Kickoff values are occupational exposure values for substances without a formal exposure limit, but with (limited) health information (H3##statements). The original kickoff values from 2005 have been updated using Hphrases, 2013 OEL databases and improved Control Banding systems. Below you will find the draft results. We presented the draft results at OH2014, the BOHS conference 2014. Our presentations at this conference are here.
Summary
Below are the draft versions of the kickoff values. They are based on the Control Banding Scheme of COSHH Essentials for exposure to dusts/aerosols and IFA’s GHS Spaltenmodell for the exposure to gases/vapours.
Comments Round closed
In the period April through November 2014 we held a commentary round for all interested parties. Meanwhile, the commentary round is closed and we process the comments and improvement in the final version of the kickoff values in 2014.
We want to thank all contributors for their comments. The final version of the kickoff values will improve significantly!
What are kickoff values?
In 2005 a method was published for the determination of 8hour exposure limits for substances without formal limits, but with risk phrases (Rphrases) concerning health or toxicological effects. These were called kickoff limit values, because they were a conservative starting point for any further development of a healthbased in company exposure limit , based on an analysis of all healthbased scientific information (animal studies and epidemiology) available at that time.
In the 90s of the last century “Control Banding” schemes grouped Rphrases of substances into socalled hazard categories were developed. The kickoff values 2005 were based on the distribution of OELs of substances in Rphrasehazard categories. It turned out that kickoff values based on the German TRGS 440 hazard categories were much more robust than the others (COSHH, ECETOC, ILO, [DUTCH] SOMS), due to the differences between the distribution of OELs of this scheme. The proposed OELs in het COSHH Essentials scheme turned out to be too optimistic.
The kickoff values were defined as the 10percentile of the OEL distribution of the substances in a hazard category. This means that 90% of the OELs for substances in that hazard category have a higher value than the kickoff. This also means that the absolute value of a kickoff value is on the conservative side.
If limiting the exposure to below the kickoff value is technically feasible, the need to conduct (expensive) additional studies for a more unbiased and probably higher healthbased limit value is less necessary.
The kickoff value method has been received positively enthousiastic: It has have been given a sound place in the realm of the Dutch exposure assessment policy. The Dutch Labour Inspectorate accepts kickoff values for compliance testing.
Why upgrading the kickoff values?
Since 2005 much has changed. First, the Rphrases are replaced soon by H/EUHcodes (as part of the new CLP system). Furthermore, the DOHSBase Compare database with OELs has more than doubled in size. Additionally the international tendency to set OELs at lower levels may have shifted the OEL distribution downwards. Finally, the “Control Banding” schemes we used in 2005 have also been modified (Hstatements instead of Rphrases) and improved by the respective authorities.
Translation from DSD to CLP
Since the CLP Regulation has entered into force, the use of Rphrases will decrease. For single substances its use is not allowed anymore. This means that the dataset used to calculate the kickoff values should be converted from Rphrases to Hphrases.
One of the challenges is the fact that the ‘translation’ of Rphrases for the new Hstatements is not a simple 1on1 replacement. The reason for this is that the borders of the groups of the toxicological groups/categories are not the same for the old (DSD) and the new (CLP). For example, see the diagram below.
Lethal; route  LD50 (mg/kg) 
Rphrase DSD 
CLP hazard class 
H Phrase CLP 

