part0032

In this chapter, I provide an overview of sex offender treatment, beginning with a brief history, description of contemporary treatment models, and then the outcome evaluation literature and the many methodological issues and debates in this arena. I focus on the literature pertaining to identified offenders, most of whom have been adjudicated. In the next chapter , I discuss prevention and early intervention efforts aimed at individuals who have either offended but not been detected or who have not offended, as far as is known. Because most of the treatment literature focuses on treatment for sexual offending generally, I necessarily focus on the broader literature, but I review studies about offenders against children specifically whenever possible.

Sex offender treatment was not widely provided before the 1980s, when the relapse prevention approach was adopted from the addictions field, along the lines that both types of behavior could be experienced as compulsive, that it could persist despite negative consequences, and that powerful motivations to approach problematic situations needed to be regulated. Relapse prevention quickly became the dominant model, until the null results reported by the widely followed Sex Offender Treatment Evaluation Project in California (Marques, Wiederanders, Day, Nelson, & van Ommeren, 2005). Although many programs still identify their approach as relapse prevention (McGrath, Cumming, Burchard, Zeoli, & Ellerby, 2010), more programs now identify as espousing a self-regulation or good lives model (GLM), reflecting the emergence of a self-regulation and then pathways model from Ward and colleagues (Ward & Siegert, 2002; see also Stinson, Sales, & Becker, 2008, describing a therapy that is also based on enhancing self-regulation).

Marlatt and Gordon (1985) described the relapse prevention approach, which is based on cognitive behavioral principles, as (a) identifying situations in which the individual is at a high risk for relapse; (b) identifying lapses, that is, behaviors that do not constitute full-fledged relapses but approximate drug-taking and that may be a precursor to a relapse (e.g., spending time in bars as a precursor to drinking alcohol); (c) developing strategies for avoiding high-risk situations or triggering cues; and (d) developing coping strategies, which are used in high-risk situations that cannot be avoided and in responding to lapses that occur. In the context of sexual offending against children, lapses might include behaviors such as masturbating to sexual fantasies about children, and high-risk situations might include spending time alone with a child. Janice Marques (personal communication, March 23, 2006) has pointed out, however, that there is wide variation in the format and content of programs that describe themselves as using a relapse prevention approach. Probably the only common theme is that cognitive behavioral techniques are used to increase the offender’s ability to detect and respond to potentially risky situations.

In the previous edition, I reviewed the results of the Sex Offender Treatment Evaluation Project (SOTEP) in more detail, as it was a big deal in the field with the publication of the final evaluation results in Marques et al. (2005). SOTEP’s distinctive features include random assignment of volunteers to treatment and no-treatment conditions after being matched for age, criminal history, and type of offense; an intensive, 2-year cognitive behavioral treatment program based on relapse prevention principles; a one-year aftercare program in the community; and a program evaluation that included both proximal (within-treatment) and ultimate outcomes. The final SOTEP report found that the program had the desired effects on within-treatment goals, such as greater acceptance of responsibility and reduced atypical sexual arousal, but it had no impact on recidivism. Indeed, the seeming trend for those who victimized children to be more likely to reoffend after treatment (22% for treated offenders and 17% for volunteer controls) compared with the opposite trend for rapists (20% for treated offenders and 29% for volunteer controls) was statistically nonsignificant, indicating no group differences. Nearly three quarters of the eligible sex offenders had victimized children, and those who victimized children (especially boys) were more likely to volunteer for treatment. This suggests a high proportion of pedophilic offenders, and therefore the SOTEP results have direct bearing on the treatment on pedophilic sex offenders.

The SOTEP report was a watershed moment because the sexual offending treatment field had to figure out what to do next. Some responded by adopting a skeptical or even pessimistic view on the possibilities of sex offender treatment. Others focused on criticizing details about SOTEP, even though it was carefully designed, comprehensive, and impressive in its scope and intensity, or even going so far as to suggest that randomized clinical trials were not necessary (W. L. Marshall & Marshall, 2007), to which I and others responded with some heat (Seto et al., 2008).

Marques et al. (2005) discussed reasons that the SOTEP program did not have the desired impact on recidivism and suggested several ways in which they would change the program if it was offered again. These changes included recruiting more high-risk offenders, conducting pretreatment assessments on all sex offenders and regularly monitoring treatment progress to ensure that participants were learning the concepts and skills being taught. Other problems with the evaluation, in hindsight, included that relapse prevention was designed to assist individuals who had completed treatment and who needed assistance to avoid relapse (using substances again). It was therefore designed to be a maintenance intervention rather than treatment to reduce risk. Also, the relapse prevention model is predicated on the person being motivated to refrain from relapse; it did not speak to someone who was not genuinely motivated to change. In response to these criticisms and to preliminary results from evaluations of relapse prevention programs, other approaches to sex offender treatment have emerged.

