- air concentrations
- air-water partition coefficient
- animal data
- chemical food contaminants
- chemical properties
- chemical safety assessment
- chemical safety report
- chemical substance
- hazard banding
- hazard categories
- human data
- Occupational exposure limit (OEL)
- octanol-air partition coefficient
- public health risk
- read-across approach
- respiratory dose
- risk assessor
- risk phrase
- sensory irritation
- targeted risk
TR 101 : Guidance for setting occupational exposure limits: Emphasis on data-poor substances | October 2006
In the absence of sound human exposure data, existing procedures for setting occupational exposure limits (OELs) for chemical substances are generally based on a no observed adverse effect level from repeated dose animal studies, with application of appropriate assessment factors to account for uncertainty and variability in the data set. These procedures are reviewed briefly in this report.
Contrary to these ‘data-rich’ substances, for which adequate data are available, no clear procedures exist for the derivation of OELs of ‘data-poor’ substances. In this report, six methods for setting OELs for such substances have been proposed and evaluated. Worked examples are provided.
Hazard banding seems to be a promising method to set OELs for data-poor substances with EC risk phrases. These risk phrases are grouped following ECETOC criteria into four categories or hazard bands for gases/liquids and solids, each corresponding to a specific OEL range.
The maximum tolerated dose in long-term studies can be used to derive an OEL. If not known, the maximum tolerated dose can be predicted from the acute oral toxicity (lethal dose in rats) and the octanol-water partition coefficient.
Four-hour lethal concentrations from rat inhalation studies can be used directly for calculating OELs.
Current (quantitative) structure-activity relationships for predicting toxicity are insufficiently reliable, and therefore of limited value for setting OELs. It is recommended to search for substances with similar structures and known toxicity, and then read the data across.
If an OEL is to be based on sensory irritation, it can be predicted from the so-called respiratory dose, i.e. the concentration in air which reduces the breathing rate of mice by 50%. If not available, the respiratory dose can be calculated from the octanol-air partition for substances from a homologous series.
Finally, the principle of threshold of toxicological concern (normally for food contaminants) can be used for deriving OELs if less conservative safety factors are applied.
For certain substances none of the proposed methods will be applicable. For others, one or more of the methods might be appropriate, but could lead to different results.
In conclusion, therefore, it is proposed that an integrated approach based on the six methods proposed can be used to set a provisional OEL for the data-poor substance concerned. However, for the value to be reliable, experienced toxicological expertise is required in the interpretation of the results.