Elderly Medicare Advantage members with multiple chronic conditions attained a survival benefit from more cost-effective care when a private plan developed gainshare and monetary risk-bearing arrangements with its contracted providers.
Objectives: In Medicare Advantage (MA) with its CMS Hierarchical Condition Categories (CMS-HCC) payment model, CMS reimburses private plans (Medicare Advantage Organizations [MAOs]) with prospective, monthly, health-based or risk-adjusted, capitated payments. The effect of this payment methodology on healthcare delivery remains debatable. How value-based contracting generates cost efficiencies and improves clinical outcomes in MA is studied.
Study Design: A difference in contracting arrangements between an MAO and 2 provider groups facilitated an intervention-control, preintervention—postintervention, difference-in-differences approach among statistically similar, elderly, community-dwelling MA enrollees within one metropolitan statistical area.
Methods: Starting in 2009, for intervention-group MA enrollees, the MAO and a provider group agreed to full-risk capitation combined with a revenue gainshare. The gainshare was based on increases in the Risk Adjustment Factor (RAF), which modified the CMS-HCC payments. For the control group, the MAO continued to reimburse another provider group through fee-for-service. RAF, utilization, and survival were followed until December 31, 2012.
Results: The intervention group’s mean RAF increased significantly (P <.001), estimating $2,519,544 per 1000 members of additional revenue. The intervention increased office-based visits (P <.001). Emergency department visits (P <.001) and inpatient hospital admissions (P = .002) decreased. This change in utilization saved $2,071,293 per 1000 enrollees. By intensifying office-based care for these MA enrollees with multiple comorbidities, a 6% survival benefit with a 32.8% lower hazard of death (P <.001) was achieved.
Conclusions: Value-based contracting can drive utilization patterns and improve clinical outcomes among chronically ill, elderly MA members.
Am J Manag Care. 2017;23(2):e41-e49
This study tested the hypothesis that payer-provider risk contracting promotes high-value care.
In caring for traditional Medicare beneficiaries, primary care physicians will need to transform their clinical practice and assume more fiduciary risk. CMS recently published “new policies to address and incentivize participation in alternative payment models” (APMs).1 An APM broadly can be defined as any reimbursement model other than strict fee-for-service (FFS). Currently, 30% of traditional FFS Medicare is reimbursed through APMs, with the goal of 50% involvement by 2018.2,3 Clinicians who want to become qualifying APM participants can expect to “bear more than a nominal amount of risk for monetary losses.”1 Through APMs and increasing risk assumption, these new policies aspire to promulgate high-value healthcare, as defined by “better care, smarter spending, and healthier people.”1,2 Whether FFS divestiture in favor of APMs and increasing capitation can generate cost efficiencies and also improve clinical outcomes remains debatable.4,5
Medicare Advantage (MA) provides an alternative to traditional FFS Medicare. It has been a commercial success, accounting for 17.5 million (30.6%) of all Medicare enrollees and $204.7 billion (28.9%) of Medicare’s 2017 gross spending budget.6,7 Because it is regulated by its own federal statutes,8 MA is classified as an “Other Payer APM” and excluded from CMS’ Proposed Rule for Medicare FFS.1 For over a decade, MA has used the CMS Hierarchical Condition Categories (CMS-HCC) payment model to reimburse private plans (Medicare Advantage Organizations [MAOs]) with prospective, monthly, risk-adjusted or health-based capitated payments for the care of MA enrollees. The value of subsidizing MA often has been challenged.9-14 Consequently, on October 3, 2016, CMS’ Innovation Center (CMMI) announced its Medicare Advantage Value-Based Insurance Design (VBID) model to test whether new initiatives can “improve health outcomes and lower expenditures for Medicare Advantage enrollees.”15
CMS adopted the CMS-HCC payment model with the concept that MAO recompense should reflect the disease and related cost burdens of the pertinent population and, thus, fundamentally changed how MAOs are reimbursed.16-19 In return for providing healthcare benefits to MA enrollees during the calendar year (CY), MAOs receive risk-adjusted payments during the following payment year (PY), modified by a Risk Adjustment Factor (RAF). Along with demographic data, specific diagnosis codes grouped into HCCs and associated with increased future expenditures impact the RAF. Certain HCC-HCC and demographic-HCC interactions have further additive effects on the RAF. Recalibration of the model occurs every 2 years so that the “typical” FFS Medicare beneficiary’s total RAF is 1.00.20 Therefore, when compared with an FFS Medicare beneficiary, a healthier MA enrollee’s total RAF would be less than 1.00 and a sicker one’s would be greater than 1.00.
