Agent-based model for Clostridium difficile transmission in hospital.
The epidemiological model is implemented in C++ and packaged in a R library named abmcdiff.
src: C++ code of the epidemiological model.Rlibrary: scripts to build the R libraryabmcdiffsimul: R scripts that run simulations using the compiled libraryabmcdiff.doc: model documentation
First, the R library abmcdiff that wraps the C++ model must be built. Execute build_library in the Rlibrary folder.
The architecture of the rooms for each health care setting simulated must be specified in a CSV file. See room_hosp.csv for an example.
Then, the parameters defining the population modelled in the health care setting must be defined in another CSV file. These includes the number of HCW staff, the stay duration distribution parameters, the admission rate, contact rates between individuals, contamination decay rates, etc. See prm_hosp.csv for an example.
If there are more than one health care setting modelleed (e.g., a hospital and a long term care facility), the movement between the facilities can be defined in the form of a matrix of transfer rates (with 0 on the diagonal). See prm_mvt_hcs.csv for an example.
Finally, the simulation parameters -- like number of Monte Carlo iterations, the time horizon and time step -- must be defined in a separate CSV file. See prm_sim.csv for an example.
The R library abmcdiff must be loaded.
Once all CSV files have been saved to define the model parameters, they can be imported in a R script and loaded as R objects (typically as lists).
One of the main R/C++ wrapping function, abmcdiff_one_simul(), takes those parameter objects as inputs and run the disease transmission simulations in the health care settings defined. See run.R for an example.
The analysis is performed separately, loading the simulation outputs saved in a RData file (e.g., simul.RData).
We want to investigate the cost-effectiveness of a potential vaccine against C. difficile currently in stage 2. Its main feature is to reduce the risk of symptomatic infection (does not provide full immunity), and provides this partial immunity for a limited period of time (2 years?).
There are two vaccination strategies envisaged:
- Vaccinate some patients before admission. Only applies to non-emergency admissions.
- Vaccinate patients who had a CDI during their stay. The simulation must keep track of them if/when they are re-admitted.
For the cost effectiveness analysis, the costs must be recorded at the individual level, with a date attached (for discounting). The costs associated with CDI are:
- isolation
- treatment (drugs)
- potentially longer hospital stay