Brief Report: An Exploratory Study Comparing Diagnostic Outcomes for Autism Spectrum Disorders Under DSM-IV-TR with the Proposed DSM-5 Revision
The proposed revision for Autism spectrum disorders (ASDs) in the Diagnostic and Statistical Manual of Mental Disorders—Fifth Edition (DSM-5) represents a shift from the Diagnostic and Statistical Manual of Mental Disorders—Fourth Edition, Text Revision (DSM-IV-TR). As the proposed DSM-5 criteria require a higher minimum number of symptoms to be present compared to DSM-IV-TR, there have been some concerns about the impact that this will have on diagnostic outcomes. Therefore, the current study aimed to compare diagnostic outcomes using both DSM-IV-TR and DSM-5 criteria for 132 children. Of the 111 participants who received an ASD diagnosis under DSM-IV-TR, 26 did not meet DSM-5 criteria. The majority of these had received a DSM-IV-TR PDD-NOS diagnosis. Implications of the results and the proposed DSM-5 changes to the ASD criteria are discussed.
KeywordsAutism spectrum disorder Asperger’s Disorder Autistic Disorder PDD-NOS Assessment Diagnosis DSM-IV-TR DSM-5 Social Communication Disorder
The proposed revision for Autism spectrum disorders (ASD) in the Diagnostic and Statistical Manual of Mental Disorders—Fifth Edition (DSM-5, accessible at www.DSM-5.org) has created some controversy in recent times, most notably in relation to the possibility that some individuals who would currently meet criteria under the Diagnostic and Statistical Manual of Mental Disorders—Fourth Edition, Text Revision (DSM-IV-TR; American Psychiatric Association 2000) will no longer meet criteria under the proposed DSM-5.
DSM-IV-TR places ASDs under the category of pervasive developmental disorders (PDDs), which also includes two non-ASD PDDs i.e. Rett’s syndrome and childhood disintegrative disorder. DSM-IV-TR outlines twelve possible “symptoms” of ASD within three areas of impairment, i.e. social interaction, communication and restricted and repetitive interests, activities and behaviors (RRBs). There are three possible ASD diagnoses in DSM-IV-TR i.e. Autistic Disorder, Asperger’s Disorder and Pervasive Developmental Disorder-Not Otherwise Specified (including Atypical Autism; PDD-NOS). These three categories are set out in a hierarchical fashion in DSM-IV-TR whereby Autistic Disorder requires the highest minimum symptom set (i.e. at least six characteristics in total across all three areas with at least two from the social interaction area, one from communication area, and one from RRBs and abnormality apparent by age 3 years). A diagnosis of Asperger’s Disorder can be considered once Autistic Disorder is ruled out, requiring at least two social interaction characteristics and at least one characteristic from RRBs. There can be no cognitive or adaptive delay and there must be evidence of functional impairment. If an individual does not meet criteria for Autistic Disorder or Asperger’s Disorder, then PDD-NOS can be considered. This is the least stringent of the ASDs outlined in DSM-IV-TR and requires evidence of severe social impairment accompanied by either communication impairment or RRBs. There is no specific minimum number of symptoms outlined.
The proposed revision for ASDs in DSM-5 consists of a single category, i.e. Autism spectrum disorder. The rationale for the removal of the three category system outlined on the DSM-5 website by the DSM-5 Neurodevelopmental Work Group is that, although differentiation of ASD from typical development and other childhood disorders is done reliably and validly, the distinction between the three categories has been found to be inconsistent and to vary across sites (American Psychiatric Association 2011). Certainly, this is the case in the Australian context. In our clinical experience some assessment centers have established their own minimum number of symptoms in order to meet criteria for PDD-NOS; some require at least some sub-threshold impairment across all three areas while others do not. Other centers do not necessarily follow the hierarchical process and diagnose Asperger’s Disorder in any individual of average cognitive ability who presents with social impairment and restricted interests. In addition to the variation in application of the current criteria in relation to the three subtypes, the vast majority of research to date has not been able to identify meaningful differences between the subtypes of ASD as outlined in DSM-IV-TR once cognitive ability and language level have been controlled for (e.g. Allen et al. 2001; First, 2008; Howlin, 2003; Mordre et al. 2011).
Another proposed change in DSM-5 is reducing the domains of impairment from three to two by merging the social interaction and communication domains into a single domain. This effectively means that the current eight criteria in these areas will be reduced into three criteria. RRBs will be the second DSM-5 domain and will consist of four criteria which specifically include unusual sensory behaviors. In order to meet criteria for Autism spectrum disorder under the proposed DSM-5 revision, a child must meet all three of the social communication impairments and at least two of the RRBs. According to the DSM-5 Neurodevelopmental Workgroup, the rationale for requiring at least two symptom manifestations from RRBs is to improve specificity (American Psychiatric Association 2011).
