A Systematic Review of Multisystemic Therapy in Adolescent Sex Offenders

Discussion

To our knowledge, this is the first literature review assessing the effectiveness of MST in clinical trials of JSOs. Our findings, which are based on the literature and included six randomized controlled trials (two of which are extensions of primary studies), yielding 403 observations of 231 participants, have important clinical and research implications. We found significant differences in juvenile sexual offense measures between MST and the alternative treatments reviewed. CBT and IT were found to be no better than TAU for the measures studied. MST was shown to outperform TAU, CBT, and IT.

For the measures employed, the beneficial effect of MST was substantial. Among the six studies reviewed, five showed significant reductions in the measures used for sexual offense outcomes. A single study of forty participants, in which 52.5 percent received MST, did not separate from TAU (n = 40; MST: 21, TAU: 19) in problematic sexual behaviors as measured by the ASBI and police records.¹⁰ One study revealed 83 percent fewer arrests for sexual crimes as well as 80 percent fewer days in detention facilities.⁹ Another study found 13.13 times fewer sexual offenses (P < .01) and 8.27 times fewer arrests (P = .02) for MST recipients.¹²

While the scope of this project was to assess the effectiveness of MST in a particular cohort of JSOs, one question raised by these robust findings is how MST faired among the non-JSO conduct disorder population. A metanalysis assessing the efficacy of MST in a heterogeneous population of conduct disordered individuals revealed surprisingly modest treatment effects compared with that observed among the JSO population studied here.³ One explanation for this may be the heterogeneity of measurements. Measures of substance use, theft, and nonsexual violence are difficult to compare with many of the JSO measures. In instances in which the measures were similar, as was the case for arrest and detention, the authors theorize that authorities may be more inclined to pursue arrest and detention of JSOs, as well as to impose longer durations of confinements, than for non-JSOs. If such an idea is valid, we would anticipate untreated JSOs would show higher counts on arrests and other measures than their non-JSOs counterparts, and hence exhibit a pronounced opportunity for MST treatment to demonstrate its effectiveness, as opposed to ineffective treatments or placebo. One means of addressing this question may be for authors to publish MST findings with a secondary analysis comparing these two populations. Barring that, the authors theorize MST may be more effective for JSOs than for CD patients in general.

Of the studies in our review, two employed arrest for sexual offense as an outcome measure, another two employed adjudication or diversion, one used problematic sexual behavior (measured by Adolescent Sexual Behavior Inventory), and one used subsequent commission of a sexual offense. The measures employed are both heterogeneous and display varying advantages. In the circumstances in which offense behavior or problematic sexual behavior is employed, the report of the participant is typically the source. While participant reports may be acceptable for some diagnoses, the JSO population is particularly challenging in this regard, and any intervention which increases underreporting may be misinterpreted as treatment response. When utilizing the report of a family member or guardian, a potential bias of concealment of sexual offenses should be expected and would likely have a similar effect. Alternatively, when employing any law enforcement activity as a surrogate, the equity, partiality, and priorities of the criminal justice system become independent variables. Complicating this, many of the studies employ search of electronic police records as the source. Generally, the legal reporting mechanisms and law enforcement procedures may influence the data's integrity. Any activity which undermines accurate or uniform entry by law enforcement may alter the findings. We acknowledge that there are limited data on these measures and no widely accepted “gold standard” measure. Clinicians may wish to be mindful of the precise measure employed in a study when comparing findings across studies, interpreting study findings, or generalizing findings to their practice setting.

The enduring nature of MST effects on JSOs is also notable. Among those receiving MST, the mean treatment course approximated seven months across studies (7.0 to 7.8 months; weighted mean: 6.84 months) with each study employing a post-treatment assessment at the conclusion of the treatment course. The earliest measure for the studies reviewed occurred at month six while the longest occurred at 24.9 years following treatment conclusion. The beneficial effects of MST relative to the comparator arm were evident by six months in all but one of the studies and continued to separate statistically as much as decades later.¹⁰ One may reason that MST has a relatively rapid onset of action, while also demonstrating a sustained benefit following treatment cessation. As the studies employed a six-to-seven-month treatment duration, an intensive treatment structure, and a single treatment course, it is unclear whether lengthier, more intensive, or repetitive MST would yield any further benefit. Research employing MST of different durations, repeated courses of MST or MST administered with varying degrees of intensity may be informative. Clinicians might consider extended, intensive, or repeated MST courses for repeat offenders or those with severe symptoms because of a lack of conclusive data on optimal MST duration. The literature suggests the treatment timeframe of four to six months.¹³ While most studies observed an average MST duration of seven months, the effectiveness of prolonged MST remains undocumented. It is important to note the financial and logistical challenges posed by MST's intensity, potentially affecting the feasibility of extending this treatment approach. When examining study design and methodology, employing a cross-over study design might pose challenges. Starting with TAU, CBT, or IT before transitioning to MST is plausible. But the reverse could be problematic because of MST's prolonged onset and lasting treatment impact. The significant duration for which MST remained effective, years or decades following treatment conclusion, may make a treatment wash-out period impractical.

