Using the Autism Diagnostic Interview-Revised and the Autism Diagnostic Observation Schedule-Generic for the Diagnosis of Autism Spectrum Disorders in a Greek Sample with a Wide Range of Intellectual Abilities
We studied the interrelationship between the Autism Diagnostic Observation Schedule-Generic (ADOS-G), the Autism Diagnostic Interview-Revised (ADI-R) and DSM-IV clinical diagnosis, in a Greek sample of 77 children and adolescents, referred for the assessment of a possible pervasive developmental disorder (PDD) and presenting a wide range of cognitive abilities. The agreement of the ADOS-G and the ADI-R with the clinical diagnosis was estimated as satisfactory and moderate, respectively, while both instruments presented with excellent sensitivity for the diagnosis of autistic disorder along with satisfactory specificity. ADOS-G/ADI-R agreement was estimated as fair. Our results confirm the discriminant validity of ADI-R and ADOS-G in diagnosing pervasive developmental disorders in children and adolescents with a wide range of intellectual abilities.
KeywordsADI-R ADOS-G Autism spectrum disorders Diagnosis
Reliable and valid diagnosis of the pervasive developmental disorders (PDD) has always presented a challenge for both clinicians and researchers. Since diagnosing PDD may sometimes be complex, the clinical assessment should involve several components including extensive information from the child, parents, school, or other important sources (Volkmar et al. 1999, 2004). Diagnostic instruments have played an important role during this process; information regarding the behavior of a child can be collected in a standardized way which is critical for diagnostic and classification purposes. However, instruments may only be deemed as valuable if they measure PDD in a valid and reliable way; it is important therefore, to study the interrelationship between instruments and the clinical diagnosis. During the last 20 years, a combination of instruments has been developed for the assessment of Autism Spectrum Disorders (ASDs) in children and adolescents: the Autism Diagnostic Interview-Revised (ADI-R), (Le Couteur et al. 1989; Lord 1997; Lord et al. 1994; Rutter et al. 2003), and the Autism Diagnostic Observation Schedule-Generic (ADOS-G), (Lord et al. 2000). Both instruments operationalize the three main dimensions of DSM-IV-TR (American Psychiatric Association 2004) and the ICD-10 (World Health Organisation 1992) criteria for autism in terms of specific behavioral features. The ADOS-G and ADI-R were initially developed during the nineties for research purposes, yet during the last few years they have been increasingly utilized in everyday clinical practice. When used together, data are available regarding both current behavior and also, the history and development of the child. Since ADI-R and ADOS-G have been developed primarily for research, more data are required on their use in clinical settings in a wide sample of patients.
Lord et al. (1993, 1994) have reported high sensitivity and specificity and good interrater reliability of ADI-R for preschoolers, as well as for people of a wide range of ages (Lord 1997). In a report by Cox et al. (1999) the ADI-R was found to have good specificity but poor sensitivity at detecting autism at the age of 20 months whereas the stability of diagnosis from 20 to 42 months was good. Lecavalier et al. (2006) identified factor structure, internal consistency, and convergent validity of the ADI-R algorithm in a large sample of children and adolescents with pervasive developmental disorders (PDD).
ADOS-G has shown excellent interrater reliability between domains, internal consistency and sensitivity/specificity for autism and Pervasive Developmental Disorders Not Otherwise Specified (PDD-NOS) relative to non-spectrum disorders (Lord et al. 2000). A revised algorithm consisting of two domains, Social Affect and Restricted, Repetitive Behaviors, has further improved its diagnostic validity (Gotham et al. 2007).
