Baroness Murphy (CB)

My Lords, in this group, we return to the issue of serious harm and risk. Clause 5, on the grounds for detention, implies that the risk factors for detention under Part 2 are identifiable and that risks are readily assessed. A number of clauses in the Bill are all about the same thing, which is why there are so many amendments in this group. They imply, again, that risks are quantifiable and predictable—if only. Academic research has often stressed how difficult it is to predict episodes of violent behaviour in individuals, because they are rare. Risk assessments given as likelihoods are of limited use when the base rate for violence in a population, particularly serious violence, is low. The same is true, by the way, for suicide and suicidal thoughts. It has been calculated, using the average of all the current tests and rating scales that have been carefully assessed in research studies, that if 5% of the patient population were in a high-risk category, the tests would correctly identify eight in 100 people who would go on to commit acts of violence, but misidentify as violent another 92 people. In fact, less than 1% of community patients will commit serious violence over the period of a year, which means the tests would correctly identify only three patients out of 100. Homicides occur at a rate of one in 10,000 patients suffering from a psychosis per annum, which makes prediction more or less impossible. A number of factors are statistically associated with later violence at a group level. Even the most effective predictive combinations of variables constructed by statisticians perform poorly, except at group level. So making statements about individual risk based on the use of these tools is complex, and some would say unsafe and unethical. Structured risk-assessment systems can be useful in routine clinical practice; indeed, most people use them in day-to-day team thinking about what these risks are. When employed by staff properly trained in their use, they are useful pointers. They perform better than unaided clinical judgment in predicting future violence, but again, at a group level. On an individual level, these checklists need to be part of a detailed understanding of a patient’s mental state, life circumstances and thinking, which is a major contributor to the prevention of harm. This is best achieved, as always, by well-trained professionals operating in a well-resourced environment where staff know well the patient’s history, response to treatment and life circumstances. The wording of the Bill encourages pseudo exact probabilities and predictions of individual behaviour, which are not possible. This exerts pressure on psychiatrists, particularly at tribunals, to make predictions that may be seriously wrong. Unreasonable expectations of what can be predicted lead to defensive practice—to detaining people where it is not justified by the unreliable evidence. There are several places in the Bill where the impression is given that risks are reliable and predictable, but it is not so. As I say, the same problem arises with suicide and suicidal behaviours. I ask the Government to look at whether the wording of these clauses is reasonable, given the evidence, and to substitute these certainties with something more flexible, indicating that a more rounded, holistic and comprehensive assessment is necessary. I beg to move.