Intellectual disability is defined as an impairment in intellect and adaptive functioning that begins during the neurodevelopmental period and persists throughout life.Reference McBride, Heslop, Glover, Taggart, Hanna-Trainor and Shevlin1,Reference McKenzie, Milton, Smith and Ouellette-Kuntz2
The National Institute for Health and Care Excellence (NICE) recommends that individuals with intellectual disabilities who are in crisis due to their mental health or behaviour receive support from a multidisciplinary intensive support team (IST).3 ISTs aim to reduce the occurrence and length of in-patient admissions by providing crisis care, in-reach support within in-patient units to facilitate discharge, and positive behaviour support in the community.3–Reference Hassiotis, Kouroupa, Hamza, Morant, Hall and Marston5
NHS England's recommendations for ISTs include the provision of 24/7 face-to-face crisis support, multidisciplinary support delivered by specialists in the management of challenging behaviour, and integration between ISTs and specialist community teams that deliver routine care.6,7 There is little evidence regarding which service-level or individual participant characteristics of ISTs are associated with effective service delivery, whereas this is better established for crisis services for the general adult population and for older adults.Reference Hassiotis, Robotham, Canagasabey, Romeo, Langridge and Blizard8–Reference Coleston-Shields, Challis, Worden, Broome, Dening and Guo10 This information is important in guiding the commissioning of services.Reference Hassiotis, Walsh, Budgett, Harrison, Jones and Morant4,Reference Hassiotis, Kouroupa, Hamza, Morant, Hall and Marston5
The Intensive Support Teams for Adults with Intellectual Disabilities and Challenging Behaviour (IST-ID) national study investigated service-level characteristics and individual patient outcomes in existing ISTs.Reference Hassiotis, Walsh, Budgett, Harrison, Jones and Morant4,Reference Hassiotis, Kouroupa, Hamza, Morant, Hall and Marston5 Stage 1 was a cross-sectional study that surveyed 73 ISTs in England and identified two types of such service provision: independent, where the IST is separate from the local community intellectual disability service, and enhanced provision, where the IST is integrated within it.Reference Hassiotis, Walsh, Budgett, Harrison, Jones and Morant4 Stage 2 was a cohort study comparing clinical outcomes between the two IST types at baseline and 9-month follow-up. The study did not find any differences in levels of challenging behaviour or any other secondary outcomes between the two types of IST provision, neither did it show significant cost differences. It concluded that local variations in need may well determine which model is adopted, in the absence of other requirements, including model fidelity.Reference Hassiotis, Kouroupa, Hamza, Morant, Hall and Marston5
The present study complements the previous project by exploring how individual and service characteristics relate to clinical outcomes, generating hypotheses about potential critical components of IST care. This is very important because it brings into consideration the theoretical fidelity of ISTs, opening the discourse on the intervention theory behind such teams in the community care of people with intellectual disabilities.
Aims
To investigate whether IST service-level characteristics relate to reduction of challenging behaviour measured by a validated instrument through the secondary analysis of data collected in the IST-ID study.
Method
Data from participants who had enrolled in stage 2 of the IST-ID study were included in this secondary analysis. In stage 2, a random sample of ISTs was selected out of the 73 ISTs that had taken part in stage 1 of the study.Reference Hassiotis, Walsh, Budgett, Harrison, Jones and Morant4,Reference Hassiotis, Kouroupa, Hamza, Morant, Hall and Marston5 To be included in stage 2, the IST must have been operating for at least 12 months, funded for the duration of the study and offering intensive support to adults with mild to profound intellectual disabilities.Reference Hassiotis, Kouroupa, Hamza, Morant, Hall and Marston5
Full details of the procedures involved in data collection in the two stages of the IST-ID study can be found in the relevant publications.Reference Hassiotis, Walsh, Budgett, Harrison, Jones and Morant4,Reference Hassiotis, Kouroupa, Hamza, Morant, Hall and Marston5
One individual (L.T.) reviewed all items from the survey administered in stage 1 of the IST-ID study to identify the most clinically important and cross-referenced them against published standards for ISTs.3,Reference Lloyd-Evans and Johnson11,12 The results of this initial screening were reviewed by A.H., resulting in a shortlist of 27 items. This shortlist of items was discussed in a team meeting with all authors, when these were further refined and items that were considered to be less clinically relevant were removed (the longlist of items is provided in the supplementary material available at https://dx.doi.org/10.1192/bjb.2023.94). This resulted in a list of 16 items. We then combined some categories to produce binary or categorical variables with fewer categories. Finally, six items were removed owing to a lack of variation where all ISTs fell into one category, giving rise to a final list of ten service-level characteristics. This final list of variables is shown in Table 1.