Noterdaeme et al. (2002) studied children with autism and children with a severe specific language disorder; the results indicated that the diagnostic accuracy of the ADI-R and the ADOS-G is very good and that these instruments are valuable aids for the diagnostic process. de Bildt et al. (2004) examined the interrelationship between ADI-R, ADOS-G, and a clinical DSM-IV-TR classification in a mentally retarded population. The level of agreement between the diagnostic classification of ADI-R and ADOS-G was fair, yet the interrelation between the instruments and the clinical diagnosis was very good in this population. Additionally, Bolte and Poustka (2004) described satisfying agreement of the ADOS-G with the clinical diagnosis when used in a German population and fair agreement in-between the interviews. Ventola et al. (2006) compared ADOS-G, ADI-R, CARS and clinical judgment when applied to toddlers. They found that there was consensus between the ADOS-G, CARS, and clinical judgment, yet not with the ADI-R. Risi et al. (2006) compared single and combined ADI-R and ADOS-G algorithms to best estimated diagnosis in four samples in two studies. Both ADOS-G and ADI-R had excellent sensitivity while their specificity varied significantly depending on the sample and the study. Much higher sensitivities and specificities were obtained when strict criteria for an autism diagnosis were applied and when both instruments were used. Sensitivity and specificity were lower for non-Autism Spectrum Disorders. Tomanik et al. (2007) studied the classification agreement of the ADI-R and the ADOS-G and found adequate concordance between the two instruments, with 75% of participants being correctly classified using the ADOS-G, while the use of adaptive functioning improved classification accuracy. In their report, Mazefsky and Oswald (2006) evaluated the use of ADOS-G and ADI-R in a clinical setting and results indicated a 75% agreement between the instruments and clinical diagnosis. ADI-R and ADOS-G diagnostic classifications were also compared to consensus clinical diagnoses when assessing 209 preschool children with developmental delay (Gray et al. 2007). Children with a clinical diagnosis of autism scored significantly higher on all algorithm domains of the ADI-R and ADOS-G. The ADOS-G performed better than the ADI-R in comparison to consensus clinical diagnosis. Fair agreement was observed between the ADI-R and the ADOS-G. Chawarska et al. (2007) reported that the agreement between the ADOS-G and clinician assigned autism but not PDD-NOS diagnosis was high, whereas sensitivity of the ADI-R diagnostic classification of autism was poor in a sample of children in their second year of life. In another sample of preschool children, findings demonstrated a good agreement between the instruments especially for children with core autism (Le Couteur et al. 2008). The instruments appeared to have a complementary effect in aiding diagnosis and confirm the importance of a multidisciplinary assessment process with access to information from different sources and settings. The presence of repetitive behaviors during the ADOS-G appeared to be of diagnostic significance.
Finally, Wiggins and Robins (in press) report that the agreement of ADI-R with ADOS-G was fair in the evaluation of toddlers and improved when the ADI-R behavioral domain was excluded.
The aim of the present study was to investigate the agreement among ADI-R, ADOS-G, and clinical diagnosis based on the DSM-IV (American Psychiatric Association 1994) criteria in a Greek sample of children and adolescents referred for a clinical assessment to a PDD clinic. We also examined the sensitivity, specificity, and positive predictive value of these instruments as used in our sample using the DSM-IV diagnosis as a gold standard. There is no previous study regarding the use of ADOS-G and ADI-R in a Greek population either for research purposes or in a clinical setting.
The sample consisted of 77 patients, 58 boys and 19 females. They were all referred for clinical assessment to an outpatient PDD clinic during a two year period and were also screened for inclusion in a molecular genetics study of autism. Referral sources included primary care clinical settings, independent professionals, schools and parents. Of the sample population, 22% had had no previous contact with a child psychiatric service. They ranged in age from 33 months to 22 years (mean = 83 months, SD = 44 months). Their non-verbal IQ ranged from 40 to 146 (mean = 83, SD = 23). All of them were Caucasians. Seventy-five were of Greek nationality and two were of another European nationality.
Measures and Procedure
A multidisciplinary team was involved in the assessment of each case. Each patient underwent a child psychiatric evaluation including developmental history and clinical examination separate from ADI-R and ADOS-G, cognitive assessment by tests, such as Leiter-R (Roid and Miller 1997), Raven Progressive Matrices (Raven 1960), WISC III (Georgas et al. 1997), speech and motor skills assessment, neurological examination and a thorough diagnostic work up including a hearing test, electroencephalogram, neuroimaging, fragile X, karyotyping, and blood and urine metabolic tests.