The self-regulation model builds on the pathways model of sexual offending described in Chapter 4 (this volume; see also Ward & Siegert, 2002). In a nutshell, the self-regulation model views offenders as following different pathways to committing their crimes. Some individuals are trying to avoid offending, which is in line with the relapse prevention presumption, but others are recognized to have approach motivations, in which the person actively or passively accepts opportunities to offend, in line with the motivation–facilitation model. Offenders in different pathways differ in their primary and secondary deficits, which can include problems with both general and sexual self-regulation.

The self-regulation model has some support, with research supporting the importance of self-regulation problems. For example, Kingston, Yates, and Firestone (2012) showed that sex offenders could be reliably classified into different pathways on the basis of their offending patterns and that offenders in different pathways differed in their risks and needs. Offenders in the active approach pathway were higher in actuarially estimated risk to reoffend, and pathway was associated with victim type: Offenders against related children were more likely to be in the avoidant-passive pathway, whereas offenders against adults were more likely to be in one of the two approach pathways.

However, the extent to which assigning offenders to offending pathways can lead to more individualized treatment has not been fully explored. Also, and most important, no process or outcome evaluations have shown how well treatment based on a self-regulation model can do, and how this would compare with relapse prevention or other treatment models. We suggested in Seto et al. (2008) that an ethically, logistically, and scientifically justifiable alternative to a randomized clinical trial assigning offenders to treatment or no-treatment conditions—given the resistance to this idea—was random assignment to alternative treatments, such as relapse prevention (the SOTEP results suggest relapse prevention is inert) and treatment based on the self-regulation or GLMs.

The GLM builds on the self-regulation model by considering what goals sexual offending has fulfilled and how those goals could be satisfied in prosocial ways instead. The GLM built on the positive psychology movement and recent recognition of the potential influences of strengths and protective factors in addition to risk factors. Ward, Yates, and Long (2006) suggested the GLM was a more holistic approach than previous treatment approaches, because it explicitly considered the contexts in which offending took place and the human needs that sexual offending and related behavior (e.g., masturbating to paraphilic fantasies) fulfilled. Ward et al. also suggested that the more holistic approach of the GLM, compared with relapse prevention, included more attention on the therapeutic relationship and individualization. Treatment still involves relapse prevention planning, but not simply on coping or avoiding risky situations.

The treatment evaluations for this model are limited and proximal, without randomized trials or matched comparison studies examining recidivism outcomes. Harkins, Flak, Beech, and Woodhams (2012) compared relapse prevention and GLM programs and found no difference between the two approaches in attrition or in pre–post treatment change on self-report measures, suggesting no advantage or disadvantage for these proximal outcomes. The unanswered and critical question is whether the focus on strengths, fulfilling needs, and a more holistic perspective leads to less recidivism.

The GLM approach has received some pushback. Andrews, Bonta, and Wormith (2011) argued that many of the putatively unique elements of the GLM are already incorporated in the Risk Need Responsitivity (RNR) model of offender rehabilitation. This critique was met by Ward, Yates, and Willis (2012), who suggested the GLM could be viewed as an adjunct rather than replacement of RNR. I do not have a pony in this race, but I point out the much larger evidence base for RNR, and the logic (and evidence) to support the idea that a CBT program addressing general criminogenic needs could have desirable impacts on recidivism among sex offenders, especially if effective interventions were developed for atypical sexual interests. I also agree that many RNR informed programs could do more in terms of explicitly recognizing strengths and building up protective factors.

I continue to believe that developments in sex offender intervention will benefit greatly by drawing from the more established literature on correctional interventions (Andrews et al., 1990; Farrington & Welsh, 2005). Some psychological treatments are effective in reducing recidivism among offenders in general. Some treatment advocates have argued that sex offenders require specialized treatment (W. L. Marshall, 2006). However, of the four randomized clinical trials reviewed by Hanson et al. (2002), the two that showed positive and significant effects were designed for all offenders. Borduin, Henggeler, Blaske, and Stein (1990) found a large positive effect size in a small sample of adolescent sex offenders, whereas D. Robinson (1995) showed a positive effect size for adult sex offenders who participated in a general offender program, although only general recidivism was reported. Borduin and Schaeffer (2002) replicated the Borduin et al. (1990) study by showing that multisystemic therapy—a treatment approach for serious juvenile offenders that focuses on criminogenic risks, targeting problem-solving and other skills rather than knowledge or abstract principles, involves multiple systems, and carefully attends to program fidelity—had a large impact on sexual recidivism in a different sample of 48 juvenile sex offenders, even though it was not designed specifically for this group (see also Letourneau et al., 2009). In contrast, the one randomized clinical trial of a specialized sex offender treatment (the SOTEP evaluation of a relapse prevention program) found no significant effect of treatment (Marques et al., 2005). One interpretation is that it is easier to shift antisociality risk factors (facilitations) than it is to shift atypical sexual interests (motivations).