In a recent press release, the American Psychiatric Association asserted that this improvement in specificity is not at the expense of sensitivity. Specifically, they stated that data from DSM-5 field studies do not indicate any reduction in the number of people “receiving care for Autism spectrum disorders in treatment centers” (American Psychiatric Association 2012). However, the DSM-5 criteria are more stringent than DSM-IV-TR, requiring a minimum of five out of seven symptoms to be present. This is compared to the lowest threshold for a diagnosis of an ASD under DSM-IV-TR criteria which is currently evidence of social impairment (without any specific symptom count) and either communication impairment or RRBs or even sub-threshold presentation across all three areas. Therefore, there has been some concern about the impact of the new criteria on diagnostic outcomes.
Several recent studies have indicated that the proposed DSM-5 criteria will have a significant effect on diagnostic rates for autism i.e. that individuals who would meet criteria under DSMIV-TR will no longer meet criteria under proposed DSM-5. McPartland et al. (2012) conducted a re-analysis of the 933 participants evaluated for possible pervasive developmental disorder during the DSM-IV field trials. When they applied the proposed DSM-5 criteria to this group, they found that 60.6 % of individuals who met criteria under DSM-IV would meet DSM-5 criteria but 39.4 % of those who met criteria under DSMIV would no longer meet criteria for a diagnosis of an Autism spectrum disorder. They found that specificity of the proposed DSM-5 was very high, with a true negative rate of 94.9 %. When comparing cases that continued to meet and failed to meet DSM-5 criteria, there were no differences in gender or age. However, there was an effect for intellectual ability and DSMIV subcategory i.e. individuals with an intellectual disability were more likely than those without an intellectual disability to meet DSM-5 criteria and individuals with Autistic Disorder were more likely to meet DSM-5 criteria than those with Asperger’s Disorder or PDD-NOS.
The original information for this re-analysis was gathered by clinicians guided by DSM-IV so it is possible that clinicians did not gather and/or record all information that would be considered pertinent to DSM-5 criteria. Specifically, they would not have included questions and/or investigation of symptoms that will be included in DSM-5 but which are not included under DSM-IV-TR criteria (e.g. indifference to pain/heat/cold, deficits in understanding nonverbal communication). Nevertheless, these results suggest that the proposed DSM-5 criteria may significantly reduce the numbers of individuals diagnosed with ASD. Similar findings in terms of reduced sensitivity for DSM-5 compared to DSM-IV criteria were also found in an earlier epidemiological study in Finland (Mattila et al. 2011). In this study, only 46 % of cases with an IQ ≥ 50 that had been identified according to DSM-IV-TR criteria met criteria under the DSM-5 draft criteria. It is important to note, however, that this study used the DSM-5 criteria posted by the APA in February 2010, which has since undergone some refinement.
In another recent study, Worley and Matson (2012) used checklists to compare ASD symptoms in 121 children classified according to both DSM-IV-TR and DSM-5 diagnostic criteria. They found that there was a subset of children (59 out of 180 or 32.7 %) who met diagnostic criteria for an ASD under DSM-IV-TR but no longer met criteria under DSM-5. Furthermore, they found no significant difference in terms of symptom severity between those who met DSM-5 criteria and those who met DSM-IV-TR but not DSM-5 i.e. participants who will not meet the proposed future criteria had similar symptom severity to those who will continue to meet criteria. It is important to note that this study was based solely on parent-informant checklists rather than on a comprehensive diagnostic assessment. Despite this limitation, this study also provides some support for the prediction that fewer individuals will meet the proposed DSM-5 Autism spectrum disorder criteria.
There is, therefore, some uncertainty around the impact of the proposed DSM-5 changes to the ASD diagnostic criteria. While some preliminary research findings suggest a reduction in the number of individuals who will meet ASD criteria under DSM-5, the APA have indicated that the field studies have not found any such reduction (American Psychiatric Association 2012). Similarly, Frazier et al. (2012) found similar sensitivity and improved specificity for proposed DSM-5 criteria compared to current DSM-IV-TR. Thus, the present study aimed to further examine the impact of the proposed revision by comparing diagnostic outcomes for children undergoing comprehensive ASD assessment by trained clinicians cognizant of the proposed DSM-5 criteria as of January 2011 (American Psychiatric Association 2011). The primary research question under consideration was whether the same clinician, faced with the same information about a particular child, would come to the same diagnostic conclusion under both DSM-IV-TR and DSM-5 criteria. In addition, the researchers aimed to explore the reasons for any discrepant outcomes under the two diagnostic criteria.
