+ Part VI · Breathing gas

Breathing gas, and the recall.

A breathing machine sends the patient's air past its materials every night. On its way to the mask, room air passes a filter, a blower lined with sound-damping foam, a heated water chamber and a hose, and each of them can add particles or gases to air that a patient may breathe for years. Chapter 19 sets out how that air is judged: under the breathing-gas standard, ISO 18562, a laboratory measures what the device adds and compares it with limits built from the dose arithmetic of Part V. Chapter 20 follows the case this book is built around, the 2021 recall of sleep and breathing machines whose sound-damping foam could break down. It traces the case through the records of Philips and of FDA, and marks the links in Philips's evaluation that the record shows did not hold.

19 — What the patient breathes

What the patient breathes.

+ The questionHow are a breathing device's air paths judged?

Every part upstream adds to the air

Part V judged what leaves a material by extracting it in a liquid. Air is not a solvent: it takes from a plastic only what evaporates, and it carries away pieces that break off (Chapter 4). The parts from intake to mask form the gas pathway of Chapter 1, and a substance released by any of them reaches the patient's airway unless a part downstream removes it. These parts are therefore judged together, by what the assembled device adds to the gas it delivers.

ISO 18562 sets out that evaluation in four parts. Part 1 places it within a risk management process, part 2 tests the gas for particles, part 3 for volatile organic compounds, and part 4 for leachables in condensate, the water that collects in the path. The four parts first appeared in 2017 and reached a second edition in March 2024. FDA recognizes parts 2 and 3 of the 2024 edition in full and part 1 only in part. These recognitions sit on a list that FDA's notice in the Federal Register made applicable on 24 June 2024, and the same notice replaced FDA's recognitions of the 2017 parts.

Part 3's scope ends at the gas, as FDA's recognition record gives it. A surface that touches the patient, such as a mask cushion on the face, stays under ISO 10993, and contamination already in the gas supplied to the device lies outside part 3. The same record states that part 3 applies over the device's expected lifetime in normal use, including the processing intended for the device.

Each part of the running device's path can add something of its own. The foam beside the blower is a polyether polyurethane, this book's choice in Chapter 5, and as it ages it can shed particles and release breakdown products. The hose is a thermoplastic elastomer, chosen in Chapter 3, and it can release volatile substances (Chapter 4). The water chamber, of polycarbonate by the same choice, holds heated water that can carry what it draws from the plastic into drops in the hose and mask. Each addition goes to the part of the series that tests it.

46 — What each part of the air path can add, and the test that catches it
ROOM AIR what the room air already carries: outside the standard INTAKE AND FILTER BLOWER AND FOAM WATER CHAMBER HOSE MASK particles and breakdown products, as the foam ages condensate can carry what heated water draws from the plastic volatile substances cushion on the face: under ISO 10993, not this standard the patient's airway PART 2 — particles PART 3 — volatile organic compounds PART 4 — leachables in condensate every addition upstream reaches the airway unless a part downstream removes it Schematic. Scope: FDA's recognition record for ISO 18562-3 (contact surfaces under ISO 10993; contamination in the supplied gas out of scope). Parts of the series: test laboratories' accounts.
Device materialReleased substanceGas or fluid pathFocal detail
Plate 46 — Whatever a part upstream adds reaches the airway unless a part downstream removes it, so ISO 18562 tests the gas the assembled device delivers: particles in part 2, volatile organic compounds in part 3 and leachables in condensate in part 4.

Particles are limited by size

Particles are judged by size, because size decides where they go. The airways sort inhaled particles by their aerodynamic diameter, and the smaller ones reach deeper (Chapter 9). Part 2 of ISO 18562 therefore measures two classes named for such sizes. PM10 is the mass of particles with a diameter of 10 µm or less in a cubic meter of gas, and PM2.5 the mass of those of 2.5 µm or less. The standard sets its limits by particle size, not by chemistry, following the approach of the US Environmental Protection Agency, as FDA's recognition record for the 2024 part 2 states.

