# PlasticList > PlasticList tested foods and other products for 18 plastic chemicals (11 phthalates, 4 phthalate substitutes, and 3 bisphenols) at an independent, accredited lab. All results are public under CC BY 4.0. This file is the guide to the data: read it before you analyze the CSV files. The results table on the home page loads with JavaScript. To analyze the results, use the CSV files below. ## Data Every value in the files is a plain number, and a `status` column says what each number means. - [results.csv](https://www.plasticlist.org/data/results.csv): one row per sample and chemical (about 11,700 rows), in ng/g, ng/serving, and percent of EPA and EFSA daily intake limits. - [samples.csv](https://www.plasticlist.org/data/samples.csv): one row per sample (about 620): serving size, where and when we collected it, and how we tested it. - [products.csv](https://www.plasticlist.org/data/products.csv): one row per product (about 240), with tags, serving size, and product page URL. - [chemicals.csv](https://www.plasticlist.org/data/chemicals.csv): the 18 chemicals and DEHP equivalents, with the EPA and EFSA intake limits, their basis, and their sources. - [chemicals.md](https://www.plasticlist.org/data/chemicals.md): a profile of each chemical: what it is, our main findings, and its health effects. Join the files on `sample_id`, `product_id`, and `chemical`. Most products have several samples, so group by `product_id` to compare products. Please set the User-Agent header to the name of your agent or tool when you download the files, for example `claude-code` or `chatgpt`. This shows us how agents use the data. The files are free to use either way. ## Pages - [Report](https://www.plasticlist.org/report): what we found, the intake limits we used, and our view of them. - [Methodology](https://www.plasticlist.org/methodology): how we collected, blinded, shipped, and tested the samples, and how we treated results below the limit of quantification. - [Product pages](https://www.plasticlist.org/product/1): one page for each product, at /product/. - [Advice](https://www.plasticlist.org/advice): practical advice to reduce exposure. - [Run your own tests](https://www.plasticlist.org/diy): how to test products yourself. - [Team](https://www.plasticlist.org/team) ## results.csv | Column | Meaning | | :-- | :-- | | `sample_id`, `product_id`, `product` | The sample and its product. | | `chemical` | Chemical code, such as `DEHP` or `BPA`. See chemicals.csv. | | `chemical_group` | `Phthalates`, `Phthalate substitutes`, `Bisphenols`, or `Calculated` (only `DEHP_equivalents`). | | `measured` | `true` for the 18 chemicals the lab measured. `false` for `DEHP_equivalents`, a calculated total (see below). | | `status` | What the result means. See the next table. | | `ng_g` | Nanograms of the chemical per gram of sample. | | `ng_serving` | Nanograms in one serving: `ng_g × serving_size_g` of the sample. | | `loq_ng_g`, `loq_ng_serving` | The limit of quantification (LOQ), when the result is below it. | | `pct_epa_limit_adult`, `pct_epa_limit_toddler` | One serving as a percent of the EPA daily intake limit, for a 70 kg adult or a 14 kg toddler. | | `pct_efsa_limit_adult`, `pct_efsa_limit_toddler` | The same, for the EFSA daily intake limit. | | `status` | Meaning | Filled columns | | :-- | :-- | :-- | | `detected` | The lab measured the chemical. | `ng_g`, `ng_serving`, and the `pct_` columns | | `below_loq` | Not detected above the LOQ. The true value is between 0 and the LOQ. | `loq_ng_g`, `loq_ng_serving` | | `above_range` | Above the highest level the lab measured. | `ng_g`, `ng_serving`, and the `pct_` columns, as lower bounds | | `no_result` | The lab has no result for this sample. | none | ### Results below the LOQ Most results (about 84%) are `below_loq`. Do not treat them as zero or as missing without a reason. "Detected" in the report means `detected` or `above_range`. For product averages, we used this rule for each product and chemical: 1. If no sample of the product has the chemical detected, the average is 0. 2. If at least one sample has it detected, use LOQ / 2 for each `below_loq` sample, and the measured value for the other samples. 