Autism and the draft mental health bill in England and Wales: Unintended consequences?

The Draft Mental Health Bill for England and Wales proposes a range of changes to the Mental Health Act 1983. Among these is a proposal to amend the definition of ‘Mental Disorder’, which would mean that people with a primary diagnosis of autism or learning disability would be excluded from most of the civil sections of the Act. If enacted, this will be the first occasion in English and Welsh law where a legal definition of autism is created. While we argue that there are inherent difficulties in creating legal definitions of clinical problems, this ‘legal Autism’ appears to be substantially disconnected from a clinical understanding of autism, and its breadth could potentially encompass conditions beyond those currently identified as autism. Moreover, numerous potential unwanted practical consequences may arise from these changes, partly because of uncertainty about the legal position of presentations that share features with autism, uncertainty about the process of assessment of ‘legal Autism’, and potential perverse incentives placed on local authorities. We argue that particular groups of people likely to be impacted by these changes are those who not only present with profound clinical impairments, but also who pose significant risks to others. We urge serious consideration is given to these issues in the legal debate of the Bill.


What does the Draft Bill propose in relation to Autism?
While there is clearly scope for a fuller review of the extent to which the Draft Bill meets the aims and aspirations of both the independent review and the White Paper, one immediate apparent change is the proposal to alter the definition of mental disorder, so as to add autism to learning disability as a category of mental disorder, which would be excluded from many provisions of Part II of the Act (clause 1(7) of the Bill). More specifically, this proposal would mean that people with autism as well as those with a learning disability could not be detained under civil legislation, except for time-limited assessment under s.2 of the Act. Potentially indefinite detention under s.3 would no longer be possible. Somewhat perplexingly, no such exclusion to the definition of mental disorder would apply for people detained under the criminal sections of the Act (Part III). This proposal would create a significant change to the status quo, where the definition of mental disorder outlined in the Act (s.1(2)) is both broad and inclusive, that is, 'any disorder or disability of the mind'. Presently, only two exceptions to this definition are applied: one is for dependence on alcohol or drugs (s.1(3)) and one is outlined in s.1(2A), which provides that for many of the key sections of the Act (s.1(2B)), a learning disability is not to be considered a mental disorder 'unless that disability is associated with abnormally aggressive or seriously irresponsible conduct on his part'. We note that this proposal does not seem to arise from any of the recommendations made by the independent review, which in fact argued against an amendment to the definition of mental disorder 'for people with learning disabilities, autism or both'. This article, therefore, offers a critical reflection on the potential impact of this specific proposed change regarding autism, considering both a legal and practical clinical perspective. We acknowledge at the outset that many of the concerns identified in this review could equally apply to the proposal to remove learning disability; and indeed, some of our largest concerns relate to the group of people who might have profound impairments possibly as a result of both conditions. At the outset, we note that the proposed introduction of new definitions of mental disorder in the Draft Bill are arguably out of step with the principles of the MHA and mental health jurisprudence generally, where there is a tendency to specifically avoid definitive legal interpretations of mental disorder, thus preserving flexibility for practitioners. The recent trend has been for an increasing emphasis on assessment and treatment, not classification.
Before considering the specific issues at hand, it is of particular note that these proposals appear entirely novel in the context of English and Welsh law, and almost without international precedent. The only common-law jurisdiction to have done anything similar appears to have been New Zealand, which, in 1992, established that 'intellectual handicap' was a specific exclusion that 'will alone not lead to compulsory treatment and detention' (Mental Health (Compulsory Assessment and Treatment) Act 1992, s.4). Subsequent legislative changes have been required to bridge this legislative gap, 7 which arose in particular for a group of people with these conditions who also presented with significant behavioural risks, but where it was not appropriate -largely due to the individual's own vulnerability -for the person to remain within the custodial system, and for whom community accommodation was unavailable or unsuitable. It is our view that it is this same group -individuals with profound impairments who present with high risks -that might be most disadvantaged by the proposed changes in England and Wales. We hope this reflection illuminates why we believe this is far from a purely academic issue.

