Quality, Health, and Spending in Medicare Advantage and Traditional Medicare

ABSTRACT

Objectives: To compare Medicare Advantage (MA) and traditional Medicare (TM) performance on quality, health, and cost outcomes in peer-reviewed literature published since 2010.

Study Design: Systematic review of peer-reviewed papers published between January 1, 2010, and May 1, 2020.

Methods: To identify relevant research papers, we searched MEDLINE, EBSCO, and ProQuest. We excluded any studies that did not meet several inclusion criteria. Titles, abstracts, and full-text articles were independently reviewed by 1 author and several trained research assistants. Disagreements were resolved through discussion. We also reviewed the bibliographies of included studies and consulted subject matter experts to identify additional papers.

For each eligible study, we extracted the first author, year published, study design, data sources, study years, sample sizes, relevant measures, and study quality. To ensure consistent and complete data extraction, each article was reviewed by 2 reviewers. Study quality was assessed using a modified Newcastle-Ottawa Scale.

Results: Thirty-five studies including 208 analyses were included. All included studies were observational. Two-thirds of studies were of high methodological quality for observational studies, and 49% addressed selection bias. Analyses compared quality of care (41%), health outcomes (44%), and spending (15%). Overall, 65% of analyses found a statistically significant relationship: 52% favored MA and 13% favored TM.

Conclusions: More than half of recent analyses comparing MA and TM find that MA delivers significantly better quality of care, better health outcomes, and lower costs compared with TM.

Am J Manag Care. 2021;27(9):395-400. https://doi.org/10.37765/ajmc.2021.88641

Takeaway Points

This systematic review revealed that more than half of analyses found that Medicare Advantage beneficiaries experienced better quality of care, better health outcomes, and lower spending across a wide range of settings of care and research questions.

• One-third of studies were of low quality, and more than half did not account for selection bias.
• Future studies should look to leverage experimental and nonexperimental study designs to isolate the benefit of Medicare Advantage for beneficiaries and identify what aspects of the Medicare Advantage benefit add the greatest value.

Policy makers have a longstanding interest in understanding the value of Medicare Advantage (MA) relative to traditional Medicare (TM). On one hand, MA plans may be more efficient than Medicare; in addition, they may offer disease management, care coordination, and supplemental benefits not offered by TM, such as dental, vision, and hearing, that could lead to better health outcomes and lower utilization of other health care services.¹ On the other hand, MA plans are allowed to use prior authorization, physician networks, and cost sharing, all of which can discourage utilization.

Previous reviews that compared MA and TM prior to the passage of the Affordable Care Act (ACA) reported mixed results. Studies found that TM patients tended to report more favorable experiences of care, whereas MA patients were more likely to receive recommended preventive care services and have fewer adverse events.² Past studies have noted the difficulty in comparing MA and TM due to scarcity of available claims data.³ Given the release of MA encounter data and other multipayer data sources, we updated the literature by reviewing recent studies comparing MA and TM on measures of quality of care, health, and spending.

METHODS

Search Strategy and Study Eligibility

We searched MEDLINE, EBSCO, and ProQuest for original research papers that compared MA and TM on quality of care, health outcomes, and/or health spending. In consultation with a research librarian, we developed a search strategy that used key words and phrases and MeSH terms. The search was limited to English language papers published since January 1, 2010 (see methods in the eAppendix [available at ajmc.com]). We also reviewed the bibliographies of included studies and consulted subject matter experts to identify additional papers.

We updated our search results prior to submission of this review and performed a supplemental search using the same protocol from December 2019 to May 1, 2020.

To identify relevant studies, we excluded studies that did not meet all of the following inclusion criteria: (1) evaluated 1 or more administrative measure of quality of care, health, or cost of care; (2) directly compared MA and TM; and (3) used study data collected after January 1, 2010. We excluded studies and measures that examined (1) differences in utilization of services but did not assess whether utilization reflected better quality or health; (2) facility-level quality measures;
(3) patient-reported satisfaction and experience measures; and (4) the price of services but did not examine the patient’s or plan’s spending.

Titles, abstracts, and full-text articles were independently reviewed by 2 trained research assistants, an author (V.G.), and a senior author (E.D.). Disagreements regarding inclusion in the final review were resolved through discussion.

