On July 1, 2013 the peer-reviewed journal Environmental Health published the scientific review article “Dispelling urban myths about default uncertainty factors in chemical risk assessment –sufficient protection against mixture effects?” (Martin et al. 2013). The researchers from Brunel University scrutinize the well-established application of uncertainty factors in risk assessment of chemicals. Uncertainty factors have been introduced to extrapolate toxicological data from animal experiments to humans in chemical risk assessment. Uncertainty factorsarealso referred to as safety factors or assessment factors. They are often considered conservative enough to also account for the potential mixture effects of exposures to chemicals with different modes of action. The authors conclude that this is not the case, based on the evidence they reviewed.

How are uncertainty factors used in chemical risk assessment? Acceptable or tolerable daily intakes (ADI, TDI) are the tolerable doses for human life time exposures to a given chemical. The ADI is calculated from the No Observed Adverse Effect Level (NOAEL), the highest dose generating no measurable toxicological effect in animal experiments; thereby, the NOAEL is divided by a default uncertainty factor of 100 and the resulting value is the ADI.

ADI levels are considered ‘without appreciable risk’. This does not imply absolute safety, but rather that the risk arising from a specific reference dose is negligible: for example, there is 95% confidence that one additional case of cancer will be caused by a specific chemical at the given ADI, in a population of 1’000’000 people. As Martin et al. write, “Currently there are no legally binding quantitative definitions of desired level of protection for non-carcinogenic chemicals”, however there is considerable debate about how the exact level of protection is derived and where it shall be set.

Additionally, the benchmark dose level (BMDL) has been introduced as a valuable alternative to NOAELs. The BMDL is defined as the lower statistical confidence limit of the dose resulting in a predetermined response, for example an effect occurring in 10% of the exposed test animal population. It accounts for the shape of the dose-response curve, the quality of the study design, and it is not restricted to doses tested experimentally.

History of the uncertainty factors. Originally, the default uncertainty factor was chosen arbitrarily. It consists of the sub-factors of 10 each for inter- and intraspecies differences, which again are subdivided into toxicokinetic and -dynamic factors (Fig. 1).



Interestingly, these uncertainty factors were historically understood to also cover mixture effects. This understanding was not based on the intrinsic conservativeness of the uncertainty factors, but rather because the complexity of the issue resulted in a general lack of data to quantify mixture effects more appropriately. The review of relevant literature carried out by Martin and colleagues demonstrates that overall safety factors are broadly distributed ranging between 1.2 and 7’500. For particularly sensitive populations they may even amount to a factor of 10’000.

Large differences between species. Aspects concerning inter- and intraspecies differences were discussed in detail, emphasizing the need to reevaluate current regulatory practice concerning default safety factors. Regarding interspecies differences, the authors of the review paper point out that many physiological functions correlate better with body surface or caloric demand (allometric scaling) than with the generally used dose metrics based on bodyweight. Yet, certain non-allometric scaling differences, such as relative brain weight and oxygen consumption may also result in a misjudgment of applicable doses. For example, the effects of polychlorinated biphenyls (PCBs) differ between rats, mice and monkeys by several orders of magnitude, making allometric extrapolation inappropriate.

Interspecies differences of toxicokinetic parameters were also found to exceed the default value of 10. Accordingly, the lethal dose of 2,3,7,8-tetrachlorodibenzop-dioxin (TCDD) for 50 percent of the tested animals (LD50) differed by four orders of magnitude between hamsters and guinea pigs. The researchers conclude from the quantitative evaluations of interspecies differences that smaller animals appear to be less sensitive than larger ones, and that a significant portion of chemicals surpasses the default safety value of 10.

Largest uncertainty associated with variability in the human population. Concerning intraspecies variability, Martin and colleagues argue that laboratory animals are genetically relatively uniform and may therefore exhibit uniform reactions. This contrasts sharply with the genetic, age and gender variability in humans and is further worsened by acquired susceptibility factors such as diseases, diet and chronic exposures. Therefore, the default factor of 10 for intraspecies differences is doubted to adequately represent human variability. A study cited in the review article suggests that inter-individual variability is greater than the default value of 10 for 8% of the 490 chemicals considered in acute lethality tests carried out in inbred rats (Weil 1972). Moreover, the list of crucial aspects to be considered in chemical safety assessments is extended by gender differences, genetic polymorphisms, and increased susceptibility of neonates, infants and children as well as health-impaired and elderly populations.

Summarizing, the authors firstly conclude that the default uncertainty factors neither are, nor were intended to be worst-case scenarios, and that their conservativeness can only be assessed in relation to a defined level of protection. Secondly, reference databases with which the adequacy of safety factors can be tested are of poor quality, partially due to the intractable ethical issues related to testing humans directly. Thirdly, they argue that according to available data a default uncertainty factor of 100 cannot be considered conservative for all cases. Fourthly, the regulatory framework should encourage chemical- or pathway-specific factors instead of default uncertainty factors to generate better data. Finally, Martin and colleagues determine that there is a need to prioritize research and regulation for the safety assessment of mixture effects, which are not included adequately in the current chemical safety assessment procedure.


Martin, O. et al. (2013). “Dispelling urban myths about default uncertainty factors in chemical risk assessment –sufficient protection against mixture effects?”. Environmental Health (published online June 21, 2013).

Weil, C.S. (1972). “Statistics vs safety factors and scientific judgment in the evaluation of safety for man.” Toxicology and Applied Pharmacology, 21(4):454-463.