19 Leaders Support CMS Medicare Advantage Proposed Payment Changes using data from United Health Group’s Study

Brief Explanation of MA Advance Notice Comment Letter from 19 Industry Leaders 

In our comment letter [below] on the 2024 MA Advance Notice we offered strong support for CMS to finalize their proposed approach to changes in the MA Risk Adjustment system. The letter included an extensive analysis of a two- sided full risk contract study by UHG/Optum Authors that we believe illustrates the urgent need to improve how Medicare pays MA Plans. A briefer summary of our analysis of that study follows.

Our Conclusions:

1. This study demonstrates that MA plans and practices operating under two-sided full risk contracts have powerful incentives to increase coding that results in higher risk scores, premiums and profits sufficiently to increase CMS payments by at least 34% more than they would be if based on the actual FFS spending.
2. Health Plans purchase provider practices to capture both the insurance and provider profits from these contracts ultimately spending much less than 85% of premium on actual healthcare services for Medicare beneficiaries
3. Any comparison of MA and FFS hospital use through analysis of claims is invalid because MA plans deny and downgrade up to 30% of inpatient stays rendering the data incomparable .

Background:

United Health Group (UHG) authors published a study of the experience of two comparable sets of Medicare beneficiaries, one with traditional FFS Medicare coverage seen by varied providers, one with MA coverage seen in physician practices operating under “two-sided full risk contracts” with MA Plans. Neither the insurer nor the practices were identified although the MA practices were described as part of a large national health care delivery organization which sounds similar to UHG’s Optum subsidiary but was not identified.

The populations were matched for age, sex and state. The Study included a table describing the matched populations of 158,156 beneficiaries each with the frequency of various Hierarchical Condition Categories (HCC’s) which are used to create the risk scores CMS uses to adjust payments.

Our Analysis:

• These are comparable populations demographically
• The FFS population was somewhat sicker with more cancer and transplants
• The reported rates of reporting of acute serious, unequivocal HCC’s are very similar
• The MA population was coded with almost twice as many HCC’s per person (3 vs. 1.56)
• The differences in HCC reported rates was concentrated in codes that are more discretionary. The MA reported rates were multiples of the FFS rates:
  • 5.7x substance use disorder (10.0% vs. 1.7%)
  • 3.5x the prevalence of psychiatric disorders
  • 2.8x for non-diabetes metabolic disorders(23.6%vs.8.4%)
  • 2.2x the of rate COPD (21.4% vs. 9.5%)
  • 3.6x the rate of Vascular disease – 50.7% vs. 14%

• The resulting risk score was 45% higher in the MA Population – 1.4 vs. .98. This is the minimum increase because we could not model several other factors that would increase the risk significantly
• This produces 34% higher potential CMS payments for the same population

How much extra money would this produce for 2024 under a “two-sided full risk contract” and where does it go? We cannot say with certainty how high the additional payments were given the complexity of the Medicare bidding system. We created a model using the MedPAC description of average 2023 bids to estimate the potential financial flows yielding these estimates:

• CMS would pay Plans $300 Per Member Per Month (PMPM) more
• Practices would get $255 PMPM more
• Plans would get $45 PMPM more to use for profits or improved benefits
• If the Plan/subsidiary owns the practices they get the full $300 ultimately therebyincreasing parent profits markedly
• In insurer owned practices the dollars spent on actual healthcare services will likely bewell below the expected 85% and possible less than 70%.
• For the study population of 158,156 the total dollars would be:

UHG/Optum Study Authors Conclusions:
1. The practice population had lower hospital use that reflected better care and outcomes.

Our Response: The comparison of hospital use is invalid because the MA Plans decrease reported hospital use through denials and downgrades of inpatient stays to observation stays.

2. CMS HCC Risk adjustment system may appropriately align reimbursement with health burden of the population.

Our Response: This is rebutted by the data showing that their doubling of HCC’s results in a 45% higher risk score and 34% higher payment for the same health burden.

