An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
National Guideline Alliance (UK). Mental Health Problems in People with Learning Disabilities: Prevention, Assessment and Management. London: National Institute for Health and Care Excellence (NICE); 2016 Sep. (NICE Guideline, No. 54.)
Lack of early recognition of mental disorders in people with learning difficulties leads to negative consequences for the person affected, and for their family. Factors that influence this poor recognition include: a lack of knowledge in health and social care staff, and families of carers, about the signs and symptoms of mental disorders; diagnostic overshadowing by the learning disabilities or physical illness; how well paid carers know the person and how well information is shared within and across paid carer teams; and difficulties the person with learning disabilities may have communicating their distress. This lack of recognition can lead to no or ineffective treatment or inappropriate resource-use. Effective treatment requires an assessment and the development of a care plan to ensure that the best available interventions are provided.
Family members and carers of people with learning disabilities are likely to be sensitive to any changes in, for example, behaviour, loss of skills, requiring more prompting, or other indications of distress. Therefore, when families raise such concerns staff should investigate them. Paid carers are less likely to know the person they support as long as, or as well as a family carer, and may support several people with diverse needs; hence, they may be less sensitive to such changes than family carers, Staff and paid carers need to also be aware that a person with learning disabilities might have more than 1 mental health problem. Effective training and supervision for paid carers, policies on sharing information within and across support teams, documentation of skills and needs, and support should be geared towards promoting effective recognition, because of the importance of early identification in optimising treatment outcomes.
Facilitating communication is fundamental to all aspects of the identification and assessment of mental health problems (and the care and support that follows) for people with learning disabilities. Communication can be influenced by (a) the person’s developmental level, (b) additional neurological, sensory, or mental health needs, (c) confidence, (d) medication side effects, (e) being brought by someone else for health and not being used to speaking up, and (f) the accessibility of the communication style of the healthcare staff; (g) social circumstances; and (i) setting/environment.
The first point of contact for the family or paid carer of a person with learning disabilities is usually the GP, and primary care is responsible for 90–95% of treatment of mental health problems in the general population (Meltzer, 1995). However, the in-depth, comprehensive assessment needed to identify a suspected mental health problem in people with learning disabilities is not possible in primary care. A GP therefore typically seeks to establish the following: whether or not there is a medical problem; the risk of diagnostic over-shadowing; that family members and carers of the person with learning disabilities might not have all of the information about the person’s symptoms and distress; and that mental health problems can present differently to those presented by the general population with the same mental health problem. Once obvious physical health causes are ruled out, GPs may investigate and initiate treatment for general health needs and common mental disorders in adults, and also refer to secondary services or the specialist learning disabilities service (depending upon local care pathways) for assessment of complex presentations in adults and young people with learning disabilities.
In secondary care, time is typically scheduled for more comprehensive assessments, using information from multiple sources. Assessments are often multidisciplinary, to access the expertise of different professional groups, and the sharing of information and coordination of care is important. Professionals will involve family members or paid carers as well as the person with learning disabilities unless there is a specific reason not to, such as safeguarding or the expressed preference of the adult or young person with learning disabilities. Assessments without carer input may lead to omission of information and inadequate risk assessment. Information collected from interviews, mental state examination, physical examination, investigations and case-note reviews are integrated and interpreted (Cooper et al., 2006). This includes the effective measurement of psychopathology, including distinguishing it from long-standing traits and considering all possible psychopathology. Differential diagnosis is undertaken, taking account of multiple-morbidity, phenotypes (e.g. thyroid disorders and sleep apnoea in Down syndrome), high prevalence of epilepsy, polypharmacy, and other physical conditions and impairments which might mimic mental health problems, affect the most appropriate management of the person’s care and require treatment.
The resulting formulation will inform the management and support plan, which needs to be developmentally appropriate, and usually involves family and/or paid carers in addition to the person with learning disabilities.
The capacity of the person with learning disabilities to make decisions about treatment and give consent is fundamental to all aspects of the assessment process, and it is important to recognise that they may have capacity for making some but not all decisions. Also, their capacity is not fixed over time and can vary, such as during episodes of mental ill-health problems.
The British Psychological Society and Royal College of Psychiatrists jointly published guidance on assessment, diagnosis and management of dementia (British Psychological Society, 2015). They found that standard approaches are not appropriate in people with dementia due to pre-existing cognitive impairments, and that diagnosis requires comparison with what is normal for that person, rather than by a neuropsychological assessment or questionnaire rating scale. Differential diagnosis is essential, in view of the under-recognition of multiple morbidities. Because of the high prevalence of dementia in people with Down’s syndrome, where there is a prevalence of 40% over 50 years of age (Oliver & Holland, 1986; Prasher, 1995), some services offer baseline neuropsychological assessments to people with Down’s syndrome in their early adult life. Other services view such assessments as potentially intrusive and overly labour-intensive. Functional assessments of skills are necessary to inform all care planning by paid carers and in learning disabilities services, and, if formalised and documented, can be useful in subsequent dementia assessments. Primary health care services usually have regular contact with people with Down’s syndrome who are under their care because of their known predisposition to other physical conditions. These consultations, as well as annual health checks, provide opportunities for the brief assessment of alert signs of dementia, to trigger secondary care referral for assessment when indicated.
The review protocol summary, including the review question and the eligibility criteria used for this section of the guideline, can be found in Table 8. A complete list of review questions and review protocols can be found in Appendix F; further information about the search strategy can be found in Appendix H.
Clinical review protocol summary for the review on the incidence and prevalence of mental health problems.
Due to the large number of existing systematic reviews considering the incidence and prevalence of mental health problems, and the volume of work that would be required to conduct a de novo search, the GC agreed to use existing reviews as a source of evidence and to update these papers. Primary papers identified in the existing reviews were obtained and extracted because the existing reviews did not describe specific characteristics of the included studies that the GC were interested in examining (such as, the degree of learning disabilities and the source of the sample).
However, because a large number of relevant systematic reviews of varying quality were identified, a pragmatic approach was taken to only consider systematic reviews considering a number of different mental health problems as these were considered by the GC to be the potentially most informative. Systematic reviews focusing on only one mental health problem were not considered.
Furthermore, only cross-sectional or cohort studies with population-based or administrative samples (samples identified through contact with a range of services for people with learning disabilities) were included. Studies with specific populations of people (such as an inpatient or outpatient population) were excluded because of the likely overestimation of rates of mental health problems in these studies, unless they were on genetic conditions (for example, all participants in a register of people with Fragile X syndrome).
While the relative risk of a mental health problem in a population with learning disabilities compared with a non-learning disabled population was the preferred outcome, there were very few studies which included a non-learning disabled population and reported these outcomes. As a result, rates of mental health problems in people learning disabilities were also extracted and presented. Where more than 1 study reported a particular mental health problem and when the data did not appear to be skewed, the range of rates was presented. Intraquartile ranges were presented if 4 or more studies reported a particular mental health problem.
There were 4 existing systematic reviews (Buckles et al., 2013; Einfeld, 2011; Kerker, 2004; Whitaker & Read, 2006) which were used as a source for papers and a new search was conducted to identify papers examining a number of different mental health problems which were published after the search date of the reviews:
The update search identified 22 cross-sectional and cohort studies that were included in the evidence synthesis:
All studies were published in peer-reviewed journals between 1968 and 2014. Further information about included and excluded studies can be found in Appendix J.
