Medicaid best price reforms to enable innovative payment models for cell and gene therapies

ABSTRACT

Background

Cell and gene therapies promise durable benefits but face financial challenges from the uncertainty in their performance. Value-based purchasing arrangements (VBPAs) can address uncertainty but have been inhibited in the US by the Medicaid Drug Rebate Program (MDRP) approach to determining Medicaid Best Price (MBP) rebates. The likely effectiveness of MDRP reform proposals enabling VBPAs is examined in this study for durable cell and gene therapies.

Methods

Monte Carlo simulations examined three potential reforms (Multiple Best Prices, Bundled Sales, and National Pooling) to determine the impact on payment misalignment, the required payer size to participate, and the percentage of total lives covered by a VBPA.

Results

Simulation results suggest that 11% to 54% of commercial US lives would be feasibly covered by a VBPA depending on reform type and condition size. MPB reform achieved the highest commercial contracted percentage and lowest misalignment for commercial payers compared to National Pooling and Bundled Sales. State Medicaid plan results suggest lower extreme misalignment across all successfully contracted instances than commercial payers.

Conclusions

The Multiple Best Prices will likely enable VBPAs for many durable cell and gene therapies and larger payers. Further reforms may be needed to extend VBPAs to ultra-orphan conditions.

KEYWORDS:

• Cell and gene therapies
• Gene Therapies
• health economics
• health policy
• Medicaid best price
• Medicaid drug rebate program
• outcomes-based contracts
• patient access
• payment innovation
• value-based purchasing agreement

1. Introduction

1.1. The historical Medicaid best price challenge

Cell and gene therapy developers offer the promise of durable benefits, even cures. These therapies can span products that (i) use or modify DNA or RNA; (ii) are administered in a single course; and (iii) are expected to deliver at least 18-months average benefit [1]. The FDA Center for Biologics Evaluation and Research (CBER) provides an instructive definition for gene therapy products and Chimeric antigen receptor (CAR) T cell products [2,3].

Despite the potential transformative benefit of cell and gene therapies, patients may face barriers in accessing these treatments due to their uncertainties and financing challenges [4]. Healthcare needs payment innovation to parallel the therapeutic scientific innovation to overcome these challenges and ensure patients can access necessary treatments [5–8]. Value-based purchasing arrangements (VBPAs: outcomes-based contracts where payment is tied to reaching pre-determined goals – sometimes clinical, but not necessarily – and reducing payer risks [9]) can address uncertainty but face multiple implementation challenges, from outcomes measurement to regulatory barriers in the US and worldwide [8,10–15].

Medicaid Best Price rules actually set rebates, not prices, that a developer must pay to a state Medicaid plan for each unit of the therapy used by its Medicaid beneficiaries. That rebate, in most cases, is the larger of either (i) the statutory rebate: 23.1% of Average Manufacturer Price (AMP; the volume-weighted average price for all unit sales in a given quarter; also referred as ‘the statutory rebate’), or (ii) the ‘best price’ rebate: the difference between AMP and the lowest price paid by any commercial buyer [16,17].

The Medicaid Drug Rebate Program’s (MDRP) best price reporting rules have been one of the primary impediments to commercial payers using innovative payment models because of (unintended) disincentives they create for therapy developers [15,18,19]. Commonly referred to as Medicaid Best Price (MBP), these rules have historically:

• Extended any outcomes-based rebates received by a single commercial payer with only a single-treated patient who also experiences a poor therapeutic outcome, to all Medicaid patients (regardless of their individual outcomes), i.e. an ostensibly ‘one-off’ rebate could be applied to all Medicaid patients receiving the same product.

• Limited the time horizon for making performance-based payments to 3 years because of reporting mechanisms. This prevents therapy developers from offering long-term outcomes rebates to commercial payers who desire it.

• Interpreted an initial payment as the only payment, thereby creating inappropriately high immediate rebates, i.e. a multi-payment contract could have the first installment interpreted to be the entire price for calculating the rebate. Each remaining payment would result in a reduced rebate but would require a clawback from each state Medicaid program – a process not well supported in the MDRP systems.

In December 2020 CMS finalized a new rule to address these and other issues for drug Value-Based Purchasing arrangements which took effect on 1 July 2022 [20]. Additional reforms to the MDRP also have been proposed [16].

EU and Asia share some of these challenges within their regulatory frameworks. For instance, timeframes can be limited by regulation, constitution or supra-national (EU) treaty [21,22]. While the discussion below focuses on U.S.-specific mechanisms, the issues often apply more broadly with the statistical volatility of value-based contracting for rare conditions applying universally.