List of service-level variables identified from the IST-ID national survey
IST-ID, Intensive Support Teams for Adults with Intellectual Disabilities and Challenging Behaviour; ADLs, activities of daily living.
The primary outcome was the mean change in challenging behaviour, measured by the total score on the Aberrant Behavior Checklist-Community version 2 (ABC-C) from baseline to 9-month follow-up.Reference Aman, Singh, Stewart and Field13 The secondary outcomes were the mean changes in each of the ABC-C subscale scores within the same time frame.
The scores from other validated questionnaires at baseline and demographic data were included in the multilevel linear regression models as covariates. The validated questionnaires included the affective/neurotic disorder and psychotic disorder subscales of the Psychiatric Assessment Schedule for Adults with Developmental Disabilities (PAS-ADD) Checklist, the Short Adaptive Behavior Scale (SABS) as a measure of adaptive functioning, the Threshold Assessment Grid (TAG) as a measure of clinical riskReference Moss, Prosser, Costello, Simpson, Patel and Rowe14–Reference Slade, Powell, Rosen and Strathdee16 and the presence of attention-deficit hyperactivity disorder and/or autism spectrum disorder. Demographic data included age group (18–24 and ≥25 years), ethnicity (White, Black, Asian and minority ethnic), gender and accommodation type.
Analysis of the primary and secondary outcomes
All statistical analyses were completed using Stata version 17 MP for Windows. We used descriptive statistics to describe the services and population.
Multilevel linear regression was used to model the effect of service-level characteristics on the mean change in ABC-C total score from baseline to 9-month follow-up, while controlling for covariates. These analyses were repeated for each of the secondary outcomes.
To ensure that the assumptions of multilevel linear regression were met, the residuals were plotted on a histogram and a standardised normal probability plot to assess for normality and were plotted against the predicted values to assess for homoscedasticity.
These analyses were completed as a series of three steps. In step 1, separate models were constructed for each of the service-level variables identified in the screening process described above. Within each model, the change in ABC-C total or ABC-C subscale score from baseline to 9-month follow-up was the dependent variable, the service-level variable was included as a fixed-effect independent variable. The purpose of this step was to identify which of the service-level variables had P < 0.10 and should therefore be included in the final model in step 3.
In step 2, the participant-level variables were included as fixed-effects independent variables.
In step 3, the model was constructed as in step 2 with the addition of the service-level variables identified in step 1 as fixed-effects independent variables. The purpose of step 3 was to determine whether the addition of the service-level variables affected the statistical significance of the model.
Ethics and data protection
This secondary analysis of data from the IST-ID study involved the processing of anonymised data from stages 1 and 2 of study, of which A.H. is the guarantor. These data were stored and processed on a password-protected desktop computer at University College London, Division of Psychiatry, in compliance with General Data Protection Regulation (GDPR) policy and the Data Protection Act 2018. All data were kept strictly confidential.
The IST-ID study was performed in accordance with the Declaration of Helsinki. The Health Research Authority reviewed and approved the study and all amendments (substantial and non-substantial). Ethical approval for the IST-ID study was granted by the London Bromley Research Ethics Committee (reference 18/LO/0890). Further ethical approval was not required for the current study since it involved the secondary analysis of existing anonymised data obtained in the IST-ID study.
Consent of study participants to use their data was covered in the main ethical application for review. Outcomes of the analysis do not allow re-identifying participants and we did not transfer data to facilities outside of the UCL Division of Psychiatry.
Results
Data regarding 226 participants across 21 ISTs were analysed. Tables 2 and 3 present service- and individual-level characteristics. In all the multilevel linear regression analyses, the residuals were normally distributed and there was no evidence of heteroscedasticity.
Baseline characteristics of the included intensive support teams (ISTs)a
a. See Table 1 for clarification of the characteristics (service-level variables).