There were 34 cross-sectional or cohort studies that examined the incidence or prevalence of mental health problems in adults with learning disabilities.
Of the included studies:
Summary of findings can be found in Table 9 (population-based studies), Table 10 (administrative-based studies), and Table 11 (studies with specific populations considering genetic conditions). The methodological quality of all papers is reported in Table 12.
Summary of findings table for the review of the prevalence (and some incidence) of mental health problems in adults with learning disabilities (population-based studies).
Summary of findings table for the review of the prevalence of mental health problems in adults with learning disabilities (administrative-based cross-sectional studies).
Summary of findings table for the review of the prevalence of mental health problems in adults with learning disabilities – genetic conditions (specific population cross-sectional studies only).
Quality assessment of the studies included in the review of the incidence and prevalence of mental health problems in adults with learning disabilities.
There were 3 studies that examined the risk of mental health problems in adults with autism (N=19, 420): 1 population-based (Melville et al., 2008) and 2 administrative-based (Bakken et al., 2010; Jacobson, 1982). Rates for these studies are reported in the footnotes in Table 9 and Table 10 below.
Most studies reported the rates of mental health problems across all degrees of learning disabilities; some reported rates for different levels of learning disabilities. This is summarised in the summary of findings tables below.
Only 2 studies included a non-learning disabled population and were able to provide data on relative risk (all population-based): (Gostason, 1985; Melville et al., 2008). Most other studies reported prevalence and some reported incidence (whether it is incidence or prevalence is indicated in the footnotes).
20 studies (3 cohort, 17 cross-sectional) examined the prevalence of mental health problems in children and young people with learning disabilities.
Of the included studies:
Summary of findings can be found in Table 13 (population-based studies), Table 14 (administrative-based studies), and Table 15 (population-based studies considering genetic conditions). The methodological quality of all papers is reported in Table 16.
Summary of findings table for the review of the prevalence of mental health problems in children and young people with learning disabilities (cross-sectional and cohort population-based studies).
Summary of findings table for the review of the prevalence of mental health problems in children and young people with learning disabilities (cross-sectional and cohort administrative-based studies).
Summary of findings table for the review of the prevalence of mental health problems in children and young people with learning disabilities – genetic conditions (population-based cross-sectional studies only).
Quality assessment of the studies included in the review of the prevalence of mental health problems in children and young people with learning disabilities.
No studies examined the risk of mental health problems in children and young people with autism in addition to learning disabilities.
Most studies reported the rates of mental health problems across all degrees of learning disabilities; some reported rates for different levels of learning disabilities. This is summarised in the summary of findings tables below.
Only 5 studies included a non-learning disabled population and were able to provide data on relative risk (Emerson, 2003b; Emerson et al., 2010; Emerson & Hatton, 2007; Linna, 1999; Rutter et al., 1970). All other studies reported prevalence.
Knowing the incidence and prevalence of mental health disorders in people with learning disabilities may lead to better prediction, identification (and thus more timely management) and possibly prevention of mental health problems in this population and has therefore, indirectly, potentially important resource implications. However, this is an epidemiological review question and economic analysis is not applicable.
The review protocol summary, including the review question and the eligibility criteria used for this section of the guideline, can be found in Table 17. A complete list of review questions and review protocols can be found in Appendix F; further information about the search strategy can be found in Appendix H.
Clinical review protocol summary for the review on methods/instruments for case identification of mental health problems in people (children, young people and adults) with learning disabilities.
No evidence was found on case identification methods or instruments for mental health problems for people with learning disabilities. In the context of the particular profile of difficulties, including communication difficulties, experienced by people with learning disabilities the GC deemed it inappropriate to consider indirect evidence from case identification tools used in non-learning disabled populations as the diagnostic accuracy and psychometric properties were likely to be significantly different. Due to the lack of evidence, the GC agreed to use the nominal group technique in order to develop recommendations. The nominal group technique method used in this guideline is described in Chapter 3.
Key issues in case identification for this population were identified from the available literature (for example Cooper & Simpson, 2006), including the competence frameworks covering the identification of mental disorders (Roth et al., 2011) and from discussions during the GC meetings. These were used to generate nominal statements to be rated by the GC. Nominal statements relating to accessibility were designed to cover a range of factors that make identification of mental health particularly challenging in this population. An example of a statement that was rated highly by the committee is: ‘Changes in behaviour that could indicate the presence of a mental health problem in a person with a learning disability include behaviour that challenges, social withdrawal, avoidance and agitation’.
Questionnaires were distributed (round 1), and completed and returned by 13 of 17 committee members. Percentage consensus values were calculated, and comments collated, for each statement. The rankings and comments were then presented to the GC members and used to inform a discussion of the issues raised by members’ comments in relation to the draft recommendations. One completed questionnaire was received after the results had been collated and was therefore not presented to the GC. The ratings presented below are the final ratings and include all completed questionnaires. A sensitivity analysis was also conducted to ascertain whether these additional ratings would have dramatically influenced the results. A second round of ratings was conducted by the GC, immediately following the presentation and discussion of results, and the results from round 2 were collated and presented to the GC. A brief summary of the process is provided in Table 18 below. The full list of statements and ratings can be found in Appendix U whilst blank copies of the questionnaires used can be found in Appendix T.
Summary of nominal group technique process followed for the development of recommendations on case identification of mental health problems in people with learning disabilities.
No studies assessing the cost effectiveness of methods/instruments used for case identification of mental health problems in people with learning disabilities were identified by the systematic search of the economic literature undertaken for this guideline. Details on the methods used for the systematic search of the economic literature are described in Chapter 3.
The GC agreed that:
No evidence on the cost effectiveness of methods/instruments used for case identification of mental health problems in people with learning disabilities is available.