1.2. An MDRP Medicaid best price illustration

Figure 1 illustrates a key challenge for value-based purchasing arrangements: a single high rebate, for a single commercially covered patient, would set the best price for all patients in Medicaid.

Figure 1. Medicaid Best Price (MBP) rebate example for a $1 million list price durable therapy with a value-based purchasing agreement reported by the drug developer.

Figure 1 illustrates the MBP rebate mechanics for a durable therapy under a VBPA with two possible contract outcomes. The highest contract outcome – the green circles – is set at the $1 million list price (WAC: Wholesale Acquisition Cost) for adequate response and a $100 K net price for patients for whom the therapy does not achieve the contract outcome – the red circle. The illustration further assumes that three payers participate in this program and in the quarter illustrated had 5, 8, and 2 total treated patients, respectively. Of these, only one patient, for Payer 3, experienced the poor outcome – seen in the red circle.

Under MBP, the drug developer would report the lowest price realized in the quarter, in this case $100 K for Payer 3: the payment for failure, shown by the single red circle. The developer would also report an AMP of $940 K – i.e. the average price where payment was made in full for 14 patients was $1 m and the price when one payment was not made full was $100 K. Medicaid would then receive the larger rebate of: the statutory rebate (23.1% of AMP) or the ‘best price’ rebate (AMP minus lowest price reported by developer). In this example, the statutory rebate would be $217 K and the ‘best price’ rebate would be $840 K ($940 K – $100 K). These situations in which the actual rebate amount exceeds the statutory rate times AMP is called ‘breaking Best Price.’ Whereas the $100 K price was realized in only one commercial contract for one patient, the $840 K MBP rebate would apply to all Medicaid patients (a distinct contract is not needed). If Medicaid pays the list price with no other discounts or negotiated supplemental rebates, Medicaid’s net price would be $160 K ($1,000 K – $840 K) for all beneficiaries.

This simplified example does not include any pharmacy dispensing fees nor the potential additional MBP inflation rebate. The MDRP program applies only to drug costs and does not include, nor require provider rebates for, any medical costs associated with the therapy.

In summary:

• The MBP process uses commercial market sales prices to set a Medicaid program rebate, not to set a Medicaid price.

• The MBP rebate then applies to ALL units of a therapy used by Medicaid beneficiaries.

• The primary MBP rebate is either the statutory rate applied to AMP or the difference between AMP and the lowest price paid by any single commercial buyer – whichever is greater. Note that an additional rebate exists should prices increase faster than inflation; this is not considered here.

• Two historic reasons inhibiting uptake of VBPAs are the¹⁵:

  • Risk of setting a high MBP rebate due to a deep discount for a single patient receiving low therapeutic benefit in a small commercial plan; and

  • Resulting quarter to quarter volatility of the MBP rebate;

1.3. The current landscape of durable therapies and precision financing models

To date, 10 cell or gene therapies with durable effect – defined here as treatment effect lasting at least 18 months after a single administration – have received FDA approval [23]. Recent projections suggest 55 to 75 durable therapies will gain approval by 2031 [1,2]. These one-time treatments hold the promise of avoiding future costs of disease progression and alternative therapies [24].

A drug with a single administration but durable effect creates financing challenges under traditional payment models. First, payment does not align with benefits accrual: the payer will be required to pay the entirety of the price upon administration, even though clinical, economic, and other benefits may be years or even decades in the future. Second, the benefits carry performance uncertainties: the product may not meet efficacy or durability expectations for a particular patient or population. Traditional up-front payment approaches possess limited mechanisms for risk-sharing with the drug developer. Third, there are actuarial estimate risks regarding the quarterly number of eligible patients for the usually rare conditions addressed by gene therapies. Especially for small payers, a few such patients in an unlucky quarter can create significant income statement volatility and generate solvency concerns.

Innovative value-based payment approaches have been proposed to address these challenges, and used in other contexts [25–27]:

1. Milestone-based contracts: a type of performance-based contract in which a pharmaceutical company guarantees to refund the cost of therapy (partially or fully) to the payer if an agreed outcome is not achieved.

2. Warranties: a pharmaceutical company purchases a patient-specific warranty policy that reimburses treatment-related costs for suboptimal performance to payers over an agreed time period. The value is related to covered healthcare costs and is not a refund for the cost of the treatment.

3. Performance-based annuities: a type of performance-based contract in which payments for a cell or gene therapy are spread over multiple years and linked to therapy performance. If a therapy fails to deliver an agreed outcome, no further payments are made.

4. Subscription model: a pharmaceutical company provides treatment for a set fee regardless of the number of patients treated or a set price per patient [28].