Data Extraction and Synthesis

Data extraction was performed independently by 1 author (V.G.) and 2 trained research assistants and was evaluated by a senior author (E.D.). To ensure consistent and complete data extraction, each article was reviewed by 2 reviewers. A modified Newcastle-Ottawa Scale (see eAppendix for scoring details) was used to assess the methodological quality of each paper. We considered a paper that received a score of 6 or greater to be of high quality (maximum numbers of stars were 7 for observational studies and 9 for cohort studies).

RESULTS

Study Characteristics

From 539 titles identified in our initial search, 35 studies were included in the final review (Figure 1 and eAppendix Tables). We excluded papers that did not include comparisons of MA and TM; did not include original research; did not measure quality-of-care processes, health, or spending using administrative data; or used only data from prior to 2010.

Most studies (53%) were published between 2017 and 2020. More than 75% of studies used study data spanning multiple years. Eighteen (51%) studies measured quality of care, 7 (20%) measured costs of care, and 27 (77%) measured health outcomes. Fourteen (40%) papers assessed hospital readmissions, which were classified as a measure of health. All but 2 studies used multivariate regression modeling.^4,5 Sample sizes ranged from 323 participants in a secondary analysis of a clinical study to more than 40 million observations.

Twenty-four (69%) papers earned 6 or more stars using a modified Newcastle-Ottawa Scale, indicating high quality for observational studies. Most papers used cross-sectional designs (n = 19), followed by difference-in-differences (n = 6) and time series studies (n = 5); 5 studies used a cohort study design. Seventeen studies (49%) used propensity score methods, differences in differences, or other approaches to address selection bias.^6-19 Among the 18 studies that did not explicitly address selection bias, most used multivariate regression models; 2 papers reported unadjusted differences in outcomes.^5,20 Data sources used varied across studies and varied by study goals. Although most studies relied on the same data set to identify MA and TM individuals, 9 studies (26%) collected their primary MA and TM samples from different data sources.^{6,7,9-12,15,21,22}

Quality of Care

A total of 17 studies (49%) measured quality of care using 85 separate analyses of established performance measures, as well as study-specific measures. The most common area of study was end-of-life care (27%) (Table), assessing hospice use, discharge from hospice, and number of transitions between settings of care in the last 3 days of life.^5,21,23 The second most common area of study was postacute care quality (26%) (Table), which largely drew on quality measures collected from Medicare postacute care assessment data sets, including for pain, pressure ulcers, and flu vaccinations.^24-26

Several studies found some evidence for a correlation between MA and better quality of care. Among analyses focused on quality of care, 45 (53%) analyses found that MA delivered statistically significantly better care, 11 (13%) analyses found that TM delivered significantly better care, and 29 (34%) reported no significant difference between MA and TM (Figure 2).

Five studies examined end-of-life care quality, and two-thirds (65%) of analyses found evidence in the direction of MA. Using the Health and Retirement Survey, Chen and Miller²⁷ estimated that the odds of dying in the hospital were significantly lower among MA beneficiaries who were continuously enrolled but did not find a difference among those who disenrolled into TM. Gidwani-Marszowski et al²¹ examined hospice use among veterans continuously enrolled in MA, those who switched, and those enrolled in TM, and found that the odds of at least 1 day of hospice were significantly greater among veterans who were ever enrolled in MA. However, the risk-adjusted analysis found fewer overall days in hospice among MA beneficiaries than among those enrolled in TM, as well as lower likelihood of hospice use among the last 3 or 7 or more days of life among MA enrollees. Teno et al⁵ used 100% Medicare claims for decedents and found somewhat higher utilization of hospice at time of death among MA beneficiaries in 2011 and 2015 compared with TM but did not find differences in hospice use within the last 3 days of life in 2015. Among potentially burdensome types of care at the end of life, such as multiple health care transitions and mechanical ventilation, Teno et al found significantly lower rates of use in the MA population compared with TM.