| Recommendations |
|
|---|---|
| Relative values of different outcomes | The GC discussed the importance and relevance of various outcomes in the evidence. The group were interested in the difference in risk of mental health problems between people with learning disabilities and people without learning disabilities. If this was not available in the literature, they were also interested in the rates of incidence or prevalence of mental health problems. The group were particular interested in which mental health problems are more common in people with learning disabilities and in which populations (that is, different degrees of learning disabilities, children and young people or adults, underlying genetic disorders). This could then inform which case identification tools might prove most useful and guide staff behaviour when identifying mental disorders in this population. When assessing tools or methods for case identification, the GC agreed that using a tool which accurately identified cases, and therefore chose sensitivity and specificity as the primary outcomes. While sensitivity and specificity are usually the most valuable for case identification instruments, the psychometric properties of validity and reliability were also chosen as important outcomes to help assess these characteristics of tools in the learning disabilities population, particularly those which have been validated and found reliable in the general population. |
| Trade-off between clinical benefits and harms | While no evidence was found on case identification tools in this population with the criteria specified in the protocol: with no more than 3 items or lasting no longer than 10 minutes), there was evidence on the prevalence (and some incidence) of mental health problems in this population. Unfortunately, there were only 11 papers which included a comparison with the population of people without learning disabilities in order to examine the relative risk in people with learning disabilities. These studies did not cover the rates of degrees of disability or usually the different types of mental health problems. The GC appreciated the difficulty in determining any additional potential risk of mental health problems where this was not reported or calculable from the existing evidence. However, in the absence of comparative data, they considered the rates of prevalence and incidence in light of what is reported in the relevant NICE guidelines for different mental health problems. The GC agreed that in addition to separating the evidence by types of mental health problems, the evidence should also be presented separately for children and young people and adults, by source of sample (that is, population versus administrative samples), and were available for specific phenotypes. The intention was to be able to identify specific groups at higher risk of specific mental health problems, and so on. However, this limited the data further as many studies did not report the rates in the desired ways (for example, by degree of learning disability). The GC noted that it was difficult to draw conclusions from the evidence presented about higher rates of mental health problems by the above groupings. This view was based on both the quantity of the data, the way the evidence was reported, and the confidence in the estimates reported (see ‘Quality of the evidence’ overleaf). The GC noted that for some conditions, the rates reported in the evidence are higher for adults with learning disabilities than those without learning disabilities including schizophrenia, dementia, and depression. Rates of specific mental health problems were also notably high in adults with autism. Rates of conduct disorder, ODD, and ADHD also appeared from the literature to be higher in children and young people with learning disabilities than those without learning disabilities. However, the group noted that the relative rates for some conditions and the prevalence or incidence rates from some studies where relative rates were high did not appear to be higher than the general population. The GC noted also the lack of data on mental health problems in people with genetic conditions. They noted that data on some conditions which their experience indicates have higher rates of mental health problems was not demonstrated in the evidence. The GC particularly discussed the harms of not identifying mental health problems in a person with learning disabilities in that those who are not identified as having mental health problems will go untreated. However, as there was no evidence identified on tools to help aid this identification, they discussed the need for more research into the development and validation of these tools. While the GC were unable to identify specific populations at higher risk of specific mental health problems from the evidence to inform recommendations, they were of the opinion that people with learning disabilities are an underserved population whose mental health problems often go unnoticed (as their behaviours are often thought to be related to the learning disabilities) and therefore untreated. As such, they agreed that case identification is an area that should be addressed in the guideline and so the GC developed recommendations using formal consensus methods (the nominal group technique, see section 3.5.7) in the absence of evidence. |
| Trade-off between net health benefits and resource use | Identification of mental health problems in people with learning disabilities may have modest resource implications. The GC estimated that initial assessment of a suspected mental health problem in a person with learning disabilities may last about 30 minutes and can be done by professionals with skills and knowledge in the care of this population. The GC agreed that identification questions already recommended in NICE guidelines should be used for this purpose, adjusted, as appropriate, for people with learning disabilities. The GC considered that a percentage of people with suspected psychosis or suspected dementia need to be referred to specialist services, in line with existing NICE guidance for people with suspected psychosis or dementia, and, specifically, to a psychiatrist with expertise in assessing and treating mental health problems in people with learning disabilities. The GC acknowledged the additional resource implications relating to such referrals and also to any adjustments required in the process of identification of mental health problems in this population. However, the GC expressed the view that having the same identification pathways with necessary adjustments ensure equality of care between people with learning disabilities and those without learning disabilities assessed for a suspected mental health problem. Furthermore, the GC considered the current under-detection of mental health problems in people with learning disabilities and, based on their experience and evidence in the general population, advised that undiagnosed, and thus often untreated, mental health problems incur much higher health and social care costs compared with the costs associated with the initial assessment of a suspected mental health problem. The GC expressed the opinion that identification of mental health problems in people with learning disabilities is going to lead to more timely, effective and cost-effective assessment and management of the mental health problems. This, in turn is expected to result in better outcomes for people with learning disabilities, their family and carers as well as in potential cost savings to health and social care services, as mental health problems will be managed appropriately and at earlier, and possibly less severe, stage. |
| Quality of evidence | The GC were concerned about the source of the samples from which studies were based including the fact that studies with administrative samples may overestimate the rate of mental health populations. However, the GC agreed that removing studies with administrative samples completely would leave minimal population-based studies for which to consider. They agreed that administrative studies should be presented in a table, supplementary to the population-based studies, with an added caveat of quality. While population-based studies are ideal, administrative populations (those defined by the participants having learning disabilities) can potentially miss participants who are not using learning disabilities services, particularly those with milder learning disabilities. As such, the studies from administrative populations might not represent the learning disabilities population as a whole. However, the GC noted that most people with learning disabilities, particularly those who may at any point call upon the use of services suitable for people with learning disabilities who have mental health problems, are likely to be included within administrative populations; they agreed that presenting them was informative. The GC noted that where the relative risk was reported, the confidence intervals were often very wide, covering both the possibility that the true rate of mental health problems in the studies was lower or higher than the general population. The GC noted as well some inconsistency within the existing studies with the results being heterogeneous with wide ranges of rates of mental health problems across different studies. However, it was difficult to formally assess the reasons for this. The methodological quality of the evidence ranged from weak to strong among the included studies. Given the variation in the quality of the included studies, the GC was not very confidence in the results. |
| Other considerations | The GC decided, on the basis of the outcome of the nominal group technique, to develop recommendations in the following areas:
While the group initially agreed through the nominal group technique that a referral should be made to a specialist healthcare professional for assessment after identification, on further discussion they agreed that this may not always be necessary. They agreed there could be an initial triage assessment after identifying a potential problem, which may be sufficient to develop an effective care plan prior to considering referral. Given the overlap between content of the 2 types of assessment initially proposed (brief and comprehensive) the GC agreed to remove this distinction from the guideline recommendations. The triage assessment should include assessment review of the nature, severity and duration of the problem and lead to the development of an action plan including possible referral for further assessment and intervention. Given the lack of evidence on case identification tools to aid the identification of mental health problems in people with learning disabilities and the persisting difficulties in identifying mental health problems in this population, the GC recommended research in developing reliable and valid tools for case identification. They noted that brief case identification tools (for use in any setting) exist for the identification of common mental health disorders in people without learning disabilities and so, they recommended future research into this area. |
Develop or adapt reliable and valid tools for the case identification of common mental health problems in people with learning disabilities, for routine use in primary care, social care and education settings.
The review protocol summary, including the review question and the eligibility criteria used for this section of the guideline, can be found in Table 19. A complete list of review questions and review protocols can be found in Appendix F; further information about the search strategy can be found in Appendix H.
Clinical review protocol summary for the review on assessment of mental health problems in people (children, young people and adults) with learning disabilities.
There were no studies found which addressed the most appropriate structure for an assessment of mental health problems. As such, the nominal group technique was used to determine general principles for assessment. The method of the nominal group technique used in this guideline is described in Chapter 3.
The assessment process was originally split into brief and comprehensive assessment and a separate questionnaire was developed for each area. However, later this delineation was discarded as the nominal statements were reviewed (there was very considerable overlap on agreed statements) and recommendations developed (see discussion in the recommendations and link to evidence in section 4.6). Key issues for each area were similar, and were identified through the same sources; available literature, GC discussion and feedback from the service user focus groups. Nominal statements relating to brief and comprehensive assessment were developed from these sources in order to be distributed to the GC. Separate questionnaires were developed and distributed for each of brief and comprehensive assessment. Both sets of statements were designed to address a range of relevant areas including methods of data collection, risk assessment, competencies of staff and outcomes from each type of assessment.