This paper describes the extent to which the 2020 CMS Rule reforms, and additional proposed MBP reforms, address legacy MBP challenges for payment innovation. We specifically seek to understand whether reform approaches may encourage VBPA adoption, align payments with therapeutic benefits, and allow developers to assume more financial risk while simultaneously holding Medicaid harmless, financially and operationally.

We first inventory the reform approaches and qualitatively assess their ability to resolve the MBP challenges for durable cell and gene therapies. We then model the impact of the reforms on drug developers, commercial payers and state Medicaid programs via Monte Carlo simulations. We conclude with suggestions for further research and for additional policy actions the simulation results suggest would improve Value-Based Purchasing feasibility and fairness.

EU and Asia share some of these challenges within their regulatory frameworks. For instance, timeframes can be limited by regulation, constitution or supra-national (EU) treaty [21,22]. While this paper focuses on US specific mechanisms, the issues often apply more broadly with the statistical volatility of value-based contracting for rare conditions applying universally.

2. Methods

We conducted a targeted literature review using PubMed, Google Scholar, and Google to identify peer-reviewed, published literature and gray literature. Search topics were (i) terms for value-based, performance-based, or outcomes-based risk-sharing and contracting, and (ii) terms for CAR-T and gene therapies. The search was conducted in 2020 and restricted to the previous 5 years and English-language results. We developed a framework of MBP regulation issues and reform challenges from qualitative analysis of the identified literature. This framework was used to construct a discussion guide for semi-structured interviews with experts from: Medicaid and Managed Medicaid Organizations (2 experts); developers (4); commercial payers (4); academia, patient advocacy, or policy organizations (4); and law firms (1).

The interviews were used to update and refine the findings from the literature and to develop an inventory of financing innovations, the MBP challenges they face and the proposed MDRP reforms. The findings from the interviews are expanded on in the supplementary appendix. We then qualitatively assessed the degree to which the reforms might mitigate the challenges.

We also developed four linked, Excel-based Monte Carlo models to simulate: VBPA outcomes (payment misalignment, volatility, use of VBPAs, and the number of patients); for select MDRP reforms (Multiple Best Prices, Bundled Sales, and National Pooling); across key VBPA terms (contract performance, rebates, treated patient population size); and structural parameters (insured pool size [29], tolerable risks of triggering Medicaid Best Price, and minimum required numbers of treated patients for payers and developers to contract).

We defined the VBPA outcomes as follows:

• Payment misalignment – what a payer would pay under each proposed reform vs. payments tied perfectly to treatment response.

• Volatility – the quarter-by-quarter variation in MBP for a contract;

• Entry threshold – the maximum number of plans or payers that would enter a VBPA contract

• Number of lives covered – the largest possible number of insured lives covered based on the use of VBPAs.

We developed 144 unique scenarios (3*3*2*4*2 = 144) based on three rebate levels (23.1% upon inadequate performance; 100%; and a 2-tier structure meant to more closely mirror the real world both with multiple tiers of rebates tied to patient performance and one in which larger payers are able to negotiate larger rebates – i.e. a 35% or 50% rebate for partial response and 66% or 100% for fully inadequate response for medium and large plans respectively), three treated population sizes (200 patients; 1,000; 10,000), two therapy performance levels (90% adequate response which is similar to AAV gene therapies; 60% adequate response which is similar to CAR-Ts), four MDRP reform approaches (Bundles Sales; National Pooling; Multiple Best Prices with Select Participation; Multiple Best Prices with Total Participation), and two AMP approaches (current AMP rules; AMP reform).

For each scenario, a 10,000 iteration Monte Carlo simulation was run in each of three steps. Results were simulated for individual commercial plans/payers and Medicaid state (including Managed Medicaid Organization) plans. In step one, we simulated impacts on developer decisions regarding which plans or payers to whom to offer VBPAs based on: (1) the developer’s risk of breaking Best Price and (2) at least a 25% chance of encountering a treated patient each quarter (i.e. 1 patient per year) in a VBPA to justify the transaction costs of negotiation and administration. In step two, for each of the participating plans from step one, we simulated payment misalignment in the commercial market as well as AMP and the resulting MBP rebate. Finally, in step three, we used that AMP and MBP rebate to determine payment misalignment in state Medicaid plans. Detailed specification and parameterization of all of the models are available in Supplementary Appendix Table A-3 to Table A-6.

