OBJECTIVES : There are currently a number of software tools in R for prediction and simulation of multi-state models. Our aim was to compare them and assess their applicability for cost-effectiveness analysis.

METHODS : We evaluated open-source software for prediction and simulation of multi-state models. Three R packages (mstate, hesim, and heemod) were considered. They were compared on seven criteria: (i) computational efficiency, (ii) facility for cohort and individual-level modeling, (iii) incorporation of patient heterogeneity, (iv) use of probabilistic sensitivity analysis (PSA), (v) support for flexible survival modeling, (vi) whether time is continuous or discrete, and (vii) integration with cost-effectiveness analysis.

RESULTS : The hesim package was the fastest and could simulate a parametric individual-level “clock reset” multi-state model with 1,000 patients and 100 PSA iterations thousands of times faster than mstate. Both heemod and hesim can be run for a population of heterogeneous patients, while mstate is designed to simulate a single covariate profile at a time. Likewise, hesim and heemod natively support PSA, while mstate does not. However, heemod is not easily integrated with flexible survival models (e.g., parametric, splines, fractional polynomials, Cox regression), while both mstate and hesim are. Similarly, mstate and hesim are in continuous time while heemod is not. mstate and heemod support both cohort- and individual-level models, while cohort-level models are still under development in hesim. Finally, only hesim and heemod have functions that can be used to directly perform cost-effectiveness analyses and represent decision uncertainty.

CONCLUSIONS : R packages can be used for prediction and simulation of multi-state models. These packages can be used for model-based cost-effectiveness analysis, which can enhance the integration between statistical and economic modeling. However, packages vary in computational efficiency, flexibility, and precision, which should be considered when developing an economic model.