The microsegmentation of the autism spectrum: research project

1 Executive Summary

Introduction

1.1 The Microsegmentation Project was funded by the Scottish Government through Scottish Autism to take forward key recommendations of the Scottish Strategy for Autism (Scottish Government, 2011).

1.2 The project arose with particular reference to Recommendation 5 in the Scottish Autism Strategy: ‘It is recommended that Knapp’s work on the economic costs of autism is analysed and applied to the Scottish context to inform strategy and planning on what interventions lead to positive impacts both for individuals and for the economy as a whole.’ In order to provide a basis for this, it was essential that more accurate and more detailed economic costs should be formulated than were currently available, and that these should relate specifically to the ASD population of Scotland.

1.3 A primary purpose of doing so was to provide a reliable foundation for identifying those costs of autism which may be ‘escapable’, that is, those which would not be incurred with appropriate interventions for individuals on the spectrum. This was taken forward by carrying out a ‘microsegmentation’ of the autism spectrum, its co-occurring conditions and its associated problems, so that a conceptual map of the spectrum might be constructed. Each segment was associated with a range of possible life outcomes, illustrating the types of issues and challenges likely to be faced by the individuals concerned.

1.4 Following an extensive scoping exercise to identify key issues from the current literature and make preparations for collection of data, three main studies were conducted. Study 1 comprised a systematic review and meta-analysis of English-language studies of prevalence of the autism spectrum from across the world. This provided more methodologically robust prevalence data to inform more accurate economic analysis. In terms of demographic mapping, all relevant and available Scottish data pertaining to the prevalence of ASD were examined and compared with all data gathered for the study from other sources.

1.5 Study 2 comprised a systematic review and meta-analysis of intellectual ability levels across the autism spectrum population, as a key factor moderating outcomes for individuals. This provided more accurate information on this variable which is central to any study relating to economic impact, and in doing so generated new figures for the proportion of the ASD population who have intellectual disability.

1.6 Study 3 comprised a fieldwork exercise conducted by way of a detailed and extensive Scottish Autism Survey, which generated a unique dataset of information pertaining directly to the ASD population of Scotland with a final analysis based upon responses relating to 950 individuals. This served to illuminate life trajectories across the lifespan in relation to the impact of presentation of autism, its co-occurring conditions and its associated features, together with the implications for service provision. This was then mapped on to the most accurate available demographic data that can be established for the population of Scotland in order to provide a rational basis for planning the services and supports that will be required to meet the needs arising, and for assessing economic impact.

Preliminary Scoping Exercise

1.7 Chapter 3 describes the parameters of the scoping exercise in terms of five preliminary questions: 1 What does research evidence tell us about outcomes and life trajectories in ASD? 2 What are the main co-occurring conditions of ASD, other associated features of the ASD profile and any other factors relevant to outcomes or acting as moderators of outcome? 3 How do the various outcomes and life trajectories in ASD translate into economic implications? 4 How do these economic implications map on to the population of Scotland? 5 What is the relationship between outcome and type of intervention received?

1.8 Key points arising from these five questions were:

• While more recent studies have shown more favourable outcomes for individuals with autism spectrum disorders than earlier studies, largely because of the diagnosis of larger numbers of less severe cases, autism may still be viewed as a lifelong neurodevelopmental disorder of a pervasive nature, with disabling aspects affecting key areas of independence and quality of life.
• Autism is associated with many co-occurring conditions and other features including intellectual disability, epilepsy, attention deficit hyperactivity disorder, schizophrenia, obsessive compulsive disorder, Tourette’s Syndrome, anxiety and depressive disorders, sleep problems, challenging behaviour, eliminatory disorders and gender identity issues.
• While any of these co-occurring conditions and other features may have impact as moderators, the single most important moderator in terms of outcomes and their translation into economic implications is the presence or absence of intellectual disability.
• Existing data on prevalence and intellectual disability were not sufficiently accurate as a basis for calculating economic consequences for the population of Scotland, necessitating a fresh analysis of both for the purposes of this study.
• Regarding the relationship between intervention and outcome, a key conclusion arising from the extensive literature on autism interventions was that it currently provides an insufficient basis for any economic evaluation. Proposals relating to interventions must therefore be based on considering where key aspects both of the needs of this population and of the economic consequences lie, and asking what avenues of intervention may offer the greatest impact in terms of addressing the most important needs.