In the Medicaid simulation, we examined two scenarios for implementing the Multiple Best Prices approach under the new 2020 CMS VBPA rule – Select Participation and Complete Participation. Under Select Participation, states can choose between participating in a Multiple Best Prices VBPA or receiving the standard MBP rebate based on non-VBPA reported prices. For Select Participation, we assumed that only the nine current states with an explicit VBPA state plan amendment (SPA) agreement with demonstrated infrastructure to implement a VBPA would participate in the Multiple Best Prices approach. All other states accept the standard MBP rebate. This is likely to be a conservative or worst case. Under Complete Participation, all states which exceed the 25% chance of encountering a treated patient in a quarter participate in the Multiple Best Prices VBPA arrangement, with none receiving the traditional MBP rebate. This is a maximum case to bound the potential VBPA performance.

The simulations allow estimates of MBP rebate volatility (e.g. small population problem), the impact on VBPA availability, and the degree of payment misalignment across reform approaches.

3. Study results

3.1. Qualitative assessment of legacy MBP challenges and MBP reform approaches

A targeted literature review and qualitative analysis produced 54 unique concepts across seven themes (see Supplementary Appendix Table A-1). Further qualitative analysis of the literature reviews and the interviews condensed the themes into five issues impeding payment innovation, two primary impact mechanisms (volatility and magnitude), and six alternative proposed MBP reforms. The research team then judged whether each of the six reform proposals addressed each of the five impediments and also whether the proposed reform also could maintain Medicaid rebates at current levels. The resulting summary of the qualitative assessment is shown in Supplementary Appendix Table A-1 and shows the research team’s judgment, informed by the literature and interviews, regarding the effectiveness of a reform approach to address an issue.

Complying with legacy MBP reporting rules inhibits, and may prohibit, implementing value-based precision financing solutions through five specific challenges as follows.

3.1.1. Single-patient problem

Single-patient problem which is that a large discount for a single patient in a commercial, Medicaid MCO, or Medicare Advantage plan may trigger that discount being mandated for all treatments within Medicaid for that product. This problem occurs irrespective of the number of patients in a plan (see Figure A-1 in the Supplementary Appendix). The impact is that VBPAs with significant risk-sharing are unlikely to be developed.

3.1.2. Small-population problem

Small-population problem is related to the single-patient problem but is also related to the number of covered lives in a payer plan. Here, a large discount for a single or a small number of patients with a rare condition in a commercial, Medicaid MCO or Medicare Advantage plan may trigger that discount being mandated for all treatments within Medicaid for that product. The impact is that Precision financing solutions would not be available to patients with rare diseases.

3.1.3. 3-year horizon problem

Three-year horizon problem is caused by the MDRP 3-year limit on back-adjusting rebates and payments, which precludes any performance-based agreements longer than that. Since performance uncertainty for durable therapies almost certainly extends beyond 3 years, it may not be adequately mitigated by contracts limited to this 3-year period.

3.1.4. First-payment problem

First-payment problem occurs when a first installment payment to a commercial (and possibly a Medicaid MCO or Medicare Advantage plan) could be considered the entire price for the therapy, potentially also triggering the ‘best price’ and making just the first installment the de facto price for all treatments within Medicaid for that product. This has prevented multi-period annuity payment innovations from being offered directly by drug developers.

3.1.5. Unit-price problem

Unit-price problem results from the requirement that fixed-fee payments for drug access (such as in subscription models) be reported to the MDRP as a price paid per dose sold in each quarter. Fluctuating volumes, particularly with a high-prevalence disease, will mean a lower price per unit in some quarters, resulting in variable and potentially higher mandatory Medicaid rebates. The impact has been that no drug developer offers subscription style arrangements to commercial payers.

Due to the underlying calculations of AMP and Medicaid Best Price rebates, these challenges impact stakeholders in ways not well elucidated nor quantified.

These challenges impact on stakeholders vary according to the precise financing innovation, the therapy modality or site of care for administration, the condition population size, the payer or plan pool size and each product’s performance (efficacy and durability). These factors all contribute to the performance rebate magnitude and volatility, developer revenues, and Medicaid net costs.

In an attempt to address the challenges posed by the MDRP, CMS finalized a rule change in December 2020 (‘2020 CMS Rule’) to better enable VBPAs [20]. These reforms allow developers to report multiple best prices for a set of performance tiers (Multiple Best Prices) or report a single best price from a weighted average of treated patients under a VBPA (Bundled Sales). These reporting approaches along with additional proposed reforms are explained in more detail in Box 1.

