Comparing co-production approaches to dynamic risk assessments in a forensic intellectual disability population: outcomes of a clinical pilot

Abstract

Purpose

A co-produced clinical practice that aims to improve outcomes through a partnership with service users is becoming increasingly important in intellectual disability (ID) services, yet these approaches are under-evaluated in forensic settings. This study aims to explore and compare the feasibility of two approaches to co-production in the completion of dynamic risk assessments and management plans in a secure setting.

Design/methodology/approach

A convenience sample of adults admitted to a secure specialist forensic ID service (N = 54) completed the short dynamic risk scale (SDRS) and drafted risk management plans under one of two conditions. In the first condition, participants rated the SDRS and risk management plan first, separately from the multidisciplinary team (MDT). In the second condition, participants and MDTs rated the SDRS and risk management plan together.

Findings

In total, 35 (65%) participants rated their risk assessments and 25 (47%) completed their risk management plans. Participants who rated their risk assessments separately from the MDT were significantly more likely to complete the SDRS (p = 0.025) and draft their risk management plans (p = 0.003). When rated separately, MDT scorers recorded significantly higher total SDRS scores compared to participants (p = 0.009). A series of Mann-Whitney U tests revealed significant differences between MDT and participant ratings on questions that required greater skills in abstraction and social reasoning, as well as sexual behaviour and self-harm.

Originality/value

Detained participants with an ID will engage in their dynamic risk assessment and management plan processes. The study demonstrates the impact of different co-production methodologies on engagement and highlights areas for future research pertaining to co-production.

Citation

Morris, D.J., Webb, E.L., Stewart, I., Galsworthy, J. and Wallang, P. (2021), "Comparing co-production approaches to dynamic risk assessments in a forensic intellectual disability population: outcomes of a clinical pilot", Journal of Intellectual Disabilities and Offending Behaviour, Vol. 12 No. 1, pp. 23-36. https://doi.org/10.1108/JIDOB-08-2020-0014

Introduction

The absence of reciprocal and equitable relationships with service users and carers have been cited as key contributing factors to poor care and abuse in inpatient and ID services (Flynn, 2012; Social Care Institute for Excellence, 2015). Forensic services, which function in the context of significant restrictions on liberties, hold considerable influence over service users’ lives. The restrictive legal and policy frameworks within such services (Lewis-Morton et al., 2017) typically hinder collaborative approaches, and their use of restrictive practices have been subject to increasing scrutiny (Brophy et al., 2016; Cusack et al., 2018).

By contrast, co-production has been defined as an approach that focuses on “developing more equal partnerships between people who use services, carers and professionals” (National Collaborating Centre for Mental Health [UK] and Social Care Institute for Excellence [UK], 2013). It emphasizes the importance of reciprocal relationships with health-care providers (Vennik et al., 2016), where service users are considered capable of and central to making care-related decisions (Boyle et al., 2006a). As such, co-production defines health-care professionals and service users as experts with different and complimenting knowledge and skills, which, together, lead to better outcomes, ostensibly through the “relocation of power” (Le Boutillier et al., 2015). It is often formulated as a tripartite process of designing, developing and delivering services in equal partnership between stakeholders (NESTA, 2013).

Within community ID services, co-production approaches are highly relevant given the disenfranchised nature of the population and high levels of family involvement in care (Ham and Davies, 2018; Roberts et al., 2012). In these settings, co-production is evident in service development (Doherty et al., 2018), staff training (Billon, et al., 2016), research domains (Armstrong et al., 2019) and care planning (Ham and Davies, 2018). Yet, co-production is less evident in inpatient forensic services, where detained individuals experience significant restrictions on their liberties and involvement in decision-making.

Co-production and forensic services

Co-production has been formulated as a human rights-based approach (Roberts et al., 2012), empowering marginalized groups (Rycroft-Malone et al., 2016). As such, it enjoys a complex relationship with forensic settings that are typically hierarchical (Lewis-Morton et al., 2017), with cultural and legal barriers impeding change. In this context, commentators have questioned whether a co-production approach, especially “shared agendas” can truly exist (Pilgrim, 2018).

