© Cambridge University Press, 1993 This article was originally published in The Journal of Child Psychology and Psychiatry and Allied Disciplines, Vol. 33, No. 5, pp. 813-842, 1992. It is provided to Asperger Syndrome Coalition of the United States, Inc. for use on its website with the express written permission of Cambridge University Press. It may be downloaded and printed for personal reference, but not otherwise copied, altered in any way or transmitted to others (unless explicitly stated otherwise) without the written permission of Cambridge University Press. Hypertext links to other Web locations do not constitute any endorsement or authorization by Cambridge University Press.

Abstract:

Recent autism and autism-related research from Gothenburg is surveyed. In indigenous families, typical autism seems no more common now than 10 years ago. Genetic factors play a part in causing autism and Asperger syndrome. Certain medical syndromes carry a relatively high risk of concomitant autistic symptoms. Evidence for nonspecific brain dysfunction is often found in autism and autistic-like conditions. The search for the underlying clue to the riddle of autism may be futile. Autism might be best conceptualized as a behavioral syndrome reflecting underlying brain dysfunction which shades into other clinical syndromes. A new class of disorders of empathy is proposed.

Introduction

When in late 1989 Dr. Ann Gath wrote to ask me to present the Emanuel Miller lecture, I felt honored and immediately agreed to do so. I was a bit surprised that it was such (relatively) short notice; May 1 was only a few months away. However, as is often the case, a typing error had caused confusion. Ann Gath's letter had led me to believe that I was in for a 1990 ordeal and I began preparations for the lecture at once. When, a month later, her next correspondence made it clear that I was not expected until 1991, I had already made a detailed draft of what I was going to say in my head. The extra year that I was given in preparing this paper allowed me to conclude a number of studies which have shed new light on many of the issues I had planned to discuss. These studies have also, as always, complicated the overall picture so much that my first draft has had to be rearranged several times. I hope that the final product would have met with Dr. Miller's approval. Unlike some of the previous Emanuel Miller lecturers, I never met the man behind the lecture. However, from the little I have read about him I have the general impression that he must have been a man interested in many different aspects of psychiatry. This paper draws on several different studies of several different aspects of autism and (closely or distantly) related conditions. Therefore, to the extent that I manage to convey a multifaceted, yet coherent approach to the understanding of an important clinical problem, I shall have managed, at least in some minor way, to achieve one of the implicit objects of inviting people to present papers of this kind. It is now almost 50 years since Leo Kanner published his first paper on children with "autistic disturbances of affective contact" (Kanner, 1943). That paper continues to be one of the most cited (possibly the most cited) papers in autism literature. It is unusual for a first paper in a new field to have such a long and active life. In a preliminary survey of all autism papers published in this journal, the Journal of Autism and Developmental Disorders and Developmental Medicine and Child Neurology from mid-1988 through mid-1990, I found a reference to Kanner (1943) in 26 out of 102 publications (25%). Why is Kanner's paper so popular? Is it because of its crystal-clear description of the children? Yes, possibly to some extent, but other "first authors" have been equally clear without achieving the same degree of acclaim and interest. Is it because autism is such a fascinating condition? This can hardly be the case: the original writings on a number of other fascinating neuropsychiatric conditions (such as Asperger syndrome, Kleine-Levin syndrome and elective mutism to mention but a few) have not been cited with equal emphasis. Or is it because Kanner, in that first paper, managed to get so close to the true nature of autism that the progress that has been made after its publication comes down to very little in comparison? There is no easy answer to these questions, but there can be little doubt that Kanner's 1943 paper remains enormously influential in current thinking in autism. What if that paper had never been written? Would know-how in the field be substantially different today? We shall never know. However, merely by formulating the question, one touches on an important issue: did Kanner's paper set our minds thinking in fruitful or fruitless directions? Has it contributed to open-minded or one-track-mind thinking? Has pluralistic thinking in the field been damped down or even hindered? Based on some recent examples of stagnation in thinking about autism I would say that it might well have done just that. Let me try to spell out why I think it has and why I have chosen this point of departure for my brief survey of autism research in our centre and how it related to that of other groups. Many authors continue to quote Kanner as defining autism in terms of a disorder of affective contact. They use this quotation as an argument for the study of an affective primary deficit in autism as if the mere fact that Kanner said it was affective makes it more (indeed much more) likely that an affective primary deficit will in the end be found. We are constantly reminded that Kanner said it was affective and that for some reason the fact that Kanner, in 1943, thought it was affective was more important than research in intervening years, suggesting that it may not be affective. It was this same Kanner, who, in 1960, was quoted by Time magazine as saying that children with autism were the offspring of "parents, cold and rational, who just happened to defrost long enough to produce a child" (Steffenburg & Gillberg, 1989). It was also Kanner who said that the children were not neurologically impaired (in the face of evidence to the contrary in his own original writings). He asserted that they were of potentially superior intelligence (without citing any evidence for this). He thought the families were upper class. All of this assumptions have been questioned or disproven, often after decades of unproductive word-game debates. And yet, in spite of the fact that at least 10 population studies (Brask, 1970); Wing 1980; Anderson and Wadensj–, 1981; Gillberg & Schaumann, 1982; Bohman, Bohman, Bj–rk & Sj–holm, 1983; Steffenburg & Gillberg, 1986; Cialdella & Mamelle, 1989; L–gdahl, 1989; Ritvo et al., 1989; Gillberg, Steffenburg & Schaumann, 1991a) have shown normal distribution of social class and only one has shown a (slight) upper social class bias (Lotter, 1966). Some authors appear to want to cling to the notion of high social class in autism. Research in our centre has shown that children with autism are not neurologically normal. We have demonstrated over a long period of time that autism is often associated with specific medical conditions. Other groups have obtained similar results. On the basis of these strong associations we have, for many years, argued for the need of an extensive neuropsychiatric and medical work-up in all children with autism (Coleman & Gillberg, 1985). Not many have heeded this recommendation. Our group has obtained results supporting the forceful evidence from Lorna Wing's studies that "Kanner autism" is not a discrete disease entity with one etiology, but rather is one of several syndromes on a spectrum of autism and autistic-like conditions. We have not been able to find any support in the literature for "Kanner autism" being more valid than other variants of the "triad of social, communication and imagination impairments" (Wing, 1989a) characteristic of most people with autism and autistic-like conditions. I have not been able to find a shred of evidence for a clear distinction between "Kanner autism" and the other variants, including some cases with so-called Asperger f "Kanner autism" or what they consider to be "true autism". Why does this go on? I think it is because the doctors read that paper! Kanner's writings have dazzled us for so long. His concept of autism has had overwhelming power, forcing our thinking onto very narrow paths. Some of Andreas Rett's notions about Rett Syndrome (Rett, 1966) were wrong and diagnostic criteria and research were soon modified accordingly. It is high time we did the same with "Kanner autism", not because of any wish to be derogatory about the brilliant Leo Kanner, but because some of his autism concepts were wrong.

