Article Text

Download PDFPDF
Autism spectrum disorder: diagnosis and management
  1. Anne O’Hare
  1. Section of Child Life & Health, University of Edinburgh, Edinburgh, UK
  1. Correspondence to Professor Anne O’Hare, Section of Child Life & Health, University of Edinburgh, 20 Sylvan Place, Edinburgh EH9 1UW, UK; aohare{at}


Autism spectrum disorders are of high prevalence and have a potentially complex range of presentations within the core impaired domains of social communication, reciprocal social interaction, imaginary thought and restricted and repetitive behaviours. Paediatricians need to recognise the possibility of these conditions among the high-risk populations of children with whom they work. This includes those presenting in the preschool years to child development clinics with delayed acquisition of language or general development delay or those presenting in the school years with coordination, academic, peer interaction and behavioural difficulties. In addition, paediatricians are essential members of the multidisciplinary teams charged with specialist assessment and their clinical history and examination can direct investigations for aetiology. This is a fast moving field with a challenging range of “grey evidence” causes and interventions. The approaches to managing these areas of work are discussed with an emphasis on recognition, important features in the history and clinical examination to aid differential diagnosis and investigations, interpreting the “grey evidence” and understanding intervention and prognosis.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Autism spectrum disorders (ASD) is now the preferred term for the neurodevelopmental disorders of empathy that include autism, Asperger syndrome, atypical autism and pervasive developmental disorder – not otherwise specified (PDD-NOS).1 ASDs are a syndrome complex characterised by serious impairments in social communication and reciprocal social interaction, with restricted and repetitive behaviours and impairments of imaginary thought.2 3 The heterogeneity of ASD presentations arises from the wide variation in affected children’s intellectual and language development, as well as the potential “obscuring” comorbidities such as attention deficit disorder or learning disabilities.4 Thus, there may be diagnostic challenges at many points on the spectrum. One example is the overlap in stereotypical repetitive movements that can occur in conditions of severe learning disability,5 or alternatively, the example at the other end of the intellectual distribution where children may present with an ASD-like phenotype that fails to reach diagnostic breadth and severity.6 Therefore a multidisciplinary diagnostic approach is recommended for security of diagnosis and also for characterisation of an individual child’s strengths and difficulties in areas such as their cognition and language, which are essential to inform the role of investigations and the relevance of interventions.7 8

Many paediatricians work in specialist diagnostic multidisciplinary teams and need to develop skills to recognise ASD in all its clinical “guises” and thus be able to employ specialist ASD diagnostic and assessment tools. However, the condition is now recognised to be of such high prevalence in the general population at around 1 in 100 children,9 that all paediatricians as well as colleagues working in child and family mental health services (CFMHS) and education need to be “ASD aware”. They have a particular responsibility to consider the possibility of ASD in the context of their work with high-risk populations of children, as there are many “perceived” advantages to early diagnosis and intervention although it is difficult to prove that earlier intervention is more effective. Paediatricians are well placed to recognise when onward referral for specialist assessment is appropriate as they are often in a position to be able to respond to parent and carer concerns.10

Paediatricians also need to understand the role of investigations in ASD and to know which approaches and strategies are likely to be helpful. Importantly, if they are informed about the debates and evidence bases around the controversial hypotheses in the causes of ASD, then they can contribute towards families feeling empowered to make sensible choices of treatments and interventions for their children.11 A striking example of this is when paediatricians can revisit with families the lack of a link between the measles, mumps and rubella (MMR) vaccine and autism, and therefore encourage them to complete their child’s vaccination programme. While ASD is seldom curable in a conventional sense, there is a substantial body of evidence that interventions can be very helpful, particularly for supporting communication and ameliorating maladaptive behaviours.12 In addition, families and young people themselves who are affected by the condition identify positive benefits that follow diagnosis that make their lives easier, in terms of their difficulties being understood and supported.7

ASD in the early years

The majority of preschool children presenting with an ASD will have delayed language and speech milestones; it is less common for Asperger syndrome, in which language is relatively unaffected, to present in this age group.13 Mothers recollect that they worried about their child’s lack of speech development from as early as 18 months.14 Delayed language development is the most common developmental difficulty seen by paediatricians and affects 6% to 7% of the preschool population.15 Although parents will often recognise difficulties in their infant’s social responsiveness, they may be less prone to voicing this and in addition the maladaptive behaviours that might be considered pathognomic of autism from the clinician’s point of view, such as ritualised and repetitive behaviours or mannerisms, may be absent or difficult to recognise in the preschool age group. Therefore, it can be very helpful to use a specific surveillance instrument such as the Modified Checklist for Autism in Toddlers (MCHAT)16 (table 1) or a clinical checklist such as that developed by the National Autism Plan for Children (NAPC) group and subsequently restated in the Scottish Intercollegiate Guidelines Network (SIGN) evidence-based guidelines (tables 2 and 3).1 7 17 Although paediatricians are used to assessing the clinical needs of children within the wider context of their family, some of these presentations can be particularly challenging if there is a complex psychosocial or child protection background. In this situation an opinion from colleagues in CFMHS can assist, particularly when interpreting whether an attachment disorder is accompanying the ASD or is the correct differential diagnosis.18

