The state of Vermont closed its only institution for persons with intellectual disabilities in 1993 and moved to a totally community-based model of services. Here we describe the characteristics of the near exhaustive statewide sample of adult male sex offenders with intellectual disabilities (N = 103) who received these services between 1993 and 2004, discuss these services, and examine the sexual recidivism rates of the sample. Over an average follow-up period of 5.8 years, 10.7% of the sample was identified as having sexually reoffended. Most reoffenses were noncontact, and most victims were staff members, relatives, or housemates of the abuser. Results are discussed in terms of their clinical, policy, and research implications.
Deinstitutionalization of persons with intellectual disabilities has resulted in a variety of challenges for community service providers. Among the most significant of these is management of individuals who have committed sex offenses (Ward, Trigler, & Pfeiffer, 2001). A sex offender is defined here as someone who has engaged in sexual behavior that would constitute a criminal sexual offense regardless of whether it resulted in criminal prosecution. Intellectual disability as used here is the same as mental retardation (Diagnostic and Statistical Manual of Mental Disorders—DSM-IV-TR; American Psychiatric Association, 2000).
Whether people with intellectual disabilities are overrepresented among sex offender populations is still a matter of debate (Day, 1994; Thompson & Brown, 1997), but there is considerable evidence that sex offenders score slightly lower in IQ than non-sex offenders (Cantor, Blanchard, Robichaud, & Christensen, 2005). Regardless, persons with intellectual disabilities who commit sex offenses are being increasingly identified, treatment programs for them are proliferating, and treatment effectiveness is being examined.
In the United States, the Safer Society Foundation's first nationwide survey almost 20 years ago identified only 136 programs treating adult sex offenders with intellectual disabilities (Knopp, Rosenberg, & Stevenson, 1986), whereas the most recent survey conducted in 2002 identified 399 such programs, 85% of which were community-based (McGrath, Cumming, & Burchard, 2003).
Despite the dramatic increase in the number of programs for sex offenders with intellectual disabilities, Courtney and Rose (2004) were recently able to identify only 31 studies that reported outcome. They found that most of the studies were methodologically flawed. Several were single-subject case studies, few had control groups, and definitions of intellectual disability and sex offending varied considerably. Likely as a result of these problems, major reviews of this literature have not provided any aggregate quantitative analyses of base rates of sexual reoffending or treatment impact among sex offenders with intellectual disabilities (Courtney & Rose, 2004; Lindsay, 2004; Thompson & Brown, 1997).
In contrast, several meta-analytic studies of the general sex offender literature exist and provide reference points for examining the results of the current study. Hanson, Morton, and Harris (2003) found that the average 5-year sexual reoffense rate (i.e., new charge or conviction) across 10 studies of treated and untreated sex offenders was 14% (95% confidence interval of 13%–15%). Meta-analyses of the treatment literature indicate that intervention, primarily cognitive–behavioral in nature, typically results in a significant reduction in sexual recidivism. Follow-up periods average about 5 years. For example, Losel and Schmucker (2005) examined 69 studies with a combined sample size of 22,181 and found a sexual recidivism rate of 11% for offenders receiving treatment and 17% for those in the control groups; a 6-percentage point difference, reflecting an overall 37% reduction in recidivism. These findings are consistent with those of other recent meta-analyses (Aos, Miller, & Drake, 2006; Hanson et al., 2002).
Our goal in this study was to contribute to the knowledge base of managing sex offenders with intellectual disabilities in community programs. We examined nearly all sex offenders with intellectual disabilities in a statewide developmental services program over an 11-year period. Characteristics of participants and services are described, and reoffense data are reported.
Vermont, a rural state with a population of approximately 623,000 (United States Census Bureau, 2006), closed its only institution for individuals with intellectual disabilities, the Brandon Training School in 1993 and began operating a totally community-based system of services (Shoultz, Walker, Hulgin, Bogdon, Taylor, & Moseley, 1999). A single state agency, the Department of Disabilities, Aging and Independent Living, funded supports and services delivered through a statewide system of community developmental service agencies for all individuals with intellectual disabilities, including those who had committed sex offenses.
Participants (N = 103) were nearly all male sex offenders with intellectual disabilities age 18 and older who received any services from state-funded developmental services programs between 1993 (the year the Brandon Training School closed) and 2004. A sex offender was defined as someone who was known to have committed a sexually abusive act that would constitute a criminal sexual offense in Vermont. Records on an estimated 5 additional men were not available. Seven female sex offenders served by the developmental services programs during this time period are not described here nor are offenders served who did not meet the DSM-IV-TR criteria for mental retardation. Sex offenders with intellectual disabilities who were incarcerated at the time of the study (<5) were not included, but individuals on probation or furlough or who had maxed out of a prison sentence were included as were individuals whose reoffense resulted in incarceration during the period covered by the study. Other participant characteristics are reported in the Results section.
