OBJECTIVES: Communication of decision analytic models is limited by their complexity and cumbersome computation requirements, lack of end-users' experience and familiarity with modeling techniques, and/or access to dedicated software applications. We explored whether second-order Monte-Carlo simulation (MCS) and regression techniques could be used to simplify complex decision trees and facilitate their use and modification by end users. METHODS: This analysis was based on a previously developed 4-cycle Markov-chain model projecting the one-year cost utility of a continuous combined hormone replacement therapy (CCHRT) vs. no therapy in women with menopausal symptoms. Using the model, we conducted 1000 MCS to project the net incremental QALYs, costs, and net monetary benefit (NMB) associated with CCHRT. Results of the MCS were summarized using OLS regressions, in which incremental QALYs and costs were separately predicted using the model's inputs. These predictors and their coefficients were then used to back-predict the NMB of CCHRT. Multiple tests were conducted to determine the predictive abilities of the regression. RESULTS: The net cost and QALY regressions were associated with R2 of 0.998 and 0.880, respectively. The model and the regression estimated the net costs associated with CCHRT to be $400.92 and $402.53, respectively. The net estimated QALY gains were 0.0380 and 0.0356, respectively. In simultaneous runs of the model and the regression, both approaches produced concordant conclusions regarding the cost-effectiveness of CCHRT in >95% of cases (kappa >0.78; P<0.001). CONCLUSIONS: MCS and regression techniques can be combined to meaningfully and successfully simplify complex decision trees into reduced, empirical forms. More broadly, the regression representation may be more universally adopted by end users, who can readily adapt it, using basic algebra, to approximate the results of the original decision tree to their own settings. Hopefully, this representation could expand appreciation of cost-effectiveness analyses on the part of end users.