OBJECTIVES: The Institute for Clinical and Economic Review (ICER) published a report of the cost-effectiveness of peanut allergy immunotherapy treatment AR-101 (Peanut (Arachis hypogaea) Allergen Powder-dnfp, “PTAH”) compared to avoidance alone in 2019. PTAH was approved by the FDA in 2020 and two-year extension trial data is now available. Our objective was to replicate and update the previously developed ICER model for PTAH to include extension trial data and updates to drug costs.

METHODS: We replicated the five-state Markov model (treatment + avoidance/avoidance alone, untreated with peanut sensitivity, peanut desensitized, peanut tolerant, and death), estimating costs and outcomes from the payer perspective over a lifetime horizon. Estimates from the 2019 ICER model were used as inputs for year one, and year two clinical inputs were derived from the PALISADE extension trial. Costs were in 2022 dollars and a 3% discount rate was applied to costs and outcomes.

RESULTS: The model estimated increased QALYs for the intervention arm vs. avoidance alone (26.56 vs. 26.06) and increased costs ($220,200 vs. $7,300) over the lifetime horizon. This yielded an incremental cost-effectiveness ratio of $442,200/QALY based on deterministic results. The one-way sensitivity analysis indicated that the utility values were primary drivers of the results, followed by the cost of PTAH. The probabilistic sensitivity analysis indicated that PTAH had a 0.06% chance of being cost-effective at a willingness-to-pay threshold of $100,000/QALY.

CONCLUSIONS: Model results indicate that PTAH is not cost-effective at the current price according to commonly accepted cost-effectiveness thresholds. The results are highly dependent upon child utility values, indicating that the assumptions used to derive these values could have significant implications for decision-making. Additionally, food allergy is a condition that affects both parents and children, so consideration of family spillover effects is crucial and is in development for future versions of this model.