Prevalence

1.9 Chapter 4 describes our work on prevalence. Previous attempts to estimate the prevalence of autism spectrum disorders in Scotland have been based on inadequate methodology and have therefore not provided a basis for determining accurate figures or for economic analysis or service planning. They have yielded figures which are very far below any prevalence levels which might be expected on the basis of sound methodological approaches.

1.10 Our systematic review and meta-analysis of English-language studies of prevalence of the autism spectrum from across the world generated a pooled prevalence estimate of 1.035% (103.5 per 10,000) on the basis of the studies using the most robust research methodology. We have recommended that this figure should be used as the most accurate prevalence estimate available. We noted that there is no credible way currently of establishing prevalence for separate diagnostic subgroups within the autism spectrum.

Intellectual ability and disability

1.11 Chapter 5 describes our work on intellectual ability and disability. Since presence of intellectual disability is the single most important moderator of outcomes and of costs, accurate figures are crucial to any economic study. We concluded that the currently available estimates for the proportion of individuals on the autism spectrum with an intellectual disability lacked clinical validity, with the most widely used figures being too high. This has major implications for economic estimates and service planning.

1.12 Our systematic review and meta-analysis of English-language studies of autism and intellectual ability from across the world generated a pooled estimate of 32.7% with intellectual disability on the basis of the very small number of studies using a sufficiently robust methodology. This is very different from the figures used previously. We have recommended that this figure should be used as the most accurate estimate available.

Prevalence and intellectual ability: the Scottish context

1.13 Chapter 6 describes the results of mapping our data on prevalence and intellectual disability on to the population of Scotland. On that basis it is now possible to provide accurate data for the number of individuals with ASD, together with numbers with and without intellectual disability, in every age range for the whole of Scotland and for every Council or Health Board area.

1.14 Table 1.1 shows these estimates in relation to the Scottish population, with population figures statistically adjusted to take account of longevity in terms of the available ASD research in this field.

Table 1.1 Prevalence of autism in Scotland by age and intellectual disability

Scotland	ASD population			Total population b
	with ID	without ID	Total
Children (0-1)	380	781	1,161	112,100
Children pre-school (2-4)	593	1,220	1,813	175,138
Children primary school (5-11)	1,394	2,867	4,261	411,638
Children secondary school (12-15)	735	1,512	2,247	217,041
Adults (16-67 a)	12,345	25,406	37,751	3,647,409
Total	15,445	31,786	47,231	4,563,326

^a The age range for which data is reported here reflects findings from longitudinal ASD studies. For further details see para. 6.2, and for data relating to the total population see Table 6.2. ^b Total population statistics taken from ONS (2017).

The Scottish Autism Survey

1.15 Chapter 7 describes our work in designing and conducting a large-scale online survey in Scotland of individuals with autism and their parents and carers, and provides a full breakdown of responses across every aspect of data collected. This was essential not only to obtaining more accurate data relating to the economic costs of autism in Scotland but also to support our aim of constructing a meaningful segmentation of the autism spectrum. The survey was lengthy and detailed, and covered topics including age, extending from early childhood to later adulthood, diagnosis, co-occurring conditions, educational placement, service provision, intellectual and educational status, residential arrangements and employment status.

1.16 The number of individuals who initiated a response to the survey was 1,604, with 950 of these providing complete or sufficiently complete data for the purpose of analysis. Responses were obtained from all 32 of the Council areas in Scotland. The results of the survey, together with the figures available from a range of national and international databases, have allowed the most comprehensive and accurate costs to be estimated across every relevant variable.