The round 1 questionnaire for brief assessment was completed and returned. Percentage values were calculated, and comments collated, for each statement. The rankings and comments were then presented to the GC members at a GC meeting and used to inform a discussion of the issues raised by members’ comments in relation to the draft recommendations. Following this process, 2 further completed questionnaires were received and were not included in the data presented to the GC. These data points are included in the data presented. An example of a statement that was rated highly by the GC is: ‘Staff conducting a brief assessment should be aware that what presents as a mental health problem might be caused by an underlying physical health problem’.
There was a high level of agreement with many of these statements; however, a second round of rating to further clarify some issues raised through the GC’s comments was undertaken. Nominal statements were developed using this feedback, and presented to the GC for rating in questionnaire format at a GC meeting immediately following the discussion of the results from round 1. Percentage agreement values and comments were again collated, and presented and discussed at a subsequent GC meeting. A brief summary of the process can be found in Table 20 below. The full list of statements and ratings can be found in Appendix U whilst blank copies of questionnaires used can be found in Appendix T.
Summary of nominal group technique process followed for the development of recommendations on brief assessment of mental health problems in people with learning disabilities.
The round 1 questionnaire for comprehensive assessment was completed and returned by GC members. Percentage values were calculated, and comments collated, for each statement. The rankings and comments were then presented to the GC members at a committee meeting and used to inform a discussion of the issues raised by members’ comments in relation to the draft recommendations. An example of a statement that was rated highly by the GC is: ‘A comprehensive assessment should be undertaken in a collaborative manner and maximise the contribution of all people involved’. A second round of ratings was not considered to be necessary as it was agreed that any comments made by the GC could be incorporated adequately into the wording of recommendations. A brief summary of the process can be found in Table 21 below. The full list of statements and ratings can be found in Appendix U whilst blank copies of the questionnaires used can be found in Appendix T.
Summary of nominal group technique process followed for the development of recommendations on comprehensive assessment of mental health problems in people with learning disabilities.
In total, 30 studies met the eligibility criteria for this review, covering 19 instruments. The included studies are summarised narratively below. Further information about both included and excluded studies as well as the methodology checklists can be found in Appendix K.
The identified studies predominantly included adults. One study, which focused solely on a general measure of mental health problems in children and young people, used a sample which excluded adults (Einfeld & Tonge, 1995). However, 2 studies used mixed populations of young people and adults: 1 (Verbrugge et al., 2011) investigated a range of risk assessment tools and another investigated the Psychiatric Assessment Schedule for Adults with Developmental Disabilities (PAS-ADD) interview (Moss et al., 1997). Both studies were presented in the review for adults only as most of the participant were adults.
Identified studies which met the inclusion criteria covered a general measure of mental health problems, depression, dementia, and the assessment of outcomes. No studies were identified which met the inclusion criteria for other areas (including anger, anxiety disorders, personality disorders, PTSD, and the assessment of communication).
Studies which included the pre-specified reference standard, full psychiatric or psychological assessment, were included in this review (except for studies considering the risk of an event). However, many studies compared results against another tool instead of this reference standard so a 2-staged approach was taken:
As studies reporting data on the same tool did not often report sufficient data to synthesise the results (for example, reporting data for the 2×2 table used to calculate sensitivity and specificity) it was not possible to perform meta-analysis of the results for any tool. As such, the results are summarised narratively below. Where there was sufficient data reported (true positives, true negatives, false positive, false negative or a 2×2 table to calculate these figures), receiver operating characteristic (ROC) curves were produced; these are found in Appendix O.
There were 15 studies (N=4069) covering 8 tools which met the eligibility criteria for stage 1 of this review: Charlot et al. (2007); Devine et al. (2010); Gerber and Carminati (2013); Glenn et al. (2013); Gonzalez-Gordon et al. (2002); Gustafsson (2005); Hove and Havik (2008); Janssen and Maes (2013); Matson and Smiroldo (1997); Matson et al. (1997); Moss et al. (1997); Moss et al. (1998); Prosser et al. (1998); Sturmey et al. (2005); Swiezy et al. (1995); van Minnen (1994). See Appendix O for ROC curves.
No studies met the eligibility criteria for stage 2 of this review as there were no studies using tools found adequate in stage 1 as a reference standard.
The DASH-II is a questionnaire measure designed to identify psychopathology in adolescents and adults with severe or profound learning disabilities. The measure has 84 items across 13 subscales; anxiety, depression, mania, PDD/Autism, schizophrenia, stereotypies, self-injury, elimination, eating, sleeping, sexual, organic and impulse control. It is rated by a staff or family-member familiar with the individual and takes around 10–15 minutes to complete. The DASH-II costs £192.
Matson 1997 and Matson & Smiroldo 1997 investigated the utility of the DASH-II to identify psychopathology in adults with learning disabilities. Of the 13 subscales, the authors found that only the mania subscale had adequate sensitivity and specificity to identify mental illness in people with severe to profound learning disabilities and a Diagnostic and Statistical Manual of Mental Disorders (4th edition) (DSM-IV) Axis 1 diagnosis (n=22; sensitivity=92.3%, specificity=99.5%). The sensitivity of the anxiety subscale was 21.2% whilst the sensitivity of the depression subscale was 73.3%. The internal consistency (Cronbach’s alpha) for the DASH-II was 0.79 whilst the convergent validity was r=0.43–0.91.
The MASS is a free, clinician-administered, semi-structured interview, based upon DSM-IV-TR (Text Revision) criteria that is designed to identify mood and anxiety difficulties in people with learning disabilities. It consists of 36 items and takes between 30 and 60 minutes to administer.
Charlot 2007 investigated the utility of the MASS in adult inpatients (n=93) with all degrees of learning disability. They found that the MASS could be used to identify any anxiety disorder with 96% sensitivity and 81% specificity, of GAD with 73% sensitivity and 69% specificity, of depression with 92% sensitivity and 73% specificity and of mania with 60% sensitivity and 100% specificity (see Appendix O for ROC curves).
The PAS-ADD interview is a clinician-administered measure designed to identify psychopathology, providing either ICD-10 or DSM-5 (DSM 5th edition) diagnoses, for use in people with learning disabilities and a level of expressive language that enables them to make a verbal contribution to the interview. The core interview consists of 66 items, takes roughly 3 hours to administer, and can be supplemented with other assessments where necessary. The PAS-ADD interview costs £225 and clinicians are required to attend training in both interviewing and coding before using it.
An older version of the PAS-ADD interview was assessed in 2 studies: Gonzalez-Gordon 2002 (n=80), assessed the PAS-ADD interview in adult populations whilst Moss 1997 investigated its’ effectiveness in a mixed population of young people and adults (from ages 16 to 69) (n=95). Gonzalez-Gordon found that the PAS-ADD interview identified neurotic symptoms with 63% sensitivity and 84% specificity. In their study, Moss 1997 found that the PAS-ADD interview identified anxiety disorders across all levels of learning disabilities, when compared with psychiatric diagnosis alone, with 100% sensitivity and 97% specificity. The PAS-ADD interview had lower sensitivity and specificity for symptoms of schizophrenia (sensitivity=76.1%, specificity=88%), depression (sensitivity=75% specificity=88%), and any psychiatric disorder (sensitivity=75.9%) in this study (see Appendix O for ROC curves).
