How We Investigated UNOS’s Liver Allocation Policy and Inequities in the Liver Transplant System

Two investigations from The Markup/The Washington Post

---

This repository contains code to reproduce the findings featured in our investigation on liver transplants, which is described in detail in our methodology. It also contains code to reproduce the findings of a subsequent investigation on racial inequities in the liver transplant system, described in detail in a separate methodology.

For the sake of convenience, we've already commited the outputs of our analyses, which you can view in the 02_analysis/output folder. If you want to rerun the analysis and verify the steps yourself, see the Run analysis section.

Data

In this repository the data used as the input for analysis is stored in the 01_data folder, along with the code used to clean the raw data. We've included the raw data as well as our cleaned data, found in 01_data/raw and 01_data/clean, respectively. Some files will need to be uncompressed first. To do that you can run:

find . -name '*.gz' -print0 | xargs -0 gunzip --keep

Within the 01_data/raw folder, you can find the following raw data sources:

Data source	Description
00_misc/cb_2021_us_state_500k	A folder of geographic datasets from the U.S. Census Bureau defining the boundaries of states and territories.
00_misc/state_code_mapping_20221022.csv	A crosswalk we created outlining each state's name, FIPS, and Census Bureau region.
01_hospitals/22F138_Institutions.csv	A data table with information on name, location (city, state, zip), provider ID, and type of each center (hospital, lab, OPO) involved with transplants. Referred to as "institution data" in the methodology.
01_hospitals/DONORHOSPITAL_22F168Records(1of3)-redacted.csv	A data table identifying which donors were associated with which hospital, lab, and/or OPO. Referred to as “donor-hospital crosswalk” in the methodology.
01_hospitals/DONORHOSPITAL_22F168Records(2of3)-redacted.csv	A data table containing the latitude and longitude of each facility. Referred to as “facility coordinate data” in the methodology.
02_candidate/Updated_CAN_LIIN_22F218Records(1of2)-redacted.csv	A dataset of information on candidates on the waitlists for liver and intestine transplant. Referred to as the "candidate data" in the methodology.
03_donor/22F123_Records_Redacted.csv	A dataset of information on liver donors. Referred to as the "donor data" in the methodology.
04_transplant/TX_LI_22F218Records(2of2)-redacted.csv	A dataset on liver transplants. Referred to as the "transplant data" in the methodology.
05_cdc_wonder/causes_of_death	A folder of data files from the CDC Wonder database on major causes of death.
05_cdc_wonder/liver_disease_rates	A folder of data files from the CDC Wonder database on the prevelance of chronic liver disease and cirrhosis.
06_don_disp/DONDISPOSITION_22F167Records-redacted.csv	A dataset of information on organ donor disposition. Referred to as "donor disposition data" in the methodology.
07_optn_tx_counts/OPTN-Transplants_in_the_U.S._by_State-20230216.csv	A data table from OPTN on the number of liver transplants performed. Downloaded on 02/16/2023.

Within the 01_data/clean folder, you can find the following clean data sources (produced with 01_master_organs_data_clean.R):

Data source	Description
all_candidate_clean.rds	A cleaned version of the candidate data, including candidates under 18 years old.
candidate_clean.rds	A cleaned version of the candidate data, excluding children. The primary "candidate data" used for analysis.
cdc_wonder_clean.rds	A cleaned version of the data from raw/05_cdc_wonder/liver_disease_rates.
don_disp_clean.rds	A cleaned version of the donor disposition data.
donor_clean.rds	A cleaned version of the donor data.
optn_tx_counts_clean.rds	A cleaned version of the data table from raw/07_optn_tx_counts.
transplant_clean.rds	A cleaned version of the transplant data.

Setup

To run the analysis

• Install the following packages in RStudio using install.packages:
  • tidyverse
  • tidylog
  • janitor
  • lubridate
  • here
  • patchwork
  • gridExtra
• To run the Jupyter notebooks with R Kernel, use the IRkernel package with IRkernel::installspec() (see here for more)

Run analysis

You can find our analysis within the 02_analysis folder. Each analysis is contained with in its own Jupyter notebook, which you can review and run to replicate our analysis.

These notebooks include:

Notebook	Description
01_transplants_by_state.ipynb	An analysis of the state-level trends in transplants performed per year in each state.
02_per_capita_rates.ipynb	An analysis of the per-capita rates in each state of chronic liver disease deaths, liver transplants, liver donations, and additions to the liver waitlist.
03_liver_import_export_flow.ipynb	An analysis of the flow of livers imported and exported between states.
04_donor_candidate_distance.ipynb	An analysis of the distance traveled by livers from donor to candidate.
05_discards.ipynb	An analysis of discarded livers.
06_donor_cause_of_death.ipynb	An analysis of the major causes of death among liver donors and in the United States broadly.
07_comp_tx_chg_across_periods.ipynb	An analysis comparing the number of transplants performed based on variable time period selection.
08_covid.ipynb	An analysis of the impacts of COVID-19 on the number of transplants performed nationally.
09_racial_discrepancies.ipynb	An analysis, by racial group and state, of how deaths from liver disease compare to transplant and transplant waitlisting (added with second investigation).
10_ihs_unmet_need_reports.ipynb	An analysis of deferrals for liver-related care by the Indian Health Service (added with second investigation).

Each notebook will generate analysis outputs (figures and tables), which can be found within 02_analysis/output.

Update log

• 2024-02-08: For our second investigation, into racial inequities in the transplant system, we produced two additional analysis notebooks, 09_racial_discrepancies and 10_ihs_unmet_need_reports.ipynb, described in the table above. These were added to this repo in the existing 02_analysis folder. We also added data obtained from IHS regarding unmet service requests, which are analyzed in the 10_ihs_unmet_need_reports.ipynb notebook - see the 01_data/raw/08_ihs_unmet_need folder. For more, see this pull request.