Free exam prep

ABT DABT Diplomate of the American Board of Toxicology

13.1 Uncertainty Factors (UFs), Interspecies/Intraspecies Extrapolation & Chemical-Specific Adjustment Factors (CSAF)

Q: Four full 10-fold EPA uncertainty areas would multiply to 10,000. Which statement matches EPA 2002/IRIS composite-UF practice rather than treating the product as a statute?

UFs are multiplied, overlap among duration, database, LOAEL, and species gaps is recognized, and EPA often treats 3,000 as a practical ceiling—collapsing two areas into one 10 rather than printing 10,000—and may decline to derive an RfD if still more tens would be needed; 3,000 is common practice, not a statute. Composite UFs are products. EPA’s 2002 RfD/RfC review and subsequent IRIS practice often cap a chronic composite near 3,000 by combining overlapping areas instead of stacking four independent tens to 10,000, and may conclude the database is too thin if still more tens would be required. That 3,000 figure is practice and a panel recommendation, not a statute. Adding tens, requiring 10,000 by statute, or deleting UFH after HED all misstate the method.

Q: At the same 10 mg/kg oral dose, a PBPK model predicts AUC_rat = 20 µg·h/mL and AUC_human = 40 µg·h/mL. Using the WHO IPCS interspecies split (default 4 TK × 2.5 TD), which CSAF treatment is correct?

CSAF_AK = 40/20 = 2 replaces the default TK 4; times residual TD 2.5 gives an interspecies composite of 5 instead of 10. EPA DDEF uses the same idea (EFAK as the external-dose ratio that matches the internal metric). If the PoD was already converted to an HED that matches this AUC, UFAK = 1 and you do not divide by 2 again.. Chemical-specific TK ratios replace the named TK default. An AUC_human/AUC_rat of 2 yields CSAF_AK = 2, times IPCS TD 2.5 = 5 for interspecies. EPA DDEF is the same logic with EFAK/EFAD names. If HED already matches the internal metric, the TK factor has been spent. Keeping the default 4 despite data, adding 2 to UFH, or treating higher human AUC as lower sensitivity reverses the method.

Q: A rat systemic oral PoD is 10 mg/kg-day. Rat BW = 0.25 kg, human BW = 70 kg. Which conversion and residual UFA match EPA 2011 oral systemic practice rather than FDA 2005 mg/m2 first-in-human practice?

HED = 10 × (0.25/70)^0.25 = 10 × 0.244 = 2.44 mg/kg-day; after that default BW^0.75 HED, residual UFA is typically 3, not a second full 10. FDA 2005 uses BW^0.67 / rat factor about 0.16 for MRSD, which is a different playbook.. EPA 2011 oral systemic HED equals animal mg/kg times (BWa/BWh)^0.25. For 0.25 kg and 70 kg that factor is 0.244, so 10 × 0.244 = 2.44 mg/kg-day, and residual UFA is typically 3. Inverting the body-weight ratio scales the wrong direction. FDA Km 6/37 ≈ 0.16 is the mg/m2 FIH path from Chapter 12, not the IRIS 0.75 default. Local portal-of-entry lesions do not get the systemic 0.75 default, but that exception does not cancel HED for systemic liver effects.

Key Takeaways

EPA default uncertainty factors are often 10-fold each: UFA (interspecies), UFH (intraspecies), UFS (subchronic-to-chronic), UFL (LOAEL-to-NOAEL), and UFD (database); a leftover half-log is recorded as 3.
UFs are multiplied, never added. EPA 2002/IRIS practice often treats a composite of 3,000 as a practical ceiling when four full tens would make 10,000; that ceiling is common practice, not a statute.
UFA and UFH split into toxicokinetic and toxicodynamic halves. EPA often uses about 3 × 3; WHO IPCS interspecies default is 4 (TK) × 2.5 (TD). CSAF and EPA DDEF replace a named half when chemical-specific ratios exist.
After default oral BW^0.75 HED or inhalation HEC dosimetry, residual UFA is typically 3, not a second full 10 on the same TK difference. Human PoDs take UFA = 1.
EPA 2011 oral systemic scaling uses BW^0.75 so mg/kg HED = animal dose × (BWa/BWh)^0.25. FDA 2005 first-in-human conversion uses mg/m2 (BW^0.67; rat Km 6/37 ≈ 0.16). Portal-of-entry lesions do not get the oral 0.75 default.