The Gothenburg studies

Several population-based studies from the Gothenburg Department of Child and Adolescent Psychiatry have bearing on the understanding of various aspects of autism and autistic-like conditions: (1) three population screening studies of autism and autistic-like conditions performed in (a) 1980, (b) 1984 and (c) 1988 in Gothenburg (a) and in Gothenburg and the rural county of Bohusl”n (b + c) (Gillberg, 1984; Steffenburg & Gillberg, 1986; Gillberg et al. 1991a); (2) a population-based twin study of autism in the Nordic count Syndrome (Asperger, 1944; Wing, 1981; Gillberg, 1991a). Yet many groups insist on getting back to the old narrow (allegedly better defined) syndrome ories performed in 1983-85 (Steffenburg, et al., 1989); (3) a population-based study of mental retardation in 1983 (Gillberg, Persson, Grufman & ThemnÈr, 1986); (4) a population screening study of deficits in attention, motor control and perception (DAMP) launched in Gothenburg in 1977 (Gillberg, Rasmussen, Carlstr–m, Svenson & Waldenstr–m, 1982; Gillberg, 1983); and (5) a total population study of anorexia nervosa launched in Gothenburg in 1985 (RÂstam, Garton & Gillberg, 1989). All of these studies, except the Nordic twin study, include longitudinal follow-up, so far for 2-14 years. In addition to these studies, a number of clinic referral based investigations have also contributed new knowledge. The possibility of comparing data obtained in clinic studies with basic information from the population studies has enabled a better subgrouping of findings into those that are likely to be representative and those that may reflect referral bias rather than facts about the condition as such. The Main Issues to be Discussed This paper will deal briefly with the following issues in relation to the above-mentioned studies against a background of research by other groups: (1) prevalence, (2) etiology, (3) recent neurochemical and neurophysiological findings, (4) differential diagnosis, (5) course and outcome, and (6) autism as one clinical syndrome among a broader group of empathy disorders including certain treatment/management aspects. Is Autism Exceedingly Rare? Kanner thought that autism was rare. Rarity, of course, depends on how you define it. In the case of autism it is also dependent on how you define autism. Most authors before 1980 reported "nuclear", "classic" or "Kanner" autism prevalence figures of 0.7-2.0 per 10,000 children (Lotter, 1966; Treffert, 1970; Brask, 1970; Wing & Gould, 1979). Our group in the mid-1980s reported that "infantile autism", meeting Rutter (1978) and DSM-III (APA, 1980) descriptions, was encountered in 2.0 per 10,000 children under age 18 years in the region of Gothenburg, Sweden (Gillberg, 1984). As in all our studies of "infantile autism", all the children with infantile autism showed autistic aloneness and an obsessive desire for the preservation of sameness, the two symptoms thought by Kanner to be necessary and sufficient for an autism diagnosis (Kanner & Eisenberg, 1956). The study reported a population-based screening performed at the end of 1980. Four years later, in 1984, Suzanne Steffenburg and I (Steffenburg & Gillberg, 1986) reported a new, population-based study of autism in Gothenburg. The methodology and definitions used were the same (except that in the new study only children under age 10 years were included) as in the previous study and the research team was the same. The rate of "infantile autism" was now 4.7 per 10,000. Three years ago, in 1988, our group again performed a population-based study (Gillberg et al. 1991a) and the autism prevalence was 7.8 per 10,000. Except for the "nuclear" autism cases, there were also a number of cases with "other psychoses" or "autistic-like conditions". These possibly correspond to "non-nuclear autism" according to Lotter (1966) or to "pervasive developmental disorder NOS" according to the DSM-III-R (APA, 1987). They constituted 1.9, 2.8 and 3.4 per 10,000 respectively in the three population groups. Even though they too increased in number, their relative contribution to the whole group of "autism and autistic-like conditions" decreased steadily. These "autistic-like conditions" were very similar in symptomatology to those with "infantile autism", but they either had an onset after age 30 months or had some uncharacteristic traits such as clinging behavior, exceptional degrees of hyperactivity, extremely frequent seizures, etc. On the whole, however, similarities rather than differences prevailed when they were compared with those who received a diagnosis of "infantile autism". They differed clearly from the autism group only with regard to sex ratio. The boy:girl ratio was high in the autism group but almost equal in the group with autistic-like conditions. One reason for the low rate of typical autism in females could be the fact that the gestalt of the autistic syndrome has been invoked from the male prototype (Kopp & Gillberg, 1991). The same type of condition could occur in girls, but with a slightly less typical "male" symptomatology. Given our current diagnostic criteria, these girls would not receive a diagnosis of classical autism but rather one of "autistic-like condition", leading to a higher girl:boy ratio in this group. There was also a tendency for epilepsy to be more common in the autistic-like group. This could mirror similar diagnostic practices. Children with very early onset epilepsy, in whom it is impossible to tease out clearly whether autistic symptoms preceded or ensued the onset of epilepsy, may be less likely than those without early onset epilepsy to receive a diagnosis of classical autism. Over the last 10 years, a number of other researchers have reported higher rates for autism prevalence (Bohman et al., 1983; Bryson, Clark & Smith, 1988; Tanoue, Oda, Asano & Kawashima, 1988; Ciadella & Mamelle, 1989, Sugiyama & Abe, 1989; L–gdahl, 1989). It has been impossible to determine the reason for this increase. Discrepancies in diagnostic criteria may have been important, though, since several of the cited studies have referred to different diagnostic systems. There has recently been growing concern that the introduction of the "autistic disorder" concept in the DSM-III-R (APA, 1987) might have inflated autism prevalence figures. However, this could not account for the high prevalence in the published epidemiological studies, since they did not use this diagnosis at all. Nevertheless, gradually looser autism concepts could have contributed to the steady prevalence rise. The Gothenburg studies go some way in offering possible explanations. Figure 1 shows how the prevalence of autism and autistic-like conditions in 1980, 1984 and 1988 was distributed according to IQ and immigrant status of the child's family. The rate of autism (and the rate of autistic-like conditions) associated with mild mental retardation has remained at almost exactly the same level throughout the whole period. Autism associated with immigrant status, severe mental retardation and near normal or normal intelligence, on the other hand have all shown an increase. In the case of near normal/normal intelligence this has been the case in spite of the fact that the relative contribution of autism associated with near average/average IQ to the whole group of autism cases has decreased. The cases Kanner described mostly came in the mildly-moderately retarded IQ range. Thus, the rate of typical "Kanner autism" in Gothenburg appears to be stable. FIGURE 1 GOES HERE~~~~~~~~~~~~~~~~~~~ There has been a very considerable increase in the number of cases diagnosed with "infantile autism" among children with severe mental retardation. This has been interpreted by us as showing the influence of education on those involved in screening. From 1980 to 1988 autism know-how has increased in Gothenburg. A state-wide diagnostic centre came into existence in Gothenburg in 1985 following an intensive period of lectures and spreading of information about autism in other ways (books, booklets, posters distributed to hospitals and well-baby clinics, newspaper articles, slide series and videos). Since 1985, the educative interventions have been further intensified. The realization that autism can - and does exist among those with severe mental retardation has led to increased autism sensitivity among staff and parents involved with severely mentally retarded children, which, in turn, has resulted in higher yields in this subgroup of the population in our screening studies. The greatest increase of autism-severe mental retardation-prevalence occurred from 1980 to 1984. On the basis of the finding of only a relatively small increase from 1984 to 1988 a period during which education about autism increased dramatically in the region it seems that we may now have "reached the ceiling" with regard to the severely retarded autism group and that the numbers ascertained may be close to the true prevalence in the population. The number of cases of autism with near or normal intelligence has increased, in spite of the fact that the relative contribution of this group to the whole autism prevalence has decreased. The reason for the increase appears to be better awareness in the general population (including parents and teachers) that normally intelligent children with autism do exist. The reason for the decline in relative proportion is the even larger increase in cases of severe mental retardation and in autism cases in immigrant families. In 1980, few immigrants from far-away countries lived in Gothenburg. In the 1980s, the number of such immigrants increased year by year. They still contributed relatively little to the total population of Gothenburg (c. half a million inhabitants), but their childbearing rate was higher than that of native Swedes. Furthermore, it seems clear that the rate of autism in their offspring is considerably higher than in the offspring of parents born in Sweden (Andersson & Wadensj–, 1981; Gillberg, Steffenburg, B–rjesson & Andersson, 1987a). This high rate has been interpreted as possibly showing the contribution of environmental factors in utero (such as viral infections including rubella). As is clearly demonstrated in the figure, both the absolute and relative number of autism cases contributed by immigrant parents increased over the 8-year period. The possibility of a contribution from intrauterine viral infections was indirectly supported by findings in a recent study from our group that children with autism in Gothenburg were more likely to be born in March and that the March-born cases came from poor social conditions, increasing the risk of pregnancy viral infections (Gillberg, 1990). The rate of autism among children of immigrant parents appeared to be high only in those whose parents had arrived in Sweden from far away countries. In summary, it appears that the only "type" of autism that has not increased in prevalence over the last 10-year period is the nuclear variant described by Kanner. Except for a real increase contributed by children of immigrant parents, the increase among the other two subgroups, those with autism with severe mental retardation and those with relatively high-functioning autism, is likely to be apparent rather than real and brought about by better detection in 1984 and 1988 than in 1980. Thus, it seems that autism is more common than previously believed but not more common than it used to be 10 years ago, at least in the native Swedish population. Asperger syndrome cases have not been included in the prevalence figures cited. If Asperger syndrome is seen as synonymous or at least partly overlapping with high-functioning autism, the prevalence figures constitute an underestimate. Much of the available data, including some from our own studies of children with DAMP (some of whom have associated Asperger syndrome), indicates that Asperger syndrome might be considerably more common than Kanner autism. A number of other conditions, including other cases of DAMP (other than those also meeting Asperger criteria), obsessive compulsive personality disorder, Tourette syndrome and anorexia nervosa also show all, many or some of the features of autism. If those cases with such conditions meeting full criteria for autism were included in autism prevalence figures, the population prevalence would definitely be higher than presently believed. Wing's London studies have indicated that the triad of social, communication and behavioral impairment may be 5-10 times as common as "nuclear autism" (Wing & Gould, 1979). Our studies of children with mental retardation in Gothenburg (Gillberg et al. 1986) have corroborated these findings. So, in answer to the question whether autism is exceedingly rare or not, I would say that autism might not be at all that uncommon and in fact the number affected may be several times that previously reported in the literature. "Kanner autism" appears to be as rare as once believed to be. However, this is true only if we limit our concept of Kanner autism to apply to those who traditionally are most likely to receive this diagnosis, viz. those with mild mental retardation. Few researchers seem to realize that thinking along these lines, although superficially reasonable and straightforward, may carry a high risk of running into circular augmentation. Our epidemiological studies, and those of Lorna Wing, suggest that even if the criteria for autism are kept rather narrow, disorders similar or identical to those described by Kanner (and by Rutter and DSM-III descriptions) exist in a considerable number of cases that would not have been thought of as Kanner autism just because they happen to come in a different costume different with respect to IQ, sex, personality and other features. These matters will be further discussed in the section on differential diagnosis. Are Children with Autism Neurologically Intact? One of the few studies before 1985 which did not appear to be heavily influenced by the tradition pursued ever since Kanner's first paper was published by Lorna Wing and Judith Gould in 1979. These authors studied social impairment among a population of handicapped children under age 15 years in southeast London. They found that severe social impairment tended to cluster with communication impairment and restrictions in the behavioral repertoire. This "triad" of social, communication and behavioral impairment tended to co-occur in some children regardless of whether the clinical gestalt of Kanner autism could be invoked or not. Furthermore, they found that clear or indirect evidence of neurological dysfunction was very common in the whole of the socially impaired group, both in those classified as Kanner autism and those with other clinical syndromes comprising the triad. Long before Lorna Wing, the literature was full of direct or indirect evidence of brain dysfunction and neurological impairment in subgroups of children with autism (e.g. Ritvo et al., 1970; Rutter, 1970). The fact is that even some of Kanner's first 11 patients showed hard evidence of neurological dysfunction (Kanner, 1943; Kanner, 1971). Very briefly, it has been clear for several decades that there are quite a number of children with classical Kanner autism who are not neurologically intact. The Gothenburg studies have demonstrated that some kind of neurological/neurophysiological impairment is the rule rather than the