Creating a concept of 'legal Autism'
The legal definition of autism adopted for the purposes of the Draft Bill is 'a lifelong developmental disorder of the mind that affects how people perceive, communicate and interact with others'. This definition appears to be almost identical to language used in the National Strategy for Autism in England 8 and very similar to the definition adopted by the National Autistic Society, 9 though in both these cases, it is described not as a 10 In the context of the Draft Bill, this definition of autism is, of course, a legal and not a clinical one. In the messier world of clinical practice, two diagnostic frameworks are currently typically used to assess psychiatric conditions including autism (Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5) and International Classification of Diseases, eleventh revision (ICD-11)). While the definitions within these frameworks bear similarities to the legal definition, there are key differences. Some components considered essential in these clinical diagnostic frameworks, but not appearing in the proposed legal definition, are the requirement to consider restricted or repetitive patterns of behaviour, the requirement for the symptoms to cause a significant impairment in functioning, and the requirement to ensure that disturbances are not better accounted for by an intellectual disability. When taken together with the shift in diagnostic frameworks towards clearer codification of the impact of any diagnosis on functioning (for instance, DSM-5 has taken the approach of codifying the level of impact of severity of autism along a dimensional framework as 'requiring support', 'requiring substantial support', or 'requiring very substantial support'), this raises the possibility that many people who would not warrant a clinical diagnosis could come under the scope of the legal definition. Furthermore, the legal definition could, taken at face value, and at least in principle, be applied to a range of conditions other than autism; personality disorder could not unreasonably be argued to be a 'developmental disorder of the mind' (which certainly would also impact on how people 'perceive, communicate and interact' with others) as could, arguably, learning disability or indeed many types of developmental neurological impairment. Certainly, the definition, applied literally, appears far more encompassing than any definition that would be applied in clinical practice. This is presumably not an intended consequence of the proposed change.
The requirement to identify whether (or not) somebody has 'legal Autism' leads to a question of how such a diagnosis would or could be assessed in practice, and who would be responsible for such an assessment. Presumably, anybody with an existing clinical diagnosis of autism would be encompassed within the legal diagnosis, though this is far from clear. However, relying on an established diagnosis has problems. First, data suggest that many people who might warrant a clinical diagnosis are not formally diagnosed, 10 and there is also a wider concern about social inequality being linked to access the relevant assessments; for example, well-educated parents from higher socioeconomic groups are more likely to request and receive a diagnosis for their child 11 compared to parents from more deprived backgrounds. It is not entirely clear how those who might be suspected to warrant a diagnosis -or who indeed do warrant a diagnosis, but who do not yet have one -might be treated in law. This issue can perhaps be seen to be an inevitability given that autism is anchored to the concept of a 'spectrum', 12,13 which implies, among many other things, that what autism looks like in one person will likely be very different to what it looks like in another. While the 'spectrum' might be an imperfect way of representing our best clinical understanding of what autism 'is', the breadth of what is encompassed by the term is illustrated by a quote on the website of the National Autistic Society: '[s]ome people feel the spectrum is too broad, arguing an autistic person with 24/7 support needs cannot be compared with a person who finds supermarket lights too bright' (p. 9). This surely creates a challenging context in which to connect highly relevant consequences to a binary decision as to whether somebody does, or does not, have a diagnosis -particularly given that the proposed legal diagnosis is even broader than that used in clinical practice. Hence, it is arguably inherently difficult to create any legislative definition of autism, and indeed, as already noted, previous legislation has avoided attempting to incorporate such a definition for these very reasons (for instance, the explanatory notes to the Autism Act (2009) (s.1(11)) explain that the decision to not include a definition of autism in the Act itself was deliberate, reflecting the breadth of the 'spectrum' and the fact that the 'range of characteristics will change' in response to 'research and experience').