Box 1: MDRP Proposed Reforms to enable drug Value-Based Purchasing

Pre-2020 CMS Rule: The historic Medicaid Best Price approach as illustrated in Figure A-1. 2020 CMS Rule – Multiple Best Prices: Developers report the best price available via contract (e.g. claim is not required to trigger discount) in a quarter in each performance tier across all commercial entity contracts using a particular VBPA arrangement design. Medicaid plans must a priori opt-in to the VBPA arrangement design to obtain the financial terms. However, the developer determines whether to report under Multiple Best Prices or Bundled Sales. 2020 CMS Rule – Bundled Sales: Developers report the lowest average price realized in a quarter in any single contract with a commercial entity using a particular VBPA arrangement design. The developer determines whether to report under Multiple Best Prices or Bundled Sales. Proposed Reform – National Pooling: Developers report the best average price realized in a quarter across all VBPA contracts with commercial entities using any VBPA arrangement design. Medicaid plans do not need to track outcomes because the VBPA reductions are included in the standard rebate calculation. Proposed Reform – Best Deal: State Medicaid plans may participate in the best commercial outcomes-based contract offered by a drug firm [30]. Proposed Reform – Exemption: Developers do not need to report prices into the MBP systems for outcomes-based or other innovative payment arrangements for durable cell and gene therapies [30,31].

Based on the qualitative results and our judgment, we determined that the Best Deal and the Exemption reform approaches are unlikely to enable any financing innovations that benefit both developers and Medicaid. The Best Deal proposed reform likely entails the highest implementation complexity for both stakeholders. Further development of practical mechanics could make this approach more attractive in the future. The Exemption proposal would fully enable private-sector payment innovations but create significant negative financial impacts to Medicaid. From the qualitative results, we selected the Bundled Sales, National Pooling, and Multiple Best Prices MBP reform approaches to quantitatively model.

The simulation results indicate the relative impact of each reform approach on the fraction of covered lives potentially eligible for VBPA contracting and on the payment misalignment for commercial payer and Medicaid VBPAs.

3.2. Simulation – estimated uptake of VBPAs in the commercial market

Figure 2 shows the percentage of commercial lives under a VBPA for each of the MBP reform approaches and model scenarios. White space represents no contracts for VBPAs; solid coloring represents successes in terms of contracts and coverage, with the length of the bars reflecting fraction of the lives covered. Percentages above each cluster of bars represent the average percent of lives included in VBPAs for that cluster (e.g. on average, 11% of possible lives are included in VBPAs across Bundled Sales scenarios for a condition with 200 total US treated lives annually). The vertical axes display all scenarios run as a part of the simulation model with each bar representing a discrete scenario.

Figure 2. Simulation results: impacts of MBP reform approaches on covered lives.

Note that, in Figure 2, data reflect the modeled plans, which only included the largest identified 57 plans, comprising 71% of covered medical lives. The remaining lives are assumed to be covered within plans that generally would not pass the thresholds for VBPA contracted lives.

Simulation results suggest that performance is highly variable. The Multiple Best Prices approach supports the broadest use of commercial VBPAs (32% to 54% of the number of lives potentially covered under a VBPA from 200 to 10,000 annual total US treated lives compared to 19% to 49% for National Pooling and 11% to 28% for Bundled Sales); National Pooling misalignment and contracted number of lives covered are close to those of Multiple Best Prices when there are 1,000 or 10,000 annual total US patients treated (46% to 49%) but supports significantly less VBPA use than Multiple Best Prices when there are 200 annual total US patients treated (ultra-orphan conditions) (19% to 32%). The Bundled Sales approach supports the lowest VBPA utilization across the reforms, but, similar to other VBPA approaches, the utilization of VBPAs increased as the number of treated patients grew (11% to 28% across the scale of patient population size).

Several simulation factors drove VBPA coverage across the scenarios. The rebate level did not impact the number of lives covered under the Multiple Best Prices reform approach, but it did affect the number of lives covered under Bundled Sales and National Pooling reform approaches due to non-contracting to avoid Best Price violations. Drug performance similarly did not affect VBPA availability under the Multiple Best Prices approach. However, for both Bundled Sales and National Pooling VBPAs were more widely available in the 90% drug success rate than in the 60% success rate scenarios. Finally, including administrative services only (ASO) lives (also known as third party administration) in the covered lives pool dramatically expanded the availability of VBPAs under all reform approaches.

3.3. Simulation – commercial payment misalignment

A heatmap of payment misalignment across all scenarios, for select commercial plans, is shown in Figure 3. Each cell is an individual plan’s payment misalignment for a given scenario. The cells in the heatmaps show one of three values: Cells denoted NC-P (Not Contracted – Patient) indicate scenarios in which contracts would not be executed because the 25% threshold for a case in a quarter is not met through a combination of plan size and condition population. Cells denoted NC-BP (Not Contracted – Best Price) indicate scenarios in which contracts would not be executed because there is a greater than 1% risk of breaking Best Price. Finally, the payment misalignment percentage. Colors represent closeness to the perfect 0% misalignment (dark green). The highest misalignment is shown in dark red (Greater than or equal to 21% or less than or equal to −21%) with lower misalignment levels shown in colors per the legend. Perfect alignment at 0% takes a health system’s perspective to matching payment with value of the benefit received by a patient. Payment misalignment greater than or equal to 21% represents overpayments (i.e. health plans paying too much relative to patient health outcomes). Payment misalignment less than or equal to −21% represents underpayments (i.e. health plans paying too little relative to patient health outcomes).