exception in autism. Some of the main findings from these studies were recently reviewed in Steffenburg (1990). At the present stage, with regard to etiology, one can discern at least four broad groups in autism: (1) familial autism, (2) autism associated with specific medical condition, (3) autism with unspecific signs of brain dysfunction without family history of diagnosed medical condition, (4) autism without any clear evidence of autism family history or brain dysfunction. In one of the Gothenburg population studies, Steffenburg (1991) found 8.5% in group (1), 37% in Group (2), 46% in group (3) and 8.5% in group (4). In an even more recent study in which the three population groups were pooled together with a large number of referral cases, Gillberg and Gillberg (personal communication) found almost identical proportions belonging to the four groups (Fig. 2). That autism might have genetic roots has been suspected with varying degree of emphasis ever since 1943. However, it is only in the last decade that hereditary factors in autism have come to be regarded as possibly important. The evolution in this respect has run parallel with the landslide developments in the field of new genetics. However, already in 1977, Michael Rutter's team (Folstein & Rutter, 1977), presented evidence from the first systematic twin study FIGURE 2 GOES HERE~~~~~~~~~~~~~~~~~ (*) Medical syndrome: Moebius syndrome Laurence-Moon-Biedl syndrome Fragile X syndrome Rett syndrome Other chromosomal syndrome Congenital hydrocephalus TS Williams syndrome NF MCA/MR syndrome of autism that some kind of hereditary cognitive factor associated with autism played an important role in the development of some cases of autism. The fact that the sibling rate of autism in families in which one child had been given a diagnosis of autism was (and remains) much higher than in the general population also supported a genetic basis for some autism cases. Our own population-based twin study of autism showed very high concordance rates in monozygotic pairs, but no concordance in the dizygotic pairs (Steffenburg et al., 1989) (Figs.3 and 4). The monozygotic-dizygotic difference remained after exclusion of fragile X positive twins. Our data could be interpreted as showing an autism-associated hereditary factor in some cases of autism, but there was no strong evidence that this factor was cognitive rather than social or social cognitive. Bolton and Rutter (1990) recently concluded that there is a "significant familial loading for specific cognitive disorders" (particularly of speech, language and reading). They based this on a survey of six studies of siblings of children with autism. Only two of these studies made use of a comparison group of (Down syndrome) siblings of children without autism. The matter is in need of further investigation, but the evidence can hardly be regarded as conclusive. In our own studies (Steffenburg, 1991), we looked at the rate of learning disorder (mental retardation or disorders of speech, language or reading requiring remedial measures) in mothers, fathers and siblings of children with autism and similarly aged comparison cases with and without deficits in attention, motor control or perception who had been drawn from the general population of children and found no clear evidence of differences. However, the rate of Asperger syndrome in mothers, fathers and siblings appeared to be increased, even though numbers were too small for differences across groups to reach statistical significance. Altogether 9% of children with auti sm had a father with Asperger syndrome or Asperger traits. In 6% of the autism cases this was the only clue as to pathogenic mechanisms. Major problems pertain to studies of this kind, such as varying sex ratios among proband and sibling groups, the choice of autism "control" group (normal, Down syndrome, DAMP or other handicap), the developmental course of the disorders studied and hence the problems associated with ascertainment at different ages, diagnostic criteria varying from one study to another, and the methods used for identifying cases. FIGURE 3 & 4 GOES HERE~~~~~~~~~~ My contention is that there is considerable support for the existence of a hereditary subgroup in autism, but that data about the nature of the genetic predisposition and about how much of the autism variance that can be explained by genetic factors are just not sufficient to suggest plausible answers. Based on our own studies, the "purely genetic" subgroup could encompass anything from 6% to 63% of the whole autism group. Perhaps the most surprising finding for some is the relatively high number of cases with an identifiable medical condition. Table 1 reviews most of the medical conditions that have been reported to be associated with autism by ourselves or other groups. Table 1. Associated medical conditions in autism documented in at least two studies Medical condition Important reference Fragile X syndrome Other sex chromosome anomalies Marker chromosome syndrome Other chromosome anomalies Tuberous sclerosis Neurofibromatosis Hypomelanosis of Ito Goldenhar syndrome Rett syndrome Moebius syndrome PKU Lactic acidosis Hypothyroidism Rubella embryopathy Herpes encephalitis CMV infection Duchenne muscular dystrophy Williams syndrome   Hagerman, 1989 Hagerman, 1989 Gillberg et al., 1991b Hagerman, 1989 Hunt & Dennis, 1987 Gillberg & Forsell, 1984 ‰kefeldt & Gillberg, 1991 Landgren, Gillberg & Str–mland, 1991 Coleman & Gillberg, 1985 Ornitz, Guthrie & Farley, 1977 Friedman, 1969 Coleman & Blass, 1985 Gillberg et al., 1992 Chess, Korn & Fernandez, 1971 Gillberg, 1986 Stubbs, 1978 Komoto, Udsui, Otsuki & Terao, 1984 Reiss, Feinstein, Rosenbaum, Borengasser-Caruso & Goldsmith, 1985 Some authors have reported considerably lower frequencies of associated medical conditions in autism than we have. This, as usual, will lead some critics to speculate that the Gothenburg material might be uncharacteristic in terms of the autism diagnosis at the symptom level and that other experienced authors would not have diagnosed autism in some of these cases. A recent blind inter-rater study from our centre based on case records judged independently by myself and Professor Michael Bohman (Steffenburg et al., 1989) suggests that this would be an unlikely explanation. If a full neurobiological work-up, including CAT-scan, EEG, chromosomal culture, auditory brainstem examination, ophthalmological, audiological and otological examinations as well as CSF-analysis had not been performed, only about 18% of the whole group (as compared with 37%) would have been shown to have a medical condition. This figure accords better with those of other authors. I therefore propose that the high rate of associated medical conditions in the Gothenburg material can be attributed to the comprehensiveness of the work-up rather than any particular bias in the material as such. However, regional differences might account for some of the variability. For instance, it appears that the fragile X syndrome might be considerably more common in western Sweden than in other regions (Wahlstr–m, 1991, personal communication). Since the fragile X syndrome sometimes underlies autism this would lead to an increased rate of autism associated with known medical conditions. In our material we have found a relatively high rate of autism associated with neurocutaneous disorders (altogether 8% of all children and adolescents receiving a diagnosis of autism in our centre had tuberous sclerosis, neurofibromatosis, or hypomelanosis of Ito (AhlsÈn, 1991; Gillberg & Forsell, 1984; ‰kefeldt & Gillberg, 1991). At least half of these cases would not have been diagnosed in connection with the diagno sis of autism had not the neurobiological work-up been exhaustive. Again, this underscores the need for a comprehensive medical and laboratory examination of all children who receive a diagnosis of autism. I have presented detailed evidence in favour of this view elsewhere (Gillberg, 1991b), and refer those interested to that publication. Here, I think it suffices to summarize the essential work-up of children with autism in a table (Table 2). Unfortunately, there are still those who maintain that this is a much too "ambitious" work-up, referring variously to the expensiveness of the investigations, the low yield, and the impossibility of performing many of the examinations in children with severe behavior problems. In respect of the expensiveness, the lifetime cost for the family and society in cases of autism is such that the cost for the medically and psychologically important work-up is negligible. Furthermore, why should the work-up of a lifelong handicap such as autism not be allowed to be as expensive as that of other severe handicaps, such as epilepsy and severe mental retardation? That the yield of a comprehensive work-up in autism is reasonable should not be a matter of dispute regardless of whether 12% - as recently reported by Ritvo et al. (1990) or 37% - as in our studies have associated medical conditions. That the behavioral problems exhibited by the child should be a major obstacle is clearly refuted by the fact that in our centre almost all cases with autism undergo a full medical work-up without major problems. However, admittedly, there can be problems for the laboratory staff to cope with the odd behaviors shown by children with autism. This is best dealt with by organizing the work-up in ways which will allow the same doctor or technician to manage all (or most) cases with autism, rather than having a whole host of different people manage only the occasional case. Also, in connection with general anesthesia which is often necessary in young children undergoing CAT-scan or MRI-examinations regardless of diagnosis it is often possible to perform several other examinations such as fundoscopy, blood sampling, etc. Table 2. Relevant laboratory analyses in all medium/low-functioning, and certain high-functioning, cases with autism and autistic-like conditions Analysis Finding Reference Chromosomal Fragile X q27.3 Hagerman, 1989 (including in a folic-acid-depleted medium) XYY Gillberg, Winnerg”rd & Wahlstr–m, 1984 Deletions, e.g. 15p Kerbeshian, Burd, Randall, Martsolf & Jalal, 1990   Marker chromosome Gillberg et al., 1991   Other Hagerman, 1989       CAT-Scan/MRI-Scan Tuberous sclerosis Gillberg, Steffenburg & Jakobsson, 1987b   Intrauterine infections Chess et al., 1971   Neurofibromatosis Gillberg & Forsell, 1984   Hypomelanosis of Ito ‰kefeldt & Gillberg, 1991   Other Tsai, 1989       CSF-protein* Progressive encephalopathy Wing & Gould, 1979       EEG Tuberous sclerosis Steffenburg, 1990   Subclinical epilepsy Gillberg & Schaumann, 1983   Epileptogenic discharge Gillberg & Schaumann, 1983   Other         Auditory Brainstem Brainstem dysfunction Coleman & Gillberg, 1985       Response Examination           Ophthalmologist Poor vision, fundus Steffenburg, 1990       Oto-laryngologist Poor hearing, anatomy Smith, Miller, Stewart, Walter & McConnell, 1988 (including hearing test)           Blood         phenylalanine high Friedman, 1969     uric acid high Coleman & Gillberg, 1985     lactic acid high Coleman & Blass, 1985     pyruvic acid high Coleman & Gillberg, 1985     herpes titer seroconversion Gillberg, 1986       24-hour urine         metabolic screen,         including muco-polysaccharidosis   Coleman & Gillberg, 1985     uric acid high Coleman & Gillberg, 1985     calcium low Coleman, 1989 Lumbar puncture for CSF-analysis is a safe and relatively non-traumatic procedure. In the Scandinavian countries, it is always considered in the work-up of young children with severe developmental disorders. If there is a nearby laboratory doing CSF-amino acids (phenylalanine in particular), CSF-monoamines and CSF-endorphins, these tests should be considered since the chlid has to have a lumbar puncture anyway (to exclude progressive encephalitis/encephalopathy). FIGURE 5 & 6 GOES HERE ~~~~~~~~~~ The "non-specific brain-dysfunctional group" with autism remains the largest in our studies, even though the rates of clearly familial and medical syndrome autism have increased steadily over the last 15 years. This group comprises cases with epilepsy, CAT-scan abnormalities without a known cause, and major abnormalities on EEG-, ABR- (auditory brainstem response), CSF- and fundus examinations. The demonstration of abnormal brain function on such examinations can sometimes be helpful in planning specific treatments (such as in the case of subclinical epilepsy). It is usually very helpful from the psychological point of view in helping families come to terms with the fact that their child is handicapped. The association of autism with even unspecific brain dysfunction makes the handicapping condition less mysterious in the eyes of most laymen (Figs. 5 and 6). There is sometimes a tendency among "experts" to disregard this aspect of the medical work-up. Even though no clear cause can be found, the mere fact that one can demonstrate any relationship at all between autism and brain dysfunction tends to make the autistic symptoms more "acceptable". Epilepsy in autism is very common, affecting 30-40% of all cases before c. 30 years of age (Olsson, Steffenburg & Gillberg, 1988; Volkmar & Nelson, 1990). Conversely, even though no systematic study has been published specifically examining autism symptoms in epilepsy, it appears that autistic-type behavior is very common in epilepsy. There seem to be at least two peaks for age of onset of seizures in autism, one in early childhood and another in the teenage period. The old notion of a single peak (in adolescence) may have been produced by old diagnostic practice based on Kanner's statements about neurological intactness in autism. Even though diagnostic criteria did not include this, it is quite possible that in older studies, children with early onset epilepsy and autism were withdrawn on the dubious basis of an implicit diagnosis hierarchy in which autism appeared at the bottom (after exclusion of all other indications of brain problems or sensory deficits). In such a hierarchy, early onset epilepsy would be a sign of brain dysfunction and a diagnosis of autism would not be considered. Adolescent onset epilepsy, it seemed, could not invalidate a diagnosis of typical Kanner autism which might have been made 10 years ago. Nevertheless, the opaque logic of diagnosing autism only in the absence of signs of brain dysfunction stands out as unreasonable in the light of this kind of confused practice in the field of autism and epilepsy. In the Gothenburg studies, complex partial seizures were very common in autism (Olsson et al., 1988; Gillberg, 1991c). It is well known that such seizures often signal underlying temporal lobe dysfunction. It is quite surprising the way the connection between autism and epilepsy, documented from Kanner's early writings on autism, has attracted extremely limited scientific attention. Many studies of autism are still launched on the notion that the underlying cause of autism will be found. Claims are often made that in scientific study autism must be "pure" in order for the cause to be revealed. "Pure" autism is then, often rather muddle-headedly, thought of as autism without mental retardation, autism without a known medical condition, autism without neurological dysfunction, etc. Does anybody believe that there is one cause for "pure" epilepsy? I do not think so. In discussing the boundaries of autism (see below) it will become even more apparent that the search for the clue to the riddle of autism is likely to be futile. This is certainly not to say that in research it may not be fruitful to subdivide the autism group on the basis of some reasonable method, such as IQ, sex, known familial loading or associated medical condition. All it implies is that there is no good scientific reason for saying that one subgroup is purer than another. Are there Neurochemical or Neurophysiological Abnormalities in Autism? I am not going to review here the extensive literature on neurochemical and neurophysiological studies of autism. The interested reader is referred to a recent excellent overview of autism neurochemistry by Cooke (1990) and to a forthcoming review of autism neurophysiology and neuroanatomy by Gillberg and Coleman. I shall instead discuss briefly some of our own recent findings, some of which are in press and others which are in the process of being written up. I shall do this against the background of some of the relatively well established findings and hypotheses concerning neurochemistry, neurophysiology and neuroanatomy in autism. Serotonin in the blood is raised in 25-35% of patients with autism, but it is not yet clear whether the association is with low IQ only or whether it is more "specific" to autism. In our own studies, 5-HIAA (the metabolite of serotonin) in the CSF has been significantly raised only in "non-nuclear" autism cases and only in the subgroup with moderate or severe mental retardation. Abnormalities of dopamine metabolism have been invoked in urine and CSF-studies of autism. In our own studies of CSF-monoamines, HVA has been consistently higher than in normal comparison cases and cases with non-autistic mental retardation. One study from our centre has suggested that the CSF-ratio of homovanillic acid to hydroxymethoxyphenylglycol might be increased in autism (Gillberg & Svennerholm, 1987). Endorphin hyperfunction has been suggested by results from a few groups, including our own (e.g. Gillberg, Terenius & L–nnerholm, 1985). In a more recent study, beta-endorphins tended to be low in the CSF of children with autism and Rett syndrome as compared with various comparison groups, most of whom were of adult age (Gillberg, Terenius, Hagberg, Witt-Engerstr–m & Eriksson, 1990a). There were no age trends within the autism or Rett groups. In another new study from our centre (Fig. 7), an astroglial cell protein, the so-called GFA (Glial Fibrillary Acidic) protein in the CSF was raised 3 times above the level in a normal age-matched group (AhlsÈn et al., 1992). Speculatively, this abnormality, if upheld by further research, could implicate increased turnover of glial cells, such as could occur in gliosis, which, in turn, might implicate unspecific brain damage (Aurell, 1989; Crols, Saerens, Noppe & Lowenthal, 1986). Another possible interpretation would be increased synapse turnover resulting from primary regulatory abnormalities or from unspecific damage to the brain. Synapses are rich in astroglial material and high GFA protein levels would result from synapse breakdown regardless of underlying cause. Similar high levels of CSF GFA protein have been reported in epilepsy (Hamberger et al., 1991). Also, the GFA protein content of adenoma sebaceum tissue in tuberous sclerosis tissue has been reported to be high. FIGURE 7 GOES HERE~~~~~~~~~~ Auditory brainstem response examinations in autism have often yielded abnormal results, but contradictory evidence also exists. CAT-scan abnormalities are relatively common but possibly unspecific. Recent interesting findings using CAT- or MRI-techniques have suggested brainstem, cortical developmental and cerebellar abnormalities in autism. Abnormalities of the Purkinje cell fibres have also been demonstrated. In the largest-scale study of auditory brainstem responses to date in autism, we have again found that about one third of all cases with infantile autism have either a prolongation of the brainstem transmission time, a pathological increase of internaural time difference or a pathological response variability, or combinations of these (Rosenhall, AhlsÈn & Gillberg, 1991). Children with these pathological findings are likely to have clear neurophysiological or neurostructural abnormalities at the brainstem level. Clinically they are slightly to moderately hypotonic, show extremes of odd sound reactions, often dislike music and have relatively poor language comprehension skills. In a study of 27 high-functioning cases with autism, single positron emission computed tomography technique (SPECT) revealed hypoperfusion of the temporal lobes (Gillberg, Gillberg, Bjure & Uvebrandt, 1991). Many of the cases in this study had epilepsy. Normal comparison groups without any suspicion of neurological/psychiatric disorder have not been examined. For these reasons, the conclusions must be guarded. Nevertheless, there was the same type and degree of hypoperfusion in the autism cases with and without epilepsy, indicating that epilepsy per se would be unlikely to account for the findings. There is no distinctive trend or pattern of overlap in these "neurostudies" that we have been able to find at the present stage. Nevertheless, all the findings are compatible with an overall model for the development of autistic symptoms of dysfunction in brainstem/cerebellar/vestibular areas, temporal lobes and prefrontal structures. The biochemical abnormalities that have been documented might well reflect abnormal neural activity in these areas. Dysfunction in the areas mentioned could result in symptoms of autism, and, depending on exactly which of these areas mentioned were dysfunctional and on whether any additional area was involved, other symptoms also (mental retardation, epilepsy, hypotonia, etc.). Some of the findings (for instance those concerning endorphin abnormalities) have inspired specific therapeutic trials, the results of which are not yet available. Is Autism a Highly Specific, Easy-to-Diagnose Condition? Most authors now agree that autism as a distinct disease entity does not exist (Andersson et al., 1989). The reasons for this conclusion are too many to list here, but some of the most important are (1) the fact that a number of different medical conditions can cause a behavioral syndrome which is indistinguishable from "Kanner autism" (Steffenburg, 1991), (2) the finding that children with and without demonstrated neurological signs do not differ in respect of autistic symptom profile (Garreau, Barthelemy, Sauvage, Leddet & Lord, 1984) and (3) the observation that "Kanner autism" overlaps both at the pathogenetic and symptomatic levels with a number of other conditions showing the triad of social, communication and behavior impairments (Wing, 1989a,b). Even with a diagnosis of classic autism and hence a core syndrome as described by Kanner, symptoms, behaviors and personality traits can vary considerably. Autism cases with the fragile X syndrome are behaviorally different than, for instance, most autism cases with Rett syndrome (Gillberg, 1991b). Once the underlying etiology is disclosed, differentiating features at the behavioral level can be discerned. Without knowing the etiology, the "extra-autism" symptomatology that may be typical of a particular medical condition cannot be extracted. Rett syndrome has provided a striking illustration of the confusion which arises if researchers continue to insist on the purity of the autistic syndrome. One study from our centre revealed that of all Rett syndrome cases, 80% had first been given a diagnosis of pure autism (half of the cases) or autistic-like conditions, by doctors other than ourselves (Witt-Engerstr–m & Gillberg, 1987). Also, we have published a number of case histories meeting criteria for infantile autism, autistic disorder or autistic features on the one hand, and all or almost all of the criteria for Rett syndrome on the other. How should they be diagnosed? I have no difficulty in saying that if they meet criteria for both disorders, they should be diagnosed as autism and Rett syndrome. Other authors have, for reasons unclear to me, argued differently that if Rett syndrome is present this should be the only diagnosis regardless of whether criteria for autism are met or not. If a child has epilepsy and the medical work-up reveals an underlying brain tumor, it would not be appropriate to say that the child has a brain tumor and not epilepsy. Autism was first described in children of some years of age. The current diagnostic criteria, to a considerable extent, are based on the clinical picture of autism as presented in 3-6 year old children (boys in particular). Autism before age 3 years may have a different symptom profile. In a recent study in our centre of 28 children referred before age 3 years with a preliminary diagnosis of autism or ""childhood psychosis" (Gillberg et al., 1990b), 20 (possibly 21) were found to meet the criteria for autistic disorder after age 3 years. This means that even at a specialized centre, the number of false-positives identified in infancy is considerable. Some turn out to have mental retardation without autistic features and others to have more benign attention disorders. Nevertheless, the majority of cases referred in infancy turn out to be true positive cases after several years' follow-up. In the case of autism associated with other medical syndromes I can see no problem in making multiple diagnoses. The scientific community seems to agree that autism, mental retardation, and epilepsy can all be diagnosed in one and the same patient. It should be equally easy to agree that autism can also be diagnosed in conjunction with the fragile X syndrome, Rett syndrome, hypothyroidism, etc. Several authors have commented on the overlap between autism and Asperger syndrome (e.g. Frith, 1991). Some children first receive a diagnosis of autism but are later thought better to fit the Asperger syndrome stereotype (Wing, 1989a, b). Autism and Asperger syndrome sometimes occur within the same family (Bowman, 1988; Gillberg, 1989; Steffenburg, 1991; Gillberg, 1991a). A number of reports in the literature on the co-occurrence of Asperger and Tourette syndrome have implied the possibility of some kind of a connection between these syndromes. There are also occasional mentions of families with autism and Tourette in different individuals as well as of Tourette and autism coexisting in the same patient (Comings & Comings, 1991). Asperger syndrome is currently most often conceptualized as a variant of high-functioning autism, but there is no complete consensus in the field. The observations that expressive language at the formal level is usually better developed in Asperger syndrome and that motor skills are relatively better in autism have led to speculation that they may not exist on a continuum of "autistic conditions". Asperger syndrome shows considerable overlap with semantic-pragmatic disorder (Bishop, 1989). Semantic and pragmatic problems are also common in children with deficits in attention, motor control and perception (so called DAMP (Gillberg 1983)). Children with DAMP quite often show "triad" symptoms and some have marked autistic-type symptoms. In reanalyzing the data from one of our population studies of DAMP, an important minority of severely affected children were shown to fulfil research criteria for Asperger syndrome both at age 7, 10 and 13 years (Gillberg & Gillberg, 1989). Thus, quite often children with severe DAMP (constituting 1.2% of all seven-year olds according to Gillberg et al., 1982) exhibit social impairments, semantic pragmatic problems and restricted-stereotyped-repetitive-obsessive behavior patterns of a milder variant, but of the same type as those encountered in autism (Gillberg, 1983) (Fig. 8). More than half of the children with DAMP have been observed to have speech and language delay already at age 4 years (Rasmussen, Gillberg, Waldenstr–m & Svenson, 1983). Some of the children with DAMP (about one third according to our studies) also meet the criteria for generalized or pervasive hyperkinesis (Gillberg & Gillberg, 1988) and may be presented clinically as suffering from "the hyperkinetic syndrome". The hyperkinetic children may or may not show autistic features. Many of them develop severe dyslexia. Overall, there are no clear boundaries between autism, Asperger syndrome, semantic-pragmatic disorders and DAMP (often with dyslexia). FIGURE 8 GOES HERE~~~~~~~~~~~~ In 1985, we reported the concurrence of autism and anorexia nervosa in certain families (Gillberg, 1985). Since then, scattered reports about such a connection have appeared in the literature. There is at least one reported case of anorexia nervosa in autism (Stiver & Dobbins, 1980). Comings and Comings (1991) have described an extended family with different individuals affected by Tourette syndrome, obsessive-compulsive disorder, autism and anorexia nervosa. RÂstam (1990) (see also RÂstam & Gillberg, 1991) in our centre, performed a mixed population-based and clinic referral study of anorexia nervosa. The population-based part of the study is clearly the most comprehensive prevalence study of anorexia nervosa ever performed. Every child in a population of c. 4,300 was physically examined and a number of other screening measures were included. It was shown that under age 18 years, the anorexia prevalence in girls was 1.1% and in boys 0.1%. In the "mixed" study of 51 anorexia cases and 51 sex-, age- and school-matched cases, 1 of the 3 boys with anorexia had Asperger syndrome and 3 of the 48 girls had autistic-like conditions. Tourette syndrome, tics and obsessive-compulsive personality disorders (all diagnosed on the basis of de-identified case records by a blinded experienced clinician) were extremely common in the anorexia nervosa and very rare in the comparison groups (Fig. 9). The DSM-III-R diagnostic criteria for obsessive-compulsive personality disorder (not for obsessive-compulsive disorder) in several instances overlap with those of autistic disorder. An interesting, originally unexpected find in the study of anorexia nervosa was that, as compared with the age-, sex- and school-matched group, the teenagers with anorexia showed dysdiadochokinesis, a finding often reported in DAMP (Gillberg, Gillberg & Groth, 1989) and commonly encountered in Asperger syndrome also. Dysdiadochokinesis in regarded as one important neurological sign of cerebellar dysfunction. Cerebellar dysfunction has been reported in several studies of autism in recent years (Ritvo et al., 1986; Courchesne, Yeung-Courchesne, Press, Hesslink & Jernigan, 1988). FIGURE 9 GOES HERE~~~~~~~~~~~~ In summary, it seems that severe and typical nuclear Kanner autism may be relatively easy to diagnose (even though at meticulous work-up even this group will be shown to consist of at least the four subgroups described in Fig. 2), but that the boundaries are far from clear in respect of "other triad conditions", autistic-like conditions, so-called pervasive developmental disorders, Asperger syndrome, and severe DAMP. Even conditions hitherto not considered in relation to autism, such as anorexia nervosa, obsessive-compulsive personality disorder, and Tourette syndrome, may show some (even considerable) overlap with autism (Fig 10, see below).

