Early Diagnosis in Autism and Asperger Syndrome


This paper describes ongoing work in the evaluation of early assessment materials but highlights, too, the difficulty of identifying any one measure that would be equally sensitive towards all children given the range of behaviours and of levels of severity of symptoms that exist within the population of children who are or who might be diagnosed with autism.

The final section also highlights the critical significance of language development in terms both of early identification of autistic symptoms and of the impact upon the course of development in a range of domains.


M.J.Connor                                                                                                August 2005



Problems in Diagnosing Asperger Syndrome in Very Young Children


The review notes by McConachie et al (2005) describe the reasons why relatively few children with possible AS are identified early, including weaknesses in the assessment instruments and the range of variation in the characteristics required for valid identification of this condition. 


They recognise that the age of diagnosis has gradually reduced with the expectation now that clinical judgement guided by a combination of multi-disciplinary assessments, parental interviews, and direct observation of abilities including social communication skill, will achieve a valid diagnosis at or before the age of 3 years.


However, experience suggests that the most likely candidates for referral to specialist clinics are children with poor language development or general developmental delay. Those children who have no evident delay in the onset of spoken language may not be referred or identified with AS until a later stage, especially if they pass the routine and early screening tests.

This is despite the consistent reports ( as highlighted by Gilchrist et al 2001) that parents of more able children with AS are able to look back upon concerns about their children’s performance at a very early age, certainly before 3 years of age, and most frequently about overactive behaviour, lack of social interest, and dislike of changes to routines.


There is also a growing consensus that it is more difficult to diagnose children with atypical features, and there is some question about the age at which a stable diagnosis of AS can validly and reliably be achieved.

For example, the main diagnostic features of AS relate to the inappropriateness of social behaviours, friendships, reciprocal communication, and special interests in the context of apparent adequacy and maturity of language and general cognitive ability.


Alongside the criterion of no significant delay in language skills, the issues to explore in seeking a diagnosis of AS include the level of self help skills, adaptive behaviour, and curiosity and the degree to which these aspects are (in)consistent with general intellectual development in the first three years.

McConachie et al note, however, that these characteristics may be difficult to quantify.


Similarly, motor weaknesses or clumsiness which have been associated with AS

( albeit not as a specific diagnostic feature ) are difficult to assess in 2 or 3 year old children, and the skills concerned show very wide variations, anyway, among the age group at large. 


Meanwhile, there is evidence that repetitive behaviours may show a different developmental pattern to that shown in respect of communication or social skills and that such behaviours may be less evident at age 2 to 3 than at age 4 to 5.

There is a suggestion ( see, for example, Shao et al 2003 ) that this different developmental course is indicative of a separate genetic source for the repetitive behaviours compared to the other social and communicative elements of AS. 


In their own study, the authors recruited a large sample of 104 children from community health centres. Their ages ranged from 24 to 48 months, and the inclusion criterion was the presence of complex social and/or communicative difficulties which could indicate ASD, although many cases were still in the process of assessment and did not yet have a clear diagnostic label.


In the first study, 46 children with ASD or language disorder were observed for precursors of executive dysfunction ; and the second study involved 58 children with suspected or diagnosed ASD in an assessment of the value of parental training and intervention.


The measures included the Autism Diagnostic Interview where the parents were the sources of information about the child’s developmental history, current skills, and behaviours. 

Any reports about developmental changes in repetitive behaviour were particularly noted with regard to sensory motor areas such as the use of objects or unusual sensory behaviours ; “resistance” which included issues such as a dislike of change or compulsions ; and interests which concerned unusual pre-occupations or attachment to certain objects.

The Autism Diagnostic Observation Schedule, a play-based rating scale, was used by trained examiners who presented children with various materials and play activities and observed the level of communication, social interaction, imagination, and repetitive behaviours.  

To gain an estimate of intellectual abilities, the Mullen Scales of Early Learning were used with ratings obtained on visual reception, receptive language, and expressive language.


