The Scottish Government

« Previous | Contents | Next »

Listen

SERIOUS VIOLENT AND SEXUAL OFFENDERS: THE USE OF RISK ASSESSMENT TOOLS IN SCOTLAND

CHAPTER TWO: RISK ASSESSMENT TOOLS IN USE IN SCOTLAND

INTRODUCTION

2.1 This chapter provides an overview of the risk assessment tools and other approaches to risk assessment that were employed by different professional groups who participated in the audit and in the research interviews. Whilst the list is clearly not exhaustive (being influenced by the individuals and organisations who responded to an invitation to participate in the research), it nonetheless provides an indication of the varying approaches to risk assessment that are in place and an indication of how approaches to risk assessment vary between different professional groups.

2.2 The purpose of this chapter is not to provide a detailed description of the features of different risk assessment tools and the types of risk they purport to assess. Such a description is provided in the report by Kemshall (2002) to which readers of this report are referred. However, to ease understanding of this report, brief details of some of the more widely used risk assessment tools (drawn largely from Kemshall, 2002) are provided in an Annex to this report ( Annex Two).

OVERVIEW OF TOOLS IN USE

2.3 Table 2.1 presents a summary of the risk assessment tools in respect of which individual audit forms were completed. The most commonly used tools were the RAGF and the LSI-R, with more specialist tools employed relatively infrequently and usually then by psychologists in prisons or healthcare settings. The most striking feature of the data in Table 2.1 is the absence of a consistent approach to risk assessment across the different professional groups.

2.4 Social work departments were mostly making use of non-specialist tools in their risk assessments. The RAGF was mentioned by 25 social work respondents, including the prison social work units that responded separately to the survey while LSI-R was included in 19 social work returns. Matrix-2000 was the most common 'specialist' tool being used by social workers, being employed by four local authorities. Three local authorities were employing Crime Pics in their assessment of serious violent and sexual offenders and two indicated that they made use of the Dunscore. The latter instrument is an actuarial tool that was developed to assist social workers in assessing offenders' risk of having a custodial sentence imposed by the courts.

2.5 One specialist sex offender project (The Tay Project) had developed its own risk assessment tool - TAYPREP30 - which focused upon 14 historical (static) factors and 16 contemporary (dynamic) factors. This tool - which has still to be validated - was also increasingly being used by other local authority social work departments. Four social work departments other than the one in which the project is based reported making use of the Tay Project assessment tool, though not all completed an individual audit form for this tool.

Table 2.1: Overview of tools is use (number of returns which mentioned the tool)

2.6 A second specialist project, which worked with adolescent abusers and which participated in interviews but not in the audit, had also developed its own risk assessment form based upon tools used by other projects working with this younger age group. The initial risk assessment would be undertaken over a period of three months, involving a minimum of 12 sessions with the young person.

2.7 A different approach to risk assessment was adopted by police forces. Completed audit returns suggested that little use was being made of structured risk assessment tools. Instead, forces had developed their own risk assessment frameworks and protocols, sometimes in collaboration with local social work departments:

"Police utilise a police developed risk assessment tool"

"Not named but is used for risk assessment of sex offenders"

2.8 In interviews police officers reported that a national working group had developed a standard document for use in risk assessments that essentially served as a tool for information gathering. Risk assessments drew upon information from a range of sources, including home visits to the offender. They were normally undertaken first within two weeks of an offender's release from prison and the document completed again on each subsequent visit. Police officers emphasised the importance of gathering 'intelligence' as part of their risk assessments and to inform risk management strategies, for example:

"…where does the person live? Who does he live with? Who lives by him? What vehicle does he drive? What does he do for a living? All sorts of stuff that we need to manage while he is in the community that can reduce his risk as well. E.g. if he is living next to a vulnerable family we would have to take action to get him moved, or maybe disclose to them to give them some protection."

For this reason, existing risk assessment tools, such as Matrix 2000, were viewed as providing only some of the information the police required.

2.9 Police respondents sometimes referred to their approach to risk assessment as 'holistic', drawing upon information provided by a range of agencies (though most commonly social work) involved in sex offender risk assessment and management. For example:

"…a holistic approach to R/A [risk assessment] in conjunction with SW RA3+4 and clinical rating form where these are available."

2.10 Several police respondents pointed to the absence of a consistent national approach to risk assessment as a weakness of the current arrangements (making it more difficult, for example, to share information with other forces if offenders moved address). This issue was being addressed by ACPO (Scotland) who were considering the possibility of the Tay Project Assessment Tool being introduced as a standardised approach across each of the Scottish forces.