1.17 Table 1.2 shows the respondent characteristics of those who provided sufficient data to be included in the analysis. A small number of responses (5%) were submitted by individuals who were not on the spectrum or parents or carers of those who were.

Table 1.2 The Scottish Autism Survey: respondent characteristics

Respondent Type	n (%)
Parents and Family Carers	754 (79)
Non-related Carers	33 (4)
Individuals with ASD	114 (12)
Professionals	36 (4)
Others a	13 (1)
Total	950 (100)

^a This category included close friends and volunteers who worked with people with ASD

1.18 The age range of the final sample was from early childhood to 86 years. A total of 694 individuals (73%) were under the age of 21 (335 age 0-10, 359 age 11-20), while 256 (27%) were age 21 or over. Older adults with ASD were poorly represented, with only six individuals age 65 and over. The sex ratio was 735 (77%) male, 214 (23%) female. In terms of diagnosis, 217 (23%) reported a diagnosis of autism (or autistic disorder), 426 (45%) Asperger’s Syndrome (or Asperger’s Disorder) or ‘high-functioning autism, and 307 (32%) other or unspecified ASD diagnoses (atypical autism, autism spectrum disorder). One third of the sample reported the presence of at least one co-occurring condition other than intellectual disability of which the most prevalent were anxiety and depression, ADHD, epilepsy and obsessive compulsive disorder.

1.19 Statistical analysis was in three stages: first, a detailed examination of demographic, diagnostic and service use data to characterise the sample of respondents and to inform an understanding of the lives of those with ASD living in Scotland; second, multivariate analysis to identify and model the relationships between the factors from the survey and outcomes in education, employment, relationships, independent living and mental health; and third, a qualitative analysis of additional free text comments provided by the respondents.

1.20 The findings revealed the impact of ASD diagnoses, sex, intellectual disability and other co-occurring conditions, education placement and support. Statistical modelling identified age, sex, intellectual disability and diagnoses of depression and of ADHD as significant predictors of educational placements and levels of support provided to the individuals with ASD. Type of ASD diagnosis was a significant predictor of educational qualifications. Age, ability to travel independently and relationship status were significant predictors of employment status. Age, type of ASD diagnosis, diagnosis of depression and employment status were significant predictors of relationship status. Age, diagnosis of mood disorder, ability to travel independently and relationship status were significant predictors of residential status (independent living). The individual’s sex, ASD diagnosis, co-occurring conditions, relationship status and relationship status were significant predictors of service use.

1.21 The impact of caring for an individual with ASD was also investigated, with age and the ability of the individual with ASD to travel independently significant predictors of the extent to which carers themselves can be in employment, training or education.

1.22 Thematic analyses of the free-text comments from the nine individuals with ASD who responded revealed concerns about support and service provision, stress and anxiety linked to day-to-day life, employment or education, and issues relating to diagnosis.

1.23 Similar themes emerged from the analysis of the comments from 68 parents and carers, who also noted the stress and anxiety experienced not only by the individuals with ASD themselves but also by parents and carers, with financial concerns a factor; impact on families; and social issues such as difficulties on the part of the individuals with ASD in socialising and maintaining employment, and forensic history.

The economic impact of autism in Scotland

1.24 Chapter 8 describes the estimation of the economic impact of autism in Scotland using the survey and the literature review. First, the annual cost of supporting an individual with autism was estimated and described for children and adults, according to ASD diagnoses. Costs were higher for individuals with a diagnosis of autism than Asperger’s Syndrome (with the cost of other ASD profiles being intermediate between them), and slightly higher for children than for adults. Second, the incremental lifetime cost was estimated for individuals with and without intellectual disability, at £1.6 million and £0.89 million respectively (2013/14 price levels). Third, the incremental annual national cost of autism was estimated at £2.2 billion.

