Background/Purpose: The treatment and prognosis of moderate to severe rheumatoid arthritis has improved considerably due to the advent of biological therapies. But at the same time, the high costs of treatment are a threat to affordability and access to patients, and highlight the need for cost-effectiveness models that can help align prices with value. Unfortunately, there is no consensus in the literature on the proper mathematical structure for these models and estimates of value vary widely. Moreover, new evidence on the efficacy of biologics is rapidly evolving, so estimates of value must be updated as the evidence base evolves. Transparent, flexible, and accessible cost-effectiveness models that shed light on the implications of different modeling approaches are needed.

Methods: We developed an open-source “family” of cost-effectiveness models that reflects a number of scientifically defensible approaches and variation in the perspectives of decision makers. The code is publically available on GitHub, a version control repository, and the model has been released as an R package. We also created a user-friendly web application where users can modify parameter values or structural assumptions and run the model online. We used the models to compare the cost-effectiveness of a standard sequence of 6 biologics treatments to conventional disease-modifying antirheumatic drugs (cDMARDs) using a number of different modeling assumptions.

Results: Under various plausible assumptions, incremental cost-effectiveness ratios (ICERs) ranged from less than $100,000 to around $300,000. Biologics were predicted to be more cost-effective when they had a larger effect on the Health Assessment Questionnaire (HAQ) Disability Index score at 6 months, increases in the HAQ score were assumed to lead to larger increases in mortality, the drop in clinical efficacy after each treatment failure was smaller, the HAQ score on cDMARDs progressed more quickly over time, biologics were assumed to be discounted more from their wholesale acquisition cost, and productivity losses were larger. For example, under our baseline assumptions, the ICER was approximately $120,000 when the effect of treatment on HAQ did not decline after the first treatment failure and over $200,000 when we assumed no clinical efficacy after 1st line treatment. In addition, subgroup analyses conducted using our baseline assumptions indicated that biologics were more cost-effectiveness in younger patients, with ICERs ranging from $100,00 at age 35 to $165,000 at age 70.

Conclusion: Estimates of value vary considerably across different modeling assumptions, suggesting that steps must be taken to shrink diversity in modeling approaches and evidence considered to the greatest extent possible. These include new research studies and processes to encourage collaboration.

« Back to ACR/ARHP Annual Meeting 2017

ACR Meeting Abstracts - https://acrabstracts.org/abstract/open-source-conensus-based-models-to-improve-the-cost-effectiveness-of-rheumatology-care/