Towards risk coherence

Background

We live in an age of risks, where the aim is to be aware of the various risks in our environment, and to control and manage our personal and societal risk levels. Risk management may indeed be considered to be a desirable attribute of modern industrialized society. Yet the information concerning these risks, from all quarters, is rarely coherent. What this means is that it is difficult to compare risks from different sources, and even more difficult to make trade-offs between these different risks. It is therefore difficult, at a personal and societal level, to manage risks in a coherent fashion. The result may be incoherent risk strategies, whether at a personal or societal level.

Risks tend to be assessed within an industry, using metrics that have evolved within that industry via the companies involved and their respective regulators. The various metrics used in one industry are not always compatible with metrics from other industries. Furthermore, as concern over environmental issues increases generally in society, there is a need to consider trade-offs between risks and environmental damage. For example, an environmental campaigner might argue for less air travel due to noise and pollution, but less air travel might lead to more travel by road and rail. There could thus be a net increase in risk (fatalities), and even a net increase in environmental harm. But there is rarely a proper interface between these two considerations (risk of injury/death and environmental harm) even though they both ultimately impact on quality and quantity of life, and even though they are both considered during planning phases for large new installations or transport systems.

Furthermore, in more ‘local’ situations, there is much information on risks associated with food, medical matters, and other areas, but the amount and variety of information makes it difficult to analyze these risks coherently. Some members of the public no doubt feel bewildered by the various statistics and counter-statistics, the pronouncements of experts and other experts who disagree with them, and the ‘scientific’ pronouncements of government ministers which subsequently turn out to be completely wrong.

It is therefore difficult for society, and the public, to judge and balance relative risks in the environment. The workshop aimed to explore the coherence of the communication of different types of risk and aims to facilitate comparative judgment, and determine what can be done to achieve a greater ‘risk coherence’. With more coherence there is a potential for a more balanced approach towards risks and risk reduction, leading to a safer and healthier environment.

Note: the workshop focused on involuntary risks, as discussed below, and not voluntary risks associated with personal lifestyles etc. Ideally even such risks would be included, but this would probably have stretched the workshop beyond an ability to derive useful conclusions and coherent insights.

Below, some key issues are considered, leading towards a specification of what type of expertise is required for the workshop.

A risky existence…

When you get up in the morning, risk questions very quickly materialize, at breakfast, for example. How many eggs is it safe to eat in a week? Do they increase or decrease cholesterol? Does the breakfast contain any GM soya, and does it matter? How should one travel to work? If one has to go to a conference abroad, which is the safer method of transport? Energy is used all the time. Which is the safest and cleanest form of energy production? How safe are the vaccinations for one’s young daughter? What are the impacts of living in the middle of a big, polluted city? And so on.

The long list of questions is matched by a bewildering array of statistics, from a range of sources. The statistics themselves come in all shapes and sizes, but all of them seem to say that their product or process is safer than any other, and there is generally nothing to worry about. But they can’t all be equal risk, can they? And what of exposure rates? These are not usually factored into readily available statistics, but the exposure rate often dramatically affects the real risk level.

Public perception of risk: bias, or poor information?

Previous work suggests that people’s risk perceptions and preferences are biased, for example against nuclear power. But what if this is more due to a lack of coherent information on relative risks of different energy production systems? Can the public be expected to make a reasoned judgment if the reasoning material is not available? Is the public really biased or does it make another judgment? Do we ignore them, attempt to change them, or allocate risk-reduction resources with these judgments in mind? The latter options do not appear to have been much considered.

Sources of risk information

There are numerous sources of risk information. Firstly there is information produced by the manufacturer or producer or corporate body associated with the product or process. This obviously will put things in a positive light. For some of these industries, there may be “opposition” groups, locally formed or more widely organized, who aim to present a different picture of the risks to the public. Then there are government sources, from various ministries, for food, health, or bodies regulating the safety and environmental impacts of various industries. Such governmental bodies aim to provide public information on risks, but some of them may have a vested interest in the activity or the product. Then there are the media organizations, which occasionally aim to highlight new risks, or significant changes in established risks, and which also sometimes aim to expose organizations which are perceived by the media to be mis-representing the risks to the public. So, who do we believe?