The sample consisted of 132 children and adolescents referred to a tertiary-level autism specific assessment service for an initial diagnostic assessment for autism. Participants ranged in age from 2 to 16 years (M = 6.06 years, SD = 3.38 years). Eighty-one percent of participants were male (n = 107) and 19 % were female (n = 25).
The Autism Diagnostic Observation Schedule (ADOS; Lord et al. 2002) is a semi-structured assessment tool that measures communication ability, social interaction, and play or imaginative skills. The ADOS consists of standardized activities that allow the clinician to observe behaviors that have been identified as important in the diagnosis of Autism spectrum disorders at different developmental levels and chronological ages. The ADOS has four modules. The clinician selects the module that is most appropriate for the particular child or adult based on their chronological age and expressive language level. Administration of the ADOS took approximately 45 min. Notes are taken during the administration of the ADOS and the clinician assigns ratings at the completion of the assessment. These ratings assist in the formulation of diagnosis through the use of the diagnostic algorithm provided for each module.
The Autism Diagnostic Interview-Revised (ADI-R; Rutter et al. 2003) is a standardized semi-structured interview that is administered to parents/caregivers regarding the developmental history and current behavior of the individual being assessed. The ADI-R consists of 93 items and focuses on three function domains (language and communication, reciprocal social interaction, and RRBs). The administration time for each interview was approximately 2 h. The ADI–R produces an algorithm which assists in determining whether or not an individual meets criteria for a diagnosis of an ASD.
The study protocol was approved by Autism Spectrum Australia’s Ethics Committee. Informed consent was obtained from parents prior to participation in the study. The research was conducted during a day-long assessment for possible Autism spectrum disorder.
The assessment consisted of informal observations, the ADOS, and the ADI-R. The relevant ADOS module was administered and, in the case of all children, an ADI-R was completed with their parents. Assessments were conducted by Psychologists and Clinical Psychologists of a specialized autism assessment service. All clinicians have been trained in the use of the ADOS and ADI-R for research purposes and administer them as a routine component of all assessments.
Firstly, diagnosis was made according to DSM-IV-TR criteria, with children being classified as having Autistic Disorder, Asperger’s Disorder, PDD-NOS, or non-ASD. The diagnostic decision was based on clinical judgment, which was informed by scores from the ADOS and ADI-R, along with information gathered from background reports and teachers or other professionals. This included results of previous cognitive assessments and/or information regarding current level of academic functioning, which were used to infer general cognitive ability for the purposes of establishing the appropriateness of an Asperger’s Disorder diagnosis. The individual’s presentation was then considered, utilizing the same information (i.e. ADOS, ADI-R, previous assessment reports and teacher background information) according to the proposed DSM-5 revision for Autism spectrum disorders by the same clinician and a diagnosis of ASD or non-ASD was applied. Age of onset was determined from ADI-R items pertaining to first concerns and onset as perceived with hindsight.
Thus for each child, the same clinician using the same information recorded two diagnostic outcomes i.e. one using DSM-IV-TR criteria followed by one using proposed DSM-5 criteria. Where clinicians noted discrepancy between DSM-IV-TR and DSM-5 outcomes, notes were recorded outlining why the child did not meet criteria under DSM-5.
The three ASD diagnostic categories under DSM-IV-TR were collapsed into one group (ASD). Those who did not receive a diagnosis under DSM-IV-TR were placed into another group (non-ASD). These DSM-IV-TR diagnostic outcomes were then compared with DSM-5 diagnoses. Participants were then divided into two groups—those cases where diagnostic outcome on DSM-IV-TR was different to outcome on DSM-5 and those cases who had the same outcome on DSM-IV-TR and DSM-5. An independent samples t test was then utilized to explore the relationship between the groups and age, and a Chi-square test was used to explore the relationship between gender and the two groups. Sensitivity was calculated as the proportion of children who met criteria under DSM-IV-TR who also met criteria according to DSM-5. Specificity was calculated as the proportion of children who did not meet criteria under DSM-IV-TR and who also did not meet criteria under DSM-5. Notes by the clinician were also examined to provide further information regarding the reasons for discrepant outcomes on DSM-IV-TR and DSM-5.