Nonetheless, an initial study exploring service user experiences in high secure settings found that inclusion strategies were valued by staff and service users, fostered relationships with care teams and improved service user well-being and self-esteem (Lewis-Morton et al., 2017; McKeown et al., 2014). They were also reported to improve role satisfaction and counteract “alienation” from self-image experienced by staff (McKeown et al., 2014). Additionally, co-produced psychological formulations in forensic settings have generated important themes of shifting power, meaningful collaboration, co-reproduced understanding, motivating and empowering service users and an agent to support organizational change (Lewis-Morton et al., 2017). Yet, there is a lack of empirical evaluation of co-production initiatives in forensic settings, which is incongruent with current policy and practice guidance that advocates co-production in all aspects of clinical care.

The assessment and management of risk are underexplored in relation to co-production (Boardman & Roberts, 2014), despite being advocated by the Department of Health (2009) and Royal College of Psychiatry (2017). Indeed, developing meaningful collaborative approaches to risk assessment and management poses a challenge given the compulsory context of care. Nonetheless, challenging the traditional dominance of clinicians is key to supporting service to users fully engage in recovery (Roberts et al., 2012). Such an approach also aims to increase service user awareness and insight into risks, maximize opportunities to develop adaptive coping skills (Boardman & Roberts, 2014) and reduce the “nanny state approach” (Better Regulation Commission, 2006). Given the current focus on reducing restrictive practices, exploring the feasibility and effectiveness of co-produced risk assessments and plans are clinical priorities.

Nevertheless, the inherent challenges of implementing co-production in secure settings are compounded by the lack of consensus for a definition of co-production (Kaehne et al., 2018) and the absence of empirical evaluations of different co-production approaches. Co-production is not a unitary construct, and evaluations of the outcomes of different co-production approaches are wanting. The absence of guidance and empirical evidence outlining how to implement co-production (Bee et al., 2015) arguably contributes to their lack of presence in risk management. Accordingly, attempts to embed co-produced approaches may be boosted if they are underpinned by methods that have been subject to empirical evaluation.

Additionally, the lack of accessible tools to facilitate co-production of risk assessment in an ID population may also present as a barrier to its implementation. For co-production to be a viable approach to the assessment and management of risk within ID services, tools must be accessible and responsive to the capacities of those with an ID. One such tool, which has been developed as an accessible risk assessment tool for an ID population is the short dynamic risk scale (SDRS Quinsey, 2004), placing lesser cognitive demand on raters through the inclusion of fewer items and simplistic language and item structure. In consideration of the barriers that those with an ID face in relation to cognitive functioning, it is essential that appropriate tools, such as the SDRS, be used when exploring the feasibility and effectiveness of a co-produced approach to risk assessment.

Consequently, in consideration of the current evidence, a number of priorities are apparent. Firstly, it is important to establish whether service users will engage in their risk assessment when given the opportunity. Furthermore, an evaluation of varying methods of co-production and their impact on service users’ engagement is warranted.

Aims of the current study

The current study reports on the implementation of co-produced dynamic risk assessments and management plans in a specialist ID forensic service. The following questions were explored:

Q1.

Would service users engage in and co-rate their dynamic risk assessment and risk management plans?

Q2.

Do different types of co-production methods impact on service user engagement in co-rating their dynamic risk assessment and risk management plans?

Q3.

Do different types of co-production methods impact on participant and multidisciplinary team (MDT) risk ratings?

Method

Design

Staff and service users of a medium-secure unit, low-secure unit and locked rehabilitation unit completed the SDRS under one of two co-production conditions. In condition one, participants (service-users) rated their SDRS and drafted risk management plans separately from the MDT (“rating separately”). In condition two, participants rated their SDRS and drafted risk management plans with the MDT present in their clinical update meeting (“rating together”). Both conditions were implemented at all levels of security in the service. Condition allocation was pragmatic, determined by the units in the service at the point when co-production approaches to risk assessments were introduced, and reflected the preferences of the clinical teams at that time.

