Risk, time and reason

Abstract

Over time, a number of alternative approaches to risk have developed and, while these co-exist, they structure time in different ways and are grounded in different combinations of cognitive rationality and affect. The initial conceptualization of risk, which remains prominent, was based on the use of knowledge from past events to provide the context for choices which minimize harm in the future. It underpins structured approaches to decision making based on use of statistics as a means of calculating the probability of future outcomes. This approach has been challenged by the development of a more forensic approach in which the emphasis is on identifying the specific causes of disasters using hindsight and from the analysis of specific cases making recommendations to prevent future disasters. While this approach has a rational basis, it also addresses the collective emotional responses to disaster and provides a cathartic function. In the late twentieth century a more precautionary approach has emerged, in which the fear of future is given precedence over evidence or lack of evidence of past harm. The precautionary approach is future oriented and casts the future principally in negative, potentially catastrophic terms.

Keywords:

• Risk
• time
• choice
• decision making
• forensic risk
• precautionary principle

Introduction

The nature of risk has, as Douglas (1990) described, changed over time and this change has important implications for the conceptualization of time and the relationship between past, present and future. The original conceptualization of risk in the seventeenth century was highly technical and closely linked to the development of statistics as a means of calculating the probability of future outcomes. This original approach was challenged in the late nineteenth century with the development of a more forensic approach in which chance or accidental elements are discounted and the prime focus is on the identification of the causes of harm in specific circumstances, especially disasters. From the late twentieth century, a more precautionary approach has emerged, in which the fear of future harm influences the management of risk. If a sense of risk is historically bound up with the emergence of probabilistic thinking and an orientation towards the future, that orientation has become less open-ended in a precautionary approach that casts the future principally in negative, potentially catastrophic terms.

Risk: A technical activity to predict and manage the future

Wharton has traced the etymological roots of risk to the Arabic risq which indicated a favourable outcome due to divine intervention. It was subsequently secularized in the process of absorption into Greek (Wharton 1992: 4). By the seventeenth century, risk was generally used to mean ‘the probability of an event occurring combined with the magnitude of losses or gains that would be entailed’ (Douglas 1990: 2). Practically, probabilistic thinking is bound up with the development of double entry bookkeeping, and the insurance industry; primarily in the very risky world of seafaring (Giddens 1999). Intellectually, this approach to risk influenced the development of the systematic study of chance. In 1654, the mathematicians Pascal and Fermat collaborated to solve the hypothetical problem of how the kitty of an unfinished game of chance between two players should be fairly divided when one is ahead, and in doing so developed the theory of probability. A hundred years later, Bayes' ‘Essay Towards Solving a Problem in the Doctrine of Chance’ went further by providing insight into the ways in which new information modifies estimates based on the previous information (Bernstein 1996: 3, 131).

Statistics provide a rational mechanism for linking the past, present and future. By carefully collecting information on defined events within specified periods and situations, it is possible to make predictions about such events in the future, especially those that may lead to undesired outcomes. If the future is the same as, or better than the present, then there is little need and incentive to predict it. Thus risk involves predictions based on:

the probability that a particular adverse event occurs during a stated period of time, or results from a particular challenge (Royal Society 1992: 2).

If such predictions are to be meaningful then it is important that the context is clearly specified. For example, while the risk of being killed in a hang-gliding accident during a week is small for a randomly selected person living in the UK, it will increase if that person is an active member of a hang gliding club (Royal Society 1992: 3).

This approach to risk has become the dominant way of managing the future and underpins individual and collective decision-making. Central to modernity is choice, the concept that people individually and collectively should select between different courses of action, and that such selections have consequences. When these consequences are particularly important they can be seen as fateful decisions and the time at which they are made are ‘fateful moments’ (Giddens 1991: 113 – 114). The role of risk in such choices can be difficult to identify as the use of knowledge from past events may be hidden. Individuals often use ‘commonsense,’ i.e. their intuitive decision-making is based on tacit, taken-for-granted knowledge built up from personal experience. In organizations much decision-making is ‘routine’ and based on knowledge developed through shared experiences and embedded in custom and practice (Lam 2000). Despite this often hidden process, increased choice involves increased accountability and pressure to justify and make explicit choices. This is especially the case in public services where officials are making fateful choices on behalf of others, such as in the release of potentially dangerous individuals from mental hospital, or surgical operations on vulnerable young children. The response to such pressure has been to make the choice process more transparent by using and recording risk predictions and management to minimize risk and harm. Such explicit systems of decision-making grounded in risk management are evident in the UK in the care programme approach in mental health (Alaszewski 2006), probation risk assessment (Kemshall 2000, 2002) and clinical governance in health care (Alaszewski 2003). Whether greater transparency necessarily improves outcomes remains open to question, however, and the general scrutiny of difficult expert judgement is necessarily problematic.

