Approaches to risk assessment 1

Published on 24/05/2015 by admin

Filed under Psychiatry

Last modified 22/04/2025

Print this page

rate 1 star rate 2 star rate 3 star rate 4 star rate 5 star
Your rating: none, Average: 0 (0 votes)

This article have been viewed 1814 times

Chapter 8 Approaches to risk assessment 1

As can be seen from the quotation, risk assessment is nothing new. It starts from the first contact that a clinician has with the patient and continues every time the patient is seen until discharge. Risk assessment allows clinicians to paint a picture of where the risk for the patient lies within the context of their illness and their environment. It helps in planning management of both the illness and risk behaviours.

In the general clinical setting, screening, assessment and management of risk will all be occurring simultaneously. Often, the assessment process will be therapeutic in its own right and becomes management. Often, after talking about their suicidal preoccupation, patients say, ‘I’m feeling better now that I’ve told you that. I was scared to let anybody know how I really felt.’

The chapters on clinical skills are written in such a way that an assumption may be made that each skill can be separated out in the clinical situation and applied incrementally. The presentation of risk assessment and management as separate clinical skills is artificial and not intended to reflect clinical practice. The skills and exercises in successive chapters build on those practised in the previous chapters.

The process of risk assessment also allows for the development of rapport with the patient and should always be seen as a component of treatment, not just an exercise undertaken to fulfil bureaucratic requirements. As is increasingly the case with all clinical work, there should be an expectation that documentation will be shared with the patient. Every time the patient is seen, there should be an assessment for the imminence and previously identified patterns of risk. On an inpatient ward, this may happen every few minutes, whilst in an outpatient setting, it may only occur every 3–6 months when a stable patient is reviewed.

It is a process of combining the ‘here and now’ assessment with the knowledge of the patient’s past to try and predict the future. The ‘here and now’ assessment will include the patient’s mental state, the patient’s current circumstances, systemic issues including resources available, geographical factors and consideration of the clinician’s own response to the situation.

The stages of risk assessment/management do not change and are fairly straightforward: screen for whether the risk exists or not, assess the seriousness and imminence, assess the risk factors and address them. Risk assessment is an evolutionary process, not an outcome. It allows a group of clinicians to defend the defensible by documenting the systematic assessment of an individual. As clinicians become more knowledgeable about a patient, they can add more information about the risk factors and the function of the risk. As some risk is inherent for all patients in one form or another, clinicians should constantly be thinking about possible risks. If the full clinical assessment has been done adequately, the risk assessment should develop more easily.

A full clinical history depends on other sources of information as well as the history from the patient. Risk assessment is no different. This includes accounts from relatives and other informants. In practice, the general mental health worker will often have concerns raised by family and friends and these concerns will be used in conjunction with the clinical assessment. Family or friends who live with a patient are often the first to notice problems.

The other vital component of risk assessment is knowledge of risk factors known to increase the likelihood of the risk behaviour. Access to standardised lists of risk factors is a requirement for good risk assessment. See pages 80–81 and 84–85 for examples.

Risk assessment does have a predictive quality to it but primarily because of the dynamic factors, it can never be 100% accurate. Risk prediction tends to be more accurate in the short-term. As with routine clinical practice, when there is uncertainty, taking advice from colleagues or asking for second opinions or discussing it in the MDT meeting is good practice.

The process of risk assessment can be very quick and easy with some patients where the risk is felt to be minimal, whereas for other patients risk assessment and management turns into a very detailed undertaking making use of specialised assessment tools and may occur over a protracted period of time. This tiered approach2 of basic assessment at one end of a continuum to extremely detailed assessment at the other end will be well known to most clinicians and has been advocated as best practice in the UK. Depending on the level of detail required, the documentation will be quite varied.

This book does not cover clinical skills about how to ask specific questions relating to the risk of suicide or violence. These skills are covered in detail in some guidelines for suicide and violence and in books on clinical skills, and should be taught to all clinicians during their training.

Types of risk assessment

There are four major approaches to risk assessment but there is overlap within each.

Actuarial tools

Actuarial tools were primarily developed for research purposes rather than for clinical practice and so they do not always meet the needs of everyday situations. Actuarial risk assessment tools have also been criticised as being less sensitive than clinical risk assessment to individual differences since actuarial tools on specific risks are developed from data on large populations and are therefore not specific to the individual person.

