Treatment of High-Risk Sexual Offenders: An Integrated Approach

Introduction

This book is a reflection on where we have got to as both researchers and clinicians working with high-risk, high-need sexual offenders for many years. We do not intend this text to represent a new theory in relation to sexual offending, though we will spend time discussing some of the more contemporary theories related to working with sexual offenders. Nor do we consider this text a detailed clinical manual for the treatment of high-risk sexual offenders. That said, where appropriate, we will certainly discuss techniques we have used with sexual offenders both in the institution and in the community. In practice, we think of this text as a very practical guide for professionals who work with sexual offenders in one of any number of capacities, be it as case management or parole officers, mental health professionals or as students interested in pursuing this area of psychology. Given the audience for whom this text is designed, where possible, we will avoid discussing detailed statistical findings in favor of descriptions of relevant research and the results of these studies. We will refer the reader to the relevant articles or books as appropriate. The purpose of this text is specifically to address the needs of professionals working with high-risk sexual offenders and specifically those sex offenders who might be deemed psychopathic or who might be referred for commitment proceedings in jurisdictions that have enacted such legislation (e.g., various states in the US). In particular, the focus of this book will be on how to manage high-risk sexual offenders in a comprehensive system of assessment and treatment from intake to the end of their sentence and beyond.

We believe that the orientation provided in this book will be of value to those working with high-risk violent offenders more generally as well. Much of the information contained herein is relevant to the treatment of psychopathy and multi-recidivist violent offenders more generally. Although we will detail what is meant by the terms risk, need, and psychopathy shortly, it seems reasonable to begin with an explanation as to why we decided to write this text in the first place.

Although it may seem an odd thing to say, neither of us wanted to write this book and we have assiduously avoided writing a book on the treatment of high-risk offenders for several years. We kept on hoping that somebody else would take up the mantle and write a practical guide to working with high-risk sexual offenders. Further, we hoped that the person(s) writing the book would have sufficient practical experience of managing sex offender treatment programs both in institutions and in the community to make the advice that they were offering germane to those who work with these populations on a day-to-day basis. Although we have seen some articles on the treatment of high-risk sexual offenders, we have not seen a text that directly confronts these issues, written from the perspective of someone who works with such groups on an ongoing basis. It is for this reason that we have decided to write this book.

We have become increasingly concerned about some newer approaches to the treatment of sexual offenders, in particular the Good Lives Model (GLM). Although we believe there are some very positive aspects to the GLM, we also believe that, as applied to high-risk offenders typically seen in the institution and the community, this approach may obscure the focus from specific treatment targets that have been identified in the literature over the past number of years. The focus of treatment, in our view, becomes less on specific criminogenic needs from a GLM perspective. Rather, several laudatory but poorly defined goals, such as achieving happiness, become the emphasis within this model. To some degree, we think that the proponents of the GLM have thrown the baby out with the bath water as it were. In the following chapters, we will outline what we mean in detail by the this statement.

However, prior to a discussion related to issues associated with the theory and practice of working with high-risk violent offenders, we believe that some introductory information is necessary. First, we will begin by describing our background in working with sexual offenders in both the institution and the community. We will then present some definitions of the terms that will be used throughout this text. Following this, we will describe the populations of sex offenders with whom we have worked in some detail. There are a variety of reasons why we feel it important to discuss the nature of high-risk, high-need populations beyond simply providing relatively dry and technical comments alone. First, the definitions that we will provide, although important, do not do justice to the full clinical picture that we are typically presented with. Even some of the best texts that have been written (e.g., Andrews & Bonta, 2003), though they clearly discuss the complexity of human behavior, proceed to focus on individual risk factors for offending without detailed discussion of the complexities actually seen in individual cases. Further, there has been relatively little discussion in the literature as to how to work with the variety of risk factors that have been shown to be related to sexual offense recidivism as they present in high-risk, high-need offenders. Although some authors have made an attempt to discuss the complexity of working with high-risk sexual offenders (e.g., Whitehead et al., 2007), these attempts have, to some degree, highlighted the failures that clinicians have had working with such populations. Further, we are not aware of any research group that has completed a series of long-term outcome studies on large groups of high-risk sexual offenders treated both institutionally and in the community. The approach suggested in this text is based on a series of studies by the authors that have demonstrated long-term efficacy in the treatment of high-risk sexual offenders both institutionally and in the community. We will argue for the need for an integrated system of assessment and treatment when working with groups of high-risk offenders.