The capitated revenue for MAOs is dependent on how its contracted providers document and code. For example, based on calculations derived from CMS’ most recent policy update and data on both MA enrollment and risk-adjusted benchmark rates,6,21,22 each 0.1 of the RAF has a nationally averaged valuation of $74.85 per member per month (PMPM), or $898.20 per member per year (PMPY). If providers document and code for a specific diabetic complication in CY2016, an MAO could anticipate a risk-adjusted revenue in 2017 (PY2016) that is 3-fold greater ($238.02 PMPM or $2856.28 PMPY) than if diabetes were coded without any specified complication ($77.84 PMPM or $934.13 PMPY). Assuming diagnostic coding accuracy, an MAO would be reimbursed at a higher rate during the PY for enrolling sicker MA members during the CY.
The CMS-HCC payment model thus should provide incentives to MAOs that reward continuous, high-value care from its contracted providers.19,20 Of all the different types of MA plans available, coordinated care plans are the majority, of which health maintenance organizations (HMOs) account for 52% of all MA plans.6 In applying for an HMO-type MA plan, the MAO must include “a network of providers that are under contract or arrangement with the organization to deliver the benefit package approved by CMS.”8 How MAOs reimburse providers has gone without much input from CMS, and most providers prefer FFS reimbursement.8,23,24 For example, in Oregon, the HMO-type MA plan denoted by CMS contract #H3805 has capitated agreements with only a quarter of its contracted providers, and the rest are reimbursed on an FFS schedule (eAppendix [eAppendices available at www.ajmc.com]). If the MAO’s capitated reimbursement from CMS is greater than its FFS disbursement to providers, there might be no incentive to encourage high-value care.9,10,13,14
The primary aim of this study tested the following hypothesis: value-based contracts augment the CMS-HCC payment model’s ability to generate cost efficiencies and improve clinical outcomes. Because local provider groups remain the core business unit for both MAO revenue generation and overall cost-of-care management, a secondary aim was to delineate the specific clinical practice transformations implemented by a provider group pursuant to such contracting changes.
Study Population, Study Design, and Provider Groups
The study population consisted of community-dwelling MA members 65 years or older, enrolled in 1 HMO-type MA plan (CMS contract #H3805),6 and ascribed to 1 of 2 provider groups for their primary care in the Portland, Oregon, metropolitan statistical area (MSA). These members had to be enrolled entirely through CY2008, which was the preintervention period (eAppendix and eAppendix Figure 1). The intervention occurred in CY2009 when the MAO’s contract with Provider Group A differed from that with Provider Group B (Table 1). Effective CY2009, Provider Group A became financially responsible for most healthcare services (full-risk capitation) for intervention-group members. In turn, starting in 2010 (PY2009), the provider group would receive most of the CMS-HCC risk-adjusted monthly capitated payment (RAF gainshare). In response, Provider Group A appointed an “HCC Physician Champion” (HPC) in CY2009. The HPC’s mission was to improve HCC documentation and coding, as well as to develop cost-effective approaches to primary care delivery. The contracting change and HPC assignation were considered a single intervention—all initiated in CY2009. In the control group, the MAO continued FFS reimbursement for Provider Group B. The postintervention period began on January 1, 2009, and ended on December 31, 2012, after which a revised CMS-HCC model (V22) was introduced and would have produced nonuniform changes in the RAF.25,26
Based on the timing of this intervention, an intervention—control, preintervention–postintervention, difference-in-differences (DID) approach among matching cohorts sought to determine the intervention’s effect on the RAF, utilization, and survival. This approach within 1 MSA inherently controlled for variations in care because of geography, time of enrollment, MAO plan administration, provider groups, and enrollee characteristics.27-29 A smaller, well-controlled study at the provider-group level with greater internal validity was chosen over one that compiled and summarized large datasets, which, although may provide greater external validity, would obfuscate exploring the specific clinical practice transformations implemented at the provider-group level. Provider Group A had 7 clinic locations, and Provider Group B had 5 locations, all within 2 to 18 miles from one another. Provider Group A had 25 primary care specialists (16 in internal medicine, 9 in family medicine) and Provider Group B had 19 (5 in internal medicine and 14 in family medicine). By the end of the study period, both provider groups were Oregon Health Authority—certified Patient-Centered Primary Care Homes (PCPCHs). Six of Provider Group A’s 7 clinic locations were PCPCH-accredited in December 2012, with the last one certified in 2014. All of Provider Group B’s clinic locations were accredited slightly earlier, in June 2012.