However, an actuarial statistical approach to risk assessment can enhance the predictions based on clinical assessments and can communicate the degree of risk in qualitative terms. Actuarial tools do have their place in that factors known to be of relevance for the risk being assessed — for example, criminal history, substance abuse, impulsivity and marital status — can be rated for their presence or not. Actuarial scales tend to focus on static risk factors, less on the dynamic ones. It is the dynamic factors which make the job of risk prediction so difficult for mental health professionals compared to other branches of medicine where there are often fewer changing variables. Actuarial data and standardised assessment tools should inform the risk assessment, not substitute for it. In recent years, standardised scales have increasingly included items for dynamic factors and treatment issues. With this shift, the use of standardised scales which focused only on static factors is lessening.

Standardised scales are discussed further in Chapter 18.

Structured clinical judgment (SCJ)

The unaided clinical versus actuarial debate has led to the development of risk prediction instruments which adopt a combined approach and recognise the importance of both static (and dynamic) actuarial variables and the clinical/risk management items that clinicians normally take into account in risk assessments of individuals. The combined approach is called structured clinical judgment (SCJ) and represents a composite of empirical knowledge and clinical/professional expertise. Structured risk assessments act as aides-memoire and make sure that all relevant information is collected.5

Structured clinical judgment is a process which blends structured assessment of risk factors with clinical judgment, including the identification of the patterns of risk behaviours. On top of the usual history and mental state examination, the static and dynamic factors known to be of empirical relevance in the context of the patient’s illness are identified. This is then put together into a plan which addresses the risks. Structured clinical judgment is currently the recommended baseline modality of assessment. If the risk assessment is completed using a structured format, communication will be enhanced and be more transparent as the documentation will be clearly laid out and easily read. If all clinicians in a service use the same format, individual biases and personal opinions are less likely to intrude on the process.

Several structured instruments have been developed to assess risk in clinical contexts. These include the Historical/Clinical/Risk Management 20-item (HCR-20) scale6 for assessing violence and more recently a scale for the assessment of suicide risk developed by Bouch.7 These structured instruments are discussed in more detail in Chapter 18.

Advantages and disadvantages of SCJ using standardised rating scales

A difficulty with the use of structured instruments which include standardised rating scales is that best practice recommends they be validated. Unfortunately, for the general clinician seeing a wide variety of patients with differing risks, there is no one instrument which is currently available which has been validated. As yet, the literature has only generated one standardised assessment tool (the START)8 which attempts to cover a wide variety of risks, albeit with a rather thin evidence base currently. This is perhaps not surprising given the focus of research (mostly from forensic services) which tends to be on specific risks and the development of tools to help identify each risk.

Most rating scales require specific training which tends only to occur for staff in forensic settings and a few clinicians in general settings. Using rating scales also takes time which is often a hindrance to compliance with documentation for clinicians.

Given this, it is perhaps not surprising that ‘locally developed, unstandardised, unvalidated schemes are often used in preference to standardised tools with demonstrated validity’.10

For the busy general clinician, risk assessment on a day-to-day basis may vary from assessment of violence with one patient to assessment of suicidality in the next and then to assessment of self-harm or assessment of harm from others. To utilise different standardised assessment tools for each of these patients when the risk may well turn out to be of minor significance would be pragmatically impossible and clinicians might be justified in complaining that the standardised assessment tools are too unwieldy. As Carroll (2007) comments when discussing the difficulties for general clinicians using standardised rating scales:

To compound matters even further, the initial documentation of risk may well vary depending on whether the first assessment occurs in an acute setting or takes place in the context of a routine ‘cold’ assessment. In the acute assessment, the level of risk is likely to be much higher with the emphasis being on the here and now. Documentation in this setting may be limited to an identification of the risk with basic information about risk factors. More information will be gathered about risk factors once the immediate risk has been contained. For the routine assessment, the level of risk is likely to be lower and the emphasis will be on identification of risk factors which will need to be considered when the patient presents in the future.

In both these settings, the likelihood is that the risk assessment will not be completed during the initial meeting with the patient but will become an evolutionary process with more information added to it as the patient becomes better known. In forensic settings, the risk behaviour has usually occurred already. The patient will have been incarcerated in prison or admitted to hospital and the risk assessment can occur over a period of time. The focus is often on prediction of future episodes as well as prevention.