Trust in sources

It is difficult to know which of the above sources to believe, as they may not agree. Even the governmental sources sometimes get it wrong, as BSE in the UK dramatically showed. Eggs are another good example, since over the past few decades advice has ranged from eating many, to only 4 per week to avoid cholesterol problems, to 2 per week (well-cooked!) during the salmonella scare, and now more recently to new evidence suggesting that perhaps eggs actually can reduce cholesterol! How can a source, whose frequent unreliability is only matched by its continual supreme confidence, be trusted?

And as for the company, the “anti” lobby groups, and the media, they all have clear interests (if only in selling papers!), so some degree of distortion can be expected.

Trust in data

Distortion is enabled by the imprecision of risk predictions, particularly for very low risks. Such risks include uncertainty ranges, and different sources can quote from anywhere in these ranges, which can themselves be quite large.

Additionally, with low probability events and hazards, new data may cause radical updates to the risk estimates. This can happen in the form of a new insight brought about by an accident, which revealed a new risk or cause of risk, which was previously thought to carry no risk. This has been seen in many industries, such as the airframe industry, the chemical and offshore industries, etc. This means that our risk estimates do not exhibit the property of stability over time, as seen by the public. This is a continual paradox in the risk and safety domains, because the risk analysts do not know what they do not know, usually until we all find out via some form of accident, or if lucky, a non-fatal incident.

Industry risks and uneven playing fields

Some industries are regulated to very strict criteria compared to others. For example, in the 1970s, there was a rush to gain North Sea oil and gas in the UK. The fatality rates were high (for example associated with commercial diving activities) during this period. However, in the early 1980s there was then a considerable push to reduce the level of fatalities (from up to 20 per year to less than 2), which was achieved. At around the same time, the fatal accident rates “allowable” in various industries differed markedly. Nuclear power stations were being designed with a risk criterion of 1E-6, whereas in the chemical and offshore industries, the allowable fatality rate was much higher (1E-4 to 1E-5). Similarly, whilst large construction projects are often designed to have very low risks for their operational lifetimes, they nevertheless lead to a number of fatalities during the construction and commissioning phases. Yet how is it that we reduce the probability of reactor meltdowns, but can’t stop people falling off a ladder?

There is clearly not a “level playing field” across different industries or aspects of normal life. It is sometimes assumed that this is in effect due to the public themselves, who for example may trust nuclear power less than coal-generated power. But even if this assumption about public perception of risk were true, what is that perception based on? Is it based on a true reflection of life cycle risks, or on media-reported events? Every day around the world, 50 miners lose their lives. It usually doesn’t make the news. Perhaps if the pubic really knew this, perceptions would change. Even if they did not, at least the influence of that perception would be an informed one, attributable to social values rather than media “hype” and company and other source “spin”. On the other hand, maybe the public is aware of these differences in risk, but there are other factors that lead them to accept conventional energy production better than nuclear.

Risk presentation

Risks can be portrayed in myriad ways, but perhaps the most useful way is largely ignored. A risk implies that a decision has to be made, and implies trading off between one risk and another. Therefore, because the essence of risk evaluation concerns trade-offs, risks are best presented comparatively, rather than absolutely. But such a comparison must not be abstract (e.g. the risk is equivalent to being struck by lightning), but should be contextual, i.e, in the context of the likely decision that has to be made. Travel is a good example. Why not show examples that people can relate to. E.g. 10 international business trips per year around Europe, via plane and via train and via car — with the attendant risks calculated. Or, even better, the composite risks of the complete travel trip, including driving to the airport, or walking to the station, etc. This could than be balanced for instance against time consumed during the trip.

Risk compartmentalisation: divide and ignore?