Propensity Score Model and Nearest-Neighbor Matching
Plan enrollment and provider-group election are by enrollee choice and subject to selection bias. Covariates known to affect healthcare utilization and expenditure include age, sex, ethnicity, original reason for Medicare entitlement, and disease burden.30 By not applying the appropriate preprocessing methodologies to ensure comparison of statistically similar groups,31 previous conclusions on MA healthcare delivery vis-à-vis traditional FFS Medicare may have been compromised.27,28,32-35 For example, past studies relied on self-reported survey data for health status and were subject to recall bias.27,28,32,34,35 This study’s proprietary access to full encounter claims data for MA enrollees facilitated 2 objective measures of disease burden: the Charlson Comorbidity Index (CCI) and the number of CMS Chronic Conditions Warehouse (CCW) categories (eAppendix). Using the aforementioned covariates, the logistic regression model created propensity scores for each member.31 Nearest-neighbor matching based on propensity scores created 2 well-balanced, statistically similar groups for subsequent analysis (eAppendix Table 1 and eAppendix Figure 2), as delineated in Table 1.
Statistical and Economic Analyses
Primary outcomes included: 1) RAF data at the member level obtained from CMS’ Monthly Membership Report, 2) utilization based on full-encounter claims data, and 3) survival data obtained from CMS’ Daily Transactional Reply Report (eAppendix). A linear regression model evaluated the RAF as a continuous variable. Poisson regression models examined utilization as count data. DID analyses compared postintervention (CY2009-CY2012) with preintervention (CY2008) data. For survival, the log-rank test—using either time-on-study or age of the enrollees on the time scale—evaluated Kaplan-Meier survival curves for the 2 groups.36,37 Permutation testing (randomization inference) validated the DID and survival analyses (eAppendix).38 Statistical analyses were performed within the R statistical computing environment (R Foundation for Statistical Computing, Vienna, Austria).
RAF-based revenue was estimated using CMS’ published rates for Multnomah County for each year (eAppendix).22 FFS Medicare expenditures extracted from the Medical Expenditure Panel Survey data files provided cost-of-care data according to place of service (eAppendix). Since the last CMS-HCC-derived capitated payment was disbursed in 2013 (PY2012), all amounts were indexed to 2013 dollar values using the appropriate component of the Personal Healthcare Expenditure Index (eAppendix).
Effect on RAF
The HPC’s sequential endeavors included: 1) in the beginning of 2009, focusing on coding specificity for diabetes and its complications; 2) toward the end of 2009, auditing all charts for disparities between clinical documentation and HCC coding; and 3) in the beginning of 2010, 1-on-1 mentoring of colleagues whose panel of intervention-group members had a mean RAF in the lower 50th percentile. Figure 1 summarizes the mean RAF for both groups over the 5-year study period (top panel) and the changes from preintervention (CY2008) values (bottom panel).
Based on the DID analysis, in 2009, the intervention group increased the RAF by 4.1% compared with the control group (P = .031). In other words, a control-group member whose RAF was 1.000 in 2009 would have increased its documented RAF to 1.041 if in the intervention group. Subsequent yearly analysis demonstrated the following multiplicative increases for the intervention group, relative to control: 1.072 from 2009 to 2010 (P <.001); 1.071 from 2010 to 2011 (P = .002); and 1.069 from 2011 to 2012 (P = .010). Randomization inference confirmed that the intervention group increased the RAF at a greater rate than control (P <.001). Over the postintervention period, the increase in the RAF generated an additional $2,519,544 per 1000 intervention-group members (eAppendix Table 2), averaging $629.89 PMPY.
Effect on Utilization
During the preintervention period, the intervention group had lower office-based utilization (6713 visits/1000 members) than the control group (7220 visits/1000 members). In 2009, the intervention group increased its office visits (7642/1000 members), comparable to the control group (7637/1000 members). Subsequently, relative to the control group and adjusted to 1000 members, the intervention group had 346 more office-based visits in 2010 (P = .016), 417 more in 2011 (P = .008), and 415 more in 2012 (P = .026). DID analysis yielded an exponentiated coefficient of 1.112 (P = .005), and randomization inference confirmed this statistical significance (P <.001). If reimbursed on a Medicare FFS schedule, additional expenditures for the increased office-based utilization for the intervention group would have been $258,334 per 1000 members (eAppendix Table 3), which Provider Group A assumed because of full-risk capitation. Furthermore, Provider Group A used this optimized RAF data to manage the increased workload. An index analysis validated Provider Group A’s risk-stratification process (Table 2); by clustering intervention-group members into 3 groups based on 2010 office-based visits, statistically different mean RAF, CCI, and subsequent office-based utilization counts were observed for all terciles (Table 2).
