A dose the body tolerates.
+ The questionHow is a tolerable intake worked out for one compound?
A dose counts per kilogram and per day
Toxicologists start from a dose at which a study saw little or no harm, and then divide it. The division turns a result in animals into a daily dose judged tolerable for people, and the amounts that Chapters 12 and 13 measure in a device's extract are weighed against that dose. An amount cannot be judged by itself, because the same amount makes different doses. A lighter patient takes a larger dose from it, as Chapter 1's solved example shows, and an amount taken in one day makes a larger daily dose than the same amount spread over a month.
In this book a dose is the amount of a substance that enters the body each day for each kilogram of body weight. Counting per kilogram puts a rat and a person on one scale, but body mass does not capture the differences between species or between people, and uncertainty factors cover those. The unit is the microgram per kilogram per day, µg/kg/d, or the milligram per kilogram per day, mg/kg/d, which is a thousand times larger and suits the amounts fed to animals. ISO 10993-17:2023, the part of the device series on toxicological risk, states doses in µg/kg/d for effects throughout the body, damage to genes, cancer and harm to reproduction or development, according to 2024 slides by presenters from FDA, a device maker and a toxicology consultancy (Chapter 1). Irritation, an effect at the site of contact, is counted per area of skin instead, in µg/cm², and Chapter 11 follows OECD in calling such an amount a dose per area.
The point of departure comes from one study
A toxicity study gives groups of animals different daily doses, usually with one group given none, and records the harm in each group. Two doses are read from the results. The NOAEL, the no-observed-adverse-effect level, is the largest dose at which the study saw no adverse effect, and the LOAEL, the lowest-observed-adverse-effect level, is the smallest dose at which it saw one. Both can only be doses the study chose to test, so the spacing of those doses limits how closely they mark where harm begins.
A study of DEHP shows that limit. Carpenter and colleagues fed DEHP to guinea pigs at two doses for one year and published the study in 1953. At the lower dose, 19 mg/kg/d, the females already had heavier livers for their body weight, as the US Environmental Protection Agency, EPA, reports in its summary of the study. The study therefore has a LOAEL and no NOAEL.
The dose a calculation starts from is the point of departure, which ISO 10993-17:2023 defines as a "low point on a toxicological dose-response curve established from experimental or observational data". A NOAEL or a LOAEL can serve, and so can a third kind, the benchmark dose, which uses every group of the study. A curve fitted through all the groups' responses gives the dose that would cause a chosen small response, such as a 5 % change. The calculation starts from a dose below it, the benchmark dose low or BMDL, which the standard defines as the "lower one-sided confidence limit" of that dose. The benchmark dose need not be one of the doses tested, and its lower limit falls further below it when the responses scatter more widely.
EPA's draft assessment of D4 siloxane, octamethylcyclotetrasiloxane, shows a benchmark dose in use. D4 is a ring-shaped siloxane that remains in silicone polymers, elastomers among them, as Environment Canada and Health Canada noted in 2008. In Chapter 13's illustration it is the compound confirmed in the water extract of the running device's silicone cushion, and so this chapter works out its tolerable intake. EPA released the assessment in September 2025 as a draft for peer review.
WIL Research, a contract laboratory, reported the assessment's principal study in 2001. It was a two-generation study in which parent rats were exposed to D4 in air before and during mating and pregnancy, and the critical harm was fewer live pups in the second generation. EPA modeled the D4 in the rats' blood over time, set a 5 % benchmark on that internal dose, and converted its lower limit into doses for people with a model of how the body takes up and clears D4. For repeated exposure those doses are 3.60 mg/kg/d by mouth, and for contact with skin 179 mg/kg/d where the skin is covered and 326 mg/kg/d where it is not.
A compound studied many times offers many candidate points of departure, and the standard says which to take: "The most critical (e.g., the lowest) clinically relevant POD shall be selected", as the 2024 slides quote it. Across different harms the lowest governs, because one tolerable dose must protect against all of them. For a single harm studied several times, the slides allow the highest of the NOAELs. The choice of study is a judgment, and it moves the result: for DEHP, EPA and EFSA, the EU's food safety authority, started from different studies of different harms (solved example below).
Uncertainty factors cover what the study could not see
A point of departure describes animals in a study, and the patient differs from them in ways the study did not measure. ISO 10993-17:2023 covers each difference with an uncertainty factor, a number by which the point of departure is divided, and the 2024 slides give the edition's defaults. The first two cover the step from animals to people. A factor of 10 from animal to human applies "unless otherwise justified", because people may be more sensitive than the test species. A factor of 10 from person to person allows for the people who, through age, illness or the way their bodies handle a substance, respond to doses that most people tolerate. Together the two make 100. For infants of 6 months or younger, the slides list a further factor of 3 with the animal-to-human factor. Efor, a consultancy, ties a factor of 3 to a different group: children exposed to substances that leave the body mainly through the liver or kidneys.
The slides give defaults for three more factors. A short study used for a long exposure adds 6 when it is subchronic, lasting a few months, and 10 when it is subacute, lasting a few weeks. A study by mouth used for a route that bypasses the gut, such as the bloodstream, adds a factor set by how much of an oral dose the gut absorbs. The factor is 100 when less than 1 % is absorbed, because the study's animals then took in less than a hundredth of their dose, and it falls to 1 at 90 % or more, with 10 when no data exist. Data of reduced reliability or relevance add 10. The edition also has a factor for a LOAEL used in place of a NOAEL, according to Gruber of the Johner Institute, a consultancy. The slides give no default for that factor.
