OBJECTIVES: To compare and contrast the structures and cost-effectiveness results for two different modeling approaches in advanced COPD. The analyses each compared once-daily fluticasone furoate/umeclidinium/vilanterol (FF/UMEC/VI) 100/62.5/25μg with twice-daily budesonide/formoterol (BUD/FOR) 400/12μg, based on the FULFIL trial (NCT02345161) outcomes.

METHODS: Two different model structures were reviewed: a linked risk equation model (GALAXY, Zhang et al. VIH 2017) versus a two-part model, combining a decision tree for the trial period with a Markov model for long-term extrapolation (Fenwick et al. VIH 2018). Incorporation of treatment effects, costs of events and clinical status, utility weights, survival projections, model findings and assessments of uncertainty were compared.

RESULTS: The two models differ in the method of estimating declining lung function and risk of exacerbation among the model cohort, including differences in the links between exacerbation and marginal lung function decline. Differences were noted in the costing structure and assignment of utility decrements. Both models included separate handling of moderate and severe exacerbations and discontinuation parameters; however, only the two-part model included pneumonia events. Despite differing approaches, both models showed FF/UMEC/VI as cost-effective compared with BUD/FOR (with incremental costs, incremental QALY, and incremental cost-effectiveness ratios [ICER]; of £1,652, 0.49, and £3,357/QALY gained for the GALAXY model, and £677, 0.50, and £1,325/QALY gained for the two-part model). Both models included probabilistic analyses varying risk prediction and treatment effect inputs and scenario testing of trial subgroup efficacy; conclusion of neither model were significantly impacted by uncertainty.

CONCLUSIONS: Despite differences in the model structures and the handling of treatment effect inputs, the two models provided similar results and demonstrated the cost-effectiveness of FF/UMEC/VI compared with BUD/FOR. It further supports the robust results of the FULFIL trial. Additional considerations regarding availability of market-specific data may impact model selection in future adaptation and submission work.