Course and Outcome
Autism associated with severe mental retardation diagnosed already before age 5 years carries a gloomy prognosis in respect of psychosocial adaptation. Also, a recent review from our centre (Gillberg, 1991c) indicated that mortality is likely to be increased in this group. Of Leo Kanner's first 11 cases, 1 died before age 30 years. Even though there is no direct proof yet, increased mortality might well be associated with severe underlying medical problems such as tuberous sclerosis.

Autism associated with mild mental retardation or near average intelligence levels has a much more variable prognosis. Unfortunately, even in this group, about half do poorly psychosocially in adult age and do not hold jobs or lead independent lives in other ways. However, a proportion of cases in this group has a relatively favorable prognosis and will be able to be self-supporting as adults. Only a few percent are likely ever to be married or to engage in marriage-like relationships.

In the very high functioning group of cases with autism or Asperger syndrome, the overall prognosis is much better. Oddities of social style, communication and interests are likely to remain, but the majority from this group hold down jobs and it seems that a large proportion get married and have children. Systematic prospective long-term follow-up studies of these cases so far have only been published in connection with reports on the whole of the "classic autism" group, which means that only 15-30% were relatively high functioning. Therefore there is much less detail with regard to the outcome picture in the high-functioning group than in the groups with concomitant mental retardation.

Wing (1983) has suggested that the overall clinical picture in autism changes across time and that by adult age, three broad groups can be identified. One remains "autistic". Another group is passive and friendly and will exhibit insistence on sameness and other typical autism symptoms only under stress. The third group is best described as active but odd. Our own studies, both those which have included long-term follow-up of population cases of autism diagnosed in childhood, and the cross-sectional studies of adolescents and adults in a mental handicap hostel have provided considerable support for Wing's subgrouping.

Rutter (1970) was one of the first to suggest that pubertal deterioration might occur in a sizable minority of children with autism. He did this on the basis of results obtained in a clinic referral group of autism and autistic-like conditions. Our group, on the basis of population screening studies of autism found a high rate of symptom aggravation in adolescence. A full 22% of the whole group of children followed through age 16-23 years showed clear signs of deteriorating in adolescence (Gillberg & Schaumann, 1981; Gillberg & Steffenburg, 1987). Follow-up at age 23-30 years, currently in progress, reveals that most cases taking a deteriorating course in adolescence never return to preadolescent levels of language, communication and social functioning.

