The need for further air quality improvement can be illustrated by reference to the results of the cost-benefit analysis of the EU’s CAFE (Clean Air For Europe) Programme (see Café Site). This shows that the total loss of life annually currently linked to particle exposure in the EU25 is of the order 2.5 million life years, despite existing legislation having led to a major reduction in exposure to pollution over the last 50 years.


In Europe this legislation takes several forms, including environmental quality standards, emission standards, fuel quality standards and fiscal incentives, such as differential fuel duties. This paper is mainly concerned with the future development of air quality standards. It is first necessary to understand that current standards do not represent no-effect thresholds, as thresholds have not been identified for the pollutants currently of most concern.


Instead, they represent levels that have been considered by experts to represent an acceptable level of risk, and can feasibly be achieved at reasonable cost.


Given that this has been done on a pollutant by pollutant basis, there is inconsistency between the standards with respect to the level of protection offered and the costs of compliance. It is also necessary to understand that current standards have been set for a series of pollutants individually, with limited attention paid to balancing risk across several pollutants taken together.


What is proposed here is a framework which aims to reduce overall risk from air pollution rather than the risks from a series of individual pollutants by prescribed amounts. This would offer greater flexibility in meeting standards, improving cost-effectiveness for achieving desired health and environmental outcomes. This, in turn, makes it possible to go further in protecting health and the environment.

When considering how future air quality standards should develop, many issues need to be considered. The approach proposed here seeks to take an integrated approach across pollutants in the following areas:


  • The costs and efficiency of pollutant control options;
  • The range of effects linked to each pollutant;
  • The risk per unit exposure.

At its simplest, the risk model would take a set of pollutants considered to have broadly similar types of effect, for example:


  • ‘Traditional’ air pollutants – particles, SO2, NO2, ozone, etc., or
  • Carcinogens – benzene, PAHs, nickel, cadmium, etc.

    Risk would then be assessed across an effect commonly reported across the group of pollutants (mortality would seem to be the obvious starting point for the ‘traditional’ pollutants listed above, and cancer incidence for the carcinogens). Each pollutant in a mixture would be weighted by the product of concentration at a given location and available concentration-response function (CRF) for the effect of concern:

    Overall_risk = [ConcPM10 x CRFPM10] + [Conc03 x CRF03] + [ConcSO2 x CRFSO2] + …


    This could be expanded in several ways, the following ranked more or less in order of added complexity:

  • Differentiate the effects of different pollutants according to knowledge of the severity of impact. For example, for carcinogens it may be considered appropriate to weight not just by the probability of getting cancer from a particular pollutant, but also by the probability of dying from that cancer.
  • Differentiate between types of particle: This is problematic because current epidemiology does not provide risk functions for different types of particle. However, expert judgement could be used to move in this direction.


  • Introduce population weighting to the equation. This would increase focus on the greatest good for the greatest number, rather than protection of those at greatest risk as the current system of air quality standards is focused on those living in areas with the highest concentration. This may be a small fraction of the population, and it is quite possible that greater benefits to society as a whole could be achieved through looking at exposure of the whole population. This leads to some conflict in terms of equity (should we accept a situation where some individuals are at significantly greater risk than others?) and social efficiency (should we target legislation on gaining the greatest improvement across society as a whole?).


  • Combine the risk assessment across groups of pollutants (e.g. combine the ‘traditional’ pollutants with the carcinogens). This would require a means for weighting different types of effect against one another, though techniques are already available using (for example) economic methods, multi-criteria assessment or the DALY (disability adjusted life year) approach.


  • Add in effects other than mortality and cancer incidence, for example, hospital admissions for respiratory ill-health, GP visits, asthma attacks and so on. Again, economic techniques, MCA or DALYs could be used.


    To illustrate further, Table 1 provides a worked example where the risk posed by four carcinogens at a specific location for which air quality data are as shown, is aggregated. Risk rates shown are adjusted pro-rata from lifetime risk to annual rates to facilitate comparison with cost data which are typically expressed on an annualised basis.

    Table 1. Aggregation of concentration and risk data for a hypothetical location.

    Pollutant Conc µg.m-3 Annual risk factor cancers/[million people.µg.m-3] Concentration x risk factor cancers/million people
    Benzene 5 0.07 0.36
    PAHs 0.001 1243 1.24
    Arsenic 0.02 21 0.43
    Cadmium 0.005 26 0.13
    Nickel 0.03 1.43 0.04
    Total     2.20

    From the table, it might seem at first sight appropriate to target benzene, as it is the pollutant present at highest concentration (column 2). However, the risk per unit exposure is quite low for benzene (column 3) with the result that we see in column 4 that the greatest risk at the location in question (column 4) is associated with PAHs, the pollutant present at lowest concentration.

    However, most people would not be concerned with the risk associated with each pollutant individually. Instead, they would be concerned primarily with the overall risk of getting cancer. On this basis, they would focus on the total risk shown at the bottom of the table. For setting a standard they would want to know how much it would cost to reduce this risk by varying amounts. The response taken would then set about introducing the most cost-effective means for reducing total risk to an acceptable level. This freedom means that action can be targeted on the most efficient ways of reducing overall risk. In the present system action could be targeted on a pollutant simply because it is present at levels above its standard, not because it is the pollutant (or one of the pollutants) whose control offers the greatest benefit at least cost.

    Whilst the proposed system is new in the context of ambient air quality standards and local air quality management, it is largely an extension of methods in cost-benefit analysis, integrated assessment modelling and regulation of industry. For example, in many ways the approach outlined here is not that different to the way that much legislation is currently developed at the European level. Under the CAFE (Clean Air For Europe) programme of European Commission DG Environment, for example, cost-benefit analysis integrates risk across a series of pollutants within a unified framework to help define optimised emission control strategies.


    Another example is provided by the Environment Agency’s H1 Methodology for IPPC assessment. This guides users in making decisions on what constitutes BAT for specified industrial facilities by balancing risk across a large number of burdens, including emissions to air, land and water, noise, accidents, visual impact, odour and waste.

    In conclusion, the proposed system has a number of advantages over current practice for setting environmental quality standards:

  • It would increase cost-effectiveness of controls.
  • It equates a reduction in pollution directly with health improvement.
  • It introduces greater flexibility to air quality risk management.
  • It would permit better consistency in the treatment of pollutants.

    The possible drawbacks of the approach are:

  • There could be some loss of transparency in going from directly measurable standards to computed standards. However, the current system, based on several different averaging periods and with a number of permitted exceedences for short-term averages that varies between pollutants is possibly far less transparent.
  • Limited availability of data to permit ready definition of the risk attributable to each pollutant in a consistent fashion.
  • The subjectivity of weighting systems (if considered necessary). However, the current system is also subjective in parts, for example, through the need to define ‘acceptable levels of risk’ for a series of pollutants.
  • The need to periodically redefine the risk attributed to each pollutant as new data become available. However, this could be said to apply to any regulation or methodology.

    Our feeling is that the advantages listed here substantially outweigh the possible disadvantages, several of which could be disputed in any case. It is stressed that the paper does not dispute the value of past and current legislation that has led to substantial improvements in air quality. The system proposed here is designed to complement that work, not to replace it. A good place to start would be consolidation of the air quality standards for carcinogens.

    By Mike Holland of EMRC.

    Mike.holland@emrc.co.uk

    http://www.emrc.co.uk

  • Action inspires action. Stay ahead of the curve with sustainability and energy newsletters from edie

    Subscribe