Table 1

Modified checklist for autism in toddlers16

Table 2

New York State Department of Health Early Intervention Programme17

Table 3

National Autism Plan For Children (NAPC): plan for the identification, assessment, diagnosis and access to early interventions for preschool and primary school-aged children with autism spectrum disorder (ASD)1

This active surveillance in high-risk children is drawing on our understanding of the critical importance of disrupted joint attention in autism.19 20 Joint attention is essentially the activity between two people when they attend to the same object and it results in one person checking the focus of attention of the other. Infants will look in the direction of another’s head and eye movements from about the age of 6 months and by 18 months they can reliably follow the gaze of others. The development of these skills relates to the onset of speech.21 Therefore autistic infants show delay in “protodeclarative pointing”, which is when a point is used to share interest rather than to make a request. Observations of this feature are captured in specialised direct clinical structured observations for the assessment of ASD such as the Autism Diagnostic Observation Schedule (ADOS-G).22 Another important early social orienting feature is that of turning to your name.23 Therefore it is not uncommon for infants to present firstly to paediatric audiology services and so again this is an opportunity to think about the possibility of an ASD diagnosis as an explanation for these concerning behaviours.

Finally, another high-risk group in which to consider the possibility of ASD when concerns are voiced is that of siblings, as the risk of ASD in siblings is around 8% overall. The risk is even higher where there are clearly identified genetic conditions such as tuberous sclerosis that result in autism as part of the severe developmental impairment.24 Research into the development of baby siblings of children affected with ASD has identified that diverging features start to emerge after the age of 1 year although this area remains a very active research endeavour and some of the subtle baby features that are going to be predictive of later autism spectrum disorder remain to be fully delineated.25 Such features appear to include delays in receptive and expressive language which are predated by reduced babbling and are accompanied by subtle anomalies of social responsiveness such as difficulty in emerging gesture, joint attention, pointing, imitation and response to name.

ASD in school-aged children and young people

Some children are not diagnosed as having an ASD condition until they start school,26 when the pressures to conform to the teacher’s agenda and work in groups and establish friendships with peers all become more of an expected focus. Children who present in this age group seldom have impaired acquisition of language, in contrast to children who present in the preschool years and are often diagnosed as having Asperger syndrome but it is important to remember that all children with ASD have pragmatic impairments of language, which is the term used to describe functional difficulties in the way that they use their language.27 In this school-age group, this pragmatic impairment may be manifest by difficulties in conversing, such as failing to take account of the listener’s needs or by problems understanding non-literal language such as idioms and metaphor. In addition, the child’s speech may be unusual due to abnormal prosody.28 Prosody is the suprasegmental aspect of speech that results in variations in pitch, fundamental frequency, loudness, pausing, intonation, stress and rhythm and thus affects the meaning of what you say. It has an important function of conveying emotion and even in able young people with ASD it can be strikingly unusual, such that their speech sounds robotic and can present a significant social barrier.

These pragmatic language impairments may be voiced by the referrer but alternatively a school-aged child’s presentation can appear entirely behavioural. Comorbid conditions such as oppositional behaviour, aggression and depression are common4 and may be more impairing of function than the underlying ASD.29 These disorders can have “physical” symptoms and presentations such as anorexia and so paediatricians who are not directly involved in child development work per se (eg, working in a feeding clinic will still need to know how to recognise the possibility of an ASD).

