For the next blog post in my series reviewing papers in the Economics of Cell and Gene Therapies collection in Pharmacoeconomics, I review ‘Using Real-World Data to Inform Value-Based Contracts for Cell and Gene Therapies in Medicaid’ found at the following link: https://link.springer.com/article/10.1007/s40273-023-01335-x.
The premise of this study is to demonstrate that real-world data can be used to inform value-based contracts. Two gene therapies for haemophilia A (valoctocogene roxaparvovec) and B (etranacogene dezaparvovec), respectively, act as the case study.
The study aimed to 1. Estimate the cost impacts of standard-of-care in haemophilia 2. Use the standard-of-care costs as inputs in a simple cost calculator to estimate cost offsets when eligible patients are modelled as receiving gene therapy for haemophilia. This enabled the authors to determine the cost savings of eligible patients receiving gene therapy and the time period over which the costs of gene therapy can be recouped. The three workstreams conducted to meet these aims are summarised below:
1. Targeted literature reviews were conducted to identify cost inputs for the subsequent cost analysis. This included reviewing recent economic evaluations for resource utilisation and resource costs;
2. The Colorado Department of Health Care Policy and Financing database was accessed to determine real-world resource use and costs between 2018-2022. Specific criteria based on number of claims per year were used to identify patients who would be eligible for gene therapy. To be classified with moderate and severe haemophilia B, patients required at least four insurance claims and severe haemophilia A patients required six insurance claims. The two respective patient populations were pooled across the time horizon, and average annual costs were estimated for haemophilia A and haemophilia B, respectively;
3. A cost calculator in the form of a partitioned survival model with two health states (alive and dead) was developed separately for haemophilia A and haemophilia B. This enabled calculation of cost offsets if patients were to switch to gene therapy from standard-of-care. The durability of gene therapies is a key source of uncertainty; as this is a basic costing model, assumptions were made regarding the proportions of individuals who would discontinue gene therapy at certain time points in place of more robust methods to calculate treatment effect durability. The two costing studies conducted in workstreams 1 and 2 acted as model scenarios.
The literature review identified costs from relevant economic evaluation which demonstrated the wide range of estimated annual costs (e.g., standard half-life FVIII in haemophilia A and alprolix in haemophilia B). Claims cost analysis demonstrate that annual costs are lower (around four times) than published costs with haemophilia A having larger annual costs. The authors’ model suggests that it would take around 8 years to recoup costs for treating patients with haemophilia A and around 6 years for haemophilia B when real-world data are used. Using published costs, this is around 5 years for both. The probabilistic analysis indicated that uncertainty had a large influence on the results with only around a 50% probability of payers breaking even at 10 years in either therapy area. The number of patients eligible for gene therapy had a large impact on the time to recoup costs.
The authors highlight that the key limitation of this study is its use of Colorado claims data which limits the representativeness of the findings.
However, the study has other limitations which are not clearly discussed. Firstly, the simple model structure is arguably too simple to capture the clinical impacts of gene therapy which are relevant to understanding its overall long-term cost impacts. This includes relevant clinical outcomes beyond that of death with an example being durability of treatment effect. The authors capture this as a proportion of patients discontinuing but this is rather crude and it is not clear in which direction the bias acts. A further example is eligibility criteria which, in the real-world costing, was based on a minimum number of claims for standard-of-care treatment, but this is not the only clinical consideration of eligibility for gene therapy (and could explain some of the differences in costs between the claims data and the published evidence). To some extent clinical outcomes may be captured within the annual costs incorporated in the model. In effect, the model captures only half the picture.
In addition, the examination of the impact of using claims data versus published evidence is helpful and it demonstrates the stark difference in the two sources. However, although the authors discuss reasons for the differences, there is no acknowledgment of how this then impacts the conclusions they have drawn regarding the design of value-based contracts. Specifically, in a situation where the costs are so stark between the claims data and the published data (and within the published data), which source is providing the most accurate estimation of costs? You may say the claims data as these are costs borne by the payer, but where assumptions are required around eligibility criteria for gene therapy and are dependent on data availability, this could be either an over-estimation or an under-estimation of costs. This limitation may be confined to specific disease areas dependent on data availability and would need to be carefully considered were this method to be tested in other disease areas.
In terms of implications for value-based contracts, the authors state three findings that can inform value-based contracts:
- Differences in real-world and published cost estimates;
- Treatment eligibility impacts the recoup time period;
- Cost of pre-treatment is a source of uncertainty in estimating recoup time period.
Essentially, more patients receiving gene therapy leads to higher cost for payers which will take longer to claw back with any additional costs adding to this; there is uncertainty as to which data sources to use to inform the design of value-based contracts.
Implications of this study to the design of value-based contracts which are not discussed in the publication are the following:
- Careful defining of the patient population who are eligible for the gene therapy is necessary to contain costs and to enable a reasonable recoup period as anything beyond 5 years is unlikely to be attractive for payers or manufacturers;
- Similarly, the relevant comparators of gene therapy need to be identified and validated within any database being used to help estimate the recoup period. This study did not validate the identification of eligible patients with clinicians or patient groups; guidance to this effect may be helpful to payers;
- Determination of a recoup period requires a consideration of health outcomes. If a patient’s health does not improve or worsens, the cost impact would be different to that of a patient whose health improved. The individual clinical impact of gene therapies is important in estimating the recoup period and thus should ideally be determined on per-patient basis and averaged across the pooled data. This would contribute to a more accurate estimation of recoup period;
- Payer budgets are commonly determined on a 5-year basis meaning that if recoup periods are longer than this, value-based contracts may be unpalatable to payers. Although it may also incentivise a re-consideration of budget time frames.
In conclusion, this paper provides a good case study for estimating the recoup time period for gene therapy from a payer perspective and demonstrates the importance of carefully considering data source for estimating the recoup time period.
Leave a Reply