The PAS-ADD checklist is a 25-item screening questionnaire for psychopathology in adults with learning disabilities, designed to be completed by individuals who know the person with learning disabilities well. The PAS-ADD checklist produces 3 sub-scores; affective or neurotic, possible organic condition (including dementia) and psychotic disorder. The scoring system includes threshold scores which can be used as indication for further assessment. The PAS-ADD checklist costs £60 for a pack of 20 forms.
The PAS-ADD checklist was assessed in 3 studies of people with mixed degrees of learning disabilities: Moss 1998 (n=201) and Sturmey 2005 (n=226) assessed the PAS-ADD checklist in adult populations and Gerber 2013 which investigated a French version of the PAS-ADD checklist in adults (n=126) with all levels of learning disability. Sturmey 2005 found that the sensitivity of the PAS-ADD checklist affective/neurotic disorders subscale was 66%, and the specificity was 70%. In their study Moss 1998 found that the PAS-ADD checklist had Spearman’s rank correlation inter-rater reliability of 0.79 and Cohen’s kappa inter-rater reliability of mean 0.42 across the subscales. The internal consistency of the PAS-ADD checklist subscales ranged from 0.6–0.84 in these 2 studies.
The Gerber 2013 study reported that the total score on the French version of the PAS-ADD checklist had poor sensitivity and specificity compared with the ICD-10 (sensitivity=59%, specificity=59%; see Appendix O for ROC curves). The internal consistency of the organic, psychotic and affective/neurotic subscales ranged from 0.72–0.81, whilst the inter-rater reliability ranged from 0.66–0.73.
The mini PAS-ADD is measure of psychopathology in adults with learning disabilities. It is designed to be used by staff who have received training in the measure, but who do not need to have a background in applied psychology or psychiatry. The mini PAS-ADD has 86 items and produces scores relating to depression, anxiety, expansive mood, OCD, psychosis, unspecified disorder (typically dementia and other organic disorders) and autism spectrum disorder. Threshold scores are intended to be used in combination with clinical judgement to make decisions about the provision of diagnoses. The tool costs £179.95, and staff must undergo 2 days of training prior to using the instrument.
Devine 2009 investigated the mini PAS-ADD in adults with mild-moderate learning disabilities (n=96), whilst Janssen 2013 (n=467) and Prosser 1998 (n=68) investigated the mini PAS-ADD in adults across the full spectrum of the degrees of learning disabilities. Devine found that the mini PAS-ADD had 100% sensitivity and 77% specificity. Janssen, with a Dutch version of the measure, found high levels of variability in sensitivity and specificity within both a general and a clinical sample. The anxiety subscale appeared to be the most sensitive identifying 60% of cases correctly in the clinical sample. The unspecified disorder subscale had the highest reported specificity at 100%. The internal consistency of the subscales ranged from 0.46 to 0.81 in this study. Prosser 1998 found that the sensitivity and specificity of the mini PAS-ADD in individuals with moderate to profound learning disabilities was higher when scored by a psychiatrist than a member of the community support team (psychiatrist: sensitivity=87%, specificity=100%; Community support: sensitivity=57%, specificity=83%). See Appendix O for ROC curves.
The PIMRA is a 56-item questionnaire and is a general measure of mental health problems for use in adults with learning disabilities. It is designed to be completed by an informant familiar with the person with learning disabilities and produces 7 subscale scores on the basis of DSM-III (DSM 3rd edition) criteria; schizophrenia, affective disorder, psychosexual disorder, adjustment disorder, anxiety disorder, somatoform disorder and personality disorder, as well as a measure of inappropriate adjustment. The PIMRA costs £163.
The PIMRA has been investigated by 3 authors; Gustafsson 2005 (n=83) in adults with mild to severe learning disabilities, van Minnen 1994 (n=89) in adults with mild learning disabilities and Swiezy 1995 (n=65) in adults with mild to moderate learning disabilities. Gustafsson 2005 found that the 41-item total score of the PIMRA had 68% sensitivity and 84% specificity, whilst the psychosis, adjustment/anxiety and hallucinations subscales ranged in sensitivity from 53–86% and in specificity from 76–94%. The internal consistency of the total score scale was 0.81 and test-retest reliability value was 0.35. Similarly, van Minnen 1994 found that the internal consistency of the subscales ranged from 0.68 to 0.9, whilst the convergent validity of the affective disorders subscale was 0.74. Swiezy 1995 reported inter-rater agreement values of 78–81% for the schizophrenia, depression and dysthymia subscales, and internal consistency values for these subscales ranging from 0.16–0.43.
The P-AID is a measure designed for use in adults with learning disabilities. It consists of 18 different checklists.
Hove 2008 investigated the agoraphobia, social and specific phobia subscales of the P-AID in an adult population with all levels of learning disability (n=35), finding that the internal consistency of these scales was 0.83–0.87 and the inter-rater reliability was 0.65–0.75. In a population of adults with mild learning disabilities (n=35) the GAD and panic disorder subscales had internal consistency of 0.89 and inter-rater reliability of 0.49–0.55. In adults with a history of psychopathology (n=111) across all levels of learning disability the P-AID had sensitivity of 30% and specificity of 89%.
The SDQ is a free, commonly-used, brief behavioural screening questionnaire designed for use in children and young people aged 3–16 years. There are 3 versions that exist: a parent-report (ages 3–16), a teacher-report (ages 3–16) and a self-report (ages 11–16). There is also an impact supplement that provides additional information for clinicians about the severity and chronicity of a young person’s difficulties. The SDQ consists of 25 items across 5 different subscales and covers emotional symptoms, conduct problems, hyperactivity and inattention, peer relationship problems and prosocial behaviour, and additionally produces a total score.
Glenn 2013 investigated the usefulness of SDQ to identify psychopathology, compared with the ICD-10, in a population of adults with Down’s syndrome (n=125). They found that the mean value of the area under the curve (AUC) was 0.689 (standard error [SE] =0.056).
1 study (N=115) covering 1 tool (Child Depression Inventory) met the eligibility criteria for this review: (Meins, 1993).
1 study (N=65) on 1 tool (Glasgow Depression Scale for people with learning disabilities) met the eligibility criteria for stage 2 of this review as it used the Mini PAS-ADD as the reference standard: (Cuthill et al., 2003).
The CDI (now the CDI-2) is a 28-item questionnaire-measure of depression symptomatology in children and young people aged 7–17 years. There are 3 full-length versions that exist: a parent-report version, a teacher-report and a youth-report. There is also a 12-item short-form version. The CDI produces 2 scales; emotional and functional problems, and 4 subscales; negative mood, negative self-esteem, ineffectiveness and interpersonal problems. A starter kit for the CDI-2 costs $400.
Meins 1993 evaluated the effectiveness of the CDI at identifying depression in an adult population (n=115), with 3 items relating to school attendance removed. When a cut-off of 13 was used the CDI had 83% sensitivity and 93% specificity, whilst with a cut-off of 17 they found a sensitivity of 75% and specificity of 98%. The CDI had an internal consistency of 0.86 in this population and inter-rater agreement of 82%.
The GDS-LD is a free 20-item questionnaire measure designed to identify depression in people with mild-moderate learning disabilities. It takes around 10–15 minutes to complete. Items are rated on a 3-point Likert scale ranging from 0 ‘no’ to 2 ‘a lot’. A score of over 13 indicates probable depression. A carer supplement (GDC-CS), consisting of the 16 items considered to be directly observable, is also free.