3. CMS payments can support the infrastructure necessary to improve care efficiency and quality.

Our Response: MedPAC has documented for years that it is impossible to demonstrate better quality in MA Plans. Estimates of the cost of investments for quality and efficiency are highly varied. We suggest $52 PMPM above already reimbursed administrative care costs would be a generous amount. MA Plans receive on average about $40 PMPM in Quality bonuses. The extra $300 PMPM in payments dwarfs any incremental amount of spending related to improving care.

How do Plans and Practices Increase risk scores?

Here is an example used in a MA industry firm’s on-line brochure. The narrative in the headers is ours, the HCCs and scores for the same person are from the firm’s slides.

March 6, 2023

Honorable Xavier Becerra
Secretary of Health and Human Services Washington, DC

Ms. Chiquita Brooks-LaSure
Administrator, Centers for Medicare and Medicaid Services Washington, DC

Docket Number CMS-2023-0010-0001
Dear Secretary Becerra and Administrator Brooks-Lasure,

The policy changes for Medicare Advantage (MA) payments proposed by CMS in the “Calendar Year 2024 Advance Notice” constitute important advances. These improvements are long overdue and badly needed to assure appropriate financial payments and stewardship for MA Funds, fair payments to enable excellent care for sicker patients, sustainability of the overall Medicare program and security for all beneficiaries. We support CMS’s finalizing these propose MA payment changes.

MedPAC has estimated that in 2023 there will be $27 billion in excessive and unwarranted payments to MA plans. Other have projected that these overpayments will cost taxpayers $600 billion over the next 8 years. Beneficiaries will ultimately directly shoulder approximately 15% of these costs, almost $90 billion in increased Part B premiums. This is a direct transfer of funds from beneficiaries to MA plans.

Summary Comments

On February 1, 2023, CMS released an “Advance Notice of Methodological Changes for Calendar Year (CY) 2024 for Medicare Advantage (MA) Capitation Rates and Part C and Part D Payment Policies”1 that proposes substantial changes to the MA Part C risk adjustment methodology. We support CMS’s approach and urge you to finalize this methodology for 2024.

Our reasons for support include:

1. The HCC Risk Adjustment system has allowed plans to in effect set their own premium by incessantly creating, hunting for and submitting more diagnosis codes to CMS with resulting overpayments that are projected to total more than $600 billion over the next 8 years. (2)
2. Medicare Beneficiaries will pay out of their pocket for about 15% of these overpayments, or more than $90 billion, by way of increased Part B benefits.(3) This represents a direct transfer of wealth from seniors to insurance companies and investors.
3. A recent study by authors from UnitedHealth Group / Optum (discussed further below) demonstrates that MA Plans can generate approximately twice as many HCC’s per person as there are in an equivalent FFS population despite the health burden being somewhat greater in the FFS population.
4. Because the financial value for each HCC is calculated from the FFS Medicare data, which have fewer codes, the submission of more codes from MA Plans results in marked overpayments. In the case of the study by UHG / Optum authors our estimate is that this creates an opportunity for at least a 34% increase in payment from CMS.
5. As demonstrated in the below analysis of the study by UHG/Optum authors, Plans have used percentage of premium contracts to entice providers into helping them drive more coding, higher premiums, and more profits for all by simply arbitraging the difference between FFS and MA coding rates to make their population appear sicker.
6. Insurer owned primary care practices using these contracts give plans the opportunity to collect even more profits masking the reality that the actual medical loss ratio for benficiaries in these practices may run less than 70%, well below the 85% minimum MLR required for Plans.
7. This Risk Adjustment arbitrage has created a gold rush of investment, driven directly by these MA overpayments, that is distorting the delivery of primary care.
8. The MA marketplace is highly concentrated; the top two for-profit MA contracting organizations, UnitedHealth Group and Humana, have 46% market share and the top 10 have 79%. These large plans have become the most adept at-risk coding and as a result are taking an even greater proportion of premium to drive greater market share to the disadvantage of smaller, frequently non-profit, community-based health plans.(4)

We believe CMS’s approach will significantly and properly recalibrate MA payments thereby decreasing these overpayments and hopefully beginning to reestablish investment patterns driven by the objective of improving care not non-care related diagnosis coding. As a result of the reduction in risk-adjustment diagnoses codes, and the elimination of highly abused HCC’s like Vascular Disease and Protein Malnutrition, dollars will be redistributed across other more appropriately utilized diagnosis codes.