Last updated: September 2026

Uncertainty factors turn a laboratory dose into a public-health number

Handbook III.C.1 H and III.C.3 G ask you to take a point of departure (PoD) from Chapter 12—a NOAEL, LOAEL, or BMDL, sometimes already converted to a human equivalent dose (HED) or human equivalent concentration (HEC) as in Chapters 11–12—and divide it by factors that cover gaps between that study and the people you are protecting. Independent OpenExamPrep teaching in this section covers U.S. Environmental Protection Agency (EPA) default uncertainty factors (UFs), the toxicokinetic (TK) versus toxicodynamic (TD) split of interspecies and intraspecies factors, World Health Organization International Programme on Chemical Safety (WHO IPCS) chemical-specific adjustment factors (CSAFs), EPA data-derived extrapolation factors (DDEFs), allometric scaling (body weight^0.75 versus mg/m² / body weight^0.67), and when those defaults still apply after an HED or HEC is already on the table. This material is not an ABT, EPA, WHO, or IPCS product and does not claim official approval, review, or partnership with those bodies.

Domain III.C (dose–response) is 9% of the examination. Items here punish stacking every 10-fold factor that can be named, treating a composite UF of 3,000 as a law Congress wrote, or applying BW^0.75 and a full UFA of 10 to the same interspecies TK difference.

What a UF does

A UF is a unitless divisor. For an oral noncancer value:

RfD = PoD / (UF_composite × MF)

If the PoD is 25 mg/kg-day and the composite UF is 100, the RfD is 0.25 mg/kg-day. The factor does not convert a 10% extra-risk BMDL into a 0.1% extra-risk BMDL by biology. It does not prove that the true human threshold sits at PoD/100. EPA’s own RfD definition still says the estimate has “uncertainty spanning perhaps an order of magnitude.” A UF is science-policy coverage of a named gap: different species, different humans, shorter studies, a LOAEL instead of a NOAEL, or a thin database.

Modifying factor (MF). Older IRIS write-ups allowed an extra 1 to 10 professional-judgment MF (completeness of the critical study, steepness of the dose–response, or other residual issues not captured by the named UFs). In most modern files MF = 1. EPA’s 2002 Review of the Reference Dose and Reference Concentration Processes (EPA/630/P-02/002F) encouraged folding leftover judgment into the named UFs, especially UFD, rather than inventing a sixth silent 10. If an item still shows MF = 3, multiply it in; do not ignore it and do not add it to the UFs (10 + 3 is wrong; 10 × 3 is right).

UFs are multiplied, never added. UFA = 10 and UFH = 10 is 100, not 20.

EPA’s default five (often 10 each)

Factor	Name	Default when the gap is fully present	Typical partial value	What it covers
UFA	Interspecies (animal to human)	10	3 after default HED/HEC, or 3 for one of TK/TD	Humans may be more sensitive than the tested species
UFH	Intraspecies (human variability)	10	3 when TK or TD is partly quantified; 1 only with strong sensitive-subgroup data	The RfD must protect sensitive people, not only the mean adult in the study
UFS	Subchronic to chronic	10	3 when duration is intermediate or the effect is not expected to progress	The critical study is shorter than the lifetime the RfD claims to cover
UFL	LOAEL to NOAEL	10	3 when the LOAEL is mild and closely spaced; 1 if a NOAEL or BMDL is the PoD	The PoD is already in the adverse range
UFD	Database deficiency	1 to 10	3 is common when one major study type is missing	Missing developmental, reproductive, or other studies that could have produced a lower PoD
MF	Modifying factor	1	1–10 in older files	Residual professional judgment

UFA (interspecies). Default 10 when you start from an animal administered dose with no dosimetric conversion. EPA often splits that 10 into TK and TD, each about a half-log. 10^0.5 ≈ 3.16, which EPA rounds to 3 when a half-factor is used alone. After BW^0.75 oral HED (EPA 2011) or RfC inhalation HEC (EPA 1994 methods), the TK half is treated as addressed, and residual UFA is typically 3 (remaining TD plus leftover TK). Human epidemiological PoDs take UFA = 1—there is no animal-to-human jump.

