Industries / Health / Medical devices / Medical sterilization
Emissions
This work is about the EtO that leaves sterilizing plants through vents, leaks and freshly sterilized goods, how much of it reaches neighbors, how risky that is, and the rules and lawsuits that follow.
7
Milestones achieved, from 1994 to 2025.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1994
Limits on plant vents
On 6 December 1994 the US EPA set national emission standards for commercial EtO sterilizers under the Clean Air Act Amendments of 1990. By the CSB's summary, plants using 10,000 pounds or more of EtO a year had to remove 99% of it from each vent (chamber exhausts, back vents and aeration rooms), and most chose oxidizers to do it.
LessonA control device can ignite the gas it treats; after explosions involving EtO and oxidizers, the EPA suspended the back-vent rule in 1997 and dropped it in 2001, while California and some other states kept requiring back-vent controls.
Credit
US EPA
2007
Canada's 99% guideline
From 1 January 2007, Environment Canada's guidelines under the Canadian Environmental Protection Act apply to any site that uses 10 kg or more of EtO a year. Commercial sterilizers and aerators should send their exhaust to controls that remove at least 99% of the gas, hospitals should follow the CSA Z314.9 standard, and each site was to report its sterilizers and controls to the minister. The guidelines say "should" throughout; they are not binding limits.
Credit
Environment Canada, now Environment and Climate Change Canada
2016
A thirtyfold higher risk
In December 2016 the EPA's Integrated Risk Information System (IRIS) judged EtO "carcinogenic to humans" when breathed, based mainly on studies of exposed workers by Steenland and colleagues. It set an adult cancer unit risk, the extra lifetime cancer risk per µg/m³ breathed, of 3 × 10⁻³ (5 × 10⁻³ with an adjustment for exposure in childhood). The EPA's 1985 value was 1.0 × 10⁻⁴, by Olaguer and colleagues' account, so the adult value is 30 times higher.
LessonThe new value puts the one-in-10,000 risk level at 0.02 µg/m³, which laboratories did not report being able to detect by 2023, and about a tenth of the background measured upwind of a Michigan plant in 2019.
Credit
the EPA IRIS program; Steenland and colleagues
2019
Measured from the street
In August 2018 the EPA's 2014 National Air Toxics Assessment flagged Michigan's largest medical sterilizer, in Grand Rapids, as adding more than one cancer per thousand people nearby. Michigan's air agency took 24-hour air samples at 16 sites in early 2019 and worked back from the readings and the wind to the plant's emissions: 594 pounds a year, within about 41% of the plant's own estimate, which the authors judged consistent.
LessonEvery site, even upwind ones, read above Michigan's screening level, and the method put the city's background at 0.247 µg/m³, so a single monitor near a plant cannot tell the plant's share from the rest.
Credit
Eduardo P. Olaguer and colleagues, Michigan Department of Environment, Great Lakes, and Energy
2019
Illinois demands 99.9%
The Matt Haller Act, Public Act 101-0022, took effect on 21 June 2019. Within 180 days it required every EtO sterilizing plant in Illinois to capture all its EtO and cut what leaves each exhaust point by at least 99.9%, or to 0.2 parts per million, and to prove it by test. Hospitals, clinics and laboratories are excluded.
Credit
the Illinois General Assembly (Senate Bill 1852)
2022
A $363 million verdict
In September 2022 a Cook County, Illinois, jury found that EtO from Sterigenics's Willowbrook plant caused the breast cancer of Susan Kamuda, who had lived nearby for 20 years, and awarded her $363 million. It ordered Sterigenics to pay $220 million, its parent Sotera Health $100 million and Griffith Foods, which MDDI calls Sotera's corporate predecessor, $5 million. In January 2023 Sterigenics and Sotera agreed to pay $408 million to settle more than 870 claims, without admitting liability.
LessonResults split: in the next Willowbrook trial, Teresa Fornek's, a jury found Sterigenics and Sotera Health not liable in November 2022, and in November 2026 a Georgia court dismissed nine suits claiming Sterigenics's Atlanta plant lowered home values.
Credit
the Cook County jury; Susan Kamuda and her lawyers
2025
The 2024 limits put off
The EPA's rule of 5 April 2024 was to cut EtO from the 88 US commercial sterilizers by more than 90%, require continuous emission monitors at most of them and give the 28 largest users two years to comply. On 17 July 2025 President Trump's Proclamation 10959 exempted the plants listed in its annex for two years past each deadline, saying the technology to comply "does not exist in a commercially viable form" and that EtO sterilizes about 50% of sterile US devices.
LessonUntil their new deadlines in 2028 or 2029, the exempted plants follow the rules in force before 2024; Harvard Law School's tracker counts 41 of them and a lawsuit filed in January 2026 counts 40.
Credit
the US EPA; President Donald Trump
Watch for
Exemptions tested in court
Four groups claim that the technology to meet the 2024 rule, including controls for room air and continuous EtO monitors, is commercially available and already in use, as the EPA itself found, so the President exceeded his authority in exempting 40 sterilizers.
Achieved whenthe court rules on whether the technology was available and the exemption lawful.