Next, the effect of increasing office-based visits on hospital-based services was analyzed. To this end, incident rate ratios (IRRs) for each place of service utilization were normalized to member-months. As with the previous analysis adjusted to 1000 members and as demonstrated in Figure 2 when adjusted to member-months, intervention-group members were more likely to have increased office-based utilization (P <.001). The intervention concomitantly reduced emergency department (ED) visits and inpatient hospital (IP) admissions by 11.2% (P = .001) and 11.9% (P = .002), respectively (Figure 2). Randomization inference confirmed that the reduction in IP admissions was statistically significant (P = .037), whereas the reduction in ED visits approached statistical significance (P = .086). These 154 fewer ED visits per 1000 members yielded $100,915 in estimated savings, and 143 fewer IP admissions per 1000 members saved $1,756,869 (eAppendix Table 3). Because the intervention reduced both readmissions (numerator) and death-censored incident admissions (denominator), the actual readmission rate was not statistically different compared with the control group. The intervention also influenced certain preventive care measures. As evidenced in Figure 2, intervention-group members were almost 3 times more likely to undergo preventive care visits (P <.001), and intervention-group women 74 years or younger were 28% more likely to undergo screening mammography (P = .012). Randomization inference confirmed these findings for preventive care visits (P <.001) and screening mammography (P = .005). Although the intervention group had higher rates of screening colonoscopy (for those 75 years or younger), the difference was not statistically significant.
Effect on Survival
The intervention group’s overall survival rate was 82%, and the control group’s was 76%. This 6% survival benefit first became apparent at 16 months after the intervention (Figure 3, left panel), coinciding with the first year that the intervention group had higher office-based utilization than the control group. Age provided a natural time-scale for calculating the hazard of death for this elderly population with multiple comorbidities and a higher risk of all-cause mortality.36,37 Intervention-group members had a 32.8% lower hazard of dying (P <.001). The survival benefit was more apparent among those aged 82 to 96 years (Figure 3, right panel). Randomization inference confirmed these survival data, whether time (P <.001) or age (P <.001) was the time scale.
In this study of MA enrollees, a contracting change between the MAO and a provider group leveraged the core initiatives proposed for reforming the Medicare FFS program, namely APM participation and increasing risk assumption. Key findings included: 1) MAO—provider collaboration optimized the RAF, 2) RAF optimization supported a risk-stratification process in the effective triage of office-based care, 3) intensive office-based care concomitantly reduced hospital-based services, and 4) this shift in healthcare delivery improved survival. The survival benefit first became apparent in CY2010, coinciding with the completion of the HPC’s mentoring program and the first year of significantly increased office-based utilization. This sequence of events leads to the assertion that improved survival is related to and attributable to enhanced CMS-HCC data and value-based contracting, which then transform primary care delivery.
In redesigning primary care practices, a CMMI-funded study noted that the focus of most clinicians is to identify “new funding streams for their program activities.”39 The CMS-HCC payment model attempts to account for anticipated future costs through an assessment of disease burden by coding HCCs during face-to-face clinical encounters. HCC coding is an onerous task for physicians, especially with each revision of the CMS-HCC model.26 For example, in V12 of the CMS-HCC model that was in effect during the study period, 2935 of the 14,567 (20.2%) unique International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes were aggregated into 70 HCCs that impact the RAF. In V22 that was phased in CY2013, 8667 of the 69,823 (12.4%) ICD-10-CM codes were grouped into 79 HCCs.
During the gradual implementation phase of the CMS-HCC payment model (2004-2007), Provider Group A began advancing its HCC documentation and coding practices, in contrast to most of the MAO’s other contracted provider groups (as exemplified by Provider Group B). Although the 2 groups had similar disease burdens based on CCI and CCW categories, the RAF scores always were different because of Provider Group A’s eagerness to learn about HCC during CMS-HCC’s implementation phase (Figure 1, top panel).
With recognition of this potential new funding stream and the goodwill established from CMS-HCC revenue previously generated for the MAO, Provider Group A effectively argued for an RAF gainshare arrangement. The intervention further increased RAF scores relative to the control group (bottom panel, Figure 1), thus optimizing CMS-HCC’s value in assessing the intervention group’s disease burden and prospective healthcare expenditures. Such gainshare arrangements also benefit MAOs because federal statutes constrain MA plan profitability since the Medical Loss Ratio must be 85% or greater.40 By including a full-risk capitation clause to the RAF gainshare, the MAO locked into a fixed rate of return and thereby limited both its upside and downside risk, analogous to the financial instrument of interest rate swaps.
MAOs receive monthly capitated prospective payments from CMS and become financially responsible for Part A and Part B benefits for their MA members.8 Cost-of-care management agreements between an MAO and its provider networks are not mandatory. Instead, the Code of Federal Regulations meekly states that MA plans “may [emphasis added] include mechanisms to control utilization, such as referrals from a gatekeeper for an enrollee to receive services within the plan, and financial arrangements that offer incentives to providers to furnish high-quality and cost-effective care.”8 When assuming fiduciary risk, MAOs reflexively create their own programs for utilization management, care coordination, and disease prevention. Alternatively, risk-contracting with clinicians can reduce the MAO’s clinical footprint. By enabling “incentives to providers to furnish high-quality and cost-effective care,”8 there is a statutory ease of administration for MAOs to risk-contract with providers, based on percentages of the CMS-HCC capitated revenue and member premium. These percentages can range from 30% to over 80%, depending on the magnitude of monetary risk providers agree to accept.