The overall outcomes suggested that, of all the 104 children recruited at age 2 or 3 years with suspected or already diagnosed ASD or language disorder, only 3 children clearly met criteria for a potential diagnosis of Asperger syndrome.

Language and intellectual development were in normal limits, but criteria were met in the areas of reciprocal social interaction, and repetitive behaviours on the Autism Diagnostic Interview, and in the area of reciprocal social interaction on the Autism Diagnostic Observation Schedule.


The children were too young for meaningful measurement of abnormal development in motor, adaptive, or self help skills in the terms set down in ASD criteria, and there was confirmation, anyway, of the wide variability in “normal” development.

In fact, of 79 children thought likely to have autism or ASD, only 10 could be assessed as within the normal range of intellectual ability and as having achieved expression of a range of single words by 24 months.



Concerns is expressed lest children are likely to be missed, but the current authors review other studies and suggest that it is unlikely that many children who might eventually gain a diagnosis of Asperger syndrome will be referred for assessment before the age of 4 years.


Similarly, children in their own study identified with pervasive but non-specified difficulties, usually on the basis of language anomalies,  might in time be recognised as having a broader range of social impairments than could be identified at this age and stage. 

The authors acknowledge the complexity of using rating scales or diagnostic interviews and note that experienced clinical judgement may lead to a more stable diagnosis than strict application of DSM-IV criteria.


They refer to problems such as the reliance upon the child’s mood and performance on a given occasion or his/her level of willingness to interact with an unfamiliar adult in a cognitive assessment situation ; or upon the parental reports which may under- or over-estimate the child’s progress and weaknesses. 

This latter problem may arise particularly in the case of children with no significant language delay, where parents adapt to the child’s idiosyncratic style, and where the child has not yet been exposed to the social and other demands of admission to a school setting.


It is held that an urgent need exists for greater clarity in the defining of autism and ASD with some cut-off point for diagnosis agreed according to outcomes on the existing rating procedures.


With regard to pointers which might lead to earlier referral, the authors referred to signs of restricted or repetitive and stereotyped behaviours and noted how many parents remembered unusual and repetitive actions during the child’s first year. 


Other studies cited have indicated how parents have reported language, social development, play, and general behavioural problems in about equal proportions as giving rise to their first concerns, although there is some indication that over-activity may be a significant initial sign.

The suggestion is that attention would usefully be directed towards general management difficulties such as the effects of over-activity and poor sleep especially in association with difficulties arising in respect of language, play, or social interaction.


It was also noted that repetitive behaviours characteristic of autism do not appear prominent at 2 or 3 years of age, although where parents did report such behaviours among their relatively able children this did appear to be a significant element in the eventual diagnosis. 

The problem is that not all types of repetitive behaviour increase, and the observations in this present study suggested that resistance to change reduced with time …. although this could reflect the way in which the parents are able to adapt to the children’s behaviours and to tolerate rituals, or prepare the child well in advance for any changes. 


McConachie et al concluded that their evidence suggests that, among community referred samples, an early diagnosis of Asperger syndrome should not be assumed.  Rather, the majority of children will continue to come to attention when they enter educational settings and when friendship anomalies, poor self help skills, circumscribed interests, and inappropriate communication styles will highlight their likely special needs within the autistic spectrum.  


A Review of Rating Scales for Detecting Asperger Syndrome


In introducing his evaluation of 5 existing rating scales for AS, Campbell (2005) notes there is still controversy about the validity of AS as a separate entity distinguishable from autism, especially from high functioning autism.


The 5 scales are ….