Claimed by
CleanAIRE NC, Sustainable Newton, Virginia Interfaith Power & Light and the Natural Resources Defense Council, in a complaint filed in the US District Court for the District of Columbia on 28 January 2026
Steri-Tech held to 99.99%
On 21 September 2026 the US Department of Justice lodged a proposed consent decree under which Steri-Tech, which runs a sterilizing plant in Salinas, Puerto Rico, is to meet a 99.99% EtO removal standard and pay a $700,000 civil penalty, settling a Clean Air Act suit.
Achieved whenthe court enters the decree and a performance test at the plant shows 99.99% removal.
Claimed by
the US Department of Justice, September 2026
Open
01
Reported against measured
One open analysis that takes every public air-sampling campaign run around a US EtO sterilizer, pairs it with the emissions the plant reported for the same months, and works back from the readings and the wind to an emission estimate with error bars, using one method for all of them. The result is a table of reported against inferred emissions, plant by plant.
The briefCredit · unclaimed02
Background EtO, rechecked
A pooled reanalysis of every public ambient EtO result in the EPA's Air Quality System since 2018, grouped by laboratory, canister type, the ion the lab used to measure EtO, and how long the sample sat before analysis. It estimates how far each setup shifts the reported reading and gives a corrected background for each region, with error bars.
The briefCredit · unclaimed
Open 01 · the brief
Reported against measured
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
One open analysis that takes every public air-sampling campaign run around a US EtO sterilizer, pairs it with the emissions the plant reported for the same months, and works back from the readings and the wind to an emission estimate with error bars, using one method for all of them. The result is a table of reported against inferred emissions, plant by plant.
Why it is still open
Each campaign has been judged alone. Michigan's 2019 study inferred one plant's emissions and found them consistent with its report. A University of Utah and Utah Department of Environmental Quality study (Spooner and colleagues, 2023) compared modeled and measured EtO around a Salt Lake Valley plant and found six of eight sites significantly different, with no correction factor to be drawn. South Coast AQMD, the Los Angeles area's air district, measured EtO from a vehicle near 14 facilities and from July 2022 monitored three large sterilizers for two years, and an Aerodyne Research team reported in 2022 a monitor precise to a few parts per trillion that traced plumes back to two emitters. Searches found these and Georgia's 2020 forward modeling of a Cobb County plant, but no study that puts several campaigns through one method. Stack monitors would settle much of this, but the exempted plants need not run them until September 2028, and the EPA's March 2026 proposal would drop the requirement for larger plants.
pmc.ncbi.nlm.nih.govapi.openalex.orgepa.govamt.copernicus.orgbakerdonelson.comeelp.law.harvard.edu
Built with what exists
monitoring reports from state agencies (Michigan, Georgia, Illinois, Utah, California's air districts) and the EPA; plants' reported releases from the EPA's Toxics Release Inventory and state permit files; AERMOD, the EPA's free dispersion model; Python with SciPy, which Michigan's team used for its regression; an AI model that reads scanned agency reports and pulls their tables with page references.
First step
Rebuild Michigan's second-phase regression from the readings published in its paper and check that it returns about 594 pounds a year.
Done looks like
At least 10 campaigns, each with the reported and inferred emissions, the uncertainty, and the inputs needed to rerun it. The method reproduces Michigan's published estimate of 594 pounds a year to within 10%. Code and data sit in a public repository.
Open 02 · the brief
Background EtO, rechecked
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A pooled reanalysis of every public ambient EtO result in the EPA's Air Quality System since 2018, grouped by laboratory, canister type, the ion the lab used to measure EtO, and how long the sample sat before analysis. It estimates how far each setup shifts the reported reading and gives a corrected background for each region, with error bars.
Why it is still open
The EtO found in city air far from plants sits above the one-in-10,000 risk level, and how much of it is real is unsettled. EPA researchers reported in May 2021 that EtO in samples held in electropolished steel canisters rose 7 to 10 times over four to five weeks, while silicon-ceramic-lined canisters stayed below detection. The Interstate Technology and Regulatory Council's 2023 guide notes that laboratories use different ions to measure EtO and did not report detection limits at or below the 0.02 µg/m³ risk level. Utah's study found the summer background almost 2.5 times the winter one. West Virginia's air agency averaged the national EtO results held in the EPA's Air Quality System in a February 2023 report, at 0.122 parts per billion with results below detection left out, then redid the averages for the national trends stations alone after EPA staff questioned the first figures; it grouped nothing by laboratory or canister. Searches found that report, agency memos, a Georgia presentation and a 2022 comparison with a laser instrument, but no national reanalysis by laboratory and method.
Built with what exists
the EPA's AirData downloads and AQS API; agencies' published quality plans to fill in canister type and laboratory; Python or R with mixed-effects models; EPA Method 327 (2024) and laser readings, where agencies used them, as cross-checks.
First step
Download one year of ambient EtO results from the national air toxics stations, label each with its laboratory and canister type, and compare the spread between groups.
Done looks like
At least 20,000 samples placed in groups, with the shift for each laboratory and canister type and its error bars. The analysis shows the canister effect in the same direction as the EPA's 2021 memo, and gives a corrected background for each EPA region with the code to rerun it.