These findings suggest that psychological treatments that target general dynamic risk factors associated with criminal behavior could have a positive impact on sexual or general reoffending. These general dynamic risk factors—often referred to in the criminological and correctional literatures as criminogenic needs —include antisocial attitudes and beliefs, associations with antisocial peers, and substance abuse (for reviews, see Andrews & Bonta, 2010; Lalumière, Harris, Quinsey, & Rice, 2005; Quinsey, Skilling, Lalumière, & Craig, 2004). Given the important role of such factors in sexual offending against children, as I discussed in Chapter 4 , it is probable that the foundation of any effective sex offender treatment program will be the techniques and skills that have been demonstrated to be effective in general offender outcome research, with some additional considerations addressing paraphilic sexual interests or excessive sexual preoccupation.

Interventions can take many forms in addressing the problem of sexual offending against children, beyond individual or group treatment for identified sex offenders, although much of this chapter focuses on this approach to the most visible and pressing intervention group. The response can include prevention efforts aimed at at-risk individuals (e.g., those who are likely or have developed pedophilia), at potential victims (e.g., school-based sexual abuse prevention programs), at potential guardians (parent and caregiver education and safe-guarding programs), and situationally (environments, institutions, online). I discuss these prevention approaches in the next chapter . Recent evidence about birth and parental factors are similar for sexual and nonsexually violent offending suggests that prevention efforts aimed at improving maternal health and pregnancy outcomes could lift many boats, reducing both offending and other negative outcomes (Babchishin et al., 2017).

The intervention response can also take place outside of the criminal justice system, even though that is now where most sex offender treatment is provided. As I discuss in the next chapter , programs have emerged for individuals who have offended but are mostly unknown to the criminal justice system (Dunkelfeld Prevention Project) or for individuals who self-identify concerns about their risk to offend against children (Dunkelfeld, Stop It Now!). People concerned about their child pornography use can find online resources (croga.org, troubled-desire.com ) and peer support groups for persons with pedophilia committed to not offending (Virtuous Pedophiles). Last, the response cannot encompass only treatment but can also comprise the many opportunities for education, skills training, and environmental changes to reduce sexual offending against children.

In the next sections, I review what is known about sex offender treatment, which relies on prison and community programs for identified sex offenders who are involved with the criminal justice system because they are serving sentences or are on probation or parole supervision. Here, I discuss intervention in the context of my conceptualization that pedophilia is like a sexual orientation regarding age (see Chapters 1 and 5 , this volume). This does not mean pedophilia is immutable, but the controversies and lack of evidence that gender orientation can change suggests that clinicians do not currently have a technology to change sexual attraction to prepubescent children. Instead, current interventions are aimed at more effectively managing pedophilic and hebephilic attractions, to prevent sexual offenses against children. Instead of a “cure,” the focus of treatments for nonoffending individuals with pedophilia or hebephilia is the development of more effective self-management, to prevent sexual offending. This can include cognitive behavioral techniques for riding out urges; shifting cognitions and behavior; sex drive reducing medication for those who experience intense and distressing sexual thoughts regarding children; and situational crime prevention efforts, through school-based education for children, parent and caretaker awareness, and institutional policies for youth-serving organizations.

Given the importance of the question regarding the effectiveness of sex offender treatment, it is probably not surprising that multiple meta-analyses have been reported, even though they mostly cover very similar sets of studies. 1 These meta-analyses generally have positive conclusions, although their emphases differ. Of particular concern are variations in which studies are included and how they are weighted on the basis of their methodological qualities. I spend a great deal of time talking about treatment outcome evaluation designs, the meta-analyses, and important methodological considerations because the answer to the question of whether treatment works is not simple; it is highly contested and conditional.

