| Objective This study was to optimize the multi-criteria decision analysis (MCDA)model for benefit-risk evaluation of medicines by the Monte Carlo simulation. Withfexofenadine, loratadine and cetirizine treatment allergic rhinitis model as an exampleto verify the feasibility of the optimized MCDA model. Trying to establish astructured, transparent and can communicate methodology of benefit-risk evaluation.Methods This study analyzed the existing methods for benefit-risk evaluationof medicines, and selected the most likely to be widely applied method——MCDAmodel. There were nine steps as following:(1) Establishment of the decision context: identification of the indication, the dosageand population for that indication.(2) Identification of the benefit and risk indicators: this study identified theindicators according to primary endpoint of the pivotal clinical trials and guidelines.(3) Data collection and processing: We should give preference to the high level ofevidence data, such as system evaluation. The model also could select therandomized controlled trials and then pooled data by Meta-analysis.(4) Scoring of the indicators: Setting the optimal value and worst values of eachindicator, then the values translated into0-100score by using fixed scales.(5) Assignment of a weight to each indicator using swing-weighting method.(6) Calculation of the weighted scores: According to the formula, the Hiviews3software calculated the weighted scores of benefit, risk and total benefit-risk of eachmedicine. (7) Conduction of a sensitivity analysis: The relative weight of the risk or benefitwas changed between0-1, and observed the results of the total benefit-risk value.(8) Conduction of the uncertainty analysis: We run the Monte Carlo simulation toanalysis the uncertainty in the Excel2003by the crystal ball software. Assumptioncells was benefit and risk indicators, forecast cells was the value difference ofbenefit, risk and total benefit-risk.Run3000times, then calculated the95%confidence interval of weighted score difference.(9) Overall conclusion.To calculate the95%confidence interval of benefit-risk difference valuesbetween the two medicines, the optimized MCDA model analysis the95%confidence interval of each data. If the confidence interval didn’t contain the zero,you could think benefit-risk values between medicines have statistical significance.If the confidence interval contained the zero, you could not think have statisticalsignificance, but you could calculate the probability of medicine Y was superiormedicine X by running probability simulation. So stakeholders could make adecision according to the probability.For fexofenadine, loratadine and cetirizine for treatment of allergic rhinitis asan example, verifying the advantages of the optimized MCDA model.ResultsBenefit value: The benefit value of fexofenadine, loratadine and cetirizine were24,21and30, respectively. The differences value of fexofenadine and loratadinewas3,95%confidence interval was (-3,9), so fexofenadine and loratadine has notstatistically significant. The differences value of fexofenadine and cetirizine was-6,95%confidence interval was (-13,-1). The differences value of loratadine andcetirizine was-9,95%confidence interval was (-16,-3). So cetirizine was overallmost preferred. The risk value: the risk value of fexofenadine, loratadine and cetirizine were-1,-7and-6. The differences value of fexofenadine and loratadine was-6,95%confidence interval was (-8,13). The differences value of fexofenadine and cetirizinewas5,95%confidence interval was (-8,14). The differences value of loratadine andcetirizine was-1,95%confidence interval was (-11,11). So the risk had nostatistical significance among the three medicines. The simulation results showedthat fexofenadine was superior to loratadine and cetirizine for the probability were70.3%and69.4%, and cetirizine was superior to loratadine was a50.7%chance.The total benefit-risk value: the total benefit-risk value of fexofenadine,loratadine and cetirizine were17,13and20. The differences value of fexofenadineand loratadine was4,95%confidence interval was (-2,8), so the total benefit-riskvalue had no statistical significance between fexofenadine and loratadine; thedifferences value of fexofenadine and cetirizine was-3,95%confidence interval was(-10,2), so the total benefit-risk value had no statistical significance betweenfexofenadine and cetirizine. The simulation results showed that cetirizine wassuperior to fexofenadine for the probability was90.9%, and fexofenadine wassuperior to loratadine for the probability was85.3%. The differences value ofloratadine and cetirizine was-7,95%confidence interval was (-12,-1). So cetirizinewas superior to loratadine.Conclusion (1) The optimized MCDA model can calculate the95%confidenceinterval of differences values between medicines and evaluate benefit-risk accordingto the interval. Comparing to the MCDA model which only obtain the results on thebasis of the point estimates were more reasonable and more accurate.(2) Theoptimized MCDA model could calculate the probability that medicine Y wassuperior to medicine X, so it provided difference treatment options to differencedecision maker, and helped stakeholder (patients, doctors, regulators and insurance institutions) make their own medical decision.(3) The study verified the feasibilityof the optimized MCDA model by the histamine medicines evaluation. |
PDF Full Text Request