The sample studied is notable in several respects. There was a significant representation of Blacks and Caucasians, as well as males, in these studies (range: Blacks: 27.1% to 54%, Caucasians: 44% to 72.9%; weighted means: Blacks: 45.73%, Caucasians: 51.52%). This sample mirrored that of broader literature on juvenile delinquents.¹⁴ While sexually offensive behaviors, convictions, or arrests were significant events, studies of sexual offenders have revealed these to be relatively infrequent for any given offender. Based on these observations, there seems to be a need for creation of more refined measures for assessing sexual offense treatment, including the use of long catchment periods to observe for sentinel measures, assessing the practices of law enforcement over time and across regions, and conducting further research assessing whether, and to what degree, race and gender factor into juvenile sexual offense metrics. Effective assessment of sexual offense treatment requires a multifaceted approach, blending individual evaluations with broader societal indicators. Direct measures capture changes in deviant sexual interests, while risk assessment tools predict reoffending probabilities. Other scales gauge the offender's understanding of the impact on victims. Feedback from therapists and the community adds depth to the evaluations, and monitoring post-treatment behavior gives insights into reintegration success. A combination of these ensures a holistic understanding of treatment outcomes.

Regarding the role of gender, some have argued that females may be under recognized because of societal tendencies to dismiss women as potential victimizers, among other factors.^15,16 Additionally, sex offenders have historically been overwhelmingly thought to be males, a finding consistent with the sample studied.¹⁶ While the literature highlights male-predominant samples in studies, gender as an independent factor influencing the treatment outcomes should be considered when interpreting the results.¹⁷ Blacks have historically been highly represented in literature on conduct disorder and some have argued that diagnostic bias has increased the likelihood of clinicians making the conduct disorder diagnosis in that race.¹⁸ Racial disparities pervade numerous areas of mental health and the justice system. Historically, diagnostic practices have sometimes differed across racial lines, potentially leading to misdiagnoses or over-diagnoses among certain racial groups. Additionally, within the juvenile justice system, Black and Hispanic youth often face disproportionate rates of prosecution and more severe sentencing compared with their White counterparts.¹⁹

A less controversial factor of note is the requirement for involvement by family or guardians in MST. The presence of accessible familial support may serve as a potential confounder in evaluating MST participant outcomes. We note that our sample mirrored that of the literature on the subject generally, and that some caution may be needed when applying these findings to populations poorly represented in the samples reviewed (such as youth from geographic regions other than those studied and certain racial groups), as well as to populations from the regions with varying sociopolitical attitudes toward the rehabilitation of sex-offending youth.¹⁴

Additionally, age of study participants was reviewed and was notably uniform (mean age range: 13.4 to 14.6 years; weighted mean: 14.34 years). Given the uniformity of age, it is unclear whether, and to what degree, participant age may be an independent factor in treatment response. Among the studies reviewed, precise data on individual study participants was unavailable. Although a consistent age range is evident in multiple studies, the justification for selecting a specific age bracket is not elaborated in the literature. It could be hypothesized that as adolescents approach adulthood, the impending transition to independence and the potential for adult court evaluations regarding offenses may factor into the considerations for study parameters. Furthermore, younger children may be more susceptible to being victims rather than offenders because of their developmental immaturity. Considering that numerous offenders have experienced victimization in their past, such experiences may influence them to display offending behaviors, as suggested by our mean age group data. Consequently, definitive age-response relationships cannot be ascertained from this dataset. Moreover, the representation of participants in any given age is unclear aside from participants uniformly being under 18 years of age at study initiation. Subsequent research may benefit from authors employing secondary analysis of age-response relations, including data for specific age populations in their studies and sampling of participants of varying ages.