The Asperger Syndrome Diagnostic Scale - ASDS ( Myles et al 2001 )

The Gilliam Asperger Disorder Scale - GADS ( Gilliam 2001 )

Krug Asperger Disorder Index - KADI ( Krug and Arick 2003 )

Autism Spectrum Screening Questionnaire - ASSQ ( Ehlers et al 1999) *

Childhood Asperger Screening Test - CAST ( Scott et al 2002) *


( * The latter two are largely confined to research studies )


Campbell also stresses that the matter of diagnosis, especially differentiating AS from high functioning autism, is complicated by differing diagnostic systems and by changing criteria over time. 


AS has been defined in a range of ways from the original description set down by Asperger in 1944 to the set of criteria contained in the DSM-IV. This leads to some practical difficulty in making a clear diagnosis, and Campbell feels that the cart has been put before the horse in that attempts to produce clear and agreed criteria for diagnosing AS have been made before AS has been fully validated as a separate and distinct condition.


The very task of establishing satisfactory diagnostic criteria for AS is that, as yet, there are no clinical or neuropsychological or behavioural markers which can reliably differentiate between cases of AS and cases of high functioning autism. 

The quest for reliable “phenotypic” differences between the two categories has produced inconsistent findings.  For example, the view that AS children have a verbal vs non-verbal cognitive advantage has been supported by some studies and refuted by others.


Given this, Campbell suggests that it is hardly surprising that all the rating scales reviewed showed significant limitations, and there is a question over their capacity to diagnose AS as a distinct condition.


Specifically, he describes how the authors of the ASDS, GADS, and KADI all report the capacity to differentiate AS from high functioning autism, but no information has been offered about the cognitive functioning of the sample of children with autism.

Campbell argues that matching of the two samples on cognitive functioning is important if one is to seek differential diagnoses ; otherwise the argument becomes tautological in that differences observed between groups would simply confirm that they are not directly comparable in terms of cognitive and language functioning.


Reference is also made to the limitations in standardisation and norming procedures given that there has been a dependence upon survey methods without independent confirmation of diagnoses of AS, with the possibility that users of these scales cannot be sure that the normative sample consisted only of children with AS.

Further, it may not be known by what precise means or with what definition the normative sample was diagnosed with AS in the first place, and the children may actually represent a heterogeneous group.


Campbell’s conclusion is that, reviewing at all the data, The KADI showed the strongest psychometric properties and most thorough item selection process, and the ASDS was the weakest among the published measures.

Both the research measures were seen as promising with the ASSQ showing good reliability albeit some weaker validity, and the CAST showing good predictive validity in the absence of published reliability data.


All the scales, however, showed shortcomings and the recommendation is that they should be used only with due caution when seeking to identify AS or to distinguish AS from high functioning autism.  It is appropriate to ensure that any such measures are just one part of a wider set of assessment processes within a multi-professional framework ensuring information from a range of sources.


                                         *          *          *          *          *


A similar theme has been pursued by Rellini et al (2004) in examining how two rating scales concur and conflict with the DSM-IV criteria for autism.


These authors, too, refer to the original description of the disorder by Kanner in 1943,  and to the autistic triad, but they note the difficulty in classifying the unique features of childhood autism in order to set down valid and reliable guidelines for diagnosis.


The rating scales under consideration were the Childhood Autism Rating Scale –CARS ( Schopler and Reichler 1980 ) and the Autism Behaviour Checklist – ABC 

( Krug et al 1980 ), both well established as means for screening for autism and for assisting in forming a diagnosis ; and the present authors set out to check the agreement between the DSM-IV criteria and the total scores on these rating scales.


A sample of 65 children were involved as participants, all of whom had been referred to the psychiatric and psychotherapeutic unit at a children’s hospital in Rome.  There were 58 boys and 7 girls aged between rising 2 and 11 years ( mean = 4.9 ; sd 2.2 ).

All were assessed on a battery of procedures including metabolic screening, blood tests, allergy tests, hearing assessment, and were clinically diagnosed by means of clinical interviews, developmental histories, ratings on the Vineland Adaptive Behaviour Scales, and observation of, or interviews with, the children.