Prior to 2008, many of Provider Group A’s other contracts required it to reimburse all external specialists and for the plans to pay all hospital claims. The difference then was split equally between the provider group and the plans (50/50—shared risk). With this past experience, Provider Group A agreed to the challenges of taking on full monetary risk for healthcare expenditures—including facility charges—with only a few exceptions or “carve-outs” (eg, out-of-area, vision, hearing aid, most transplantation services). In contrast, Provider Group B remained on a strict FFS schedule, with no risk assumption of other services.
Despite previous risk-contracting, Provider Group A still had a steep learning curve to potentiate high-value care. Both provider groups were PCPCH-certified, and—similar to previous reports on patient-centered medical homes—this certification did not spontaneously transform clinical practice.41,42 Because risk-stratification can cultivate novel care coordination protocols and workplace efficiency, CMMI has funded studies on targeting high-risk patients.43,44 A novel finding of the current study was that, after optimizing HCC coding, intervention-group enrollees at greatest risk for hospital-based services could be identified and then managed with intensive office-based care. Provider Group A standardized care with: 1) a triage system so that frail, complex patients (as determined by higher RAF scores) had immediate access to their primary care physicians or nurse practitioners (Table 2); 2) improved care coordination that booked clinic visits soon after hospitalization, as the appointment dates proposed by hospital-based physicians often were overdue; and 3) scheduling members with specific HCCs (ie, heart failure [HF], chronic obstructive pulmonary disease [COPD], and diabetes) at regular intervals because elective office visits by members with these HCCs often were prodromal of imminent hospital-based services. The current study thus provides preliminary evidence that the RAF can direct primary care. In parallel, MAOs that participate in CMMI’s upcoming VBID demonstration project can formulate different plan benefits for enrollees with the following HCCs: HF, COPD, diabetes, cerebrovascular disease, hypertension, coronary artery disease, and mood disorders.15 Unlike the VBID demonstration project, in this study, the providers who best knew the enrollees designed these value-based protocols and not a third-party payer. Provider Group A’s clinical practice transformation increased office-based utilization and reduced hospital-based services, with an estimated cost benefit of $2,116,118 per 1000 members, or $529.03 PMPY.
One clinical vignette exemplified the type of care intervention-group members received. The HPC cared for an 80-year-old man with ischemic cardiomyopathy, whose HF worsened and resulted in acute renal failure. Given the poor prognosis, the patient’s cardiologist recommended hospice care. After the patient beseechingly asked about alternatives, the HPC embarked on an intensive schedule of weekly to fortnightly office visits with continual medication readjustment. The patient lived for another 3.5 years with good quality of life, only 1 hospital admission, and no admissions for HF. Evidence-based medicine has determined that exacerbation of HF is an “ambulatory care—sensitive condition”; nevertheless, ED visits for HF are commonplace and a significant drain on Medicare’s budget.45,46 Provider Group A shifted its practice standards after full-risk contracting. Respected clinicians within a provider group (eg, the HPC) can advocate for unique critical pathways, such as linking the CMS-HCC fiduciary model to innovative healthcare delivery.
By comparing disparate populations of MA and traditional Medicare beneficiaries and analyzing incomplete utilization data, previous reports have attempted to demonstrate that MA promulgates high-value care.9,27,28,31 By comparing statistically similar groups of MA enrollees and analyzing full encounter claims data, the current study adds further credible evidence that value-based contracting generates cost efficiencies and improves health. Following this shift in utilization patterns, intervention-group members experienced a 6% survival benefit. Because a DID approach can underestimate confidence intervals and lead to erroneous conclusions about healthcare policy reform measures,38,47 randomization inference was employed to validate that the intervention’s effects were statistically significant with adequate statistical power to support the current conclusions (eAppendix). There were no deaths for the first postintervention year (Figure 3, left panel) in either group. Study inclusion required both a minimum age of 65 and survival during the entire preintervention period. Although the average life expectancy in the United States is 78.8 years, those aged 65 years can expect to live another 19.3 years, and those aged 75, another 6.6 years.48 The study requirement of living the entire preintervention period most likely preselected for subsequent survivability.