Baseline characteristics of the included participants
SABS, Short Adaptive Behavior Scale; ADHD, attention-deficit hyperactivity disorder; ASD, autism spectrum disorder; TAG, Threshold Assessment Grid; PAS-ADD, Psychiatric Assessment Schedule for Adults with Developmental Disabilities Checklist; ABC-C, Aberrant Behavior Checklist-Community version 2.
The univariate analysis conducted in step 1 demonstrated that working hours was the only variable found to be significantly associated with a change in ABC-C score (P < 0.1) (Supplementary Table 1).
Table 4 presents the results of the multivariable analyses in steps 2 and 3 for the primary outcome, change in ABC-C total score from baseline to 9-month follow-up. Results of the multivariable analyses in steps 2 and 3 for the secondary outcomes are shown in Supplementary Tables 2–6.
Multivariable analyses in steps 2 and 3, for the primary outcome, where change in ABC-C total score from baseline to 9-month follow-up was the dependent variable
ABC-C, Aberrant Behavior Checklist-Community version 2; PAS-ADD, Psychiatric Assessment Schedule for Adults with Developmental Disabilities Checklist; SABS, Short Adaptive Behavior Scale; TAG, Threshold Assessment Grid; ADHD, attention-deficit hyperactivity disorder; ASD, autism spectrum disorder.
In step 1, working hours had a P-value <0.10 for the primary outcome and for some of the secondary outcomes (ABC-C irritability, lethargy/social withdrawal and hyperactivity/non-compliance subscales), so it was included in the multivariable analyses. However, for one of the secondary outcomes (inappropriate speech subscale), the exemptions and care coordination variables had a P-value <0.10 and therefore were included in the multivariable analysis for this secondary outcome.
For the primary outcome, change in ABC-C total score from baseline to 9-month follow-up, the TAG score, a measure of the severity of mental illness, was the only independent variable that was statistically significant (1.52, 95% CI 0.49–2.55). This was found in both the unadjusted and adjusted models.
As regards the secondary outcomes, change in the irritability subscale score at 9 months was significantly associated with living in an adult social care setting (−4.25, 95% CI −7.63 to −0.86) and the affective subscale of the PAS-ADD Checklist (4.44, 95% CI 0.59–8.29). Change in the hyperactivity subscale score was found to be significantly associated with the affective subscale of the PAS-ADD Checklist (4.65, 95% CI 0.46–8.83) in both the adjusted and unadjusted models.
Change in the ABC-C lethargy/social withdrawal subscale was found to be significantly associated with gender (2.98, 95% CI 0.04–5.93) when the model was adjusted for working hours.
There were no associations between the stereotypical behaviour and inappropriate speech subscales and demographic or clinical variables.
Discussion
To our knowledge, there has not yet been any study examining the active ingredients of intensive support teams (ISTs) for adults with intellectual disabilities who display challenging behaviour. Prior to the IST-ID study, studies of IST service characteristics and patient outcomes had been conducted within individual ISTs, included small samples of ISTs or had been conducted within individual regions.Reference Hassiotis, Robotham, Canagasabey, Romeo, Langridge and Blizard8,Reference Martin, Costello, Leese, Slade, Bouras and Higgins9,Reference Lowe, Felce and Blackman17–Reference Coelho, Kelley and Deatsman-Kelly19 Davison et al conducted a cross-sectional study that collected data from community teams that supported people with intellectual disabilities displaying challenging behaviour, but it did not investigate patient outcomes or any associations between teams and behaviour.Reference Davison, McGill, Baker and Allen20
We did not find any IST-level characteristics to be associated with changes in behaviour (improvement) as measured by the ABC-C total score. However, we saw a pattern emerge whereby participants’ clinical variables appeared to influence outcomes. The TAG score (a measure of the severity of mental illness), accommodation and affective status were significantly associated with change in primary and secondary measures of challenging behaviour.
Our results indicate that a higher TAG score at baseline is associated with an increase in ABC-C total score from baseline to 9-month follow-up. This association is supported by the well-documented relationship between mental illness and challenging behaviour in people with intellectual disabilities, although what mediates this relationship is unclear and likely to be multifaceted.Reference Westlake, Hassiotis, Unwin and Totsika21–Reference Reyes-Martín, Simó-Pinatella and Font-Roura24 Individuals with intellectual disabilities who are mentally unwell may display challenging behaviour as a secondary or atypical presenting feature.
