What the record shows of the foam.
+ The questionWhat happened inside the recalled machines, as the records show it?
A foam to quiet the blower
The foam in the recalled machines had a mechanical job: it was there to reduce sound and vibration, in the words of FDA's recall page (opening). The page names it polyester-based polyurethane foam, PE-PUR. Philips Respironics, the firm whose plant FDA inspected, places such foams "in the air inlet path" of its Trilogy 200 ventilators, in reports that FDA's Form 483 quotes. The foam therefore sat in the air that the machine delivered, and a foam offers that air a large area of thin cell walls for its weight (Chapter 5). Chapter 1's drawing shows the same part in the running device, and Chapter 19's plate of the air path shows what such a part can add to the air.
Philips Respironics announced the recall on 14 June 2021. It covered various models of the firm's ventilators and of its CPAP and BiPAP sleep therapy machines made between 2009 and 26 April 2021, and FDA's page counts about 15 million devices worldwide. FDA identified the recall as Class I in July 2021. About 10.3 million of the CPAP and BiPAP machines were in the United States, in FDA's recall database. The company's notice said that the foam "may degrade into particles which may enter the device's air pathway" and "may off-gas certain chemicals", with "possible toxic and carcinogenic effects". FDA posted the notice as a public service and did not endorse it.
Heat, humidity and ozone, each in its own source
Several records name what broke the foam down, each in its own words. The firm's own documents of 2016 to 2019, which FDA's form quotes, name heat and humidity (opening). Chapter 5 sets out the chemistry: water cuts the ester links of a polyester polyurethane, and heat speeds the reaction. FDA's form describes an email among the firm's staff dated 24 August 2018. The email states that testing confirmed that the affected foam breaks down in high heat and high humidity environments, in agreement with complaints about Trilogy ventilators. The same email records the firm's decision not to change the design and to keep the polyester polyurethane foam in its Trilogy platform. FDA's recall page names hot and humid conditions as well.
The word ozone appears nowhere in the Form 483, and in the foam's record it enters with the recall. Philips's notice of 14 June 2021 named ozone before heat and humidity, in the company's words: "The foam degradation may be exacerbated by use of unapproved cleaning methods, such as ozone," and high heat and high humidity "may also contribute". A Philips page of February 2022 cites an earlier FDA safety communication, of 27 February 2020, on the risks of cleaning CPAP machines with ozone and UV light. FDA's order of 10 March 2022 states that ozone cleaners "may exacerbate the breakdown of the foam" and requires Philips to maintain "prominently displayed information on the risk of using ozone cleaners". FDA issued the order under section 518(a) of the Federal Food, Drug, and Cosmetic Act. FDA's recall page says that the breakdown "may be worsened by using ozone cleaners or other cleaning methods not recommended by the manufacturer".
The company's own figures put a number on ozone's part. In Philips's results of 16 May 2023, devices whose users reported cleaning them with ozone were "14 times" more likely to show significant visible degradation of the foam than devices without. The ratio rests on cleaning histories that users reported themselves, and the release does not give the numbers of devices behind it. The ratio therefore compares likelihoods and gives no share of devices. The same release reports that, in the company's laboratory, diethylene glycol, one of the foam's building blocks (Chapter 5), became detectable among the emitted VOCs after 200 ozone cleaning cycles, in tests of up to 500.
The record, in order
Three years before the recall, the firm was already dealing with the foam's failure in the field. An investigation opened on 12 April 2018 concerned Trilogy units returned from the field whose foam "was degrading, and getting into the motor/air path, causing at least 1 Trilogy unit to fail". The investigation closed in June without becoming a formal corrective action. Its correction, a preventive maintenance procedure for Trilogy ventilators, went out as a field communication dated 12 June 2018. FDA observed that no verification of the correction's effectiveness was performed, and that such changes to maintenance and servicing were not adequately verified, reviewed or validated before implementation. FDA also observed that the correction, made to reduce a risk to health, was not reported in writing to FDA.
In 2021 the recall, a wrong foam and FDA's inspection ran side by side. Philips told FDA in April of its intent to conduct a field action and announced the recall on 14 June. On 15 April an incorrect polyester polyurethane foam, not the one specified, had reached the production line of Trilogy Evo ventilators from the raw foam supplier. FDA's form records the episode under purchasing controls: a test report dated 9 August documents failed VOC and ISO 18562 testing of related Trilogy Evo ventilators, and a correction and removal was approved on 15 September. Chapter 23 follows the episode as a supplier case. FDA's investigator inspected the Murrysville plant from 26 August to 9 November and issued the Form 483 on the last day. The form lists observations and states that they are not a final Agency determination (opening).
