PURPOSE: • To help researchers understand how published Kaplan-Meier curves can be augmented with other statistics (e.g., event counts, numbers at risk, median time-to-event, hazard ratio) to improve estimation of time-to-event data, and • To illustrate how various approaches to generating analyzable data from the curves perform in parametric fitting/projection and comparative analyses.  Examples will be used to demonstrate and compare the various approaches, and to highlight their strengths and limitations. Attendees will be asked to respond to questions based on the presented material via a web-enabled application; a summary of responses will be shown and used for further discussion and clarifications, where necessary. DESCRIPTION:  Economic evaluations often rely on published data for model inputs. In particular, time-to-event distributions are often taken from publications and subjected to parametric fitting analyses to project unobserved portions of the curve. A standard, but somewhat crude approach is to perform least-squares regressions on coordinates derived by digitizing the published curves. Two approaches are available, however, to derive, virtual patient-level (VPL) data in the form of interval censored, or specific event/censoring times. VPL data can be used for parametric fitting to project incomplete Kaplan-Meier curves or derive comparative measures in pooled analyses of data from different curves. Analyses can also produce standard errors of parameter estimates, which are used in probabilistic sensitivity analyses. The approach used to generate data from curves can affect the accuracy of derived estimates. For instance, analyses of coordinates using least-squares regression is prone to bias with all types of distributions, while actual event times may be more accurate than interval censored data for log-normal data. Examples will be used to demonstrate the VPL data generation process from curves for which actual patient data are available. Virtual and actual patient data will be analyzed to compare the accuracy of approaches.