Of the 65 children, 54 were diagnosed with autism, 9 with ASD ( 5 with Asperger syndrome and 4 with pervasive disorder unspecified ), 1 with ADHD, and 1 with language disorder.


The CARS material assesses behaviour in 14 domains and includes a rating for general impressions of autism ; and the ABC rates the children against a series of behavioural descriptors concerning sensory activity, relationships, stereotypical actions and object use, language, self help skills and social performance.


All the children had CARS and ABC scores although the clinical diagnosis and the study with the CARS and ABC were independent.


The findings were that the 54 children with autism as defined by DSM-IV all had a CARS score above 30, the cut-off point for diagnosis of autism, so that the sensitivity of the CARS was seen as full. 

The 5 cases of Asperger disorder, the 4 cases of PDD, and the cases of ADHD and language disorder, were all scored below 30 on the CARS.  It was held that this scale does not appear to recognise Asperger disorder and does not distinguish them from non-autistic individuals.


With the ABC, 29 of the 54 autistic cases had a score that was at or above the cut-off point for a diagnosis of autism.  The remainder of the 54 were variously categorised as having severe emotional disorder ( 2 cases), severe learning difficulty (19 cases), and sensory disabled ( 4 cases ).


The authors concluded that the CARS does seem effective in differentiating children with autism from individuals with other disorders … but it does not differentiate between individuals with other autistic difficulties like Asperger syndrome or PDD unspecified. 

However, this is unsurprising since the CARS ( like the ABC ) was devised before Asperger syndrome was included within formal diagnostic descriptors of pervasive developmental disorders.


The ABC is seen as a less discriminating scale than the CARS particularly with respect to mild/moderate cases of autism for which there is a relatively high rate of false negatives.


On a slightly tangential theme, the authors described data concerning the children of families who had immigrated from non-european countries and whose  behaviour was unusual in terms of seeking close body contact and total absence of imitation and all forms of communication, albeit showing hyperactivity and attention deficits of a kind similar to that observed in the rest of the sample of diagnosed autistics.


The implication suggested by the authors was that environmental influences could be significant in at least some forms of autism. 

It was speculated that the critical issues included the isolation of the immigrant families and the upheavals in their way of life and the associated probability of depression and loss of affective references ; and that such issues could act as supplementary causal factors in what is a (largely) neurobiologically pre-determined condition.


In any event, the authors recommended the use of the CARS rather than the ABC when rating scales are to be used, but they agree that a wide ranging diagnostic system is what is required and that a reliance upon single assessment measures is inappropriate and could underlie inconsistencies in reported prevalence rates.


Meanwhile, one might return to the review of early rating scales and the advice produced by the New York Sate Department of Health (1999) which recognises, too, that no single device is likely to be all-sensitive in identifying features of autism given that it is a spectrum condition involving various levels of severity in multiple developmental domains.


It is noted that the Checklist for Autism in Toddlers appears effective as a screening instrument to identify children with possible autism from 18 to 36 months. However, it is impractical to suggest a universal screening programme, so the advice is for the use of the CHAT with those children for whom there has been an expression of concern or where there are existing clinical clues suggesting possible autism.  In other words, the CHAT is primarily useful in helping to determine the further assessment of a child rather than in establishing an actual diagnosis.


With regard to the Autism Behaviour Checklist (ABC), the advice does refer to the advantages of a standardised method for assessing children with autism, but there is confirmation that sensitivity is relatively low and, in any event, the material appears more suitable for children above the age of 3 years. 


The Autism Diagnostic Interview- Revised is considered a generally useful device for use in a structured interview with parents since it covers a wide range of areas concerned with autism, and it does link to the current diagnostic criteria contained in DSM-IV. 

However, the administration can be very time-consuming, and the advice has it that this material would be most useful as part of a wider assessment of children for whom initial screens or expressions of concern have suggested a distinct probability of an autistic diagnosis.


