Subtypes of Pervasive Developmental Disorder: Clinical Characteristics*

Previous analysis of data from 505 preschool children with disordered communication, falling into the a priori categories of Pervasive Developmental Disorder (PDD), Developmental Language Disorder, and Nonautistic Low IQ, showed that PDD is statistically distinct from nonPDD. Current analyses of the 194 children with PDD also showed the existence of two subgroups within PDD. Children in these two subgroups are sharply divided on the basis of overall cognitive level; children in both groups share major symptom areas, but specific behavioral manifestations differ. Differing developmental trajectories into school age validate the distinction. A clinically usable algorithm for classifying PDD children into the two subgroups, based on social developmental level and degree of social abnormality, is provided. The findings suggest that highand low-functioning individuals with PDD should be conceptualized as essentially distinct and should be studied separately for etiology, pathophysiology, course, and treatment. Classification of Autistic Spectrum Disorders Since the mid l970s, a variety of researchers have called for exploration of possible subgroups within the autistic spectrum (Lotter, l974; Reiss & Freund, l990; Szatmari, l992; Volkmar & Cohen, l986). It has been repeatedly pointed out that identification of possibly distinct subtypes will be a necessary step in delineating the etiology, pathophysiology, course, and treatment options for children on the autistic spectrum (Roux, Garreau, Barthelemy, & Hameury, 1994). Following DSM-IV usage, we will hereafter refer to the broad autistic spectrum as ‘Pervasive Developmental Disorder’ (PDD) and restrict the term ‘autistic’ to the more tightly defined Autistic Disorder as per DSM-IV (APA, 1994). Many studies done prior to the publication of DSM-IV have subject samples that were termed ‘autistic’, but that were defined more broadly or with different criteria than the current ‘Autistic Disorder’. When describing these studies, therefore, we will refer to their samples as having PDD. At present there are two general forms of subgrouping within the Pervasive Developmen2 DEBORAH FEIN ET AL. tal Disorders, as follows: (1) empirical classification conducted on varying subsets of features of the disorder, and (2) clinical consensus and field trial studies conducted on the complete set of diagnostic symptoms. Empirical subgrouping Some empirical classifications have focussed on behaviors and symptoms: Wing and Gould (l979) classified children with PDD on the basis of sociability (aloof, passive, active-but-odd), a classification which has been supported recently (Borden & Ollendick, 1994; Castelloe & Dawson, l993; Volkmar, Cohen, Bregman, & Hooks, l989); Allen (1988) classified children on the basis of communication and play. Other studies have attempted classification on the basis of cognitive differences, such as overall cognitive strengths and weaknesses (Fein, Waterhouse, Lucci, & Snyder, l985) or specific language deficits (Allen & Rapin, l992; Simmons & Baltaxe, l975). Other attempts at classification have focussed on presence or absence of EEG abnormalities (Tsai, Tsai, & August, l985) age of onset, or developmental course (Burd, Fisher, & Kerbeshian, l989; Percy, Gillberg, & Hagberg, l990; Prior, Perry, & Gajzago, l975; Volkmar, l992), although no firm consistent findings have yet emerged relating onset or course to final behavioral outcome. Some investigators have used a variety of empirical and statistical methods to define PDD subgroups. In an early study, Prior, Perry, and Gajzago (l975) found two clusters, one representing early onset and Kanner’s type symptomatology, and the other representing later onset and more varied symptoms. Siegel, Anders, Ciaranello, Bienenstock, and Kramer (l986) performed a cluster analysis and identified four groups of children with PDD; although no measure of IQ was included, one of these types seems to correspond to high-functioning autism, and one to retarded autism, whereas the other two were marked by schizotypal features, and by negativism and anxiety. Bagley and McGeein (l989) found four clusters, related to mutism, speech pathology, and social responsiveness, with the mute and unresponsive subjects showing particularly poor outcome after 4 years. Overall and Campbell (l988) identified anger, hyperactivity, speech deviance, and autism as the behaviors most able to discriminate subtypes of autism. Eaves, Ho, and Eaves (1994) identified four clusters of children with PDD, differing in behavioral profiles and cognitive level. Fein, Waterhouse, Lucci, and Snyder (1985), also using cluster analysis, identified eight cognitive profiles accounting for 51/54 children with PDD; these patterns were related to handedness, but not to autistic symptomatology. Important beginnings have been made in identifying etiologies that may correspond to certain behavioral or cognitive profiles. Gillberg (l992) addressed this most directly, comparing subgroups of children with PDD with a defined etiology on a set of behaviors and symptoms, cautioning that etiological classification does not necessarily map well onto behavioral subgroups. Specific behavioral features that seemed to occur disproportionately with different biological conditions (although sample sizes per condition were too small for tests of significance) included IQ level, a somewhat remitting course after age 7, and a variety of specific symptoms. Other investigators have suggested subgroups marked by a variety of etiologies and biological markers (Coleman, l990), although others present evidence that only a small minority of individuals with PDD have known medical conditions (Rutter, Bailey, Bolton, & LeCou-