6.3.3 Approach used in derivation of guideline values for chemicals
The guideline value for each organic and inorganic chemical is the concentration that, based on present knowledge, does not result in any significant risk to the health of the consumer over a lifetime of consumption and is consistent with water of good quality.
The health-based guideline values are very conservative, and are calculated using a range of safety factors. They always err on the side of safety, particularly where scientific data are inconclusive or where the only data available are from animal studies.
Where aesthetic considerations, including taste and odour, corrosion, and stains on sanitary ware and laundry, dictate a more stringent guideline than that required to protect health, both values are quoted. Health considerations may be of less concern in such cases (although they must still be considered), because water that is aesthetically unacceptable is less likely to be consumed.
For most chemicals, it has not been possible to estimate the higher concentrations that would affect health over shorter periods, so short-term guideline values have generally not been set. However, given the very conservative nature of the Guidelines, deviations from the guideline values over a short period do not necessarily mean that the water is unsuitable for consumption. The amount by which and the period for which any guideline value could be exceeded without causing concern will depend on the chemical involved and other factors, such as the risks and benefits to public health.
Each excursion beyond a guideline value should, however, be a trigger for further action.
Chemicals fall into two categories based on health effects:
those where the effects are observed only above a certain threshold dose, with no effects observed at doses below this threshold;
those that do not appear to have a threshold.
Sources of data used
Human data
There is little information on the effects of human exposure to organic and inorganic compounds, including pesticides, at the concentrations likely to occur in water. Occasionally, there are useful epidemiological data, and where available, these have been the primary consideration in setting the guideline value.
Animal data
In the absence of human data, experiments on laboratory animals provide toxicological data on the effects of exposure to chemical agents. Ideally, these are long-term studies involving ingestion of the compound dissolved in water or present in food, rather than inhalation or dermal exposure studies. For expediency, such studies are conducted at concentrations that are relatively high in comparison to the concentrations likely to be found in drinking water. Furthermore, the most sensitive animal species, and the most sensitive group within that species, are used in order to increase the likelihood of observing a toxicological effect.
Effects of exposure to chemicals in experimental animals are generally classified in the following broad categories:
organ-specific;
neurological/behavioural;
reproductive/developmental;
carcinogenic/mutagenic.
Effects may be prolonged or short term, reversible or irreversible, immediate or delayed, single or multiple. The nature, number, severity, incidence and prevalence of specific effects generally increase with increasing dose. Adequately designed and conducted experimental studies in animals can usually provide an exposure level below which adverse effects are not seen.
Interpreting these data and extrapolating from them to human populations can be difficult, as health effects vary with dose, route of exposure (e.g. ingestion, inhalation, skin absorption), frequency or duration of exposure, and the species, sex and age of the exposed population. This can require appropriate expertise and prudent judgment (e.g. see IPCS 1978).
Derivation of guideline values for substances for which a threshold exists
Where appropriate human data are available, these have been used in the derivation of the guideline value.
In the absence of human data, the guideline value is generally based on the highest dose that causes no adverse effects in long-term experiments on laboratory animals. It is calculated using the following formula:
where:
In using this equation, it is necessary to make assumptions about the amount of water consumed per day, the average body weight and the proportion of total intake that can be attributed to water consumption, and to decide on an appropriate safety factor. Clearly the figures selected will all affect the guideline value, and varying one or more of them could raise or lower the resultant value by a factor of 10 or more. Any guideline value will thus have a degree of ‘fuzziness’ surrounding it; however, the assumptions made in calculating these guideline values are generally very conservative, and always err on the side of safety.
Animal dose
The animal dose is usually the ‘no observed effect level’ (NOEL); that is, the highest amount of the compound that does not cause observable effects in repeat dose studies on experimental animals. If this is not available, then the dose often used is the ‘lowest observed effect level’ (LOEL); that is, the lowest amount of the compound that causes observable effects in studies on experimental animals. (In some cases, the ‘no/lowest observed adverse effect level, NOAEL/LOAEL, is used – that is, the highest amount of the compound that causes no observable adverse effects, or the lowest amount that causes observable effects, in repeat dose studies on experimental animals). If the LOEL (or LOAEL) is used, an extra safety factor is applied.
The dose data can come from drinking water studies or feeding or force-feeding studies. Dose is expressed as milligrams of compound per kilogram of animal body weight per day.
Human weight
It has been assumed that the average weight of an Australian adult is 70 kg. This is the figure used in Canada and other developed countries. The World Health Organization (WHO) uses a value of 60 kg, which reflects the lower adult weights in developing countries. The heavier weight assumed here will slightly increase the magnitude of the guideline value.
Where there is a specific need to protect young children, the average weight of a child at 2 years of age is assumed to be 13 kg. The same figure is used in other developed countries, such as Canada. The WHO uses 10 kg.
Proportion of intake from water
The animal dose data are assumed to encompass all sources of exposure. It is thus necessary to estimate the proportion of total human intake of a compound that is derived from water. Intake from air is generally negligible compared with other sources, but intake from food, pharmaceuticals and other products can be significant.
For chemicals that are used commercially or industrially, it is assumed, in the absence of other information, that water contributes 10 per cent of intake. For compounds that are not used commercially or industrially, a higher proportion of intake (usually 20 per cent but sometimes 80 per cent or 100 per cent) is assumed to come from drinking water. These figures are regarded as conservative (assuming a higher proportion deriving from drinking water would result in raising the guideline value), and the approach is consistent with that adopted by the WHO and by other countries.