UFH (intraspecies). Default 10 to cover children, elderly people, genetic polymorphisms, disease, and pregnancy. It can be split about 3 TK × 3 TD. EPA 2002 warned against shrinking UFH merely because “adults vary by less than 10.” Reduction needs data on the susceptible group you claim to have covered, not a wish. A Food Quality Protection Act (FQPA) extra 10× for infants and children is an Office of Pesticide Programs statutory overlay on some pesticide risk cups. It is not a sixth default IRIS UF for every industrial chemical.

UFS (duration). A 90-day (subchronic) rodent study supporting a lifetime RfD typically draws UFS = 10. A 2-year rodent bioassay supporting a chronic RfD draws UFS = 1. A 6-month or 1-year study, or a subchronic effect that is not expected to worsen with time (some adaptive liver-weight changes), may support UFS = 3. UFS is not applied because the chemical has a long half-life; that is a TK problem for UFA/UFH or a PBPK model.

UFL (LOAEL). If the PoD is a NOAEL or BMDL, UFL = 1. If you must start from a LOAEL because no NOAEL was identified, default 10. A minimal LOAEL (mild, reversible, tightly spaced above a true no-effect region) may support 3. BMD modeling is how you avoid UFL: the BMDL is already a bound on a defined response, so you do not also take a LOAEL-to-NOAEL 10 on the same endpoint.

UFD (database). A package with chronic, developmental, and reproductive studies and a well-chosen critical effect often gets UFD = 1. Missing a prenatal developmental or two-generation reproductive study commonly draws 3. Several missing study types, or a critical effect that looks like the wrong organ because the right study was never run, can draw 10. UFD is the factor people forget, then double-count by also inflating UFS.

Half-logs and the composite

EPA uses 3 as the stand-in for √10. Arithmetic consequences:

One leftover half-factor with tens: 10 × 10 × 3 = 300. Keep the 3. Do not “round 300 to 100” or to 1,000.
Two half-factors together are treated as a full log: 3 × 3 is recorded as 10, not 9. So UFA = 3, UFH = 10, UFS = 3 becomes 100, not 90.
Never add 10 + 10 + 3 = 23.

Worked composite. Animal BMDL, no HED: UFA = 10, UFH = 10, UFS = 1 (chronic study), UFL = 1 (BMDL), UFD = 3 (no developmental study). Composite = 10 × 10 × 3 = 300.

Same BMDL after BW^0.75 HED: UFA = 3, UFH = 10, UFD = 3 → 100 (two 3s counted as 10, times the remaining 10).

Do not pile 10s without thought: the 3,000 practice

Multiplying four full tens gives 10,000. Five tens give 100,000. EPA practice does not treat that product as automatic. The 2002 RfD/RfC review recommended limiting the total UF on a chronic reference value to 3,000. The reason is overlap: duration, database, and LOAEL gaps are not independent of species and human-variability gaps. When four full areas would mathematically be 10,000, IRIS-style write-ups often combine two areas inside one 10, leaving a composite of 3,000.

This 3,000 figure is common IRIS practice and a technical-panel recommendation. It is not a statute. Congress did not enact it. An older file with UF = 10,000 is not “illegal,” and a 2024 assessment that stops at 300 because the database is rich is not “noncompliant with a 3,000 rule.” If the assessor would need more than 3,000—five independent full tens—the 2002 panel’s logic is that the database is probably too thin to support an RfD at all. The grown-up move is to say no RfD, not to print 0.00001 mg/kg-day from a 28-day LOAEL in one species.