Three companies give the limits as 12 µg/m³ for PM2.5 and 150 µg/m³ for PM10. Element, a testing laboratory, gives them for the 2017 edition, and Saavalainen of Measurlabs, another laboratory, gives them in 2026 for the 2024 edition. Keller of IMT, a device development company, names no edition. FDA's record prints no value, so the figures are these companies' reading of the standard. The record does give the range of sizes counted, from 0.25 µm to 10 µm, and a reason for each end. Larger particles deposit in the nasal cavity, and smaller ones are left out because "Insufficient data exist to establish exposure limits for particles less than 0,25 µm diameter".

To measure particles, the laboratory passes the gas that leaves the device through filters that are weighed, through a particle counter, or both. Element and Keller describe the test this way, and so does Turner, who writes for the testing laboratory Medical Engineering Technologies and gives a filtration at 0.2 µm. The result is a mass per cubic meter for each size class. A particle limit is already a concentration in breathed air, so the measured value is compared with it directly.

Gases are judged by the dose they deliver

Volatile organic compounds, or VOCs, are carbon compounds small enough to evaporate at the temperature of use, the molecules that air takes from a plastic (Chapter 4). Part 3 measures them in the delivered gas while the device runs in a test chamber, at settings that the chapter's last section explains. The gas is typically drawn through sampling tubes at 20 to 100 mL a minute, as Aichner, of the laboratory group Eurofins, wrote in 2018. Heat then releases what the tubes trapped into a gas chromatograph and a mass spectrometer, which report each compound's concentration in µg/m³ (Chapter 13).

A concentration alone does not show harm, because harm follows the dose (Chapter 14). What a patient takes in of a breathed compound depends on how much air carries it in. The amount breathed in a day is the concentration multiplied by the volume of air breathed that day, and the tolerable exposure is the tolerable intake multiplied by the patient's body mass, in Aichner's account of the 2017 edition. Dividing the tolerable exposure by the daily volume therefore gives the tolerable air concentration: the most of one compound that each cubic meter of breathed air may carry. For a compound with no data, the threshold of toxicological concern of Chapter 16 takes the place of the tolerable exposure.

47 — From a tolerable intake to a tolerable air concentration
ON ISO 18562's 2017 DEFAULTS ADULT — 70 kg, 20 m³ of air a day tolerable intake: 1 µg per kg per day × 70 kg tolerable exposure: 70 µg a day ÷ 20 m³ of air a day tolerable air concentration: 3.5 µg/m³ INFANT — 3.5 kg, 2.0 m³ of air a day tolerable intake: 1 µg per kg per day × 3.5 kg tolerable exposure: 3.5 µg a day ÷ 2.0 m³ of air a day tolerable air concentration: 1.75 µg/m³ air breathed a day per kg of body mass: adult 0.29 m³ infant 0.57 m³ twice the air per kilogram: half the air limit AN UNTESTED COMPOUND, ADULT 360 µg a day, first 24 h ÷ 20 m³ 18 µg/m³ 120 µg a day, less than 30 days ÷ 20 m³ 6 µg/m³ 40 µg a day, 30 days or more ('permanent') ÷ 20 m³ 2 µg/m³ Defaults and thresholds: ISO 18562-1 and -3, 2017 editions, as test laboratories reproduce them (Element; Aichner, Eurofins, 2018; Keller, IMT; Turner, MET, 2021). The 2024 edition changed the patient groups and the threshold.
Gas or fluid pathLimit or thresholdFocal detail
Plate 47 — A tolerable intake becomes an air limit in two steps: multiplied by body mass, then divided by the air breathed in a day. On the 2017 defaults an infant breathes twice the adult's air per kilogram, so its limit is half the adult's; an untested compound's daily threshold, spread over an adult's air, allows 2 µg/m³ in permanent use, 30 days or more.

The 2017 edition gives its thresholds of toxicological concern as daily amounts that fall as use lengthens, as the thresholds of Chapter 16 do. In the accounts of Element, Aichner and Keller, they are 360 µg a day for the first 24 hours, 120 µg a day for the prolonged use that follows, and 40 µg a day for the longest use, which the edition calls "permanent". In the accounts of Element and of Markes, an instrument maker, prolonged use lasts less than 30 days and permanent use 30 days or more. By those two accounts, the gas standard's permanent band starts on day 30, a day before the long-term contact of ISO 10993-1, which starts after 30 days (Chapter 2).