FDA then acted by order, and later through the courts. The order of 10 March 2022 found that the devices "present an unreasonable risk of substantial harm to the public health" and required Philips to notify health professionals and others within 45 days. On 2 May 2022 FDA's device center proposed a second order under section 518(b), for a plan to repair, replace or refund the devices. On 4 April 2024 the Department of Justice filed a complaint on behalf of FDA, alleging that the devices were adulterated and misbranded. On 9 April 2024 the US District Court for the Western District of Pennsylvania entered a consent decree of permanent injunction, whose terms Chapter 26 sets out.
What the testing showed, and what it left open
Philips tested the devices under the standards of Chapter 19: VOCs under ISO 18562-3 and particles under ISO 18562-2. In its results of 16 May 2023 the company concluded that exposure to VOCs and to particles is "unlikely to result in an appreciable harm to health in patients". For particles it reached that conclusion even on the assumption that all the foam degrades and reaches the patient. The conclusions cover the first-generation DreamStation, System One and DreamStation Go devices, which make up about 95 % of the registered devices worldwide in the company's figure. They cover only devices that had not been cleaned with ozone. The program ran in five testing laboratories, which the company describes as independent and certified.
On 5 October 2023 FDA stated that it did not believe the testing and analysis Philips had shared to date were adequate "to fully evaluate the risks", that additional testing was necessary, and that Philips had agreed to conduct it. The statement names no test, condition or standard behind that judgment. Philips's investor FAQ of 3 November 2026 states that the additional testing "has now been completed and confirms the validity of the earlier findings", and it gives no date for its completion. FDA's foam-testing page still says that the testing is underway and that FDA will provide an update if new, significant information becomes available. The page is current as of 15 August 2024, and it does not confirm the company's account.
The replacement foam is silicone: Philips's FAQ states that the repaired and replacement devices "all contain silicone sound abatement foam". The silicone foam has a record of its own. The opening records the firm's test report of 24 August 2021, in which an A Series CPAP device with silicone foam failed VOC testing. FDA's foam-testing page tells of a "similar device marketed outside the U.S., which failed one safety test for the release of VOCs". The page states that "devices authorized for marketing in the U.S. had demonstrated acceptable results", and notes that patients had already received devices with silicone foam in the repair and replace program. FDA requested that Philips retain an independent laboratory to test the silicone foam further, and it recommends that patients with a repaired or replaced device continue to use it. As of 13 November 2026 no result of that independent testing was posted on FDA's page or reported in Philips's FAQ.
The firm's laboratory reports of January 2021 judged the degraded foam or its extract mutagenic and cytotoxic (opening). The firm's risk assessments called cancer a possible outcome: "Potential cytotoxicity and genotoxicity leading to carcinogenicity are possible outcomes". FDA's form states in one sentence that the foam "was later found to be mutagenic, cytotoxic, carcinogenic, and non-biocompatible", but it lists no test of carcinogenicity among the firm's reports. A positive mutagenicity test identifies a hazard, and whether the hazard harms a patient depends on the route and the dose (Chapter 14).
Where the chain did not hold
Each link of this book's chain, from material to response, can be checked against what the record shows the firm knew. The first link is the material chosen for its environment. FDA's form records an email of August 2018 among the firm's staff. In it the firm kept the polyester foam in the Trilogy platform after the testing it reports had confirmed that the foam breaks down in high heat and high humidity. Chapter 5 shows why a polyester breaks down in warm, humid air, and why this book gives the running device a polyether foam, which has a weakness to oxidants of its own.
The second link is aging in the conditions of use, and the record shows that the degradation was found first in use. The earliest of the firm's reports examined Trilogy foam returned from the field (opening). For the sleep therapy machines, a risk assessment that FDA's form quotes records that the firm "was made aware in May 2019 that four CPAP units were returned to a service center with degraded sound abatement foam". The company later aged the foam in the laboratory, and in its own account that aging is not intended to predict the rate of degradation in used devices (Chapter 5).