The Childhood Autism Rating Scales are seen as the most frequently used standardised early assessment material and the advice refers to the acceptable combination of practicality in casework and in supporting research.  Again, the scales would be useful as a part of a wider assessment programme and the rating of symptom severity would be valuable in periodic monitoring of the progress of the children concerned, with some need to beware of false negatives and false positives,  and the relative lack of sensitivity to the milder presentations of autism or to cases where the issue is actually about severe learning difficulties rather than autistic symptoms.


Language and Psychological Development in Autism


The paper by Pry et al (2005) highlights the significance of language in respect both of its impact upon other areas of development and upon the intensity of symptoms of autism. 


They begin by noting how, among children with autism, the initial aspects of language are nearly always significantly delayed, and in many children the autistic features may appear as early as the 30th month following an initial development seen as normal.  This apparent regression typically follows the pattern of some loss of skills acquired, notably in the expression of words, then a period of stagnation of language acquisition, and then further development but at a lower level than is typical for a child of that age.


In cases where the initial onset of (spoken) language is late, there may be unusually rapid progress in language acquisition, with phonological and syntactic development approximating normal levels even if problems exist in semantics and pragmatics. 

However, the authors report that more commonly the first stages of language acquisition are marked by clear and negative indicators such as echolalia, inversion of pronouns, as well as stereotypical speech, which may be long lasting … especially the idiosyncratic or stereotyped style of language. 

It appears that the presence of these traits is inversely correlated with overall cognitive level.


The authors emphasise from their brief review of studies that language occupies a critical position among the various functional domains ; and they suggest that the psychological development in children with autism may depend to a significant degree upon the age at which the first ( language-based ) symptoms are observed and their intensity.


Their own study involved a large sample of children, with a mean age of 5 years

( range : rising 2 years to 7 years ) all diagnosed with autism and who were further assessed on subscales of the Vineland Adaptive Behaviour Scales to gain an estimate of general (cognitive) level.


The children were monitored by experienced clinicians who met regularly to ensure maximal consistency in assessments and data gathering on a series of instruments,  with some information gained retrospectively via the use of rating scales, language measures adapted from the Autism Diagnostic Interview, and parental interviews. 

The children were subdivided into 2 groups according to early or later age of onset of language  ( before or after 18 months ) with comparisons made against clinical measures indicating the age of detection of clear signs of autism.

The data were analysed in terms of relationships between expressive language level and performance on the CARS, Vineland scores, and the derived developmental quotient.  Subsequently, a matrix of possible correlations was constructed assessing the relationships among age of onset of autistic symptoms, their severity, overall expressive language level, and level of psychological development.


The results indicated that the age at which symptoms were detected varied considerably, although 4 age groups were distinguishable … below 12 months (40%), 12 to 18 months (18%), 18 to 24 months (19%), and above 24 months (23%).

The mean was around 18 months.


The CARS scores showed a mean of 37, with 25% of cases scoring from 25 to 30

( mild autism), 30% scoring from 31 to 37 ( moderate autism ), and 45% scoring above 37 ( severe autism ).

In respect of language, about half of the sample of children did not speak or had fewer than 5 words, around 30% had speech with at least 5 different words, and around 20% had spontaneous and functional language with sentences made up of 3 or more words.


Based on the “Daily Living skills ” section of the Vineland measures, psychological development was characterised as follows …

50% of the children with no apparent language skills had an age equivalent between 10 and 24 months ( mean = 19 ) ; for the group of children with limited language, the range was 27 to 38 months ( mean = 32 ); and for the children with functional language, the range was 43 to 52 months ( mean = 46 ).


With regard to correlations, significance was found in respect of the relationship between absence of language and intensity of autistic symptoms, and between the level of language usage and psychological development ( as measured in the Vineland Daily Living Skills’ section ).


The authors concluded, therefore, that language activity appears to play a major role in the relationship between the intensity of the observed autistic behaviours and general psychological development.


There is clear evidence that overall language development is closely related to the development of cognitive and adaptive competences.