1.25 We also used data from the survey to examine whether the characteristics of individuals were associated with support costs, looking at children and adults separately. Among children with autism or Asperger’s Syndrome, those with co-occurring conditions had higher costs. For adults with autism, those living away from their parents had higher social care and total costs, while for adults with Asperger’s Syndrome, those in a relationship or with educational qualifications had lower social care costs.

Segmenting the autism spectrum

1.26 Chapter 9 describes our proposals regarding the question of microsegmentation itself. The need for segmentation arises from two considerations. First, planning for research, services or interventions in autism cannot be done on the basis of treating the whole spectrum as one entity. Second, and conversely, it cannot be done on the basis of treating everyone on the spectrum as being unique. The concept of recognising every person’s unique individuality does not over-ride the need for, and recognition of, meaningful homogeneity in clinical presentation. Identifying the key homogeneous features is a prerequisite for planning research samples, for setting up specialist provision, for providing targeted interventions and for predicting the parameters of future life trajectories.

1.27 Attempts to segment the autism spectrum have been made on the basis of diagnostic subgroups, of the nature of ASD profiles or of the presence of additional co-occurring conditions. While all of these have made some relevant contribution to segmentation, existing research has not provided an adequate foundation for segmenting the spectrum in a way that would provide a meaningful conceptual map of autism. Diagnostic subgroups have not been demonstrated as having clinical validity. ASD profiles, other than in the matter of presence or absence of intellectual disability, have not reliably predicted outcome, service needs or economic costs except in very broad terms. Co-occurring conditions, while having a significant impact, are too variable in their effects to act as stable moderators.

1.28 Service providers encounter some recurrent characteristics in the groups of people with ASD for whom they make provision, ranging from those who require lifelong 24 hour care and support, through those who have higher capabilities and a measure of independence and who do not require structured support on a daily basis, to those who are on the spectrum but have minimal support requirements. However, service needs only relate to assessment profiles in very general terms which allow of many exceptions. For example, it is not uncommon for individuals with high levels of intellectual functioning to be vulnerable to high levels of challenging, violent or offending behaviour and to require a very high tariff of support.

1.29 In considering both the general literature and the Scottish Autism Survey dataset, and examining the clinical significance of assigned diagnostic subgroup in its relation to intellectual and linguistic status and symptom presentation, we propose a model of segmentation in which intellectual ability and original symptom severity are stable moderators of outcome and co-occurring conditions are variable ‘additive risk factors’.

1.30 This allows the construction of a microsegmentation matrix, containing four segments which reflect the gradation from higher intellectual ability and lower symptom severity, commonly represented currently in the Asperger profile, through to those, currently with a diagnosis of autism or other ASD, with moderate or severe intellectual disability and higher symptom severity. These four segments reflect the stable moderators of intellectual status and symptom severity, and each is then subdivided to reflect the variable additive risk factors associated with co-occurring conditions. Thus, the matrix comprises eight segments.

1.31 Figure 1.1 shows the microsegmentation matrix in terms of these eight segments, together with indications of the gradation of outcomes from more to less independent travel, employment, independent living and long-term relationships, and economic costs ranging from low to high. As noted in the matrix, there are variable costs within each segment according to the impact of additive risks. Thus, an individual in segment 1, where outcomes would generally be more favourable and economic costs lower, may in fact prove to have a disproportionately high level of need and cost depending on the extent of impact of additive risk factors.

Figure 1.1 The autism spectrum: microsegmentation matrix

Microsegmentation and future research and provision for ASD in Scotland

1.32 Chapter 10 describes ways in which the matrix may be used to offer an evidence-based template for a structured approach to future research and provision for ASD. It may be combined with any other framework to provide microsegmentation best suited to addressing the issues which will most affect the quality of life of individuals on the autism spectrum and their parents and carers, in the key areas of planning priorities for research, resource planning, commissioning, service provision, tailoring interventions to address needs and leading to positive impacts both for individuals and for the economy as a whole.

