Can’t sign in? Forgot your password?
Enter your email address below and we will send you the reset instructions
If the address matches an existing account you will receive an email with instructions to reset your password.
Can’t sign in? Forgot your username?
Enter your email address below and we will send you your username
If the address matches an existing account you will receive an email with instructions to retrieve your username
Your password has been changed
While the COVID-19 pandemic rages, the past two years have seen another epidemic of a far different type—in financing and acquisitions of firms focused on serving Medicare beneficiaries. These firms include physician practices, notably primary care practices (PCPs); management services organizations (MSOs) that aggregate practices; and Medicare Advantage (MA) insurers. In this arena, the combined activity of private equity and venture capital firms, initial public offerings, special purpose acquisition companies (SPACs), and insurance company purchases of MA-focused firms has soared: more than $50 billion in valuation has been created in the past 18 months, dwarfing the speculative bubble for physician practice management companies in the 1990s.
One indicator of the exuberance underlying this “Medicare Gold Rush” is the amount per covered life implicit in a firm’s overall valuation. Historically, per-life valuations in MA have ranged from $4,000 to $10,500. Exhibit 1 shows per-life valuations for a sample of recent transactions. The average is $87,000 per beneficiary. Most of the firms acquired or financed are PCPs or MSOs that typically produce no margins—just an average take-home income of $240,000 per physician. The first six are participants in the Centers for Medicare and Medicaid Services’ (CMS) new Direct Contracting Model, which we shall discuss further in Part Two of this post.
Source: Authors’ analysis of publicly available information
If these valuations for PCP Practices and MSOs look hard to believe, that is because they are. Annual Medicare Part A and B spending per individual is roughly $12,000. PCPs typically receive only 5 percent of that amount. By what logic would an investor pitch in at a rate equal to almost eight times the total annual health care expenditure per capita for PCPs with no margin? Investment at this level is smoke; what is the fire?
In this two-part post, we will attempt to explain the perverse MA business model that underlies this elevated level of investment, and we will explore its connection to the Direct Contracting model now being tested by CMS. The story is complex, but we think it is worth telling because the stakes for beneficiaries, the public treasury, and our health care system are very high. This business model is distorting health care delivery, creating excessive costs for taxpayers and Medicare beneficiaries, draining the Medicare Trust Fund, obstructing the badly needed value transformation of American health care, and diverting the money needed to fund other social services and goods.
Part one of this post focuses on MA. Part two, to be published tomorrow, will discuss Direct Contracting and suggest some reforms for both MA and Direct Contracting. We also offer a broader reform agenda that calls for expanding the accountable care organization (ACO) model by working directly with providers, rather than investors.
Four main business realities drive the interest in Medicare-related acquisitions. First is the expected doubling of Medicare spending from $800 billion in 2019 to $1.6 trillion in 2028 as Baby Boomers age. Second is the reality that MA harbors an arbitrage game in which CMS consistently overpays MA Plans with no demonstratable clinical benefit to patients. Third is the heavily subsidized and distorted market dynamics that result from these overpayments. Fourth is the Trump administration’s creation of the Direct Contracting Model as a vehicle for privatizing Medicare’s projected 2028 $1.6 trillion spend.
As exhibit 2 shows, the Medicare Payment Advisory Committee (MedPAC) has documented approximately $140 billion in MA overpayments over the past 12 years. MedPAC further concludes that risk adjustment overpayments are currently increasing. Kronick and Chua have estimated savings at $355 billion over the next eight years if just risk-score related overpayments were eliminated. (Exhibit 3)
Source: The MedPAC Blog, March 3, 2021 (authors’ conversion of MA overpayments to dollars from percent of FFS payments as calculated by MedPAC)
Source: Estimating Impact of Coding Intensity Adjustment: Exhibit A.7, DECI (Demographic Estimate of Coding Intensity) p.28
MedPAC has documented MA plans’ ability to obtain overpayments by cherry picking counties with favorable benchmarks and escalating quality bonus payments through contract manipulation. These tactics add 8 percent in program costs in 2021, leading to MA payments 2 percent above fee-for-service (FFS) Medicare payments. MedPAC has made important recommendations to Congress to address these issues. We will focus primarily on the “risk-score gaming” that increases MA payments and the resulting marketplace dynamics impacting health care delivery across America.
