2.1. CSI Minimum Input Requirements
The CSI tool is currently implemented as a semi-automated worksheet application in Microsoft Excel. A customized software solution is under development with the objective of documenting hazard assignments, organizing associated metadata, automating product scoring, and minimizing user error.
The minimum data inputs needed to derive a CSI product score include: (1) the identity and percent composition by weight of components making up at least 70% of the total composition of the product to be scored; and (2) hazard assignments for each product component or a determination that data are not available to assign hazards for a component.
Component composition of the product to be scored is typically available from internal sources (e.g., internal formulation data or SDS) or external sources (e.g., information provided by a chemical supplier). If a component cannot be identified based on its Chemical Abstract Service (CAS) registry number or other chemical identifier (e.g., International Union of Pure and Applied Chemistry [IUPAC] name, International Organization for Standardization [ISO] name, common chemical name, Simplified Molecular-Input Line-Entry Specification [SMILES] notation), then it is categorized in the CSI as having “No Data Available”. If more than 30% of a product’s composition is due to the contribution of components with “No Data Available”, then that product is designated “Do Not Evaluate” in the CSI, and re-evaluated at a later time when more information may be available.
2.2. Hazard Categories Evaluated in the CSI
Selection of the appropriate hazard categories is a key component of chemical and product safety assessment, requiring a balance between comprehensiveness, complexity, and redundancy. The three broad groupings of “hazard categories” evaluated by the CSI—hazards to the environment, physical safety, and human health—are referred to as “hazard criteria”. Each criterion within the CSI includes a number of hazard categories as defined by the United Nations’ (UN) Globally Harmonized System (GHS) of Classification and Labeling of Chemicals [12
], supplemented with additional non-GHS hazard categories in the environmental criterion (Table 1
The GHS is an international system for standardizing and harmonizing the classification and labeling of chemicals [12
]. While current chemical management systems may be similar in content and approach, their differences are significant enough to require multiple classifications, labels and safety data sheets for the same product when marketed in different countries, or even in the same country when parts of the life cycle are covered by different regulatory authorities. The GHS was developed to provide greater consistency across jurisdictions. It is a voluntary international system, not a regulation or a standard, and no international implementation schedule has been defined. However, implementation has been initiated, or is actively being discussed, by several international bodies and various countries, including the U.S., have incorporated the GHS into their chemical regulatory programs. It is therefore anticipated that the GHS will continue to be implemented over the coming years and serve as a harmonizing standard for chemical hazard communication in the future. Since a key objective of the CSI is to be applicable and relevant as a global tool for product hazard evaluation, the GHS was selected as an appropriate starting point.
The CSI includes or addresses all hazards defined by the GHS [12
], with some modifications under the environmental, physical safety, and health criteria (see Table 1
and Supplementary Material). Additional environmental hazards included in the CSI, but not considered under the GHS, are “volatile organic compounds” (VOCs), “hazardous air pollutants” (HAPs), “hazardous water pollutants” (HWPs), and “endocrine disruptors”, and the use of two separate biodegradability hazard categories as opposed to one in the GHS. These two separate, but mutually exclusive, biodegradation hazard categories, “biodegradation-persistent” and “biodegradation-inherent”, are used in the CSI to be consistent with regulatory environmental testing requirements for oil and gas operations under OSPAR regulations [2
]. Also, the GHS “chronic aquatic toxicity” hazard is not scored as a separate hazard but chronic aquatic toxicants are covered by the CSI through individual scoring of “acute aquatic toxicity”, “biodegradation”, and “bioaccumulation”. Similarly, acute toxicity, biodegradation and bioaccumulation are used as separate environmental hazards to classify chronic aquatic toxicants in the GHS classification scheme when no adequate chronic aquatic toxicity data are available. Consequently, the CSI approach for chronic aquatic toxicity hazard assignment is consistent with the GHS.
