The framework and supportive calculation tool can be adopted at several moments in the lifecycle of a MEA. They can for example be used to define and compare relevant MEAs in the initial reimbursement negotiations and decisions. However, they can also be applied at a later stage, for example to evaluate implemented arrangements after the timeframe of the payment agreement has ended. The information can then be used to enter potential renegotiations. To support end users of the framework and calculation tool, a list of key decision questions per framework part is outlined in Fig. 2.

Fig. 2

Key decision questions for end users per part of the framework and calculation tool

3.1 Are the Clinical Uncertainties and Financial Challenges Pressing Enough to be Decisive in the Reimbursement Decision? Part 1

The level of uncertainty in clinical evidence (e.g. the quality of evidence, the efficacy [precision of the effect size or the durability of the effects] and length of follow-up) can be seen as a main driver when choosing what type of reimbursement model might be suitable [10, 17, 19]. In line with the previous literature, clinical uncertainty is defined as ‘any explicitly or implicitly reported unresolved shortcoming, concern, question, or issue in the clinical evidence’ [40, 41]. To determine which reimbursement model could be relevant to compare in the calculation tool, Part 1 of the framework focuses on whether there are uncertainties in the clinical evidence [7, 15, 36]. Given that not necessarily all uncertainties can be mitigated nor are relevant to the reimbursement decision, the framework questions which clinical uncertainties are pressing enough to be decisive to the reimbursement decision and if an additional data collection could be of added value [33, 34, 36]. If there is already an indication that uncertainties may not be resolved over time irrespective of the reimbursement or payment model applied, or if the clinical uncertainties do not seem to hamper the reimbursement decision, a financial-based reimbursement model could be more suitable. This model facilitates that the financial risk of reimbursing the therapy under the existing clinical uncertainties is shared and the financial-based model can be updated over time when e.g the perception about the remaining uncertainties change [42]. When a financial-based reimbursement model is deemed more suitable, the next steps of the framework do not need to be taken, and end users are advised to consider multiple discount rates based on, for instance, their cost-effectiveness thresholds to determine what they find agreeable under the circumstances.

When the reimbursement model has been defined, the type of payment model that might be relevant needs to be determined. Multiple strategies have been described in the literature [14, 16, 19, 21, 38]. If financial challenges need to be mitigated, delayed payment models should be considered. Suitable types of delayed payment models may depend on the uncertainties relevant to the reimbursement model. For example, to account for the financial risk of possible non-responders or an unexpected large deterioration in patient outcomes of the main outcome measure annuity payments would be a potentially suitable payment model. Payments will be then spread over time where the height of the payments is linked to outcomes achieved. When clinical uncertainties are extremely large, payments at outcomes achieved can be considered where payments are only made after specific results have been achieved [14, 19, 43,44,45,46,47].

3.2 What is a Suitable and Relevant Outcome Measure to Which the Payment Can be Linked and How is Success Defined? Part 2

In Part 2 of the framework, when a MEA is linked to outcomes, involved stakeholders (e.g. healthcare payers and healthcare professionals) need to come to a consensus about when a therapy is effective to define to what extent the treatment can be considered a success and is thus ‘worth’ paying for. To evaluate product performance, the different possible levels of outcomes should be outlined (binary or ordinal), and the minimum and maximum success score of the main outcome measure per subgroup should be defined [6, 18, 19]. When selecting which outcome measure is suitable, the literature denotes the importance of frequently collected outcome measures that are both clinically relevant and relevant to the patient [6, 18, 19]. To enhance the feasibility of successfully implementing an outcome-based reimbursement model, it is advised to focus on one outcome (if possible) that does not pose a large administrative burden to the healthcare provider.

Finally, the end users need to decide whether successful outcomes can/should be defined on a population, sub-population or individual level. The literature denotes that when the payment model is linked to specific outcomes, a successful implementation is most feasible at an individual patient level [13, 16]. This feasibility is specifically high when there is a small patient population, which is often the case for high-priced one-off therapies [14,15,16]. If no suitable and relevant outcome measure can be found, end users are advised to consider financial-based reimbursement models.