Health

A total of 22 studies (63%) measured health outcomes using 91 analyses, including measures of self-reported health, intermediate health outcomes, hospital readmissions, and mortality. Higher-quality studies accounted for 16 studies and 82 analyses. Among studies of health, the most common area of study was the inpatient setting (56%), in which more than 75% of measures examined hospital readmissions.^{8-10,16,20,28-31} Analyses that examined health in the outpatient context measured intermediate health outcomes, including blood pressure and cholesterol control, as well as whether individuals were discharged to the community or to other care settings from the hospital.^32,33

Among analyses focused on health, 45 (49%) found that MA was associated with significantly better health outcomes and 7 (8%) found that TM was associated with significantly better health outcomes (Figure 2). Thirty-nine analyses (43%) did not find a statistically significant relationship between coverage type and health outcomes.

Of 7 studies using 38 analyses and comparing readmission rates in MA and TM, 16 analyses found a statistically significant relationship (12 in favor of MA and 4 in favor of TM), whereas 22 did not find a statistically significant difference.^{8-10,12,16,20,29-31,33} Using the Medicare Provider Analysis and Review (MEDPAR) file to identify hospital readmissions and Healthcare Effectiveness Data and Information Set (HEDIS) to identify potentially missing readmissions in MA, the work by Panagiotou et al was the only study to find a consistent association in favor of TM beneficiaries across clinical subgroups and sensitivity tests.¹⁰ A previous study also using MEDPAR over a shorter time frame found that MA beneficiaries experienced fewer hospital readmissions than TM beneficiaries.³³

Of the 3 studies examining disparities, all examined differences in readmission rates in TM and MA by race and ethnicity.^14,28,34 Using the Minimum Data Set reported by nursing homes, Rivera-Hernandez et al reported that although MA readmission rates were lower than those for TM, the differences in readmission rates among nursing home residents by race and ethnicity were consistent between the 2 programs and did not differ by nursing home star rating or the racial makeup of the facility’s residents.³⁴ In contrast, using 2009-2012 New York State hospitalization data, Li et al reported that there were persistent racial differences in readmission rates in TM but did not find racial differences in MA or evidence of ethnic differences.¹⁴ Updating the analysis using hospitalization data from 2013, Li et al found that racial differences in readmission rates between TM and MA were not significantly different.²⁸

Three studies examined differences in MA and TM and found lower overall mortality rates following hospitalizations for 3 clinical conditions.^4,7,33 The 2 studies examining overall mortality used different approaches but found that MA beneficiaries had lower mortality rates. Using actual-to-predicted mortality, Beveridge et al found that MA enrollees had lower predicted 1-year mortality overall and that the difference between actual and predicted mortality increased as Hierarchical Condition Categories scores increased.⁷ In contrast, Newhouse et al calculated hazard ratios to compare MA and TM mortality rates. Using this approach, Newhouse et al reported that mortality in MA was lower than in TM, but the difference attenuated over time.⁴

Health Spending

A total of 8 papers (23%) examined spending using 32 different analyses. Spending was defined based upon setting of care (hospice care, spend per discharge), as well as total spending and patient out-of-pocket costs. Analyses focused most on the inpatient setting (31%), examining spending by discharge and by service line.⁸ The second most common area studied was overall measures (19%), which included total spending,^6,13,18 out-of-pocket spending,^13,35 90-day episode-of-care spending,³⁰ and projected Medicare spending.^11,33

Among analyses focused on spending, 16 (50%) found that MA was associated with significantly lower spending, 10 (31%) found that TM was associated with significantly lower spending, and 6 (19%) did not find a meaningful difference (Figure 2).

Four studies calculated total spending in MA and TM; 2 estimated TM spending as if MA utilization patterns of care were adopted.^11,33 One study, using data from a single institution, found that 90-day episode-of-care spending was significantly higher among MA beneficiaries undergoing hip or knee arthroplasty. The differences were not explained by higher rates of complication or readmission in the hospital setting; MA beneficiaries were more likely to be discharged to a postacute care facility.³⁰

Using a proprietary multipayer claims data set collected by the Health Care Cost Institute and Medicare claims, Curto et al⁶ compared total spending in MA and TM overall and by site of care and inpatient diagnosis. To address differences based on observed and unobserved factors, Curto et al used 2 approaches: weighting based on county and risk score and weighting based on predicted mortality. After accounting for selection differences, total spending in MA was 9% to 24% lower per month and out-of-pocket spending per month was 53% to 60% lower. MA spending was lower than TM in the inpatient hospital, outpatient, and skilled nursing facility settings, whereas hospice spending ranged from 25% lower to 1.8% higher. Curto et al found that lower MA spending was largely explained by lower utilization rates, not lower prices.

