OBJECTIVES : Cure models in health economic analyses, which typically assume that patients are permanently free of a full disease burden following successful treatment, have become more common with the development of potentially curative therapies. However, despite being a growing area, few economic guidelines exist for the development of such models. Here, the key considerations for developing cure models are discussed before presenting a conceptual guideline for best practices.

METHODS : Previously developed cure models were reviewed via a targeted search, primarily considering cost-effectiveness analyses. From here, the main factors impacting the validity of the cure assumptions used were compiled and compared. Based on these results, a conceptual framework and guideline for using cure models was developed.

RESULTS : After review of previous models, it was observed that valid cure assumptions will vary widely depending on the specific disease characteristics and expected treatment outcomes. However, four key areas of consideration that universally apply when deciding on the best cure approach were noted: (1) Which time cut-off should be used signaling the end of treatment and beginning of the “cure state”?; (2) Should the survival rate and other outcomes in the cure state reflect the general population, or should a disease-related adjustment still be used after a patient is “cured”?; (3) Should patients in the cure state incur different complications and associated costs than the general population, and for how long?; (4) Should patients in the cure state experience equal utility to the general population, or are there still disease-related considerations in these patients?

CONCLUSIONS : As described here, cure models are associated with a set of complex assumptions that must be carefully considered when developing a valid analysis. By methodically approaching these assumptions, the expected outcomes and costs of curative therapies can be modeled to more accurately reflect real-world treatment expectations with reduced uncertainty.