The ADI-R and ADOS-G were administered by three child psychiatrists (K. P., E. P., S. V.) all trained in their use with periodic evaluation of interrater reliability ranging over 85%. All ADOS-Gs were videotaped.
The ADI-R (Lord et al. 1994; Rutter et al. 2003) is a standardized semi-structured clinician based interview for caregivers of individuals with autistic disorder. The diagnosis of autistic disorder according to the ADI-R requires that individuals reach or exceed cut-off scores in all three ICD-10 symptoms domains: social interaction (SI), communication (CO) and repetitive behaviors and stereotyped patterns (RB); it is necessary that the abnormality was evident before the age of 36 months. In our study, we used the initial version of ADI-R which was more extensive than the current one, as some of the old items have been removed and two new items assessing aggressiveness have been added. None of these changes affects the diagnostic algorithm scores. The algorithm is based on ICD-10 and DSM-IV-TR definitions and yields a classification of either autistic disorder or non-autistic; it does not consider PDD-NOS as a possible diagnosis. For the purposes of this study however and in some of our analyses, we also considered a PDD-NOS diagnosis in accordance with the contention that children who exceed the cut-off in two of the three domains of ADI-R are likely to present with some type of PDD (Rutter et al. 2003).
ADOS-G (Lord et al. 2000) is a semi-structured, standardized assessment of communication, social interaction, and play or imagination. It consists of four modules; each module contains a schedule of activities designed for use with children or adults at a particular developmental and language level, ranging from non-verbal to verbally fluent persons. Diagnostic classification is made on the basis of exceeding thresholds on each of two domains, social interaction and communication and exceeding a threshold for a combined social-communication score. Apart from autism ADOS-G also provides an algorithm for non-autism ASDs.
ADI-R and ADOS-G were translated to Greek and then independently translated back into English. ADOS-G module 1 was administered to 19 subjects, module 2–42 subjects, module 3–17 subjects and module 4–1 subject.
The clinical diagnosis was made during a case conference independently of the ADI-R, ADOS-G study and on the basis of all available information from a variety of sources excluding the scores of the ADI-R and ADOS-G algorithms which were calculated afterwards. At that time, DSM-IV criteria for autistic disorder and other PDD diagnoses were used as the gold standard for clinical diagnosis. Because of the small number of non-autism ASDs diagnosis (notably Asperger’s syndrome, PDD-NOS and childhood disintegrative disorder), they were treated all together as the PDD-NOS group. Due to lack of resources the assignment of a blind diagnosis by an independent clinician was not feasible. In the case of disagreement in the assignment of the diagnosis, further information about clinical presentation was obtained by parents, school or direct observation of the child in order to reach a consensus.
Only subjects that were fully assessed including cognitive assessment were included in the study.
Cohen’s kappa (k) was used to detect pair-wise agreement between ADI-R, ADOS-G, and clinical diagnosis. Agreement was evaluated considering both dichotomous (autistic disorder/PDD-NOS and non-autism spectrum disorders) and tricotomous diagnosis (autistic disorder/PDD-NOS/non-autism spectrum disorders). Furthermore, agreement in trichotomous diagnosis was considered both weighted (meaning that discrepancy between autistic disorder and non-autism spectrum weights 1 while between autistic disorder and PDD-NOS weights 1/2) and absolute. Agreement was considered as excellent (k > .75), satisfactory (k = .60–.74), moderate (k = .40–.59), and fair (k < .40) based on Cicchetti and Sparrow (1981) classification of measurement of observer agreement for categorical data.
Receiving Operating Characteristics (ROC) curves were applied to evaluate sensitivity and specificity of the instruments.
Logistic regression was conducted to evaluate the impact of age, sex, and cognitive level on the agreement of the instruments.
Forty-two of our referrals received the clinical diagnosis of autistic disorder, 23 received the clinical diagnosis of an alternate Pervasive Developmental Disorder diagnosis (PDD-NOS = 18, Asperger’s syndrome = 5 and childhood disintegrative disorder = 1), and 12 received a non-ASD diagnosis. Thirty-one of our subjects received an autism spectrum diagnosis for the first-time in their life.