From a different perspective, the example of the introduction of the concept of 'Dangerous and Severe Personality Disorder' (DSPD), originally proposed in 1999 14,15 might have resonance with the issue of introducing an apparent diagnostic construct that transcends clinical data, even without the meaning of the diagnosis being spelled out in legislation. The DSPD concept was introduced prior to the failed 2002 (2)   appears to be quite different to present circumstances, where an emphasis on reducing 'unnecessary admissions' 17 has been no better illustrated than in the arguably failed 'Transforming Care' agenda. 18,19 With this supportive policy context, funding for new services and significant research activity followed the DSPD concept, and despite the 2002 Bill never coming into force, a whole new programme of interventions for people deemed to have a DSPD were developed. While a fuller review of the outcomes of the DSPD project in terms of its stated aims is outside the focus of this article, a legacy of the DSPD programme can be found within the Offender Personality Disorder (OPD) pathway. 20 For present purposes, while a definition of DSPD never entered into law, supplementary guidance was used to create an operational definition to determine who was eligible for DSPD treatment services. 21 Like the proposed definition of 'legal Autism', this definition did not reflect any existing clinical classification. On the contrary, rather than creating a very broad definition (as in the Draft Bill's attempt to define autism), the DSPD definition involved blending a highly specified clinical criterion of personality disorder (that a person needed to score above a certain level on an established clinical assessment of psychopathy and/or meet diagnostic criteria for a certain number of personality disorders) with a far more subjective 'dangerousness' criterion (the person needed to be judged 'a significant risk of committing a serious violent or sexual offence after release (within the next 15 years)'). This latter point perhaps understandably caused particular controversy. 22 One might reflect that key lesson of the DSPD 'experiment' 23 is that creating legal definitions of clinical problems is inherently difficult, particularly where the consequences attached to being deemed to have one of those problems are so potentially significant.

Assessing 'legal Autism' in the context of the MHA
One might consider the argument that the breadth of the proposed definition of autism allows the matter to be left to clinicians to determine using an appropriate clinical methodology and best judgement. Current best practice guidelines 24 require that clinicians should adopt one of a number of formal structured tests or assessments to support a diagnosis of autism. Perhaps, like the DSPD project, legislators could even go down the road of specifying particular tests and threshold scores that could be used to determine whether a person had autism. In any case, the need for additional assessments of autism would create a further un-resourced 'assessment gap' in existing specialist services (over 100,000 people are currently waiting for assessment of autism in England). 25 Even ignoring the resource issue, the decision about 'which test to use' remains a live question, 26 and we are certainly not at a stage where we are confident that a 'positive' diagnosis on one assessment tool would be consistently replicated by another. 27 Thus, any attempt to define autism by a positive outcome on any established assessment tool risks creating a group of people who might have autism by some measures, and not by others. Of course, even if a consistent tool were to be adopted for the purposes of assessing 'legal Autism', this would mean practical challenges with implementation, specifically for the existing services who did not already use this tool in practice. Even if a specific tool is not adopted and the matter is left to clinical judgement, the existing variation in practice between services (and clinicians) will create a situation where a person might receive a diagnosis of autism if they live in one area, but not in another. A more pragmatic approach, therefore, might be to create some kind of abbreviated assessment process for 'legal Autism' that diverged from clinical guidance but which was adopted consistently nationally. An obvious cost of such a proposal is that 'legal Autism' would need to be recognised as distinct from 'clinical Autism'. Nonetheless, such a proposal might be seen as attractive given, for example, the likely need for doctors approved under s.12 to make relatively speedy decisions about the need to detain in the context of an MHA Assessment (MHAA). However, clinically, such an approach could be seen to be beset with potential problems. One particular challenge might be the response to presentations where impairments or symptoms are shared between two diagnostic categories and there is uncertainty or disagreement as to which is the better explanation. It is increasingly recognised that psychiatric diagnoses have significant heterogeneity and that current approaches to psychiatric classification are limited. For instance, there are conceptual and symptomatic overlaps between the social impairments in autism, schizoid personality disorder, 28 and social anxiety disorder, 29 and some of the impairments in emotional processing identified in autism may appear hard to distinguish from psychopathic and narcissistic personality presentations. 30 In the case of an MHAA being completed by an s.12 doctor, how should they proceed if they consider one of these core impairments a key reason to recommend detention, but is unsure of the full diagnostic picture? One might argue that the Bill explicitly provides for this scenario by continuing to allow admission under s.2. This would, in theory, provide a period of 28 days under which an assessment could be completed, and a more certain diagnosis reached. While this may appear a feasible solution, the process of assessing somebody for a lifelong neurodevelopmental disorder within the context of an acute period of crisis is also problematic, at least if one is to carry out the assessment in line within the framework of accepted guidance. How valid would such an assessment be? Would a clinical team be able to arrange the necessary detailed reviews of early developmental history and interviews with informants meaningfully within this period? Moreover, how do clinicians attribute or explain impairments in communication or social functioning if their cause is unclear? What if the clinical picture remains unclear during the period of the s.2? Again, one can imagine the temptation for the implementation of a boiled-down assessment process to suit this circumstance.