A major stumbling block seems to be compartmentalised thinking, reinforced by compartmentalised government ministries and attendant agencies and regulators. Many industries are regulated to be as low as reasonably practicable (ALARP) within that industry only. They are not compared to other industries. So, for example, car travel is a big killer, but is very under-regulated compared to other modes of public transport. This is accepted generally, but not always. Many small residential communities lobby to place speed-controlling devices in their roads to ‘kill speed not kids’. This is a clear example of local rejection of risks that are overall accepted, and also a neat understanding of the difference between procedural defences (e.g. fines etc.) and deterministic defences (speed bumps etc.). But at a governmental level, if number of fatalities from transport risks is the key aspect to reduce, then road is obviously the target for radical measures. A common transport ministry, and less compartmentalised regulatory systems, would possibly help in this respect, of highlighting the obvious and leading to logical action. If the governments need encouragement for such radical considerations, they need only to look at the public health bills associated with car accidents etc., in much the same way that airlines have realized that few airlines have recently financially survived a fatal crash. However maybe the costs associated with further risk reduction, such as slower traffic or the need for expensive types of road construction would outweigh the costs of fatalities and injuries that it aims at reducing.

Risk target allocation

This leads on to two critical and often under-discussed aspects: who sets the risk targets and how are they derived? This is a question currently facing the aviation industry, for example. When an industry regularly kills people, the target is easily seen: zero fatalities. But once the industry moves into the low risk region, then targets become less intuitive. Frequently, figures are used of less than once in a million years, for example. Such figures however, do not always seem relevant or realistic. Such stringent criteria can in fact lead to risk analysts either saying a new plant (designed to be safer than existing ones) cannot be built, or else ‘massaging’ the figures to achieve the criteria. Risk analysis then becomes a ‘numbers game’ — the criteria are unrealistic, so the numbers can also be unrealistic, because we ‘know’ the plant is safe. Yet still such targets are used, despite several engineers and risk analysts pointing out that such figures are unrealistic and optimistic in the face of incident statistics and common mode failure phenomena. In such industries, there is also concern over the “cliff edge” scenario, wherein an industry appears to be safe, but in fact is merely heading for an — admittedly very unlikely — accident whose consequences will be on a catastrophic scale. Such failure modes can be over-looked in the fight to reach the 1E-6 or 1E-7 target criteria for a system.

Risk gradients: rich and poor societies’ risks

In some countries, risks are seen very differently. There is not concern over whether eggs increase cholesterol, but rather concern over when, or even if, the next meal will occur. Certain countries are regularly plagued by recurring natural disasters, such as flooding in Bangladesh, drought and famine in many countries, malaria. The amount of money spent in the industrialized nations on risk assessment alone, must be seen by some as wholly obscene when natural disasters are still able to claim the lives of millions every year. Taking the idea (or ideal) of risk coherence to its logical extreme, with quantity and quality of life as the goal, then such naturally occurring risks should feature in a “global risk landscape”, and might result in a re-thinking of risk priorities.

Aims of the workshop

• To contrast risk assessment and acceptance approaches across a number of different domains (energy, transport, food) to see the different approaches and to assess their strengths and weaknesses, and their compatibility.

• To consider whether the approaches are indeed incoherent, and if so, to consider ways to improve matters.

• To elaborate an approach towards more coherent risk presentation across industries and public sectors.

Approach

What is needed is a look across a number of different industry and public sectors associated with risk, and listening to a number of different people involved in dealing with risks in technical and public domains. Ideally some of these perspectives will themselves be comparative. Additionally, there clearly does need to be an update of the “public perception of risk” angle, but with a more pragmatic focus in terms of what to do with these different perceptions in decision-making. The media also need to be represented, and governmental agencies and companies involved in risk presentation to the public, as well as at least one “lobbying group”.

The goal of achieving risk coherence should be one of informing acceptance and safety/health investment strategies, i.e. deciding where most risk lies, and how to improve it, to lead to the higher level goals of more life and more quality of life. The overall goal should be a general and measurable reduction of risk, and not merely exporting of risk from one domain to another.

Workshop organizers

• Barry Kirwan, Eurocontrol

• Ben Ale, TU Delft

Return to the list of workshop topics…