OBJECTIVES: Until now, the input to CE simulations was primarily the probability of events and costs associated with them. The probabilities are based on raw frequency data available or logistic regression models. The goal of this research improve simulations by incorporating Cox proportional hazards (CPH) analyses into these models to increase their validity and usefulness to policy makers. METHODS: A CPH analysis estimates the coefficients of a linear combination of predictors. Raising this to the "e" power provides the hazard ratio comparing a group of subjects defined by a specific vector of predictors to an "average" referent group. Once a Cox model is estimated, we define a hazard ISOBAR as the set of values that make the linear combination of predictors equal to a constant "c". "C" can be varied freely. This "c" can be used as the criteria for intervention assignment. The higher the "c" the smaller the sample to receive intervention but it is expected that the cost-effectiveness ratio would be higher. RESULTS: A simulated data set representing a 10-year study is generated with three predictors including a risk factor "R" as well as an outcome "D" and time to event. A logistic regression model was estimated to predict probabilities used as inputs to a fixed cost per event Markov model. Selection of sample to receive intervention used hazard ISOBARS as cutpoints and is compared to selection based on simple risk factor cut points. ISOBARS for c = 0, 1,...4 were used. It is useful to note that when c = 0, "e" to the c equals one and splits the sample into those above average risk and those below average risk. Intervention assignment based on risk alone used break points of top 50th, 25th and 10th percentile. CONCLUSIONS: Hazard ISOBAR cutpoints for interventions yielded larger cost-effectiveness ratios than those generated by risk cutpoints alone.