Anthony Giddens, who has been influential in the thinking of New Labour, sees Britain as living in a post-traditional society in which the "Risk Society" is the context of politics. Over recent years, risk has become an academic growth area. Some of the stimulus for this increase has been the work of Mary Douglas, in particular her book Risk and Blame, published in 1992, in some ways regarded as ahead of its time. The subtitle of her book is Essays in cultural theory, which indicates that her work is at least partly about method. What I want to look at in this presentation, like Mary Douglas, is how we view risk, in particular in relation to mental health.

I am from Norwich. As elsewhere, people in the Norwich area have a certain way of living in relation to institutions. There is a cultural debate about issues. Earlier in the year, Norwich City Council became Radio 4's Today programme's favourite source of amusement amongst local authorities. This was because of several decisions it made which seemed to be lacking in commonsense, although they were all motivated by the avoidance of risk.

For example, in June this year the City Council threatened to fell seven horse chestnut trees because of the risk posed by their conkers. To quote from the report in BBC News online:-

This example demonstrates that there is always a political question about what is acceptable risk. Not all councillors agreed with Harry Watson. Nor did many of the local residents when there was an outcry in the media, apparently condemning the decision as daft. The Council had to apologise for not consulting the public and the trees were given a reprieve.

There is a debate about the balance between risk taking and risk aversion. If anything, what the Risk Society seems to mean is a shift towards the risk aversion end of this relationship. The word risk has been pre-empted to mean bad risks. The promise of a good political outcome is couched in other terms. Yet any society which did not take risks would not be making the most of its opportunities for growth. Over-cautious risk-averse behaviour can be crippling.

Just to prove that conkers are not such an unusual example for risk analysis, there was another report on the dangers of conkers in BBC News Online about a year ago. This followed the report in the Times Educational Supplement of some research from Keele University. The headline was that schools are banning playtime activities, such as conkers, because headteachers are afraid of being sued by parents in the event of an accident. To quote from the researcher's warnings:-

How does this concern about the safety of conkers tie in with mental health? I hesitate to say that I have demonstrated that it is possible to be bonkers about conkers. What I mean is that mental health policy is subject to the same sort of cultural debate about risk as that about conkers. Even if the issue of conkers may seem more trivial, the structure of the debate about mental health is the same. This is basically about authority, and oscillates between pressures to move on from the old institutional constraints, and pressures to sustain the institutions in which authority and solidarity reside.

The important point is that we need to be aware that debates about risk in mental health are political debates. When New Labour came to power there was an increased emphasis on public safety in mental health in the context of suggesting that community care had failed. Even though this rhetoric may be less commonly used now by a more experienced administration, we need to see government concern for public safety for what it is.

Of course, we should be aware of this from the history of mental health practice. For example, it is only since the 1950s that psychiatric hospitals have been unlocked. Any history of the opening of the asylum doors, such as that by David Clark of Fulbourn Hospital, highlights the risks that were taken at the time, and the emotional agony that ensued from taking such decisions. Nonetheless overall, the changes were therapeutic for patients. Institutionalisation was recognised as a danger to mental health.

Similarly, the more recent debate about community care, and whether it has failed, can be seen in terms of risk analysis and political consequences. The primary risks of closing the traditional psychiatric hospital have been noted to be homelessness and murder in the community. This is despite the evidence that only a small proportion of people discharged from psychiatric hospital end up on the streets and that the number of murders by mentally ill people have stayed constant during the period of dehospitalisation.

We need to be bold enough to point out that regarding risk assessment in mental health as a purely clinical activity is not sensible. It has become very common following inquiries into homicides by psychiatric patients to hear the recommendation that risk assessment needs to be improved and that staff should receive further training in risk assessment. But will this really lead to any improvement? Does merely knowing about risks improve practice?

A guiding influence in Government mental health policy over recent years has been the National Confidential Inquiry (NCI) into Suicide and Homicide by People with Mental Illness. The NCI was initially funded by the Department of Health. It makes recommendations on clinical practice with the aim of reducing the risk of suicide and homicide by people under mental health care. Its Director, Professor Louis Appelby, is the National Director of Mental Health, commonly known as the Mental Health Czar, an important advisor to government.

In a press release earlier in the year, the then Health Minister, John Hutton, welcomed the authority that the latest NCI report Safety First gave to the Government's proposals for reform of the Mental Health Act 1983, in particular the introduction of community treatment orders (CTOs). Currently, compulsory treatment can only be provided in hospital. The new Mental Health Act proposes extending powers of treatment to the community. Mr Hutton states that the NCI "clearly demonstrates that the Government's proposals ... will help save lives".