The primary reason for the inclusion of additional environmental hazards is that the CSI uses a broader definition of environmental hazards than the GHS. Under the CSI, the health criterion focuses on worker health (note that although there is a focus on worker health, many of the hazard categories included under the health criterion are relevant to general population exposures as well). In contrast, hazards that may result from a chemical being released into the environment, migrating off-site and subsequently affecting the general population that is exposed to the resulting environmental conditions are included in the environmental criterion. Accordingly, the environmental criterion in the CSI includes both ecological hazards and certain types of hazards that may ultimately affect human health. For example, HAPs, HWPs, and VOCs are not included as environmental hazards in the GHS, but are considered in the CSI because they relate to the health hazards to which populations may be exposed in the event of a release and off-site migration of components through environmental media. “Endocrine disruptors” were included in the CSI, but not in the GHS, based on a concern for potential endocrine-disruptive properties of product components to humans and wildlife. The inclusion of these additional environmental hazards in the CSI allows for a more thorough identification of potential hazards and is intended to result in a comprehensive hazard identification that is appropriate for a typical operational setting in the oil and gas industry.
The physical hazards evaluated by the CSI mirror those defined by the GHS, with two minor exceptions. The GHS considers “explosives” more broadly by including “pyrotechnic components”, while the CSI evaluates these as separate physical hazard categories (Table 1
). Also, the GHS “flammable aerosols” hazard is not included in the CSI because these kinds of components are not typically used in the oil and gas industry.
Chemistry Scoring Index (CSI)—Hazard Categories.
Chemistry Scoring Index (CSI)—Hazard Categories.
|Environmental Criterion||Physical Criterion||Health Criterion|
|Acute Aquatic Toxicity (Category 1, 2, and 3)||Explosive||Carcinogenicity |
(Category 1 and 2)
|Chronic Aquatic Toxicity (Category 1, 2, 3, and 4) 1||Pyrotechnic||Mutagenicity|
|Biodegradation—Persistent 2||Flammable Gas||Reproductive Toxicity|
|Biodegradation—Inherent 2||Oxidizing Gas||Sensitizers|
|Bioaccumulation||Gases Under Pressure||Acute Toxicity |
(Category 1, 2, 3, and 4)
|Ozone Depleting Substance||Flammable Liquid (Category 1, 2, 3, and 4)||Corrosivity (Category 1 and 2)|
|Volatile Organic Compound (VOC) 3||Flammable Solid||Acute Target Organ Toxicity|
|Hazardous Air Pollutant (HAP) 3||Self-Reactive Substance||Chronic Target Organ Toxicity|
|Hazardous Water Pollutant (HWP) 3||Pyrophoric (Liquids and Solids)||Aspiration hazard|
|Endocrine Disruptor 3||Self-Heating Substance|
Emit Flammable Gases in Contact with Water
Corrosive to Metals
While the CSI health hazards are largely based on the GHS, the CSI includes a number of modifications. The “reproductive toxins” hazard contains multiple categories in the GHS (i.e., reproductive toxicity categories 1A, 1B, and 2; there is also a separate category for effects on or via lactation), but is represented by only one category in the CSI, thus treating all reproductive toxins as equally hazardous. Similarly, the “mutagens” hazard contains multiple categories in the GHS (i.e., germ cell mutagenicity categories 1A, 1B, and 2), but is represented by only one category in the CSI, thus treating all mutagens as equally hazardous. The CSI has two categories for carcinogenicity: Category 1 encompasses the GHS carcinogenicity classification 1A and 1B, while Category 2 corresponds to GHS carcinogenicity classification 2. GHS “category 3 skin irritants (mild irritants)” are not scored in the CSI because some countries/regions have not adopted the class 3 category and they are of minimal public health concern. Also, “serious eye damage/eye irritation” (a distinct health hazard in the GHS) is evaluated jointly with skin corrosion/irritation under “corrosivity” in the CSI. And finally, the CSI does not include GHS “category 5 acute toxicants” because category 5 acute toxicants are considered to have very low acute toxicity and the GHS, based on animal welfare concerns, discourages animal testing in this range unless there is a reason to believe a chemical (at this dose magnitude) may pose a hazard to a vulnerable population. Overall, the modifications to the GHS scheme implemented in the CSI increase the likelihood of identifying a health hazard and therefore represent a conservative approach to hazard assessment.
