Traditional methods of risk assessment tended to focus on singular exposures and health effects.  Many of the issues that are the focus of integrated environmental health impact assessments, however, concern more complex situations, where people are exposed to multiple hazards either in succession (e.g. on different days or at different times of life) or in combination (i.e. as mixtures).  Examples of exposure mixtures include the classical situation of multiple chemicals, but also the combination of chemical and other exposures (e.g. ozone in outdoor air and allergens in the home). 

It is not easy to estimate the effects of mixtures, for the necessary data on the collective relationships between exposures and response (i.e. exposure-response functions for mixtures) are often lacking.  Various methods have, however, been devised to allow for these types of exposures in epidemiological and toxicological studies, and to a varying degree these can be used as a basis for health impact assessment. 

Indicators

One method that has been widely used in epidemiological studies is to select an individual indicator (e.g. pollutant species), which characterises the mixture of interest.  Ozone might thus be used as an indicator for photochemical smog; PM₁₀ is used to represent the wide variety of atmospheric particulate matter; trihalomethanes (THMs) are employed as a proxy for the much wider mix of disinfection by-products that occur in drinking water (see example); distance from source has been widely used as an indicator of exposure to complex pollution mixtures (e.g. from road traffic, landfill sites). 

Using indicators is simple and has many advantages, but the approach clearly works well only when the indicator represents the mixture equally well in all the different settings of relevance (i.e. if it has a close and consistent relationship with the exposure mixture), which is not always the case.  It is therefore sometimes better to use a number of indicators, selected to represent different sources or environmental processes and settings (e.g. PM₁₀, N0₂ and O₃ as indicators of general air pollution).  This has a danger, however, for it can result in ‘double-counting’, because each indicator is being treated as though it had an independent (and additive) effect, whereas in practice at least part of the apparent effect seen in epidemiological studies might be shared.  More generally, it always has to be remembered that the species used as the indicator might not be (and probably is not) the causative agent.  This can have serious limitations when indicators are used in impact assessments, because it is likely to be misleading to assume that changes in the exposure indicator (e.g. under a policy scenario) would imply changes in health impact.

Interaction effects

Another well-established method in epidemiological studies is to allow for interaction effects in the quantitative analysis.   Ideally, this is done by designing the studies specifically to detect interaction effects - e.g. by studying populations variously exposed to the pollutants both separately and in combination.  This is often difficult to do, however, so more commonly some form of post hoc analysis is done, by testing for interactions in the statistical analysis - though the power of this is nevertheless limited by the small sample size of many studies. 

Because of these difficulties, detailed information on interactive effects is rare.  It exists for a few, well-established mixtures - e.g. asbestos and smoking, radon and tobacco smoke, diesel and some allergens - but in most cases the information needed to allow for interactive effects in health impact assessments is lacking

Toxicological approaches

For practical reasons it is often easier to explore interactive effects using toxicological methods, where greater control can be exerted on the exposure mixtures being studied.  Interactions can then be quantified in the form of relative potency factors, of which toxicity equivalence factors (TEF) are a special case.  This approach is most azppropriate for the well-defined class of agents that operate through a common mode of action for the same health outcome.  Examples include the relative potency factors for some carcinogenic polycyclic aromatic hydrocarbons (PAHs) and the TEF for dioxin-like compounds.  The difficulty, however, is in translating these measures into dose-response functions that can be applied to large and heterogeneous population groups, and varied exposures, as part of an impact assessment (see example of THMs in drinking water). 

Biomarkers

Biomarkers  offer a further approach to assessing effects of mixtures.  Different types of biomarker may be useful in this respect.   Biomarkers of exposure are useful when exposures occur though multiple environmental media, such as air, food, drinking water and absorption through the skin.  These allow the collective exposure through all the different routeways to be assessed, without the need to measure or model each one separately.  One example where this has been documented is for lead (Pb): exposure-response functions for lead are thus typically formulated as the effect (e.g. on learning abilities) per unit concentration of lead in blood, which integrates the different exposure routes.  Biomarkers of early, physiological effect can also be used.  These are especially important when a large number of components are present in the exposure mixture, such that their interactions are too complicated to be modelled explicitly. 

While biomarkers are valuable as part of the epidemiological or toxicological analysis used to develop understanding of dose-response relationships, however, their use in health impact assessments is inevitably more limited, because of the need (in many assessments) to estimate effects of changes in exposure in response to some form of intervention (e.g. a change in policy).   To enable this,  models are also needed, indicating how the measured levels of the biomarkers might vary under the assessment scenarios.  This, in turn, usually implies an ability to model the different sources and exposure pathways.