1.33 Chapter 10 also considers the question of the ‘escapable costs of autism’ in the light of the lack of a robust evidence base linking interventions to outcomes or demonstrating links between interventions and economic impacts. Following the practical approaches adopted in the ‘Menu of Interventions’ devised in relation to the Scottish Strategy for Autism and published by the Scottish Government, it focusses on those factors arising from this study which are currently associated with high costs both economically and in terms of reduced quality of life, and the potential impact of supporting individuals with these difficulties towards more optimal life outcomes.

1.34 In particular, reference is made to the potential economic benefits that may arise from the following: ensuring access to multi-disciplinary teams for the timely identification and assessment of autistic adults; the availability of early interventions for autistic children, both with and without intellectual disability; supported employment schemes, particularly for autistic adults without intellectual disability; the availability of parent training and support programmes for families of autistic children; the provision of cognitive behaviour therapy appropriate to the needs of both autistic children and adults; the availability of interventions that emphasise personalised approaches; and regular health checks for the entire autistic population.

1.35 It is not possible in terms of the current evidence base to quantify the savings that might be achieved in relation to any particular intervention with potential economic benefits. By way of illustration, a number of examples are presented in Chapter 10 to indicate what savings would be achieved annually in Scotland in terms of several different scenarios involving cost-effective interventions for children and for adults, with and without intellectual disability, and for the total autistic population.

1.36 In terms of the total autistic population, for each percentage point by which evidence-based interventions reduced total costs there would be potential savings of around £22,000,000 annually in Scotland. A reduction in costs by five percentage points would bring annual savings of around £111,000,000, while if a 10% reduction could be achieved there would be annual savings of around £223,000,000.

1.37 In very many major reports a large number of recommendations have been made regarding the needs of people on the autism spectrum, the services required to address these needs and the principles of good practice for professionals working in this field. It is not the remit of this report to reiterate these recommendations but rather, in line with the purpose of the report as set out above, to provide a reliable foundation for identifying the escapable costs of autism. The recommendations that follow are therefore those which relate to strategies and interventions designed to improve the quality of life of the whole autistic population of Scotland and their parents and carers and which in doing so also have evidence of potential economic benefits.

Recommendations

Prevalence and intellectual disability

Recommendation 1

It is recommended that a prevalence figure of 1.035% (103.5/10,000), of whom 32.7% would be likely to have a learning disability, should be used as a basis for planning autism provision and services.

The microsegmentation matrix

Recommendation 2

It is recommended that the microsegmentation matrix should be adopted as a template for a structured approach to future research and provision for ASD in Scotland.

Quality of life and potential economic benefits

Recommendation 3

It is recommended that every NHS Scotland Health Board should have, or should have access to, a multi-disciplinary team to identify and assess autistic adults.

Recommendation 4

It is recommended that, while economic gains have not at this stage been clearly evidenced, there should be an increased focus on the potential value of parent-mediated and other evidence-based early interventions for autistic children, both with and without intellectual disability.

Recommendation 5

It is recommended that there should be a key focus on supported employment schemes for autistic adults, particularly those without intellectual disability, together with a focus on supporting such adults to travel independently where required.

Recommendation 6

It is recommended that there should be an extension of parent training and support programmes for the families of autistic children and adults, both with and without intellectual disability.

Recommendation 7

It is recommended that cognitive behavioural therapy ( CBT) should be made universally available to autistic children and adults without intellectual disability who have anxiety and other mental health disorders.

Recommendation 8

It is recommended that autism-specific training should be made available to cognitive behavioural psychotherapists with a view to modifying the standard CBT protocol to suit the needs of children and adults on the autism spectrum.

Recommendation 9

It is recommended that there should be an increased focus on personalised approaches which tailor interventions to the individual needs, strengths and personal preferences of autistic children and adults.

Recommendation 10

It is recommended that regular health checks should be made available to the whole autistic population.