Agreement of ADOS-G and ADI-R with clinical diagnosis and in-between considering trichotomous weighted and absolute diagnosis
% Agreement, kappa measure of agreement (p-value)
81.82, k = 0.49 (p < .001)
67.53, k = 0.42 (p < .001)
74.68, k = 0.34 (p < .001)
57.14, k = 0.23 (p = .002)
83.77, k = 0.60 (p < .001)
72.73, k = 0.53 (p < .001)
83.33, k = 0.58 (p = .001)
66.67, k = 0.42 (p = .006)
88.89, k = 0.73 (p < .001)
83.33, k = 0.70 (p < .001)
76.83, k = 0.39 (p = .001)
58.54, k = 0.24 (p = .017)
81.71, k = 0.53 (p < .001)
68.29, k = 0.46 (p < .001)
64.71, k = 0.08 ns*
47.06, k = 0.08 ns*
85.92, k = 0.65 (p < .001)
76.47, k = 0.61 (p < .001)
These data show the moderate to satisfactory agreement of ADOS-G with clinical diagnosis, as well as the moderate agreement of ADI-R with clinical agreement. For both analyses, ADOS-G modules were evaluated both in total and separately (module 4 was conducted in only one case, so it was not included in the analysis). It appears that agreement with clinical diagnosis was moderate to satisfactory in all ADOS-G modules, showing the high diagnostic validity of ADOS-G for the diagnosis of all autism spectrum disorders in all developmental levels. In-between agreement of the interviews was estimated as fair, both weighted and absolute, when chance was measured. The best agreement of ADOS-G with ADI-R was in ADOS-G module 1 (moderate to satisfactory), whereas it was not statistically significant in module 3.
Agreement between ADOS-G and ADI-R with clinical diagnosis and in-between considering dichotomous diagnosis
% Agreement, kappa measure of agreement, (p-value)
79.22, k = 0.57 (p < .001)
68.83, k = 0.35 (p = .001)
84.42, k = 0.68 (p < .001)
72.22, k = 0.43 (p = .033)
88.89, k = 0.77 (p < .001)
73.17, k = 0.43 (p = .003)
82.93, k = 0.65 (p < .001)
58.82, k = 0.13 ns*
82.35, k = 0.64 (p = .004)
These results show again that ADOS-G has a satisfactory to excellent agreement with clinical diagnosis, for all the three modules analyzed. Agreement of ADOS-G with ADI-R was fair when considering all three modules. When analyzed separately, modules 1 and 2 showed moderate agreement with ADI-R, while agreement between module 3 and ADI-R did not even reach statistical significance.
Logistic regression was applied for sex, age, and non-verbal IQ. No effect of sex and age was detected while non-verbal IQ affected only ADOS-G/ADI-R agreement (OR = 0.97, 95% CI: 0.949–0.995). Children with higher non-verbal IQ had significantly less chance to present ADI-R/ADOS-G agreement; an increase of 10 points in IQ would result in a 33% decrease in ADI-R/ADOS-G agreement.
Sensitivity, specificity and positive predictive value of the instruments were estimated considering diagnoses according to the diagnostic algorithm typically designed for each instrument.
ADI-R and ADOS-G: sensitivity, specificity, positive predictive value and ROC curves for different comparisons
Autism versus PDD-NOS & non-spectrum
Autism versus PDD-NOS
Autism versus non-spectrum
PDD-NOS versus non-spectrum
Autism & PDD-NOS versus non-spectrum
Positive predictive value
ROC curve (AUCa, 95% CIb)
Additionally, ROC curves were conducted for both instruments to evaluate sensitivity and specificity in accordance with clinical diagnosis considering the whole range of scores instead of the cut-off scores. In our study, the size of the area under the curve (AUC) in both instruments was indicative of a very good correlation between sensitivity and specificity. The lower value was for the diagnosis of PDD-NOS versus non-spectrum disorders when using ADOS-G.
