OBJECTIVES: The cost-effectiveness acceptability curve (CEAC) based on the net benefit (NB) is defined as a graphical method for representing uncertainty in economic evaluation. The decision-making using the CEAC is probability that the new treatment is the cost-effectiveness if NB >0. However, the NB is not binary data but continuous, therefore the information is used partially as a positive or not. The quantity CEAC is proposed aiming to estimate overall impact based on the sum of individual NB and to suggest the graphical result. METHODS: We generated individual cost and effect data based on three scenarios. First, individual incremental cost-effectiveness ratio (ICER) was taken a sample from the normally distributed. Others, the tails of the distribution were not equally weighted; left-skew and right-skew with normal distribution. We calculated to compare the minimum Willan’s estimator (the probability of positive result = 0.5) on the CEAC and the point of zero on the sum of individual NB given WTP (= ICER) in three scenarios. The non-parametric bootstraps were used to explore the statistical inferences that involved estimators. RESULTS: A total of 1,000 patients were generated their costs and effects by three scenarios. Also, we performed 10,000 bootstrapping to estimate 95% confidence interval of estimators. In normal distribution, the Willan’s estimator was approximated by the ICER. However, in skew data, it was more robust than the ICER as a median’s property. The graphical result was represented overall impact under zero on the sum of individual NB given WTP. CONCLUSIONS: The minimum Willan’s estimator has a robustness but it could not reflect overall impact by means of using binary outcome. The sum of individual NB was indicated distribution of given WTP among total patients. When the Willan’s estimator is different from the ICER, the subgroup or multivariate analysis will be needed for decision making.