The factors multiply, because each is a ratio between the dose a study saw and the dose it stands for. When a short rat study by mouth is used for a device that delivers into the blood for years, the animal, person, duration and route factors all apply together. The standard calls their product the modifying factor, the "mathematical product of uncertainty factors" in its definition. A product can grow large enough to mislead: the 2024 slides warn that a modifying factor above 10,000 for the general population can overestimate the risk.
The tolerable intake is what the division leaves
Dividing the point of departure by the modifying factor gives the tolerable intake, which ISO 10993-17:2023 defines as an "estimate of the daily exposure of an identified constituent ... that is considered to be without appreciable harm to health". Two words in that definition carry its meaning. "Estimate" says the value is computed, not measured, and "considered" says it rests on judgments about studies and factors. Because the value comes from a division, a tolerable intake lies below the point of departure by the whole product of the factors, not at the edge of harm. A dose above it falls inside that allowance, so exceeding the tolerable intake does not show that harm will follow. Chapter 15 turns the comparison into a ratio.
- EPA's starting point. EPA's assessment, last revised in 1987, starts from the LOAEL of Carpenter and colleagues, 19 mg/kg/d. It divides by 10 from animal to human, 10 from person to person and a third factor of 10, which EPA's summary ties to a study shorter than a lifetime with a minimally adverse effect: 1,000 in all.
- EPA's value. 19 divided by 1,000 is 0.019 mg/kg/d, which EPA rounds to 0.02 mg/kg/d, or 20 µg/kg/d, and calls a reference dose.
- EFSA's value. EFSA set its value in 2005, and the UK Committee on Toxicity, COT, printed its basis in 2011. EFSA started from a NOAEL of 5 mg/kg/d, as COT prints it, for harm to the testes and to development. The study behind it was a multigeneration feeding study in rats, which Wolfe and Layton reported in 2003 and COT cites as an unaudited draft. Divided by 100, the NOAEL gives 0.05 mg/kg/d, or 50 µg/kg/d. In 2019 EFSA set the same 50 µg/kg/d, on a temporary basis, as a group limit for DEHP and three other phthalates together, expressed as DEHP (Chapter 15).
- The result. The two values differ by a factor of 2.5, 50 against 20 µg/kg/d. EFSA started from a dose about four times lower, 5 against 19 mg/kg/d, but divided by a factor ten times smaller, 100 against 1,000, so its value ends higher. EPA's third factor of 10 stands for what its study lacked: a dose without effect and a lifetime of exposure.
- Assumptions. Both values were set for intake by mouth, not for a device, so another route may need a route factor.
EPA's draft for D4 replaces one default with data: it divides by 3 from animal to human and by 10 from person to person, 30 in all. Dividing the draft's human dose for repeated exposure by mouth, 3.60 mg/kg/d, by 30 gives 0.12 mg/kg/d, or 120 µg/kg/d. EPA's draft uses the factor of 30 as a benchmark for comparing exposures with the point of departure. The 120 µg/kg/d is therefore this book's division of EPA's draft numbers by the standard's method, not a value EPA set. The same division applied to the draft's two values for skin gives the tolerable intakes that Chapter 15 needs for the cushion. For a substance that can cause cancer, the 2023 edition offers a second route. There an accepted level of added lifetime risk sets the tolerable intake instead of a division by factors, and Chapter 16 gives that risk level with the thresholds built on it.
Body mass enters on the patient's side
Because a tolerable intake is counted per kilogram, a light patient and a heavy one share one value, and body mass enters on the patient's side, through the dose the patient receives. The 2002 edition of ISO 10993-17 worked differently: it had a second limit, the tolerable exposure, set for the patient and the use. "The notion of Tolerable Exposure (TE) thus disappears from the 2023 revision", Efor wrote in 2023. The patient and the use now enter the estimate of exposure, which divides what the patient receives each day by a body mass. The body masses are the defaults of Chapter 1's solved example. Gruber advises a lower mass for lighter patients, so the running device is judged at its lightest user's 30 kg (Chapter 15).
FDA recognizes the 2023 edition in part
The 2023 edition is the standard's second, and it cancels and replaces the edition of 2002. FDA recognizes the 2023 edition in part. It entered the edition in its database of recognized standards on 18 December 2023, and the record of that entry leaves out four items. Among them are a unit in a table printed as g/d where µg/d is meant, and a table in Annex E that is "not supported by the normative text". The record also states that FDA will accept declarations of conformity to the 2002 edition, in support of premarket submissions, until 20 December 2026. Chapter 26 sets out how FDA's recognition of a standard works.
For each compound on an extract's list, record the point of departure with its study, species, route, duration and harm, each factor with its reason, and the tolerable intake with the authority behind it and whether that value is final or a draft. Prefer a study by the route of the device's contact, or add the route factor. Where an authority's factors differ from the standard's, recompute its value. For the running device's cushion, which rests on skin, carry the skin values of EPA's draft for D4 siloxane, and their draft status, into the estimate of exposure (Chapter 15).
- A dose is the amount of a substance entering the body each day per kilogram of body weight, in µg/kg/d or mg/kg/d.
- The point of departure is a NOAEL, a LOAEL or a benchmark dose from one study, chosen as the most critical relevant value.
- Uncertainty factors for animal to human, person to person, duration, route and data quality multiply into a modifying factor, which starts at 100 for the first two.
- The tolerable intake is the point of departure divided by that factor, an estimate per kilogram that holds for light and heavy patients alike, and body mass enters through the patient's dose.
Let's Talk