Participants

Participants were a convenience sample of 54 adults detained to a secure specialist forensic ID service. This Tier 4 service provides specialist forensic care to working-age adults who predominantly have mild ID (World Health Organization, 1992) and comorbid complex mental health disorders. At the time of the study, the service could support up to 64 service users. It consisted of two medium-secure, two low-secure and two locked units, each consisting of a male and female unit.

Table 1 illustrates that 57% of the sample were male. The most common developmental disorder diagnosis was mild ID (78%) and personality disorder was the most common co-morbid psychiatric diagnosis (57%). Just over half of participants (53.7%) were detained under Section 3 of the Mental Health Act (1986, Amended 2007) and 50% were detained to medium-secure services. In total, 31 participants (57.4%) were given the opportunity to complete their risk assessment tools under condition one (“rated separately”), whilst 23 (42.6%) were given the opportunity to complete them under condition two (“rated together”). Analyses revealed no significant differences between conditions in relation to demographic variables such as age, level of security, legal status and type of diagnoses.

Multidisciplinary team raters

Across the six units, the MDT providing ratings for the risk assessments and developing risk management plans consisted of multiple clinical staff members from a range of disciplines with varying lengths of experience. Across all units, the MDT comprised of a consultant psychiatrist (1–20 years’ experience), clinical psychologist (2–10 years’ experience) or forensic psychologist (≤1 years’ experience), an assistant psychologist (≤1 years’ experience), social worker (<1–10 years’ experience), occupational therapist (<1–16 years’ experience) and a senior member of the nursing team (≥2 years’ experience).

Materials

The short dynamic risk scale (Quinsey, 2004).

The SDRS (Quinsey, 2004) is an eight-item scale that measures problems with taking responsibility, coping skills, anger, anxiety, obnoxious or insulting behaviour, a lack of concern for others, self-care and living space. Three additional items developed for using the tool in forensic settings assessing aggression, sexual behaviour and self-harm were also included (Brewer et al., 2016). Items are rated on a five-point Likert scale, with increasing scores representing increasing problems and responses are summed to determine an overall score (0–44), with higher scores indicating greater dynamic risk. The tool is completed based on the behaviour of the individual for the month prior to the assessment.

Risk management plans.

The risk management plans were developed within the service and included the following six sections to be completed in an open-ended format; “early warning signs/triggers for risk behaviours”, “what makes the behaviour less likely to occur”, “behaviours to focus on increasing (positive behaviours to shape)”, “plan to achieve this”, “behaviours to reduce (risk behaviours)” and “plan to achieve this”.

Procedure

Participant training in short dynamic risk scale completion

Prior to rating the SDRS, all participants completed a training package developed by the service (Radcliff & Staniford, 2016). The package was delivered across as many individual sessions as required by the participant. The training pack included an overview of risk, the SDRS and its scoring instructions, practice opportunities for completing the SDRS and how to use the tool to inform risk management plans. The training also included adaptive and maladaptive examples of behaviours rated on the SDRS. Participants were also given access to a visually-adapted version of the SDRS developed by the Speech and Language Therapy team using “widget” (Widgit Symbols, 2012), which included a pictorial Likert scale of green smiling face for adaptive behaviour (no problems over the past month) through to amber and red faces for the increasing severity of problem behaviours. MDTs also completed a training session, supported by a booklet that covered the same areas as the participant (service-user) training. All training sessions were facilitated by registered psychologists and assistant psychologists working in the service. The current evaluation reports on the first administration of the SDRS within the service.

Completion of the short dynamic risk scale and risk management plans

Condition 1: rated separately from multi-disciplinary team.

