Archives ofDisease in Childhood 1991; 66: 737-741 737

CURRENT TOPIC Arch Dis Child: first published as 10.1136/adc.66.6.737 on 1 June 1991. Downloaded from

Childhood : its diagnosis, nature, and treatment

Sula Wolff

Leo Kanner,I although not the first to describe epidemiological prevalence studies have found this serious and intriguing developmental dis- two to four autistic children in every 10 000, order of early childhood,2 so precisely captured more if severely retarded children with autistic its essential features that the clinical account of features are included.2 his first 11 cases has never been bettered. The There have been several long term follow up salient characteristics he mentioned were an studies of autistic clinic attenders but only two 'extreme autistic aloneness' from the beginning of total population samples.5 6 Gillberg and of life, delay and abnormality in language Steffenburg followed up 23 autistic children and acquisition with echolalia and pronomial rever- 23 with 'other psychoses', often associated with sal, and 'an obsessional desire for the mainte- organic handicapping conditions, to the age of nance of sameness' in the presence of islets of 16-23 years.6 In 17% of both groups outcome normality, in particular an excellent rote mem- was 'good' or 'fair', and in 44% of autistic and ory. He saw the primary deficit as an inborn dis- 70% of other psychotic children it was 'poor' or turbance of affective contact and later described 'very poor'. Among this still quite young group similar, although very mild features of emotio- of subjects only one was self supporting, 23 nal coldness and obsessionality in the parents. lived with parents, 20 in institutions for the Since then research into autism and related handicapped, and one in a psychiatric clinic. developmental disorders has flourished, there is Altogether 35% of the total group (29% of autis- a journal entirely devoted to them and, although tics) had developed epilepsy, 50% between 13 they are so rare, their elucidation promises to and 14 years. One half had deteriorated symto-

shed light on many unanswered questions both matically in early adolescence with no difference http://adc.bmj.com/ about normal development of language and between the two groups. A perceptive observa- sociability and about the nature of a number of tion was that, while as children the autistics clinical conditions. were often particularly attractive to look at, Recent growth areas of research have been after puberty their appearance tended to the genetics of autism, the relation between aut- coarsen, often becoming clearly deviant. As in ism and allied clinical syndromes, the nature of previous studies, the best predictors of outcome

the psychological dysfunctions of autistic peo- were a preschool measured intelligence quotient on September 29, 2021 by guest. Protected copyright. ple, and the effects of specific treatment inter- (IQ) of 50 or more and the presence of com- ventions. There have also been changes in how municative speech before the age of 6. Epilepsy the condition is perceived and classified in the and fragile X chromosome abnormality were newer systems of diagnostic classification. associated with a worse outcome. Before reviewing recent progress, some basic Although many early symptoms-echolalia, facts must be recalled. social withdrawal, and stereotypies-tend to disappear with age, severe social impairments and abnormalities of language use remain, and- Age of onset, sex, social class incidence, the overall outlook for autistic children is poor. prevalence, and prognosis Nevertheless, that minority whose intelligence Though most cases, as Kanner himself noted, is within the normal range can do surprisingly develop the disorder insidiously from birth, in well despite serious childhood abnormalities. some there is an early period of apparently nor- Szatmari et al found that among 16 autistic chil- mal social and language development followed dren with mean IQ of 92, four had 'recovered' by a set back. Autism is distinguished from at a mean age of 26 years, living independent other pervasive developmental disorders, as we lives and with long term relationships.7 One had shall see, by its onset before 30 to 36 months of married. Eight in all were totally independent, age. More boys than girls are affected (a ratio of six needed only minimal supervision, and seven University Departmen. about 2-5-3:1). The sex ratio is more equal had obtained university degrees. On the other of , when autism is associated with organic brain hand, even among these rare well functioning Royal disease or severe mental retardation. The social autistics, one was profoundly retarded at follow Hospital, Morningside Park, class distribution of families with autistic chil- up, one had chronic schizophrenia, three had Edinburgh EH1O 5HF dren has been found upwardly skewed in some hallucinations, and two paranoid thinking. Two Correspondence to: studies, especially in well functioning autistic were unemployed, three in sheltered work- Dr Wolff. children,3 but not in others.4 A number of shops, and nine had never dated. 738 Wolff