Our results indicated that in both dichotomous and trichotomous diagnosis, the ADI-R had a moderate agreement with clinical diagnosis, while the ADOS-G had a moderate to satisfactory agreement consistent for all modules, along with a high sensitivity and specificity for the diagnosis of the autistic disorder. The interrelation between ADOS-G and ADI-R was fair.
The ADOS-G sensitivity and specificity findings were somewhat lower but on the whole comparable to the ones found by Lord et al. (2000) in the original paper describing its properties. ADOS-G showed low sensitivity in diagnosing PDD-NOS versus non-spectrum disorders in our sample, which once more highlights the difficulties embedded in the diagnosis of PDD-NOS. Also ADOS-G was less effective in differentiating autistic disorder from PDD-NOS in accordance with the report by Lord et al. (2000). Bolte and Poustka (2004) reported lower agreement of ADOS-G and clinical diagnosis compared to our findings in a German sample; sensitivity was similar in both samples, although specificity was higher in our sample.
The ADI-R sensitivity and specificity in our sample were also somewhat lower but comparable to those reported by the original studies by the authors (Lord 1997; Lord et al. 1993, 1994); they were higher however, compared to those found by Cox et al. (1999) and Ventola et al. (2006) possibly as their samples consisted of much younger subjects. Our results are also in accordance with other reports (de Bildt et al. 2004; Gray et al. 2007; Mazefsky and Oswald 2006) showing satisfactory agreement of both interviews with the clinical diagnosis. In their samples, Risi et al. (2006) reported sensitivities for both instruments comparable to our findings but lower specificities.
In our study, ADI-R presented its best agreement with clinical diagnosis when dichotomous diagnosis was applied. This was partly to be expected, since the algorithm has been developed for the diagnosis of autistic disorder. The fact that the agreement with the clinical diagnosis remained moderate even when a trichotomous diagnosis was considered, supports the contention that the children who exceed cutoffs in two domains of the ADI-R are likely to present with some form of PDD (Rutter et al. 2003). However, difficulties in the assessment of PDD-NOS have been extensively studied in the literature, indicating that a sole clinical opinion lacks validity in the majority of the cases (Mahoney et al. 1998). Since revision of ADI-R diagnostic algorithm has been discussed by many researchers, mainly focusing on using less stringent algorithm, changes to the algorithm in order to be able to assign a PDD-NOS diagnosis could also be considered.
A finding that remains consistent among studies is the low degree of agreement between the ADOS-G and the ADI-R (Bolte and Poustka 2004; de Bildt et al. 2004; Gray et al. 2007; Ventola et al. 2006; Wiggins et al. in press). Le Couteur et al. (2008) though showed good kappa level of agreement for a best estimate of autism in preschoolers which is in accordance with previous research showing that concordance between the two instruments is greater in younger children (de Bildt et al. 2004). This was also the case in our study where kappa level of agreement between ADOS-G and ADI-R was moderate in module 1, whereas in module 2 and 3 which are usually administered to older subjects, or in all three modules analyzed together it was fair or even statistically non-significant. There are many possible reasons for this finding.