Because of the observational nature of this study, survival differences might be attributable to factors not examined, such as environmental or social ones. Focusing the study within a single MSA and matching cohorts with propensity scores attempted to control for such factors. In a randomized trial, the propensity score is a known function. On the other hand, in an observational study, it is always unknown because differences in unobserved baseline characteristics may influence observed outcomes. Confirming the survival analysis with randomization inference rejoined many theoretical issues of relying solely on propensity score matching in this observational quasi-experiment. Of course, the most notable limitation is whether these findings can be applied to the broader MA population. Future studies must investigate how MAOs and their differing arrangements with providers interact with ethnic and geographic variations in healthcare delivery.28,29
Value-based contracting between MAOs and providers generate cost efficiencies and improve clinical outcomes in MA, which is the ultimate aim of the current initiatives for Medicare FFS reform. Empiric economic analyses—ignorant of differing MAO—provider contracting arrangements—have challenged the value of subsidizing the MA program and calculated a rather poor return on the taxpayers’ investment.9,10,13,14 CMMI’s VBID model thus has been aptly timed, beginning on January 1, 2017, and running for 5 years.15 In the interim and with minimal effort, MAOs and provider groups can alter contracting arrangements. In turn, providers organically develop innovative primary care strategies that are cost-efficient and improve clinical outcomes to the benefit of all MA program stakeholders.
The reporting and interpretation of these data are the sole responsibility of the authors. The authors wish to thank: 1) Joanna Martson, Director of Market Consultation (OR, WA, AK, HI) at Optum, for her additional insights on the Oregon MA market; 2) Larry Kunz, Director of Research, and his staff at SavvySherpa, Inc for their independent audit of our data files, assistance with our econometric analysis (Philip R. Vande Kamp, PhD), and fastidious statistical analytical support (Daniel J. Halterman, MS); and 3) Eric M. Peterson, Deputy General Counsel at Optum, for his input during the editing of this paper. Dr Howell’s current affiliation is with the Heritage Provider Network.
Author Affiliations: Optum, UnitedHealth Group (AKM, GKT, RVF, SCH), Santa Ana, CA.
Source of Funding: None.
Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (AKM); acquisition of data (AKM, GKT, RVF); analysis and interpretation of data (AKM, SCH); drafting of the manuscript (AKM, SCH); critical revision of the manuscript for important intellectual content (AKM, GKT, RVF, SCH); statistical analysis (AKM, SCH); provision of patients or study materials (RVF, GKT); obtaining funding (AKM); administrative, technical, or logistic support (AKM, GKT, RVF, SCH); and supervision (AKM).
Address Correspondence to: Aloke K. Mandal, MD, PhD, Optum, Mail Code CA152-0243, 3110 W Lake Center Dr, Santa Ana, CA 92704-6917. E-mail: firstname.lastname@example.org.
1. Proposed rule: Medicare program; Merit-Based Incentive Payment System (MIPS) and Alternative Payment Model (APM) incentive under the physician fee schedule, and criteria for physician-focused payment models. Federal Register website. http://federalregister.gov/a/2016-10032. Published May 9, 2016. Accessed December 31, 2016.
2. Better care. smarter spending. healthier people: improving quality and paying for what works. CMS website. https://www.cms.gov/Newsroom/MediaReleaseDatabase/Fact-sheets/2016-Fact-sheets-items/2016-03-03-2.html. Published March 3, 2016. Accessed December 31, 2016.
3. Burwell SM. Setting value-based payment goals—HHS efforts to improve U.S. health care. N Engl J Med. 2015;372(10):897-899. doi: 10.1056/NEJMp1500445.
4. Gosden T, Forland F, Kristiansen I, et al. Capitation, salary, fee-for-service and mixed systems of payment: effects on the behaviour of primary care physicians. Cochrane Database Syst Rev. 2000;(3):CD002215. doi: 10.1002/14651858.CD002215.
5. Flodgren G, Eccles MP, Shepperd S, Scott A, Parmelli E, Beyer FR. An overview of reviews evaluating the effectiveness of financial incentives in changing healthcare professional behaviours and patient outcomes. Cochrane Database Syst Rev. 2011;(7):CD009255. doi: 10.1002/14651858.CD009255.
6. Medicare Advantage/Part D contract and enrollment data. CMS website. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/MCRAdvPartDEnrolData/. Updated December 6, 2016. Accessed December 31, 2016.
7. HHS FY 2017 budget in brief—CMS—Medicare. HHS website. http://www.hhs.gov/about/budget/fy2017/budget-in-brief/cms/medicare/index.html. Updated February 12, 2016. Accessed December 31, 2016.
8. Part 422—Medicare Advantage Program. US Government Publishing Office website. http://www.ecfr.gov/cgi-bin/retrieveECFR?gp=&SID=5942b7e3c726955e6601f4af927598b7&mc=true&r=PART&n=pt42.3.422. Updated December 29, 2016. Accessed December 31, 2016.
9. Pizer SD, Frakt AB, Feldman R. Nothing for something? estimating costs and value for beneficiaries from recent Medicare spending increases on HMO benefits and drug benefits. Int J Health Care Finance Econ. 2009;9(1):59-81. doi: 10.1007/s10754-008-9047-x.