1

Background and Definitions

Both of the authors have had extensive experience in the assessment and treatment of sexual offenders. Jan Looman (J.L.) has been the Clinical Director of the Regional Treatment Centre High Intensity Sex Offender Treatment Program (RTCSOTP) since the mid-1990s. He has overseen several updates to the RTCSOTP treatment manual (e.g., Looman & Abracen, 2002), including a recent version which was submitted to an international panel of experts as part of accreditation procedures for program development in the Correctional Service of Canada (CSC). Jeffrey Abracen (J.A.) worked at the RTCSOTP from 1995 to 2001 and then, in 2002, began working in the Toronto, Ontario, area with sexual offenders released to the community. From 2005 to 2009, J.A. was the Clinical Director of the National Maintenance Sex Offender Treatment Programs operated in Central District (Ontario), which includes the greater Toronto area. Recently, J.A. has taken positions as the Chief, Community Correctional Research, with Research Branch at CSC and currently works as the Chief Psychologist in Central District (Ontario) Parole. Both J.L. and J.A. have been involved in the assessment and/or treatment of sexual offenders for approximately 20 years. Both of us have been employed by the CSC on a full-time basis since the early to mid-1990s.

This is all to say that we have been lucky enough to have accumulated a wide variety of experience in working with sexual offenders in a number of contexts. We have also adopted the position that if you are going to invest the effort in treating high-risk groups of clients than you should also determine the efficacy of the work that is being done. In the area of forensics, perhaps the most significant indication of whether treatment is useful is if it reduces the risk of recidivism. We believe that the results of our research, as well as the results of a number of other dedicated teams, all converge on the same conclusion. That is, contemporary approaches to sex offender treatment appear to have a clear and significant impact on recidivism in the hoped-for direction. In short, appropriate treatment does seem to reduce the risk of recidivism, even among high-risk offenders. We will discuss the evidence in support of this conclusion below. However, before moving on to the topics outlined earlier, we think it important to define some of the terms that will be used throughout this book. What follows is a list of some of the more commonly used terms in this book and a discussion of the basic concepts associated with these terms.

Throughout the text, we will be referring to high-risk, high-need sexual offenders. As a shorthand manner of describing this population we will typically only use the term high-risk populations or refer to offenders treated at the RTCSOTP, who, for the purpose of this discussion, represent a group of high-risk sexual offenders. When referring to risk, we are referring to assignments based on the results of actuarial assessment instruments specifically designed to assess risk of sexual or violent recidivism. Actuarial instruments are measures that have a specific set of items and clear directions for scoring those items. The scores on the individual items are tallied in a pre-defined manner such that the assessor arrives at an overall risk score for general, violent or sexual recidivism. The best of these measures have been shown in a number of studies, using a variety of different groups of offenders, to be accurate predictors of risk (e.g., Hanson & Morton-Bourgon, 2009; Hare, 2003; Quinsey et al., 2006). A few of the better known (and more extensively researched actuarial instruments) are the Static-99/99R, developed by Hanson et al. (2000; Helmus et al., 2012), the Violence Risk Appraisal Guide (VRAG), and the Sex Offender version of the VRAG (SORAG; Quinsey et al., 1998, 2006).

The Hare Psychopathy Checklist-Revised (PCL-R; Hare, 1991, 2003) was developed as a measure of personality to assess the characteristics thought to be prototypical of this condition. The measure initially included the assessment of two factors, the first of which is thought to be related the personality traits associated with psychopathy in the literature. These so-called Factor 2 traits include such features as glibness/superficial charm, grandiose sense of self-worth, conning and manipulative behaviors, lack of remorse, lack of empathy, and failure to accept responsibility for one’s own actions. Factor 2 items are related to criminal lifestyle issues. Examples of Factor 2 traits include the need for stimulation/proneness to boredom, parasitic lifestyle, impulsivity, juvenile delinquency, and revocation of conditional release. Recent research suggests that the 20 items that comprise the PCL-R are best conceptualized as either three (see Cooke et al., 2006) or four factors (Hare, 2003). Each of the 20 items are scored either 0, 1, or 2. Items are scored based on whether the individual exhibits traits that are similar to the descriptions provided in the manual for that item (in which case the individual would receive a score of 1 or 2) or not (in which case the individual would be given a score of 0). Individuals who score above 30 are typically considered to meet the diagnostic criteria for being a psychopath. When referring to psychopathy or individuals with psychopathic traits, we are referring to the Hare PCL-R score in this text. Although terms such as sociopath are similar in nature to the term psychopath, these terms are not interchangeable.

