Summary: | This document sets out software options for evidence synthesis that are compatible with probabilistic cost-effectiveness analysis, which is the preferred methodology for the NICE reference case. Four possibilities are discussed: 1. Evidence synthesis by Bayesian posterior estimation, and posterior sampling. Other parameters of the cost-effectiveness models can be incorporated into the same software platform. Bayesian Markov chain Monte Carlo simulation methods with WinBUGS software are the most popular choice. 2. Evidence synthesis by Bayesian posterior estimation. Posterior samples are exported to another package where the other parameters are generated and the costeffectiveness model evaluated. 3. Frequentist methods of parameter estimation followed by forward Monte Carlo simulation from the maximum likelihood estimates and their variance-covariance matrix. 4. Bootstrap re-sampling - a frequentist simulation approach to parameter estimation. When multiple parameters are estimated from the same synthesis model, we emphasise the need to choose a method that propagates the parameter correlation structure through the costeffectiveness model. A table is provided that shows the possible approaches and the restrictions on their use. Software packages for evidence synthesis are listed. Technical issues relating to software are covered in an Appendix. Finally we mention software suitable for transferring data between different software packages, and software that provides user-friendly interface for integrated software platforms. These offer investigators a flexible way of examining alternative scenarios
|