CMS estimates that this risk model revision and normalization, along with other changes and the expected continued increase in MA risk scores will result in an average 1.3% payment increase for 2024. The impact will not be even across MA plan organizations and will generally redistribute MA premium away from the largest and most aggressive coders to community-based plans that have traditionally served low-income and minority populations.

America’s Physician Group(APG) has criticized the CMS proposed approach and raised the specter of their members decreasing care or leaving lower income communities. Many APG members have provided outstanding care in their communities for many years and did so well before the onset of risk adjustment and coding games. Many are non-profit firms that have long term commitments to serve lower income populations and no doubt will continue with their missions. Many have also become highly dependent on risk coding to improve care models and drive higher profits.

Others are new provider entities created explicitly to profit investors by creating Medicare Advantage overpayments. The resulting overpayments by CMS make clear that these are not “Value Based Contracts,” they are value consuming contacts that will more rapidly exhaust the Medicare Trust Fund. Many of these providers have already been acquired by the major MA plan organizations including UnitedHealth Group, Humana, Aetna, and Cigna. These leading MA plan organizations have also purchased and/or signed high-profile deals with software vendors whose inflated values are based on their ability to identify “missing” risk adjustment diagnosis codes; others have developed in-house capabilities. As demonstrated by the study we will examine below, there are extraordinary profits being made by aggressive coders. Many of these profits end up leaving the healthcare system as dividends and stock buybacks to benefit investors, not patients.

The proposed CMS changes, while significant, leave many opportunities for aggressive coders. Along with the inflated benchmarks MedPAC has long cited, MA plan organizations will continue to have ample funding that will allow them to continue to provide care and coverage for lower income communities. They just may need to adjust their profit expectations and create more effective care models that produce real savings, not illusionary savings from the use of discretionary diagnosis codes.

It is true that in the MA industry plan organizations are compelled to try and increase their risk scores to be competitive. Some are much more aggressive than others. We know that plans can’t just unilaterally stop. But all plan leaders know it is a broken system that is creating overpayments. For those who claim to be committed to delivering high value care, addressing the Social Determinants of Health and helping the country address other social conditions, now is the time for them to stand up and support this CMS effort to begin to improve the system.

Recommendations: The chart below includes our overall recommendations along with a brief explanation for each.

In the remainder of this comment letter, we present data from a recent article written by UHG/Optum employees that we believe illustrates the above points dramatically. Thank you for this opportunity to comment on CMS’s important initiative to improve the accuracy of payment for Medicare Advantage plans and the sustainability of the Medicare Trust Fund.

Background

On February 1, 2023, CMS released an “Advance Notice of Methodological Changes for Calendar Year (CY) 2024 for Medicare Advantage (MA) Capitation Rates and Part C and Part D Payment Policies” (5) that proposes substantial changes to the MA Part C risk adjustment methodology. Three risk-adjustable HCCs will be eliminated entirely. More than 2,200 diagnosis codes, from the three eliminated HCCs and portions of the remaining HCCs, will no longer impact risk adjustment payments. The remaining HCCs will be split into more-granular HCCs, renumbered, and renamed to create a new list of HCCs. Risk adjustment factors (RAFs) for the new HCCs will then be calculated using 2018 diagnoses and 2019 expenditures.

CMS determined that the 2,200+ diagnoses codes and 3 HCCs “should be reclassified based the relative coding in MA versus in FFS and on clinical input regarding the degree of discretion to code each condition.” CMS discussed the rationale for these changes in its 2021 “Report to Congress: Risk Adjustment in Medicare Advantage”.(6)

In the MA Money Machine Part 1 and MA Money Machine Part 2, Gilfillan and Berwick (among letter signatories below) described how MA plans used two-sided full risk contracts to incent providers to find, gather and submit more diagnosis codes to increase premium and payment. Citing evidence from MedPAC and many others they showed an industry wide increase in coding in MA vs. FFS and the resulting overpayment of MA plans. However there has been no clear quantification of how coding intensity varies at the HCC level or at the provider level that would allow confirmation of the Money Machine Model or the actual magnitude of the profits involved. (7)