2.11 The most disappointing response to the risk audit was from health agencies. Some of those who did respond indicated that they were not involved in risk assessments of serious violent or sexual offenders or, if they were, only on a very occasional basis. However what is evident from the health respondents - including the prison psychologists and psychologists in forensic/secure settings - is their greater reliance upon standardised and/or validated specialist tools such as the HCR-20, PCL-R and SARA (Spousal Assault Risk Assessment).

2.12 The two-stage SPS risk assessment procedure involves an initial collateral file review, which may trigger a more detailed psychological risk assessment, undertaken by a forensic psychologist under the supervision of a chartered forensic psychologist. The detailed risk assessment includes the administration of the HCR-20 and other psychometric tools as required, though SPS acknowledge the importance of the resulting psychological reports being considered in conjunction with other relevant sources of data (Scottish Prison Service, 2001).

2.13 This was confirmed in interviews conducted with social workers and psychologists in forensic settings and in prisons, who also variously made mention of the SVR 20, CARE (Child Abuse Risk Evaluation), the VRAG and SORAG, though one respondent suggested that the latter two instruments would only be employed to supplement risk assessments because of their actuarial nature. Another respondent stressed the value of combining structured clinical and actuarial methods, with the latter providing a cross check for the structured clinical opinion:

"If the two measures are out of sync you would have to go back and look at your structured opinion again so it's quite a good check to see if the methods are all consistent."

2.14 The same respondent suggested that Scottish research had shown that actuarial measures could have similar predictive ability to structured clinical assessments, though the former were less helpful in informing multi-disciplinary risk management plans.

TYPE OF TOOLS IN USE

2.15 Respondents were asked to indicate whether each of the tools included in the audit was actuarial (i.e. using a formal objective procedure such as an equation), structured clinical (i.e. based on clinical assessment of risk factors) or a combination of both. In two-thirds of returns (64) the tools were described involving as a combination of actuarial and structured clinical methods, in 23 they were described as structured clinical and in 9 they were classed as being solely actuarial ⁸. However it also appears that respondents had some difficulty classifying instruments in this may, since the same tool was often categorised differently by different respondents. For instance, The RAGF was described as actuarial by one respondent, as structured clinical assessment by 6 respondents and as a mixture of both by 17. Likewise, LSI-R was categorised by 16 respondents as involving actuarial and clinical approaches, by two as involving actuarial methods and by one as involving structured clinical assessment.

2.16 In the majority of audit returns (90) the tools were said to assess both static risk factors (i.e. factors not amenable to change, such as age or criminal history) and dynamic risk factors (i.e. factors amenable to change, such as offender attitudes). In only four reruns were static factors alone said to have been assessed by the tools while in three returns only dynamic factors were said to have been assessed ⁹.

WITH WHOM THE TOOLS ARE USED

Types of offenders

2.17 The types of offenders with whom the tools were employed are summarised in Table 2.2. Overall, risk assessments were most often undertaken with sexual offenders. However in 43 returns tools were described as being 'generic' in their use, that is applied to all types of offenders. This was true of tools designed for general offender populations, such as the RAGF, LSI-R and Crime Pics.

Table 2.2: Types of offenders with whom risk assessment tools were employed

2.18 With the exception of the police - whose risk assessments were confined to sexual offenders - the other professional groups were utilising risk assessment tools with a range of offenders. However, the proportions of different types of offenders assessed varied across the groups. Social workers were most likely to use the tools in the assessment of sexual offenders while the tools used in prison were most often applied to violent offenders and those used by health professionals were most often applied to the assessment of mentally disordered offenders. Some reservations were expressed about the use of generic tools, such as the LSI-R, in assessing risk among particular categories of offenders, including sexual offenders:

"Acknowledge that this tool is not helpful with sex offenders, serious violent offenders, domestic violence and often women."

"LSI-R is pretty useless for sexual offenders and domestic violence and there are other categories that it is not perfect for - women offenders, road traffic in some cases."

"Concerns about use with sex offenders and young people have led to considerations of using other tools."

2.19 LSI-R may be more limited in its applicability with these groups of offenders because the relative significance of the factors it draws upon to derive a composite risk score may vary according to the type of offence and may also differ between men and women (McIvor et al., 2001).

2.20 Although the current risk assessment tools being used in prisons were considered by psychologists to be the best currently available, there were some adult prisoners - for example those convicted of very serious crimes who have no prior history of offending and no evidence of psychopathy - for whom they were said to be not "hitting the mark". This perceived strengths and weaknesses of different types of risk assessment tool are discussed in more detail in Chapter Four.