3.3 Is There a Data Infrastructure in Place to Register the Main Outcome Measure and is that Outcome Measured in Such a Way That is Useable for the Payment Arrangement? Part 2

To capture the collected data correctly, efficiently and in a trustworthy manner so that they are usable for payment schemes, it is important to ensure that an infrastructure is in place [18, 19, 48, 49]. In the literature, the importance of aligning and sharing data between involved stakeholders to increase the feasibility of successfully implementing outcome-based reimbursement models is underlined [18, 19, 47]. Moreover, given that reimbursement and payment negotiants often have a tight timeframe, there might not be enough time to set up a new register. Therefore, the literature denotes the importance of having an existing registry when entering an outcome-based agreement [18, 19, 48]. If a registry is not in place, the feasibility of setting one up within the timeframe of the first outcome measurement should be explored, or the possibilities of collecting, registering, and sharing data through other means or sources. Therefore, in Part 2 of the framework, a key decision point is establishing this feasibility, for which experts in the field should be consulted. If there is no data infrastructure in place, and the feasibility of creating one is considered low, implementing financial-based reimbursement models is more feasible.

3.4 Is a Realistic Timeframe to Spread Out the Payments Possible and Can the Clinical Uncertainty Be Lowered Within this Timeframe? Part 2

Determining at what moment the main outcome measure should be measured depends on its application in clinical practice. In Part 2 of the framework, healthcare providers should be consulted on when the outcome is measured in clinical practice. Additionally, if it is not a routinely collected outcome measure, patient representatives can be consulted on how likely patients are willing to come back for non-routine appointments for measurements necessary for the payment model. The frequency of how often this main outcome is measured will determine the frequency of the spread payments during the timeframe of the payment agreement. In the literature, it is emphasised that the timeframe of the payment agreement should be long enough to allow for a reliable clinical assessment and adequate data collection, but, at the same time, must not be so long that the agreements become difficult to enforce or execute. In light of this, it is often noted that the timeframe should preferably be no longer than 5 years [16, 19, 50,51,52,53].

3.5 What Input is Needed to Perform the Calculations to Compare the Different Models? Part 3

Various input parameters for the calculation tool need to be defined in Part 3 of the framework. End users should know that in principle most steps can be completed using (publicly) available information to stakeholders involved in the reimbursement process of a health innovation (e.g. available health technology assessment [HTA] reports, clinical trials, European public assessment reports and/or available effectiveness data). Nonetheless, some information might be context specific or country specific and experts in the field, for example clinicians and patient representatives, need to be asked to provide input. Moreover, national procedures, political or societal discussions, and stakeholder preferences will shape the inputs and results of the calculated payment models and scenarios.

To determine how patients respond to the therapy, it was chosen to express the calculation of benefits in quality-adjusted life-years using information frequently available in reimbursement dossiers. Nevertheless, if other outcome measures are relevant, the calculation tool allows end users to choose their preferred country-relevant approach. To calculate the benefits gained within the timeframe of the payment agreement, transition probabilities (e.g. per month) and utilities (e.g. per year) need to be defined for each subpopulation for the best supportive care and the therapy under consideration. If further assumptions are relevant to determining how patients respond, this can be added to the tool (e.g. the transition between the health states depends on specific response classifications).

To calculate the associated costs expressed in euros (or any other applicable currency) per payment model, the (public list) price of the therapy needs to be defined, whereafter (outcome-related) discount rates are applied to this (list) price. In the framework, the simple payment model is further specified by determining a suitable discount rate applied over the defined therapy’s price to establish which amount of the treatment costs will be paid in the upfront payment. To select this discount rate, end users can consult previously recommended discount rates in HTA reports for similar health innovations or apply any rate they find relevant or suitable given the circumstances of the health innovation under consideration. For the innovative more complex payment model, the discount rates for each predefined outcome measure score need to be determined (e.g. depending on how success is defined in the therapy). By setting discount rates per outcome measure score, the payments can be adapted categorically with a fixed decrease in payment if patient responses drop below a certain threshold in a stepped manner and will be made after the patient response has been measured [19, 54, 55]. Consequently, it can be ensured that the payments made will relate to the outcomes achieved.