Firstly, the ADI-R algorithm score relies mainly on the 4–5 age period while the ADOS-G algorithm on current behavior. For children aged five and over, the ADI-R algorithm specifies that on some items the prototypical autistic behavior is seen at the age 4–5. Abnormalities might have become more subtle by the time they are observed by ADOS-G (de Bildt et al. 2004) especially when subjects are more able (Baird et al. 2003). This is supported by our finding that children with higher IQ had significantly less chance to present with ADI-R/ADOS-G agreement. Secondly, ADI-R relies on parent-reported behaviors while the ADOS-G is based on the observation of the subject. Discrepancies between clinical and parental views particularly when referring to past preschool years are a common finding in the literature, for older, verbal subjects with adequate functioning (Noterdaeme et al. 2002). Agreement of parental reports with clinical observations is greater for non-verbal young children due to parents’ recollection difficulties later on (Risi et al. 2006). This is in accordance with the findings of our study where ADI-R/ADOS-G agreement became non-significant for ADOS-G module 3 which is administered to verbal and usually older subjects, while it was moderate to satisfactory for module 1. Discrepancies could also be accentuated in our sample by the fact that we assessed many families whose child had no previous clinical diagnosis of autism. First-time parents tend to underestimate abnormalities in social and language development (Lord 1997). Thirdly, the ADI-R algorithm score for autistic disorder is based on exceeding the cutoffs in three areas of development namely communication, social interaction and restricted interests while in ADOS-G the algorithm is based only on the scores for the communication and social domains, excluding restricted and repetitive behaviors. The exclusion of the ADI-R behavioral domain seems to improve the agreement of ADI-R with other diagnostic instruments, such as the ADOS-G in toddlers (Wiggins and Robins in press). It remains to be seen whether the implementation of the revised algorithm for ADOS-G (Gotham et al. 2007) where restricted repetitive behaviors are included, could also increase the agreement between the two instruments.
An important limitation of our study is the fact that it was not possible to have a blind diagnosis assigned by an independent clinician. Information from ADI-R and ADOS-G was available to the clinicians who assigned the diagnoses. In order to retain some independence, however, the algorithm scores were not available to them and also there were many other sources of information. It has been also reported (Risi et al. 2006) that the consensus diagnosis is more likely to agree with ADOS-G than ADI-R classification, when clinicians are present during the ADOS-G which was the case in many of our patients where a one way mirror was used. This might have biased our results towards a better performance of ADOS-G compared to ADI-R. Finally, another limitation of the study is its cross sectional design, meaning that predictive validity of the instruments could not be obtained.
Despite the limitations, our results illustrate the discriminant validity of ADI-R and ADOS-G in diagnosing pervasive developmental disorders in a Greek sample of children and adolescents with a wide range of intellectual abilities in a setting that receives referrals both for clinical assessment and research screening. By being consistent with the findings of levels of sensitivity and specificity reported in the original studies (Lord et al. 2000; Lord et al. 1994, 1993), our findings validate the use of these instruments in our population. Also, they reveal once more the need to combine information from both ADI-R and ADOS-G and from clinical assessment in order to assign a diagnosis within the autism spectrum disorders.
We wish to thank the parents and children that participated in this study. We also thank our colleagues Margarita Bouranta, psychologist, Sofia Giannopoulou speech therapist, Dimitrios Arvanitis and Konstantina Argyrou occupational therapists, and Vassiliki Dre health visitor who contributed to the clinical assessment of our patients. We also thank Debbie Spain, Cognitive Behavioral Therapist, for the careful editing of the final manuscript.
- Association, American Psychiatric. (1994). Diagnostic and statistical manual of mental disorders (DSM-IV). Washington, DC: American Psychiatric Press Inc.Google Scholar
- Association, American Psychiatric. (2004). Diagnostic and statistical manual of mental disorders forth edition-text revision (DSM-IV-TR). Washington, DC: American Psychiatric Press Inc.Google Scholar
- Cox, A., Klein, C., Charman, T., Baird, G., Baron-Cohen, S., Swettenham, J., et al. (1999). Autism spectrum disorders at 20 and 42 months of age: Stability of clinical and ADI-R diagnosis. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 40, 719–732. doi:10.1017/S002196309900400X.PubMedCrossRefGoogle Scholar
- de Bildt, A., Sytema, S., Ketelaars, C., Kraijer, D., Mulder, E., Volkmar, F., et al. (2004). Interrelationship between autism diagnostic observation schedule-generic (ADOS-G), autism diagnostic interview-revised (ADI-R), and the diagnostic and statistical manual of mental disorders (DSM-IV-TR) classification in children and adolescents with mental retardation. Journal of Autism and Developmental Disorders, 34(2), 129–137. doi:10.1023/B:JADD.0000022604.22374.5f.PubMedCrossRefGoogle Scholar
- Georgas, D. D., Paraskevopoulos, I. N., Bezevegis, I. G., & Giannitsas, N. D. (1997). Greek WISC III Wechsler intelligence scales for children. Athens: Ellinika Grammata.Google Scholar
- Gray, K. M., Tonge, B. J., & Sweety, D. J. (2007). Using the autism diagnostic interview-revised and the autism diagnostic observation schedule with young children with developmental delay: Evaluating diagnostic validity. Journal of Autism and Developmental Disorders. doi:10.1007/s10803-007-0432-y.