The nature of risk in such systems is evident from the decision support systems which underpin their use. Dowie (1999) provides a common sense example based on the hypothetical case of the nursery rhyme character Humpty Dumpty faced with the fateful decision of whether or not to sit on a wall. As Dowie points out in this context, risk is used to refer to a number of different aspects of the process of choosing a course of action. The dimensions of choice include the probability or chance of different outcomes, the outcome itself, the consequences (especially undesired) of such outcomes, the expected utility or value of different outcomes, and their acceptability. Dowie provides a diagram of the ways in which existing information can be organized to provide a framework for the fateful choice (see Figure 1).

Figure 1. Summary of information to inform Humpty Dumpy; choice to sit on the wall (after Dowie 1999: 44).

Linkages between past, present and future underpin this diagram. The key information contained in the diagram has to be obtained from the past. The probabilities should be derived from epidemiological studies of previous similar situations. The source of utilities and values is more complex. Personal and individual values can be linked to individual biographies through the accumulated knowledge of outcomes from which they get pleasure or satisfaction (Zinn 2005). More collective values can be accessed through systematic investigation, such as willingness to pay for risk reduction (Akter et al. 2008) or Quality-Adjusted Life Years, years gained through treatment adjusted for pain and disability (Phillips and Thompson 2003). The information summary provides the immediate context for choice making and the diagram maps display the desirability and probability of future consequences of each course of action. This approach to risk, linking evidence from past events to current choices and future outcomes, is grounded in cognitive rationality and systematic calculations based on evidence. It emphasizes the need for accurate information and formal technologies to interpret and use information to achieve the optimum outcomes.

Risk, inquiries and the allocation of blame through hindsight

In the nineteenth century a different approach to risk developed, which Douglas (1992) calls the forensic approach in which the prime focus is on the past rather than the future. Forensic risk focuses on specific past events and seeks to explain why they have occurred and to identify the role of human agency. As Green (1999) notes, underpinning this development is a shift away from the concept of the accidental. In the forensic approach there is little scope for accidental, i.e. chance random events beyond the control of any individual; all events are essentially foreseeable and therefore preventable. This emphasis on prevention can reinforce the more traditional approach to risk. As Green has noted, preventative strategies are grounded in ‘techniques to calculate statistical correlations of risk factors’ and form the basis for interventions which reduce the aggregate level of risk (Green 1999: 33). It can also involve a different approach, if the emphasis is on a single discrete case or the focus is on the human actions or inactions that caused specific injury or harm. Each case is examined in its own specific terms to identify where the fault occurred. This has a rational motivation to learn the lessons and prevent reoccurrence, but also tends to introduce an emotional dimension of enabling those who have suffered a loss to understand why it has happened, and identify and punish those responsible. It tends to decontextualize both actors and actions from the wider systems of which they were a functional part. They tend to be seen exclusively from the perspective of the harm with which they are retrospectively linked.

The development of the forensic approach to risk can be most clearly seen in the use of inquiries to investigate ‘avoidable harm.’ In health services, these inquiries have been incorporated into everyday practice, being used to investigate not only incidents that have resulted in significant harm such as maternal deaths during childbirth and homicides by patients discharged from mental hospital, but also to incidents which might have resulted in harm, so-called near misses (Alaszewski 2007). The nature of forensic risk can be seen most clearly in the inquiries into disasters; events involving unusual and unacceptable levels of harm. While the level of harm is clearly important in defining an event as a disaster, other factors are involved especially the perceived horror and sense of threat stimulated by the event. For example, the inquiry into the sinking of the Marchioness on the 20 August 1989 with the loss of 51 lives noted that ‘many members of the public were profoundly shocked that such an event could happen in the centre of London’ (Marchioness/Bowbelle Inquiry 2001: 1). The sense of the Marchioness constituting a disaster worthy of major inquiry is at least partly derived from particular cultural norms and expectations rather than only more objectively defined criteria. The consequences of disasters are not restricted to those immediately involved, but spread beyond. There is a broader perception that disasters challenge the established order and ontological security; that is the confidence in ‘constancy of the … social and material environments of action’ (Giddens 1990: 92). Disasters are a major challenge which require some sort of collective response. The editor of the BMJ, commenting on the consequences of the events that led to the appointment of the Bristol Royal Infirmary Inquiry, entitled his editorial ‘All changed, changed utterly’ (Smith 1998). Whitney, in his introduction to the report of 9/11 Congressional Inquiry, compared the Al-Qaeda suicide attack to the traumatic disaster of Pearl Harbour:

That crystal blue morning changed the world, shocking the United States into realizing that it had been drawn into a global war with brutal suddenness … and made Americans see their place in the world in a new way (Whitney 2004: ix).