The CSI approach allows sufficient flexibility that addition or removal of hazard categories could be relatively easily implemented by other users, if deemed necessary.
2.3. CSI Hazard Assignments Using LOLI
The CSI assigns hazards to product components based on their presence on a “Minimum Set of Lists” contained within the LOLI database (see Supplementary Material). The LOLI database is considered a highly reliable hazard data source because it is based on determinations made by regulatory authorities or other widely recognized authorities from around the world. Also, the LOLI database is updated on a weekly basis, and licensed users of the database are notified of changes/updates on a quarterly basis.
The Minimum Set of Lists was defined for the purposes of hazard assignments within the CSI. It contains lists that have been populated with chemicals in accordance with the GHS definitions for each of the GHS hazard categories. For example, the GHS defines an oxidizing gas as “any gas which may, generally by providing oxygen, cause or contribute to the combustion of other material more than air does as determined through specified ISO test methods.” The CSI implements this definition by identifying a number of regulatory lists in the LOLI database that categorize gases using this same definition, including lists published by authorities from the European Union, Japan and New Zealand (see Supplementary Material). Similarly, for non-GHS hazard categories (i.e., biodegradation/persistent, biodegradation/inherent, VOCs, HAPs, HWPs, and endocrine disruptors), relevant lists were identified within LOLI and included in the Minimum Set of Lists. The Minimum Set of Lists is updated annually to reflect changes to the LOLI database.
There are sometimes instances where various lists on the Minimum Set of Lists provide conflicting hazard classifications for a particular component. For example, a component may be listed as a hazard on a list issued by one country but not another list issued by a different country, or different lists may categorize components in different (higher or lower) categories for the same hazard (e.g., carcinogenicity category 1B versus
carcinogenicity category 2). In such cases, the CSI conservatively scores the component in the more stringent hazard category unless there is compelling evidence to suggest that the hazard assignment is inappropriate. These determinations require professional judgment using the weight-of-evidence approach outlined in the GHS [12
]. When decisions on hazard assignment are made in this manner, the ultimate basis is always documented in the CSI product scoring materials to aid in transparency and allow for future reference. Note that weight-of-evidence decisions are typically not needed when the hazard assignment is based directly on analytical measurements (e.g., “bioaccumulation” assignment based on an experimentally-derived bioconcentration factor or octanol-water partitioning coefficient).
While the CSI relies on the Minimum Set of Lists and LOLI as its primary data sources for making hazard assignments, it also provides users with the flexibility to rely on additional non-LOLI data sources to make hazard assignments, such as internal toxicity or other test data, additional database sources not included in LOLI, SDS information, etc. In such cases, the user is encouraged to select sources with the highest reliability. The same weight-of-evidence approach used for LOLI data sources, including documentation of expert judgment decisions, is used when non-LOLI data are used to assign hazards. While currently not implemented in the CSI, a future version may incorporate data reliability flags associated with component hazard assignments, and ultimately, confidence levels associated with CSI product scores.
When hazard data are lacking for a product component, the component is assigned to the “No Data Available” category. However, for several hazard categories (i.e., ozone depleting substances, endocrine disruptors, VOCs, HAPs, and HWPs), absence of a component from the Minimum Set of Lists shown in the Supplementary Material is used to classify that component as “No Hazard Identified” (referred to as “No Hazard” hereafter). For example, if a component is not on the VOC Minimum Set of Lists, then that is not considered a data gap, but evidence that it is not a VOC.
2.4. CSI Hazard Scoring Using the CSI Scoring Matrix
Following hazard assignment, the CSI assigns hazard scores based on the percent composition by weight of each component in the finished product using the CSI scoring matrix (Table 2
). The matrix assigns numeric values for hazard categories that vary based on (1) nature and severity of the hazard; and (2) the concentration of a particular component within a product.