Diagnostic classification, diagnostic from disintegrative disorders with their much instruments, diagnostic criteria poorer prognosis (DSM-III-R, still uses the 30

The diagnostic criteria for autism have changed months cut off between early and 'late onset' Arch Dis Child: first published as 10.1136/adc.66.6.737 on 1 June 1991. Downloaded from in emphasis over time8 as the condition has autism). come to be viewed not so much as a psychosis specific to childhood, as in the Intemational Classification ofDiseases, 9th revision (ICD-9),9 Differential diagnosis: the borderlands of but rather as a deviation of multiple develop- autism mental processes. This gave rise to the DSM-III Half of severely retarded children (IQ less than (Diagnostic and Statistical Manual ofMental Dis- 50) have symptoms of autism.'6 Although their orders, 3rd edition) and DSM-III-R (revised educational and treatment needs resemble those edition) concept of 'pervasive developmental of autistic children and they need much one to disorder'"0 11 and contributed to the envisaged one teaching to compensate for their social changes for ICD-1I.'2 Here too autism is classi- unresponsiveness, for research purposes a clear fied as a pervasive developmental disorder. This distinction must be made. For research and concept, like that of the specific developmental practice it is also important to differentiate disorders,'2 involves an onset very early in life, autistic children from those rare cases of disin- impairment of functions related to biological tegrative disorder after a period of at least three maturation, a steady course without appreciable years of apparently normal development. While remissions and relapses, improvement with age, the aetiology of this condition generally remains an increased incidence in males, and an often unknown, it may be caused by progressive, familial association with similar or related dis- degenerative brain disease such as cerebral orders. This concept has justly been criticised as lipoidosis or leucodystrophy. The end state is not in itself logical nor fully applicable to child- usually one of severe mental handicap with no hood autism'3 (autism does not affect all available language. Rett's syndrome may be developmental areas, its course is not always symptomatically indistinguishable from autism steady from birth and, while symptoms because of its early onset, but occurs predomi- change over time, only a few cases improve with nantly in girls and is associated with normal pre- maturation). The diagnostic criteria for autism natal and perinatal development. Rett's syn- itself, however, are clear and there is good evi- drome is also distinguishable by a decrease in dence for their validity.8 A detailed autism head growth after 5 months of age, loss of hand diagnostic observation schedule and an autism skills between 6 and 30 months, and increasing diagnostic interview of excellent reliability and gait and trunk apraxia and ataxia.'7 Doubt has validity, have recently been developed for recently been cast on the assumption that the research purposes by an international research fragile X chromosome abnormality, perhaps the group.'4 15 second most common cause of mental retarda- The diagnosis of autistic disorders, as they tion, is also specifically linked to autism. Payton are now called, rests on three main groups of et al found a 2 4% incidence of this abnormality http://adc.bmj.com/ symptoms: in 85 autistic males, similar to that in a mentally (1) Qualitative impairment and deviance of retarded male population.'8 reciprocal social relationships with abnormali- The differential diagnosis of autism from ties of gaze and social responsiveness to other schizophrenia, specific developmental language people, especially other children. disorders, and Asperger's syndrome (also refer- and red to as schizoid or schizotypal personality dis- (2) Deviant development of language order) is less problematical in clinical practice, other communication skills, with either no lan- issues to be discus- on September 29, 2021 by guest. Protected copyright. guage at all, early onset followed by loss of but raises many aetiological speech, or delayed onset and characteristic lan- sed next. guage features such as immediate and delayed echolalia, pronomial reversal, metaphorical Aetiology expressions and, especially, a failure in language Goodman has clearly summarised the argu- use (in the 'pragmatics' of language), relative to ments in favour of multiple coexisting factors in available comprehension and language skills. the causation of autism, although he does not (3) Repetitive, stereotyped patterns of make the point that the condition may be so rare behaviour, which include Kanner's 'obsessive because it requires the coincidence of more than desire for the maintenance of sameness', but one, perhaps quite common, causal agents.'9 also prolonged attachments to particular Goodman focuses especially on the coexistence objects, the formation of rows of toys irrespec- of distinct constellations of functional impair- tive of the needs of the rest of the household, ments necessary for autism: impairments of extreme food fads, repetitive questioning often general intelligence, deficits in mechanical lan- with insistence on ritualistic 'to and fros' with guage skills, as in the developmental dyspha- others, and motor stereotypies. sias, and deficits in social relatedness, play and While age of onset distinguishes well between non-verbal communication as in Asperger's syn- autistic disorders and early onset schizophrenia, drome. A difficulty here is that a number of which rarely begins before puberty, childhood children with specific developmental language disintegrative disorders as they are to be called disorders also have social deficits resembling in ICD-10 (the former disintegrative psychoses) those of Asperger/schizoid/schizotypal dis- tend to manifest only after the age of 3 and aut- orders,20 and almost half the boys with these ism usually before 30 months. In this classifica- disorders had serious or multiple specific tion age of onset, the fourth criterion, has been developmental delays of language related skills set to under 3 years to improve differentiation (S Wolff, unpublished observations). Childhood autism: its diagnosis, nature, and treatment 739