The shortcomings of CMS’s Hierarchical Condition Category (HCC) risk adjustment system have been well described since its full implementation in 2006. Simply stated, MA plans can draw enormous overpayments by submitting diagnosis codes that create more HCCs per person. While the codes are, presumably, accurate, the dollar coefficients used in MA payment logic are inflated because they were modeled using markedly under-coded FFS data. “Risk-score gaming” overpayments come from inaccurate pricing of HCCs. Congress and every administration since 2006 have avoided fixing this inaccuracy, in part because of plans’ enormous political clout.
Exhibit 4 illustrates how the MA bid model rewards increased coding. (Part D costs are excluded). Total CMS Premium includes two pieces. One is the Plan bid to provide A and B services including profit and administration, multiplied by the risk score. The other is a rebate Medicare pays to the plan, calculated on average as 65 percent of the amount it bids below the risk adjustment benchmark; CMS retains the other 35 percent. Both pieces increase as the risk score goes up. The A and B Medical Expense in each column is unchanged since the population is the same.
Column A shows the resulting financials for a 2021 average plan described by MedPAC. Despite the 1.0 Risk Score, Medicare pays roughly 1 percent more than FFS, due to the benchmark and quality issues noted above. Column B illustrates the theoretical results for a very highly competitive market where the Plan uses most extra revenue to increase rebates, not profits. CMS overpayments increase by $58 million annually per 100,000 beneficiaries, with beneficiaries paying $12 million more in Part B Premiums.
Source: Authors’ model. MLR includes additional expense for code collection.
Researchers in multiple studies have demonstrated that MA markets are not so competitive, and that Plans tend to use additional revenue to improve profits more than member benefits. Plan bid documents are not public, so we cannot show this directly. Columns C and D in exhibit 4 result from our more conservative model based on MedPAC’s 2021 Bid Analysis and Jacobs and Kronick’s empirical analysis. The resulting estimates are that for each 0.1 increase in risk score, an average plan would use roughly $11 PMPM for profit and $14 to improve premiums and benefits. For each 0.1 increase, estimated profits increase about 25 percent, Medicare overpayments for 100,000 beneficiaries increase by $58 million, of which $8 Million will be paid by Part B beneficiaries. Individual plans’ actual use of risk score revenue will vary widely depending on their strategic weighting of profitability vs. growth.
The Demographic Estimate of Coding Intensity (DECI) estimates in exhibit 3 include a 2021 MA coding intensity difference of approximately 0.13. Projected across the MA population of 26 million, each 0.1 increase in risk scores in our model results in an additional $15 billion in overpayments and $3.5 billion in additional MA plan profits at current enrollment levels. CMS would pay $13 billion of the overpayments and Medicare Part B beneficiaries would pay the other $2 billion in inflated Part B premiums. Risk-score gaming creates a major transfer of wealth from taxpayers and Medicare beneficiaries to MA plans, and it lies at the heart of the business model for most MA plans.
Supporters of MA point to the program’s growth as evidence that the privatized model works. The reality is that MA grows because the structural and risk-score gaming overpayments subsidize MA plans to offer some improved benefits, lower Part D costs, an average $5,000 out-of-pocket cap, and underutilized supplemental benefits. Low-income beneficiaries remain underinsured and subject to significant copayments and deductibles. As plans code more, risk scores go up, CMS provides more subsidies, benefits and premiums get better, and buyers choose the improved plans that cost taxpayers more. This is one distorted dynamic in the MA marketplace: the costlier the plan is to the payer (CMS), the easier it is to sell it to the customer, and the greater the profit.
This subsidized marketplace is the major reason that over the past 15 years MA plans have been by far the most popular form of health insurance company start-ups. Firms that initially targeted other segments, such as the exchanges (viz. Bright and Oscar) or Medicaid (viz. Centene and Molina), have all found their way to MA as their preferred business opportunity. Most recently new MA startups have been prominent, including Clover Health, Devoted Health, and Alignment Health.
One potential restraint on risk-score gaming is that as risk scores go up plans begin approaching the 85 percent Minimum Loss Ratio requirement under the Affordable Care Act. Plans have found a solution for that, which we label the “MA Money Machine,” the next major component of the distorted MA Marketplace.