Figure 3. Payment misalignment results across all scenarios, commercial payers.

These are repeated for the main scenarios of interest: reform approach, performance levels, treatable patient population size, rebate level, and whether self-insured or only fully insured lives were included. These are mapped across five plans that span the different covered lives. Subsequent figures show the simulation payment misalignment results across Medicaid states with current AMP rules and with AMP reforms.

The model assumes that plans and developers will only consider entering into VBPAs when in each quarter there is: 1) at least a 25% chance of case occurring and 2) less than a 1% risk of the VBPA exceeding the statutory Best Price rebate. While arbitrary cutoffs they reflect that: a minimum number of expected patients are needed to justify the negotiation, tracking and adjudication of VBPA costs; and developers will guard against a windfall Medicaid rebate due to small population statistical variability breaking Best Price.

Each method possesses the same pattern of NC-P cells. The NC-P rule is applied first and then the NC-BP rule in Figure 3. In each row, any white NC-P cells to the left of black NC-BP cells would also fail the 1% risk threshold. Bundled Sales 5i generates the most NC-BP scenarios, whereas Multiple Best Prices generates none.

The Multiple Best Prices reform approach, which cannot impact Best Price, achieved the most contracted scenarios (114 of 180 possible scenarios, 63%) with the Bundled Sales 5i approach achieving contracts in only 24% of scenarios (43 of 180 cells). The National Pooling reform approach for 5i products met contracting criteria in 53% of displayed scenarios (96 of 180 cells).

When contracted, the Multiple Best Prices reform approach performed the best (possessed the most scenarios with low payment misalignment) with 60% of contracted scenarios having ≤9% misalignment (green cells; 68 of 114 cells). In its few contracted scenarios, Bundled Sales 5i performed less well with only 45% of contracted scenarios (19 of 43 cells) having ≤9% misalignment. National Pooling performed between the two reform approaches with 51% of contracted scenarios with ≤9% misalignment (49 of 96 cells).

The 23.1% discount scenarios performed poorly for the 60% drug performance scenarios because the maximum 23.1% discount was insufficient to fully cover the 40% performance shortfall.

Multiple Best Prices performed the best with larger plans (>500 K covered lives) having near zero payment misalignment when contracted using the 100% and two-tier discount designs. Adding ASO lives to their fully insured population under a single reportable arrangement can enable contracting feasibility (avoids NC-P classification). Smaller plans usually had higher payment misalignment than larger plans and never had lower payment misalignment.

The National Pooling reform performed as well as the Multiple Best Prices reform in scenarios with 90% product performance, excepting four scenarios in which the risk of breaking Best Price exceeded the 1% threshold. However, in scenarios with 60% product performance, National Pooling did not perform as well as Multiple Best Prices in over half of scenarios due to exceeding the 1% risk threshold or generating greater payment misalignment.

Compared to the other two reforms, the Bundled Sales method never generated lower payment misalignment and had the most instances of not contracting due to exceeding the 1% risk threshold. Simulation results did not substantially change when the risk of exceeding the best price increased to 10%.

3.4. Simulation – Medicaid payment misalignment

Figure 4 presents payment misalignment results for ten select state Medicaid agencies ordered by enrollee size. Compared to the commercial plans, Medicaid plans see less extreme misalignment across scenarios (5% [33 of 720 Medicaid cases] versus 25% [63 of 253 commercial contracted cases] dark red cells) but also fewer scenarios with strong alignment (2% [11 of 720 Medicaid cases versus 26% [67 of 253 commercial contracted cases] dark green cells). The lower Medicaid payment misalignment is partially due to the statutory minimum 23.1% rebate regardless of VBPA arrangement performance. Note that state 48 (Washington) has a state plan amendment (SPA) in place; the others do not.

Figure 4. Payment misalignment results across all scenarios, Medicaid payers.

The Total Participation Multiple Best Prices reform scenarios in which all states choose to use VBPAs performed the best as measured by payment misalignment across state Medicaid programs, regardless of the patient or the state population size. However, payment misalignment was dependent on the discount and contract performance. As with the commercial plans, the scenarios with a 23.1% discount and 60% product performance performed poorly because the maximum 23.1% discount was insufficient to cover the 40% performance shortfall.

