and acute care utilization. For example, the difference in utilization rates between MA and TM beneficiaries was larger among MA beneficiaries attributed to primary care organizations reimbursed under 2-sided risk arrangements than those cared for under less advanced value-based payment models. Value-based payment models in TM have been associated with reduced acute care utilization for beneficiaries with frailty⁵ but not other complex care needs,^6,7 suggesting that differences in the design and implementation of value-based payment models may influence acute care utilization.

We also found evidence that plan design moderated the association between MA enrollment and acute care utilization. The difference in utilization rates between MA and TM beneficiaries was larger for MA beneficiaries enrolled in HMO plans than for those enrolled in PPO plans. This difference could be explained by the fact that HMO plans place a greater incentive on in-network care and an accountable primary care relationship. Although we were not able to assess the influence of other potential levers, previous research has suggested that the disease management and care management programs used by MA plans may reduce acute care utilization and improve outcomes for beneficiaries with chronic illnesses.^27-29

This study has several limitations. First, we used a specific claims-based taxonomy to define our study population; therefore, our findings may not be generalizable to other cohorts of Medicare beneficiaries with complex needs. Second, our outcomes were focused on measures of acute care utilization, and we did not compare postacute and outpatient care utilization or measures of care quality and access. Third, we were unable to observe outcomes for MA or TM beneficiaries who disenrolled or switched to another plan during the study period. Fourth, our MA population was drawn from plans offered by a single insurer, which may limit the generalizability of our findings to other MA populations. Fifth, our results are subject to confounding by nonrandom beneficiary selection into MA on unobservable characteristics. Sixth, we were unable to explore all mechanisms by which MA enrollment may influence care utilization, such as care management, care coordination, and supplemental benefits. Additional research should focus on specific elements of payment and delivery reform within MA that may impact outcomes for beneficiaries with complex care needs.

In this cross-sectional study of Medicare beneficiaries with complex care needs, those enrolled in MA had lower rates of hospital stays, ED visits, and 30-day readmissions than similar beneficiaries enrolled in TM. These findings suggest that managed care activities in MA may influence the nature and quality of care provided to Medicare beneficiaries with complex care needs.

Accepted for Publication: August 16, 2022.

Published: October 7, 2022. doi:10.1001/jamahealthforum.2022.3451

Open Access: This is an open access article distributed under the terms of the CC-BY-NC-ND License. © 2022 Drzayich Antol D et al. JAMA Health Forum.

Corresponding Author: Brian W. Powers, MD, MBA, Tufts University School of Medicine, 136 Harrison Ave, Boston, MA 02111 (bpower03@tufts.edu).

Author Contributions: Mr Schwartz and Ms Caplan had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Drzayich Antol, Casebeer, Erwin, Powers.

Acquisition, analysis, or interpretation of data: Drzayich Antol, Schwartz, Caplan, Erwin, Shrank, Powers.

Drafting of the manuscript: Drzayich Antol, Erwin, Powers.

Critical revision of the manuscript for important intellectual content: Drzayich Antol, Schwartz, Caplan, Casebeer, Shrank, Powers.

Statistical analysis: Schwartz, Caplan, Casebeer, Powers.

Administrative, technical, or material support: Drzayich Antol, Shrank, Powers.

Supervision: Drzayich Antol, Erwin, Shrank, Powers.

Conflict of Interest Disclosures: Dr Shrank reported membership on the board of directors for GetWell. Dr Powers reported previous employment by Anthem and Fidelity Investments. No other disclosures were reported.

Disclaimer: The views expressed in this article are those of the authors and not necessarily the views or policies of their respective affiliated institutions.