Within each hazard criterion, professional judgments were made regarding the relative importance of certain hazard categories by assigning different maximum scores to different hazard categories. In general, the hazards judged as most significant under each of the criteria are assigned a maximum value of 100 points (Table 2
). For example, for the health criterion, the maximum score is assigned to known human carcinogens and category 1 acute toxicants. Similarly, the maximum score is assigned to components with acute aquatic toxicity in category 1 in the environmental criterion and components that are considered explosive and organic peroxides in the physical criterion. While these categories represent very different hazards to potentially different receptors, these endpoints were judged to be of similar impact in terms of evaluating intrinsic hazard in the CSI. Under the environmental hazard criterion, the highest maximum scores are given to the categories that describe toxicity rather than component properties that may or may not result in subsequent toxicity, such as biodegradation or bioaccumulation. For the physical hazard criterion, the highest maximum scores are assigned to the explosives and organic peroxides, as components in these categories have the greatest potential to endanger physical safety. For human health, higher scores were given to categories that have the potential for more significant adverse outcomes (e.g., fatality, cancer, birth defects). Components for which data are lacking (“No Data Available”) for all hazard categories within a particular criterion are conservatively assigned a maximum value of 100 points for that criterion, thereby assuming that they may be as hazardous as the most severe hazard categories. This approach ensures that potentially hazardous components that are unreported, or for which hazard data are not available, do not result in a favorable score due to this lack of information.
The same types of relative scoring judgments also extend to the maximum scores assigned to hazard categories within a given hazard criterion. While the maximum score of 100 is assigned to the hazard(s) within each criterion judged to have the greatest impact, and to the “No Data Available” category, other hazards within a criterion are assigned lesser maximum scores. For example, the maximum score assigned to a known or presumed human carcinogen is 100, which is 10 times the maximum score assigned to an irritant. By assigning higher values to hazard categories of greatest impact or components for which data is lacking, components which have a mode of action that results in serious adverse effects under multiple categories or for which hazard data are unavailable will be highlighted by receiving a higher hazard score.
As noted, the scores assigned by the CSI scoring matrix also vary based on the concentration of a component in a product. The CSI weighting matrix assumes a defined quantitative relationship between the concentration at which a component is found in a product and degree of hazard. For example, in the environmental hazard category “acute aquatic toxicity category 1”, the CSI tool assumes that a component which makes up 5%–9.9% of a product is five times more hazardous than a component present at greater than 0.1%–0.9% (Table 2
). This represents a simplified approach given that a defined quantitative relationship between a component’s percent composition and degree of hazard is a function of component-specific toxicity information and properties. Nevertheless, the generic approach taken in the CSI is preferable to assuming that a component’s hazard is equal at all concentrations. Furthermore, this approach allows for distinctions to be made between products that contain a high percentage of a hazardous component and those containing a low percentage of the same component.
The GHS also considers the concentration of a component in a product but relies on generic percent composition cut-offs/concentration limits below which components are not considered hazardous. The cut-off values used in the GHS are 0.1%, 1%, or higher (for irritants), depending on the health and environmental hazards (cut-offs are not used for physical hazards). In contrast, the CSI evaluates and assigns a score to every product component regardless of its concentration in the product. However, that score may be zero if the percent composition is below the percent-based threshold for the hazard category or if the component was assigned “No Hazard” for all categories in a hazard criterion (Table 2
). For example, carcinogenicity is not considered in the GHS at concentrations below 0.1%, whereas a component present at a concentration of less than 0.1% and classified as a carcinogen category 1 would receive a score of 25 in the CSI. Therefore, the CSI approach is more inclusive in assigning hazards in the low concentration range than is the GHS approach.
Chemistry Scoring Index (CSI)—Hazard Scoring Matrix.
Chemistry Scoring Index (CSI)—Hazard Scoring Matrix.