Multiple Best Prices under Select Participation, in which only the 9 SPA states participate (Select States), has the second-lowest misalignment, outperforming National Pooling and Bundled Sales. The Complete Participation approach, adding the 80% of non-SPA state Medicaid plans (All States) would improve (lower) misalignment for the 60% product performance scenarios but worsen misalignment in the 90% product performance scenarios. In other words, state Medicaid plans would financially perform better, but the overall system’s payment misalignment would be worse (i.e. further away from 0%).

An alternative presentation of results can be found in Supplementary Appendix Figure A-2 which is color-coded from the perspective of a state Medicaid agency in which the cells most beneficial to Medicaid (less than or equal to 0%) are highlighted in dark green.

3.5. AMP impact and potential need for reform

In contrast to the commercial market, Bundled Sales outperforms National Pooling Sales in most scenarios because National Pooling has a secondary effect of lowering AMP. A lower AMP leads to a lower absolute rebate amount, increasing payment misalignment. AMP drops under National Pooling because: more commercial VBPAs exist (fewer NC-BP black scenarios); the VBPAs have lower net prices (especially in the 100% performance discount and 2-Tier performance discount designs); the VBPAs are then included in the AMP calculation (most durable therapies classified are 5i drugs).

Thus, under current AMP rules for 5i products, Medicaid plans directly participating in Multiple Best Prices VBPA arrangements would achieve better payment alignment than those using the Medicaid rebate due to the lower AMP under the National Pooling 5i or the Bundled Sales 5i methods. States not directly participating are also better off with the Multiple Best Prices approach, under current AMP rules for 5i products, because they will receive larger absolute rebates as the Multiple Best Prices approach excludes commercial VBPA arrangements from the AMP calculation.

To better understand the impact of AMP on the National Pooling approach, we repeated the payment misalignment analysis for Medicaid changing the National Pooling AMP calculation to mimic that of retail drugs. With this new calculation, VBPAs are excluded from AMP calculations, thus avoiding the drop that occurs under the 5i AMP calculation. Figure 5 presents results of this AMP reform analysis.

Figure 5. Payment misalignment results with AMP Reform under national pooling approach, Medicaid payers.

Reforming the National Pooling AMP calculation to mimic AMP calculations for retail drugs improves National Pooling misalignment for the 60% product performance scenarios to roughly match the misalignment of Multiple Best Prices with the 9 SPA states. Those two methods already performed similarly without AMP reform for the 90% product performance scenarios. The nine individual states with SPAs perform better than the National Pooling; however, our model assumes risk aversion to best price exceeding the statutory minimum, and the results could improve if developers go above this benchmark.

Alternative presentation of results can be found in Supplementary Appendix Figure A-3 which is color-coded from the perspective of a state Medicaid agency in which the cells most beneficial to Medicaid (less than or equal to 0%) are highlighted in dark green.

4. Discussion

4.1. Summary of findings

Our analysis suggests that of the MBP approaches examined, the Multiple Best Prices method under the 2020 CMS Value- Based Purchasing arrangement (VBPA) rule, with all Medicaid states directly participating, generates the best payment misalignment outcomes of the MBP reform approaches modeled. The Multiple Best Prices approach avoided rebate volatility that was assumed to prevent developers from offering a VBPA to any plan whose statistical risk of exceeding the statutory rebate exceeded 1% in a quarter. Even so, the Multiple Best Prices approach does not work well in all scenarios. Designs limiting the deepest discount to the statutory rebate (23.1% for adult products) had large payment misalignments in the 60% product performance scenarios in which the 40% nonperformance rate exceeded the statutory discount level. Transaction costs further limited the Multiple Best Prices approach for small state Medicaid plans and for smaller commercial plans for conditions with 10,000 drug-treated cases per year, even with the generous assumption that only a 25% chance of encountering a solitary case in a quarter was sufficient to justify the negotiation, outcomes tracking and administration costs for a VBPA arrangement. Creative single contracting to include ASO managed lives in addition to fully insured members to reach at least 500,000 contracted lives expanded VBPA utilization among mid-sized payers.

We found that the Bundled Sales approach suffers the largest number of ‘no contract’ failures due to continued volatility ‘breaking Best Price’ and never achieves the better (closer to 0%) payment misalignment categories for commercial payers. However, for Medicaid payers the Bundled Sales approach induces no transaction costs from negotiation, outcomes tracking or administrative changes, and achieves approximately the same or better payment misalignments in most scenarios.