Given the dollar magnitude, risk-score gaming becomes a central part of every MA plan’s strategy. The starting point is to get as many diagnosis codes as possible. An entire industry been created to do just that, leading to billion dollar valuations for firms, like Signify Health, that provide analytical tools to enable coding efforts or make home visits for plans and providers. Most Plans now use Artificial Intelligence (AI) HCC Tools to identify coding opportunities.
In a recent investor call, United Health remarked on the importance of home visits, noting that, as the COVID pandemic waned, their HouseCalls nurses were back in the home collecting diagnoses that should lead to improved profits in their MA plans. MedPAC and the HHS Inspector General have identified these home visits as key drivers of overpayments. But MA plans know that the best sources of more codes are providers. They have developed three well-established schemes to get more codes directly from providers, which we call “Deal 1,” “Deal 2,” and Deal 3.”
Some plans pay providers to code more diagnoses by using pay for performance metrics like HCC Gaps Closure and Recapture Rates and using AI tools to direct their efforts. Clover Health simply pays MA (and now Direct Contracting) physicians $30 per visit to use its “Clover Assistant” AI platform, which identifies coding opportunities.
Many MA insurers distort the value-based care (VBC) contract model to make it a vehicle to drive more coding. Column A in exhibit 5 illustrates the results of a legitimate value-based contracting approach similar to that used in Medicare ACOs. A medical expense target is set based on historical experience. If actual costs are less than historical costs, the provider keeps a portion of savings, contingent on quality-of-care metrics. ACOs operating under VBC models have saved CMS $1.9 billion in 2020 and more than $4 billion over the past eight years.
Column B, C, and D in Exhibit 5 show how MA plans distort VBC contracts to increase CMS costs. These MA “non-value-based care” contracts sets the target based on a percentage of the premium the plan receives for a provider’s panel of patients. The provider has good reason to focus on increasing the risk score. In column B, the risk-score increase of 0.1 drives higher premium, the target goes up, and the resulting contrived “medical cost savings” of $56 PMPM, without any actual change in costs or care, “drop through” to become provider profits. Insurer profits increase as well, since insurers collect 15 percent of a larger premium. CMS ends up paying $120 million more and beneficiaries pay for $14 million in Part B premiums.
Columns C and D show how each 0.1 RAF increase creates $87 PMPM more in Part A and B revenue, with $71 going to the provider and $15 to the plan, which can use it as profit or to improve benefits. Rebates go up $14 PMPM as well, due to the benchmark increase. Column D also shows that, any decrease in medical costs just becomes additional profit. CMS shares in none of the savings, and costs still go up $370 M in total per 100,000 beneficiaries.
These MA Percentage of Premium contracts create a continuous “Money Machine,” that allows the provider firm and plan to harvest a financial windfall just by finding more codes. As a result, providers look hard for diagnoses using various AI-enabled platforms. A current popular tactic, for example, is to screen beneficiaries for peripheral vascular disease (HCC 108), which delivers an extra $2,800 per year per patient, by ordering carotid ultrasound studies, even though the US Preventive Services Task Force recommends against such screening for the general population.
Source: Authors’ analysis
Key points comparing Deal 1 provider payment tactics (in exhibit 4) to Deal 2 Percentage of Premium contracts (in exhibit 5) include the following:
Recognizing that the largest share of the MA Money Machine profits goes to providers, some insurers have decided to own the providers outright. This tactic ensures optimal use of their sophisticated AI coding by employed staff. The parent collects both the insurance profits and the Money Machine profits. MedPAC raised the issue of whether plans with Deal 3 arrangements may be inaccurately reporting related provider incentive payments in ways that overstate medical expenses. The final row in exhibit 5 demonstrates that if such claims, which could ultimately result in profits for the plan, were excluded, the actual MLR could be in the low 70s.
United Healthcare, the most profitable of the large national MA Plans, seems to have used Deal 3 for seven years following the purchase, through its OptumHealth subsidiary, of Monarch and Applecare PCP Networks in 2014. With over 50,000 physicians owned or in affiliated independent practice associations (IPAs), United may today be the largest employer of physicians in America, and it plans to add 10,000 more physicians in 2021. The Money Machine model was described as a core driver of profitability in a recent United Healthcare C-Suite fireside chat: “OptumHealth . . . revenue per consumer served increased 29 percent for 2020 driven by expansion . . . in value-based care arrangements and increasing acuity of the care services provided.”