A new device's emissions also fall with time: volatile substances leave its surfaces fastest at first (Chapter 4). Because both the threshold and the emission fall, the gas is sampled more than once over the period of use. In Aichner's account, a prolonged use is sampled at the start, after 24 hours and at the end, and a permanent use at the end of day 29. Long-term sampling continues until emissions fall below 40 µg a day, in Turner's account.

Solved example: the air concentration ISO 18562's 2017 defaults allow
  • The defaults. The 2017 edition, as test laboratories reproduce it, gives an adult a body mass of 70 kg and 20 m³ of air breathed a day; the group it calls pediatric, 10 kg and 5.0 m³; and an infant, 3.5 kg and 2.0 m³. The 2024 edition's foreword states that it expanded the patient groups and changed the threshold, so these are the 2017 values only.
  • An untested compound, adult. The threshold is already a daily amount, so it is divided by the air: 360 ÷ 20 = 18 µg/m³ on the first day, 120 ÷ 20 = 6 µg/m³ for the prolonged use that follows, less than 30 days, and 40 ÷ 20 = 2 µg/m³ from day 30.
  • A compound with a tolerable intake. Each microgram per kilogram per day of tolerable intake allows 70 ÷ 20 = 3.5 µg/m³ for the adult, 10 ÷ 5.0 = 2.0 µg/m³ for the pediatric patient and 3.5 ÷ 2.0 = 1.75 µg/m³ for the infant.
  • The result. On these defaults an adult in permanent use, 30 days or more, may breathe at most 2 µg/m³ of one untested compound. In Aichner's account, a compound below that level need not be identified. The infant's limit for a compound with a known intake is half the adult's, because an infant breathes twice as much air per kilogram.
  • Assumptions. The defaults are set for groups, not measured in patients. The newborn group is left out, because the laboratories disagree on its daily air, giving 0.2, 0.21 or 0.23 m³.

The tolerable intake in this arithmetic comes from studies that gave the compound by some route of exposure (Chapter 14). Two clauses of the 2024 part 1 let a laboratory with no inhalation data use data from parenteral studies without modification. In a parenteral study the compound is given by injection or infusion. FDA's record states that the two clauses conflict with ISO 10993-17:2023 on extrapolating from one route to another, so FDA's recognition of part 1 does not extend to them.

Condensate and cleaning add to the gas in their own ways

Water adds a third kind of release. The running device's water chamber warms and moistens the air, and air that cools along the hose can leave drops in the hose and mask that carry what the water drew from the plastics. Part 4 applies "where the device may form condensate during clinical use", in Element's words. A laboratory extracts the path under simulated use and analyzes the water for organic leachables by chromatography and mass spectrometry and for elements by ICP-MS (Chapter 13). In Saavalainen's account, organic compounds are then compared with limits derived as in part 1, and elements with ICH Q3D, the international guideline on elemental impurities in medicines.

How much condensate reaches the patient sets how much of each leachable the patient receives. Under the 2017 edition a laboratory that cannot establish the volume may assume 1 mL a day, in Turner's account. The 2024 edition instead asks for the volume that can reach the patient in 24 hours under worst-case conditions, in the accounts of Saavalainen and of two staff of the testing firm NAMSA, Smiraldo and Ruch. The NAMSA authors write that the default "can no longer be assumed". With a heated water chamber, that volume depends on the device's design and settings, so it is found at the worst case of both.

Cleaning can put into the path a gas that no material released. Ozone cleaners are sold for these machines, and Chapter 7 sets out what FDA has said of them. Ozone is also an oxidant, so a cleaner can change the material whose emissions parts 2 and 3 measured: this book's polyether foam is vulnerable to oxidation (Chapter 5). A cleaner that the maker does not recommend lies outside the processing intended for the device, and Chapter 20 shows what the record says of such cleaners and the recalled foam.

