Replication data and code for Food contaminant exposure sources and screening risk across US population groups

This repository contains the replication data, analysis scripts and supplementary outputs supporting the manuscript “Food contaminant exposure sources and screening risk across US population groups” by Jiawei Xu and Hongrui Zhao. The analysis links food contaminant measurements from the US Food and Drug Administration Total Diet Study for fiscal years 2018–2020 with dietary recall and participant data from NHANES 2017–2018. It examines total arsenic, cadmium, mercury and lead across 11 selected foods and 14 food–analyte pairs. The package includes Python analysis scripts, the food code crosswalk, derived participant-level exposure data, survey-weighted population and consumer estimates, subgroup analyses, benchmark comparisons, influence diagnostics, reporting-limit sensitivity analyses, concentration bootstrap results, mean-concentration scenarios, and the candidate food inventory with screening counts. README.txt provides instructions for obtaining the public source files, arranging the project directories and reproducing the outputs. FILE_MANIFEST.csv provides file sizes and SHA-256 checksums. The estimates describe exposure from the selected foods rather than the complete diet. Consumer percentiles are based on two dietary recall days and should not be interpreted as usual long-term intake. Small consumer groups are subject to reporting suppression rules documented in the manuscript and accompanying files. Original FDA and CDC source datasets are not redistributed in this archive. Their official download sources and required filenames are provided in README.txt. The participant-level file contains derived records linked to publicly released NHANES data.