A Multistate Markovian Model of the Economic Burden for Allergy Immunotherapy

The incidence of allergic rhinoconjunctivitis due to pollinosis is increasing in Western countries. The first-line therapy (No-AIT) typically involves the administration of antihistamines and corticosteroid sprays to manage symptoms. Immunotherapy repre- sents an alternative treatment option, as it promotes desensitization to allergens. However, it is associated with significant costs. Currently, two types of allergen immunotherapy (AIT) are prescribed: subcutaneous immunotherapy and sublingual immunother- apy. This article compares these two therapeutic options with No-AIT. The comparison is conducted through a cost-effectiveness analysis (CEA), which evaluates health-related outcomes by estimating the incremental cost per unit of change in a composite outcome that combines morbidity and quality-of-life metrics. To perform the analysis, we developed a realistic multistate model describing the progression of a cohort of patients undergoing the three therapeutic approaches. The model was designed to be sufficiently flexible to account for treatment-related challenges commonly observed in real-world settings, which are often inade- quately represented in randomized controlled trials. By employing a novel two-dimensional framework, we tracked the proportion of the cohort transitioning between health states during each cycle while simultaneously capturing the origin and destination of each transition. This approach enabled the integration of structural features that are typically overlooked, such as early treatment discontinuation, transition rewards, nonstationarities associated with the usual termination of immunotherapy after three years, and differential protection against severe complications (e.g., asthma) depending on whether immunotherapy was completed or not. Deterministic simulations were conducted using standard input parameters, supplemented by probabilistic simulations to generate CEACs for each of the three strategies. The results from our model indicate that AIT is not cost-effective unless the payer exhibits a moderately high willingness-to-pay. These findings have important implications for the pharmaceutical industry involved in the production of AIT drugs.