Alternatively, therefore, an s.12 doctor might resolve this uncertainty by reaching for the 'non-Autism' diagnosis and pragmatically avoiding making a provisional diagnosis of autism. That would allow detention under s.2 and s.3 -that is, potentially indefinitely. However, what happens if autism is subsequently diagnosed after detention? Or if the responsible clinician becomes aware of a historically made diagnosis, hitherto unknown? Presumably, if autism becomes the only or primary diagnosis, it could lead to the person being discharged since they would no longer have a mental disorder within the meaning of the Act. This is itself a potentially significant concern for the patient, who might suddenly then find themselves without a completed assessment, without care and support, or placed at risk as a consequence of an unplanned discharge. However, there is also a potential concern for the service who had initiated the detention. Bearing in mind that autism is considered a lifelong developmental disorder, and so the identification of autism during an admission has the implication that it existed prior to the admission, this raises a potential question as to the validity of the original decision to detain. While s.139 of the Act limits an individual's liability in relation to decisions made under the MHA, National Health Service (NHS) Trusts (s.139(4)) are specifically excluded from this provision. In such circumstances, could a claim for arbitrary detention be directed towards the service? Might a fear of such claims fuel defensive practice?
If there is this risk, then the clinician's anxiety about 'missing' a diagnosis of autism, or the service's anxiety about civil liability arguably risks creating a situation where routine assessments of autism, quite possibly unnecessary and certainly un-resourced, are completed to rule out this possibility. The resource implications mean that these could only likely be abbreviated and quite possibly meaningless 'screening' assessments, completed purely to demonstrate the clinician exploring the possibility of autism existing. The fact these assessments would be completed by clinicians without specialist skills in autism assessment would only add to their likely clinical invalidity. From a patient perspective, this situation creates a significant potential incentive for patientsand particularly patients whose primary diagnoses might overlap with features of autism -to seek a diagnosis of autism in preference to other potentially overlapping diagnoses. Would financially able patients be motivated to instruct private assessments to support such a diagnosis, or to challenge an equivocal diagnosis made within the clinical team? What does a clinician do when a patient 'strongly suspects' that they always had autism but were never properly diagnosed? In extreme cases, might this potentially risk detained individuals fabricating an Autism presentation to facilitate discharge?
This raises another potential issue, which is how professional disagreements and disputes in the diagnostic process might be dealt with. Whose opinion has priority? In particular, how does a responsible clinician respond when they are confronted with a low quality but apparently conclusive assessment suggesting an autism diagnosis was warranted? The responsible clinician might have strong grounds to disagree, but one can imagine the difficulties that this might cause, not least in the context of a tribunal setting where the patient is represented by a proactive legal advocate. The responsible clinician might also, despite their better judgement, determine that the risk of detaining such a patient, who they believe does not have autism, is too great and hence instruct discharge, to avoid any possibility of inadvertently detaining a person who might later be identified to have autism.
