A recent article by Ben et al. (2023) provides an R tutorial for implementing economic evaluations (often cost-effectiveness analysis) using data from clinical trials and analyzed using R. The article begins by providing summaries of the key issues that researchers face when conducting these economic evaluations:
- Missing values. Missing data are common in clinical trials, whether due to disenrollment, limited follow-up, or nonresponse. What are some methods to address this? The authors write: “Naive methods, such as mean imputation of missing values and last observation carried forward, are discouraged because they do not take into account the uncertainty of the imputed observations. The most robust methods for handling missing and/or censored data are multiple imputation (MI), inverse probability weighting (IPW), likelihood-based models, and Bayesian models. Of these, MI is the most widely used and is a valid method when missing data is related to observed data (e.g., missing at random, MAR) in economic evaluations.“The R package relevant to MI is mice.
- Biased data. Cost data are often skewed to the right, with most observations around the median, but a nontrivial number of very high cost outliers. The authors cite a scoping review (El Alili et al. 2022) and state that appropriate methods for handling skewed cost data include: “nonparametric bootstrapping, generalized linear models (GLM), hurdle models, and Bayesian models with a gamma distribution.”
- Correlated costs and effects. Sometimes treatment effects may be correlated (positively or negatively) with costs. Approaches to handling costs and correlated effects include “seemingly unrelated regressions (SUR), pairwise bootstrap costs and effects, and Bayesian bivariate models.”
- Initial imbalances in trial characteristics.. Even when individuals are randomized in a trial, randomization may be imperfect and trial characteristics may be unbalanced. Some approaches to address these differences include: including regression-based adjustment, propensity score adjustment, and matching.
Here is sample code to implement each of the 4 approaches.
Missing values. The R package relevant to MI is mice.
Address skewed data and correlated costs with bootstrapping methodology and seemingly unrelated regressions (SUR). The authors use the boot function provided by boot Package R. The bootstrap function is used to resample the data and for each bootstrap sample a SUR model is fitted using the system setting function. [The authors note that rather than using SUR, a linear mixed model (LMM) could be fit instead using the lme4 or nlme R packages].
Relevant statistics of interest can then be extracted as follows:
Additional instructions are given on how to create a profitability plane and a profitability acceptability curve. You can read the full article. here.