Although exposure to chemical agents in water is predominantly through drinking the water, skin absorption during bathing or inhalation in a shower can also occur. Such exposures may increase the proportion of the chemical derived from drinking water, but the lower proportion (10 per cent or 20 per cent) is used for calculating the guideline value because it provides a higher margin of safety.
Volume of water consumed
The amount of water consumed by an adult each day is assumed to be 2 L. If the guideline value is based on the weight of a child, 1 L per day is assumed. Consumption can vary with season and climate; however, both figures, which are the same as those used by the WHO, are believed to be appropriate, on average, for Australian conditions. Some colder countries use different values: Canada, for example, uses 1.5 and 0.75 L per day.
Safety factor
Safety factors are used because of the uncertainty inherent in extrapolating from animal studies to human populations, or from a small human group to the general population. Safety factors generally applied are:
a factor of 10 for variations between animals of the same species (because some animals within a species may be more sensitive to the effects of a chemical than the group tested);
a factor of 10 for variations between species (because the animal species tested may be less sensitive than humans, and in many cases human sensitivity is unknown);
a factor of 10 if data from a sub-chronic study are used in the absence of reliable data from chronic studies (this factor can be less if chronic studies are available and indicate that no other effects occur, or that other effects are mild);
a factor of up to 10 if adverse effects have been observed at the lowest doses (usually the data used are based on the highest dose at which no adverse effects are seen).
The individual factors for each of the points listed above are multiplied together to give an overall safety factor. A safety factor of 100 to 1000 is common; higher values may be used on occasions.
Occasionally, individual safety factors lower than 10 are used where there is additional information to justify a reduction. This can occur, for instance, where information is available to clarify the mechanism of the effects on humans, where human epidemiological data are available, where the adverse effects observed are regarded as being relatively minor, or where large amounts of animal and human data are available.
Guideline values for carcinogenic compounds that act only above a threshold dose are determined in the same way as for non-carcinogenic compounds, but with an additional safety factor for carcinogenic effects.
Derivation of guideline values for substances where no threshold has been demonstrated
With compounds for which no threshold can be demonstrated, it can be expected that, as the level of exposure decreases, the resultant hazard similarly decreases. The risk associated with exposure to very low concentrations may be extrapolated using a risk assessment model, often over many orders of magnitude, from the dose–response relationship observed at higher doses. A number of uncertainties are involved, but the calculations used tend to overestimate rather than underestimate the risk, and so provide a greater margin of safety: it is possible that the actual risk from exposure to low concentrations may, in fact, be lower than the estimated values by more than an order of magnitude.
This approach can be applied for genotoxic carcinogenic compounds, and has been used by the WHO for this purpose.
Benchmark dose (BMD) approach
In a few cases, a slight variation to the above approaches for setting guideline values has been used. This variation, known as the benchmark dose (BMD) approach, has been used in dealing with both cancer and non-cancer end points. It is described in Environmental Health Criteria 170 (WHO 1994) and a modified version for use with carcinogenic soil contaminants was described by the NHMRC (1999).
The benchmark dose corresponds to a predetermined increase (between 1 and 10% but commonly 5%) of a defined effect in a test population. Mathematically it is the statistical lower confidence limit on the dose that corresponds to that predetermined increase, although some agencies are using a best estimate rather than a lower confidence limit (IEH, 1999).
Guidance on rounding
The vast majority of numerical guideline values in the Guidelines are rounded to a single significant figure. Consistent with standard rounding convention, mid-way values are rounded up. For example, 1.5 is rounded to 2 and 25 is rounded to 30. Trailing zeros in numbers where there is no decimal point should not be taken as significant (e.g., nitrate, 50 mg/L).
Practically all of the health-based guideline values were established using data and assumptions with a precision of one significant figure (e.g., volume of water consumed by an adult = 2 L/day). Furthermore, the vast majority include the incorporation of safety factors, which are applied at the precision of ‘order of magnitude’ (e.g., 10 for interspecies extrapolation and 10 for intra-species variation).
Quoting more significant figures misrepresents the degree of calculated precision and may lead to unfounded concern when guidelines are exceeded at the second or third significant figure.
It is noted that exceptions to this may be necessary for some chemicals. These will be considered on a case-by-case basis and the reasons for the deviation from the convention of rounding to a single significant figure will be explained in the fact sheet.
It is noted that aesthetic guidelines are generally based on direct information on palatability to consumers, including appearance, taste and odour, and so do not need to be rounded.
Interaction between chemicals
Guideline values are calculated for individual chemicals without specific consideration of the potential for each to interact with others in the water. Normally, the majority of chemicals will not be present in concentrations at or near the guideline value, and the large margin of safety incorporated in the majority of the guideline values is considered to be sufficient to account for potential interactions with other substances.
Guidance on parent compounds and environmental transformation products
Fact sheets and health-based guideline values are established for the form(s) of the chemical that may be present in drinking water. Where the chemical form present in drinking water is an environmental transformation product, toxicological data on the transformation product(s) should be evaluated to derive a health-based guideline value if available.
Use of screening assays
For some chemicals, the accepted analytical method is a screening assay that is able to detect any of a group of chemicals, with the measured value derived from the members of the group that are present. For example, in analytical chemistry, a residue method involves breaking down the sample and measuring a breakdown product (for example, subjecting the sample to acid digestion and measuring carbon disulphide, CS₂). Also, some bioanalytical methods detect groups of chemicals based on their biological activity, for example estrogenic compounds. Using such methods, it may not be possible to determine the parent compound(s) present in the original sample. In this case, the result should be expressed in units of the most potent compound (or a well-established method reference compound) and compared to the relevant guideline value, or follow-up analytical techniques should be applied that measure the chemical(s) of interest directly.
Last updated