DISCUSSION

In 35 studies comparing MA and TM that were conducted since the passage of the ACA, more than half of analyses (52%; 106 of 208) found that MA was statistically significantly associated with better performance on measures of quality, health, or spending. Analyses focused on quality of care were more likely to report that MA enrollment was associated with better quality of care. We did not observe a consistent pattern in the relationship between coverage type and quality, health, or spending within focus areas. This may be due in part to the heterogeneity in the definition of outcome measures, as well as variation in data sources and sample definitions.

The MA program gives private plans an opportunity to deliver care to Medicare beneficiaries through a managed care structure. Since the ACA, policy makers have shifted away from MA as a vehicle to reduce Medicare spending and focused on achieving parity with TM spending. Through the use of care management tools, benefit design flexibilities, and the ability to provide benefits outside the TM benefit and benefits aimed at addressing health-related social needs, MA plans may be able to better manage beneficiary care needs. MA plans also have tools available that could limit access to care, which is why studies of care, health outcomes, and spending in MA compared with TM are critical for policy makers and beneficiaries.

All included studies used observational study designs. Fewer than half of the studies used propensity score methods or other approaches to account for selection differences between MA and TM. Matching methods are designed to balance groups based on observed characteristics; however, these approaches cannot match individuals on unobserved characteristics such as patient activation or motivation. In addition, studies using claims typically do not observe the presence of supplemental coverage through Medigap or other retiree benefits, which can modify health-seeking behavior.³⁶ It is important to note that not all studies were looking to create comparable groups; for example, Teno et al reported observed differences between the 2 programs providing insights into differences in care.⁵

One of the motivations for this analysis was to understand what data sets researchers were using to compare MA and TM. We found that researchers have used the CMS postacute care assessment data sets and hospice claims, which use the same collection forms for both populations. To identify inpatient stays, investigators have used MEDPAR data to identify MA inpatient stays at hospitals reporting “shadow claims” for MA beneficiaries. A strength of MEDPAR data to identify MA inpatient stays is that the data source—the hospital claim—is the same for MA and TM. Nine papers, accounting for 45 analyses, leveraged different data sources to identify MA and TM samples, which can create challenges in ensuring that measures are calculated consistently. For example, MA contracts report HEDIS measures, which are often calculated using submitted claims and chart reviews. Similar measures calculated using TM fee-for-service claims alone could underestimate services received. The comparison of HEDIS measures with TM fee-for-service claims was first used by Brennan and Shephard³⁷ and used by several researchers in this review.^10,12,38 Additional work is needed to better understand the bias introduced by directly comparing HEDIS and TM calculated claims measures and how to account for these differences in data collection.

The relatively lower quality-of-care performance of TM in many of these studies is concerning. More than 30 million Americans rely on TM for their care coverage. President Joe Biden campaigned on expanding the number of people who could rely on TM by lowering the age of eligibility from 65 to 60 years. An alternative model to MA, accountable care organizations (ACOs) offered in TM through the Medicare Shared Savings Program and other Centers for Medicare and Medicaid Innovation initiatives, was not directly compared.³⁹ It may be the case that integrated care delivery, such as through ACOs, particularly those in 2-sided risk arrangements, may be more comparable with MA than traditional fee-for-service Medicare.

Policy makers interested in addressing the gap between MA and TM quality, health, and spending outcomes could consider using more outcomes-based measures within the context of pay for performance under the Merit-based Incentive Payment System and inpatient value-based payment programs. The Medicare Payment Advisory Commission has recently suggested focusing more on patient outcomes and comparisons at local geographic levels and less on process quality-of-care measures in MA.⁴⁰

