1 Introduction

It is worth beginning by reporting that there has been a lively discussion in the UK over the past two years about risk tolerability criteria which has culminated in the publication of 'Reducing risks, Protecting people' (Ref. 1) which describes the UK's Health and Safety Executive' decision making process. This publication is an update of the hitherto standard reference 'The Tolerability of Risks from Nuclear Power Stations (Ref. 2) originally published in 1988 and revised in 1992.

It says in the introduction:

A major purpose of this document was to set out an overall framework for decision taking by HSE which would ensure consistency and coherence across the full range of risks falling within the scope of the Health and Safety at Work Act 1974.

2 Common Risk Tolerable Criteria

Tolerability to risk is clearly very subjective and involves many economic, ethical and social considerations. In everyday life, we make decisions about risk based upon a myriad of considerations; by no means all of which are conscious, let alone numerical. For example, a private individual making decisions which concern him and his immediate family rarely uses any formal criteria.

However, when the decisions affect a larger society, the decision making process in a democratic society is frequently called into public question and there is an increasing need to have criteria on which to make, and to justify, the decision. It must be stressed here that considerations of risk and adherence to numeric criteria are only seen as a help to the decision making process.

There are many ways in which human (i.e. life safety) risk can be measured. The most frequently used measures are:

• Individual risk
• Societal risk and
• Collective risk.

Each of these three is used to determine the level of risk in its own way and is used for different purposes. Individual and Societal risk are rather specific and are used for judgement whether safety aspects of the system in question are acceptable.

Collective risk, which is less specific than the first two, provides information on an average risk likely to occur due to a given system operations and can be also considered as a sub-class of societal risk (Ref.3). It is very often utilised in a decision-making process justifying the use of safety specific enhancements. The authors are unaware of any attempt, and there is probably no need, to set up acceptance criteria for collective risk.

When discussing risk, there are usually three zones.

• An unacceptable zone where the risks are too high and the activity should not be undertaken
• A tolerable (or broadly acceptable) zone where society tolerates the risk
• A zone in between where society has to make choices about the cost of reducing the residual risk

In this paper authors share their thoughts on, and show problems associated, with the tolerability criteria via the use of the most popular measures used when carrying risk assessments.

The pros and cons of using criteria to evaluate Individual and Societal risk are discussed below.

3 Individual Risk

Individual risk is relatively well understood. It is normally defined as the risk to a hypothetical member of the public either using a facility (say a road tunnel), living outside the boundary of another facility (say a nuclear power station) or participating in a particular recreational activity (say rock climbing).

The criteria for individual risk depend heavily on whether the activity is voluntary or not. For example, the risk of a fatality involved in rock climbing would be about 1 in 10 per year (intolerable risk) if carried out as a full time occupation. In contrast, the risk from an office-based job would normally involve a 'tolerable risk' of 1 in 500,000 per year.

The criteria for individual risk are really only applicable to activities where the person is carrying out the activity for non-recreational purposes, that is to say the risk is involuntary. It is interesting to consider whether travel by road or rail is involuntary or not. Some may argue that the decision to travel is voluntary. However, it would be difficult to live in 21st century western culture without travelling for work and refusal to travel would be likely to damage promotional prospects. For non-work activity, the convenience afforded by motor cars usually far outweighs any 'statistical' safety disadvantage.

The concept of individual risk incorporates average risk, which is defined as the total risk divided equally into those experiencing the risk. It is not a particularly useful concept as it is very rare for those at risk to share it equally.

Different nations arrive at roughly the same acceptability limits for involuntary individual risk in different ways and it is worthwhile exploring two approaches. In the UK, the risk of death for deep-sea fisherman is 1 in 1,000 per year. It is one of the riskiest professions in the UK and is generally recognised as representing the upper limit of tolerability.

Road users run an annual risk of death of around 1 in 10,000 per year. It must be acknowledged that some aspects of road usage involve a voluntary aspect, as many journeys are not strictly necessary. Another important aspect of road fatalities is that they tend to occur in ones and twos and so do not tend to attract much media attention. We will return to this conundrum later in the paper.