INSIGHT: Scientists Claim EU Uses Unrealistic Chemical Exposure Models for Safety Calls

Models being used by European regulators to assess the safety of chemicals in commerce could give skewed results because they do not align with a fundamental principle of physics. The exposure models are unreliable, calling into question the justifications for chemical safety calls in the bloc.

One of the fundamental laws of physics is conservation of mass—i.e., matter cannot disappear or be created spontaneously.

In exposure sciences, mass balance is the minimum requirement for understanding mass flows of a pollutant from its sources to human receptors. Because the European Chemicals Agency (ECHA) endorses exposure models not aligned with this physical law, the results may sometimes be unrealistic. Without using this first-principle, it is not possible to understand whether any modeling result is right or wrong.

The European Chemicals Agency, located at Helsinki, Finland, recommends in the Registration, Evaluation, Authorization, and Restriction of Chemicals (REACH) legislation the use of non-physical models, namely Stoffenmanager® (StM) and the Advanced REACH Tool (ART), for regulatory safety decision-making. In StM and ART, the mass is not conserved, and instead of using physical quantities, they rely on obscurely designed scores and indices to calculate an exposure score that is converted to an inhaled exposure level (milligrams/cubic meter) using a calibration factor.

This leaves the models difficult to understand and more difficult to explain compared to physical models.

Reality and Exposure Models

The figure below of a welding operation shows how easy an exposure scenario can be generalized and simplified for mathematical analysis. Suffice it to say that ventilation flows, emission rates, and near-field and far-field concentrations are all physical quantities that can be measured to determine the ambient concentration of a pollutant both close to the source and in the surrounding area.

Since publication of StM and ART during 2008 and 2011, respectively, the models have been developed, tested, validated, and calibrated in more than 40 peer-reviewed studies. But a year ago it was found that the general ventilation multipliers were incorrectly calculated. These multipliers are always present in the models. In a model based on physics, the errors would have been possible to identify simply by checking a mass balance—the first validity check when developing exposure models.

The model developers did not see this as an issue because StM and ART are calibrated using thousands of workplace measurements. But does this quality control process actually work? The bottom line is, can the models predict exposure levels within reasonable limits if the models are not based on the underlying physical and verifiable exposure determinants?

Poor Precision?

Recent studies have shown that the predictive power of exposure models, i.e. the ratio of measured versus modeled concentration, is in the range of 100,000-fold for StM and 100-fold for ART. This means that if we have a true 1 milligram per cubic meter concentration in the workplace, the concentration predicted with StM or ART ranges from 0.001 to 100 and 0.1 to 10 milligrams per cubic meter, respectively.

This precision is poor when compared to near-field and far-field model accuracy, (see graphic of a welding operation below) which typically ranges from 0.5 to 2 fold —or 0.5 and 2 milligrams per cubic meter in our example. Thus, it seems that invoking calibration factors is not as scientifically sound as it could be.

The StM and ART developers are aware of the poor predictability and that the models can underestimate or overestimate the exposure. Thus, they decided to use the upper 90th percentile of the exposure distribution to overestimate the exposure and make precautionary safety decisions.

However, the ranges of these models’ uncertainties are so large that StM still underestimates the exposure level in over 10% of the cases. This could be helped by using 95th or 99th percentile, but would that be reasonable? It seems that StM and ART are not as useful as they could be for safety decision-making. Because they do not conform to physical laws, they cannot be used to pin down the factors behind the exposure.

As long as StM and ART are non-physical models, their development is stuck. Exposure model development depends on how well the mass-balance model parameterization reflects the reality and how well their uncertainties are known. Only a scientifically well-justified exposure model can be used for reasoned regulatory safety decisions. Physical models based on mass balance can be easily interpreted, operated, and updated to replicate reality for the purpose of exposure estimates. They are not limited only to occupational environments but are applicable to any built environment and can be coupled with atmospheric environmental models.

Accurate Exposure Predictions Needed

There is an urgent need for reasoned safety decision-making in indoor air pollution models, which is evidenced by StM being used by over 33,000 users, a number that is increasing weekly by some 50 new users.

However, before any models are applied to regulatory safety decision-making, their rationale should be reviewed similarly as it has been done for the near-field and far-field, indoor-air model of breathing zone concentrations by Jayjock et al. presented in the figure. Considering StM and ART, it is recommended that this review occur before the models are subjected to judicial review.

This column does not necessarily reflect the opinion of The Bureau of National Affairs, Inc. or its owners.

Author Information

Joonas Koivisto is an exposure scientist at the University of Helsinki’s Institute for Atmospheric and Earth System Research. Tareq Hussein is a professor of atmospheric sciences in the Department of Physics at the University of Jordan. Michael Jayjock is an exposure consultant and principal of Jayjock Associates, LLC, in Langhorne, Pa.