Each test has its own worst case

A device passes these tests only under the conditions in which it was tested, so each test runs under the conditions that make its result worst, and the two gas tests run at opposite flows. In Saavalainen's account, VOCs are sampled at the lowest clinically relevant flow, because less air dilutes the same emission into a higher concentration, and at the highest foreseeable use temperature, because heat raises emission. Particles are counted at the maximum clinically relevant flow, which Keller gives as the particle test's worst case. One set-up therefore cannot serve both tests.

48 — Two tests, two opposite flow settings
PARTICLES — part 2 VOLATILE ORGANIC COMPOUNDS — part 3 lowest maximum weighed filters particle counter maximum clinically relevant flow: the particle test's worst case sizes counted: 0.25 to 10 µm result: µg/m³ of PM2.5 and PM10 lowest maximum highest foreseeable use temperature thermal desorption, GC-MS sampling tube, typically 20 to 100 mL a minute lowest clinically relevant flow: less air dilutes the same emission result: µg/m³ of each compound each test at its own worst case: opposite flows both: new devices and devices aged and cleaned as the instructions set Schematic. Settings as test laboratories state them: maximum flow for particles (Keller, IMT); lowest flow and highest temperature for VOCs (Saavalainen, Measurlabs, 2026); sampling rate (Aichner, Eurofins, 2018). Size range: FDA's recognition record for ISO 18562-2.
Device materialReleased substanceGas or fluid pathFocal detail
Plate 48 — The particle and VOC tests set the device's flow at opposite ends. Particles are counted at maximum flow, the particle test's worst case; VOCs are sampled at the lowest flow and highest use temperature, where less air dilutes the emission into a higher concentration.

Age sets the second condition. Part 3 applies over the device's expected lifetime, and the NAMSA authors read the 2024 edition as placing "an emphasis on the effects of aging/life cycle of the device". An aged foam can shed particles and release breakdown products that a new foam does not: a polyester foam breaks down by hydrolysis in warm, humid air, and this book's polyether foam by oxidation, with ozone from a cleaner among the oxidants (Chapter 5). Accelerated aging models heat but neither ozone nor cleaning (Chapter 7). An aged sample therefore stands for use only after it has also been through the cleaning that the instructions set.

The patient's own use sets the third condition, because it may leave out parts that the designers counted on. FDA's Form 483 of November 2021, the inspection record that this book's opening tells, quotes Philips's 2018 health hazard evaluation for its Trilogy ventilators: "To reach the patient, particulate would have to pass through the device, through the humidifier and through the patient circuit. If a bacteria filter was in place, the particulate would not reach the patient". FDA observed that neither the humidifier nor the bacterial filter is required for use of those ventilators. FDA also observed that the evaluation described a typical and healthy lung and did not consider patients with a tracheostomy, an opening into the windpipe through the neck that bypasses the nose (Chapter 9). A worst case counts only on the parts that every patient uses.

In practice

List each part of the gas pathway, what it can add and the part of ISO 18562 that tests it. Test new devices and devices aged and cleaned as the instructions set, each test at its own worst case. For the running device, count the aged foam's particles with the water chamber empty. Because the running device's users weigh over 30 kg, and the lightest falls between two patient groups, set the limits of compounds with a tolerable intake on the stricter of the two (Chapter 1). Record the edition behind each limit, and label the 2017 defaults, where used, as those the 2024 edition replaced.

+ What this chapter established
  • ISO 18562 judges the gas the assembled device delivers: particles in part 2, volatile organic compounds in part 3 and leachables in condensate in part 4.
  • Particles are limited by size, at 12 µg/m³ for PM2.5 and 150 µg/m³ for PM10 in three companies' accounts.
  • A volatile compound's air limit is a tolerable daily amount divided by the air breathed a day, which on the 2017 defaults allows an adult in permanent use, 30 days or more, 2 µg/m³ of an untested compound.
  • Each test runs at its own worst case, particles at maximum flow and VOCs at the lowest flow and highest temperature, on aged devices and without counting on parts a patient may not use.
Created with Claude (Anthropic).
Educational reference material, not engineering, regulatory or clinical advice. Product and company names referenced are the property of their respective owners.