Children with Asperger syndrome may have poor coordination and have often received prior diagnoses of conditions such as developmental coordination disorder (DCD), dyspraxia, benign hypotonia, hypermobility syndrome, congenital ataxia and deficits in attention, motor control and perceptual abilities (DAMP). These can predate the recognition of the social communication difficulties and there is considerable overlap of symptomatology between DCD and Asperger syndrome. Both conditions can present with emotional difficulties and poor social skills and children with DCD have similar difficulties recognising emotional states from facial expressions.30 Children with Asperger syndrome also frequently experience difficulties in school from poor handwriting which can be compounded by dyslexia and difficulties with imaginary thought that affect composition. Alerting signs and symptoms of possible ASD in school-aged children and young people were highlighted in the recent SIGN 98 guidelines (table 4).7

Table 4

Sign guidelines

The diagnosis of ASD in these older age groups of children can present particularly difficult diagnostic challenges and so again it is critical that information is gathered from different settings outwith the clinic. There are a number of structured questionnaires that can be completed by parents and teachers that can be very useful in this regard.31


Regression in communication affects up to a third of preschool children with ASD and usually starts between the ages of 12–18 months.32 It can have a very dramatic or insidious onset and often leads to early paediatric referral. This regression is rare among children with specific language impairment and so it is an important symptom that points to autism. Language outcomes in adolescence may be similar for those with and without language loss but nevertheless is poorer than the outcome that might have been otherwise expected for children who had earlier language acquisition before the onset of their regression.33 However, there is no increase in epilepsy.34 The timing of the regression and the features of the language impairment are important considerations to orientate the paediatrician to the possibility of autism as opposed to other conditions that are associated with loss of speech (table 5).35 36 37 38 39 40 Electroencephalography (EEG) study is not indicated in autism unless it is part of the clinical work up for epilepsy.

Table 5

Differential diagnosis

Autistic regression is very rare before the age of 1 year and so if the clinician is faced with this timing, they need to think of other precipitants such as the autistic-like regression that can accompany infantile spasms with tuberous sclerosis and which usually comes on between 4–10 months.41 Alternatively, regression after the age of 3 should alert the clinician to the possibility of Landau–Kleffner dysphasia and both of these conditions may have very subtle outward signs of seizures so that the possibility that this is a paroxysmal disorder may be overlooked and, in the case of Landau–Kleffner dysphasia, there may be no seizures evident at all. Some of the features of Landau–Kleffner dysphasia in the expressive language, such as naming difficulties, verbal paraphrasias, telegraphic speech and neologisms, have a different quality to the stereotyped echolalic speech of ASD but there is some degree of overlap and mutism can occur in both regressions. The clinician also needs to keep in mind rare but treatable conditions such as biotinidase deficiency where there can be very early onset regression42 as well as important features in the clinical examination such as the trajectory of reducing occipital frontal diameter and onset midline hand stereotypes that can characterise Rett syndrome.43 There can also be developmental “setbacks” that can emerge as early as 16 months for children contending with a profound visual impairment.44

Clinical examination and the role of investigations

The SIGN 98 Guidelines for ASD advocate physical examination that pays particular attention to neurological abnormalities, dysmorphic features and the presence of learning disability to inform the role of investigations.7 In everyday practice, the likelihood of diagnosing an underlying aetiology is at best around 10%45 46 and the majority of these “syndromic” ASD diagnoses are genetic. However, none of these known genetic causes explain more than 1% to 2% of cases.47 Most evidence-based guidelines present an approach to clinical investigations in this way but it leaves the paediatrician with possible uncertainty as to where to go next in their investigation framework as this could be potentially enormous. The situation here is somewhat analogous to the investigation of global developmental delay/learning disability in which there is similarly no single major genetic cause and a vast range of potential causes. Rates of learning disability are over 70% for autism and over 50% for the whole wider autism spectrum.9 Although it is important to acknowledge that evidence-based guidelines for the investigation of global developmental delay in preschool children48 are not similarly evidence-based for learning disability comorbid with ASD, they can prove a useful starting point in thinking one’s way through the non-genetic and genetic aetiologies. The presence of the dual conditions does however lower the chances of finding an underlying aetiology.49

First line investigations for learning disability and developmental delay48 may still be relevant in the context of ASD and might include the investigation of urea and electrolytes, creatine kinase, lead, thyroid function tests, urate, full blood count, ferritin and biotinidase. There can be potential hypothyroidism in 22q11 deletion and a low calcium in William syndrome and both of these conditions have a higher rate of ASD than the general population.50 Dietary intakes in ASD can also be faddy and nutritionally inadequate.51

Presently, the SIGN evidence-based guidelines does not support the routine use of MRI brain imaging but it is important to integrate information from aspects of the physical examination such as the occipitofrontal diameter (OFC) to inform this. There can be early accelerated overgrowth in ASD52 and hence there will be a need to interpret the degree of macrocephaly in the context of autism as it may well not indicate the need for an MRI. Therefore, clinical features such as microcephaly might indicate an aetiology such as intrauterine cytomegalovirus infection and the latter can now be retrospectively confirmed from identification of viral load on the “Guthrie card” dried blood spot.53