In the first meta-analysis, of 43 English language studies of psychological treatments comprising a total of 9,454 sex offenders, treated sex offenders had lower sexual recidivism rates than sex offenders in comparison conditions, 17% versus 12% (Hanson et al., 2002). The three studies that examined sexual recidivism using randomized assignment, the strongest inference study design, showed no effect of treatment, whereas the 17 studies that used what was described as “incidental assignment” designs indicated a positive effect of treatment. Hanson et al. (2002) considered a study to be incidental assignment if the reasons for group assignment did not appear to be related to offender risk to reoffend, for example, no treatment spots were available at the time or the offender’s sentence did not correspond to when the treatment program was offered. The comparison groups included offenders released before implementation of a treatment program, offenders who were matched on risk factors from archived records, offenders who received earlier versions of a treatment program, and offenders who received no treatment or treatments judged to be of lower quality due to administrative reasons, such as program unavailability or insufficient time left in their sentences. Hanson et al. (2002) considered these designs to be informative in the absence of more randomized clinical trials evaluating sex offender treatment. The results of this meta-analysis have been cited as evidence that sex offender treatment is effective in reducing recidivism.

The meta-analysis by Hanson et al. (2002) was criticized, however, by Rice and Harris (2003) in terms of the study design quality ratings and other analytic decisions. Rice and Harris pointed out that all but three of the 12 studies identified as incidental assignment studies of contemporary treatments included men who would have refused or dropped out of treatment—if treatment had been offered to them—in the comparison group but excluded men who refused or dropped out of treatment from the treatment group. Given that men who refuse treatment or drop out once treatment begins are more likely to reoffend (Hanson & Morton-Bourgon, 2004), this decision created a selection bias, independent of any possible treatment effect, that increased the likelihood of finding fewer reoffenses among the treatment group. Rice and Harris also identified methodological issues about the three remaining incidental design studies. At the same time, Rice and Harris pointed out that two studies rated as lower in methodological quality could have, in fact, been informative because the treatment and comparison subjects were matched on a number of risk factors (Rice, Quinsey, & Harris, 1991; Quinsey, Khanna, & Malcolm, 1998). Neither of these two studies found a beneficial effect of treatment. As another example of disagreement about study coding, Hanson et al. (2002) reported that cognitive behavioral interventions produced larger effects than other forms of treatments. Treatments were deemed to be “contemporary” if the treatment was being offered at the time of the meta-analysis or if it was a cognitive behavioral treatment that had been available since 1980. Rice and Harris (2003) observed that one of the studies that was not identified as being contemporary reported data on cognitive behavioral treatment delivered between 1974 and 1983, with the majority of offenders treated after 1980 (Perkins, 1987); this study found a large negative effect of treatment.

The debate about the interpretation of the Hanson et al. (2002) meta-analysis results, and disagreements about what constitutes acceptable study quality, highlights the importance of having strong methodological designs that most (if not all) reviewers and knowledge users can agree are informative about treatment outcome. Focusing on randomized clinical trials that provide the strongest possible inference about the effect of treatment, a Cochrane Collaboration review identified nine sex offender outcome studies, seven of which evaluated the impact of psychosocial interventions on proximate targets, that is, treatment targets other than recidivism (Kenworthy, Adams, Bilby, Brooks-Gordon, & Fenton, 2004). The other two studies (Romero & Williams, 1983; Marques, Day, Nelson, & West, 1994) evaluated the impact of treatment on recidivism and were included in the Hanson et al. (2002) meta-analysis.

Most of these studies followed mixed groups of adult male sex offenders. Only a few studies reported specifically on pedophilic sex offenders (e.g., Hucker, Langevin, & Bain, 1988), but 52% of the combined sample examined by Kenworthy et al. (2004) had sexually offended against children. For example, Anderson-Varney (1992) randomly assigned 60 sex offenders against children to cognitive behavioral therapy or no treatment conditions; the outcome measures were sexual attitudes, knowledge, and self-reported behavior, social avoidance, and empathy. Overall, Kenworthy et al. concluded there was no significant impact of treatment on these proximal targets.

Lösel and Schmucker (2005) conducted a quantitative review that included evaluations of biomedical treatments, accepted a broader definition of outcome than recidivism, and included studies published in a non-English language. They calculated 80 comparisons from 44 published and 25 unpublished studies: 35 comparisons of cognitive behavioral programs, seven comparisons of behavioral interventions, 18 of nonbehavioral interventions (insight-oriented, therapeutic community, other psychosocial treatments), six of drug treatments, and eight of surgical castration. Six comparisons used randomized clinical trials and an additional six comparisons used matching or statistical controls to make the treatment and comparison groups equivalent. These latter two sets of comparisons found no group difference in recidivism.

Schmucker and Lösel (2015) recently updated this 2005 meta-analysis. Searching for additional studies since the earlier meta-analysis, they identified 29 comparisons that met their new eligibility criteria, representing 4,939 treated and 5,448 comparison sex offenders. All the comparisons evaluated psychosocial treatments (mainly cognitive behavioral in orientation) because none of the biomedical treatments met the new eligibility criteria. The mean effect size for sexual recidivism was smaller than in their 2005 review, but it was still statistically significant; the odds ratio of 1.41 translates to a reduction from 13.7% to 10.1%, or a relative reduction in recidivism by approximately one quarter. The significant effect was heterogeneous, but not affected by outliers.