The relationship identified between the affective subscale of the PAS-ADD Checklist and the irritability and hyperactivity subscales of the ABC-C may have arisen since those symptoms can be transdiagnostic and therefore present in many different mental disorders.Reference Eaton, Tarver, Shirazi, Pearson, Walker and Bird25–Reference Sturmey, Laud, Cooper, Matson and Fodstad27 Furthermore, the relationship identified between gender and the lethargy/social withdrawal subscale may be reflective of female gender having been identified as a risk factor for depression and mental ill health more generally among adults with intellectual disabilities.Reference Hsieh, Scott and Murthy28,Reference Chester, Chaplin, Tsakanikos, McCarthy, Bouras and Craig29
There are potentially several explanations as to why service-level features did not appear to influence patient outcomes in this study. First, it could be that individual characteristics are more important than service-level characteristics in underpinning outcomes, as has been found in a previous study of predictors of readmission in mental health services.Reference Osborn, Lamb, Canaway, Davidson, Favarato and Pinfold30 Second, there may have been service-level characteristics that we did not measure and that may have been important in this context, such as area deprivation, intensity of support and specific input to crisis triage. Finally, it could be that the quality of care provided had a greater impact on patient outcomes and that this was not captured by the service-level characteristics we identified. For example, NICE and NHS England recommend that staff working within ISTs must be skilled and competent in delivering interventions to reduce risks associated with challenging behaviour and that these interventions should be delivered in a way that is person centred and is in line with relevant treatment manuals.3,6,7,31 In addition, NICE recommends that the clinical competency of staff should be regularly evaluated.31 These recommendations may be more difficult to measure, albeit they are an important factor in determining the clinical outcomes of patients supported by ISTs. A previous systematic review identified longer opening hours and inclusion of psychiatrists within the staff skill mix as central factors in implementing crisis resolution teams to prevent hospital admissions.Reference Wheeler, Lloyd-Evans, Churchard, Fitzgerald, Fullarton and Mosse32 However, in our study, our analysis did not support this.
Limitations
It is important to highlight that this study has a number of limitations that must be considered when interpreting these findings. The independent variables included were binary and categorical and therefore may not have been sensitive enough to represent more subtle variation in how the ISTs were organised and structured. In addition, this secondary analysis focused on one clinical outcome, the ABC-C and its subdomain scores, although many other clinical outcomes may have been important indicators of service efficacy. It is also important to recognise that this was an exploratory study; although we sought to include the most clinically relevant service-level variables, we did not test any predetermined hypotheses.
About the authors
Lucretia Thomas is a Foundation Year 1 Doctor at Homerton University Hospital, London, UK. Brynmor Lloyd-Evans is Professor of Mental Health and Social Inclusion, Division of Psychiatry, University College London, London, UK. Louise Marston is Professor of Clinical Trials Statistics, Department of Primary Care and Population Health, University College London, London, UK. Angela Hassiotis is Professor of Psychiatry of Intellectual Disability, Division of Psychiatry, University College London, London, UK.
Supplementary material
Supplementary material is available online at https://doi.org/10.1192/bjb.2023.94.
Data availability
Anonymised data that support the findings of this study are available on request from the corresponding author, A.H.
Author contributions
L.T.: development of the research protocol and data analysis plan, data handling and analysis with statistical software, data analysis and interpretation of results, writing the manuscript draft, approval of the final manuscript. B.L.-E.: study conceptualisation, development of the research protocol and data analysis plan, data analysis and interpretation of results, comments regarding the manuscript, academic supervision, approval of the final manuscript. L.M.: study conceptualisation, development of the research protocol and data analysis plan, statistical analysis, data analysis and interpretation of results, comments regarding the manuscript, academic supervision, approval of the final manuscript. A.H.: study conceptualisation, development of the research protocol and data analysis plan, data analysis and interpretation of results, comments regarding the manuscript, academic supervision, approval of the final manuscript.
Funding
This secondary analysis of data from the IST-ID study received no specific grant from any funding agency, commercial or not-for-profit sectors.
Declaration of interest
None.
Open access
eLetters
No eLetters have been published for this article.