Limitations

First, it is possible that we did not identify all eligible papers. To limit this risk, we collected potentially eligible citations from multiple databases and consulted with experts to identify missing papers. Second, 28 of 35 studies conducted multiple analyses examining alternative specifications of their primary outcome, as well as multiple outcomes across different domains or in different disease subgroups. Often papers had mixed results across these models, making it difficult to assign a single result to each paper. To address this, we collected data at an analysis level instead of at a paper level. It is unclear if this approach would favor MA or TM. Third, this review did not include patient experiences of care or measures of utilization. Patient-reported measures can provide unique insights into the care delivery process, particularly patient-centeredness of care and health outcomes. Lastly, we also excluded measures of utilization, such as number of days in the hospital, which are difficult to categorize as “good” or “bad.” Without clear guidance from the researcher or other sources on the appropriateness of greater (or lesser) utilization, we decided to exclude these measures. Understanding the differences in utilization between MA and TM is important for understanding why health care spending patterns may differ between these programs and is an important area for further research.

CONCLUSIONS

In this review, we found heterogenous results, but more than half of analyses found that MA beneficiaries experienced better quality of care, better health outcomes, and/or lower spending. Notably, one-third of studies were of low quality and more than half did not account for selection bias. Our study suggests that the MA model adds value for Medicare beneficiaries and identifies gaps in the field for researchers.

Acknowledgments

The authors acknowledge the thoughtful feedback of Courtney Brown, Adrianne Casebeer, Teresa Rogstad, William Shrank, and Zhou Yang on an earlier draft of this article.

Author Affiliations: Berkeley Research Group (ED, RT, TD, VG), Washington, DC; University of Maryland School of Public Health (ED), College Park, MD.

Source of Funding: Humana.

Author Disclosures: Dr DuGoff, Mrs Tabak, Mr Diduch, and Ms Garth are employed by Berkeley Research Group, which has several contracts with Medicare Advantage plans to address policy and operational issues. Dr DuGoff reports receiving speaking fees from Zimmer Biomet.

Authorship Information: Concept and design (ED, RT, TD); acquisition of data (ED, VG); analysis and interpretation of data (ED, RT, VG); drafting of the manuscript (ED, RT, TD); critical revision of the manuscript for important intellectual content (ED, RT, TD); statistical analysis (VG); obtaining funding (ED); administrative, technical, or logistic support (TD, VG); supervision (ED, TD); and literature review (ED, VG).

Address Correspondence to: Ruth Tabak, MPP, MPH, Berkeley Research Group, 1800 M St NW, Second Fl, Washington, DC 20036. Email: rtabak@thinkbrg.com.

REFERENCES

1. Willink A, DuGoff EH. Integrating medical and nonmedical services—the promise and pitfalls of the CHRONIC Care Act. N Engl J Med. 2018;378(23):2153-2155. doi:10.1056/NEJMp1803292

2. McGuire TG, Newhouse JP, Sinaiko AD. An economic history of Medicare part C. Milbank Q. 2011;89(2):289-332. doi:10.1111/j.1468-0009.2011.00629.x

3. Brennan N, Ornstein C, Frakt AB. Time to release Medicare Advantage claims data. JAMA. 2018;319(10):975-976. doi:10.1001/jama.2017.21519

4. Newhouse JP, Price M, McWilliams JM, Hsu J, Souza J, Landon BE. Adjusted mortality rates are lower for Medicare Advantage than traditional Medicare, but the rates converge over time. Health Aff (Millwood). 2019;38(4):554-560. doi:10.1377/hlthaff.2018.05390

5. Teno JM, Gozalo P, Trivedi AN, et al. Site of death, place of care, and health care transitions among US Medicare beneficiaries, 2000-2015. JAMA. 2018;320(3):264-271. doi:10.1001/jama.2018.8981

6. Curto V, Einav L, Finkelstein A, Levin J, Bhattacharya J. Health care spending and utilization in public and private Medicare. Am Econ J Appl Econ. 2019;11(2):302-332. doi:10.1257/app.20170295

7. Beveridge RA, Mendes SM, Caplan A, et al. Mortality differences between traditional Medicare and Medicare Advantage: a risk-adjusted assessment using claims data. Inquiry. 2017;54:46958017709103. doi:10.1177/0046958017709103

8. Henke RM, Karaca Z, Gibson TB, et al. Medicare Advantage and traditional Medicare hospitalization intensity and readmissions. Med Care Res Rev. 2018;75(4):434-453. doi:10.1177/1077558717692103