Vermont's 14 not-for-profit community developmental service agencies, as part of their contractual obligation to the Department of Disabilities, Aging and Independent Living provided demographic, offense profile, and sexual reoffense data on participants in the study. Criminal record checks were used to identify sexual reoffense data for individuals who were out of developmental services during any part of the study period. In addition, the first and second author scored participants on the Rapid Risk Assessment for Sex Offense Recidivism—hereafter called the Rapid Risk Assessment (Hanson, 1997), a 4-item actuarial measure that scores the probability of sexual reoffending at 5- and 10-year intervals. Scoring criteria adapted for persons with intellectual disabilities were used (Harris, Phenix, Hanson, & Thornton, 2003, p. 17).
Participants received a range of services commensurate with their risk, treatment needs, capacity for independent living, and legal status. The program is detailed elsewhere (Vermont Agency of Human Services, 2005a) and reviewed here briefly.
Services were built around the individual's residential setting, of which there were several options. Typically, between 1 and 3 offenders lived in a private or staffed home with paid caregivers or a supervised apartment. Residences for offenders who presented a high risk to elope were fitted with security features, such as alarms and Plexiglas windows. Residential providers supervised residents up to 24 hours a day and taught social, daily living, community participation, and sexual risk management skills. Residents often had jobs and participated in other activities, including sex offender treatment. This treatment, primarily skills training and cognitive–behavioral group therapy, was provided by a network of contracted mental health professionals. As an individual lowered his risk, supervision could fade to give the person graduated periods of time alone. Some lower risk individuals were placed in supervised apartments, their own home, or in residence with their natural family, and supervision was provided by scheduled visits, drop-in visits, and phone check-ins.
In terms of legal status, some participants (see Results) were committed to the care of the Commissioner of Department of Disabilities, Aging and Independent Living under Act 248, Vermont's civil commitment statute for persons with mental retardation enacted in 1988 (13 V.S.A. Section 4823 and 18 V.S.A. Sections 8839 et seq.). Individuals so designated had committed a sexual offense, had mental retardation, had been deemed dangerous, and had been found “not competent” to stand trial. Remaining individuals had either been criminally convicted of committing a sex offense and were supervised by a probation or parole officer, had committed a sexual offense substantiated by adult or child protective services, had been enrolled in services by their guardian, or had voluntarily entered services.
The mean age of the sample was 34.6 years (SD = 12.5; range = 18 to 70). The mean IQ of participants was 61.8 (SD = 6.7). Divided into quartiles, the IQs of 23% of participants were 45 to 56; 24% scored 57 to 62; 26%, 63 to 66; and 27%, 67 to 74. Consistent with Vermont's population demographics, fewer than 5% of participants were non-White. The average time-at-risk of the sample (number of years individual lived in the community following placement under the care of the Department of Disabilities, Aging and Independent Living until the time of follow-up) was 5.8 years (SD = 3.7, range = 0.2 to 11.0). Forty-six men (44.7%) had one or more DSM-IV-TR Axis I diagnoses. Thirty-eight participants (36.9%) were known to have been sexually abused as a child. Participants' Rapid Risk Assessment scores indicated that 30 men (29.4%) were at low risk to sexually reoffend (score = 0 or 1); 61 (59.8%), moderate risk (score = 2 or 3); and 11 (10.8%), high risk (score = 4 or 5).
Table 1 shows participants categorized by primary offender type using definitions established by the Association for the Treatment of Sexual Abusers (Gordon et al., 1998). Based on these definitions, a child victim was defined as someone age 15 or younger and adult victim, as 16 or older. Table 1 also shows offense severity characteristics of participants.
About half the participants (53.4%) had a history of committing more than one type of sex offense, a phenomenon called crossover offending. To illustrate, as shown in the first row of Table 2, of the 10 men who had a history of sexually assaulting an adult male, 4 had a history of also sexually assaulting an adult female (40%) and 1 had molested a male child (10%).
The level and range of services provided to participants were diverse (see Table 3). Further, the annual cost of these services per individual in 2004 varied considerably (M = $81,539, SD = $42,701; range = $1,020 to $224,740). As might be expected, costs were higher for individuals who had higher levels of supervision, F(2, 87) = 15.0, p < .001, and those who received sex offender treatment, F(1, 88) = 6.2, p < .05.
During the 11-year follow-up period, 11 individuals (10.7%) sexually reoffended (i.e., they were known to have committed a sexual misbehavior that, if prosecuted in Vermont, would constitute a criminal sex offense). As a consequence of reoffending, 1 of the 11 recidivists (9.1%) was civilly committed under Act 248, 4 of the 11 (36.4%) were convicted of a criminal sex offense, and 6 of the 11 (55.0%) received no criminal or civil legal sanctions.