Two types of causal agents have been identi- together with single minded pursuits of particu- fied: those associated with brain damage and lar interests or behaviours. There have also been

genetic factors. It is known that autistic chil- a few case reports of families with both an autis- Arch Dis Child: first published as 10.1136/adc.66.6.737 on 1 June 1991. Downloaded from dren have somewhat higher rates of paranatal tic child and a child with Asperger's syndrome, abnormalities and that they are prone to develop and series of cases have been reported in which epilepsy, not so much in infancy and early a very early onset of schizophrenia was preceded childhood as are mentally retarded and severely by a clear cut autistic syndrome.27 birth injured children, but in early adolescence. Autism is likely to be a heterogeneous condi- In a minority of cases there is a specific associa- tion, with different causal factors operating in tion with some types of cerebral disorder and different groups of cases.2' In a very few the cli- not with others (with congenital rubella and nical syndrome is associated with chromosomal tuberous sclerosis for example, but not with or single gene abnormalities: fragile X chromo- Down's syndrome or cerebral palsy).2 More- some, phenylketonuria, tuberous sclerosis, or over, when monozygotic twins are discordant neurofibromatosis; in others with infections: for autism, it is the autistic twin who more often congenital rubella; and in yet others with evi- had evidence of paranatal injury. dence of diffuse brain dysfunction. Although There have been three population based twin symptomatically similar, severely mentally studies of autism: two in the UK, including a handicapped and well functioning autistic peo- follow up of one set of twins into adult life, and ple may differ in the causal agents and patho- one in Scandinavia (A LeCouteur, et al; paper logies involved. The genetic basis too may differ presented at First World Congress on Psychiat- in different groups of cases and may involve ric Genetics, Cambridge, 1989).21 2 The con- single or multiple genes. cordance rate for autism itself in monozygotic So far, while brain lesions have increas- pairs ranged from 36 to 91% and in dizygotic ingly been found with the newer technologies in pairs from 0 to 10%. The concordance rates for autistic children, no consistent pattern has other cognitive and social impairments, includ- emerged and children with and without such ing global and specific developmental delays lesions do not differ symptomatically.28 No spe- and deficient or unusual social responsiveness cific metabolic abnormalities have been identi- (which tended to get worse with age), ranged fied. High serotonin concentrations have been from 82 to 91% in monozygotic pairs and from 0 found in one third of autistic children, but also to 30% in dizygotic pairs. Clearly, what is inher- in severely retarded non-autistic children,29 and ited is either the full syndrome or a lesser the results of treatment with fenfluramine to variant consisting of both cognitive and social counteract such high concentrations have been impairments. equivocal. Further evidence for this comes from the stu- dies of siblings and other biological relatives of autistic children. The evidence, still requiring The nature of the deficits confirmation, is of an excess of cognitive delays The early work of Hermelin and O'Connor, in http://adc.bmj.com/ among the siblings of severely retarded autistic which psychological functioning of autistics was children, and an excess of minor social impair- compared with that of IQ matched retarded and ments of the Asperger/schizoid/schizotypal kind normal children, showed impaired encoding in the parents and other biological relatives of and restructuring of information.30 Although high functioning autistics.23 24Piven et al found short term memory, for example for random an excess of autism (3%), of 'severe social dys- words, was normal, the memory of autistics

