Objective: In drug repurposing decision analytical models could serve as a valuable tool by assessing the costs and benefits of the drug across its life cycle to aid decision-making 1,2. These models can range from quick “back of the envelope” calculations to very complex and time-consuming models 3. Deciding on the appropriate model complexity that would inform decisions adequately while meeting the time and financial constraints of the project is often challenging. For instance, early models for mechanism-based drug repurposing could involve complex modeling of patient trajectories through diagnostics and test pathways, requiring extensive resources that might not be available at early stages. Therefore, a trade-off should be made between cutting corners in the model while ensuring that it answers the research question. In this article we develop a decision tool that helps to decide on the complexity of models based on the consequences, good and bad, of 'cutting corners’.
Methods: We characterized model complexity by compiling model aspects and features informed by existing economic evaluation frameworks (population, intervention, comparator, outcomes, time horizon, perspective, and audience) and the TRUST tool for uncertainty assessment in models. We reviewed key literature, supplemented with our own expertise on different model complexity levels and its consequences for validity, transparency and time and resources needed to develop the model. Then, we developed a decision tool that helps inform model complexity.
Results: Model complexity can relate to population (heterogeneity, dynamics), intervention (impact on disease, care pathway placement, interactions with other factors), number of comparators, model outcomes (number of outcomes, downstream consequences), length of time horizon, perspective, audience (number of audiences, precision of results, decision context), model structure (aggregation level, connectivity), model approach (combination level, memory inclusion, interactions, explicit modelling of time), selection of evidence (review type, number of sources), model inputs (number of inputs, endpoints used, time dependency), model implementation (software used), and model analysis (type and number). The simplification of each model feature can produce a different consequence for the model validity, transparency and time and resources required. The SMART tool will assist modellers in selecting the complexity level and assess its consequences, for each model feature.
Conclusions: When repurposing drugs, factors such as efficacy, safety, cost-effectiveness, and regulatory requirements all play pivotal roles in decision-making. Ensuring that the chosen level of the decision model complexity captures all these factors comprehensively is important. The SMART tool for model conceptualization will help health economic modellers in deciding where and when to cut corners when developing models by laying bare the consequences of model choices.These informed choices can help optimize the use of resources and time while ensuring transparency and relevance in healthcare decision-making.