A pragmatic compromise for the general mental health clinician

The dilemma of how clinicians can practice SCJ has been considered carefully in the UK and best practice recommendations are, currently, that locally developed forms ‘should be designed with evidence-based principles in mind, stating clear and verifiable risk indicators’.12 This is a compromise allowing general clinicians to practise well without being encumbered with time consuming specialised rating scales. For the general mental health clinician:

The risk assessment and management model (hereafter referred to as simply ‘the model’) used for the exercises in this book is a structured format for assessing, documenting and managing the risk while making use of the static and dynamic risk factors known to be associated with the type of risk being assessed. It is important to have an empirically validated list of risk factors available for common risks such as violence or suicide and these can be found on pages 80–81 and 84–85. This is then combined with the clinician’s knowledge of the patient and, importantly, the patient’s own perspective.

The major function of the model is as a ‘decision support tool’14 designed to assist the general clinician manage risk within the context of the illness. The framework incorporates the basic structures of standardised tools: focus on known static, dynamic and protective risk factors as well as a focus on management. The lists of empirical risk factors for each type of risk are separate to the model and can be updated as new research is published. The model is both an assessment and management tool and attempts to meet the following requirements:

In practice, the model is easily applied and is sufficient for most situations and may lay the foundation for the administration of a standardised tool if required.

This model has changed many times over 6 years, has been refined and field tested in a mental health service of over 500 clinicians, and will continue to evolve. Feedback from a variety of sectors including Emergency Departments (EDs), general practitioners and crisis mental health teams has been of central importance in its continuing development. A major requirement in the development of this model has been the need to ensure that it integrates well with assessment and treatment documentation. It also needs to work within a tiered approach to risk.16 Documentation on risk within the clinical assessment can be ‘cut and pasted’ into this model. In a similar fashion the recommendations for interventions identified in this model can be cut and pasted into the treatment plan.

The model is weighted towards prevention and helping a clinician, who does not know the patient, manage an acute situation. In general mental health services this may be of more practical relevance. It avoids statements about level of risk. This is less common in tools derived from forensic services where the emphasis is more likely to be on prediction, level of risk and prevention in the longer term. The section of the model focussing on the function of the risk behaviour could be criticised for assuming that all risk behaviour is driven by frustration but in practice this is not always the case as it can also be coldly goal-directed and sometimes occurs in order to obtain a response from others — secondary gain. When this section has been left out of previous versions, the planning for possible interventions to manage frustration or secondary gain has often been forgotten to the detriment of the care of the patient. This section is sometimes very useful in picking up ‘signature risk signs’ specific to the patient; for example, a patient who has increased risk at the time of an anniversary.

In general mental health settings one of the most common risks identified is the risk of relapse. Risk of relapse is not usually considered to be a ‘risk’ within the risk literature. It is normally addressed within a treatment plan but the process for assessment and management is identical to that of routine risk management. Identification of dynamic risk factors, consideration of static factors which increase the propensity for relapse, and identification of early warning signs and triggers are no different. Because risk of relapse is also a risk factor for suicide, violence and so on, management of the risk of relapse is always required. The model can be utilised to identify the factors of importance for relapse prevention even when no other risks are identified. In practice, it is rare for relapse management not to be incorporated as an integral component of the risk management.

The completed form will generate interventions to be implemented in the longer term but the assessment of the level of risk in the acute/crisis setting remains the responsibility of the clinician on the spot. It gives the clinician guidance and a head start into what interventions may help but the decision about which interventions to utilise is left to the clinician and patient at the time.

If the risk documentation takes a structured narrative form, this allows the focus of inquiry to be directed and also allows risk factors to be contextualised and patterns identified.17 As a result, tick boxes have been excluded. The model allows for the level of risk to change over time and does not need updating with fluctuations of the clinical condition. If the risk plan is included in the overall treatment plan, it should not undermine in any way the overall thrust of treatment. (The treatment plan should include management of the identified dynamic risk factors.)

Examples of completed risk assessments/plans are in Appendix 3.