Beyond the difficulty of responding to the blurred boundaries of a presentation of autism, one can imagine the difficulties that might arise with a client who has a clear diagnosis of autism, which exists comorbidly with another psychiatric diagnosis. The Draft Bill would seem to allow detention on the grounds of this alternative diagnosis. However, this then risks creating a nonsensical process of responsible clinicians attempting to partial out the elements of risk and treatment need that derive from one aspect of the presentation compared to another. It ignores the more likely possibility that both factors could work together in maintaining risk; for instance, rigid thinking maintaining conviction in delusional beliefs 31 ; social avoidance and inhibition maintaining low mood in depression 32 ; or difficulties in perspective taking maintaining interpersonal impairment consequent to personality disorder. 33 How would a Mental Health Tribunal (MHT) resolve such a dilemma? One might imagine a particular concern in the case of a patient with Autism and another psychiatric disorder, such as psychosis, who is refusing to take medication in the community. Is this better explained by a lack of 'insight' (a problematic concept in itself, but one typically attributed to psychosis) or rigid thinking (autism)? This may be here particularly problematic in the later stages of a detention when a patient is asymptomatic and the key issue is securing compliance with medication before discharge. How easy will it be to determine which mental disorder's 'nature' is creating the need for detention? One can imagine significant time spent resolving the dilemma between an advocate seeking to attribute all elements of risk to the person's autism, and a responsible clinician trying to paint a more nuanced picture of causality.

What is the goal of the proposed change?
More generally, the proposals invite questioning of why it is considered necessary to distinguish autism in this specific way. An article by Hollins et al. 34 provides one perspective; that 'learning disability and autism are distinct from the serious mental illnesses for which the Mental Health Act is designed to be used'. At face value, this is an attractive argument. But there are problems the other way in creating a 'hierarchy' of 'mental disorder', and the practice of drawing a clear line between 'mental illness' and 'Autism' is not necessarily so easy. This is not least because the diagnostic concepts that underlie the notion of 'mental illness' are themselves not without problems, 35 a situation that is not helped by a further tendency towards reification of the associated diagnoses. 36 For example, criticisms have been made regarding the validity of key diagnostic concepts including schizophrenia 37,38,39 and personality disorder, 40,41,42 and more broadly the research in this area is moving towards an accepted understanding that overlaps between problems are common, and that dimensions (rather than categories) are typically better ways of understanding mental health experience. 43,44,45 One might perhaps argue more strongly for a 'distinctness' of learning disability, but again one can identify examples to illustrate that it is not obvious how a line might be drawn in practice. For example, one might consider a person with a borderline learning disability who later has further cognitive impairment because of an acquired brain injury. Should they be treated as having a learning disability or a psychiatric disorder? What about a person with dementia in the context of Down's syndrome and an existing learning disability? While these questions might create an untidy legal landscape, and potentially invite bigger questions about the role of diagnosis in legal decisionmaking, for present purposes they raise a more basic question about the false simplicity of arguments based on the 'distinctness' of mental health problems. The examples cited earlier about aspects of autism, which are shared by other mental health diagnoses, is another clear case in point.
One could also argue that autism and learning disability could be distinguished from other mental health conditions by the fact that no medical treatment for autism or learning disability is available; indeed, as the White Paper states 'both learning disability and Autism are lifelong conditions, which cannot be removed through treatment'. But this begs the question: is the goal of treatment to remove a condition? If this question feels familiar, it is perhaps because the very same issue was debated in previous iterations of mental health reform, but in the other direction -where emphasis was placed on the fact that that the goal of personality disorder treatment did not need to be 'cure', 46,47 and subsequently leading to a considerable widening of the definition of 'medical treatment' (s.3(4) and s.145(1) and (4)). The present definition encompasses many other types of treatment beyond medication given compulsorily, including psychological therapies, occupational therapies, or behavioural/functional interventions. There is no requirement that somebody detained under the MHA receives medication at all if such treatments are available and have a clear therapeutic aim and medication is not clinically indicated. Furthermore, autism and learning disability are certainly not the only conditions where the primary 'medical treatment' provided under the Act may not be medication. One might, for instance, look specifically at other conditions, such as dementia, eating disorders and personality disorders, as examples where in many cases the primary aspects of 'medical treatment' may not be medication.