Cuthill 2003 investigated the validity and reliability of depression diagnosis with the GDS-LD compared with the DSM-IV in adults with mild to moderate learning disabilities (n=65). They found that the GDS-LD identified depression with 90% sensitivity and 100% specificity. The carer supplement also had good test-retest reliability (r=0.98) and internal consistency (Cronbach’s alpha = 0.88).
There were 4 studies (N=515) which covered 5 tools met the eligibility criteria for this review: (Cosgrave et al., 1998; Deb & Braganza, 1999; Deb et al., 2007; Li et al., 2015).
No studies met the eligibility criteria for stage 2 of this review as there were no studies using tools found adequate in stage 1 as a reference standard.
The TSI is a clinician-administered 24-item questionnaire-measure designed to identify decline in cognitive function due to dementia in adults aged 51–91 years. The measure is not specifically designed for people with learning disabilities. It takes between 10 and 20 minutes to administer. There are 8 domains which the individual is required to respond to verbally. The TSI costs £283.
Cosgrave 1998 assessed the usefulness of the TSI at identifying cognitive decline in adults with Down’s syndrome and moderate-severe learning disabilities (n=60), compared with diagnosis using the ICD-10. They found that the internal consistency of the TSI was 0.89, and the test-retest reliability ranged from 0.74 in severe learning disabilities to 0.97 across all levels of learning disabilities. The convergent validity of the TSI ranged from 0.74 in severe learning disabilities to 0.94 in the full sample.
The DSQIID is a free carer-rated questionnaire screening measure for dementia in people with learning disabilities. It consists of 53 items, takes 10–15 minutes to complete and is appropriate for use in people with mild to severe learning disabilities. A score of 20 or above indicates possible dementia.
Deb 2007 (n=116) investigated the effectiveness of the DSQIID at identifying dementia in adults with Down’s syndrome across all levels of learning disabilities, compared with the ICD-10. The authors found that the DSQIID had 92% sensitivity and 97% specificity, internal consistency of 0.91 and inter-rater reliability of 0.9. Li 2015 (n=200) compared the DSQIID with the Diagnostic Criteria for Psychiatric Disorders for Use with Adults with Learning Disabilities/mental Retardation in adults over the age of 40 across all levels of learning disabilities, reporting sensitivity of 92.3% and specificity of 99.5%. They reported internal consistency of 0.945 and inter-rater reliability of 1. See Appendix O for ROC curves.
The DMR (now named the Dementia Questionnaire for People with Learning Disabilities) is a carer-rated screening questionnaire for cognitive deterioration resulting from dementia in people with mild to severe learning disabilities. It consists of 50 items scored on 3-point Likert scales ranging from 0 ‘no deficit’ to 2 ‘severe deficit’, and takes 15 to 20 minutes to administer. The Dementia Questionnaire for People with Learning Disabilities produces 8 subscales; short-term memory, long-term memory, orientation, speech, practical skills, mood, activity and interest, and behavioural disturbance. The first 4 subscales produce the sum of cognitive scores scale, and the last 4 produce a sum of social scores scale. The DMR costs £129.50.
Deb 1999 investigated the usefulness of the DMR, compared with the ICD-10, in a population of adults with Down’s syndrome (n=62) over the age of 35. They reported sensitivity and specificity of 92% (see Appendix O for ROC curves).
The DSDS (now Gedye DSDS) is a clinician-rated questionnaire measure of dementia symptoms designed for use in people with Down’s syndrome. It consists of 60 items, takes around 30 minutes to administer and is suitable for use in people with all levels of learning disabilities. The DSDS costs US$80.
Deb 1999 investigated the usefulness of the DSDS, compared with the ICD-10, in a population of adults with Down’s syndrome (n=62) over the age of 35. They reported sensitivity of 85% and specificity of 89% (see Appendix O for ROC curves).
There were 6 studies (N=388) which covered 4 tools met the eligibility criteria for this review: (Fitzgerald et al., 2013; Morrissey et al., 2007a; Morrissey et al., 2007b; O’Shea, 2015; Pouls & Jeandarme, 2014; Verbrugge et al., 2011).
No studies met the eligibility criteria for stage 2 of this review as there were no studies using tools found adequate in stage 1 as a reference standard.
The HCR-20 (version 3) is a structured violence risk assessment tool for use by clinicians with adults with learning disabilities. The HCR-20 does not have numerical cut-off values. It has 20 items and is used in combination with clinical judgement to assess an individual as posing low, moderate or high risk of future violence. The manual costs £75, a 50-pack of worksheets costs £75 per unit when 1–24 are ordered and £60 per unit when over 25 sets are ordered.
Fitzgerald 2013 investigated the usefulness of the HCR-20 at predicting risk in a population of adults with mild to moderate learning disabilities based within a medium-secure unit (n=25). The authors reported an AUC value of 0.73 (SE 0.11) for the risk of any physical aggression and 0.81 (SE 0.1) for severe physical aggression over 6 months. The study also reported a Spearman’s rho inter-rater reliability value of 0.6. Morrissey 2007a investigated the same tool (n=54) in individuals detained under the MCA within a national high-secure unit, with mild to moderate learning disabilities. This study reported an AUC value of 0.68 (interpersonal physical aggression) and 0.77 (verbal or property aggression), and a Spearman’s rho test-retest reliability value of 0.45 over 12 months. Morrissey 2007b investigated the HCR-20 at the same unit, again with individuals with mild to moderate learning disabilities who were detained under the MCA, over 2 years (n=73). The authors reported an AUC value of 0.69 (standard deviation [SD] 0.53–0.81) for prediction of positive progress on the basis of HCR-20 values, and of 0.49 (SD 0.28–0.70) for negative progress. O’Shea (2015) reported on the use of the HCR-20 to predict risk of violence in in adults, predominantly with serious mental illness, within a forensic inpatient setting. The authors reported an AUC value of 0.67 (95% CI 0.58–0.75) for the prediction of any aggression, and of 0.61 (95% CI 0.52–0.70) for physical aggression in this population. In a mixed community-based sample of young people and adults with mild to moderate learning disabilities Verbrugge 2011 (n=59) used an intellectual disability supplement for the HCR-20 in a population of people who were no longer in custody and reported an AUC value for general recidivism in individuals classified as moderate to high risk of 0.97 (SE 0.03), and for violent recidivism of 0.8 (SE 0.09). In this study inter-rater reliability was 0.65.
The PCL-R is a 20-item clinician-rated questionnaire-measure of psychopathy designed for use in adults. The full kit costs $460, a pack of 25 ‘QuikScore forms’ is $100 and a pack of 25 ‘interview guides’ is $160.
Both Morrissey 2007a (n=60) and Morrissey 2007b (n=66) investigated the PCL-R in a population of adults with mild to moderate learning disabilities detained in National high-security facilities under the MCA. Morrissey 2007a reported an AUC value for the prediction of aggressive incidents over 12 months from PCL-R score of 0.54 (SD 0.39–0.68), internal consistency (Cronbach’s alpha) of 0.73, Spearman’s rho test-retest reliability value of 0.11 and inter-rater reliability value of 0.8. Morrissey 2007b reported an AUC value for the prediction of positive progress after 2 years on the basis of PCL-R score of 0.69 (SD 0.53–0.81) and of negative progress of 0.49 (SD 0.28–0.70). Pouls 2014 investigated the PCL-R at predicting institutional violence and violation of conditions in detained adult patients with mild to severe learning disabilities in forensic units (n=60). They reported an AUC value of 0.68 (SD 0.52–0.84), internal consistency of 0.75 and inter-rater reliability of 0.73.