Both MST and community-based treatment showed improvement with problematic sexual behavior in one of the studies cited.¹⁰ This study was notable in several respects. The MST arm outperformed the TAU arm, though to a degree later determined not to separate statistically. While out of home placement was comparable in both the MST (n = 2) and TAU groups (n = 2), the MST group showed no conviction of sexual offense in comparison to TAU (n = 1; P value calculation not published). The TAU arm demonstrated some improvement over that of the pretreatment level of impairment. This study had the second smallest sample size of the six studies reviewed (MST: 21, TAU: 19). While caution should be applied whenever interpreting neutral results in a positive light, we suggest withholding judgment on this study until it reaches completion, and hypothesize that the small sample size, duration, choice of primary outcome measure, and possibly other factors may make this study underpowered. Clinicians are encouraged to examine further data if it becomes available from Fonagy, et al.¹⁰

Our conclusions are limited by the availability of studies. Despite employing the MST as the sole initial search criteria, an orthodox inclusion-exclusion criteria, and utilizing five databases to broaden the literature pool, only six citations in total were found, of which two were extensions of an initial group of four. We limited searches to articles published in English, though no articles were found in other languages which otherwise met our search criteria. One may reasonably argue that there are presently limited data on MST in JSOs and that further research is indicated generally. Despite these limitations, our review suggests MST is an effective treatment for JSOs. Additionally, the published studies highlight the strength of sample size, which is often a limitation in clinical trials. While efficacy remains an important feature when assessing any treatment, other factors such as treatment availability and cost were not assessed.

In assessing the cost-benefit of MST, several approaches may be considered. Sheidow et al.²⁰ examined the impact of MST on overall health-care utilization. In their cohort, Medicaid enrollees receiving MST experienced fewer hospitalizations, less utilization of emergency services and less overall medical care utilization. An inflation adjusted, per participant yearly savings of $1,617 was observed in their cohort. The current inflation adjusted cost would be $2,410 per participant per year and the corresponding inflation-adjusted savings would be approximately $2,410 for each youth in their cohort. There are also law enforcement, judiciary, and detention costs incurred among the population appropriate for MST. Dopp et al.²¹ revealed a net savings to taxpayers for these services, showing a savings of $3.31 for every $1 of MST treatment cost. Some caution may be required when assessing cost estimates. At present, however, there are no cost-benefit studies available related to use of MST in JSOs, and, besides those cited, few studies assess MST’s cost-benefit features. Alternatively, the psychological impact of sexual offenses on society defies easy measure and may justify even substantial costs.

Choosing treatment methods for sexual offenders prominently involves fairness considerations. If our findings are true of MST in general, some caution may be needed when devising subsequent research. From our findings, CBT, IT, and TAU provide minimal benefit to the participants studied. Alternatively, MST appears to be reproducibly effective. Assuming each treatment adheres to its standard practices, our comparison evaluates them based on their typical administration. MST is commonly applied over six to seven months, whereas the duration for CBT and IT can fluctuate related to clinician or patient preferences. Even if the time frames differ, it is pertinent to contrast them based on real-world application standards, ensuring the comparison is representative of practical scenarios. Future research will need to consider whether failing to apply MST, as may transpire in a prospective randomization, is ethical, given the existing evidence base for MST and its alternatives. We suggest that quasi-experimental MST studies, retrospective cohorts, or, when resources are limited, comparators with a waitlist, may be ethical solutions.

Of the studies reviewed, three were conducted in the United States (South Carolina, Missouri, and Maryland) and one in Europe (Great Britain). As each study employed a law enforcement measure, it is unclear if the findings from one region may be generalizable to another. The study conducted in Great Britain revealed no difference between MST and TAU, while the other three studies showed similar positive results with MST.¹⁰ In the Great Britain study, 40 individuals were enrolled, approximately half of whom received MST. Consequently, the study lacked sufficient statistical power to conclusively demonstrate the efficacy of MST compared with alternative treatment offered. The difference was not significant enough to establish a statistical significance, although no P value was reported in the study. Our primary contention is that the limited sample size makes it challenging to draw definitive conclusions from this study. A more robustly powered investigation might have yielded different results or provided greater confidence in confirming the effectiveness of MST compared with other treatments. Moreover, several differences between this study and the other five reviewed studies were identified, possibly influenced by variations in law enforcement, legal justice system practices, and health-care delivery among countries. In Sweden and Norway, youth offenders are primarily assisted through a child welfare approach, where legal sanctions are not imposed on individuals.⁴ Instead, they are referred to social services, leading to more frequent utilization of in-home services. This differs from the United States, where youth offenders are processed within the juvenile justice system. Hence, we propose caution about generalizing our findings to regions other than those studied.²² Clinicians who wish to employ MST should be aware that the specific practices or epidemiology of juvenile sexual offenses may be a factor in treatment effectiveness.