CSAF and DDEF: replace defaults when you have ratios

WHO IPCS (2005) chemical-specific adjustment factors and EPA (2014) data-derived extrapolation factors are the same idea with different initials. When you can measure the TK or TD ratio, you replace that half of the default, not the entire UF stack.

IPCS default interspecies 10 = 4 (TK) × 2.5 (TD). The 4 is the allometric/TK piece (clearance scaling roughly with BW^0.75). The 2.5 is residual TD. That 4 × 2.5 split is not the same as EPA’s 3 × 3. Mixing “EPA half-log 3” with “IPCS TK 4” on one line without naming the scheme is an item trap.

IPCS default human variability 10 ≈ 3.16 TK × 3.16 TD.

EPA DDEF names: EFAK (interspecies TK), EFAD (interspecies TD), EFHK (human TK), EFHD (human TD). After default oral BW^0.75 or inhalation HEC, EPA’s remaining interspecies default is the TD-ish 3, not another 4.

How a TK CSAF is built. Pick a dose metric tied to the mode of action (AUC, Cmax, amount metabolized). Find the animal external dose and the human external dose that produce the same internal metric.

CSAF_AK (or EFAK) = animal_external / human_external for that match.

If the animal PoD is 10 mg/kg-day and a PBPK model says humans reach the same AUC at 2.5 mg/kg-day, EFAK = 10 / 2.5 = 4. You then apply that 4 instead of the default TK piece. If the HED is already 2.5, you do not also divide by 4; the conversion is the TK factor, and UFAK = 1.

Worked AUC ratio (same external mg/kg). At 10 mg/kg, predicted AUC_rat = 20 µg·h/mL and AUC_human = 80 µg·h/mL. Humans experience 80/20 = 4-fold more internal exposure per mg/kg, so they are TK-more sensitive. IPCS-style CSAF_AK = 4. Times default TD 2.5 → interspecies composite 10 (here the data equal the IPCS TK default; they do not enlarge it). If instead AUC_human = 40, CSAF_AK = 2, times 2.5 = 5, which replaces default 10 for interspecies. If AUC_human = AUC_rat, CSAF_AK = 1, and only 2.5 (TD) remains—if you trust the metric.

TD CSAFs need a comparable tissue response (in vitro receptor occupancy, red-cell sensitivity, acetylcholinesterase IC50). EPA’s ethylene glycol monobutyl ether IRIS story is the teaching example of PBPK TK plus measured red-cell TD, driving UFAK and UFAD away from 10. You do not need that chemical’s numbers memorized; you need the logic: data replace the named half; they do not license deleting UFH because you liked the PBPK plot.

If humans are less TK-sensitive (higher clearance), CSAF_AK can be < 1, which increases the HED. Defaults of 10 never go below 1; data can. That is the point of CSAF.

Allometric scaling: BW^0.75 versus mg/m² (0.67)

Physiological rates (clearance, caloric use) often scale closer to body weight^0.75 than to body weight^1. Dose in mg/day then scales with BW^0.75, so dose in mg/kg-day scales with BW^−0.25:

HED (mg/kg-day) = animal dose (mg/kg-day) × (BW_animal / BW_human)^0.25

EPA 2011 (Recommended Use of Body Weight^3/4 as the Default Method in Derivation of the Oral Reference Dose) uses that 3/4-power default for oral systemic animal-to-human conversion. After it is applied, UFA residual is typically 3.

Body-surface-area / mg/m² scaling uses exponent 2/3 ≈ 0.67. Then mg/kg HED scales with (BWa/BWh)^0.33. FDA 2005 first-in-human guidance (Chapter 12) implements this with Km factors (human 37, rat 6, mouse 3, dog 20). Rat mg/kg × 6/37 ≈ 0.16. That is the FIH/chemotherapy playbook, not the EPA chronic oral RfD playbook.

Playbook	Exponent on BW for mg/day	mg/kg conversion	Typical next factor
EPA 2011 oral systemic RfD	0.75	× (BWa/BWh)0.25	Residual UFA ≈ 3
FDA 2005 FIH / mg/m²	0.67	× (BWa/BWh)0.33, or Km ratio	Safety factor often 10 on HED (MRSD), not an RfD
IPCS TK default	Built into CSAF_AK = 4	Same 3/4 idea	Times TD 2.5
REACH oral DNEL (rat)	Allometric 4	Same family	Times remaining 2.5

Worked BW^0.75. Rat 0.25 kg, human 70 kg, animal PoD 10 mg/kg-day.