Surprisingly, methodological quality was not significantly related to effect size in Schmucker and Lösel’s (2015) meta-analysis. However, some study characteristics were associated with more positive effects: Studies with smaller samples, cognitive behavioral or multisystemic interventions, medium- to high-risk offenders, and community-based treatments did better. Studies with smaller samples may reflect artisanal programs that can provide more tailored treatment (or allegiance effects of the program developers). Cognitive behavioral and multisystemic interventions are more in line with contemporary models of (sexual) offending; the risk principle (discussed later in this chapter) suggests the potential for a bigger impact on recidivism for higher risk offenders, and community-based treatments are more ecologically valid (and serve an overlapping but different client population than prison-based programs). The results of this meta-analysis suggest greater investment in cognitive behavioral or multisystemic interventions in the community, with the intensity titrated to the risk of recidivism. I do not think I can improve on Schmucker and Lösel’s (2015) conclusion:

Although our findings are promising, the evidence basis for sex offender treatment is not yet satisfactory. More randomized trials and high-quality quasi-experiments are needed, particularly outside North America. In addition, there is a clear need of more differentiated process and outcome evaluations that address the questions of what works with whom, in what contexts, under what conditions, with regard to what outcomes, and also why. (p. 598)

Grønnerød, Grønnerød, and Grøndahl (2015) recently meta-analyzed 14 studies selected and coded using the guidelines recommended by the Collaborative Outcome Data Committee (2007), the same group behind the Hanson et al. (2002) meta-analysis ( http://www.publicsafety.gc.ca/cnt/rsrcs/pblctns/cllbrtv-tcmdt-gdlns/index-en.aspx ). These studies included 1,421 adult offenders receiving treatment and 1,509 untreated controls, followed for a minimum of 3 years. Most studies were methodologically weak, and the overall effect size was small, suggesting no significant positive impact.

Together, these quantitative reviews provide some encouragement regarding the efficacy of current psychosocial treatments, with caveats and limitations. However, we still need innovative, theoretically informed interventions and methodologically strong evaluations. We need a careful consideration of how well treatment models reflect what we know about sex offender risk factors and the origins of sexual offending against children, followed by rigorous evaluations of treatment programs based on these models. To assist in this theoretical and clinical development, I provide a more detailed review of different interventions. I begin with an overview of the risk, need, and responsivity principles of effective correctional intervention.

Correctional research has repeatedly demonstrated that treatments are more effective to the extent that they adhere to the risk–need–responsivity (RNR) framework (Andrews & Bonta, 2010). The risk principle suggests that the intensity of intervention should be matched to the offender’s risk for recidivism (Andrews & Bonta, 2010). The most intensive services should be directed at higher risk offenders, and minimal levels of service should be assigned to low-risk offenders. Treatment of low-risk offenders is not cost-effective, given the limits on resources such as staff time, money, and space, because these offenders are already unlikely to reoffend. There is little room for improvement, and there is the possibility of inadvertent negative effects, for example, when low-risk offenders are exposed to the antisocial attitudes and beliefs of high-risk offenders in group interventions (see Dishion, McCord, & Poulin, 1999). In the context of sex offender treatment, this could mean combining incest offenders with sex offenders against unrelated children or combining sex offenders against children with sex offenders against adults. Following the logic of the risk principle, some sex offenders do not require treatment, because they are already at low risk for recidivism. Significant impacts will be achieved only by focusing on higher risk offenders. The typical practice, however, is to prescribe treatment for all sex offenders (Mailloux et al., 2003), even for those who are very low in risk to reoffend, and some treatment advocates have suggested that it would be unethical to refuse treatment to any sex offenders who want it (W. L. Marshall, 2006; W. L. Marshall & Marshall, 2007). This view is compounded by the political and public expectation that all sex offenders should receive treatment if they are to be eligible for parole or for an easing of supervision conditions.

As I reviewed in the previous chapter , sex offender risk assessment has advanced greatly in the past decade, with many empirically and independently validated structured or actuarial risk scales (Hanson, Morton, & Harris, 2003; Seto, 2005). The focus has shifted to the assessment of dynamic risk, protective factors, and how to best communicate risk information to influence decision making. Using risk scales could greatly increase the efficiency and effectiveness of decisions about treatment intensity and type.