function and isolation' (4A4%), and of cognitive unlike that of controls did not improve for word on September 29, 2021 by guest. Protected copyright. disorders (15%) in adult siblings of autistic chil- sequences with meaning. Autistic children dren, especially in those who had been diagno- could not use redundancy to improve their sed by Kanner himself.25 memory span. Moreover, their language deficit There is no excess of schizophrenia in the was but one aspect of a more general inability to biological relatives of autistics and no excess of use signs and symbols, and they were better at the much rarer condition of autism in the rela- processing spatial than temporal sequences; this tives of schizophrenics. The two conditions are is in line with their often remarkable jigsaw clinically distinct in their symptomatology, age solving skills. of onset, sex, and social class distributions and Since then there have been a number of associations with mental retardation and imaginative explorations of the autistic child's epilepsy.26 They may nevertheless share a gene- specific functional disabilities. Peter Hobson, tic predisposition to the schizophrenia spectrum again comparing autistic with normal and men- disorders of the Asperger/schizoid/schizotypal tally retarded subjects, examined their capacity kind. Altogether 75% of children diagnosed to match videos of people's vocalisation, facial 'schizoid' have the features of schizotypal perso- expression, and activities associated with those nality disorder in adult life and this is firmly emotions that have clear cut facial expressions part of the schizophrenia spectrum (S Wolff et (anger, fear, happiness, and sadness) against al, unpublished observations). The symptoms photographs and line drawings of faces or activi- of such children: solitariness, lack of empathy, ties expressing these same emotions.3" He found the single minded pursuit of special interest pat- autistics specifically impaired in the recognition terns, unusual modes of communication, and of vocalisations and facial expressions of emo- oversensitivity, although different, resemble tions. Even as adults autistic people are defi- those of autism in that they are all features of cient both in the recognition and expression of deficits in social and emotional relationships emotions verbally (that is, in their tone of voice) with other people and in communication skills, and facially.32 Hobson believes that the basic 740 Wolff

flaw in autism is in the biologically based capac- no child with an IQ below 50 gained useful ity, normally present in everyone, for attention skills in literacy or numeracy.

