Last reviewed 4 October 2012
Mel Cooke looks at the discrepancies between GHS classification inventories and provides some reasons as to why they might have occurred.
The EU chemical industry is all too aware of the requirements to classify chemical substances according to the Classification, Labelling and Packaging Regulation (CLP Regulation No. 1272/2008), which became mandatory from December 2010.
The pain of this transition to a new chemical classification scheme was meant to be outweighed by the benefits of using a Globally Harmonized System (GHS) for classification, on which the CLP Regulation is based — a classification made in the EU should more or less apply in any country adopting the GHS.
However, this idyllic vision has been poorly realised, and is danger of an East–West split in the practical outcome chemical classification, brought about by two influential sources of GHS classification: the REACH Registration data, and the Japanese NITE database.
The NITE database is influential particularly in Japan, but also in the wider Pacific Rim area. Classifications from NITE are consistently more severe than those provided in the EU REACH Registration data, despite the classifications being based on the same GHS criteria. The classifications are so diverse that chemical labelling might not be irreconcilable between these regions.
This article gives some examples of the discrepancies, some reasons why they might have occurred, as a possible solution.
The main sources of classifications in the EU are the following.
CLP Regulation, Annex VI. These are official classifications, and have to be adhered to in the hazard classes for which classification is given. However, they are partial classifications, and should be completed using available information (see CLP Art. 4.3). A further limitation is that some hazard classes (particularly acute toxicity and STOT RE) are regarded as a minimum classification, and a more severe classification is appropriate if the data warrants it.
REACH Registration data. As part of the registration process, applicants are required to provide the classification of their substance. This is a potentially very useful source, because the classifications are likely to be agreed by a consortium of large enterprises on the basis of pooled information.
The classification and labelling inventory. Currently, for most substances, this is a hotchpotch of diverse classifications that provide some indication of what classifications are being used by suppliers. It is hampered by not including substances that are not classified as hazardous. One hopes that it will improve over time.
The Japanese NITE database gives the classifications of approximately 1500 chemicals. The classifications were conducted by the Japanese GHS Inter-ministerial Committee in 2006 and translated into English by the Ministry of Economy, Trade and Industry (METI) in Japan. The classifications in the NITE database are not mandatory, but are influential in Japanese industry.
Discrepancies between the EU and NITE classifications
Note that the Japanese have implemented all the building blocks of the UN GHS, in contrast to the EU, where some building blocks were omitted, notably Acute Toxicity, Category 5, Skin Irritation Category 3, Aspiration Toxicity Category 2, and Aquatic Acute Categories 2 and 3. This means that the Japanese classification may include these elements of classification that are not necessary in the EU. These parts of the NITE classification have been omitted from Table 1 for the sake of clarity. Likewise, classification elements peculiar to the EU, such as supplementary labelling, have also been omitted.
Apart from these differences, the criteria for classification in other classes and categories are the same, and applying the criteria to the substance dataset should result in the same classification.
That this is not the case can easily be demonstrated by comparing EU REACH Registration classifications with those indicated in the NITE database (Table 1). These examples of large discrepancies are not isolated or difficult to find. The NITE database consistently shows more severe classification, particularly in the following areas.
Use of the STOT SE classification.
Use of the STOT RE classification.
Use of classification for carcinogens, mutagens, and reproductive toxicants (CMRs).
Substance (CAS number)
EU Classification according to REACH Registration
Classification according to NITE database
Flammable Liquids, Category 2; Eye Irritation, Category 2; Specific Target Organ Toxicity (Single Exposure), Category 3 (narcotic effects)
Flammable Liquids, Category 2; Eye Irritation, Category 2; Toxic to Reproduction, Category 2; Specific Target Organ Toxicity (Single Exposure), Category 3 (respiratory tract irritation; narcotic effects); Specific Target Organ Toxicity (Repeated Exposure), Category 2 (blood)
Flammable Liquids, Category 2; Acute Toxicity: Oral, Category 4; Acute Toxicity: Skin, Category 4; Acute Toxicity: Inhalation, Category 4; Eye Irritation, Category 2
Flammable Liquids, Category 2; Acute Toxicity: Skin, Category 3; Eye Irritation, Category 2; Germ Cell Mutagenicity, Category 2; Specific Target Organ Toxicity (Single Exposure), Category 1 (central nervous system, respiratory organs); Specific Target Organ Toxicity (Repeated Exposure), Category 2 (central nervous system, respiratory organs, kidneys, blood system, liver)
Flammable Liquids, Category 2; Eye Irritation, Category 2
Flammable Liquids, Category 2; Eye Irritation, Category 2; Germ Cell Mutagenicity, Category 1B; Toxic to Reproduction, Category 1A; Specific Target Organ Toxicity (Single Exposure), Category 3 (respiratory tract irritation; narcotic effects); Specific Target Organ Toxicity (Repeated Exposure), Category 1 (liver); Specific Target Organ Toxicity (Repeated Exposure), Category 2 (nervous system)
As a demonstration of these differences for labelling, some label elements are given (Figure 1) for acetonitrile classified as above.
* See Table 1.
If the data for the substances and the criteria for classification are the same, why are the results so different? The answer is partly cultural, because the EU has had some continuity from the older classification system under EU Directives 67/548/EEC and 99/45/EEC (for dangerous substances and preparations), so there is some reluctance to make large changes to the classifications of substances just because a new classification system has been introduced.
This can be seen for ethanol (Table 1), where there is no doubt that a “toxic to reproduction” classification is appropriate, according to the criteria, but a pragmatic approach has led to this classification being overlooked. The CLP Annex VI classifications set a precedence of what classifications should look like, but may distract the classifier from rigorously applying the criteria as presented in the CLP Regulation.
In contrast, the Japanese classification system, prior to GHS, was for historical reasons mainly concerned with environmental exposure and carcinogenicity, mutagenicity, and reproductive effects, so the governmental authority responsible for NITE has no preconceptions of the GHS classification system. It is too simplistic to assert that the NITE classifications are more severe because they have been developed by an impartial government body compared to industry classifications in the REACH registration dossiers.
Also, some of the GHS criteria is open to different interpretation. This can be demonstrated for STOT SE, where the GHS and CLP try to differentiate between classification for this effect and classification for acute toxicity. While one might think that STOT SE is meant as a classification for substances that produce toxicity in a single, specific organ, the GHS states (paragraph 184.108.40.206) that “generalized changes of a less severe nature involving several organs should be taken into account”. It is unclear how a substance that is classified for acute toxic effects, with lethality as the endpoint, could escape classification as STOT SE, as it would inevitably produce either specific or generalised organ toxicity.
With STOT RE and CMR classification, there is much reliance on “significant effects”, “weight-of-evidence”, and “expert judgement” — how should we expect two independent assessors to come up with the same classification?
United Nations list — too late already?
A simple solution to this would have been to have a single list of classifications agreed at international level by the UN from the beginning. The idea of a classification inventory has been discussed at UN level, but this currently is at an early stage, so no international list is possible for several years. It would be too late, if by then the two influential databases have become well established. Even if the UN does create a classification list, it would be non-binding, so that individual jurisdictions such as the EU and Japan would not necessarily adopt the classifications.
Unless industry, as global players, decides independently of the two databases on chemical classifications they will use globally, it might be that much of the potential of the GHS will be lost.