10. Song Z, Landrum MB, Chernew ME. Competitive bidding in Medicare: who benefits from competition? Am J Manag Care. 2012;18(9):546-552.
11. Guram JS, Moffett RE. The Medicare Advantage success story—looking beyond the cost difference. N Engl J Med. 2012;366(13):1177-1179. doi: 10.1056/NEJMp1114019.
12. Baicker K, Chernew ME, Robbins JA. The spillover effect of Medicare managed care: Medicare Advantage and hospital utilization. J Health Econ. 2013;32(6):1289-1300. doi: 10.1016/j.jhealeco.2013.09.005.
13. Duggan M, Starc A, Vabson B. Who benefits when the government pays more? pass-through in the Medicare Advantage program [NBER working paper No. 19989]. The National Bureau of Economic Research website. http://www.nber.org/papers/w19989. Published March 2014. Accessed December 31, 2016.
14. Cabral M, Geruso M, Mahoney N. Does privatized health insurance benefit patients or producers? evidence from Medicare Advantage [NBER working paper No. 20470]. The National Bureau of Economic Research website. http://www.nber.org/papers/w20470. Published September 2014. Accessed December 31, 2016.
15. Medicare Advantage Value-Based Insurance Design Model. CMS website. https://innovation.cms.gov/initiatives/vbid/. Updated December 30, 2016. Accessed December 31, 2016.
16. Brown RS, Clement DG, Hill JW, Retchin SM, Bergeron JW. Do health maintenance organizations work for Medicare? Health Care Financ Rev. 1993;15(1):7-23.
17. Riley G, Tudor C, Chiang YP, Ingber M. Health status of Medicare enrollees in HMOs and fee-for-service in 1994. Health Care Financ Rev. 1996;17(4):65-76.
18. Mello MM, Stearns SC, Norton EC, Ricketts TC 3rd. Understanding biased selection in Medicare HMOs. Health Serv Res. 2003;38(3):961-992.
19. Pope GC, Kautter J, Ellis RP, et al. Risk adjustment of Medicare capitation payments using the CMS-HCC model. Health Care Financ Rev. 2004;25(4):119-141.
20. Pope GC, Kautter J, Ingber MJ, Freeman S, Sekar R, Newhart C. Evaluation of the CMS-HCC risk adjustment model: final report. CMS website. https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/downloads/evaluation_risk_adj_model_2011.pdf. Published March 2011. Accessed December 31, 2016.
21. Announcement of calendar year (CY) 2017 Medicare Advantage capitation rates and Medicare Advantage and Part D payment policies and final call letter. CMS website. https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Downloads/Announcement2017.pdf. Published April 4, 2016. Accessed December 31, 2016.
22. Ratebooks & supporting data. CMS website. https://www.cms.gov/medicare/health-plans/medicareadvtgspecratestats/Ratebooks-and-Supporting-Data.html. Updated April 25, 2016. Accessed December 31, 2016.
23. Bodenheimer T, Pham HH. Primary care: current problems and proposed solutions. Health Aff (Millwood). 2010;29(5):799-805. doi: 10.1377/hlthaff.2010.0026.
24. Tilburt JC, Wynia MK, Sheeler RD, et al. Views of US physicians about controlling health care costs. JAMA. 2013;310(4):380-388. doi: 10.1001/jama.2013.8278.
25. Announcement of calendar year (CY) 2013 Medicare Advantage capitation rates and Medicare Advantage and Part D payment policies and final call letter. CMS website https://www.cms.gov/Medicare/Prescription-Drug-Coverage/PrescriptionDrugCovContra/Downloads/RateNoticeA.pdf. Published April 2, 2012. Accessed December 31, 2016.
26. Comments on proposed changes to the CMS-HCC risk adjustment model for payment year 2017. CMS website. https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Downloads/RiskAdj2017ProposedChangesComments.pdf. Published December 18, 2015. Accessed December 31, 2016.
27. Landon BE, Zaslavsky AM, Saunders RC, Pawlson LG, Newhouse JP, Ayanian JZ. Analysis of Medicare Advantage HMOs compared with traditional Medicare shows lower use of many services during 2003-09. Health Aff (Millwood). 2012;31(12):2609-2617. doi: 10.1377/hlthaff.2012.0179.
28. Matlock DD, Groeneveld PW, Sidney S, et al. Geographic variation in cardiovascular procedure among Medicare fee-for-service vs Medicare Advantage beneficiaries. JAMA. 2013;310(2):155-162. doi: 10.1001/jama.2013.7837.
29. Ayanian JZ, Landon BE, Newhouse JP, Zaslavsky AM. Racial and ethnic disparities among enrollees in Medicare Advantage plans. N Engl J Med. 2014;371(24):2288-2297. doi: 10.1056/NEJMsa1407273.