In addition, there has been some confusion in the literature as to whether antisocial personality disorder (APD) is synonymous with the term psychopathy. The criteria for a diagnosis of the latter, listed in the Diagnostic and Statistical Manual (DSM 5; American Psychiatric Association, 2013), serve to identify individuals who have been persistently antisocial. However, many of the individuals who meet the diagnostic criteria for APD would not meet the stricter criteria for psychopathy as measured with the PCL. With reference to forensic populations, the base rate for psychopathy (15–25%) is much lower than the base rate for APD (50–80%; Hare, 1998, 2003). As Rogers et al. (2000) noted, the DSM relegates the personality features of psychopathy (e.g., lack of concern for the suffering caused to others) to associated features of the disorder. These authors also caution that it is unlikely that the PCL-R and APD measure the same construct. Hare (2003) acknowledges that these constructs are highly correlated, but argues that this does not amount to saying that they are the same clinical disorder.

Hare (2003) has also noted that arguments have been made that question the evidentiary reliability of APD for forensic evaluations and testimony. Given the evidence in favor of reliability and validity regarding the PCL-R, as well as the very clear scoring criteria that exist for the measure, it is hard to argue with the psychometric properties of the scale (see Hare, 2003 for a detailed discussion of these matters). With reference to high-risk sexual offenders, we have found that, among those offenders treated at the RTCSOTP, offenders who scored high on the PCL-R (defined as a score at or above a cut-off of 25) recidivated at significantly higher rates than sexual offenders scoring low on the PCL-R (Looman et al., 2005b). However, we have failed to find significant differences in terms of recidivism among those with or without a diagnosis of a personality disorder (Abracen & Looman, 2006). In our view, these findings are not surprising in that the diagnosis of APD is hardly informative with reference to the RTCSOTP. Given the many convictions typically found on the official summaries of these offenders’ criminal histories and the many years of antisocial behavior that have been associated with such behavior, a diagnosis of APD could likely be applied to the majority of the offenders attending the RTCSOTP. As such, the diagnosis would be of little value in distinguishing between recidivists and non-recidivists.

A review (Seto & Quinsey, 2006) of research on treatment with psychopaths chose to discuss studies related to both APD and psychopathy. These authors argue that evolutionary perspectives may be best able to account for psychopathy and argue that psychopaths are a discrete natural class (taxon – for discussions, see Harris and Rice, 2006; Quinsey et al., 1998). Seto & Quinsey (2006) rationalize their approach by noting that there are few controlled treatment outcome studies with reference to psychopathy and that they are therefore justified in discussing the literature on both psychopathy and APD when evaluating the research on psychopathy. From our perspective, it is problematic to argue that psychopathic offenders represent a discrete taxon (i.e., they are qualitatively different), but that the literature related to the majority of offenders (i.e., those with a diagnosis of APD) is relevant to the assessment of whether psychopathic offenders can be treated.

Before we leave the issue of risk, it is important to note that when discussing risk Andrews and Bonta (1998, 2010) highlight the need to include only moderate- and high-risk clients in high-intensity programs. We potentially make low-risk clients worse when these clients are placed in high-intensity programs. For example, these clients may be exposed to certain criminal values or discussions related to deviant fantasies that may result in them developing problems that were not present when they were first incarcerated.

We will also be referring to need areas throughout this text. Need refers to criminogenic needs as defined by Andrews and Bonta (2010). Criminogenic needs are simply treatment targets that the literature has shown to be related to recidivism and which, at least in theory, are subject to modification. According to Andrews and Bonta (2010) the Big 8 criminogenic needs are as follows: Criminal history (early involvement in a number and variety of antisocial activities), criminal associates, criminal thinking, criminal personality, problematic circumstances at home (family/marital), problematic circumstances at school or work, few if any positive leisure activities, and substance abuse. Mann et al. (2010) identified dynamic risk factors specific to sexual offenders such as deviant sexual interests, emotional identification with children, and attitudes supportive of sexual assault. Non-criminogenic needs such as mental health issues are viewed as potentially important treatment targets but are not necessarily related to reductions in recidivism. Therefore, such issues as mental health and self-esteem are viewed as less relevant in the context of forensic treatment. However, it is important to note that for some higher-risk offenders, mental health issues may be seen as important in terms of predisposing an offender to criminal activities. Thus, for the high-risk sub-group, this broader statement regarding mental health issues being non-criminogenic may not be accurate (see Chapter 5 for a complete discussion). Tony Ward and his colleagues (e.g., Ward & Stewart, 2003; Ward & Maruna, 2007; Yates & Ward, 2009) have criticized the emphasis on so-called criminogenic needs and have suggested that a focus on basic human goods is also critical in the treatment of offender populations. We will discuss the Good Lives Model and its relevance to the treatment of high-risk offenders later in the text.