3. Average over the samples. Use `ng_serving` and `loq_ng_serving` for amounts per serving, or `ng_g` and `loq_ng_g` for concentrations. ### Results above the range `above_range` values are lower bounds: the true value is at least that number. There are only a few, but some are the highest results in the data. If you add them up or rank them, say that the result is a lower bound. ### Percent of intake limits A `pct_` column is empty when the agency has no limit for the chemical, or when the result is below the LOQ. The percent is `ng_serving / (limit × body weight) × 100`, with the limits in chemicals.csv. A value above 100 means that one serving is above the daily limit. The EFSA limit for BPA, 0.2 ng per kg of body weight per day, is not a typo. EFSA set it in 2023. For a 70 kg adult, the limit is 14 ng per day, which is below the LOQ of most BPA tests. So a `below_loq` BPA result does not show if the serving was below or above this limit. ### DEHP equivalents `DEHP_equivalents` is the total `1 × DEHP + 5 × DBP + 0.1 × BBP + 0.3 × DINP`, for the EFSA group limit. These four phthalates have cumulative effects on the reproductive system. The total uses only the chemicals detected in the sample. Its `chemical_group` is `Calculated` and `measured` is `false`, so that a sum of the `Phthalates` group does not count these four phthalates twice. ## samples.csv | Column | Meaning | | :-- | :-- | | `sample_id`, `product_id`, `product` | The sample and its product. | | `serving_size_g` | Serving size in grams. Usually the same as the product's, but not always. | | `triplicate_group` | Samples tested in triplicate share this value. Empty for other samples. | | `collected_on`, `collected_at`, `collection_notes` | When and where we bought or collected the sample, and how. | | `lot_no`, `manufacturing_date`, `expiration_date` | From the package, when available. | | `blinded_name`, `blinded_photo_url` | The generic name and photo the lab saw. We didn't share brand names with the lab. | | `shipped_on`, `shipped_in`, `shipment_type`, `arrived_at_lab_on` | Shipping details. | | `analysis_method_phthalates`, `analysis_method_bisphenols` | Lab method: QuEChERS extraction for almost all samples, or heated migration for receipts. | ## products.csv | Column | Meaning | | :-- | :-- | | `product_id`, `product` | The product. | | `tags` | Comma-separated tags, such as `baby_food`, `glass`, or `organic`. The `vintage` tag marks historical products, such as food from 1952 military rations. Exclude them for questions about food that people can buy today. | | `serving_size_g`, `serving_size_unit` | The usual serving, such as 240 g and `1 cup`. | | `sample_count` | The number of samples we tested. | | `url`, `photo_url` | The product page and photo on this site. | ## chemicals.csv One row for each of the 18 chemicals and for `DEHP_equivalents`: name, group, and the intake limits we used. The EPA and EFSA limits are in ng per kg of body weight per day, with the basis and source of each. An empty limit means the agency did not set one. `profile_url` links to the profiles in chemicals.md. - Phthalates (11): DEHP, DBP, BBP, DINP, DIDP, DEP, DMP, DIBP, DNHP, DCHP, DNOP - Phthalate substitutes (4): DEHT, DEHA, DINCH, DIDA - Bisphenols (3): BPA, BPS, BPF ## Example ```python import pandas as pd results = pd.read_csv( "https://www.plasticlist.org/data/results.csv", # Please name your tool, so that we can see how the data is used. storage_options={"User-Agent": "your-agent-name"}, ) # Detected BPA, as a percent of the EFSA limit for an adult, highest first bpa = results[(results.chemical == "BPA") & (results.status == "detected")] print(bpa.sort_values("pct_efsa_limit_adult", ascending=False) [["product", "ng_serving", "pct_efsa_limit_adult"]].head(10)) # Average DEHP per serving for each product, with the rule above dehp = results[results.chemical == "DEHP"].copy() dehp["found"] = dehp.status.isin(["detected", "above_range"]) dehp["value"] = dehp.ng_serving.where(dehp.found, dehp.loq_ng_serving / 2) any_found = dehp.groupby("product_id").found.transform("any") dehp.loc[~any_found, "value"] = 0 average = dehp.groupby(["product_id", "product"]).value.mean() print(average.sort_values(ascending=False).head(10)) ``` ## License and citation PlasticList data is licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). You may freely share and adapt it with attribution. ``` PlasticList. 'Data on Plastic Chemicals in Bay Area Foods'. plasticlist.org. Accessed . ``` ```bibtex @misc{PlasticListBayArea2024, title = "Data on Plastic Chemicals in Bay Area Foods", author = {{PlasticList}}, year = 2024, url = {https://plasticlist.org}, note = "Accessed " } ```