Against this backdrop, a remarkably welcome level of coding intensity transparency has been provided in the recent paper “Comparison of Care Quality Metrics in 2-Sided Risk Medicare Advantage vs Fee-for- Service Medicare Programs” by a team of UHG’s Optum employees published December 12, 2022, in

JAMA Health Policy.8 Apparently in response to the Money Machine articles, the authors sought to demonstrate the effectiveness of two-sided full risk arrangements in improving quality, reducing utilization and “aligning Medicare payment with the health burden of the population.” We believe this paper instead proves the opposite: it confirms that MA Plans use provider risk contracts to create overpayments and demonstrates the importance of CMS’s proposed changes.

The UHG/Optum Team Paper (9)

General Approach and Results: The authors’ compare the healthcare utilization of a population of MA patients cared for by “a nationwide delivery system operating under two-sided risk contracts” and a comparable matched population of Fee for Service (FFS) Medicare beneficiaries. (10) The documentation includes the distribution of HCC codes across the populations. The results are breathtaking, not for the assertion of lower MA plan utilization but rather for the illumination of MA coding practices in provider groups under two-sided risk. It appears that while the dynamics reported in the Money Machine articles were accurate, those articles significantly underestimated the ability of incented provider groups to drive extraordinarily high levels of questionable coding. The paper shows that the MA plan and delivery system documented twice as many HCC’s, almost doubled the risk HCC portion of the total population risk score and created at a minimum an opportunity for 34% higher Medicare payments versus what would be justified by the Medicare fee-for-service (FFS) risk score. Furthermore, it demonstrates clearly that the CMS proposed adjustments to the HCC coding system are essential and appropriately target several of the codes that are driving massive overpayments.

Methodology: The UHG/Optum authors describe their approach as intended to evaluate the quality and efficiency of care for MA patients seen in “practices that are part of a nationwide health care delivery organization for which they had complete data.” The MA population studied “included only members in value-based compensation programs in which physicians were at full medical risk.” They created a subset of such 2018-2019 MA beneficiaries and a matched population of FFS beneficiaries from Medicare’s 5% sample population.11 The final 158,156 MA beneficiaries and 158,156 FFS beneficiaries are explicitly matched on:

• 24 months of continuous enrollment
• Non-dual status
• Non-hospice user
• Non-institutionalized status
• Non-ESRD status
• Age 65+
• State
• Age
• Sex

Exhibit 1 in the Appendix illustrates the characteristics of the resulting populations.

The authors then report the baseline characteristics of the population including individualized HCC counts summarized using 31 categories rather than the 80+ HCCs in version 24 of the CMS-HCC risk adjustment model. This data from the article is recreated in Exhibit 2 along with the ratios of the prevalence rates for different HCC groups in the two populations. The ratios, and hence the prevalence rates of the most severe acute HCCs, are generally similar (marked in the Exhibit with one star). These conditions include heart attacks, strokes, intestinal obstructions and perforations, and leukemia. Similarity is expected based on the matching process. But it appears that the FFS population is somewhat sicker as the MA population has 14% fewer cancers and 37% fewer transplants (marked with two stars).

But the similarities stop there. Overall, the MA risk group patients were coded with almost twice (1.9 times) as many HCC’s per person versus the FFS population, 3.00 vs. 1.56. The MA prevalence for HCC groups that are less severe and provide opportunities for more documentation are stunningly high (all marked with 3 stars). For example, the MA population is coded as having 5.7x the prevalence of substance abuse disorder (10.0% vs. 1.7%), 3.5x the prevalence of psychiatric disorders, 2.8x for non- diabetes metabolic disorders (23.6% vs. 8.4%), and 2.2x the of rate COPD (21.4% vs. 9.5%). Vascular disease was coded 3.6x more often, with fully one half of the entire MA population, 50.7% so coded as having vascular disease vs. 14% in FFS. It is well-known that plans maximize the vascular disease HCC by sending staff into MA patients homes with digital diagnostic devices to try and find the slightest hint of sclerosis with little or no clinical relevance. (12) Many of these diagnoses and HCC’s are the very ones that CMS has addressed in their 2024 proposed changes to the risk adjustment system.