Metabolic investigations should only be pursued where there are concerning additional clinical features such as organomegaly, coarse facial features and a consanguineous family history. As mentioned in the section on regression, EEG should only be pursued as part of the clinical assessment of epilepsy which has a lifetime rate of 30% to 40% in individuals with autistic disorder before the age of 30 years54 and in a recent population study of the broader ASD, epilepsy affected 16.7% of children by the age of 11 years.26

In fact there are relatively few non-genetic causes of ASD which are presently recognised and those that have been cited in more than two studies such as rubella embryopathy, herpes encephalitis, cytomegalovirus intrauterine infection, retinopathy of prematurity and thalidomide embryopathy are listed by Gillberg et al.50 Population-based studies have also identified foetal alcohol syndrome55 and prenatal exposure to valpoate can result in an autistic-like picture.56

All children with ASD should have an audiological assessment as, for example, intrauterine cytomegalovirus infection can result in a late-onset deafness so that the child can pass the normal universal neonatal screen. There are also helpful diagnostic stigmata to be found potentially from fundal examination (eg, cytomegalovirus (CMV), rubella, tuberous sclerosis), as well as the need to assess vision in all children with ASD because of the risk of impairment.55

Genetic investigations

Autism has a strong genetic basis and this is a fast-moving field.57 Excellent reviews are to be found in Abrahams and Geschwind,47 and Lintas and Persico.52 A strong genetic basis is supported by the evidence that mutations and structural variations in any of several genes can dramatically alter the risk of autism and in addition the relative risk to siblings is 25 times the background population prevalence, with siblings and parents showing broader ASD phenotypes and twin studies showing higher rates in concordant compared to discordant twins.58 59 60 Gene association studies, whole genome linkage studies and recent genome-wide techniques to assess copy number variation (ie, the insertion or deletion of a relatively large DNA fragment) have identified a large number of potentially important novel candidate loci.47 Numerous susceptibility genes including 2q, 7q, 15q, 15p14.1 and X have been described.52 57

Karyotype and fragile X analysis61 62 should both be undertaken and the former may identify sex aneuploidies that are often characterised by disturbances in speech and language development and communication, clumsiness, learning and autistic-like presentations.50 They should also be considered in children who have the characteristics of Asperger syndrome.

There are a number of well recognised genetic ASD-related syndromes that can contribute up to 1% to 2% of ASD aetiologies.47 These genes have multiple molecular functions from these different causes, which suggest that there must be a common pattern to the final biological pathway that results in the ASD symptoms.

It is helpful to conduct a Woods light examination to identify the characteristic hypomelanotic macules of tuberous sclerosis as these may only be visible in this way.

Other ASD-related syndromes that explain around 1% to 2% of ASD include Angelman syndrome with 15q duplication and UBE3A and other genes associated, and 22q deletion and 16p11 deletion. High-resolution chromosome analysis and whole gene micro array studies show cryptic chromosomal deletions and duplications such as those recently reported in Weiss et al.63

There are a large number of other rare syndromes in which a high proportion of affected individuals have an ASD presentation including neurofibromatosis, Cornelia De Lange syndrome, Down syndrome, Jouberts, Sotos, Potocki–Lupski syndrome and Smith–Lemli–Opitz syndrome. It is also important to remember that some of the mutations involved in Rett disorder result in a relatively asymptomatic phenotype or mild learning disability and a so-called verbal Rett variance and therefore should always be considered in girls with ASD.52

Lintas and Persico52 outline some helpful guidelines for genetic counselling in ASD and stresses that when the child has dysmorphic features and neurological symptoms, there should be a dialogue with the geneticist so that there can be further examination of further chromosomal rearrangements, even if the karyotype appears normal. Families who have a non-syndromic child with autism still have the probability of a 5% to 6% risk in further children.

Managing the “grey evidence” aetiologies

This is a vexatious issue as it is understandably of great interest and importance to affected families but it is an enormous literature and can be difficult to address in the context of busy clinical practice. This literature ranges from that in which the evidence is still evolving from early scientific enquiry, such as immunological mechanisms,64 65 66 and those exploring hypotheses such as epigenetic factors67 through to the potentially misleading literature where causes and interventions are asserted as fact when the evidence base does not support that.68 It is equally important to be aware of evidence bases that confidently refute hypotheses on the mechanisms of autism (eg, the demonstration that opioid peptides cannot be used as a marker of autism or as a guide to response to casein and gluten exclusion diets).11 The SIGN 98 guidelines listed all the biomedical search terms in their appendices and this can be a helpful starting point for the clinician who is asked about non-mainstream causes and treatments in the clinic, for if the search term is listed and there is no accompanying recommendation, then the clinician can conclude that there was no evidence of an acceptable level at that time.