Over the past 15 years, the MA Money Machine has been growing as an essential business model component for many prominent physician groups, IPAs, PCP/MSOs and even some integrated systems. United Healthcare and Humana today control 12 million MA lives, almost 50 percent of the national total. Both are rapidly expanding their use of the MA Money Machine. Humana reports that two-thirds, or 2.4 million, of its individual beneficiaries are in these models. They have relied on Deal 2 historically, but recently announced the creation of Centerwell as their new Money Machine Deal 3 vehicle. United Healthcare, with a particular focus on acquiring non-profit physician groups like Reliant and Atrius Health, has said that Optum now has 2 million of its MA lives in “Value-Based Contracts” and is rapidly increasing that number. Thus at least 4.4 million people, or 17 percent of all MA members, and almost $60 billion, are involved in Deal 2 or Deal 3 Money Machine contracts today, with rapid growth ahead.
Exhibit 6 (modeling a hypothetical physician’s panel of 400 MA patients) illustrates how these contracts turn break-even primary care practices into very profitable “assets” that have attracted the attention of private investors
Source: Authors’ analysis
Over the past eight years a number of new start-up venture capital backed PCP firms have been created using the Money Machine as their core business model. They share many common features:
Recent PCP and MSO partnerships with Humana, for example, include Iora, Oak Street, Agilon, Cano Health, and Landmark. VillageMD has partnered with Aetna/Anthem, and ChenMed has partnered with Independence Blue Cross. As shown in Exhibit 1, financing for these firms has been extraordinary. Another set of PCPs and MSOs are following closely behind, including Miami Beach Family Practice and several other Direct Contracting Entities (DCEs). The primary business model for all is the MA Money Machine.
Why the rush of investors into MA primary care space? Because it is an MA Money Machine. While all can agree that we should improve compensation for primary care, these extraordinary profits are more likely to be captured by the for-profit parent entities rather than passed through to physicians delivering care.
The toll of the MA coding game, though high, has heretofore been confined to the MA portion of Medicare, that is 42 percent of all CMS beneficiaries. Under the Trump Administration, that changed. The Trump Administration avowed its intention to de-risk CMS by moving the 58 percent of Medicare beneficiaries who chose traditional coverage into MA-like full risk capitated arrangements. This full privatization of Medicare coverage would require new entities to act as financial intermediaries between CMS and non-MA beneficiaries. CMS officials decided that the same firms that benefited from risk-score gaming overpayments in MA—insurers and MA-focused Primary Care Firms (PCF’s)—should be given the opportunity to manage the $350 billion in Medicare spending for the majority of beneficiaries not in Medicare Advantage. They needed a new program to accomplish this overarching goal. The Direct Contracting Model was announced in April of 2019 as the vehicle.
Tomorrow, part two of this post will explore the shape and implications of the Direct Contacting model, and then will offer some ideas to remedy or mitigate some of the untoward consequences of the Money Machine.
Dr. Gilfillan was the CEO of Trinity Health System from 2013 until 2019. He is a Trustee for United States Pharmacopeia; a Director for the Health Care Transformation Task Force; a member of Advisory Committees for the Institute for Exceptional Care and several Robert Wood Johnson Foundation programs addressing health equity and SDOH (all uncompensated). Dr. Gilfillan also recently consulted for an integrated health system (compensated). Dr. Gilfillan was the Director of CMMI during the roll-out of several ACO Models and was involved in the development of CMS ACO regulations. He has also been a leader and member of teams that managed multiple ACOs and Medicare Advantage plans.
Dr. Berwick served as Administrator of the Centers for Medicare and Medicaid Services from July, 2010, to December, 2011, during which he oversaw the issuing of the initial CMS regulations for Accountable Care Organizations, as well as numerous other regulations devolving from the Affordable Care Act. He serves on the boards of LumiraDx (stock option compensation); Virta Health (stock option compensation); NRC Health (stipend and stock option compensation); Institute for Exceptional Care (uncompensated); CareVisor (stipend) Partners in Health (uncompensated); Results for Development (uncompensated). He also serves on Advisory Boards for the National Institute for Health Care Management Foundation, Datavant, and the Institute for Accountable Care, and on the American Medical Association Journal Oversight Committee Dr. Berwick occupies multiple committee positions with the National Academies of Sciences, Engineering, and Medicine.
Series supporters. See our Considering Health Spending hub.
Can’t sign in? Forgot your password?