The effect of saying that something has been "clearly" demonstrated is that it rules out any questioning of the statement. If it is so obvious, it is irrational to cast doubt on it. There is no need to look at the evidence. However, let's follow the rules of risk and blame that I have been talking about and look for the political argument. Making a statement like "Government policy will help save lives" is a statement about risk. It needs to be understood in cultural context and the tenacity or "obviousness" with which it is held may have less to do with rationality than the need to support a political argument.

To quote from a previous NCI report Safer Services: "It is clearly unacceptable that patients who have a history of violence, or serious aggressive behaviour, in the context of mental illness should be allowed to be non-compliant with any effective treatment or to lose contact with services, as occurred in several cases prior to homicide. The same is also true of patients at high risk of suicide." The simple argument is that treated, even forcibly treated, mentally ill people are less likely to commit suicide and murder than mentally ill people who are untreated. It may be seen as crazy to question such a reasonable supposition.

Mental health professionals understandably wish that their work was as easy as this simple statement assumes. They strive to reduce the number of deaths related to mental illness. Nonetheless they fear being blamed for deaths which may be said to have been preventable. If services can make an improvement in mental health, they can also make matters worse. This follows inevitably from the power of psychiatry to do good. Having such control and influence implies that there is also the potential for doing harm. Psychiatric interventions aim to be effective, but a good outcome does not necessarily follow logically nor in practice

Applying these principles to the specific case of prevention of suicide and homicide implies that treatment may not always be successful in reducing deaths. Suicides and homicides will never be completely eliminated. However perfect the mental health service, suicide and homicide are a reality of the human condition

Inappropriate psychiatric treatment could potentially increase the risk of deaths by suicide or homicide. However, inconceivable this may be for politicians and mental health services, it does need to be considered. As it is so difficult to think about it, only the advantages of the introduction of CTOs are proclaimed, rarely the disadvantages.

Increasing coercion in treatment makes services more custodial than therapeutic. Not that the social control function of mental health services should be devalued or minimised. The mental illness of individuals in society may need to be contained in their own or others' interests. Yet the disadvantages of making services too custodial has been amply demonstrated in the history of psychiatry. The natural tendency has always been for treatment of the mentally ill to degenerate into being degrading.

The testimony of many survivors of mental health services is that they have been made to feel worse by the lack of an understanding, humane response by the services to their condition. Some have said that the treatment they received precipitated them into attempting suicide or behaving aggressively.

Overactive intervention may therefore be counterproductive. For example, this may help to explain why about one-quarter of inpatient suicides occur in the first week after admission to psychiatric hospital. Admission to hospital may demoralise a suicidal patient more than give them relief. Similarly, inappropriate imposition of a care and community treatment order may precipitate the acting out of more risky behaviour, rather than contain the risk.

The NCI has established that even increasing the nursing observations of a suicidal inpatient to one-to-one continuous observation does not eliminate the risk of suicide. Three percent of inpatient suicides occurred despite such close observation. Although it may be common to suggest that suicide prevention can be improved by increased observation, the reality of the potential for increased risk by more intrusive and coercive interventions should be acknowledged.

The NCI suggests that the proportion of suicides and homicides that will be prevented by enforced community treatment is small. Having made the assumption that CTOs will be beneficial in non-compliant patients, it then attempts to estimate the number of deaths which can be prevented. To quote from Safer Services "If, for example, the target of such treatment were people with schizophrenia or affective disorder who in their last admission had to be detained under the Mental Health Act, and if all suicides and homicides were prevented in which non-compliance or non-attendance had recently occurred, then according to our data, 30 suicides and 2 homicides would be prevented per year." These estimates have gone up slightly in the latest report Safety First to 32 suicides and 3 homicides, although for some reason they no longer figure as key findings of the report. Even if the NCI is no longer placing as much emphasis on these estimates, the Government still is, and the evidence needs to be examined in more detail. Quoting figures like this seems to give a scientific respectability to the recommendations made.

The NCI makes various assumptions, including that the group who are going to be most suitable for a CTO are those who with a major mental illness, such as schizophrenia or affective disorder, who have been been treated under the Mental Health Act in their last admission. Supervised discharge for these people has existed since the Mental Health (Patients in the Community) Act 1995 and such new compulsory powers have not yet clearly reduced the number of deaths.

The NCI concedes that CTOs would not necessarily make people compliant or attend for treatment and that if patients had complied or attended that they would not still have committed suicide or homicide. The estimate made on its own assumptions is therefore a maximum.

The number of people who are non-compliant with treatment before suicide and homicide is much higher than the sub-group identified by the NCI report. About one-fifth of suicides were non-compliant with medication in the month before death, as were about a quarter of homicide cases who were in contact with the services in the previous year. If non-compliance is the critical factor, then the NCI could have produced much higher estimates of the apparent value for saving of lives by the introduction of CTO's.