BWa/BWh = 0.25/70 = 0.003571.

(0.25/70)^0.25: 0.25^0.25 = 0.7071, 70^0.25 ≈ 2.893, ratio 0.7071/2.893 = 0.244.

HED = 10 × 0.244 = 2.44 mg/kg-day. Using 0.24 as a rounded DAF: 2.4 mg/kg-day.

Apply UFA = 3 and UFH = 10 (chronic NOAEL/BMDL, complete database): composite 30.

RfD = 2.44 / 30 = 0.081 mg/kg-day.

Wrong double count: HED 2.44 then UFA = 10 and UFH = 10 → 2.44/100 = 0.024 mg/kg-day. That reapplies TK already spent in the 0.75 exponent.

Wrong skip: no HED, 10 / (10 × 10) = 0.10 mg/kg-day. Different number, older administered-dose style; if the item says EPA 2011 HED was used, this is incorrect.

FDA 0.67 on the same 10 mg/kg rat dose: 10 × (6/37) = 1.62 mg/kg-day. That 1.62 is an HED for FIH, not an RfD until someone applies a different factor stack. Using 1.62 inside an IRIS RfD as if it were the 2011 DAF is mixing playbooks.

Mouse DAF check. Mouse 0.025 kg, human 70 kg: (0.025/70)^0.25 ≈ 0.138. Animal PoD 10 mg/kg-day → HED ≈ 1.38 mg/kg-day. ICH Q3C F1 = 12 for mouse is a surface-area species factor (1/12 ≈ 0.083), not this EPA DAF, and not a UF you stack on top of 0.75 scaling.

HED/HEC recap (Chapters 11–12) without rerunning those chapters

You already converted inhalation concentrations with RfC dosimetry in section 11.3: regional gas dose ratio (RGDR) or particle regional deposited dose ratio, Category 1 portal-of-entry gases versus Category 2/3 systemic gases, producing an HEC in mg/m³. You already converted drug NOAELs with FDA 2005 mg/m² (0.67) in section 12.3 for maximum recommended starting dose. This chapter’s job is the UF that remains after those conversions, and the EPA oral 0.75 path that Chapter 12 did not use for first-in-human arithmetic.

Rules of thumb:

If the PoD is already an HED or HEC from default dosimetry, residual UFA is typically 3, not 10.
If a PBPK model produced the HED by matching the relevant internal metric, UFAK (TK) = 1; decide UFAD from TD data or leave the TD default.
Portal-of-entry / local GI corrosion is not a systemic metabolic-rate problem. Do not apply oral BW^0.75 as if clearance scaled the effect. Use the local PoD and a full UFA of 10 unless you have local dosimetry.
Human PoD: UFA = 1. Still apply UFH unless the human dataset already includes the sensitive subgroup the RfD must cover.
Do not run 0.67 FIH HED and 0.75 RfD HED on the same line and average them.

Scenario

A 90-day rat oral study, BMDL10 12 mg/kg-day for a systemic liver enzyme pattern, no PBPK, no developmental study. Assessor A applies EPA 2011 DAF 0.24, HED = 12 × 0.24 = 2.88 mg/kg-day, then UFA = 3, UFH = 10, UFS = 10 (subchronic), UFD = 3. Two 3s plus two 10s: EPA-style composite 1,000 (3 × 3 recorded as 10, times 10 × 10). RfD = 2.88 / 1,000 = 0.0029 mg/kg-day. Assessor B skips the DAF, uses UFA = 10, UFH = 10, UFS = 10, UFD = 10 = 10,000, and reports 12/10,000 = 0.0012 mg/kg-day. Assessor B piled four full tens. EPA 2002 practice would question whether an RfD should be derived, or would collapse to 3,000, not celebrate 10,000 as “more conservative, therefore better.” Assessor C has a PBPK HED of 4.0 mg/kg-day matching AUC of the parent and in vitro TD showing equal hepatocyte sensitivity: UFAK = 1, UFAD = 1, UFH = 10, UFS = 10, UFD = 3 → 300, RfD = 4.0/300 = 0.013 mg/kg-day. Three lawful-looking numbers; the examination asks which factors were already spent in the PoD.