Our Estimate of the Value of Aggressive Coding: The paper shows an MA population that has a lower burden of illness but is coded to make them look much sicker with presumably a higher risk score. To quantify how much higher the risk score was over the typical Medicare FFS population, we created a model, explained in Exhibit 3, using the study’s population demographics and MA-to-FFS ratios by HCC category.

A plan’s risk score includes two pieces, a demographic piece based on age and sex and an HCC piece using submitted diagnosis codes. Per our model, the demographic scores (.44) for the two populations were the same, as expected for demographically matched populations. Our model shows that, at a minimum, the MA population HCC portion of the risk score was almost double the FFS score (0.96 vs 0.54).

The total MA risk score was 1.403 for MA and .98 for TM population or a minimum of 44% higher. This result likely understates the MA score significantly because the Study used HCC groups rather than individual HCC’s. Our model used the national FFS distribution of HCCs within groups for both populations, even though we know that MA plans typically submit the more complicated and costly HCCs within groups. We also cannot use CMS’s additional coefficients for people with multiple HCCs because we don’t have individual level information. These two additions would likely add significantly to the risk scores and explains why we talk about our results as being a “minimum increases.”

UHG/Optum Author Conclusions: The authors draw three conclusions:

1. The MA model of care was “associated with improved health outcomes and care efficiency” asdemonstrated by fewer MA inpatient hospitalizations, fewer readmissions and fewer ER visits.
2. The “Medicare Advantage risk adjustment model may be meeting its intended goal by aligning the capitation payments to the health care burden of both the individual beneficiary” and the “aggregate population served“.
3. The Risk Adjustment system “may allow revenue to be deployed to develop the infrastructure that improves the quality and efficiency of care” for MA members.

Our Response to the Authors’ Conclusions: We believe that each of the conclusions is unfounded, that the paper validates the Money Machine model and that it provides a compelling imperative to reform the HCC risk adjustment model.

1. MA is associated with improved quality and efficiency as demonstrated by reduced utilization of services.

This conclusion is wholly unsupported by the data. The authors report that in the study period the MA population had 18% lower inpatient utilization and 11% lower emergency room use. They also cite another study from Aetna comparing MA and FFS that found lower inpatient cost and utilization in MA. We have previously pointed out that such studies by MA firms are not reliable because MA Plans deny or downgrade to observation status 30% or more of inpatient stays. (13) Downgrades to observations status are not a reflection of the quality of prior patient care nor of the acute care a patient needs; it is simply changing an artificial classification that allows payers to pay the provider less.

Another highly touted MA sponsored study comparing MA and FFS use of acute care utilization actually demonstrated this very point with that the total count of observation stays plus inpatient stays being higher for the MA population. (14) Most recently Beckman et al in a study of Ambulatory Care Sensitive Conditions (ACSC’s) across 10 million MA beneficiaries showed how MA Plans shift inpatient admissions to observation status. (15) They concluded that MA actually had more ACSC acute events in total and raised the question of whether ER early discharges impacted the quality of care. Claims data from MA firms cannot be used to evaluate the efficiency and quality of care unless it effectively recaptures the care that is never recorded because of their often arbitrary claims payment decisions.

2. The Medicare Advantage risk adjustment system appears to be meeting its intended goal by aligning the capitation payments to the healthcare burden of the individual beneficiary and aggregate population served.

This conclusion is not only unsupported, it is clearly refuted by the data. To make this simpler, we will disregard the higher frequency of cancer and transplants and assume that the MA population has the same health burden as the FFS population. The actual total average cost per person for the entire medicare population for 2024 is projected to be approximately $1,200 Per Beneficiary Per Month (PBPM). (16) For the FFS population in the study risk score of .98 normalizes to 2024 (17) as .85. Using the average of $1,200 Medicare’s FFS expected cost would be $1,026 (PBPM). The MA risk score of 1.40 after normalizing would be 1.22 and with CMS’s 5.9% coding intensity adjustment, the MA population score would be 1.15. The potential MA premium is $1,379 (PMPM), $353 more than the expected cost to Medicare. The Risk adjustment system results in 34% more potential premium for populations of the same risk. Annualized, the system has generated excess potential payments of $4,272 per individual and $670 M across the population of 158,156. Clearly payment is not aligned with health burden. We will explain below what happens to this additional potential premium.