Children with autistic-like conditions have a very variable prognosis, ranging from very poor psychosocial outcome in the small subgroup with so-called childhood disintegrative disorders to fair or even good in the much larger subgroup with DAMP, normal intelligence and autistic features.

Prognosis in the whole field of autistic syndromes can be more precise if we subgroup according to other known correlates in each individual case. For instance, a child with severe autism, severe mental retardation and severe epilepsy, all with onset before age 2 years, is likely to have a generally very poor psychosocial outcome. Tuberous sclerosis is a quite common cause for such a constellation of symptoms. If tuberous sclerosis constitutes the main underlying etiologic factor, then other specific problems associated with that diagnosis can be predicted with varying degree of precision. If, on the other hand, a boy of three with autism, normal nonverbal intelligence and echolalia, and has a well-functioning father with an Asperger type personality (implying a form of familial autism in the boy), outcome is likely to be relatively good.

This is not the place to extensively review the theory of mind theory. Suffice it to say that this type of thinking about autism has opened a new window on the whole field, both clinically and theoretically. Parents and other caregivers agree that here is a psychological theory that not only explains so many of the behaviors shown by children with autism, but is also able to predict their behavior, social interaction and communication in a given setting. The theory of mind theory is neutral in respect of affect in autism. It does not predict that affects are primarily dysfunctional in autism, nor does it predict that affect does not become dysfunctional in autism.

It has been my clinical experience over the years that children with autism are sensitive to affective change, that they can "feel" (unreflectingly) if an interaction is full of strong emotion (even if not shown "on the outside"), that they themselves can be happy, fearful, glad, and angry, but that they cannot make sense of the feelings of self or others. In this sense, it seems clear to me that autism cannot be primarily a basic disturbance of emotions. However, people with autism have problems, usually throughout life, when it comes to putting emotions into a cognitive perspective. Compassion can be said to be a cognitive effect and people with autism have problems in this field.

I think it makes sense to regard the compassion deficits encountered in autism as a secondary consequence of the theory of mind deficit. If you do not understand that other people have, as it were, inner worlds, how can you be expected to show compassion or sympathy.

Empathy could be defined as the ability to conceptualize other people's inner worlds and to reflect on their thoughts and feelings. Good empathy skills, of course, require a well-developed theory of mind. In fact, one way of looking at it is as having a theory of mind being synonymous with having empathy. Empathy and sympathy would not be synonymous, however. Most people know that some people have good empathy skills and yet do not demonstrate a shred of sympathy. Psychiatrists can be empathic without being sympathetic. Some psychopaths, almost by definition, have good empathy skills or, if you will, an excellent theory of other people's minds and yet show very little, or indeed no, sympathy. In order to be able to deceive other people one of the fields in which psychopaths excel you must have good empathy skills/a well-developed theory of mind. People with autism, on the other hand, have enormous difficulty in deceiving other people. Except when it comes to getting their way in the field of obsessional interests, this holds for people with Asperger syndrome also.

In the foregoing, I have made a case for the overlap of autism and Asperger with other clinical syndromes. Could it be that obsessive-compulsive personality disorders, some cases of anorexia nervosa and Tourette syndrome, and perhaps also those with paranoid disorder share a common problem with autism and Asperger syndrome in that they all have moderate to severe problems in the field of empathy? Is their development of a theory of mind delayed and/or deviant? Is autism perhaps the most severe form of empathy disorder? Is anorexia nervosa in girls in some cases the equivalent of Asperger syndrome in boys? Have we relied so much on the male prototype for autism that we have failed to understand that the extreme obsessiveness, limited capacity for social interaction and generally rather poor psychosocial prognosis (Herzog, Keller & Lavori, 1988) in girls with anorexia nervosa could signal a close relationship between the two at the psychological (and perhaps biological) level? Could so-called "alexithymia" (the inability to speak about feelings), said to be characteristic of anorexia nervosa, be a sign of theory of mind problems reflecting not just the inability to communicate about feelings but about mental states generally? The clinical leads (Bruch, 1977; RÂstam, 1990) in the field of anorexia nervosa are admittedly limited, but, I think, sufficient to warrant testing of this hypothesis. One way of exploring a possible relationship would be taking detailed developmental histories with a particular view to finding autism-associated symptoms in a series of anorexia cases and controls. We are now planning to do this by having a research assistant blind to group status interview the mothers of all cases and controls in the RÂstam (1990) study, using parts of the HBS (Handicaps, Behaviors and Skills; Wing & Gould, 1978). We also intend to subject cases and controls to higher-order theory of mind tasks to test the hypothesis that there may be associated empathy defi cits in certain cases of anorexia nervosa.

Taking the issue of overlap further, obsessive-compulsive personality disorder is typical of many cases of anorexia nervosa and of Tourette syndrome and share some important clinical features with Asperger syndrome and autism. Some people with Asperger syndrome grow up to develop paranoid tendencies. Obsessive-compulsive personality disorders, Tourette syndrome and paranoid disorders could also be further explored with regard to a possible autism relationship by studies similar to those suggested for anorexia nervosa.

Figure 10 aims to illustrate overlap of clinical syndromes in the field. The inserted square comprises the majority of the empathy disorders according to the model discussed. This overlap model could, by some, be taken to indicate an extension of the concept of autism to take in just about everything. Such concern is quite legitimate. However, my purpose is not to impossibly widen the bounds of Kanner's specific variant of autism, but rather to examine the various clinical manifestations of impairments of social interaction. Many, perhaps most, conditions in psychiatry are on a continuum, just like many physical disorders (hypertension, diabetes, etc.). A proper aim of research is to explore the boundaries as well as to try to find sub-groups that have some cohesion within the continuum.

FIGURE 10 GOES HERE!~~~~~~~~~~~

One way of looking at possible underlying mechanisms for the development of empathy disorders is illustrated in Fig. 11. Falling in the lowermost portion of the normal distribution with respect to empathy skills, having a more specific genetic trait for poor empathy skills, or sustaining damage to brain areas involved in subserving empathy functions might all result in a similar clinical picture. This line of reasoning would be very similar to that generally accepted when trying to conceptualize background mechanisms in mental retardation.

FIGURE 11 GOES HERE~~~~~~~~~~~~~~~`

Concluding Remarks
Can a concept of autism and other disorders of empathy such as that presented in this paper be of any use? I think it can, and in a number of different ways. First of all, it is possible to test out within the various clinical syndrome populations whether they, as contrasted with other groups, show specific theory of mind/empathy deficits. Second, knowledge about autism might help in the study of obsessive-compulsive personality disorder and anorexia nervosa and vice versa. Third, finding new boundaries within this conceptual framework might well lead to the establishment of more useful clinical diagnostic criteria across these groups. Finally, but by no means least, understanding more about the underlying empathy deficit has already contributed to the clinical management of people with autism. It does not seem unlikely that, if a disorder of empathy is also present in some cases of anorexia nervosa, the management of these patients in everyday practice will have to change considerably paying attention to the handicapping nature of the underlying disorder and not only to the rather more "superficial" eating disorder.

In summary, unlike those who "insist on Kanner autism", I think that we would do well to take a more balanced view and pursue some of the lines of research which present themselves if autism is viewed as a subclass of empathy disorders. Instead of putting all the money on the splitter approach, I suggest that lumping may again prove useful as a first step in the process of subgrouping on a more rational basis than that of "pure Kanner autism".

Acknowledgements
I am very grateful to Uta Frith and Lorna Wing for valuable comments on the first draft of this paper. I am also indebted to Maria RÂstam and I. Carina Gillberg for taking an active part in discussions which have led to the formulation of some of the issues raised at the end of the paper.

References
AhlsÈn, G. (1991). Tuberous sclerosis in western Sweden. Epidemiological aspects (in press).

AhlsÈn, G., Rosengren, L., Palm, A., Belfrage, M. Haglid, L., Hamberger, A. & Gillberg, C. (1992). GFA protein in the CSF of children and adolescents with autism and other neuropsychiatric disorders (submitted).

‰kefeldt, A. & Gillberg, C. (1991). Autism and hypomelanosis of Ito syndrome. Developmental Medicine and Child Neurology, 33, 737-743.

American Psychiatric Association (1980). Diagnostic and statistical manual of mental disorders (3^rd edn.). (DSM-III). Washington DC: American Psychiatric Association.

American Psychiatric Association (1987). Diagnostic and statistical manual of mental disorders (3^rd edn. revised). (DSM-III-R). Washington DC: American Psychiatric Association.

Andersson, L. & Wadensj–, K. (1981). Tidigt psykotiska barn I Malm–hus l”n [Early Childhood Psychosis in Malm–hus l”n: a Descriptive Study] (in Swedish). Lund: Socialavdelningen.

Andersson, L., Bohman, M., Campbell, M. et al. (1989). The state of the art. In C. Gillberg (Ed.), Diagnosis and treatment of autism (pp. 441-442). New York: Plenum Press.

Asperger, H. (1944). Die "autistischen" Psychopathen im Kindesalter. Archiv f¸r Psychiatrie und Nervenkrankheiten, 117, 73-136.

Aurell, A. (1989). An ELISA for determination of the S-100 protein and the GFA protein in the CSF. The use of astroglial proteins as markers of ischaemic brain damage. G–teborg: Institute of Neurobiology, University of G–teborg.