With reference to responsivity factors, Andrews and Bonta (2010) simply define this term as delivering treatment in a style and mode that are consistent with the client’s abilities and learning style. Andrews and Bonta (2010) suggest that, as cognitive-behavioral treatments have been shown to be very effective with offender populations, these are the procedures that should be employed with offenders. They also note that such issues as level of anxiety, verbal intelligence, and cognitive maturity may impact on an offender’s ability to benefit from one type of treatment program or another. These authors note that the principles of risk, need, and responsivity should be thought of as guides but that professional judgment will need to be made in particular circumstances and that our clients cannot be treated in a formulaic fashion (Andrews & Bonta, 2003, pp. 264–265).

We will also refer to the use of phallometry and phallometric assessment at various points in this book. Phallometric assessment (also colloquially referred to as PPG assessment) refers to the physiological assessment of sexual arousal to depictions involving either neutral or sexually charged stimuli. Typically the offender is placed in a room and is provided with slides depicting either clothed or naked children or adults or audio-only stimuli. Arousal to these stimuli is monitored by a device that translates changes in physiological arousal to data that can be quantified. One method of assessing physiological arousal, for example, is by means of a mercury-in-rubber strain gauge which the offender places around his penis. Changes in the circumference of the strain gauge are translated into electrical signals that are than available for analyses. Phallometric testing and related issues are discussed in detail in a later chapter.

With reference to mental health, when we refer to a mental or psychiatric disorder, we mean diagnoses and the associated criteria that are present in one of the editions of the DSM. One notable exception to this is the use of the term psychopathy by which, as noted earlier, we refer to the offender’s score on the PCL-R.

We hope that this brief outline of some of the terms that we will be using throughout this book has been of value. Our starting point in the treatment of high-risk offenders is that the perspective outlined by Andrews and Bonta (2010) is of central importance to the practice of assessment and treatment of high-risk populations. Although we agree with others, such as Ward and his colleagues, that the so-called risk–need–responsivity model outlined by Andrews and Bonta is not without its problems, we believe that these problems are surmountable.

Andrews and Bonta noted that their theory would require elaboration and that it would need to be adapted to work with particular groups of offenders. In short, a certain amount of professional discretion would be needed. We hope to offer such elaboration of their model as applied to high-risk sexual offenders. We disagree with others (e.g., Ward & Maruna, 2007) who have suggested that a new model is necessary (e.g., the Good Lives Model), especially a model that is no longer based on the assumptions of cognitive-behavioral interventions which have been shown to be the most effective techniques used to date with various groups of offenders. Before abandoning a model that has resulted in many positive changes in the treatment of offenders, clear evidence of efficacy of the competing approach(es) is necessary. At present, such evidence seems to be lacking. That being said, the model outlined by Andrews and Bonta needs to incorporate specific issues that are germane to high-risk groups of offenders. For example, as we will argue, with high-risk groups of sexual offenders, issues associated with negative emotionality probably represent criminogenic risk factors in spite of the assertion by Andrews and Bonta that mental and emotional health are not of criminogenic relevance. This does not mean that the model should be abandoned – as much as we would like to take credit for developing a new model, it only means that we need to add a few pieces to the puzzle.

2

The Regional Treatment Centre High Intensity Sex Offender Treatment Program (RTCSOTP)

Description of Group Characteristics and the Treatment Program

Before entering into a detailed discussion of the strategies that have been employed in the RTCSOTP, a description of the population and an overview of the program is necessary. The RTCSOTP was designed to be one of a series of sex offender treatment programs offered by the Correctional Service of Canada (CSC) within the Ontario Region. The CSC is mandated to supervise all offenders given sentences of 2 years or more. Sexual offender treatment programs were offered at the low-, moderate-, and high-intensity levels within the institutional context. Maintenance treatment is also offered, but typically these programs are delivered once offenders are released into the community under some form of conditional release.