Increasingly, new and potentially controversial treatments such as hyperbaric oxygen are being evaluated within randomised controlled trials.69 Another helpful avenue is the Research Autism website ( which is a charitable partnership with clinicians and researchers which has an easily accessible, user-friendly website, grading evidence as basic, intermediate and advanced so that it can be tailored at an appropriate level for the reader. The site includes objective and detailed evaluations of 100 mainstream and alternative interventions, each of which is rated according to the number and quality of scientific research studies published in peer-reviewed journal articles.


Most interventions in ASD are evaluated against outcomes such as cognition, language or amelioration of maladaptive behaviours rather than by quality of life. There are particular challenges to assessing quality of life in ASD as the child may not have the language skills to express their views or they may have difficulty identifying, discriminating and describing their emotional state. However, several possible indicators of quality of life in ASD, such as social support, academic functioning or satisfactory employment, family life and self determination are all important to keep in mind when advising parents.70

This is particularly the case if the clinician is considering pharmacological treatments as these can be effective but they have only a limited role in addressing particular symptoms and they should only be considered after a careful deliberation of the risks and benefits. They are generally considered for the treatment of comorbid psychiatric or neurodevelopmental conditions in ASD and are usually a short to medium term intervention. Pharmacological treatments by or in collaboration with clinicians who have appropriate training and experience include risperidone for the short-term treatment of significant aggression, methylphenidate for the treatment of attention difficulties and hyperactivity and melatonin for the treatment of intractable sleep problems.71

There is a substantive literature on the neuropsychological underpinnings of symptoms, signs and surface behaviours in ASD and these have made a substantial contribution to informing interventions. There is a whole range of examples that have been recently reviewed12 and they incorporate a wide range of psychoeducational interventions, the majority of which are based on behavioural theory or communication-focused therapies. Not surprisingly, many of the studies had methodological weaknesses, but the overall picture was of a positive outcome, with an impression that some form of treatment was more favourable than no treatment at all. Neuropsychological insights might allow for treatment to be adapted for children with ASD such as cognitive behavioural therapy for anxiety where the impairments of social skill, self-care, organisational skills and circumscribed interests and stereotypes can be accommodated in the way that therapy is applied.72

Similarly, neuropsychological insights have informed interventions for very young children such as targeting joint attention behaviour for infants73 and have given the impetus to identify the condition as early as possible in high-risk populations. Sometimes well established neuropsychological understanding such as that of “mentalising” that is to say the ability to impute the thoughts of others74 75 may be still under development in terms of an evidence base for efficacy but they do inform how a child’s environment might be potentially modified in order to support their difficulties.76 77

It is of course critical to pursue evidence bases for interventions, particularly as many of them are extremely intensive and expensive, but it is also important to remember that all children and young people with a disability are entitled to be supported so that they can benefit from their education and wider life experiences. In the absence of high levels of evidence, this can often mean that adaptations in educational environments and other settings have to be pragmatic, eclectic and individualised to arrive at a best attempt to adapt the child’s environment to promote their developmental progress and their quality of life.78

The literature from the charity Research Autism ( urges the professional to “recognise the value of people with ASD and the contribution they can make to society, without minimising the difficulties they and their families face”. Young people have voiced that they want to be told the truth about their ASD condition and that they found the diagnosis supportive and they expressed a preference that their peers know about the kinds of difficulties that they experienced and what might help them.7 In terms of prognosis, it is the quality of social engagement with peers that predicts better functioning and quality of life for individuals affected with ASD and more so than their cognitive functioning and academic achievements.79 However, over and above this, the outlook for living independently is much reduced if a child has an IQ below 70 and poor language development.80 Although generally speaking the dramatic regression seen in infancy is not repeated in later life, a more insidious form of regression has been described in adolescence81 and there has been the very rare but dramatic presentation described of onset catatonia in adult life.82

These are broad indicators to prognosis and children may do better than these historical groups if their condition is recognised early and they have early intervention. Prognosis can also be affected by the underlying syndromic explanation for autism if one exists. Thus discussing prognosis is complex and one may not be in a position to predict too far ahead at the time of early diagnosis but it is important to remember that parents value good quality information and also an opportunity to ask questions during the disclosure of their child’s diagnosis. They may also benefit from interventions to inform them about ASD and to help them build up their own skills in parenting their affected child. Children, young people and their families can be encouraged to continue to learn about ASD through interventions and support such as the National Autistic Society ( and the Scottish Society for Autism ( and a range of sources of further information appended in the recent guidelines (SIGN Guidelines 98).



  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.