The third link is the route to the patient. Chapter 19 showed that the firm's 2018 health hazard evaluation for Trilogy counted on a humidifier and a bacterial filter that patients need not use, and on a healthy lung. Yet the ventilators' intended use, as the form quotes it, is the care of "individuals who require mechanical ventilation". The same evaluation assigned particles to the airway's defenses by size, in its own wording. Those "greater than 5 microns" would be removed by the cilia and mucosal lining "through inertial impaction", those of "1-5 microns" "through sedimentation", and those "less than 1 micron" would be left to the alveolar macrophages. The bands are the firm's, not a standard's, and Chapter 9 sets out deposition by size from its own sources.
The fourth link is cleaning. Philips's release of May 2023 found ozone-cleaned devices more likely to show visible degradation, yet it drew its conclusions on VOCs and particles from devices not exposed to ozone cleaning. FDA's order of March 2022 required the ozone risk to be displayed prominently. In this book's chain a cleaning that the maker does not approve still belongs to the device's use, which is why Chapter 19 cleans an aged sample before testing it.
- The figures. In Philips's laboratory, diethylene glycol from the foam became detectable among the emitted VOCs after 200 ozone cleaning cycles, in tests of up to 500 cycles (the company's figures, 16 May 2023).
- Once a night. 200 ÷ 365 = 0.55 years, about 6.6 months; 500 ÷ 365 = 1.4 years, about 16 months.
- Once a week. 200 ÷ 52 = 3.8 years; 500 ÷ 52 = 9.6 years.
- The result. A user who cleans with ozone every night passes the laboratory's 200 cycles within seven months, and one who cleans weekly within four years. A cleaning habit, not the laboratory's choice of cycles, decides when a device reaches the condition at which the breakdown product appeared.
- Assumptions. The two frequencies are this book's assumptions, not data, and one laboratory cycle is taken as one home cleaning, which the release does not establish.
The fifth link is the signal from the field, and each count in the record measures something different. FDA's form records that a query of the firm's complaints from 2008 onward, for words such as foam, particles, debris and black, returned over 222,000 complaints. That figure counts keyword matches, not confirmed cases. An analysis of April 2021 confirmed 1,254 complaints related to foam degradation from 2014 to April 2021, and the form counts 110 received from 2014 to 2017. The firm's 2018 evaluation counted 17 instances, and FDA observed that the firm's assessments of that year did not "accurately reflect the known data at that time".
Reports to FDA form a separate count. From 2011 to April 2021, before the recall, Philips submitted 30 reports to FDA that it linked to the foam's breakdown. From April 2021 to 30 September 2023 FDA received more than 116,000 reports about the recalled devices, including 561 reports of death. FDA states that "The incidence, prevalence, or cause of an event cannot typically be determined from this reporting system alone".
The opening ended on what the firm's files already held, and the record shows three steps of 2018 that drew on those files. An evaluation counted on a humidifier, a bacterial filter and a healthy lung that its patients need not have, a decision kept the polyester foam in the Trilogy platform, and a correction was neither verified nor reported in writing to FDA. The opening named the step that the firm's test results needed, a decision about the device's risk. Part VII follows that step into the design file: the risk chain of Chapter 22 and the control of a changed material in Chapter 23.
For each gas-pathway material, write down the conditions of use it must survive, including the cleaning methods users adopt whether or not the maker approves them, and test aged and cleaned samples under those conditions (Chapter 19). Send every complaint that names the material's failure into the risk analysis, and record each count with what it counts: keyword matches, confirmed cases or reports to a regulator. Treat a replacement material as a new material, tested to the same standards before it ships, and carry each result, pass or fail, into a documented risk decision (Chapter 22). Report a correction made to reduce a risk to health as the regulator requires.
- The recalled machines' polyester polyurethane foam sat in the air path to quiet the blower, and the records name heat, humidity and ozone cleaning as conditions that may worsen its breakdown, each in its own source's words.
- The record dates each step of the case: the field correction of June 2018, the recall of June 2021, the Form 483 of November 2021, FDA's order of March 2022 and its proposed second order of May 2022, and the consent decree of April 2024.
- Philips concluded in 2023 that its devices' VOC and particle emissions were unlikely to cause appreciable harm, FDA judged that testing not adequate to evaluate the risks fully, and no result of the further testing of the silicone replacement foam was posted as of 13 November 2026.
- The record points at five weak links: a polyester kept in warm, humid air; degradation found in use; a 2018 evaluation that counted on a humidifier, a bacterial filter and a healthy lung its patients need not have; cleaning left out of the testing; and complaint counts the 2018 assessments did not reflect accurately, as FDA observed.
Let's Talk