Significant changes to the way in which sex offender treatment has been delivered over the last few years have resulted in the closing of the RTCSOTP in 2011–2012. Currently psychology staff are no longer responsible for the delivery of group-based sex offender programming in the CSC. Such programs are currently being offered by programs officers who are trained in the delivery of the sex offender program, but who otherwise have no specific training in iusses related to sexual offender assessment and treatment. However, individual treatment of sexual offenders presenting with significant mental health concerns is still being offered by mental health staff. Moreover, low-risk sexual offenders (as defined by scores on an actuarial assessment instrument) are currently deemed not to require any specific sex offender treatment.

Sex offenders entering the Ontario Region of the CSC were first screened at an assessment center located at Millhaven Institution (the Millhaven Assessment Unit [MAU]). Based on the results of the MAU assessment, which included a clinical interview, psychometric testing, and the scoring of risk assessment instruments, offenders were assigned to either low-, moderate-, or high-intensity treatment programs. We have previously demonstrated that those attending low-, moderate-, and high-intensity programs differ significantly from one another based on a variety of actuarial assessment instruments, including the Level of Service Inventory (LSI) and the Psychopathy Checklist-Revised (PCL-R; Mailloux et al., 2003). Those attending the RTCSOTP (at the time of writing, the only high-intensity program offered in the Ontario Region of CSC) were found to have significantly higher scores on the LSI and the PCL-R than those attending lower-intensity treatment programs.

We have published a number of studies using samples taken from the RTCSOTP. These studies highlight the high-risk/high-need nature of the RTCSOTP population. For example, Looman et al. (2005b) investigated 154 consecutive admissions to the RTCSOTP. They found that the average number of sexual convictions among those included in the study was 3.84 (standard deviation [SD] = 5.81). The average number of violent convictions listed on the Royal Canadian Mounted Police (RCMP) Finger Print Service report (FPS Sheet), which represents an official record of all charges and convictions registered in any province in Canada, was 5.35 (SD = 5.27). The sample also registered an average of 11.88 (SD = 11.68) non-violent convictions on their FPS sheets. The average PCL-R score for the sample was 22.5, which is quite close to the recommended cut-off for psychopathy when the PCL-R is used in research studies (for a discussion, see Quinsey et al., 1998). In a more recent study we observed, among a sample of rapists attending treatment at the RTCSOTP, that the mean PCL-R score for this group was 25 (n = 31) and that for a group of 31 child molesters the mean score was 19 (Abracen et al., 2006).

With reference to psychiatric history we have examined this issue in a study of a sub-sample of the RTCSOTP that scored 5 or higher on the Static-99 (n = 188, in short, a high-risk group of sexual offenders). The average PCL-R score for this sample was 23.7. The sample registered an average of 4.0 (SD = 5.2) sexual convictions on their FPS sheet. The total number of convictions listed on the FPS sheets for subjects included in this sample was 21.7 (SD = 14.8). Approximately 75% of the offenders included in this sample met Diagnostic and Statistical Manual (DSM)-based criteria for a personality disorder. Ninety-two offenders (43.6%) met the criteria for one or more paraphilias (i.e. the DSM-IV-TR based diagnoses for sexual deviations). Sixty-three (33.5%) of the offenders met the criteria for both a personality disorder and a paraphilia.

With reference to substance abuse disorders, we have administered the Michigan Alcohol Screening Test (MAST; Selzer, 1971) and the Drug Abuse Screening Test (DAST; Skinner, 1982) for many years at the RTCSOTP. Langevin & Lang (1990) demonstrated, using factor analysis in a large sample (N = 461) of male sexual offenders, that both the MAST and the DAST could be treated as single-factor tests. Alpha reliabilities for the MAST and the DAST were found to be 0.89 and 0.90 respectively. Further, both of these measures have also been found to have good reliability and validity. In a recent study (Abracen et al., 2008), we observed that 21.5% of the sample had elevated scores on the DAST, whereas 38.4% of the sample had elevated scores on the MAST. The sample consisted of 380 sexual offenders treated at the RTCSOTP.

It should be noted that we have adopted a much more conservative strategy of scoring the MAST at the RTCSOTP than was originally proposed by Selzer (1971). At the RTCSOTP, all items on the MAST are scored as 0 or 1, as opposed to the initial scoring system proposed by Selzer (1971) where offenders could be assigned scores > 1 for answers indicating symptoms prototypic of alcohol problems. This is to say that the observed scores on the MAST represent a very conservative estimate of the presence of substance abuse in the RTCSOTP population. In practice, we have observed that the majority of offenders attending the program have experienced at least minor problems related to substance abuse. More typically, those attending the program have