Overall, the 11 recidivists committed 20 new sexual offenses. Six (54.5%) committed 1; 1 (9.1%) committed 2; and 4 (36.4%), 3 committed sexual reoffenses. As detailed in Table 4, 11 of the reoffenses (55.0%) were noncontact (e.g., public masturbation and exhibitionism). Of the 9 contact reoffenses, 1 involved intercourse and the others involved sexual touching. In 11 of the 20 reoffenses (55.0%), victims were persons well-known to the offender, either a staff member, relative, or someone with whom he lived. Thus, over an 11-year period, there were 3 contact reoffenses against acquaintances or strangers.
To assess the accuracy of the Rapid Risk Assessment in predicting sexual reoffending, the area under the receiver-operating characteristic curve was used (Rice & Harris, 1995). The finding was not significant at an alpha level of .05 (receiver-operating characteristic area = .58, 95% CI = .39 to .77). Finally, chi-square analyses revealed that there were no significant differences, also at an alpha level of .05, between reoffenders and nonreoffenders on the variables described in Tables 1 and 3 or in the Participant Characteristics section of this paper.
The present study is noteworthy because it followed a nearly exhaustive sample of adult male sex offenders with intellectual disabilities within an entire geographic jurisdiction (Vermont) for a lengthy time period (11 years), beginning immediately after the jurisdiction converted to a totally community-based system of services. Identification and follow-up of individuals were enhanced because services were funded by a single agency and participants tended to be nontransient.
The rate and type of sexual reoffending among the sample, which included treated and untreated offenders over an average 5.8-year follow-up period, suggest that a jurisdiction can manage a wide range of sex offenders with intellectual disabilities in community settings in a relatively safe, cost-effective, and humane manner. Although it is difficult to compare the 10.7% sexual reoffense rate in the current study with those of other programs for sex offenders with intellectual disabilities due to methodological problems in this literature (Courtney & Rose, 2004), it is similar to previously discussed findings in the general sexual offender literature (e.g., Hansen et al., 2003; Losel & Schmucker, 2005).
Perhaps the most relevant comparison group for the sample is sex offenders without intellectual disabilities placed in the same geographic area (Vermont) during similar follow-up and time periods. Of 195 treated and untreated adult male sex offenders released from Vermont prisons between 1989 and 1997, almost a quarter (23.1%) were charged with a new sexual offense over a mean follow-up period of 5.72 years (McGrath, Cumming, Livingston, & Hoke, 2003). In a study primarily composed of probationers, 6.5% of 122 treated and untreated male sex offenders in a Vermont county were charged with a new sexual offense over a mean follow-up period of 5.24 years (McGrath, Cumming, & Vojtisek, 1998).
Of course, important variations among outcome studies should be considered when comparing reoffense rates. Participants in the present study likely received more intense supervision than did participants in other studies cited. Over two fifths of them (62.1%) received 24-hour supervision, and this limited their access to potential victims. On the other hand, the level of supervision also likely resulted in more complete identification and recording of reoffenses. Sexual offending behaviors detected in the present study were considered reoffenses whether or not criminal charges resulted.
Several other challenges of conducting outcome research with this population are evident in this study. Efforts to detect significant differences between reoffenders and nonreoffenders were likely hampered by a relatively small sample size and low recidivism base rate. In addition, the amount, frequency, quality, and duration of services provided to these participants varied considerably, and the relative impact of each could not be sorted out.
Beyond the overall reoffense rate, the type of reoffenses committed by individuals in the study is noteworthy. Whereas 83% of participants were classified as contact sex offenders, having committed primarily sexual assault and molesting offenses, over half of the reoffenses committed (55%) were noncontact, namely, exhibitionism and public masturbation. Although the goal of program interventions was to prevent any sexual reoffense, from a harm-reduction perspective (Laws, 1999), these reoffending patterns arguably represent a reduction in offense severity from participants' previous offending patterns. As would have been the case if men in this study had been institutionalized, many of the victims of their sexually abusive acts (45.0%) were the staff members who provided them services and the people with whom they lived.
The results demonstrate that an individual who has committed one type of sexual offense against one type of victim may still be at risk to commit other types of sexual offenses against other types of victims. Slightly more than half of the participants (53.4%) committed sexual offenses against victims from multiple age, gender, and relationship categories. These crossover rates are consistent with findings among sex offenders without intellectual disabilities (Heil, Ahlmeyer, & Simmons, 2003). Clearly, caution is called for in defining a sex offender's risk group solely on the basis of the identity of previous victims.
