Chapter 10 The decision-making process
Since we are unable to predict the future, the best we can do is to assess the patient and present reasonable treatment alternatives.1
For example, look at the treatment decisions to be made in the following scenarios.
• If an inpatient is granted home leave, what is the likelihood of him not returning and what might be the consequence? Should he remain in hospital?
• A woman with schizophrenia has not been seen for 4 days. Should the police be asked to look for her?
• An actively suicidal patient does not wish to be admitted to hospital and his family want him treated in the least restrictive environment. Should he be admitted involuntarily?
There is a myth that the decision-making process will determine the outcome.
Clinicians cannot control or predict all the people and circumstances the patient will meet after they have left the office. Tragic outcomes do occur even after good and robust decision-making. Clinicians can only be responsible and expect to be held accountable for the decision-making process and its implementation, not the outcome. Most decisions can be defended if there is a record of having considered the alternatives. If a clinician has conferred with a colleague, and can demonstrate this in the notes, the likelihood of the decision being defensible is even greater. Clinicians should not be held accountable for the contributing factors over which they have no control.
Heuristics
The process of decision-making using insights from the field of cognitive psychology has useful implications for risk assessment and management. Heuristics as a method of decision-making were introduced briefly in Chapter 2. Heuristics are ‘cognitive shortcuts that allow decisions to be reached in conditions of uncertainty’,2 or ‘rules that guide cognitive processing to help make judgments’.3
• Heuristics reduce the time, resources and cognitive effort required to make a decision.4
• They are a feature of mature clinical thinking5 and are useful particularly when time and information are limited.
• Heuristics are more likely to be used when there are situations of high complexity or uncertainty,6 when there is a high cognitive load or a high density of decision-making,7 and when time for individual decisions is short.8
These are all situations when decisions about risk management may be made. However, reliance on heuristics leads to cognitive bias and ‘severe and systematic errors’9 and these have now been termed ‘cognitive errors’10 or ‘heuristic biases’. The opposite process of using heuristics is one of rational-deductive decision-making in which all necessary evidence for and against any potential course of action is carefully examined and weighed. This assumes no bias on the part of the decision-maker and also assumes optimal time and resources.11 Errors are more likely to be made using heuristics than when rational-deductive decision-making is used.
Heuristics can be seen as being closely allied to the concept of intuitive decision-making based on ‘what feels right’ or so-called ‘gut feelings’. The differences between intuition and reason in decision-making can be outlined as shown in Table 10.1.12
| Intuition | Reason |
|---|---|
| Fast | Slow |
| Effortless | Effortful |
| May be emotionally charged | Emotionally neutral |
| Opaque | Transparent |
For many risk decisions made in everyday life, such as whether to cross the road or not, heuristics are used. We have neuropsychological systems responsible for such automatic choices,13 based on habits derived from practice or from cognitive associations that have been learned over time. As is readily apparent, the capacity to make choices quickly has substantial adaptive value, but if the situation has changed subtly or if there are variables that have not been factored in, heuristic biases may creep in. There is a need to check the cognitions that are used for heuristics on a regular basis otherwise errors will creep in because of personal bias. To complicate matters, the use of heuristics is characteristic of doctors with good clinical acumen14 so the best practice that is recommended is that, as far as possible, heuristics should be used consciously with an awareness of their pitfalls.15
Examples of heuristic biases, or cognitive errors16
• ‘What I have experienced is more likely.’ For example, ‘this patient is just like that man I treated last year who ended up killing his father so he must have a very high risk of violence’.
• ‘What I can remember is more likely.’ For example, ‘we have had three patients abscond in the past 6 months; this man seems likely to abscond if we let him out without an escort’.
• ‘What I can easily imagine is more likely.’ For example, dramatic episodes such as a heinous crime by a psychiatric patient can have a disproportionate influence on future decisions.
• ‘What I want is more likely.’ For example, mental health clinicians do not want bad outcomes so they may underestimate the likelihood of a patient being violent.
• ‘What I expect is more likely.’ For example, ‘he seemed very pleasant, so I didn’t ask him about plans to harm other people’.
Other forms of heuristic biases include:17
• Hindsight bias. Ignoring ‘gut feelings’ and not exploring them sufficiently.
• Ignoring relevant factors such as base rates data and contextual factors.
• A tendency to reinterpret difficult situations to resolve cognitive dissonance (see glossary). For example, a diagnosis may be modified so that the dissonance between ‘a public commitment to care for needy people’ and ‘an inability to care for this needy person because of resource constraints’ can be resolved.
The best process for decision-making is likely to be one where heuristics are utilised in conjunction with rational-deductive decision-making. It is not always going to be possible to use a slow, deductive process but, equally, clinicians are not going to be able to take cognitive shortcuts until the slower rational-deductive process has been utilised on many occasions. Once this has occurred, heuristics (i.e. the cognitive shortcuts) can be utilised more frequently, as long as possible biases are considered. When this occurs, heuristics are likely to function maximally. Utilising the risk thermostat (Figure 6.1, page 45) when making decisions about risk may facilitate increased consciousness of heuristics.
For a more complete list of heuristics and their strengths and weaknesses see Crumlish and Kelly (2009).18 For a more detailed exploration of heuristic biases (cognitive errors) when making risk decisions see Carroll (2009).19
The risk/benefit analysis (rational-deductive decision-making)
For junior practitioners or when the possibility of heuristic bias is present, the slower process of rational decision-making can be a very useful option. In this method risks and benefits are carefully considered. The risk/benefit analysis is an iterative process about deciding which treatment option to take. It may occur during the assessment or during management and should be considered as a tool to be utilised whenever uncertainty exists. When the treatment option has been chosen and implemented, the outcome will be reviewed and further options considered in another risk/benefit analysis. This follows the thinking of Carson’s (1996) ‘Risk Path’ in which a series of small ‘steps’ are taken.20 The evaluation should consider the benefits and the opportunities presented by the situation as well as the risk. There will be situations where there will be high risk but also potentially better outcomes. There will also be situations where the obligation to protect society from a potentially violent person is in conflict with the patient’s right to be treated in the least restrictive environment. These situations of cognitive dissonance need to be evaluated carefully.
There are several phrases that are commonly used to acknowledge that a risk has been considered.
• ‘The benefits of taking this course of action outweigh the risks.’
• ‘We need to take risk in the short term for long-term gain.’
• ‘I weighed up the risk factors and made a clinical judgment.’
These are all examples where a risk/benefit analysis has occurred.