and empathic responsiveness to the emotional The more recent study by Howlin and Rutter Arch Dis Child: first published as 10.1136/adc.66.6.737 on 1 June 1991. Downloaded from expression and emotional behaviour of others. of the efficacy of an intensive, home based, Simon Baron-Cohen33 and Uta Frith34 see behaviourally oriented intervention has been these difficulties as part of a more general cogni- reported in a book that has much to teach all cli- tive deficit: an impairment in the capacity to nicians who look after autistic children, because infer what is in the minds of others-that is, a it focuses also on the problems of parents and lack of a 'theory of mind'. In one of a series of the other children in affected families.37 It imaginative experiments autistic, Down's syn- makes clear that, whatever the specific technical drome, and normal children were shown two interventions, families with an autistic child are dolls, one with a basket and one with a box. The coping with particularly stressful circumstances first doll puts some marbles in her basket and and require skilled counselling in their own goes out. The second then takes the marbles right. Such families need informed recommen- from the basket and puts them in her box. The dations for special schooling for the child, and first doll returns, and the question to the experi- social work services providing access to facilities mental child is: where does the first doll look for for respite care, holidays, financial entitle- the marbles? The autistic children tend to ments, and special, often residential, settings answer: in the box, because that is where they for the autistic person at and after adolescence. known the marbles to be. Other children of Due emphasis is given to the practical help and similar IQ can put themselves into the 'mind' of emotional support that can come from joining doll number one, know that this doll is unaware the National Autistic Society or one of its local of the transfer of marbles, and correctly point to equivalent organisations. the basket as the container first to be explored. As for the specific treatment interventions, Frith sees autistic individuals as 'natural they were individually tailored to the needs of behaviourists', who do not feel the normal com- each autistic child and family, with highly speci- pulsion to weave together mind and behaviour fic treatment goals. The focus was on fostering for the sake of coherence. the development of language, play, and social Baron-Cohen believes the primary defect in interactions. Treatment also aimed to decrease autism to be cognitive; Hobson sees it as pri- the specific obsessional, ritualistic behaviour of marily emotional. Yet both sociability and pre- the autistic children, and to help parents reduce linguistic behaviour develop together towards their children's often so distressing general the end of the first year of life and, though behavioural difficulties such as sleep and elimi- linguistic-cognitive and emotional abilities are nation disorders, food fads, temper tantrums, often selectively impaired (for example, in the and screaming attacks. Parents were the thera- retarded or in children with specific develop- pists, but care was taken not to impose too time mental delays), there is no reason why these two consuming a programme on them. While types of functions, so intimately connected in behavioural techniques of operant learning with http://adc.bmj.com/ the development of the baby, should not be prompting, fading, reinforcement procedures impaired together. Frith makes clear that the and 'time out', if necessary, were taught to the specific autistic impairments, which include mothers by demonstration in their homes after a also the absence of make believe play, of role full functional analysis of the child's behaviour, modelling, and an inability to distinguish pre- the approach to the families as a whole used tence play from reality, all involve competencies psychotherapeutic methods also.

which normally begin to develop in the second In this study treated children (all with an IQ on September 29, 2021 by guest. Protected copyright. year of life. This is also the time at which cor- of 60 or more and without associated medical rect empathic responses to distress in others are conditions) were compared with matched con- first seen and is precisely the time at which the trol groups both for short term gains (at six symptoms of autism first become apparent. months) and for long term outcome (at 18 Although the basic defect is likely to be inborn, months). In the short term, the greatest it does not manifest until the time at which the improvement was in reducing the children's impaired functions normally appear. general disruptive behaviour and their ritualis- tic and other solitary activities. The overall level of language usage also improved significantly, Treatment approaches with reduction of echolalia and stereotyped There have been two excellent controlled stu- utterances, but there was little difference in the dies of treatment interventions for autistic level of language complexity between treated children, one now a classic,35 36 the other more and control groups, both of which improved recent.37 The first demonstrated clearly the over time. Mothers of children just on the superiority of a structured, behaviourally threshold of speech were best able to put what oriented teaching approach geared to the child's they were taught into practice, presumably particular developmental level, and with much because their children were the most rewarding one to one teaching and contingent reinforce- to teach. In the long term, treatment worked ments. The children progressed more both edu- best for general disorders such as temper tan- cationally and socially, with less repetitive ster- trums, aggression, sleep and toiletting difficul- eotyped behaviour in free play settings, than ties; it also reduced but never totally removed children at schools where a permissive approach specific autistic rituals and routines. Play and was emphasised and the focus was on respond- social behaviour was also improved relative to ing to possible underlying emotional problems. the controls but always remained severely res- Even with the most effective methods, however, tricted. The communicative use of language, Childhood autism: its diagnosis, nature, and treatment 741