Types of risk assessed

2.21 The types of risks assessed by the tools are shown in Table 2.3. Other 'risks' that were said to be assessed included attitudinal change, risk of custody, risks related to drug use, criminogenic needs and the existence of serious personality disorder.

Table 2.3: Types of risks assessed

2.22 Social workers were most likely to use tools that predict risk of general recidivism and risk of harm, with the latter being assessed principally through the use of the RAGF. Psychological risk assessment tools, on the other hand, were more often used to assess specific types of risk.

Age of offenders

2.23 Table 2.4 shows the ages of offenders with whom the tools were reported to be employed. The ages of offenders upon whom risk assessments were conducted could not be easily classified, since different agencies were assessing offenders in different age groups, depending upon their remit and the nature of the setting. To simplify matters, the tools referred to in the audit returns have been classified according to whether they were reported to be used with young people under 16 years of age, with young offenders (that is, those aged between 16 and 20 years of age) or with adults (that is, those aged 21 years and over). The column total in Table 2.4 exceeds the number of returns received since some in many cases the tool included in the audit was said to be used with more than one age group of offender.

2.24 It is clear that in the majority of cases tools were being used with adults. Only social workers and the police indicated in the audit that they made use of risk assessment tools with young people under 16 years of age.

Table 2.4: Age of offenders with whom the tools were used

2.25 The voluntary sector project that worked with adolescent abusers observed that the age and risk profile of their clients had changed in recent years. Whilst older adolescents were still being referred, they tended to be those presenting the highest risk. Increasingly they - and other similar projects - were working with children under 12 years of age and in some instances with children as young as four or five. A second trend that the project had observed was an increase in the number of girls being referred, many of whom had been abused themselves and were exhibiting "concerning" behaviour. Finally, this project also suggested that perhaps as many of 40 per cent of the children and young people they worked with had learning difficulties. The specialist social work project working with adults in the same area also estimated that around 40 per cent of their referrals involved people with learning difficulties.

2.26 Two of the interviewees worked with children and young people who had sexually abused others or who were displaying inappropriate sexual behaviour. Both stressed the inappropriateness of existing tools for use with children because the factors they covered were not necessarily appropriate to the younger age group. A clear need was therefore identified for a risk assessment tool that could be applied in different settings in Scotland where work is undertaken with young people who sexually abuse others. As one respondent explained:

"If I am asked to do a risk assessment on a 14 year old who has committed a sexual offence…I have a big problem in terms of doing that. I can either rely on clinical impression, which is the way it has been done for decades and it is the way it is still being done by many professionals in the NHS … or I can take the tools that are currently available and adapt them to fit that group… I think this is a better way of doing it rather than just having a subjective impression of risk, but I think it's not a good way of doing it… I have a problem with other professionals making judgements on risk and dangerousness without providing systematic evidence, without identifying all the risk factors and backing up their evidence."

2.27 This respondent suggested that the absence of risk assessment tools for use with children and young people possibly reflected the fact that intervention with adolescent offenders had traditionally been the responsibility of child psychiatry and child and family social work services, neither of which have specific expertise in forensic risk assessment. In addition, there was said to be an absence of relevant research from which structured clinical or actuarial measures might be developed.

2.28 Research is currently being conducted into the relative effectiveness of the Youth Level of Service Case Management Instrument (YLS-CMI)- a version of the LSI-R that has been developed to provide a general assessment of risk and needs among young people - and a youth version of the PCL-R. Initial indications suggest that YLS-CMI is a useful initial screening tool for boys, but that it is less so for girls because girls have different criminogenic needs. This would be consistent with the previous finding that the LSI-R has less predictive accuracy with women (McIvor et al, 2001).

2.29 The need for a robust risk assessment tool for use with adolescents was believed to be urgent in view of the proposals contained in the Criminal Justice Bill for the introduction of Orders for Lifelong Restriction. These orders could be imposed upon offenders of all ages, with decisions about their imposition informed by risk assessments, despite the available evidence to support risk assessments of adolescents still being weak.

Sex of offenders

2.30 In 30 of the returns the tools included in the audit were used exclusively with men, while in 68 returns they were used with both men and women ¹¹. The sex of the offenders with whom the tool is used will, of course, be partly influenced by the setting in which it is employed. In a male prison, for example, risk assessments will be carried out exclusively with men.

2.31 Tools employed with both men and women were more likely than those employed solely with men to assess risk of general recidivism and risk of harm. Conversely, tools employed uniquely with men were more likely than those employed with men and women to assess risk of sexual or violent offending.