Participants were given the opportunity to rate their SDRS and draft their management plans prior to their monthly care planning meeting. The SDRS and risk management plans were rated individually, supported by an assistant psychologist. The participant-rated SDRS and risk management plans were then added to by the MDT in the care planning meeting, without the participant present. Participants were offered the opportunity to discuss the SDRS further with their MDT during their care plan meeting.

Condition 2: rated together with multi-disciplinary team.

Participants rated the SDRS and risk management plan with the MDT present in their care planning meeting. Staff and participants rated the individual items at the same time, with both offering separate scores that were recorded on the same SDRS form. This approach was also used for drafting risk management plans.

Ethical considerations

The study was approved by the internal clinical effectiveness team as a service evaluation project. The SDRS and risk management plan document had been introduced for routine practice within the service and the current study was an evaluation of the tool.

Data analysis

Data was entered into IBM statistics SPSS version 25. The Shapiro Wilks test revealed that total SDRS scores were normally distributed, hence parametric analyses were used. Tests of normality also revealed that most individual SDRS item scores were not normally distributed and non-parametric analyses were used when reporting on individual items. In consideration of the small sample size and the fact that not all variables were normally distributed, median values are reported in conjunction with mean values for individual SDRS items.

To assess whether participants would co-score their dynamic risk assessments (Q1) descriptive statistics were used. To explore the impact of the condition on completion rates for the SDRS and risk management plans (Q2), Fisher’s exact tests were conducted and odds ratios were calculated. Additionally, the rate of attrition between completion of the SDRS and risk management plan was calculated in each condition. To test for differences in participant and MDT SDRS ratings (Q3) two strands of analyses were completed. Firstly, an analysis of variance was conducted to explore the difference in total SDRS scores, both across and within conditions. Where a significant difference was found within a condition, differences in individual SDRS item scores between staff and participants in that condition were explored using a series of Mann-Whitney U tests.

Results

Question 1: would service users engage in and co-rate their dynamic risk assessment and risk management plans?

All participants (N = 54) were given the opportunity to rate their own SDRS assessments. Overall, 35 (65%) participants self-rated their dynamic risk assessment. In total, 25 participants (47%) also contributed to the draft of their risk management plan. Clinical notes and feedback from MDTs and service users revealed no incidents of aggression, distress or interpersonal conflict in either condition. There was also no evidence that the completion of the tools contributed or led to ruptures in therapeutic relationships.

Question 2: do different types of co-production methods impact on service user engagement on co-rating their dynamic risk assessment and in drafting their risk management plans?

In total, 24 participants (77%) in the “rating separately” condition and 11 (48%) in the “rating together” condition completed the SDRS. A Fisher’s exact test showed that those in the “rating separately” condition were significantly more likely to complete their dynamic risk assessment (p = 0.025). The odds of completing the SDRS were almost four times greater for those in the “rating separately” condition (OR 3.74, 95% CI: 1.16–12.10).

In total, 20 participants (62%) in the rating separately condition drafted their risk management plan, compared to 5 (22%) in the “rating together” condition. A Fisher’s exact test revealed that participants in the “rating separately” condition were significantly more likely to draft their risk management plan (p = 0.003). The odds of drafting their risk management plan were six times higher for those in the “rating separately” condition (OR 6.18, 95% CI: 1.79–21.34). In the “rating separately” condition, of those who completed the SDRS (n = 24), 16.7% (n = 4) did not go on to draft their risk management plan. By contrast, in the “rating together” condition, of those who completed the SDRS (n = 11), 54.6% (n = 6) did not go on to draft their risk management plan.

Reasons for declining to co-rate the short dynamic risk scale.

Reasons for declining to co-rate the SDRS were recorded in the participant’s electronic clinical notes and reported by MDTs. The most common reason for declining to complete the SDRS when rated together with the MDT was not wanting to rate the tool in their care plan update meeting with all of the MDT present or not wanting to attend their care plan update meeting due to other reasons, precluding completion of the tool. Participants in both conditions also declined due to suspicion as to why they were being asked to do so. The remaining participants did not give a reason.