but not language competence, children II: some characteristics of the parents and improved also, children. Soc Psychiatry 1%7;1:163-73. but there was no difference at all in the overall 4 Gillberg C, Schauman H. Social class and autism: total level of cognitive capacity. The authors caution population aspects. J Autism Dev Disord 1982;12:223-8. Arch Dis Child: first published as 10.1136/adc.66.6.737 on 1 June 1991. Downloaded from 5 Lotter V. Social adjustment and placement of autistic that all treatment approaches need to be so children in Middlesex: a follow-up study. Journal ofAutism structured that the demands on the parents are and Childhood Schizophrenia 1974;4:11-32. 6 Gillberg C, Steffenburg S. Outcome and prognostic factors in reduced rather than increased, but also that, to infantile autism and similar conditions: a population-based avoid disppointment and treatment study of 46 cases followed through puberty. J7 Autism Dev frustration, Disord 1987;17:273-87. goals must be realistic. The results of this well 7 Szatmari P, Bartolucci G, Bremner R, Bond S, Rich S. controlled study will certainly contribute to A follow-up study of high-functioning autistic children. Jf Autism Dev Disord 1989;19:213-25. this. The study also demonstrates that without 8 Rutter M, Schopler E. Autism and pervasive developmental adequate controls new treatments can all too disorders: concepts and diagnostic issues. J7 Autism Dev Disord 1987;17:159-86. easily engender false hopes. 9 World Health Organisation. International classifcation of There has recently been much publicity for diseases. 9th Revision. Geneva: WHO, 1978. 10 American Psychiatric Association. Diagnostic and statistical 'holding therapy'. While this emotionally manual ofmental disorders (DSM-III). 3rd Ed. Washington, exhausting method may well lead to some DC: American Psychiatric Association. 1980. 11 American Psychiatric Association. Diagnostic and statistical increase in emotional contact between the autis- manual of mental disorders (DSM-III-R). 3rd Ed revised. tic child and the 'holding' parents immediately Washington, DC: American Psychiatric Association, 1987. 12 Rutter M. Child psychiatric disorders in ICD-10. J Child after the session, there have been no properly Psychol Psychiatry 198930:499-513. controlled outcome studies to back up the often 13 Gillberg C. Autism and pervasive developmental disorders. Annual Research Review. J Child Psychol Psychiatry 1990; extravagant claims for the efficacy of this 31:99-120. method. 14 LeCouteur A, Rutter M, Lord C, et al. Autism diagnostic interview: a standardized investigator-based instrument. What is clear, in the management of autistic J Autism Dev Disord 1989;19:363-87. children and their families, is first the need for I5 Lord C, Rutter M, Goode S, et al. Autism diagnostic observa- tion schedule: a standardized observation ofcommunicative all affected children to be diagnosed and for and social behavior.J Autism Dev Disord 1989;19:185-212. possible associated neurological or genetic 16 Wing L. Language, social and cognitive impairments in autism and severe mental retardation. J Autism Dev Disord abnormalities to be identified. Whether or not 1981;11:31-44. such pathologies exist, all autistic children 17 Gillberg C. The borderland of autism and Rett syndrome: 5 case histories to highlight diagnostic difficulties. J Autism should be notified to the educational services so Dev Disord 1989;19:545-59. that they can benefit from the special educatio- 18 Payton JB, Steele MW, Wenger SL, Minshaw NJ. The fragile X male and autism in perspective. J7 Am Acad Child nal arrangements they need and which are Adolesc Psychiatry 1989;28:417-21. increasingly available in the UK as a result of 19 Goodman R. Infantile autism: a syndrome of multiple local authority endeavours and the resources primary deficits. J Autism Dev Disord 1989;19:409-24. 