OBJECTIVES: Cost-effectiveness analysis plays a crucial role in evaluating the viability of healthcare interventions. Monte Carlo methods are widely used because they can be applied to probabilistic models (the class within we can find cost-effectiveness models) and the mathematical formalisms underlying it are easy to understand. However, this approach can present some theoretical limitation when compared to exact methods. In this study we aim to compare both approaches and analyse when each one should be applied.

METHODS: Considering some representative case studies, probabilistic models were designed and built. The tool PRISM – probabilistic symbolic model checker – was applied to study toy models and generate expected outcomes with exact methods. We stablished the fundamentals for implementing a probabilistic model on PRISM and to analyse it. In this way, we explore the differences in results obtained using exact methods with those obtained through simulation-based approaches.

RESULTS: Monte Carlo methods can provide inaccurate results by overestimating or underestimating outliers. Probabilistic models, combined with probabilistic symbolic model checking tools like PRISM, provide an exact approach to conduct such analyses. Moreover, PRISM syntax embeds a temporal logic language which may be used to infer properties about the model without the need of rebuilding it for that purpose.

CONCLUSIONS: Exact tools can be useful for Health Technology Assessment since the methods underlying their functioning are more efficient than those based in Monte Carlo methods but require more computational memory. PRISM computational tool presents flexibility, enabling users to leverage both exact and simulation-based approaches, according to the complexity of the model.