Question 3: do different types of co-production methods impact on participant and multidisciplinary team risk ratings?

Total scores for the short dynamic risk scale when rating the short dynamic risk scale separately from the multidisciplinary team.

Figure 1 illustrates that the mean total score for participants in the “rating separately” condition was 13.29 (Median = 12, SD = 9.27) and 20.71 (Median = 23, SD = 10.70) for the MDT. This difference was statistically significant F (1, 54) = 7.29, p = 0.009) with the MDT reporting higher total scores.

Total scores for the short dynamic risk scale when rating the short dynamic risk scale together with the multidisciplinary team.

The mean total scores when rating the SDRS together were 18.00 for participants (Median = 18, SD = 9.99) and 18.43 for staff (Median = 18, SD = 8.27). This difference was not statistically significant, F (1, 32) = 0.02, p = 0.90.

Interactions between rater and scoring condition on total short dynamic risk scale scores.

The results of a two-way analysis of variance (ANOVA) showed a significant effect of both rater and condition, collectively, on total SDRS scores, F (3, 84) = 2.72, p = 0.049. Further post-hoc comparisons with Bonferroni adjustment revealed that significant differences in scores lay between patients and staff in the “rating separately” condition (p = 0.035) only. There were no significant differences in total scores when comparing between staff nor between patients across conditions. As no significant differences in total SDRS scores between patients and staff in the “rated together” condition were found, differences between these groups for individual SDRS items were not explored.

Individual short dynamic risk scale item scores between participant and multidisciplinary team raters in the “rating separately” condition.

A series of Mann-Whitney U tests revealed significant differences in scores for “taking responsibility”, (p = 0.019), “obnoxious behaviour” (p = 0.012), “lack of concern” (p = 0.016), “living space” (p = 0.009), “self-care” (p = 0.030), “sexual behaviour” (0.035) and “self-harm” (p = 0.002) between patients and staff in the “rated separately” condition. Figure 2 illustrates mean SDRS item scores, by condition.

Discussion

The current study is the first to explore the feasibility and impact of different methods of co-producing dynamic risk assessments in an ID population. The findings demonstrate that co-production is indeed feasible. When given the opportunity, the majority of participants actively engaged in risk assessments and, to a lesser extent, the drafting of risk management plans. Furthermore, there was no evidence of an adverse therapeutic outcome for participants or their relationships with staff in either condition. This finding is encouraging and reassuring to clinical teams, given the importance of the therapeutic alliance in supporting service users to make progress (Rycroft-Malone et al., 2016).

Our finding that the method of co-production significantly impacted on engagement demonstrates the importance of exploring optimal methodologies in developing co-produced models of care. The significantly higher level of engagement with the SDRS when completed separately from the MDT could be accounted for by a number of factors. This approach may have comparatively lower sensory, cognitive, interpersonal and anxiety barriers to engagement. It is also possible that completing the risk assessment and management plan before the care team increased participants’ sense of value and ownership within this process. By comparison, collaborative completion of the measures with the MDT in a larger meeting scenario, in which wider care issues were also deliberated, may have been more challenging for participants. The subjective experiences of conditions should be explored in future studies to better account for this finding. Nonetheless, whilst the “rating separately” condition produced higher levels of engagement, it cannot be assumed that this approach is clinically more effective in the long term. The “rating together” condition may provide greater opportunities for informal discussions that could reciprocally educate participants and MDTs about the individual nature of risk, which, in turn, may have greater therapeutic benefits and outcomes. Consequentially, whilst the current study explored the impact of the method of co-production on engagement, further studies are needed to explore the impact of the co-production method on reciprocal education, therapeutic alliance, predictive validity of ratings and clinical outcomes.