20 Cantwell DP, Baker L, Rutter M, Mahood L. Infantile established by the parent societies for autistic autism and developmental receptive dysphasia: a compara- children. Second, all families with autistic chil- tive follow-up into middle childhood. J Autism Dev Disord 1989;19:19-31. dren need long term care and support to ensure 21 Folstein S, Rutter M. Autism: familial aggregation and gen- their access to all the treatment and back up ser- etic implications. J Autism Dev Disord 1988;18:3-30. 22 Steffenburg S, Gillberg C, Hellgren L, et al. A twin study of vices they may need as the child grows up. autism in Denmark, Finland, Iceland, Norway and These include home based Sweden. Jf Child Psychol Psychiatry 1989;30:405-16. http://adc.bmj.com/ intensive, specific 23 Wolff S, Narayan S, Moyes B. Personality characteristics of behavioural treatment approaches as outlined parents of autistic children: a controlled study. J Child by Howlin and Rutter for those children Psychol Psychiatry 1988;29:143-53. likely 24 Narayan S, Moyes B, Wolff S. Family characteristics ofautis- to respond37; short term medication with major tic children: a further report. J7 Autism Dev Disord 1990;20: tranquillisers to cope with particularly distres- 523-35. 25 Piven J, Gayle J, Chase GA, et al. A family history study of sing episodes of behaviour problems; and an neuropsychiatric disorders in the adult siblings of autistic understanding of the possible and individuals. J Am Acad Child Adolesc Psychiatry 1989;29: personality 177-83. marital difficulties of the parents and of the 26 Kolvin I. Psychoses in childhood-a comparative study. In: on September 29, 2021 by guest. Protected copyright. stresses on the siblings of the autistic child with Rutter M, ed. Infantile autism: concepts, characteristics and treatment, London: Churchill Livingstone, 1971:7-26. appropriate support and counselling. It is also 27 Watkins JM, Asarnow RF, Tanguay PE. Symptom develop- important to have social work help to ensure the ment in childhood onset schizophrenia. J7 Child Psychol Psychiatry 1988;29:865-78. families have access to their financial entitle- 28 DeLong GR, Bauman ML. Brain lesions in autism. In: ments and to holiday and other respite services Schopler E, Mesibov GB, eds. Neurobiological issues in autism. : Plenum, 1987:229-42. and more intensive back up at adolescence when 29 LeCouteur A, Trygstad 0, Evered C, Gillberg C, Rutter M. many autistic children will require day or Infantile autism and urinary excretion of peptides and protein-associated peptide complexes. J Autism Dev Disord residential units designed for their particular 1988;18:181-90. needs. Autistic children cannot generally 30 Hermelin B, O'Connor N. Psychological experiments with autistic children. London: Pergamon, 1970. benefit from work training centres set up for 31 Hobson RP. Beyond cognition: a theory of autism. In: non-autistic handicapped young people. The Dawson G, ed. Autism: nature, diagnosis, and treatment. New York: Guilford, 1989:22-48. National Autistic Society and other locally 32 Macdonald H, Rutter M, Howlin P, et al. Recognition and based parent societies for autistic people have expression of emotional cues by autistic and normal adults. Jf Child Psychol Psychiatry 1989;30:865-77. contributed much in the setting up of special 33 Baron-Cohen S. Social and pragmatic deficits in autism: schools and adult units and in offering mutual cognitive or affective? J Autism Dev Disord 1988;18: 379-402. support services for parents. 34 Frith U. Autism: explaining the enigma. Oxford: Blackwell, 1989. 35 Bartak L, Rutter M. Special educational treatment of autistic children: a controlled study. I: Design of study and charac- teristics of units. J Child Psychol Psychiatry 1973;14: 1 Kanner L. Autistic disturbances of affective contact. The 161-79. Nervous Child 1943;2:217-50. 36 Rutter M, Bartak L. Special educational treatment of autistic 2 Rutter M. Infantile autism and other pervasive develop- children: a controlled study. II: Follow-up findings and mental disorders. In: Rutter M, Hersov L, eds. Child and i1m4plicatio0ns for services. Jf Child Psychol Psychiatry 1973; adolescent psychiatry: modern approaches. Oxford: Blackwell 1985:545-66. 37 Howlin P, Rutter M. Treatment of autistic children. London: 3 Lotter V. Epidemiology of autistic conditions in young Wiley, 1987.