It is worth reinforcing here that the FFS payment level is the current aligned capitation payment for the health burden of the population. The MA population has the same health burden. The addition of more diagnosis codes does not change the health burden. Overpayment result from the process for two reasons:

1. 1. The dollars allowed for each HCC, called the coefficient, is calculated from FFS data where there are many fewer HCC’s because fewer diagnoses are submitted. The result is that every HCC is valued higher than it would be if, as in MA, every diagnosis possible was submitted. For any sizeable population, multiplying the smaller number of HCCs by the larger coefficient delivers the actual Medicare costs for the population.
   2. Every additional HCC the MA Plan submits above the FFS number is valued at the FFS derived coefficient and automatically creates a payment above the FFS amount. It is of course likely that many of the diagnoses that MA plans identify represent less intensive or long resolved conditions.

3. The current risk adjustment system allows revenue to be deployed to develop the infrastructure that improves the quality and efficiency of care for the patients enrolled in Medicare Advantage plans.

Does it require $353 PMPM more to build the infrastructure suggested? MedPac has long reported that it is impossible to analyze differences in the quality of FFS vs. MA suggesting that not all that much is going into improving quality. Studies have shown that PCP practices spend on average $40,000 per physician per year on activities related to quality documentation. (18) EHR systems many of which include care management and quality improvement capabilities are estimated to cost $1,200 per user per year. (19) Spread across an average physician panel of 2,000 patients they would add about $20 PMPM in costs. The costs of care coordination have varied widely but spread across an average medicare population at a high end might add about $30 for a total of $52 PMPM. MedPAC says that the Quality Bonuses paid to plans would add about $36 PMPM. The extra $353 above the FFS projected spending dwarfs these expenses. Where does it go?

The Real Impact of Two-sided Full Risk Contracts: The one clear conclusion from the UHG/Optum paper is that two-sided full contracts successfully incent providers to generate more diagnosis codes. The UHG/Optum paper reports 1.9x more HCCs for MA beneficiaries in a two-sided risk arrangement relative to matched FFS beneficiaries with the same health burden. The resulting average risk score is at least 44% higher. These differentials between equivalent MA and FFS populations are far higher than the differentials that have been reported by MedPAC and others for the total MA program where most providers are not at two-sided full risk. (20) The ultimate size of the extra payments will be decided through the MA Bid process. The ultimate destination of the payments is a function of ownership of the provider practices.

A simplified Bid Example: The FFS population in the study had risk score of .85 and an expected FFS cost of $1.026 PMPM. Per MedPAC the bids to provide Part A and Part B services average 85% of FFS cost. So a plan using this expected FFS cost will bid .85 times $1,026 or about $872. This would include about 15% for Administrative and Profits or $130 PMPM leading the remainder in the bid, $742, as the expected medical cost for the population. Most bids will actually become the base A&B premium that the plan is paid.

But now, working in concert with provider group under the two-sided risk model, the Plan has documented a higher risk score and presumably higher expected cost of $1,379 as described above. At 85% of the projected cost, the bid can now increase to $1,173, a 34% increase above the $872 bid. But the population is the same, the actual expected medical costs are the same. The only difference is, as described below, the plan will pay most of the extra dollars to providers as a surplus. The way the bid model works, the Plan’s increased bid will actually improve their competitive position because the risk score increase also provides more rebate revenue. To keep the illustration simple, we will only look at the A&B portion of the Bid. Suffice it to say that the higher risk score will also increase the rebate which will allow the plan to make more profits and offer some additional benefits.

Independent Practices: The two-sided full risk “value-based contract” referenced in the paper are most often a percentage of premium contract between the payer and the provider. Under these contracts, the Plan establishes a medical expense target based on a percentage of the premium (inclusive of risk adjustment) the Plan receives from CMS for a practice’s population. The target is typically about 85% of premium. If the total costs are less than the target, the provider keeps the excess. If the costs are higher than the target, the provider pays the plan the difference.

