OBJECTIVES: The aim of this study was to develop a methodological framework to account for the impact of non-adherence on the cost-effectiveness of chronic medications in the context of health technology assessments (HTA) using time-to-event outcomes.

METHODS: The framework was informed by four linked stages of this research: (1) a systematic review of methodological papers that identified 12 methods that could be used to adjust for non-adherence, and assessed their suitability for use in HTA; (2) a simulation study that assessed the performance of four adjustment methods in 90 scenarios; (3) a case study that applied two g-methods to a trial of maintenance immunosuppressants in kidney transplantation in order to produce cost-effectiveness estimates based on external ‘real-world’ non-adherence data; and (4) the development of an analytical framework for incorporating non-adherence into cost-effectiveness models.

RESULTS: We put forward a seven-stage methodological framework covering recommendations for study design and analysis to produce adherence-adjusted effectiveness and cost-effectiveness estimates. Based on the review, the simulation study, and the case study, the framework made 17 recommendations and considerations covering: (1) understanding the clinical context, (2) identifying and measuring the relevant types of non-adherence, (3) specifying data requirements and making assumptions explicit, (4) selecting the appropriate non-adherence adjustment method, (5) adjusting treatment effectiveness for real-world non-adherence, (6) considering the impact of non-adherence on direct treatment costs, and (7) estimating and reporting adherence-adjusted cost-effectiveness of treatments. To apply the framework, three prerequisites need to be met: (a) identification and validated measurement of patient non-adherence, (b) identification and measurement of baseline and time-dependent confounders, and (c) access to individual patient-level data from a trial.

CONCLUSIONS: The framework provides guidance on the steps that could be followed to account for real-world non-adherence when undertaking cost-effectiveness analyses. Future research is recommended to use the framework in different case studies for evaluation.