The PCL-SV is a 12-item screening measure for psychopathy. The PCL-SV kit costs $240, a pack of 25 Interview guides is $85 and a pack of 25 ‘QuikScore forms’ is $90.
Pouls 2014 investigated the PCL-SV in detained adult patients with mild to severe learning disabilities in forensic units (n=60). They reported an AUC value of 0.69 (SD 0.55–0.84), internal consistency below 0.6 (Cronbach’s alpha) and inter-rater reliability of 0.44 (kappa).
The Violence Risk Appraisal Guide is a free 12-item questionnaire measure designed to assess risk of violence in.
Fitzgerald 2013 (n=23) investigated the usefulness of the Violence Risk Appraisal Guide to predict aggression in adults with mild to moderate learning disabilities placed within medium-secure units. They reported an AUC value for any physical aggression of 0.87 (SE 0.08) and for severe physical aggression of 0.78 (SE 0.1). They reported a Spearman’s rho inter-rater reliability value of 0.53 for risk of any physical aggression. Verbrugge 2011 in a community sample of young people and adults with mild to moderate learning disabilities (n=59) reported an AUC value for violent recidivism of 0.79 (SE 0.07) and of general recidivism of 0.92 (SE 0.06). The inter-rater reliability was reported as 0.66.
Only 1 study (N=664) was identified in children and young people and met the eligibility criteria for stage 1 of the review (Einfeld & Tonge, 1995); no studies met the eligibility criteria for stage 2 as there were no studies using tools found adequate in stage 1 as a reference standard.
The DBC-P is a 96-item parent or carer-rated measure of emotional and behavioural problems in children and adolescents aged 4–18 years. The measure is in the process of being licensed for use. At the time of writing no information was available on the anticipated cost of the tool to services. A score of 46 or above has been identified as the optimal cut-off for this instrument. There is also a version of the DBC for adults, however we did not find evidence of sufficient quality for the utility of that version. Einfeld 1995 found that the DBC-P had an AUC value of 0.92 in young people with mild–severe learning disabilities, test-retest reliability of 0.83 (intraclass correlation, 95% CI 0.69–0.90) and inter-rater reliability between parents of 0.80 (intraclass correlation, 95% CI 0.59–0.90).
No studies assessing the cost effectiveness of key components of, and the most appropriate structure for, an assessment of mental health problems in people with learning disabilities were identified by the systematic search of the economic literature undertaken for this guideline. Details on the methods used for the systematic search of the economic literature are described in Chapter 3.
The GC endorsed statements stating that a brief assessment should:
The GC endorsed statements stating that a brief assessment should:
The GC endorsed statements stating that:
The GC endorsed statements stating that involved staff should be:
The GC endorsed statements stating that outcomes of a brief assessment should include:
The GC endorsed statements stating that, in addition to the points detailed above for brief assessment, a comprehensive assessment should:
They agreed that, in contrast to the points detailed above for brief assessment that a comprehensive assessment should not:
No evidence on the cost effectiveness of key components of, and the most appropriate structure for, an assessment of mental health problems in people with learning disabilities is available.
| Recommendations | Conducting a mental health assessment
|
|---|---|
| Relative values of different outcomes | The GC discussed the importance and relevance of the purpose and outcome of the assessment of need and of risk, as well as special considerations for this group when crisis assessments were necessary. In doing so they took into account the time needed to undertake the assessment, the skills and knowledge required to complete the assessment, the resources required and which individual might need to be involved. In addition they considered what assessment tools might help inform the assessment process, whether these tools may have value in monitoring the outcome of any interventions and the reliability, validity and sensitivity to change of the tools (validity, sensitivity, specificity were considered critical outcomes). The GC also used their expert knowledge to inform the overall structure of the assessment process with a particular emphasis on ensuring that the outcomes of the assessment were integrated with the existing care planning and reviews systems that operate in services for people with learning disabilities. |
| Trade-off between clinical benefits and harms | No evidence was found to address the most appropriate content and structure for an assessment of mental health problem, the GC agreed to develop recommendations using formal consensus methods. However, evidence was identified on the use of various tools for assessing mental health problems in this population. Many studies combined different levels of learning disabilities so it was difficult to determine the differential psychometric properties of the tools by level of learning disability. Only 1 study (Einfeld & Tonge, 1995) considering a single tool (DBC-P) specifically looked at the use of tools in children; all others considered adults. General measures of mental health problems – children The DBC-P was found to have good diagnostic accuracy at identifying emotional and behavioural problems in children and young people with learning disabilities so the GC agreed to recommend the tool. While the SDQ had poor quality in adults (see below), the SDQ has been validated in a large number studies of mental health problems including children with a range of developmental and physical disorders (Goodman, 2001) and has excellent psychometric properties in this context. In view of this, the GC drawing on their expert judgement decided to recommend the use of the SDQ in the assessment process for children and young people. It also has the advantage of having versions that can be self-completed or by a parent/carer and teacher. General measures of mental health problems – adults It was difficult to compare tools given the variety of outcomes reported in the studies. However, the evidence suggested that the DASH-II had good diagnostic accuracy on the mania subscale only. The MASS had better diagnostic accuracy for any anxiety disorder and on the depression subscales, but lower sensitivity on the GAD and mania subscales. The PAS- ADD interview had relatively good diagnostic accuracy on anxiety disorders subscale but the neurotic symptoms, depression and schizophrenia subscales were not as good. Also, the GC noted that the evidence on the PAS-ADD interview was on an earlier version of the tool that is currently available. The GC also noted that the PAS-ADD interview is a very long interview (lasting 3 hours) and that it focuses on descriptive phenomenology. However, the Mini PAS-ADD was shorter and could be used by a broader range of professional staff than the PAS-ADD interview. The evidence on the Mini PAS-ADD was generally adequate on both sensitivity and specificity, when using a psychiatric diagnosis as a reference standard. The PAS-ADD checklist had little evidence supporting its use and was viewed by the GC as essentially a screening instrument and, therefore, not suitable for assessment purposes. The GC decided, supported by stakeholder comments, that due to the limited evidence and practical issues of, for example administration time, not to recommend the PAS-ADD instruments The PIMRA was also a screening instrument and although it has reasonable psychometric properties the GC were of the view that it would not contribute much to the assessment process. The P-AID is a series (18) of symptom/behavioural checklist; it has relatively poor psychometric properties and was not viewed by the GC as likely to make a significant contribution to the assessment process. One study examined the use of SDQ in adults with learning disabilities and it performed poorly which, given that it is measure developed for children and young people, is not surprising. Depression The GC were interested in measures of symptomatology which could inform an assessment and which may also serve as a tool for routine outcome monitoring. Given the time constraints inherent in routine outcome monitoring this favours the use of measure which focus either on single domains of symptomatology (for example depression) or provide a brief global rating of outcomes. The review of the literature identified only 2 such measures both of which were measures of depressive symptoms. After reviewing the evidence on the 2 measures of depressive symptoms the GC decided to recommend the GDS-LD and the GDS-CS for use in children and young people, and adults. It had good psychometric properties and the GC took the view it could be of value in both the assessment of depressive symptoms and their routine monitoring. While the study included participants with mild to moderate learning disabilities, the GC considered that it would be reasonable to expect the carer supplement of the GDS (the GDS-CS) would be appropriate for people with more severe learning disabilities as it was being completed by a carer. Dementia The identification, diagnosis and assessment of dementia present particular challenges in people with learning difficulties. Dementia