Fewer participants contributed to their risk management plans than SDRS risk assessments, with greater attrition rates in the “scoring together” condition. Attrition between rating risk assessments and drafting risk management plans could reflect a number of individual, systemic and training variables. Firstly, completing both tasks in the same session may have placed too high a demand on the participant’s resources relating to sustained attention, emotional regulation, as well as confidence and skills. The lower completion rates may also reflect shortcomings in the participant teaching pack and support offered around risk management planning. Whilst the study design, which did not control for fatigue, precludes drawing a causal link between condition and attrition, the significantly higher attrition rates in the “completing together” condition indicates that the method of co-production may impact on willingness to complete multiple tasks in a single session.

The patterns of scores for total and individual items on the SDRS also generated notable results. The reporting of similarities and differences between staff and participants on individual SDRS items mirror findings in other studies comparing beliefs about care needs (Najim & McCrone, 2005). Overall, whilst higher total SDRS scores were yielded by MDT raters compared to participants, this difference was only significant in the “rating separately” condition. Rating separately may have reduced the impact of suggestibility, namely, the tendency to mirror the suggestions of the MDT, which may have been an effect in the rating together condition.

Differences in total scores may also reflect differences in the level of training and experience between MDTs and service users. Evidence suggests that the length of time trained in using risk assessment tools and the level of qualification impacts on health professionals’ scoring accuracy when assessing risk for violence (Teo et al., 2012). When rating separately, participants were not able to consult the wider experience of professionals who had greater experience in risk assessments, which may account for differences in ratings. Future studies exploring the impact of service user experience on their risk assessment ratings over time, as experience increases, would be of interest.

Furthermore, significant differences between MDT and participant ratings were largely noted on SDRS items that explored more abstract concepts, such as “taking responsibility for behaviours”, “teasing and obnoxious behaviour” and “lack of concern for others”. Arguably, these items place greater demands on cognitive skills such as abstraction, theory of mind and socio-moral reasoning, which can be a deficit in people with an ID (Langdon et al., 2011; Owen & Maratos, 2016; Scotland et al., 2015), which may account for differences in ratings. Additionally, it is arguably more difficult to rate oneself on constructs that are essentially socially constructed about how others experience our behaviours. Whilst deficits and challenges in these skills to do not prelude engagement in co-production, they highlight the importance of ensuring service-users are offered appropriate training and support. Consequently, differences in scores may have also reflected the quality of the participant training regarding the explanation and exemplification of abstract behaviours. Differences were also noted between MDT and participant ratings for self-harm and sexual behaviour, which may reflect differences in MDT and participants knowledge and perceptions of “normality” of these behaviours. Indeed, previous research has demonstrated significant differences in caregiver and service user perceptions of sexuality, experience and knowledge (Szollos & McCabe, 1995). Hence, in the absence of reciprocal discussions between the MDT and participants in these conditions, differences in perception were likely reflected in ratings.

By contrast, similarities in scores in the “rating together” condition could reflect a consensus view between the MDT and participants, rather than reflecting an “independent” voice of participants. As a result, it is possible that participant responses may have reflected social pressure, impression management or desirability in the presence of key decision makers. Further studies exploring optimal designs could consider exploring the impact of the presence of health-care workers on service user decision-making.

Finally, in the absence of follow-up studies, it would be presumptive to assume that MDT ratings have greater predictive validity and that differences in ratings reflect deficits in the skills and insight of participants. Indeed increased rating in staff could also reflect risk aversive decision-making and it would be presumptuous to assume objectivity in MDT decision-making (Gowensmith, & McCallum, 2019). Further follow-up studies using prospective designs are needed to explore this particular finding, not least to support service users and staff to reflect on their perceptions and beliefs around risk.

Clinical and research implications

The current study suggests that co-produced dynamic risk assessments are feasible to implement within forensic ID services, although further work is needed to identify optimal methods for promoting engagement. As such, a research agenda that identifies optimal co-production methods and training approaches to support this process is warranted. Within this, the impact on therapeutic engagement, clinical progress and outcomes need also to be considered.