Further, there is a second link between language level and symptom intensity …. it appears that it is the low level of linguistic activity, and not the adaptive level, which best explains the intensity of the syndrome at this early age.


In other words, it may be the language problems themselves that impact upon the overall disturbance within autism as illustrated by the anomalies in communication and social interaction, and by the repetitive behaviours.


Support is given to the assertion of Wallon (1942) who was among the first to highlight the significance of language for organising thought and influencing behaviour …. “ Language is the best systematised and expedient instrument of thought which handles not objects but symbols, or handles objects by means of symbols.”


Secondly, language is to be seen as having a strong emotional component ( or as the prime means of expressing emotion ).


Finally, what matters critically is the problem with the pragmatic use of language through which ideas, knowledge, and thought can be expressed and shared. 

It is at this point of development where language becomes critical, and where adequate interaction requires pragmatic skills and true two-way communication, that many children fail, or where their disabilities first become evident, because they lack this language function of directing mental orientation.


Pry et al suggest that many children with autism fail at this stage for the same reason. 


                                   *          *          *          *         *          *


M.J.Connor                                                                                                August 2005






Campbell J.  2005   Diagnostic assessment of Asperger’s disorder.   Journal of Autism and Developmental Disorders  35(1)  25-35


Ehlers S., Gillberg C., and Wing L.  1999   A screening questionnaire for asperger syndrome and other high functioning autism spectrum disorders in school age children.  Journal of Autism and Developmental Disorders  29  129-141


Gilchrist A., Green J., Cox A., Burton D., Rutter M., and Le Couteur A.  2001   Development and current functioning in adolescents with Asperger syndrome.  Journal of Child Psychology and Psychiatry  42  227-240


Gilliam J.  2001   Gilliam Asperger Disorder Scale.   Austin, Texas : Pro-Ed


Krug D. and Arick J.  2003   Krug Asperger Disorder Index.   Austin, Texas : Pro-Ed


Krug D., Arick J., and Almond P.  1980   Behaviour checklist for identifying severely handicapped individuals with high ;levels of autistic behaviour.   Journal of Psychology and Psychiatry  21  221-229


McConachie H., Le Couteur A., and Honey E.  2005   Can a diagnosis of Asperger syndrome be made in very young children with suspected autism spectrum disorder ?   Journal of Autism and Developmental Disorders  35(2)  167-176


Myles B., Bock S., and Simpson R.  2001   Asperger Syndrome Diagnostic Scale.  Los Angeles : Western Psychological Services


New York Department of Health.  1999   Assessment and Intervention for Young Children ( Autism/PDD ).  Early Intervention Programmes : Report 4215


Pry R., Petersen A., and Baghdadli  2005   The relationship between expressive language level and psychological development in children with autism at 5 years of age.   Autism  9(2)  179-189


Rellini E., Tortolani D., Trillo S., Carbone  S., and Montecchi F.  2004   Childhood Autism Rating Scale and Autism Behaviour checklist : correspondence and conflicts with DSM-IV in diagnosis of autism.   Journal of Autism and Developmental Disorders  34(6)  703-708


Schopler E. and Reichler J.  1980   Towards objective classification of childhood autism.   Journal of Autism and Developmental Disorders  10  91-103  


Scott F., Baron-Cohen S., Bolton P., and Brayne C.  2002   The CAST : preliminary development of a UK screen for mainstream primary age children.   Autism  6  9-31


Shao Y., Cuccaro M., Hauser E., et al  2003   Fine mapping of autistic disorder to chromosome 15q11-q13 by use of phenotypic subtypes.   American Journal of Human Genetics  72  539-548

Wallon H.  1942   De L’Acte à la Pensée : Essai de Psychologie Comparée.   Paris, Flammarion.




This article is reproduced by kind permission of the author.

© Mike Connor 2005.

Back to NAS Surrey Branch Welcome Page