is more common in people with learning disabilities than in the general population, and particularly so for people with Down’s syndrome. Existing cognitive deficits mean that standard instruments cannot be used. The GC therefore looked carefully at the existing measures to see how they could help in addressing these issues. The DSQIID and the DMR are both carer focused screening measures of dementia psychopathology with the evidence showing both had good psychometric properties which could be used to inform an assessment, particularly carer input into the process. Both appear to use across the range of disabilities and can be used for people with Down’s syndrome. The DSDS is a clinician-rated questionnaire measure of dementia symptoms designed for use in people with Down’s syndrome but which the GC consider may have use beyond just people with Down’s syndrome. The evidence showed it also has reasonable psychometric properties. The TSI is a clinician- administered measure of cognitive function. The measure is not specifically designed for people with learning disabilities but the GC considered that it would be useful as a measure of change in cognitive function over time in people with learning disabilities. Again the evidence showed that it has reasonable psychometric properties. Assessment of risk Several tools were identified which assessed risk but all of these were instruments designed for use in forensic environments and were particularly concerned with the prediction of violence. They did not look at risk to self or self-neglect or exploitation. Given the focus of the guideline on mental health the GC decided not to recommend these measures but instead developed the recommendations for risk assessment by informal consensus drawing on their expert knowledge and experience. Adaptive function/behaviour While the evidence was searched for a number of tools measuring adaptive function or behaviour, there was no evidence found on these tools that met the inclusion criteria specified in the protocol. However, the GC agreed that it would be inappropriate not to recommend a measure of adaptive function to monitor changes over time, particularly for dementia. As such, they developed a recommendation in this area through informal consensus that adaptive function should be measured to monitor changes over time in people diagnosed with dementia but they were unable to recommend a specific tool for this use. However, given the lack of evidence to make a specific recommendation about a tool for adaptive function, they considered that the judgement about which tool to use should be made by the specialist who are monitoring changes over time in the person with dementia. The GC noted that it was important to have tools with higher sensitivity in order to prevent the harms associated with not identifying and treating mental health problems. Many of the tools reviewed did not have high sensitivity in order to adequately identify mental health problems. Additionally the GC considered the practicality of using each tool when considering whether to make a recommendation for its’ use in clinical practice, for example time taken to administer, whether training was required and ease of use, The low sensitivity and lack of practicality of many of the identified tools led to few tools being recommended. |
| Trade-off between net health benefits and resource use | Formal assessment of a mental health problem in people with learning disabilities is likely to have important resource implications. However, the GC expressed the view that comprehensive, effective assessment followed by the development of an appropriate mental health care plan will allow more timely, appropriate and cost-effective management of the mental health problem, that is targeted to the specific needs of the person with learning disabilities and thus can result in cost-savings that offset, fully or partially, the costs associated with assessment. In contrast, inadequate assessment may lead to sub-optimal, less clinically and cost-effective care pathways and inappropriate treatments, ultimately leading to sub-optimal outcomes for the person and higher health and social care costs. Regarding the specific resource implications associated with formal assessment of a mental health problem in people with learning disabilities, the GC advised that the assessment needs to be co-ordinated by a clinician with expertise in mental health problems in this population, to ensure appropriate assessment according to the service user’s individual needs and circumstances. The GC acknowledged that the assessment may require a considerable amount of staff time and incur a range of costs that are determined by:
Assessment of mental health problems may cause distress to people with learning disabilities. The GC expressed the view that, where possible, the assessment should be conducted in a place (for example room) within the care setting where the person with learning disabilities may have been familiar with, so that the level of distress is minimised. Conducting assessments in places people with learning disabilities are familiar with, where possible, would not incur any extra resource implications. |
| Quality of evidence | The evidence was limited, studies were small in size and some had restricted samples (for example, people with Down’s syndrome). In most cases for those measures recommended the psychometric properties were acceptable. However, there were notable gaps in the evidence including: very limited evidence for children (here, the GC chose to recommend the SDQ as it is common practice in other populations with neurodevelopmental problems, such as autism). In other cases, the evidence on risk assessment was confined to violence assessment in forensic settings and there was an absence of evidence on the structure of the assessment process and so in these cases the GC used both formal and informal consensus methods to develop recommendations. |
| Other considerations | The group noted that no studies were identified which met the inclusion criteria for other areas (including anger, personality disorders, PTSD, trauma, and the assessment of communication). The GC decided not to make any recommendations other than referring to relevant NICE guidelines for those disorders where effective implementation of the assessment and intervention recommendations requires that practitioners follow advice on assessment and outcome measures provided in the guideline. Given the lack of high quality evidence demonstrating valid and reliable tools, the GC decided to recommend research into the development and validation of tools for assessment of mental health problems in this population. Brief and Comprehensive Assessment While the assessment process was originally split into brief and comprehensive assessment, after reviewing the outcome of the nominal group technique the GC decided this was not an appropriate delineation of the assessment process and was amended (see section 4.5.1 for further details). The GC noted that conducting assessments in a setting that is familiar to the person may reduce feelings of distress. They noted that this may not always be possible but that it is generally standard practice so is not likely to incur extra resources so agreed it was appropriate to include a recommendation on this. Comments were received from stakeholders and discussed by the GC in relation to the recommendations for practice. These related to the impact of sensory impairment and the person’s cultural, ethnic and religious background on the assessment process, the appropriateness of DSM-5 and ICD-10 as diagnostic classification systems for those particularly with milder learning disabilities and the importance of highlighting early warning signs of relapse. These issues were all incorporated into the recommendations. The GC decided that mental health assessment during a crisis episode was a particularly important area of practice which required NICE guidance, a view supported by comments from stakeholders during consultation. They discussed issues of particular relevance to this group, such as good understanding of the person, their learning disabilities and communication needs, the difficulties with obtaining consent and assessing capacity within this group, and the need for reasonable adjustments should acute inpatient care be required. No evidence was found in this area, and it had not been covered by the nominal group technique process. Therefore the group decided to develop a recommendation by informal consensus. |
This recommendation updates and replaces recommendations 1.5.1.2 and 1.6.2.7 in the NICE guideline on supporting people with dementia and their carers in health and social care.
This recommendation and recommendations 5, 51, 52 and 73 update and replace recommendation 1.3.3.2 in the NICE guideline on supporting people with dementia and their carers in health and social care.
Development of and validation (including diagnosis) of assessment tools for mental health problems in people with learning disabilities.
All rights reserved. NICE copyright material can be downloaded for private research and study, and may be reproduced for educational and not-for-profit purposes. No reproduction by or for commercial organisations, or for commercial purposes, is allowed without the written permission of NICE.
Your browsing activity is empty.
Activity recording is turned off.
See more...
