An example of using the moment-generating function method for health economic evaluation in R. Also includes discrete event simulation code for comparison.
The moment-generating function (MGF) method is described in
Snowsill T. A new method for model-based health economic evaluation utilising and extending moment-generating functions. Med Decis Making (under review).
Briefly, it expresses quantities of interest (e.g., total discounted costs) as functions of random variables representing event times and calculates their expected values by using the moment-generating functions for those random variables.
The example depends on some R packages, all of which are available on CRAN:
> install.packages(c("flexsurv", "tidyr", "dplyr", "purrr"))Once you have the necessary packages, set your working directory to the directory containing the R scripts and source mgf.R to run the MGF method.
> setwd("/path/to/scripts")
> source("mgf.R")You will then have a data frame (actually a tibble) called results with the results of running the MGF method. Substitute DES.R for mgf.R to run the discrete event simulation instead.
The example included is a three state model (typical for health economic evaluations of cancer). The three states are Stable disease, Progressive disease and Dead.
There are three events included in the model:
- Disease progression (from Stable disease to Progressive disease)
- Death from stable disease (from Stable disease to Dead)
- Post-progrssion mortality (from Progressive disease to Dead)
These events are assumed to have time-to-event distributions (from time of entering the state) that are Weibull, Gompertz and log-normal respectively.
There are two treatment arms: Treatment and Control. The treatment has a proportional hazards impact on disease progression.
Costs are incurred for disease progression and death (one-off costs) as well as being incurred at a constant rate while in the Stable disease and Progressive disease states.
QALY weights are calculated from a baseline function (quadratic in age) multiplied by state-specific utility multipliers.
There are no plans to update or enhance this example, but you are free to use or adapt it as you wish (see License).
This project is licensed under the MIT License - see the LICENSE.md file for details.