BACKGROUND: The purpose of PSA is to account for parameter uncertainty within a decision problem and is considered a necessary methodological requirement for any health economic model. PSA involves sampling model input parameters from their respective distributions, as apposed to using point estimate values. Guidance on the appropriate choice of distribution to assign specific inputs, and methods to parameterize such distributions, given different types of data sources, appears limited. In the absence of comparative data, modeling with RWE is becoming increasingly popular, therefore guidance on the appropriate methods for conducting PSA in such circumstances, is required.

AIMS: We sought to investigate and evaluate guidance available from the literature, and from HTA assessment bodies, on how to conduct PSA; and to determine if guidance exists on conducting PSA when modeling with RWE. We subsequently aimed to determine methods that have been utilized to conduct PSA in HTA submissions that have modeled with RWE, noting how such methods were received and revised. Finally, we intended to establish a framework for how to conduct PSA and whether differences in methodology are required when modeling with RWE.

CONCLUSIONS: Although PSA is a principal element of cost-effectiveness modeling, and PSA results are a key determinant of reimbursement decision making, detailed guidance on how to conduct PSA remains limited. Our search identified few literature-based studies on methods of PSA modeling, and recommendations from HTA bodies on how to parameterize input distributions were repeatedly vague. Specific guidance on how to conduct PSA when modeling with RWE is particularly restricted. Despite the limited guidance currently available, we have established a framework for how to conduct PSA, with specific consideration for methods that should be used when modeling with RWE. To streamline and standardize methods used in practice, this framework should be adopted when conducting PSA, in cost-effectiveness modeling.