This process may include changes in the risk assessment tools used, to ensure accessibility. Whilst the SDRS was developed for people with an ID, forensic ID services are mandated to use tools not developed for this population. For example, the Historical Clinical Risk Management-20 (HCR-20) (Webster et al., 1997), which is considerably longer with more complex language, item structure and scoring criteria, is arguably inaccessible to ID populations. Structured professional judgement tools require registered clinicians, who have already completed professional training, to complete formal training courses. The cognitive skills needed to read, process and comprehend these tools compared to the educational context of service users with an ID is an important barrier to overcome. As demonstrated by the current study, the SDRS is one of such tools, which allows service users with an ID to engage in their own risk assessment and as such could be implemented within ID services. That said, the development of accessible tools that address the full range of risk-related needs are needed.

In addition, developing an understanding of how service users make decisions about their risk is needed to support service users’ full engagement in this process. To ensure that co-production approaches evolve from tokenism to a position of informed partnership, a comprehensive training programme to maximize these skills is warranted. Exploring decision-making heuristics would build an understanding of the frames of reference for participants and their MDT and highlight the support and education needed. Such research also has the potential to lead to innovative interventions for service users that improve self-awareness and monitoring, which could have a wider impact on recovery, as well as fortifying the equitable role of service users in their care.

Limitations of the current study

The current study was a post-hoc review of a novel clinical initiative. As such, the study has a number of shortcomings, not least in the robustness of the design. Experimental condition was allocated according to service availability and preference, and was not randomized. This precludes assessing whether intrinsic differences between the units, rather than the experimental condition, account for the findings. Nonetheless, given that three units were included in each condition, the risk of bias is somewhat mitigated. The current methodology also precluded participants having a choice for the condition. This would have been more consistent with the ethos of co-production and may have increased engagement. In addition, whilst participants completed an SDRS training session, formal measures to monitor the efficacy of the training package were not used. Data relating to the therapeutic and clinical impact of the pilot was also not collected, precluding exploration of the wider impact of the two approaches and how participants experienced this process. Finally, the study did not record information about how participants made decisions about risk nor did it control for social pressure or desirability in responses.

Conclusions

The empirical evidence base for co-production is in its infancy and our understanding of the clinical utility of such approaches, and the factors driving their success, is limited. The essence of co-production is the recognition of service users as important resources who can add unique insights into health-care decision-making (Clayson et al., 2018). Moreover, the emphasis on experience, respect and dignity, which are powerful constructs, are increasingly being recognized as human rights and of particular relevance to disenfranchised and marginalized individuals. The current study demonstrates that service users are willing to be involved in this key area of care planning. Despite the limitations outlined above, the findings demonstrate the feasibility of a co-produced approach in the assessment of dynamic risk within an inherently restrictive setting.

Figures

Figure 1

Mean total SDRS scores for participants and MDTs in both conditions

Figure 2

Mean participant and MDT scores for individual items on the SDRS by condition

Table 1

Sample characteristics (n = 54)

Variable	n	(%)	M (SD)	Range
Age (years)			32.5(10.3)	18–54
Gender
Male	31	57.4
Female	23	42.6
MHA status
Section 3	29	53.7
Section 37	2	3.7
Section 37/41	17	31.5
Section 38	1	1.9
Section 47	1	1.9
Section 47/49	3	5.6
CPIA	1	1.9
Level of security
Medium secure	27	50.0
Low secure	24	44.4
Locked	3	5.6
Primary developmental disorder diagnosis (ICD-10)
Border lineintellectual disability	1	1.9
Mild intellectual disability	42	77.8
Moderate intellectual disability	6	11.1
Cerebral palsy	3	5.6
Hyperkinetic disorder	2	3.7
Primary psychiatric diagnosis (ICD-10)
No psychiatric disorder	7	13.0
Personality disorder	31	57.4
Psychosis	10	18.5
Schizoaffective disorder	4	7.4
Bipolar disorder	1	1.9
Obsessive compulsive disorder	1	1.9