In such circumstances, routine processes of investigation are deemed no longer adequate and acceptable, and governments tend to invoke the authority and expertise of independent experts, especially judges. As Drewry noted, such inquiries involve ‘borrowing authority’ and the ‘the deployment by politicians of the judges’ status and credibility to defuse matters which those politicians feel they can neither safely ignore or tackle by normal political and parliamentary methods’ (Drewry 1996: 369). Thus inquires have multiple functions. They are formally charged with investigating why certain events happened and, by implication, why individuals and systems failed to prevent the disaster. But contemporary forensic inquiries also have a cathartic function; they can be used for ‘pacifying communal feelings’ (Blom-Cooper 1996: 59), as in the case of the Aberfan (n.d.) disaster when in 1966 a coal tip slid on the village killing 144 people, 116 of them children. In the case of the Bristol Royal Infirmary, an important part of the inquiry's function was to allow parents whose children had been injured at Bristol Royal Infirmary to express their grief. As the inquiry noted:

A Public Inquiry cannot turn back the clock. It can … offer an opportunity to let all those touched by events, in our case Bristol, be heard and to listen to others. Through this process can come understanding (The Bristol Royal Infirmary Inquiry 2001: Foreword).

The forensic use of risk can be rationally justified as the systematic investigation of the past to identify the causes of harm so that they can be prevented in the future. In this sense it is similar to the more traditional approach to risk. However, there are also important differences. The forensic approach not only involves a more explicit emotional dimension but the underlying social processes may be more evident both in the identification of events as disasters and also in some aspects of the inquiry process. Some incidents involving death and injury are immediately and unproblematically accepted as disasters, and inquiries rapidly appointed. When 18 children and teachers were shot and killed on 13 March 2006 at Dunblane Primary School in Scotland, parliament voted on 21 March 2006 to establish a tribunal of investigation which reported by the end of September (Cullen Inquiry 2006). In other cases, survivors and relatives of victims have had to campaign to get a Public Inquiry appointed. Following the sinking of the Marchioness, the Conservative government decided to commission an accident investigation report rather than a public inquiry (Bopcris Website n.d.). The campaign for a full public inquiry continued and achieved success in February 2000 (Marchioness/Bowbelle Inquiry n.d.). In some other cases the government has resisted the pressure for public inquiries. Despite the loss of life of British service personnel and civilians during and following the Iraq war, for example, the only public inquiry has been one into the suicide of civil servant Dr. David Kelly (Hutton Inquiry 2004). The Butler inquiry into the intelligence on weapons of mass destruction was held in secret (Butler Review 2004).

The contested nature of forensic risk is also evident in the process of collecting and reviewing evidence. In formal inquiries, the actions of some individuals and organizations are subject to scrutiny to examine the extent to which their actions and inactions contributed to the disaster. These witnesses may not only seek to deny their actions contributed to harmful outcomes, but may argue that harmful outcomes were either accidental or unavoidable. Kewell (2006) has examined the ways in which risk was contested and constructed in the Bristol Royal Infirmary Inquiry. A central issue was the reputation of the paediatric surgical service; whether it was a centre of excellence providing safe care, or a failing or underperforming service exposing children to danger and unacceptable risk (Kewell 2006: 374). While the establishment of the inquiry clearly favoured those arguing that the service was failing, the inquiry still felt the need to prove the service was failing by obtaining evidence which showed that ‘a substantial and statistically significant number of excess deaths, between 30 and 35, occurred in children under 1 … between 1991 and 1995’ (The Bristol Royal Infirmary Inquiry 2001: 4). The inquiry concluded that the quality of the services provided by the paediatric cardiac surgical team was so poor that it exposed children under the age of one to unacceptable levels of risk, ‘beyond those ordinarily to be expected in the time and context’ (The Bristol Royal Infirmary Inquiry 2001: 131).