We saw mixed results for the National Pooling approach which had similar transaction costs benefits and fewer ‘no contract’ scenarios compared to the Bundled Sales approach but suffered from the ‘breaking Best Price’ risk in the 200 treated patient scenarios that the Multiple Best Prices approach mostly avoided. Through secondary impacts on AMP, National Pooling might also result in lower absolute rebate amounts than the Multiple Best Prices approach. National pooling provides the administrative efficiency of the Bundled Sales 5i approach and nears the payment misalignment performance of Multiple Best Prices approach. Especially, with adjustments to the AMP methodology, National Pooling might provide the best balance of VBPA benefits, significant commercial payer participation and state Medicaid accessibility.

4.2. Limitations

The four interlinked models to simulate VBPA contracting required key assumptions regarding developers being willing to accept a 1% risk of exceeding MBP statutory discounts in a quarter; and that a 25% chance of encountering a treated patient in a quarter creates sufficient benefits to overcome VBPA transaction costs. We limited the VBPA designs to two binary discount designs (23.1% or 100% discount for lower product performance) and one 2-tier discount design. More tiers, different assumed performance levels and associated discounts would affect the results. For simplicity, we modeled a total period VBPA design. Durable therapy VBPA designs with differing performance levels and discounts over multiple periods can always be averaged to such a summary design, but the statistical variability would differ and is not accounted for in these results. Payment misalignment which we employ is but one of many assessment metrics that could be used for VBPAs. Our work considers durable therapies which are only a small portion of novel therapeutics. We encourage others to further this work by examining additional designs, treated patient population sizes, assessment metrics, and therapeutic modalities, as well as to compare future real-world VBPA experience with these results to the extent possible given publicly disclosed VBPA data.

5. Conclusion

Value-Based Purchasing arrangements provide a tool to provide appropriate patient access to potentially transformative durable therapies by mitigating the financial risk they pose to payers by sharing that risk with the drug developers. The Medicaid Drug Rebate Program was not designed to facilitate these arrangements. The new CMS rule significantly addresses the core impediments posed by prior rules. Additional targeted refinements to pooling and AMP rules as well as modest investment in administrative capability building will further enable payment innovations to evolve from expectation driven pricing to value-based pricing driven by evidence of actual patient value received.

6. Expert opinion

Multiple Best Prices as described in the new CMS rule is the best reform approach of those examined – as measured by payment misalignment and the projected number of lives covered by commercial VBPA agreements. However, our qualitative analysis suggests that this approach faces challenges of operational costs. Bundled Sales, the other currently available option, is significantly limited due to population size and volatility effects, as well as unintended consequences through the AMP calculation. Targeted policy action can mitigate or even eliminate these durable cell and gene therapy challenges to VBPA arrangements by:

• Enabling National Pooling by (1) reporting best price by aggregating all VBPAs in each quarter (2) the exclusion of VBPAs from AMP calculations in a manner similar to the Multiple Best Prices approach. The National Pooling approach would eliminate State Medicaid administrative cost impediments while preserving nearly all VBPA benefits.

• Increasing outcomes tracking capacity through CMS capability building grants to the states or by directly building national real-world evidence tracking capabilities through CMS cooperation with other HHS agency efforts such as the FDA Sentinel system and RWE programs, NIH Natural History Studies and the Patient-Centered Outcomes Research Institute (PCORI).

• Amending MDRP to annual reporting for ultra-orphan conditions (<1,000 therapeutic treated patients per year) or otherwise aggregate reporting to reduce VBPA rebate volatility.

Declaration of Interest

C. Quinn: Reports financial support from a National Pharmaceutical Council research grant in support of this work. M. Ciarametaro: Michael Ciarametaro was an employee of the National Pharmaceutical Council at the time of authorship. Brian Sils: Brian Sils was an employee of the National Pharmaceutical Council at the time of authorship. S. Phares: Sharon Phares is an employee of the National Pharmaceutical Council. M. Trusheim: Reports financial support from a National Pharmaceutical Council research grant in support of this work, and other financial interests including personal fees outside the scope of this work from: Co-Bio Consulting LLC, Merck & Co. and Novartis.

The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

C. Quinn Wrote Manuscript, Designed Research, Performed Research, Analyzed Data, and Contributed New Analytical Tools. M. Ciaramatero: Wrote Manuscript, Designed Research, Performed Research, Analyzed Data, and Contributed New Analytical Tools. B. Sils: Wrote Manuscript, Designed Research, Performed Research, Analyzed Data, and Contributed New Analytical Tools. S. Phares: Wrote Manuscript, and Analyzed Data. M. Trusheim: Wrote Manuscript, Designed Research, Performed Research, Analyzed Data, and Contributed New Analytical Tools.

Acknowledgments

We would like to thank all those who participated in the MBP framework refinement primary research interviews conducted as part of this research.

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/14737167.2023.2159813.

Additional information

Funding

This work was funded by the National Pharmaceutical Council.