/ca_vaccination_impact

Reproducible code for our paper in Journal of General Internal Medicine

Primary LanguageHTML

Recent Shifts in Racial/Ethnic Disparities in COVID-19 Mortality in the Vaccination Period in California

This is reproducible code for our recent Journal of General Internal Medicine paper, “Recent Shifts in Racial/Ethnic Disparities in COVID-19 Mortality in the Vaccination Period in California”, which uses restricted-access, decedent-level death certificate data from the California Department of Public Health (CDPH) to systematically examine racial/ethnic disparities in COVID-19 deaths during the vaccination period. The full citation is:

Riley AR†, Kiang MV†, Chen Y-H, Bibbins-Domingo K, & Glymour MM, Recent shifts in racial/ethnic disparities in COVID-19 mortality in the vaccination period in California, Journal of General Internal Medicine (February 2022), doi: 10.1007/s11606-021-07380-6. †First authors contributed equally.

Issues

Please submit issues via Github or via email.

Important note about reproducibility

In accordance with our data use agreement with the CDPH Vital Statistics, we cannot share individual level data. When possible, we provide aggregated data in cases where there are more than 10 observations. This restriction means this pipeline is not fully reproducible without the restricted-access data.

Software

All analyses are conducted using R, which can be downloaded via CRAN, and the Joinpoint Regression Program, which can be downloaded from the National Cancer Institute.

We also recommend the use of RStudio when running R, which will allow users to take advantage of renv for dependency management.

Analysis pipeline

The analysis pipeline is divided into three discrete steps.

In Step (1), we clean, subset, munge, and calculate mortality rates using the raw (restricted-access) data. This results in a working dataframe that contains the data necessary for the Joinpoint Regression Program to fit our models of interest. These are held in the 01 to 04 code files.

In Step (2), the joinpoint regressions are fit in an external program (NCI Joinpoint Regression Program) and the results are exported. The ./joinpoint_analyses/age_std_rates.jps file contains our session information to reproduce our analysis and requires the ./joinpoint_analyses/age_std_rates_long.csv file generated from the step above (potentially with noise added to small cells).

In Step (3), the resulting (exported) joinpoint files are combined into a single file for plotting and to create tables. These are code files 05 to 08.

All files should be run sequentially.

Authors