Human health hazard assessment
Last update 30 November 2016
Human health hazard assessment establishes the toxicity of the chemical and identifies the set of inherent properties that make it capable of causing adverse health effects.
For chemicals with unknown toxicity (for example, new chemicals), this involves a series of animal studies investigating major biological systems, including studies on acute toxicity, repeat dose toxicity, genotoxicity and other specific endpoints such as irritation and sensitisation.
For existing chemicals, in addition to animal data, human health effects data may be available.
We assess both human and animal data in accordance with international guidelines to identify the critical health effects of the chemical and determine the dose–response relationship, with No Observed Adverse Effect Levels (NOAELs) established wherever possible.
We prefer to use good quality human data for risk assessment. We classify the chemical's health hazards in accordance with the Approved Criteria for Classifying Hazardous Substances and the Globally Harmonised System of Classification and Labelling of Chemicals.
The toxicological data may consist of studies performed with a structural analogue of the chemical, or with a formulation containing the chemical.
We take data adequacy and applicability into account when assessing available data (for example, concentrations tested in toxicological studies).
Where data gaps exist, or the notification does not require toxicology data (as with some classes of polymer), we may be able to predict the toxicological hazard from the chemical's physical properties or the characteristics of structurally related chemicals, given that factors such as volatility, water solubility and molecular weight can indicate the likely extent of absorption across biological membranes.
For existing chemicals, structurally similar chemicals may be grouped into a category based on structure and physical chemical properties, and a read across methodology applied where data from one chemical is used for a data-poor chemical in the category.
Quality of data
To ensure data are of sufficient quality for use in risk assessment, we require that all new testing must be conducted according to internationally recognised methods, for example, the Organisation for Economic Co-operation and Development's Test Guidelines and Good Laboratory Practice standards.
For many existing chemicals, data will have been generated before these guidelines and standards were created, and may be able to be sourced from academic publications using non-standard methodology. These data can still be used for assessment if valid conclusions can be drawn from them.
Evaluation and assessment requires expert judgment, and determining validity has to be both justified and transparent. In determining the quality and validity of data, matters such as completeness and scientific detail in test reports must be considered.
Relevance of data
When assessing chemicals, we consider the relevance of test data by, for example, judging if the appropriate route of exposure was used, if the most suitable species was studied and if the substance tested represented the chemical being assessed. We also consider the relevance of animal and in vitro test data for humans.
Our assessment also considers toxicokinetic and metabolism data for the chemical, in animals and humans, if available. Information on the physicochemical properties and chemical structure can be used to make predictions on the absorption, distribution, metabolism and excretion of substances. For example, physico-chemical parameters can inform the potential to cross biological membranes.
Generally, we assume that effects observed in animals occur in humans unless there is clear, well-documented evidence for a species-specific effect that would justify concluding that the effect could not occur in humans or is of little relevance.
We accept non-animal alternative test methods if they are included in the OECD Test Guidelines and are scientifically validated and have received regulatory acceptance (for example, EpiDerm or Episkin skin corrosivity test, Bovine Corneal Opacity and Permeability test, Isolated Chicken Eye test).
In vitro data alone are generally not directly predictive for effects on humans. However, highly electrophilic substances, which give positive results in genotoxicity studies in vitro, may be of concern for their potential to be mutagenic in humans at the initial site of contact (for example, the skin or respiratory tract).
Evaluation of human data
The evaluation of human data generally requires more critical appraisal of data validity.
The main types of human data are epidemiological studies, controlled studies in volunteer case reports and, in the case of sensitisation, multi-clinic data.
The strength of epidemiological evidence for specific health effects depends on matters such as the type of analysis and the magnitude and specificity of response.
Our confidence in findings is increased when comparable results are obtained from at least two independent studies on populations exposed to the same chemical under different conditions.
Criteria for assessing the adequacy of epidemiological studies include the proper selection and characterisation of the exposed and control groups, adequate characterisation of effect and exposure, sufficient length of follow up for disease occurrence, adequate control for confounding factors and proper statistical analysis.
Controlled human studies can be used by us in determining exposure levels associated with acute effects such as skin irritation. Human patch tests for skin sensitising effects can also be conducted.
Criteria for a well-designed study include using a double blind study design, including a matched control group, using a sufficient number of subjects to detect an effect and taking confounders and bias into account.
Epidemiological studies with negative results cannot prove the absence of a particular toxic effect of the chemical in humans, but good quality controlled human studies that are negative may be useful in assessing risk. Negative human data for skin-sensitising effects cannot normally be used to negate positive results from animal studies.
Evaluation of animal and in vitro studies
Most health effects information required for risk assessment will be derived from controlled studies in experimental animals and in vitro test systems.
We need to identify the adverse effects of the chemical in these studies, and judge how well the studies identify particular effects.
Generally, we need to judge if a study establishes a dose or exposure level at which the critical effect is not observed. For repeated dose studies, a NOAEL should be established or, where this is not possible, a Lowest Observed Adverse Effect Level (LOAEL) stated.
For each study, it is important for us to evaluate the study design and how the study was carried out, considering matters such as frequency and duration of exposure, appropriateness of species and strain of animals used, route of exposure and choice of doses.
When evaluating data in each study, we consider matter such as the effects in treated animals compared to control animals, causes of mortality, clinical observations during exposure, organ and body weight changes, biochemical changes, mode of action and relevance of effects on humans.
Dose response assessment
The international community generally agrees that there is a threshold dose or concentration for many adverse health effects caused by chemicals.
The threshold dose may vary considerably for routes of exposure and for different species because of differences in toxicokinetics and possibly mechanisms of action.
The observed threshold dose in a toxicity test is influenced by the sensitivity of the test system, that is, it depends on exposure concentrations and durations used in the study.
For genotoxic carcinogens, it is generally a given that thresholds cannot be identified, unless a threshold mechanism is demonstrated.
When a reliable dose–response relationship is identified, then the slope of the curve is taken into account. For a steep curve, the NOAEL is more reliable, as the greater the slope the greater the reduction in response to reduced doses.
For a shallow curve, the uncertainty in the NOAEL may be higher and must be allowed for when assessing risk.