The average per person annual cost of providing community services to sex offenders here was substantial. According to data from the Vermont Agency of Human Services (2005b), it was almost two times more than the cost for non-sex offender clients with intellectual disabilities funded by Department of Disabilities, Aging and Independent Living in 2004 ($81,538 vs. $43,198). On the other hand, participants' average annual cost was almost three times less than what the projected 2004 annual per person cost would have been if Vermont had continued to operate its institutional program at Brandon Training School ($81,538 vs. $224,165). Further, nationally, the 2002 average annual cost per resident for public institutions of 16 or more persons with intellectual disabilities was $134,619, with costs increasing dramatically for smaller institutions (Rizzolo, Hemp, Braddock, & Pomeranz-Essley, 2004), and the 2004 average annual cost of keeping someone civilly committed in an institution as a sexually violent predator in the 16 states that have such statutes was $100,000 (Lafond, 2005). Many of the sex offenders served in the present study would surely be institutionalized today at a very high financial cost if Vermont had not moved to a community-based system.
Thoughtful community placement has advantages beyond cost savings. Services delivered to individuals in their natural environment, such as home, school, and community, are typically more effective than those delivered in congregate settings, where individuals are grouped together with other antisocial and sexually abusive individuals. Negative peer influence is associated with increased rates of criminal reoffending (Andrews & Bonta, 2003). Vermont sought to avoid this problem by a system of care with an average of 1.3 persons per residence. This central component of the program—small and individualized community residences—may be one of the most challenging to implement in states that do not have individualized funding systems for developmental services.
Vermont is a state of small cities, towns, and rural areas. Offenders were moved to geographic and program placements based upon safety and programmatic needs. Rural settings at a distance from potential victims were used to provide a level of personal freedom for individuals who were not an elopement risk. High risk individuals were placed in alarmed and intensively staffed homes. Apartments in towns and cities provided opportunities to fade supervision but maintain frequent check-ins and covert supervision. Rural placements offered distance from victims that would be difficult to replicate in urban areas.
An incidental study finding concerned the predictive validity of the Rapid Risk Assessment for Sex Offense Recidivism. Although this measure has been found to predict sexual recidivism with moderate accuracy in at least 17 studies of adult male sex offenders (Doren, 2002) and at least one study of adult male sex offenders with intellectual disabilities (Tough, 2001), these results were not replicated in the present study. This may have been because men in the present study, especially those who were deemed highest risk to reoffend, received high levels of supervision and had little opportunity to reoffend. It remains to be determined whether mainstream risk instruments will work well with sex offenders who have intellectual disabilities. Consistent with the recent findings of Lindsey, Elliot, and Astell (2004), we suspect that reoffense risk factors among sex offenders with and without intellectual disabilities are quite similar, but more research is needed in this area.
Perhaps the most important types of risk factors in need of study with sex offenders who have intellectual disabilities are those that are changeable, causally linked to sexual reoffending, and, therefore, important targets of intervention. These are commonly called dynamic risk factors and include variables such as antisocial attitudes, poor emotion management, deviant sexual interests, and failure to cooperate during treatment (Hanson et al., 2003). In the present study, we did not have access to data about the relationship between participants' dynamic risk factors and treatment outcome. However, development of an instrument for assessing and measuring treatment needs and progress is a current focus of our research efforts (McGrath, 2005; McGrath, Livingston, & Falk, 2007). The goal is to improve our ability to provide the appropriate types and amount of treatment and supervision and to better evaluate the effectiveness of these services.
For over a decade, Vermont has attempted to manage its statewide population of sex offenders with intellectual disabilities in the community in a safe, cost-effective, and individualized manner, and the results of these efforts have been encouraging. The model and outcomes described here, we hope, will be of assistance to other jurisdictions interested in implementing and conducting research on this approach.
Research support for this study was provided from the Office of Justice Programs, U.S. Department of Justice (Grant No. 20001-WP-BX-0019). The opinions expressed in this article do not necessarily reflect the views of the Vermont Department of Disabilities, Aging and Independent Living. We thank the staff of Vermont's Developmental Services Programs for data collection and Jennifer Garabedian for her assistance with managing the project. All correspondence should be sent to the first author.
Robert J. McGrath, MA (firstname.lastname@example.org), McGrath Psychological Services, P.C., 105 Happy Valley Rd., Middlebury, VT 05753. Joy A. Livingston, PhD, President, Flint Springs Associates, 402 Fletcher Farm Rd., Hinesburg, VT 05461. Gail Falk, JD, Chief of Community Services, Vermont Department of Disabilities, Aging and Independent Living, 103 S. Main St., Weeks Building, Waterbury, VT 05671