2.32 In general, there was a view among those interviewed that there were no tools available that were specifically for use with women and the applicability of some existing tools for use with women was called into question.

Mentally disordered offenders

2.33 Interview respondents explained that the HCR-20 can be used with people with mental health problems and can be used with women. However the PCL-R, despite its widespread use in prisons and forensic health settings, has not been designed specifically for a psychiatric population. This, one respondent suggested, means that care needs to be taken when using it with mentally disordered offenders, since certain behaviours may be present as a result of mental health problems, rather than being an indication of underlying psychopathy.

VALIDATION STATUS

2.34 Respondents were asked to indicate whether the risk assessment tools they used had been validated. As Table 2.5 indicates, in fewer than half of returns were the tools to which the referred known to have been validated. Social workers and police were least likely to have employed validated tools while all the tools employed by prison psychologists and most used in health settings were said to have been validated in some context. It should be noted, however, that respondents' understanding of the nature, process and importance of validation may have varied across professional groups. Psychologists would be most likely to be expected to have detailed knowledge of the evaluation status of the instruments they employed while it may be assumed that other professionals had less technical expertise in this respect. The information gathered about the validation status of the instruments will consequently be based on varying levels of understanding and its accuracy will therefore be variable as a result.

Table 2.5: Perceived validation status of the tools referred to in audit returns

2.35 According to audit respondents, locally developed tools had not been validated, nor had one of the most widely used generic risk assessment tools, the RAGF. Tools what were said to have been validated included the LSI-R, Matrix 2000, Crime Pics, Static 99, SARA, HCR-20, RRASOR, PCL-R, SVR-20, VRS, VRS-50 and SOAQ. In most returns in which respondents indicated that the tool had been validated (35) it had been validated by the person or persons who developed it. In some cases tools developed elsewhere (e.g. LSI-R, PCL-R) were said to have been validated for use in the UK. Tools were thought by audit respondents to have most commonly been validated against prison populations, followed by samples of offenders subject to supervision in the community (Table 2.6).

Table 2.6: Settings in which the tool was understood to have been validated

2.36 The characteristics of the samples that were understood to have been used to validate the tools are summarised in Table 2.7. Around one quarter of the returns indicated that the tools they referred to had been validated for use with UK populations but only 15 returns indicated that they had been validated using a Scottish sample to generate local norms. Tools were, it seems, most likely to have been validated for use with adult populations, with only two returns indicating that the tools to which they referred had been validated for use with young people under 16 years of age. In one third of the returns the tool was understood to have been validated using populations of male offenders. However in less than one half of this number of returns were tools thought to have been separately validated for use with women. In only 12 returns did respondents indicate that the tool being used been validated for use with offenders with mental health problems.

Table 2.7: Samples understood to have been employed for validation of tools

2.37 In 10 returns respondents indicated that their organisation planned to validate the tool for use with sex offenders in prison (5 returns), on community supervision (3 returns) and in specialist programmes (2 returns). In five returns validation of tools for use with violent offenders was planned (in four cases using prison populations). In most returns in which respondents indicated that there were plans on place for the tool to be validated the timescales for validation were unknown, though one tool was expected to be validated within 12 months, one within two years and two within five years.

2.38 Whilst the tools employed by other professional groups had not, in general, been validated, in interview psychologists expressed a clear preference for validated instruments, while recognising that relatively few - the HCR-20, PCL-R and VRAG being the primary exceptions - had been validated against Scottish populations (Cooke et al, 2001). Psychologists admitted to having greater confidence in tools that had been validated and expressed concern at the possibility of being having their risk assessments discredited in court if they were not sufficiently evidence-based. As one respondent commented, " if it has been validated there is evidence to suggest it has merit. If it hasn't, you are in trouble."

SUMMARY

2.39 Various approaches to risk assessments were being adopted by different professional groups working in different settings across Scotland. Social workers were most likely to use tools developed to assess risk of recidivism amongst general offender populations while the police did not make use of standardised instruments in the risk assessments they undertook.

2.40 Psychologists - in prisons and in forensic health settings - were most likely to employ tools that had been developed to assess risk of sexual or violent offending (or close correlates thereof) and that had been validated, though only three of the tools used had been validated against Scottish populations.

2.41 Tools in use had mostly, it was understood, been validated against or based on research evidence derived from male populations. Particular populations for whom existing tools were considered mostly inappropriate included young offenders (under 18 year of age), women and offenders with mental health problems.

« Previous | Contents | Next »