|Hazard Criteria||Hazard Categories 1||Max. Score||Product Component Percent Range (by weight)|
|ENVIRONMENTAL 2||No Data Available 3||100||10||25||50||75||100||Do not evaluate||Do not evaluate|
|Acute Aquatic Toxicity Cat. 1||100||1||5||10||25||50||75||100|
|Acute Aquatic Toxicity Cat. 2||75||0||1||5||10||25||50||75|
|Acute Aquatic Toxicity Cat. 3||50||0||0||1||5||10||25||50|
|Volatile Organic Compounds||50||5||10||50||50||50||50||50|
|Hazardous Air Pollutants||50||1||5||10||25||40||50||50|
|Hazardous Water Pollutants||50||1||5||10||25||40||50||50|
|No Hazard 4||0||0||0||0||0||0||0||0|
|PHYSICAL 5||No Data Available 3||50||0||5||10||25||50||Do not evaluate||Do not evaluate|
|Flammable Liquid Cat. 1||75||0||5||10||25||50||75||75|
|Flammable Liquid Cat. 2||50||0||1||5||10||25||50||50|
|Flammable Liquid Cat. 3||25||0||0||1||5||10||25||25|
|Flammable Liquid Cat. 4||10||0||0||0||1||5||10||10|
|Pyrophoric (Liquids and Solids)||75||1||5||10||25||50||75||75|
|Gases Under Pressure||25||1||5||25||25||25||25||25|
|Emit Flammable Gases in Contact with Water||10||0||0||1||1||5||10||10|
|Corrosive to Metals||5||0||0||1||1||5||5||5|
|No Hazard 4||0||0||0||0||0||0||0||0|
|HEALTH||No Data Available 3||100||10||25||50||75||100||Do not evaluate||Do not evaluate|
|Carcinogenicity Cat. 1||100||25||100||100||100||100||100||100|
|Carcinogenicity Cat. 2||75||10||75||75||75||75||75||75|
|Acute Toxicity Cat. 1||100||10||25||50||75||75||100||100|
|Acute Toxicity Cat. 2||75||5||10||25||50||50||75||75|
|Acute Toxicity Cat. 3||50||0||1||5||10||25||50||50|
|Acute Toxicity Cat. 4||10||0||0||1||5||5||10||10|
|Acute Target Organ Toxicity||50||1||5||10||25||25||50||50|
|Chronic Target Organ Toxicity||50||1||5||10||25||25||50||50|
|Corrosivity Cat. 1||25||0||1||5||5||10||25||25|
|Corrosivity Cat. 2 (Irritant)||10||0||0||1||5||5||10||10|
|No Hazard 4||0||0||0||0||0||0||0||0|
In the CSI, an environmental hazard score, a physical hazard score, and a health hazard score are calculated based on each of the product’s individual components. As discussed earlier, these three hazard criterion scores for the product incorporate scores given to each component, including components for which information was not available to assign hazards (i.e.
, assigned “No Data Available” in Table 2
). This ensures that products are not identified as non-hazardous solely based on a lack of data. The environmental, physical, and health hazard criterion scores are added together to compile a total hazard or CSI score for the product. The three hazard criteria are weighted equally when added together to generate the final CSI score; i.e.
, no hazard criteria score is added in multiples when deriving the total hazard score for a product. However, caution must be exercised in comparing the health hazards of a particular product to its physical hazards or environmental hazards given that the weighting matrix and the CSI were not designed to facilitate such comparisons.
Overall, the CSI is designed to allow all of the hazards associated with one product to be compared with all of the hazards associated with other products in the same product use group. Products that score lower within a product use group are considered to have a lower intrinsic hazard compared to other products within the same use group that have higher scores, as illustrated below by a CSI example analysis for five products.
A range of sensitivity analyses were conducted on the CSI scoring matrix to evaluate the robustness of the overall relative CSI scores within a product use group (full analyses available upon request). The sensitivity analyses were conducted using eight product use groups, consisting of a total of 127 products. As part of the sensitivity analysis, several alternative approaches to scoring were evaluated. For example, we evaluated a binary scoring system (i.e., if a chemical is assigned to a hazard category, it is given a score of 1; otherwise, it is given a score of 0), different approaches for addressing components for which no data are available, use of combined hazard scores for components assigned to the same hazard category, and assigning differential weights to the individual CSI criteria scores to derive a combined total score. The impact of each approach on scoring and ranking within each product use group was then evaluated. These analyses showed that the different evaluated approaches did not significantly affect CSI scores and relative ranking of products within the use groups evaluated. That is, products that were either significantly more or less hazardous than the average product within a use group were relatively unaffected by slight modifications to the weighting matrix.