- Le Couteur, A., Haden, G., Hammal, D., & McConachie, H. (2008). Diagnosing autism spectrum disorders in pre-school children using two standardized assessment instruments: The ADI-R and the ADOS. Journal of Autism and Developmental Disorders, 38(2), 362–372. doi:10.1007/s10803-007-0403-3.PubMedCrossRefGoogle Scholar
- Lord, C. (1997). Diagnostic instruments in autism spectrum disorders. In D. Cohen & F. Volkmar (Eds.), Handbook of autism and pervasive developmental disorders (pp. 460–483). New York: Wiley.Google Scholar
- Lord, C., Risi, S., Lambrecht, L., Cook, E., Leventhal, B., DiLavore, P. C., et al. (2000). The autism diagnostic observation schedule-generic: A standard measure of social and communication deficits associated with the spectrum of autism. Journal of Autism and Developmental Disorders, 30(3), 205–223. doi:10.1023/A:1005592401947.PubMedCrossRefGoogle Scholar
- Lord, C., Rutter, M., & Le Couteur, Ann. (1994). Autism diagnostic interview-revised: A revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. Journal of Autism and Developmental Disorders, 24(5), 659–685. doi:10.1007/BF02172145.PubMedCrossRefGoogle Scholar
- Lord, C., Storoschuk, S., Rutter, M., & Pickles, A. (1993). Using the ADI-R to diagnose autism in preschool children. Infant Mental Health Journal, 14(3), 234–252. doi :10.1002/1097-0355(199323)14:3<234::AID-IMHJ2280140308>3.0.CO;2-F.Google Scholar
- Mahoney, W., Szatmari, P., MacLean, J., Bryson, S., Bartolucci, G., Walter, S., et al. (1998). Reliability and accuracy of differentiating pervasive developmental disorders subtypes. Journal of the American Academy of Child and Adolescent Psychiatry, 37, 278–285. doi:10.1097/00004583-199803000-00012.PubMedCrossRefGoogle Scholar
- Raven, J. C. (1960). Guide to using the standard progressive matrices. London: H. K. Lewis.Google Scholar
- Risi, S., Lord, C., Gotham, K., Corsello, C., Chrysler, C., Szatmari, P., et al. (2006). Combining information from multiple sources in the diagnosis of autism spectrum disorders. Journal of the American Academy of Child and Adolescent Psychiatry, 45, 1094–1103. doi:10.1097/01.chi.0000227880.42780.0e.PubMedCrossRefGoogle Scholar
- Roid, G. H., & Miller, L. J. (1997). Leiter performance scale-revised manual. Wood Dale, IL: Stoelting.Google Scholar
- Rutter, M., Le Couteur, A., & Lord, C. (2003). Autism diagnostic interview-revised. Los Angeles, CA: Western Psychological Services.Google Scholar
- Volkmar, F., Cook, E., Pomorey, J., Realmuto, G., & Tanguay, P. (1999). Summary of the practice parameters for the assessment and treatment of children, adolescents and adults with autism and other pervasive developmental disorders. Journal of the American Academy of Child and Adolescent Psychiatry, 38(12), 1611–1615.PubMedCrossRefGoogle Scholar
- Wiggins, L. D., & Robins, D. L. Brief report: Excluding the ADI-R behavioural domain improves diagnostic agreement in toddlers. Journal of Autism and Developmental Disorders, in press. doi: 10.1007/s10803-007-0456-3.
- World Health Organisation. (1992). The tenth revision of the International Classification of Disorders and related health problems (ICD-10). WHO: Geneva.Google Scholar