This study aimed to evaluate the completion rate and effectiveness of R&R2 MHP which is a cognitive skills program developed for MDOs and derived from the Reasoning and Rehabilitation program. The program was initially piloted in medium and high secure settings [27
] and the present findings support the feasibility of delivering the program to MDOs in medium and low security.
An important finding was the low drop-out rate, supporting the hypothesis that the group completion rate would be more favorable than that found in previously reported studies. The present study applied a very stringent completion rate of 80% attendance. The completion rate obtained in the present study of 78% is considerably higher than the rate of 50% reported by Cullen et al. [18
] using the original 36-session version of the R&R program and applying the same completion criteria as the current study. Their sample was drawn solely from medium security whereas the current study included participants from low security, just over one-third of whom were in the treatment condition. As treatment drop out has been reported to be associated with risk status [17
], it is possible that completion rates were inflated in the current study by the inclusion of lower risk patients who were more advanced in the rehabilitation pathway. Nevertheless comparison of group completers and non-completers in the present study showed no significant difference between groups in their motivation to engage in treatment, number of previous convictions and/or number of previous admissions to secure services, nor was there a significant difference in the number of patients who dropped out from medium and low security. Thus, as R&R2 MHP is 22 sessions shorter than its predecessor, program length and intensity of treatment may account for the favorable program retention.
A further and important influence on retention may be the specific adaptations that were made to the original program to improve responsivity. R&R2 MHP was designed to be more responsive to the needs of a forensic mental health population who are a more complex group of offenders, often presenting with severe mental illness, high rates of comorbid mental health problems, substance misuse and rigid cognitive styles. Moreover R&R2 MHP includes an individual mentoring paradigm which has been identified to be a supportive element associated with higher completion rates [22
The association between non-completion of OBPs and recidivism is worrying; indeed it seems that it is better to not attend an OBP at all, than start one and drop out [19
]. In the current study those who dropped out of treatment tended to have generally better social problem-solving skills, thus they may have perceived that a cognitive skills group was inappropriate and unlikely to meet their needs. However, the finding is inconsistent with that of a previous study reporting the reverse with poorer problem-solving skills being associated with program drop-out [39
]. The present study did not obtain PCL-R scores which have been found to be an important marker of risk associated with drop-out [17
]. It is a priority to identify predictors of treatment drop-out and develop methods to maintain engagement as this will have important implications for the selection of participants for group program and the management of offenders.
A second aim of the study was to evaluate the effectiveness of R&R2 MHP in MDO’s and, as hypothesised, significant treatment effects were found at outcome with small effects on self-reported measures of violent attitudes, rational problem-solving and anger cognitions. Improvements were endorsed by informant ratings of psychological and social functioning within the establishments. In order to reduce the load of the self-report battery only two measures were administered at follow up; one being the primary outcome measure of violent attitudes and the second, a relatively brief questionnaire, to determine locus of control. For the treatment group, significant small effects were found for these two measures at follow-up. Thus improvement was sustained over time suggesting that those who completed the intervention continued to use and consolidate the strategies learned in sessions after they finished treatment.
The present study found improvement on only one aspect of social problem-solving (i.e. rational) of the SPSI-RS. By contrast, other studies have reported post-treatment improvement in the Impulsivity/Carelessness, Avoidance and Total scales in MDO’s following treatment with the longer 36-session R&R intervention [17
] and in offenders with severe personality disorder following treatment with the 15-session R&R2 ADHD [28
]. At 12-month follow-up, Cullen et al. [18
] found the effect for Impulsivity/Carelessness was sustained but results indicated less improvement in negative problem orientation compared with controls. As noted by Cullen et al., the R&R program may have differential impact on the varied functional modalities on the SPSI-RS with problem-solving orientation (positive/negative) being more resistant to change than problem-solving style. Cullen et al. [18
] did not find a significant effect at outcome on the NAS-PI scales whereas the current study found a reduction in anger cognitions, possibly reflecting the greater focus on emotional monitoring and control strategies introduced in R&R2 MHP.
Nevertheless, in common with many multisite studies, a significant treatment effect was not found for every scale at outcome and, despite attempts to standardize the treatment and research protocols and ensure program integrity between the sites, there may have been variation in standards of delivery. Another possible explanation may be that most outcome measures were not re-administered at follow-up. The treatment effect at follow-up was sustained for violent attitudes and although there was no significant difference in locus of control between the two groups post-treatment, a small significant effect was present at follow up. Had other secondary outcome measures been repeated at follow-up, it is possible that a similar enhanced treatment effect may have extended to the SPSI-RS, NAS-PI and DBSP measures. This pattern of improvement has been reported in a randomized controlled trial of the R&R2 ADHD program delivered to outpatients with ADHD [40
], emphasizing the importance of including follow-up evaluations to assess treatment outcome.
The results of the present study indicate that the R&R2 MHP program was effective in reducing antisocial thinking and behaviour, which is a primary aim of the program. Evaluation of R&R in correctional facilities has generally applied reconviction rates as the primary outcome measure [12
]. Consistent with the findings of the present study, the ad hoc per protocol analysis of Cullen et al.’s [18
] 12-month follow-up data found a treatment effect for violent attitudes. Thus ‘softer’ measures evaluating antisocial attitudes are likely to be important early markers due to their association with offending [41
]. Thus antisocial attitudes and behaviors, together with reconviction rates, should be the primary benchmarks for evaluating OBPs in MDOs.
A strength of the study is the multi-site involvement, however participants were not randomly assigned to group condition. Thus in order to control for variance at baseline, ANCOVA was used with baseline Time 1 scores covarying for the dependent outcome scores and a more conservative ITT analysis selected over per protocol analysis. Nevertheless, high levels of staff turnover on wards meant that there were higher rates of missing informant data on the DBSP (that could be rated by the same member of staff across the two time points). One solution for future research would be to request that ratings are made collectively by the clinical team during ward rounds or clinical case conferences. Applying this method would additionally reduce informant bias. We also found that a record review was unhelpful as these were inconsistently recorded across sites; moreover critical incident records had a floor effect with most patients having no incidents recorded. Future research should consider using a prospective measure of aggression, such as the Staff Observation Aggression Scale Revised (SOAS-R) [45
]. Multi-site trials are thus not without limitations due to within and between-site variations among procedures and participants. This ‘clustering’ of data is particularly salient to our inclusion of participants from low and medium security sites.
A second limitation was that the sample was exclusively adult males with severe mental illness and therefore the findings cannot be generalized to a wider offender population. Third, other characteristics may have influenced outcome that were not investigated, in particular IQ [39
] self-esteem [47
], impulsivity [48
] and psychopathy [17
] have been associated with non-completion rates. Fourth, four of the five outcome measures were self-report and we aimed to minimize a positive bias by these being administered by researchers who had not been involved in treatment provision. Fifth, clinical indicators were used to evaluate effectiveness and follow-up was relatively short. Nevertheless, despite these limitations, the data provide a good starting point for further development and investigation. Future randomized controlled evaluations need to be conducted to further reduce the potential for confounding variables, with a longer follow-up period, and objective measures including reconviction data. Program evaluation could be extended to offenders living in the community, females and those with learning disability to establish if this shorter program is responsive to the needs of these groups.