The point is that the NCI does not want to overstate the value of CTOs but it has decided that their introduction will be of benefit, however small. The trouble is that the logic it has used to sustain this position is untenable. Evidence should precede a conclusion. Conversely, the NCI has assumed that CTOs will reduce deaths and then proceeded to estimate the potential number of deaths that would be prevented based on this premise. It could have produced a higher figure, but it chose not to. It could have produced other figures dependent on different assumptions. The fact is that whatever figure was produced is not evidence of the value of CTOs, as it is only an estimate calculated if CTOs are effective. Although John Hutton has suggested he has evidence, quoting figures based on assumptions which may be difficult to justify does not prove that CTOs will reduce deaths.

The NCI has concentrated on a specific subgroup of people. The introduction of CTOs will impact on more than this group, both to increase and decrease deaths in different subgroups. What matters in terms of outcome is the overall effect, not just the effect on one specific subgroup.

The problem in clinical practice is that suicide and homicide are uncommon even in high risk groups. Accurate prediction in risk assessment is therefore difficult and the potential number of false positives high. The danger of the introduction of CTOs is that too many people will come inappropriately under their remit. Even if this does not happen in practice, the services will continue to be unsuccessful in preventing all suicides and homicides and the number of deaths may not alter much.

Relying on the logic used by the NCI can lead to false conclusions. For example, the NCI found that about 16% of inquiry suicide cases in England and Wales were psychiatric in-patients. This is a significant number of all suicides. Admitting people to psychiatric hospital therefore does not necessarily keep them safe. It would be illogical to deduce that the number of people admitted to psychiatric hospital should therefore be reduced. Nonetheless if this did happen the number of suicides occuring in this subgroup would be likely to decrease. If the NCI had a bias against admitting people to psychiatric hospital and believed that admission to hospital increased the risk of suicide, it could make an estimate of the number of suicides that would be prevented by adopting the policy of reducing admissions. I am not making this proposal seriously, but merely attempting to show that this logic is no different from the one the NCI has used for compliance with treatment to support its bias in favour of CTOs.

Reasonable inferences need to be made from the NCI reports. The danger is that recommendations have been made beyond the evidence. Practicalities need to be taken into account. Ethical principles cannot be evaded.

For example, one of the new recommendations of Safety First is that in-patient units should remove (or make inaccessible) all likely ligature points. This may well be a sensible suggestion but it does need to be implemented with commonsense. Following a similar recommendation after an incident of suicide by hanging from a curtain rail in an inpatient ward in an NHS Trust in which I used to work, all the curtain rails were summarily removed from the dormitory taking away any privacy for patients. Although better facilities than an inpatient dormitory should have been available, the realities of ligature points in inpatient wards do need to be acknowledged and a sensible approach to removing risk undertaken.

Unthinking implementation of CTOs risks undermining patients' rights. The momentum for reform of the Mental Health Act 1983 could be seen as reflecting a lack of tolerance for mental illness in society. Unrealistic efforts to control risk may inappropriately restrict freedom of thought and action. A Government which promulgates the aim of social inclusion will be judged by its policies towards the mentally ill.

What I have looked at briefly is the role of the "Risk Society" in creating a policy of community treatment orders over recent years. I have done this through examining in particular the political influence of the National Confidential Inquiry.

Illness is a misfortune. This applies as much to mental illness as physical illness. Our natural tendency is to think it should not happen. We wish it did not happen. But it does. We could just accept it as a misfortune, but more often than not we rail against it emotionally. There must be some reason why it has happened. Something has gone wrong. Something or someone must be to blame.

Yet we have to live with dangers like illness. Illness is a risk, and there are multiple risk factors involved in causing most illnesses. And as the technology of medicine has grown this has allowed us to switch the danger of illness from natural misfortune to medicine itself which fails to be all powerful in dealing with illness. Doctors are blamed when illnesses, including mental illnesses, are not cured or made worse.

This analysis helps us to realise that talk about risk is a political process. Debate about accountability is a contest to muster support for one action rather than another. People pressurise each other in society and a conformity is created. The charge of causing risk is a stick to beat opponents.

So debate about risks to mental health is not so much about calculations of probability. We do not want to recognise the considerable uncertainty involved. This is why we get caught up in a neutral model of risk perception which leads to the conclusion that there should be more education of the misguided public.

As Ulrich Beck has advocated in his book Risk Society, science needs to stop pretending it is neutral. It needs to become more conscious of its political nature.

Mental health policy has become increasingly bureaucratic in its response to risk. It needs again to open up the debate about the balance between risk taking and risk avoidance.