Another potential driver to the proposed change likely comes from a wider ambition to address a contemporary UK concern about the institutional abuse of people who have a learning disability and/or autism. Numerous examples have been highlighted in the press in recent years, with most of these reports coming from units operated by private companies but commissioned by the NHS. Winterbourne View 48 is a particularly infamous example, although other examples including Cawston Park, 49 Cygnet Yew Trees, 50 Whorlton Hall, 51 Eldertree Lodge, 52 and Mendip House 53 can also be noted. Indeed, two of these units (Whorlton Hall and Cygnet Yew Trees) are cited in the White Paper as an indication of the policy direction: 'The exposure of abuse at Whorlton Hall in May 2019 and at Cygnet Yew Trees this summer is confirmation that some detentions cannot be justified' (p. 6). Certainly, given these failings, nobody would disagree with an ambition to avoid further instances of institutional abuse, and indeed it may be the case that to achieve this, further legislative safeguards are justified. However, it does not logically follow that detention in a hospital is, in all cases, inappropriate. Indeed, it is not even clear that these failings are essentially unique to people with autism or learning disability, a point illustrated by the recent and widely publicised failings at the Edenfield Centre in Manchester -an inpatient secure hospital providing care for people with a wide range of mental health conditions. 54 Moreover, by allowing people with autism to be subject to 58. NHS England, 'Service Specification: Low Secure Mental Health Services (Adult)', 2021, available at https://www.england.nhs.uk/publication/service-specification-low-securemental-health-services-adult/.
immediately apparent, but rather than strengthening safeguards for people with autism, by restricting access to some of the safeguards within the MHA (such as the MHT, statutory advocacy, and the Second Opinion Approved Doctor), it may actually dilute them. One might argue that by leaving the door open to MCA, the Draft Bill might ultimately have no practical effect on the number of people with autism being 'detained'. There could, however, be a specific impact specifically for people with profound impairments who present with significant risks. This is because the national standards for commissioning of Low Secure Units 58 currently suggest that only people detained under MHA should be admitted to such units. This, therefore, seems to create a situation whereby somebody with profound impairments, long-term rehabilitation need, and significant risks could not be detained under s.3 (because of the exclusion in the Draft Bill), would in practice be unable to be detained under s.37 (because in practice the case would never go to court), and could not be detained in a secure unit under an MCA framework (because of commissioning exclusions). This leaves the only possibilities for this group as detention under the MCA framework in either unsuitable residential accommodation (placing both client and community at risk) or a non-secure hospital environment primarily focused on short-term detentions of people with acute symptoms (i.e. an acute ward or locked rehabilitation ward). A final concern relates to the implications for local authorities and commissioning bodies. Currently, anybody detained under s.3 is eligible for s.117 aftercare. This places a requirement on local authorities to provide community care and support to people who have been detained under the MHA. By excluding autistic people from the provisions of s.117, does this risk creating a perverse incentive for local authorities to 'find' a diagnosis of autism? Perhaps the largest concern would be if the proposed register that commissioners would be required to keep (Clause 125D) could not only be seen as a means of supporting commissioning decisions to support people with autism, but was instead used as a way of blocking access to s.3 detentions, and hence s.117; particularly for people who may have limited autistic traits but could still come under the broad legal diagnosis of autism. This would be a more serious risk if, for example, Approved Mental Health Professionals (AMHPs) were expected to consult this register at the point of an MHAA request, at which point it is possible the register could act as a de-facto 'exclusion list' from detention and aftercare responsibilities.

Conclusion
In conclusion, while the authors welcome and support the goal of ensuring that people with autism are offered treatment that is appropriate to their needs, and agree with a broad ambition to avoid unnecessary detention, it is our view that given the inherent problems in adopting legal definitions of clinical problems, a potential number of unintended consequences may be reasonably foreseen. The attempt to move autism out of the sphere of a 'mental disorder' is arguably much less a problem of inappropriate categorisation that applies specifically to autism, and much more a problem with the notion of defining and identifying 'disorder' itself. Unfortunately, it is argued that the proposals in this regard may actually undermine some of the clearly positive and important goals of the process of MHA reform. We hope this article stimulates debate on this important issue.

Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.