Baron-Cohen, S. (1990). Autism: A specific cognitive disorder of "mind-blindness". International Review of Psychiatry, 2, 81-90.

Bishop, D.V.M. (1989). Autism, Asperger's syndrome and semantic-pragmatic disorders. Where are the boundaries? British Journal of Disorders of Communication, 24, 107-121.

Bohman, M., Bohman, I.L., Bj–rk, P. & Sj–holm, E. (1983). Childhood psychosis in a northern Swedish county: some preliminary findings from an epidemiological survey. In M.H. Schmidt & H. Remschmidt (Eds.), Epidemiological approaches in child psychiatry. Stuttgart: Georg Theme.

Bolton, P. & Rutter, M. (1990). Genetic influences in autism. International Review of Psychiatry, 2, 67-80.

Bowman, E.P. (1988). Asperger syndrome and autism: the case for a connection. British Journal of Psychiatry, 152, 377-382.

Brask, B.H. (1970). A prevalence investigation of childhood psychosis. Paper given at the 16^th Scandinavian Congress of Psychiatry.

Bruch, H. (1977). Psychological antecedents of anorexia nervosa. In R. Vigersky (Ed.), Anorexia nervosa (pp. 1-10). New York: Raven.

Bryson, S.E., Clark, B.S. & Smith, IM (1988). First report of a Canadian epidemiological study of autistic syndromes. Journal of Child Psychology and Psychiatry, 29, 433-445.

Chess, S., Korn, S.J. & Fernandez, P.B. (1971). Psychiatric disorders of children with congenital rubella. New York: Brunner/Mazel.

Cialdella, P. & Mamelle, N. (1989). An epidemiological study of infantile autism in a French department (Rhone): a research note. Journal of Child Psychology and Psychiatry, 30, 165-175.

Coleman, M. (1989). Autism: non-drug biological treatments. In C. Gillberg (Ed.), Diagnosis and treatment of autism (pp. 219-235). New York: Plenum Press.

Coleman, M. & Blass, J.P. (1985). Autism and lactic acidosis. Journal of Autism and Developmental Disorders, 15, 1-8.

Coleman, M. & Gillberg, C. (1985). The biology of the autistic syndromes. New York: Praeger.

Comings, D.E. & Comings, B.G. (1991). Clinical and genetic relationships between autism-pervasive developmental disorder and Tourette syndrome: a study of 19 cases. American Journal of Medical Genetics, 39, 180-191.

Cooke, E.H. (1990). Autism: review of neurochemical investigation. Synapse, 6, 292-308.

Courchesne, E., Yeung-Courchesne, R., Press, G.A., Hesslink, J.R. & Jernigan, T.L. (1988). Hypoplasia of cerebellar vermal lobules VI and VII in autism. New England Journal of Medicine, 318, 1349-1354.

Crols, R., Saerens, J., Noppe, M. & Lowenthal, A. (1986). Increased GFAp levels in CSF as a marker of organicity in patients with Alzheimer's disease and other types of irreversible chronic organic brain syndrome. Journal of Neurology, 223, 157-160.

Folstein, S. & Rutter, M. (1977). Infantile autism: a genetic study of 21 twin pairs. Journal of Child Psychology and Psychiatry, 18, 297-321.

Friedman, E. (1969). The autistic syndrome and phenylketonuria. Schizophrenia, 1, 249-261.

Frith, U. (1989). Autism. Explaining the enigma. Oxford: Blackwell.

Frith, U. (1991). Autism and Asperger syndrome. Cambridge: Cambridge University Press.

Garreau, B., Barthelemy, C., Sauvage, D., Leddet, I. & Lelord, G. (1984). A comparison of autistic syndromes with and without associated neurological problems. Journal of Autism and Developmental Disorders, 14, 105-111.

Gillberg, C. (1983). Perceptual, motor and attentional deficits in Swedish primary school children. Some child psychiatric aspects. Journal of Child Psychology and Psychiatry, 25, 377-403.

Gillberg, C. (1984). Infantile autism and other childhood psychoses in a Swedish urban region. Epidemiological aspects. Journal of Child Psychology and Psychiatry, 25, 35-43.

Gillberg, C. (1985). Autism and anorexia nervosa: related conditions? Nordisk Psykiatrisk Tidskrift, 39, 307-312.

Gillberg, C. (1986). Onset at age 14 of a typical autistic syndrome. A case report of a girl with herpes encephalitis. Journal of Autism and Developmental Disorders, 16, 369-375.

Gillberg, C. (1989). Asperger syndrome in 23 Swedish children. Developmental Medicine and Child Neurology, 31, 520-531.

Gillberg, C. (1990). Do children with autism have March birthdays? Acta Psychiatrica Scandinavica, 81, 152-156.

Gillberg, C. (1991a). Clinical and neurobiological findings in six family studies of Asperger syndrome. In U. Frith (Ed.), Autism and Asperger syndrome (pp. 122-146). Cambridge: Cambridge University Press.

Gillberg, C. (1991b). The medical workup of persons who suffer from autism, Asperger syndrome and other autisticlike conditions. Brain Dysfunction (in press).

Gillberg, C. (1991c). The treatment of epilepsy in autism. Journal of Autism and Developmental Disorders, 21, 61-77.

Gillberg, C. (1991d). Subgroups in autism: are there behavioral phenotypes typical of underlying medical conditions? Journal of Mental Deficiency Research (in press).

Gillberg, C. (1991e). Outcome in autism and autisticlike conditions. Journal of the American Academy of Child and Adolescent Psychiatry, 30, 375-382.

Gillberg, C., Ehlers, S., Schaumann, H., Jakobsson, G., Dahlgren, S.O., Lindblom, R., BÂgenholm, A., Tjuus, T. & Blidner, E. (1990b). Autism under age 3 years: a clinical study of 28 cases referred for autistic symptoms in infancy. Journal of Child Psychology and Psychiatry, 31, 921-934.

Gillberg, C. & Forsell, C. (1984). Childhood psychosis and neurofibromatosis more than a coincidence. Journal of Autism and Developmental Disorders, 13, 1-8.

Gillberg, I.C. & Gillberg, C. (1988). Generalized hyperkinesis: follow-up study from age 7 to 13 years. Journal of the American Academy of Child and Adolescent Psychiatry, 27, 55-59.

Gillberg, IC & Gillberg, C. (1989). Asperger syndrome some epidemiological considerations: a research note. Journal of Child Psychology and Psychiatry, 30, 631-638.

Gillberg, IC, Gillberg, C., Bjure, J. & Uvebrandt, P. (1991). SPECT findings in 32 children and adolescents with autism and autisticlike conditions (in press).

Gillberg, IC, Gillberg, C. & Groth, J. (1989). Children with preschool minor neurodevelopmental disorders. V: Neurodevelopmental profiles at age 13. Developmental Medicine and Child Neurology, 31, 14-24.

Gillberg, IC, Gillberg, C. & Kopp, S. (1992). Autism and hypothyroidism. The case for a connection. Journal of Child Psychology and Psychiatry, 33.

Gillberg, C., Persson, E., Grufman, M. & ThemnÈr, U. (1986). Psychiatric disorders in mildly and severely mentally retarded urban children and adolescents: epidemiological aspects. British Journal of Psychiatry, 149, 68-74.

Gillberg, C., Rasmussen, P., Carlstr–m, G., Svenson, B. & Waldenstr–m, E. (1982). Perceptual, motor and attentional deficits in six-year-old children. Epidemiological aspects. Journal of Child Psychology and Psychiatry, 23, 131-144.

Gillberg, C. & Schaumann, H. (1981). Infantile autism and puberty. Journal of Autism and Developmental Disorders, 11, 365-371.

Gillberg, C. & Schaumann, H. (1982). Social class and infantile autism. Journal of Autism and Developmental Disorders, 12, 223-228.

Gillberg, C. & Schaumann, H. (1983). Epilepsy presenting as in infantile autism? Two case studies. Neuropediatrics, 14, 206-212.

Gillberg, C. & Steffenburg, S. (1987). Outcome and prognostic factors in autism and similar conditions: a population-based study of 46 cases followed through puberty. Journal of Autism and Developmental Disorders, 17, 273-287.

Gillberg, C., Steffenburg, S., B–rjesson, B. & Andersson, L. (1987a). Infantile autism in children of immigrant parents. A population-based study. British Journal of Psychiatry, 150, 856-858.

Gillberg, C., Steffenburg, S., & Jakobsson, G. (1987b). Neurobiological findings in 20 relatively gifted children with Kanner type autism or Asperger syndrome. Developmental Medicine and Child Neurology, 29, 641-649.

Gillberg, C., Steffenburg, S. & Schaumann, H. (1991a). Is autism more common now than ten years ago? British Journal of Psychiatry, 158, 403-409.

Gillberg, C., Steffenburg, S., Wahlstr–m, J., Gillberg, IC, Sj–stedt, A., Martinsson, T., Liedgren, S. & Eeg-Olofsson, O. (1991b). Autism associated with marker chromosome. Six case studies. Journal of the American Academy of Child and Adolescent Psychiatry, 30, 489-494.

Gillberg, C. & Svennerholm, L. (1987). CSF-monoamines in autistic syndromes and other pervasive development disorders of early childhood. British Journal of Psychiatry, 151, 89-94.

Gillberg, C., Terenius, L., Hagberg, B., Witt-Engerstr–m, I. & Eriksson, I. (1990a). CSF beta endorphins in childhood neuropsychiatric disorders. Brain and Development, 12, 88-92.