Forensic risk also differs from the statistically grounded approach to risk whose methodology builds up knowledge from combining multiple events and comparing outcomes under different defined circumstances, e.g. incidence of lung cancer and heart disease among populations who smoke and those who do not. The forensic approach focuses more on an unusual pattern of events. Disasters are exceptional cases defined by their atypical harmful outcomes. It is possible to aggregate data from such case studies to build up a picture of how and why atypical disastrous events happen, as when the Department of Health and Social Security (1982) undertook a study of child abuse, and Turner and Pidgeon (1997) reviewed ‘man-made disasters.’ However, inquiries are essentially case studies focussing on a specific set of events and are expected to reconstruct events to identify the fatal flaw. Thus inquiries involve the application of hindsight, and bring together information that at the time was separate. So-called ‘missed opportunities’ figure prominently in such analyses. The Ritchie Inquiry (1994) identified a catalogue of failure and missed opportunity in the events that led up to ex-mental hospital patient Christopher Clunis killing Jonathan Zito. The inquiry identified phenomena whose significance only became evident in the light of future events. Christopher Clunis's early interest in knives only became noteworthy in the context of his later delusions and violence. From this perspective there was a failure of social services to identify and respond to warning signs.

From analyses of unique sequences of events in highly atypical situations, inquiries are invited to generalize about the underlying general causes of harm, and make recommendations for changes in systems and practices designed to prevent a repetition of disasters. Such recommendations often involve the development of more effective systems for identifying and managing risk. The Ritchie Inquiry (1994) singled out the repeated failure of practitioners from a range of services to effectively assess the danger posed by Christopher Clunis. It recommended that mental health patients who had a history of violence should have formal aftercare plans which included ‘an assessment … as to whether the patient's propensity for violence presents any risk to his own health or safety or to the protection of the public’ (Ritchie Inquiry 1994: para 45.1.2). In passing, one can note that, despite repeated inquiries in areas such as child abuse and homicides by ex-mental health patients, there is little evidence of successful prevention. As in other spheres, the highly formalized micro ‘risk management of everything’ has no clear relationship to improved outcomes (Power 2007). Consequently, new inquiries tend to identify the same failings as previous ones, if sometimes denouncing them in ever more strident terms. Inquiries provide models and points of departure for others inviting them to identify ‘systematic failings’ at least as clearly as their predecessors and recommending sweeping institutional reforms to make sure ‘it never happens again’ (to some extent disregarding the necessarily unique aspects of the particular disaster itself). Inquiries such as the one into the Clunis killing have become iconic, referred to in the media coverage of each new inquiry (Warner 2006).

The statistically grounded approach to risk is based on a ‘common sense’ approach in which evidence from the past is used to inform choices made in the present, thereby shaping the future, minimizing harm and maximizing desirable outcomes. By contrast, the forensic approach central to inquiries proceeds from an acceptance that harm has occurred and that it could and should have been prevented. Inquiries involve reconstructing the past to identify the key errors and missed opportunities that ‘caused’ the harmful outcome and use this to make recommendations about changes that will prevent this reoccurring. There is little scope for the random or chance elements or Perrow's concept of the normal accident that is a product of the complexity of modern systems (Perrow 1984). While the methods used are different to that in the statistically grounded approach, especially in the focus on a single rather than multiple cases, they do include a clear and systematic rational component, with evidence being tested according to accepted criteria. There are important aspects that depart from such standards however, as the process is explicitly subject to external social and political functions and criteria, both in the decision to deem an event worthy of forensic investigation and then pursue a cathartic end. If one of the functions of inquiries is to reestablish public trust and confidence, then paradoxically they create the ‘well-known “tombstone” pattern in which risk regulation functions as a monument to public emotions about past tragedies’ (Hood et al. 2004: 110) and this is one of the factors in the development of the precautionary approach.

The precautionary approach: Fear of the future shaping the avoidance of risk

In the late twentieth century, it is possible to identify the development of a precautionary approach to risk. This approach is grounded in a perception that, due to the nature and speed of technological and other changes, the past is no longer such a reliable source of guidance. This rupture with the past undermines the role of risk as a way of predicting the future and creates a situation in which the future becomes an area of uncertainty in which fears of harm shape responses to new technologies and increasingly more ‘everyday’ aspects of life. This uncertainty can be linked to sudden collective changes of risk perceptions and associated changes of behaviours such as food panics (Reilly 1999).

Beck (1992) and Giddens (1990) see social and technological changes as significant developments that underpin a radically new social formation with a changed relationship to nature, the ‘Risk Society.’ Beck has argued that the development of modern technologies has created new hazards such as radiation which in many cases are beyond our capacity even to detect, let alone influence. The nature and distribution of these hazards is the defining political and social concern in advanced industrial societies; we are now marked by the ‘distribution of bads rather than goods’ (Beck 1992: 48).