A second file is forestomach irritation in rodents after gavage of a corrosive. Applying BW^0.75 because “EPA 2011 says so for oral” treats a local effect as a systemic clearance problem.

Traps

Adding UFs (10 + 10 + 3) instead of multiplying.
Applying UFA = 10 after a default HED/HEC.
Using IPCS 4 × 2.5 and EPA 3 × 3 as if they were the same numerals.
Treating 3,000 as a legal maximum written into the US Code.
Using FDA 0.67 / Km 0.16 inside an IRIS oral RfD as the 2011 DAF, or the reverse in an FIH protocol.
Taking UFL = 10 on a BMDL.
Shrinking UFH because the workers in a small plant study “looked healthy.”
Applying BW^0.75 to a portal-of-entry lesion.

Test Your Knowledge

Four full 10-fold EPA uncertainty areas would multiply to 10,000. Which statement matches EPA 2002/IRIS composite-UF practice rather than treating the product as a statute?

The US Code requires every IRIS RfD to use a composite UF of exactly 10,000 whenever four gaps can be named, and a 2002 technical panel cannot change that number

UFs are multiplied, overlap among duration, database, LOAEL, and species gaps is recognized, and EPA often treats 3,000 as a practical ceiling—collapsing two areas into one 10 rather than printing 10,000—and may decline to derive an RfD if still more tens would be needed; 3,000 is common practice, not a statute

UFs are added, so four tens equal 40, which EPA then rounds to 30 as the only lawful composite

Once BW^0.75 HED is calculated, all UFs including UFH must be set to 1 because dosimetry has removed every gap

Test Your Knowledge

At the same 10 mg/kg oral dose, a PBPK model predicts AUC_rat = 20 µg·h/mL and AUC_human = 40 µg·h/mL. Using the WHO IPCS interspecies split (default 4 TK × 2.5 TD), which CSAF treatment is correct?

CSAF_AK = 40/20 = 2 replaces the default TK 4; times residual TD 2.5 gives an interspecies composite of 5 instead of 10. EPA DDEF uses the same idea (EFAK as the external-dose ratio that matches the internal metric). If the PoD was already converted to an HED that matches this AUC, UFAK = 1 and you do not divide by 2 again.

CSAF_AK must stay 4 even when the AUC ratio is 2, because IPCS forbids replacing the TK default with data

The AUC ratio of 2 is added to EPA’s UFH of 10 to make a composite of 12

Humans have twice the AUC so they are less sensitive; invert the ratio and multiply the animal PoD by 2 without a TD factor

Test Your Knowledge

A rat systemic oral PoD is 10 mg/kg-day. Rat BW = 0.25 kg, human BW = 70 kg. Which conversion and residual UFA match EPA 2011 oral systemic practice rather than FDA 2005 mg/m2 first-in-human practice?

HED = 10 × (6/37) = 1.62 mg/kg-day, then apply UFA = 10 because FDA Km factors are required for every IRIS RfD

Skip allometric conversion and always use UFA = 10 on local forestomach corrosion and on systemic liver effects alike, because EPA 2011 forbids HED

HED = 10 × (70/0.25)^0.25, which enlarges the human mg/kg dose, then set residual UFA = 1

HED = 10 × (0.25/70)^0.25 = 10 × 0.244 = 2.44 mg/kg-day; after that default BW^0.75 HED, residual UFA is typically 3, not a second full 10. FDA 2005 uses BW^0.67 / rat factor about 0.16 for MRSD, which is a different playbook.

Up Next

13.2 Derivation of Reference Values: RfD, RfC, ADI, TDI & Permissible Daily Exposure (PDE)

Continue learning