Gillberg, C., Terenius, L. & L–nnerholm, G. (1985). Endorphin activity in childhood psychosis. Spinal fluid levels in 24 cases. Archives of General Psychiatry, 42, 780-783.

Gillberg, C., WinnergÂrd, I. & Wahlstr–m, J. (1984). The sex chromosomes one key to autism? An XYY case of infantile autism. Applied Research in Mental Retardation, 5, 353-360.

Hagerman, R.J. (1989). Chromosomes, genes and autism. In C. Gillberg (Ed.), Diagnosis and treatment of autism (pp. 105-132). New York: Plenum Press.

Hamberger, A., Nordborg, C., Nystr–m, B., Silfvenius, H., Wang, S. & Haglid, K.G. (1991). Neuronal and glial marker proteins in human epileptiform cortex-biochemical correlation to dysplasia, gliosis and astrocytoma infiltration. Epilepsy Research (submitted).

Herzog, D.B., Keller, M.B. & Lavora, P.W. (1988). Outcome in anorexia nervosa and bulima nervosa. A review of the literature. Journal of Nervous and Mental Disease, 176, 131-143.

Hunt, A. & Dennis, J. (1987). Psychiatric disorder among children with tuberous sclerosis. Developmental Medicine and Child Neurology, 29, 190-198.

Kanner, L. (1943). Autistic disturbances of affective contact. Nervous Child, 2, 217-250.

Kanner, L. (1971). Follow-up study of eleven children originally reported in 1943. Journal of Autism and Childhood Schizophrenia, 1, 119-145.

Kanner, L. & Eisenberg, L. (1956). Early infantile autism: 1943-1955. American Journal of Orthopsychiatry, 26, 55-65.

Kerbeshian, J., Burd, L., Randall, T., Martsolf, J. & Jalal, S. (1990). Autism, profound mental retardation and atypical bipolar disorder in a 33-year-old female with a deletion of 15q12. Journal of Mental Deficiency Research, 34, 205-210.

Komoto, J., Udsui, S., Otsuki, S. & Terao, A. (1984). Infantile autism and Duchenne muscular dystrophy. Journal of Autism and Developmental Disorders, 14, 191-195.

Kopp, S. & Gillberg, C. (1991). Girls with social deficits and learning problems. Autism, Asperger syndrome or variants of these syndromes? European Child and Adolescent Psychiatry (in press).

Landgren, M., Gillberg, C. & Str–mland, K. (1992). Autistic symptoms in Goldenhar syndrome (submitted).

L–gdahl, K. (1989). The prevalence of autism in a Swedish county. Report to S–rmland County Authorities (in Swedish).

Lotter, V. (1966). Epidemiology of autistic conditions in young children I. Prevalence. Social Psychiatry, 1, 124-137.

Olsson, I., Steffenburg, S. & Gillberg, C. (1988). Epilepsy in autism and autistic-like conditions: a population-based study. Archives of Neurology, 45, 666-668.

Ornitz, E.M., Guthrie, D. & Farley, A.J. (1977). The early development of autistic children. Journal of Autism and Childhood Schizophrenia, 7, 207-229.

Rasmussen, P., Gillberg, C., Waldenstr–m, E. & Svenson, B. (1983). Perceptual motor and attentional deficits in seven-year-old children: neurological and neurodevelopmental aspects. Developmental Medicine and Child Neurology, 25, 315-333.

Reiss, A.L., Feinstein, C., Rosenbaum, K.N., Borengasser-Caruso, M.A. & Goldsmith, B.M. (1985). Autism associated with Williams syndrome. Journal of Pediatrics, 106, 247-249.

Rett, A. (1966). Ðber ein cerebral-atrophisches Syndrom bei Hyperammonaemie. Wien: Br¸der Hollinek.

Ritvo, E.R., Freeman, B.J., Pingree, M.S., Mason-Brothers, A., Jorde, L., Jenson, R., McMahon, W., Petersen, B., Mo, A. & Ritvo, A. (1989). The UCLA-University of Utah Epidemiologic Survey of autism: prevalence. American Journal of Psychiatry, 146, 194-199.

Ritvo, ER, Freeman, B.J., Scheibel, A.B., Duong, T., Robinson, H., Guthrie, D. & Ritvo, A. (1986). Lower purkinje cell counts in the cerebella of four autistic subjects: initial findings of the UCLA-NSAC autopsy research report. American Journal of Psychiatry, 143, 862-866.

Ritvo, ER, Mason-Brothers, A., Freeman, B.J., Pingree, C., Jenson, W.R., McMahon, W.M., Petersen, P.B., Jorde, L.B., MO, A. & Ritvo, A. (1990). The UCLA-University of Utah Epidemiologic Survey of autism: the etiologic role of rare diseases. American Journal of Psychiatry, 147, 1614-1621.

Ritvo, ER, Yuwiler, A., Geller, E., Ornitz, E.M., Saeger, K. & Plotkin, S. (1970). Increased blood serotonin and platelets in early infantile autism. Archives of General Psychiatry, 23, 566-572.

Rosenhall, U., AhslÈn, G. & Gillberg, C. (1991). Auditory brainstem responses in a very large sample of children and adolescents with autism and autisticlike conditions (in press).

Rutter, M. (1970). Autistic children. Infancy to adulthood. Seminars in Psychiatry, 2, 435-450.

Rutter, M. (1978). Diagnosis and definition in childhood autism. Journal of Autism and Childhood Schizophrenia, 8, 139-161.

RÂstam, M. (1990). Anorexia nervosa in Swedish urban teenagers. G–teborg: G–teborg University MD Thesis.

RÂstam, M., Garton, M. & Gillberg, C. (1989). Anorexia nervosa in a Swedish urban region: a population-based study. British Journal of Psychiatry, 155, 642-646.

RÂstam, M. & Gillberg, C. (1991). The family background in anorexia nervosa: a population-based study. Journal of the American Academy of Child and Adolescent Psychiatry, 30, 283-289.

Smith, D.E.P., Miller, S.D., Stewart, M., Walter, TL & McConnell, J.V. (1988). Conductive hearing loss in autistic, learning-disabled and normal children. Journal of Autism and Developmental Disorders, 18, 53-66.

Steffenburg, S. (1990). Neurobiological correlates of autism. G–teborg: G–teborg University MD Thesis.

Steffenburg, S. (1991). Comprehensive neuropsychiatric assessment in autism. A population-based study. Developmental Medicine and Child Neurology, 33, 495-511.

Steffenburg, S. & Gillberg, C. (1986). Autism and Autistic-like conditions in Swedish rural and urban areas: a population study. British Journal of Psychiatry, 149, 81-87.

Steffenburg, S. & Gillberg, C. (1989). The Etiology of Autism. In C. Gillberg (Ed.), Diagnosis and treatment of autism (pp. 63-82). New York: Plenum Press.

Steffenburg, S., Gillberg, C., Hellgren, L., Andersson, L., Gillberg, IC, Jakobsson, G. & Bohman, M. (1989). A twin study of infantile autism in Denmark, Finland, Iceland, Norway & Sweden. Journal of Child Psychology and Psychiatry, 30, 405-416.

Stiver, R.L. & Dobbins, JP (1980). Treatment of atypical anorexia nervosa in the public school: an autistic girl. Journal of Autism and Developmental Disorders, 10, 67-73.

Stubbs, E.G. (1978). Autistic symptoms in a child with congenital cytomegalovirus infection. Journal of Autism and Childhood Schizophrenia, 8, 37-43.

Sugiyama, T. & Abe, T. (1989). The prevalence of autism in Nagoya, Japan: a total population study. Journal of Autism and Developmental Disorders, 19, 87-96.

Tanoue, Y., Oda, S., Asano, F. & Kawashima, K. (1988). Epidemiology of infantile autism on Southern Ibaraki, Japan: differences in prevalence rates in birth cohorts. Journal of Autism and Developmental Disorders, 18, 155-166.

Treffert, D.A. (1970). Epidemiology of infantile autism. Archives in General Psychiatry, 22, 431-438.

Tsai, L. (1989). Recent neurobiological findings in autism. In C. Gillberg (Ed.), Diagnosis and treatment of autism (pp. 83-104). New York: Plenum Press.

Volkmar, F. & Nelson, I. (1990). Seizure disorders in autism. Journal of the American Academy of Child and Adolescent Psychiatry, 29, 127-129.

Wing, L. (1980). Childhood autism and social class: a question of selection. British Journal of Psychiatry, 127, 410-417.

Wing, L. (1981). Asperger's syndrome: a clinical account. Psychological Medicine, 11, 115-129.

Wing, L. (1983). Social and interpersonal needs. In E. Schopler & G.B. Mesibov (Eds.), Autism in adolescents and adults (pp. 337-354). New York: Plenum Press.

Wing, L. (1989a). The diagnosis of autism. In C. Gillberg (Ed.), Diagnosis and treatment of autism (pp. 5-22). New York: Plenum Press.

Wing, L. (1989b). Autistic adults. In C. Gillberg (Ed.), Diagnosis and treatment of autism (pp. 419-432). New York: Plenum Press.

Wing, L. & Gould, J. (1978). Systematic recording of behaviors and skills of retarded and psychotic children. Journal of Autism and Childhood Schizophrenia, 8, 79-97.

Wing, L. & Gould, J. (1979). Severe impairments of social interaction and associated abnormalities in children: epidemiology and classification. Journal of Autism and Developmental Disorders, 9, 11-29.

Witt-Engerstr–m, I. & Gillberg, C. (1987). Autism and Rett syndrome. A preliminary epidemiological study of diagnostic overlap (Letter). Journal of Autism and Developmental Disorders, 17, 149-150.