Beck (1992) emphasised the global impact of technological and economic change on environmental changes which now centre on anxiety about global warming (O'Riordan and Timmerman 2001). New and threatening changes can be identified in other aspects of social and political life such as health and health care. Important examples where healthcare might be seen in ‘Beckian’ terms are the development of new infectious diseases (30 previously unknown since 1970s including HIV/AIDS, Department of Health 2002), the spread of diseases normally associated with tropical areas such as malaria to temperate areas as a result of tourism or migration (Prothero 2001), and the development of strains of microrganisms immune to normal drug therapies such as MRSA (Campillo et al. 2001, Andersen et al. 2002). Biotechnology, particularly the development of genetically modified organisms, are seen to have major implications for ethics (Almond 2000), the environment and health (De Waele 1997, Achyra 1999). These developments have drawn attention to the role of science and expertise in predicting and managing future change and potential risks.

One response to the inherent uncertainties of prediction has been the ‘precautionary approach’ in which the onus is placed on innovators to demonstrate that proposed innovation will not have harmful effects (Kreibel and Tickner 2001) and involves a negotiated consensus over risks and benefits (Calman and Smith 2001). The development of the precautionary principle is, arguably, more directly a response to the perceived decline in traditional sources of knowledge; it is an attempt to re-establish public trust by being seen to take even apparently baseless public concerns seriously. While it remains unclear whether such a process does indeed effectively rebuild trust, it does have the consequence of giving comparable status to fears about possible risks as to evidence that such risks exist. As Burgess (2004) notes, there is no clear scientific evidence that the radiation from mobile phone masts causes ill-health yet the UK government established an inquiry into them, for example. This was, at least partially, an exercise in overcoming the perceived legacy of mistrust bequeathed by the handling of the BSE crisis, through demonstrating a new willingness to take seriously and investigate even hypothetical fears.

There are concerns that a systematic use of the precautionary principle will undermine scientific research and innovation (Holm and Harris 1999). Considered more socially, Burgess (2002) argues that in the case of mobile phones and mobile phone masts the application of a precautionary approach increased the very concerns it was designed to assuage. Despite these concerns and more practical legal worries that it should not be explicitly invoked as a binding principle there is evidence of an increasing use of the precautionary approach There is evidence that interest in this approach is spreading to health and health care (Calman and Smith 2001). It is central to the ‘new politics of risk regulation in Europe’ (Vogel 2001), and is also institutionalized in the USA, albeit in a less politicized form (Burgess 2004). The precautionary approach has become routine, at least at the rhetorical level. At the time of writing, British parliamentarians recommended that houses no longer be built near power lines because of fears over a long suggested, but never proven, link to childhood leukaemia (Randerson 2007).

The precautionary approach focuses on uncertainty rather than risk, and uncertainty is often an openly posed condition rather than the bounded and specific challenge common to the more technical conception of risk. Within the context of radical new technologies, the past is no longer a good or acceptable guide to the future. It also focuses on the less clearly determined aspects of risk, notably the perception rather than its more objectively given dimensions. The emphasis is on the emotional response to challenges, especially fear and anxiety. It is in this respect that the perception of risk from mobile phones or power lines is considered sufficient to invoke a precautionary approach, irrespective of scientific evidence. One critic of precaution describes it as a ‘law of fear’ (Sunstein 2005).

Precaution is characterized by mistrust in rationality and authoritative sources of evidence. Neither foresight nor hindsight provide a good guide to risk management and the safest approach is risk aversion. Precaution is not so much a guide to (future) action, as the recommendation that inaction is the safest course in the face of threatening uncertainty.

Foresight, hindsight and no sight: Risk in its various guises

Risk is a key concept in modernity but its uses, misuses and ambiguities have led to some commentators to argue that it obscures more than it illuminates (Dowie 1999). The statistically grounded approach to risk can be seen as the ‘common sense’ approach in which evidence from the past is used in choices made in the present to shape the future, thereby minimizing harm and maximizing desirable outcomes. By contrast, the forensic approach starts with the recognition of unacceptable harmful events and uses hindsight to identify what went wrong and who was to blame, and from this analysis to make recommendations to prevent a reoccurrence of the same sequence of events with the same harmful outcomes. While it has a technical rational dimension, it also involves emotional processes that are evident both in the contested nature of some inquiries and their cathartic functions. In the precautionary approach, the experiential and emotional components override the rational use of evidence derived from past experience as the past is no longer accepted as a trustworthy guide to the future. In the precautionary paradigm, choices and, by implication, the very organization of society, are grounded in the limited, even hypothetical possibility of catastrophic outcome.