Estimating causal effects for COPD Emergency Department visits

This repository contains code for a Bayesian spatial causal inference model. Our model estimates the causal effects of county-level exposures on the number of chronic obstructive pulmonary disease (COPD) emergency department (ED) visits. The dataset comes from the counties in North Carolina in 2023.

Packages Required

In order to get this code to run, you must have a version of R >= 3.5.0 installed. It would be ideal to have the most recent version of R. You must install the R packages that appear in the lines library(...) to run these R scripts. Most of these can be installed using the R command line or the RStudio console or GUI. You also need to install rstan. To install rstan, you must first configure your R installation to be able to compile C++ code. See here for instructions. Once the C++ toolchain is configured, you can then install rstan using

install.packages("rstan", repos = "https://cloud.r-project.org/", dependencies = TRUE)

Income as exposure

This folder contains code to implement the model with median income (log-transformed) as the exposure. All files in this folder must be saved in the SAME local directory for the code to run. The main scripts are as follows:

• Car_Stage_1_income.R: Implements the Stage 1 generalized propensity score (GPS) model for median income
• Car_Stage_2_income.R: Implements the Stage 2 outcome regression model for estimating the causal relative risk of median income on the number of COPD ED visits
• get_exposure_response_curve_income.R: This estimates the causal exposure-response curve for the average causal effect of median income on the number of COPD ED visits
• GPS_CAR_model.stan: MCMC sampler for fitting the Stage 1 GPS model
• outcome_CAR_model.stan: MCMC sampler for fitting the Stage 2 outcome regression

Vape shop density as exposure

This folder contains code to implement the model with vape shop density (zero-truncated and log-transformed) as the exposure. All files in this folder must be saved in the SAME local directory for the code to run. The main scripts are as follows:

• Car_Stage_1_vape_shops.R: Implements the Stage 1 generalized propensity score (GPS) model for vape shop density
• Car_Stage_2_vape_shops.R: Implements the Stage 2 outcome regression model for estimating the causal relative risk of vape shpo density on the number of COPD ED visits
• get_exposure_response_curve_vape_shops.R: This estimates the causal exposure-response curve for the average causal effect of vape shop density on the number of COPD ED visits
• GPS_CAR_model.stan: MCMC sampler for fitting the Stage 1 GPS model
• outcome_CAR_model.stan: MCMC sampler for fitting the Stage 2 outcome regression model

Visualization and test for spatial autocorrelation

This folder contains code to make heatmaps of the COPD ED visit rate, median income (log-transformed), and vape shop density (zero-truncated and log-transformed) for the counties in North Carolina. The code also conducts the global Moran's I test for positive spatial autocorrelation for each of these variables. The main script is:

• heatmaps_and_Morans_I.R: Makes the heatmaps and conducts the global Moran's I tests