We use the first FFS bid example above to demonstrate the financial flows under a non-percentage of premium Value Based Contract that uses a historically based medical expense target to incent providers to improve care and reduce costs. Simply put, the medical expense target for the VBC is set at $742 based on the history. If the provider can decrease the total cost by another 5%, they will get a gain of $37 PMPM.

Now the practice contract changes to an 85% of premium contract. The practice is at full risk for total medical costs. Any spending below the target becomes their gain. Any spending above the target will be their cost. But the reality is there is no risk of the spending being above the target. Because the practice increased the risk score to 1.4 the bid/premium increased to $1,173 and the 85% medical expense target becomes $997. The expected medical expense of $742 is unchanged because as shown in the Study, the population has the same severity of illness they just are coded more. Without any improvement in medical costs, the surplus for the practice is now $255 PMPM. In addition, this leaves 15% of the higher premium, or $176 PMPM for the Plan, an increase of $45 PMPM. Because the administrative expenses are minimally changed, the Plan can use these dollars for additional profits or some improvement in benefits or premium. Through the bidding process and the two-sided full risk contract the provider and the plan share $300 of the potential $353 additional revenue generated by the increased risk score.

Insurer Owned Practices: These mutual wins can be combined. If the plan parent corporation acquires provider practices, then the parent collects both the practice gain and the additional 15% administration and profit load. The contract works the same way with a percentage of premium contract between the insurance subsidiary and the practice subsidiary. The dollar flows work as described for the independent practice. The difference is that the practice profits become consolidated into the parents’ profits. One added advantage for the parent is that the insurance subsidiary has an 85% MLR requirement. The contract with the practice assures that the insurance side will meet this. However, when the profits of the practice are consolidated the reality of the overall MLR is obscured. Even if we assume that $50 PMPM are used to gather codes and pay their employed physicians more, the total medical expenses are still about $792 but the total premium is $1,173 for an MLR of 68%.

MA organizations have figured this out. Insurer owned practices are becoming the norm as evidenced by Humana’s creation of CenterWell and Aetna’s $10.8 B acquisition of Oak Street Health. UHG, however, is far ahead because they have been building Optum Health as a capitated medical group to these ends for the past 10 years. One executive explained in a recent earnings call when asked about practice acquisitions and moving their medical groups to capitation “what you are really seeing is the result of almost 10 years of building a flywheel that now has significant momentum.” (21)

Our Conclusions from the UHG/Optum Team Paper: In contrast to the authors’ conclusions, our conclusions are that:

1. It is impossible to use claims-based utilization measures to demonstrate lower hospital use in MA populations because of high rates of denials and downgrades to observation status.
2. The combination of premium based two sided risk contracts with the HCC adjustment model creates a Money Machine allowing providers and plans alike to dramatically increase profits for themselves and costs for taxpayers, Medicare beneficiaries and CMS.
3. Most increased payments under two-sided risk arrangements go to provider and the plan, not better benefits.
4. Insurers can markedly improve their profitability by owning primary care practices operating under two-sided risk contracts where premium is dependent on risk scores obscuring the reality that they may be spending only 60-70% of CMS premium on healthcare costs.

With this study, the UHG/Optum authors have documented the power of two sided full risk contracts to increase coding, CMS overpayments and corporate profits. In short, they have confirmed the reality of the Money Machine model. (We have not assumed that the “national health system operating under a two-side risk contract” described in this report is Optum Health nor that the insurer is UHG.) It is clear to us that CMS must take action to limit the impact of discretionary diagnosis codes.

The UHG/Optum authors’ paper is an important start on bringing transparency to the world of MA risk contracting. We believe it would be helpful for the authors to provide more clarity on the types of contracts, the ownership of the practices and the sources of coverage. Given the concerns raised by America’s Physician Groups we believe they should join their Optum colleagues and be transparent about their contracts, their risk scores, and the degree to which their overall results are a function of risk coding rather than improved medical care.

Footnotes: