To make informed health policy decisions, we must consider both a treatment's effectiveness and its cost. We previously developed a novel probabilistic measure of cost-effectiveness using the potential outcomes framework. This approach elucidates the probability that a patient receiving one treatment will have a more cost-effective outcome than a patient receiving another treatment, at a particular willingness-to-pay value. Our cost-effectiveness determination (CED) curve serves as a graphical tool based on this parameter that overcomes limitations of currently used visual aids such as the acceptability curve. In this talk, we present an inverse-probability weighting approach to estimate the CED for censored survival outcomes. This approach accommodates modifications to account for confounding in observational databases, as well as regression approaches to allow for subgroup discovery. Through simulations, we show that the method has desirable finite-sample properties (e.g., low bias and proper coverage). As an illustration, we use data from a large observational cancer registry to determine the cost-effectiveness of adjuvant radiation therapy in endometrial cancer patients.