Oral  ≤  5 
28 
Acute tox. 1 
300 
oral  525 
28 
Acute tox. 2 
300 

oral  2550 
25 
Acute tox. 2 
300 

oral  50200 
25 
Acute tox. 3 
301 

oral  200300 
22 
Acute tox. 3 
301 

oral  3002000 
22 
Acute tox. 4 
302 
Control Banding Systems
As in 2005, substances with an OEL are assigned to a hazard class in a Control Banding System. Compared to 2005, some of the known Control Banding schemes have been adjusted to CLP. Kickoff level assessments are performed using the following Hstatement based Control Banding schemes. Click on the name of the system for more information on the control banding systems:
▪ COSHH Essentials (United Kingdom)
▪ DGUVIFA Spaltenmodell (Germany)
▪ EMKG (Germany)
Database DOHSBase Compare
The basis of the data set used is DOHSBase Compare (version 2013/14) with nearly 3000 chemical substances with one or more national, international and/or company OELs (as TWA 8 hour): these values are used for the occupational hygiene advice and compliance verification. DNELs and DMELs were excluded but will be used in the future to test their relation with their notified CLP classification. Furthermore the following OELs were removed:
 Exposure limits with a different dimension (respirable dust, fibres, fibril, CFU, vol.%, EU/m3, glycine unit/m3), with the exemption of µg (micro), pg (pico) and ng (nano). The latter were converted to mg (milli).
 Substances that have been evaluated (for example, by the Health Council of The Netherlands and SCOEL) but for which no limit could be established.
 In case of multiple TWA 8 hr OELs for a substance, the OELs from lower hierarchy for OELs were removed
The remaining, (merely health based) OELs are linked through their CASnumber with a database containing the harmonized CLP classification (symbols and Hstatements) of nearly 8000 substances. The linking of OELs and Hstatements leads to a database with more than 900 substances with Hstatements and a TWA 8hour OEL. These are the data with which the analytical computations were performed.
Analysis
Using the substances, their TWA 8hour OELs and hazard classifications, the shape and location of the OEL distribution was determined per hazard group of the different Control Banding schemes (A, B, C, D, E). Based on the ”best fit” with the lognormal distribution, the 10%tile of the TWA 8hour OEL distribution was estimated using the Wilks’ unbiased Studentt lower tolerance limit or nonparametric for each hazard class.
De Generic Exposure Values of the Control Banding schemes and the lower limits of the Control Bands were compared with the associated OEL distribution.
The following methods were used:
 Lognormal probability plot and boxplots (HYGINIST)
 Shapiro and Wilk test for distributional goodnessoffit with Regression for censored samples (HYGINIST)
 ANOVA to test the differences between the hazard groups (BW_Stat)
Occupational Exposure Limits are, by definition, greater than zero, often have values between 0,01 and 100 mg/m3 and are rarely greater than 1000 ppm or mg/m3. OEL distributions are therefore skewed. The software program HYGINIST was used to investigate whether the statistics of the lognormal distribution could be used to estimate the lower limit of the distribution. The Geometric Mean (GM) is the measure of the location of the OEL distribution. The geometric standard deviation (GSD) is the measure of the dispersion. The percentiles are calculated using OEL% = GM * GSD^{^k}, where k is the Wilk factor for the unbiased estimate of the percentile.
Results
The distributions of the OELs in ppm and mg/m3 (yaxis) per hazard class (xaxis) for three schemes (COSHH, TRGS600 and EMKG) are shown in the tables below, together with the statistical data of each distribution. The lognormal shape fits best between the 10%90%tile. Where necessary censoring is used to anticipate deviations in the tail. The 10%90%tile is a measure for the bandwidth (range).
The bandwidth for all three schemes is fairly large for each hazard class: 2 orders of magnitude for the aerosols and to 3 orders of magnitude for the gas / vapor OELs.
All Control Banding schemes show a significant downward trend of OEL distribution and hazard group. This decrease attenuates as the hazard group becomes more toxic (from A to E). From the various statistical functions (linear, logarithmic, polynomial, exponential, power series), the negative exponential function describes the relationship between values and hazard class the best.
In the box plots (page 2), the vertical rectangle represents 80% of the limit distribution valve and the vertical line the whole bandwidth (range) again. For all three schemes the current aerosol OEL’s are limited upward to 10 mg/m3.
In each box plot the comparison of the Generic Control Value is presented for each Hazard Class. The lower yellow line in the box plots represents the Kickoff OEL level. In the table above the box plots the kickoff OELs are presented in figures.
Results of the OEL calculations :
 with COSHH Essentials (United Kingdom)
 with DGUVIFA Spaltenmodell (Germany)
 with EMKG (Germany)
Conclusions
The grouping of the H3## statements of COSHH essentials offers according to our findings the best fit for the establishment of kickoff OELs for aerosols. For gases and vapors the grouping of DGUVIFA GHS Spaltenmodell provides the best fit. See for proposed Kickoff OELs tables at the top of this page.