Some evidence indicates that the risk principle is influencing sex offender practices. Sex offender treatment standards were established in Canadian federal corrections in 2000, and these standards prescribe different amounts of treatment according to offender risk level. High-intensity programs provide between 360 and 540 hours of treatment, moderate-intensity programs provide between 160 and 200 hours of treatment, and low-intensity programs provide between 24 and 60 hours of treatment (see W. L. Marshall & Yates, 2005). Risk-titrated sex offender treatment standards have been established in the United Kingdom under the auspices of the Correctional Services Accreditation and Advisory Panel (disclosure: I am currently a member of this panel). However, W. L. Marshall and Yates (2005) also pointed out that correctional practices allow for the adjustment of treatment plans, such that sex offenders who score in the middle range of an actuarial risk scale might still be placed in a high-intensity program because they show pedophilic sexual arousal or have high scores on measures of psychopathy. No examples are given of sex offenders being placed in lower-intensity programs than indicated by their actuarially estimated risk. Mailloux et al. (2003) argued that some sex offenders may be overprescribed treatment relative to their risk to reoffend.

The needs principle suggests that interventions are more likely to have a significant impact when they target changeable factors associated with recidivism—such as antisocial attitudes, beliefs, and values; substance abuse; and self-regulation skills—in contrast to noncriminogenic needs, such as poor self-esteem, anxiety or mood problems, and subjective distress (Hanson & Bussière, 1998; Hanson & Morton-Bourgon, 2004, 2005). Dynamic risk measures, such as the Stable-2007, can be used to identify criminogenic needs, as can general offender measures of criminogenic needs.

Finally, the responsivity principle recognizes that treatments are more likely to be effective when tailored to the individual’s learning style and capacity. This often includes the use of behavioral and social learning techniques that involve modeling prosocial behavior, graduated rehearsal of problem-solving and other skills, role-playing, and reinforcement. We know more about risk factors and treatment needs than the factors that influence offender responsivity. The sex offender field will benefit by drawing from the general therapy literature on responsivity and on the influence of therapist characteristics, such as warmth, a nonconfrontational style, encouragement and rewards for treatment progress, and directiveness in therapeutic interaction on therapeutic alliance and treatment impact (see Kirsch & Becker, 2006; W. L. Marshall et al., 2003).

Hanson, Bourgon, Helmus, and Hodgson (2009) found that more adherence to RNR principles in sex offender treatment programs was associated with more positive outcome results. Many risk factors discussed in the previous chapter are shared with other offenders, and thus the core of any effective sex offender treatment program is the RNR model, which advocates for matching the intensity of cognitive behavioral interventions to risk level, addressing factors known to be associated with greater risk of recidivism, and tailoring interventions to individual learning styles. The meta-analytic support—including randomized clinical trials—is strong for the effectiveness of RNR approach to general offender rehabilitation (Andrews & Bonta, 2010). The question then becomes, What else might be needed to address distinctive risk factors, especially atypical sexual interests and/or excessive sexual preoccupation?

Randomized clinical trials show that cognitive behavioral therapy (CBT) programs targeting criminogenic needs in higher risk offenders produce larger effects than programs that are not CBT oriented (e.g., humanistic, psychodynamic) or that target noncriminogenic needs and include low-risk offenders (Andrews & Bonta, 2010; Schmucker & Lösel, 2015). CBT is usually the best option for treatment of adult offenders. Multisystemic therapy has empirical support for juveniles who have sexually offended, but the underlying theoretical framework for this treatment is still CBT, with a social-ecological perspective incorporating parents and other responsible adults.

Cognitive behavioral treatments target attitudes, beliefs, and behaviors that are believed to increase the likelihood of sexual offenses against children. Thus, cognitive behavioral treatments can vary widely, depending on which factors are considered to be the most important. As discussed in Chapter 4 , many such factors have been proposed in the clinical and research literatures, including tolerant views about sex with children, empathy deficits, disinhibition, and social skills deficits. A typical cognitive behavioral treatment program might use cognitive restructuring techniques to change views about sex with children and skills-based training to increase empathic statements, self-control of behavior, and social competence.

The latest Safer Society survey of residential and community programs in North America suggests CBT is the dominant theoretical orientation (McGrath et al., 2010). Most Canadian or American programs (86% in the United States) for either adults or adolescents selected the CBT model as one of their top three choices. Rounding out the top three were relapse prevention, self-regulation model, and GLM, all of which are also CBT based. It is interesting that, as this was the first Safer Society survey of treatment programs to include the RNR model, the RNR model was endorsed by less than one third of programs overall, even though it underlies the national standard for correctional programming in Canada and the United Kingdom.