30. National Center for Health Statistics. Analytic issues in using the Medicare enrollment and claims data linked to NCHS surveys. CDC website. http://www.cdc.gov/nchs/data/datalinkage/cms_medicare_analytic_issues_final.pdf. Published December 2012. Accessed December 31, 2016.
31. Haukoos JS, Lewis RJ. The propensity score. JAMA. 2015;314(15):1637-1638. doi: 10.1001/jama.2015.13480.
32. Lemieux J, Sennett C, Wang R, Mulligan T, Bumbaugh J. Hospital readmission rates in Medicare Advantage plans. Am J Manag Care. 2012;18(2):96-104.
33. Wong ES, Hebert PL, Maciejewski ML, et al. Does favorable selection among Medicare Advantage enrollees affect measurement of hospital readmission rates? Med Care Res Rev. 2014;71:367-383. doi: 10.1177/1077558714533823.
34. McWilliams JM, Hsu J, Newhouse JP. New risk-adjustment system was associated with reduced favorable selection in Medicare Advantage. Health Aff (Millwood). 2012;31(12):2630-2640. doi: 10.1377/hlthaff.2011.1344.
35. Rahman M, Keohane L, Trivedi AN, Mor V. High-cost patients had substantial rates of leaving Medicare Advantage and joining traditional Medicare. Health Aff (Millwood). 2015;34(10):1675-1681. doi: 10.1377/hlthaff.2015.0272.
36. Lamarca R, Alonso J, Gomez G, Munoz A. Left-truncated data with age as time-scale: an alternative for survival analysis in the elderly population. J Gerontol A Biol Sci Med Sci. 1998;53(5):M337-M343.
37. Cho H, Mariotto AB, Mann BS, Klabunde CN, Feuer EJ. Assessing non-cancer-related health status of US cancer patients: other-cause survival and comorbidity prevalence. Am J Epidemiol. 2013;178(3):339-349. doi: 10.1093/aje/kws580.
38. Kaestner R. Did Massachusetts health care reform lower mortality? no according to randomization inference. Stat Public Policy. 2016;3(1):1-6. doi: 10.1080/2330443X.2015.1102667.
39. Mathematica Policy Research. Evaluation of Health Care Innovation Awards (HCIA): primary care redesign programs—second annual report. CMS website. https://downloads.cms.gov/files/cmmi/hcia-primarycareredesignprog-secondannualrpt.pdf. Published March 2016. Accessed December 31, 2016.
40. Medicare program; medical loss ratio requirements for the Medicare Advantage and the Medicare Prescription Drug Benefit Programs; final rule. US Government Printing Office website. https://www.gpo.gov/fdsys/pkg/FR-2013-05-23/pdf/2013-12156.pdf. Published May 23, 2013. Accessed December 31, 2016.
41. Jaén CR, Ferrer RL, Miller WL, et al. Patient outcomes at 26 months in the patient-centered Medical Home National Demonstration Project. Ann Fam Med. 2010;8(suppl 1):S57-S67; S92. doi: 10.1370/afm.1121.
42. Jackson GL, Powers BJ, Chatterjee R, et al. Improving patient care. the patient centered medical home. a systematic review. Ann Intern Med. 2013;158(3):169-178.
43. NORC at the University of Chicago. First annual report: HCIA complex/high-risk patient targeting. CMS website. https://innovation.cms.gov/Files/reports/HCIA-CHSPT-FirstEvalRpt.pdf. Published November 7, 2014. Accessed December 31, 2016.
44. NORC at the University of Chicago. Second annual report: HCIA complex/high-risk patient targeting. CMS website. https://downloads.cms.gov/files/cmmi/hcia-complexhighriskpattargeting-secondevalrpt.pdf. Published March 2016. Accessed December 31, 2016.
45. Berwick DM, Nolan TW, Whittington J. The triple aim: care, health, and cost. Health Aff (Millwood). 2008;27(3):759-769. doi: 10.1377/hlthaff.27.3.759.
46. Dharmarajan K, Hsieh AF, Lin Z, et al. Diagnoses and timing of 30-day readmissions after hospitalization
for heart failure, acute myocardial infarction, or pneumonia. JAMA. 2013;309(4):355-363. doi: 10.1001/jama.2012.216476.
47. Sommers BD, Long SK, Baicker K. Changes in mortality after Massachusetts health care reform: a quasi-experimental study. Ann Intern Med. 2014;160(9):585-593. doi: 10.7326/M13-2275.
48. Table 15: life expectancy at birth, at age 65, and at age 75, by sex, race, and Hispanic origin: United States, selected years 1900—2014. CDC website. http://www.cdc.gov/nchs/data/hus/2015/015.pdf. Updated April 21, 2016. Accessed December 31, 2016.