U.S. Simulation of Lifetime Major Depressive Episode Prevalence and Recall Error

Citation: Tam J, Mezuk B, Zivin K, Meza R. U.S. Simulation of Lifetime Major Depressive Episode Prevalence and Recall Error. American Journal of Preventive Medicine. 2020. https://doi.org/10.1016/j.amepre.2020.03.021

This repository contains code to run a simulation model of major depressive episodes that adjusts for recall error in estimates of lifetime major depressive episodes (MDEs). The model is run in R and calibrated to data from the National Survey on Drug Use and Health (NSDUH) 2005-2017.

Download, clean, and harmonize NSDUH data

1. Download the full combined NSDUH dataset for 2002-2017 here.
2. Load the NSDUH data in R
3. Run data_clean.R. This produces the depprevs_2005-2017.Rda file.

Perform model calibration and parameter estimation

1. Specify the parameters to be estimated by adjusting the code in main.R (see code comments). This can also be done manually in parameters.xlsx.
2. Run calibration.R to generate parameter estimates that minimize the sum of squared distances between model and NSDUH data.

Run the model and generate Figures 2, 3, 4, S1, S2, S4:

1. Run main.R with the calibrated parameter estimates.
2. Run manuscript_figures.R.

Run sensitivity analysis and generate Figures S3 and S5:

1. Run sensitivity_analysis.R. This produces the PSE_200N_50B.Rda file.