The evidence for treatment of juveniles who have sexually offended is thin, based on two small but positive pilot trials, but those trials can be considered in the context of a larger set of studies showing multisystemic therapy is effective in reducing recidivism and other negative outcomes in juvenile delinquents at risk of out-of-home-placements (Borduin et al., 2009; Letourneau et al., 2009). An important caveat is that multisystemic therapy delivery requires certification, which is intended to ensure fidelity in implementation and high-quality control, but this means few independent evaluations have been conducted, raising some concerns about an allegiance effect.

Biomedical treatments are available for pedophilia, particularly among offenders against children. Focusing on drug options, the goal with medications is to reduce sexual urges toward children and thereby reduce the likelihood of sexual behavior directed toward children. Drugs can include antiandrogens, selective serotonin reuptake inhibitors (SSRIs), and gonadotropin-releasing hormone (GnRH) agonists.

It is a logical assumption that antiandrogens would have an impact on pedophilic sexual response, because testosterone plays a critical role in human sexuality (Davidson, Smith, & Damassa, 1977). Dampening pedophilic sexual response, in turn, would be expected to reduce sexual behavior directed toward children, although it would not necessarily eliminate it. The earliest clinical studies, conducted in Germany, were reported by Laschet and Laschet (1971), who treated more than 100 paraphilic men, most with pedophilia or exhibitionism. Today, the most commonly prescribed agents are cyproterone acetate (CPA) or medroxyprogesterone acetate (MPA), both of which interfere with the actions of testosterone. The use of either CPA or MPA in the treatment of pedophilia is off-label, meaning it is not specifically approved by regulatory bodies for this purpose. Side effects of antiandrogens can include headaches, dizziness, nausea, gynecomastia, depression, and osteoporosis.

The efficacy of antiandrogens in reducing the frequency or intensity of sexual urges has some support, but larger, better-controlled evaluation studies have still not been conducted. Gijs and Gooren (1996) reviewed the literature evaluating the effects of CPA and MPA, focusing on methodologically stronger studies that included double-blind procedures, placebo conditions, and random assignment. All four CPA studies reported that treated men had a significant reduction in sexual arousal, whereas only one of the six MPA studies showed this effect.

Hucker et al. (1988) reported one of the randomized MPA trials. Beginning with 100 men referred for assessment and treatment after being accused of sexually offending against a child, only 48 completed the assessment and 18 agreed to participate in the drug trial. However, only 11 men completed the 3-month trial, with five receiving MPA and six receiving the placebo. Dropouts significantly differed from those who completed the treatment or the placebo phase by reporting more frequent sexual fantasies about children, which would bias any group difference. Men in both the MPA and placebo conditions reported a decrease in sexual fantasies, but men who received the placebo still reported more fantasies at the end of the trial ( M = 28/month vs. 12/month for the MPA group). MPA did have the desired effect on hormones, with a large drop in testosterone among those who completed the follow-up, and no change among those who completed the placebo condition.

Focusing on two additional studies of pedophilic offenders, Cooper, Sandhu, Losztyn, and Cernovsky (1992) approached 28 persons with pedophilia to participate in a drug trial and only 10 accepted. Moreover, they reported data on only seven pedophilic offenders from their double-blind study because an additional three men dropped out during the initial placebo phase. Cooper et al. tested both MPA and CPA over 7 months. They concluded that both drugs reduced sexual thoughts, fantasies, masturbation, and sexual arousal assessed phallometrically. Bradford and Pawlak (1993) treated 20 pedophilic offenders with alternating CPA and placebo phases. They reported data from 17 of these men, assessed phallometrically and hormonally at baseline and again 2 to 3 months later. One participant was dropped from the study for noncompliance and the other two showed no phallometric response to child stimuli. CPA had a significant effect on sexual arousal to children but not on sexual arousal to adults.

It is discouraging that Amelung, Kuhle, Konrad, Pauls, and Beier (2012) found that very few Dunkelfeld participants were willing to take antiandrogen medication and that those who agreed to do so were more concerned about their risk to sexually offend and more willing to consider medication as part of their treatment. Given antiandrogen use is voluntary under biomedical ethics guidelines, this suggests medication—even if effective—is a limited option in the treatment of individuals with pedophilia or hebephilia.

