dampack: A Flexible R Package for Analyzing and Visualizing Cost-Effectiveness Analysis Results
Published in 41th Annual North American Meeting of the Society for Medical Decision Making, 2019
Purpose: To improve the accessibility, transparency, and reproducibility of state-of-the-art cost-effectiveness analysis (CEA) methods for users and developers of decision-analytic models.
Methods: We developed the R package dampack (Decision-Analytic Modeling Package) to analyze and visualize CEA results with expanded functionality over similar existing packages. Using standard CEA outputs exported from a model developed in any software, dampack facilitates both basic and advanced calculations of CEA results crucial for informing decision making. Users can input results into dampack in one of two ways: as a probabilistic sensitivity analysis (PSA) dataset or as deterministic results (e.g., expected costs and effectiveness of each strategy). The package can be used to identify dominated strategies, calculate incremental cost-effectiveness ratios, plot the efficient frontier, and visualize one- and two-way sensitivity analyses. Additionally, for PSA datasets, dampack users can construct cost-effectiveness acceptability and expected loss curves and compute the expected value of perfect, partial, and sample information. Sensitivity and value of information analysis functionality are implemented using linear meta-models based on the PSA samples. We also designed dampack to deliver customizable, publication-quality figures for all analyses. The source code of the dampack package is freely available, allowing users to easily customize and extend its functionality. To demonstrate the functionality of dampack, we analyzed the PSA outputs from a previously published evaluation of the cost-effectiveness of different treatment strategies for Clostridioides difficile (C. diff). The PSA compared 48 treatment strategies, sampling 10,000 sets of 51 parameters.
Results: CEA results generated by dampack from the C. diff PSA are shown in the 4-panel figure. The cost-effectiveness plot highlights the six strategies on the efficient frontier, while also indicating strongly and weakly dominated strategies. One-way sensitivity analysis plots show how the optimal strategy changes as model parameters are varied individually. PSA results are additionally summarized through cost-effectiveness acceptability and expected loss curves. For each, dampack also plots the frontier, indicating the optimal strategy in expectation at each willingness-to-pay threshold.
Conclusion: dampack is a powerful and flexible package for the analysis and visualization of CEA results. It can summarize key CEA results, even when comparing a large number of strategies. With user-friendly and computationally efficient implementations, dampack also facilitates the use of state-of-the-art economic evaluation methods, such as value of information analysis, to a broader user-base.