9. Kumar A, Rahman M, Trivedi AN, Resnik L, Gozalo P, Mor V. Comparing post-acute rehabilitation use, length of stay, and outcomes experienced by Medicare fee-for-service and Medicare Advantage beneficiaries with hip fracture in the United States: a secondary analysis of administrative data. PLoS Med. 2018;15(6):e1002592. doi:10.1371/journal.pmed.1002592

10. Panagiotou OA, Kumar A, Gutman R, et al. Hospital readmission rates in Medicare Advantage and traditional Medicare: a retrospective population-based analysis. Ann Intern Med. 2019;171(2):99-106. doi:10.7326/M18-1795

11. McGarry BE, Grabowski DC. Managed care for long-stay nursing home residents: an evaluation of Institutional Special Needs Plans. Am J Manag Care. 2019;25(9):438-443.

12. Timbie JW, Bogart A, Damberg CL, et al. Medicare Advantage and fee-for-service performance on clinical quality and patient experience measures: comparisons from three large states. Health Serv Res. 2017;52(6):2038-2060. doi:10.1111/1475-6773.12787

13. Mahmoudi E, Tarraf W, Maroukis BL, Levy HG. Does Medicare managed care reduce racial/ethnic disparities in diabetes preventive care and healthcare expenditures? Am J Manag Care. 2016;22(10):e360-e367.

14. Li Y, Cen X, Cai X, Wang D, Thirukumaran CP, Glance LG. Does Medicare Advantage reduce racial disparity in 30-day rehospitalization for Medicare beneficiaries? Med Care Res Rev. 2018;75(2):175-200. doi:10.1177/1077558716681938

15. Mattke S, Han D, Wilks A, Sloss E. Medicare home visit program associated with fewer hospital and nursing home admissions, increased office visits. Health Aff (Millwood). 2015;34(12):2138-2146. doi:10.1377/hlthaff.2015.0583

16. Chen M, Grabowski DC. Hospital Readmissions Reduction Program: intended and unintended effects. Med Care Res Rev. 2019;76(5):643-660. doi:10.1177/1077558717744611

17. Chung S, Lesser LI, Lauderdale DS, Johns NE, Palaniappan LP, Luft HS. Medicare annual preventive care visits: use increased among fee-for-service patients, but many do not participate. Health Aff (Millwood). 2015;34(1):11-20. doi:10.1377/hlthaff.2014.0483

18. Meyers DJ, Kosar CM, Rahman M, Mor V, Trivedi AN. Association of mandatory bundled payments for joint replacement with use of postacute care among Medicare Advantage enrollees. JAMA Netw Open. 2019;2(12):e1918535. doi:10.1001/jamanetworkopen.2019.18535

19. Park S, White L, Fishman P, Larson EB, Coe NB. Health care utilization, care satisfaction, and health status for Medicare Advantage and traditional Medicare beneficiaries with and without Alzheimer disease and related dementias. JAMA Netw Open. 2020;3(3):e201809. doi:10.1001/jamanetworkopen.2020.1809

20. Oh JJ. Analysis of hospital readmission patterns in Medicare fee-for-service and Medicare Advantage beneficiaries. Prof Case Manag. 2017;22(1):10-20. doi:10.1097/NCM.0000000000000172

21. Gidwani-Marszowski R, Kinosian B, Scott W, Phibbs CS, Intrator O. Hospice care of veterans in Medicare Advantage and traditional Medicare: a risk-adjusted analysis. J Am Geriatr Soc. 2018;66(8):1508-1514. doi:10.1111/jgs.15434

22. Weissman GE, Kerlin MP, Yuan Y, et al. Potentially preventable intensive care unit admissions in the United States, 2006-2015. Ann Am Thorac Soc. 2020;17(1):81-88. doi:10.1513/AnnalsATS.201905-366OC

23. Teno JM, Christian TJ, Gozalo P, Plotzke M. Proportion and patterns of hospice discharges in Medicare Advantage compared to Medicare fee-for-service. J Palliat Med. 2018;21(3):302-306. doi:10.1089/jpm.2017.0046