Serotonin is involved in the regulation of human sexual behavior, and SSRIs are known to reduce sexual desire in males (see Meston & Gorzalka, 1992). Some clinical investigators have gone further and suggested, albeit with little evidence, that SSRIs can specifically affect sexual arousal to children without affecting sexual arousal to adults (Fedoroff, 1993, 1995; Greenberg & Bradford, 1997; Kafka, 1991). The enthusiasm for serotonergic agents for the treatment of pedophilia is based on even weaker evidence than for antiandrogens (see Gijs & Gooren, 1996). Serotonergic agents have been evaluated in only one experimental study, by Kruesi, Fine, Valladares, Phillips, and Rapoport (1992), which compared desipramine and chloripramine in a double-blind crossover trial that was preceded by a single-blind placebo condition. Kruesi et al. reported a significant reduction of self-reported paraphilic behavior (predominantly exhibitionism, transvestic fetishism, obscene telephone calling, fetishism) with either drug, but their result is difficult to interpret because only eight of 15 paraphilic men completed the trial; four patients were dropped because they responded to the placebo (which defeats the purpose of having a placebo comparison condition), and three did not complete the drug trial. Including the patients who responded to placebo would have attenuated, and perhaps eliminated, the apparent positive effect of the drugs on self-reported paraphilic behavior.

Greenberg, Bradford, Curry, and O’Rourke (1996) reported a retrospective study of 58 paraphilic men (74% had pedophilia) treated with different SSRIs and followed for 3 months. Greenberg et al. reported that the men reported a significant decrease in paraphilic fantasies over the follow-up, with no differences between SSRIs. Again, the results of this study are difficult to interpret because 17 men dropped out: nine because they discontinued the medication (four against medical advice, three because of side effects, and two because they felt better), one changed his prescription, and three were treated with CPA, presumably because the SSRI was not considered to be sufficiently helpful in reducing their paraphilic fantasies. Therefore, one would expect a more positive result for those who remained in the study. Ignoring this selection effect, Greenberg et al. concluded that drug compliance was excellent because few men missed doses between clinic visits, which should be expected, because they had already terminated patients who stopped taking the drug! Analyses that retained dropouts in an intent-to-treat comparison—a recommended method of dealing with treatment attrition effects in outcome studies—were not reported. Approximately one quarter of the sample was noncompliant with medication.

Fedoroff (1995) suggested that a potential advantage of SSRIs over antiandrogens is that persons with pedophilia may be more willing to take serotonergic agents. Of the 59 men who acknowledged active paraphilic symptoms in his sample of 100 patients, seven opted for psychotherapy alone, 41 chose an SSRI in addition to psychotherapy, and only one patient chose an antiandrogen. Fedoroff suggested the patients preferred SSRIs because of the fewer side effects and because of less stigma associated with having the prescriptions filled.

Gonadotropin-releasing hormone (GnRH) agonists, such as leuprolide acetate, inhibit the production of testosterone by overriding pituitary regulation. Research on the effects of GnRH agonists on paraphilias has been reviewed by Briken, Hill, and Berner (2003), who identified 13 articles, representing a total of 118 men treated for different paraphilias in open, uncontrolled studies. Forty-three of these men were diagnosed with pedophilia; the sexual preferences of another 59 men from two mixed samples of sex offenders were not reported, but the sample descriptions indicate many had sexually offended against children. Most of these studies relied on self-report alone to assess outcomes.

The largest sample of persons with pedophilia in a trial of GnRH agonists was reported by Rösler and Witztum (1998). This was an open, uncontrolled study of triptorelin pamoate with 30 paraphilic men; 25 of them were diagnosed with pedophilia. Twenty-four men completed treatment for at least 12 months; among the others, two men emigrated, three had intolerable side-effects and withdrew, and one withdrew because he wanted to father a child. Men who completed treatment reported fewer paraphilic sexual fantasies, a lower frequency of masturbation, and fewer incidents of “abnormal sexual behavior” (not further defined), from a mean of five incidents per month to zero during treatment.

Briken et al. (2003) concluded that the use of GnRH agonists had some support and described an algorithm for considering antiandrogen or GnRH agonist treatment in the treatment of paraphilias. The long-term consequences of antiandrogen or GnRH agonists are unknown.

The World Federation of Societies in Biological Psychiatry has produced guidelines for the pharmacological treatment of paraphilias, including pedophilia (Thibaut, Bradford, et al., 2016; Thibaut, De La Barra, et al., 2010). The goal for these medications is to control paraphilic fantasies and behaviors that increase risk of sexual offending, control sexual urges, and reduce the person’s distress. The guidelines are based on a review of the available literature, including the studies mentioned earlier, with consideration of methodological quality. These authors also concluded that the methodological quality of available evidence is poor, so the guidelines relied on expert opinion, where the risk of potentially serious harm if paraphilias are acted upon is titrated against the effects of medications on conventional sexual behavior and potential side effects (see Table 8.1 ). As my brief review suggests, little evidence exists to encourage more use of medications to reduce sex drive without more rigorous evaluation.