Static factors on risk assessment/plans

Most risk assessment/plans in general psychiatry do not document all the static factors. The model prompts for the major static risk factors only. If clinicians are developing a complete risk management plan for a patient — for example, a patient in a forensic unit with a risk of violence who is close to discharge — the risk management plan might include a detailed list of all the static risk factors of importance, especially psychopathy, and will describe the influence of the static factors on the current risk. The decision whether to document the static risk factors separately will vary from patient to patient. On the one hand, it is always useful to be able to see the static risk factors separately but on the other hand imposing this on staff means yet more filling out of forms and makes compliance less likely. Also, ‘forms irritate the people who have to fill them in’.18 The important point is to be able to use the static factors to help inform the management of the patient in the longer term.

When developing the prompts for interventions, it is tempting to be prescriptive to ensure that all bases are covered. For example, the following prompts cover most situations:

When headings like those above are used on risk forms, it is implied that each section needs to be filled out and clinicians are tempted to write an intervention for each prompt such as: ‘Take medication’ or ‘Go to crisis respite’ and other self-evident options.

If history was completely predictive then interventions could be presented in a decision tree. A vital task when developing forms is to encourage clinicians to step outside the triage, diagnose and treatment process for risk management and support the patient by getting to know them. Find out what moves the patient, what and who influences them and then present a resource for working with them through a crisis where risk is elevated.

A resource that is prescriptive is potentially less reliable, but a resource that provides insight and information to the clinician on the spot supports interventions that can reduce the likelihood of the risk behaviour occurring.

Figure 8.1 is the complete model with commentary inserted into the boxes (in italics) where clinical information would usually be written. Many of the prompts have been taken from actuarial information. As clinicians get used to the format of the documentation, the prompts can be left out, which simplifies the form somewhat. When electronic health records are used, the prompts can be made to automatically disappear from the finished form. The form has been kept down to two A4 pages as anything longer than this seems to reduce the likelihood of clinicians completing it. When using this form, validated scales for the relevant risk should be utilised for reference.

The model, as printed over the next two pages, appears not to give sufficient room for information to be adequately documented. If there is more than one risk or if there have been several episodes of the risk behaviour previously, there would not be enough room in a hard copy

version of this form. However, most clinicians utilise this documentation from a Microsoft Word template, or similar. The template can be copied into a separate patient file and information keyed directly into the template.

Notes

1 Maden A. Risk assessment in psychiatry. British Journal of Hospital Medicine. 56(2/3), 1996.

2 Royal College of Psychiatrists 2008 Rethinking Risk to Others in Mental Health Services. Final report of a scoping group. June 2008 Royal College of Psychiatrists College Report CR 150, June, p 38.

3 Scottish Executive. Report of the Committee on Serious, Violent and Sexual Offenders. Edinburgh: Scottish Executive; 2000.

4 Bouch J., Marshall J.J. Suicide risk: structured professional judgment. Advances in Psychiatric Treatment. 2005;11:84–91.

5 Maden A. Standardised risk assessment: why all the fuss? Psychiatric Bulletin. 2003;27:201–204.

6 Webster C.D., Douglas K.F., Eaves D., et al. HCR-20: Assessing Risk of Violence (version 2). In Mental Health Law and Policy Institute. Vancouver: Simon Fraser University; 1997.

7 Bouch & Marshall, above, n 4.

8 Webster C.D., Nicholls T.L., Martin M.L., Desmarais S.L., Brink J. Short-Term Assessment of Risk and Treatability (START): The case for a New Structured Professional Judgment Scheme. Behavioural Science and the Law. 2006;24:747–766.

9 Carroll A. Are violence risk assessment tools clinically useful? Australian and New Zealand Journal of Psychiatry. 2007;41:301–307.

10 Higgins N., Watts D., Bindman J., Slade M., Thonicroft G. Assessing violence risk in general adult psychiatry. Psychiatric Bulletin. 2005;29:131–133.

11 Carroll, above, n 9.

12 Department of Health UK. Best Practice in Managing Risk. Principles and Evidence for Best Practice in the Assessment and Management of Risk to Self and Others in Mental Health Services. Document prepared for the National Mental Health Risk Management Programme, June 2007.. 2007.

13 Royal College of Psychiatrists, above, n 2.

14 McNeil D., Gregory A.L., Lam J.N., Binder R.L., Sullivan G.R. Utility of decision support tools for assessing acute risk of violence. Journal Consult Clin Psychol. 2003;71:945–953.

15 Royal College of Psychiatrists, above, n 2, Mullen, p 37.

16 Royal College of Psychiatrists, above, n 2, p 38.

17 Higgins et al, above, n 10.

18 Maden, above, n 5.