24. Chang E, Ruder T, Setodji C, et al. Differences in nursing home quality between Medicare Advantage and traditional Medicare patients. J Am Med Dir Assoc. 2016;17(10):960.e9-960.e14. doi:10.1016/j.jamda.2016.07.017

25. Goldfeld KS, Grabowski DC, Caudry DJ, Mitchell SL. Health insurance status and the care of nursing home residents with advanced dementia. JAMA Intern Med. 2013;173(22):2047-2053. doi:10.1001/jamainternmed.2013.10573

26. Waxman DA, Min L, Setodji CM, Hanson M, Wenger NS, Ganz DA. Does Medicare Advantage enrollment affect home healthcare use? Am J Manag Care. 2016;22(11):714-720.

27. Chen EE, Miller EA. A longitudinal analysis of site of death: the effects of continuous enrollment in Medicare Advantage versus conventional Medicare. Res Aging. 2017;39(8):960-986. doi:10.1177/0164027516645843

28. Li Y, Cen X, Cai X, Thirukumaran CP, Zhou J, Glance LG. Medicare Advantage associated with more racial disparity than traditional Medicare for hospital readmissions. Health Aff (Millwood). 2017;36(7):1328-1335. doi:10.1377/hlthaff.2016.1344

29. Zingmond DS, Liang LJ, Parikh P, Escarce JJ. The impact of the Hospital Readmissions Reduction Program across insurance types in California. Health Serv Res. 2018;53(6):4403-4415. doi:10.1111/1475-6773.12869

30. Yayac MF, Harrer SL, Janiec DA, Courtney PM. Costs and outcomes of Medicare Advantage and traditional Medicare beneficiaries after total hip and knee arthroplasty. J Am Acad Orthop Surg. 2020;28(20):e910-e916. doi:10.5435/JAAOS-D-19-00609

31. Kim H, Charlesworth CJ, McConnell KJ, Valentine JB, Grabowski DC. Comparing care for dual-eligibles across coverage models: empirical evidence from Oregon. Med Care Res Rev. 2019;76(5):661-677. doi:10.1177/1077558717740206

32. Figueroa JF, Blumenthal DM, Feyman Y, et al. Differences in management of coronary artery disease in patients with Medicare Advantage vs traditional fee-for-service Medicare among cardiology practices. JAMA Cardiol. 2019;4(3):265-271. doi:10.1001/jamacardio.2019.0007

33. Huckfeldt PJ, Escarce JJ, Rabideau B, Karaca-Mandic P, Sood N. Less intense postacute care, better outcomes for enrollees in Medicare Advantage than those in fee-for-service. Health Aff (Millwood). 2017;36(1):91-100. doi:10.1377/hlthaff.2016.1027

34. Rivera-Hernandez M, Rahman M, Mor V, Trivedi AN. Racial disparities in readmission rates among patients discharged to skilled nursing facilities. J Am Geriatr Soc. 2019;67(8):1672-1679. doi:10.1111/jgs.15960

35. Khandelwal N, White L, Curtis JR, Coe NB. Health insurance and out-of-pocket costs in the last year of life among decedents utilizing the ICU. Crit Care Med. 2019;47(6):749-756. doi:10.1097/CCM.0000000000003723

36. Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci. 2010;25(1):1-21. doi:10.1214/09-STS313

37. Brennan N, Shepard M. Comparing quality of care in the Medicare program. Am J Manag Care. 2010;16(11):841-848.

38. Li Q, Rahman M, Gozalo P, Keohane LM, Gold MR, Trivedi AN. Regional variations: the use of hospitals, home health, and skilled nursing in traditional Medicare and Medicare Advantage. Health Aff (Millwood). 2018;37(8):1274-1281. doi:10.1377/hlthaff.2018.0147

39. Pham HH, Pilotte J, Rajkumar R, Richter E, Cavanaugh S, Conway PH. Medicare’s vision for delivery-system reform—the role of ACOs. N Engl J Med. 2015;373(11):987-990. doi:10.1056/NEJMp1507319

40. Medicare Payment Advisory Commission. Report to the Congress: Medicare and the Health Care Delivery System. Medicare Payment Advisory Commission; 2019. Accessed July 1, 2020. http://medpac.gov/docs/default-source/reports/jun19_medpac_reporttocongress_sec.pdf