DSM-IV-TR and DSM-5 diagnostic outcomes
n (N = 132)
% of those assessed
N (N = 132)
% of those assessed
ASD (all above)
Autism spectrum disorder
Comparison of Diagnostic Outcomes Under DSM-IV-TR and DSM-5
Breakdown of discrepant outcomes in DSM-5 by DSM-IV-TR diagnostic category
Sensitivity and specificity of DSM-5
n = 21
n = 0
n = 26
n = 85
In an attempt to understand why children who met criteria for an ASD under DSM-IV-TR failed to meet criteria under DSM-5, the notes of the clinician at the time of the assessment were examined. The notes indicated that, of the 26 children who did not retain their DSM-IV-TR diagnosis under DSM-5, 14 failed to meet criteria under DSM-5 due to insufficient evidence of impairment in RRBs (i.e. only satisfying 1 of the 4 criteria). The remaining 12 cases were sub-threshold in terms of the DSM-5 social communication domain, with the majority of this group (eight children) displaying largely intact use of nonverbal behaviors, both currently and according to historical record. The final four cases did not meet full criteria for either deficits in social emotional reciprocity or developing and maintaining relationships as set out in DSM-5.
In this sample of 132 children referred for comprehensive autism assessment, 26 of the 111 (23.4 %) participants who received a diagnosis under DSM-IV-TR would not meet criteria as proposed in DSM-5. Similar to the findings of McPartland et al. (2012) the majority of those who would miss out under the revised criteria came from the current PDD-NOS group with 50 % failing to meet new criteria. The discrepancy rates were lower for Asperger’s Disorder and Autistic Disorder. For Asperger’s Disorder 16.6 % of those who received a diagnosis under DSM-IV-TR would not meet criteria under DSM-5. For Autistic Disorder the proportion of those with discrepant outcomes was 10.2 %. There was no relationship between the child’s age or gender and change in diagnostic status between DSM-IV-TR and DSM-5. Overall, specificity (or true negatives) for DSM-5 was very high at 1.0, however this was at some cost to sensitivity (or true positives) at 0.76.
When examining the reasons for discrepant outcomes between DSM-IV-TR and DSM-5, 46 % of the children who received a diagnosis under DSM-IV-TR but not DSM-5 failed to satisfy all three of the social communication impairments. These children demonstrated either relatively intact non-verbal behaviors or were sub-threshold in terms of meeting the DSM-5 criteria related to their ability to develop and maintain relationships. The most common reason for discrepancy in diagnostic outcome was the failure to meet the set requirement of at least two of the four RRBs, with 54 % of children who did not retain their DSM-IV-TR diagnosis only satisfying one of the criteria. In some cases this was despite presenting with a number of unusual behaviors such as lining up objects, hand flapping and stereotyped utterances. Under the proposed criteria all of the above-mentioned behaviors are included in just one of the four RRBs criteria, i.e. stereotyped or repetitive speech, motor movements, or use of objects.
The implication is that children with marked social and communication difficulties accompanied by several stereotypical behaviors may only meet criteria for one of the four in the RRBs and therefore not meet full criteria for an Autism spectrum disorder diagnosis under DSM-5. Therefore, children with clear support needs that are very akin to an ASD may have difficulty accessing the necessary funding, supports, and early intervention programs. This is consistent with Worley and Matson’s (2012) findings of similar symptom severity in children who will meet future criteria compared to those who meet current, but not future, criteria.
Minor changes to the proposed DSM-5 diagnostic criteria could mean that some of the children whose presentation is similar to that outlined above would be able to meet all of the proposed criteria and, consequently, access appropriate supports. For instance, our study suggests that at least some children who just fail to meet full DSM-5 criteria would meet criteria with a reduction to the requirement for at least two RRBs down to one RRB. Alternatively, the creation of one additional RRB criteria option by separating out some of the multiple behaviors that are currently grouped together (e.g. separating out repetitive use of objects from stereotyped language and motor movements) would allow for additional diagnoses.
These findings would indicate that, without any further changes, DSM-5 is likely to reduce the number of children who will be diagnosed with an ASD in the future due to the more stringent requirements in terms of the number of criteria that must be met. Results from this study suggest that the children likely to fall short of meeting full DSM-5 criteria are those who met DSM-IV-TR criteria for PDD-NOS. The findings highlight the need for comprehensive assessment, with multiple sources of information including clinical observation, and information from parents, teachers and other professionals involved with the child. This is to ensure that adequate information is obtained in order to satisfy the higher proportion of criteria that will need to be met under DSM-5.
The rationale behind many of the changes in DSM-5 is to improve diagnostic consistency and increase the stability of ASD diagnoses over time. It is possible that many of the children who would no longer meet ASD criteria under DSM-5 would instead meet criteria for a diagnosis of the proposed DSM-5 Social Communication Disorder (SCD). SCD is conceptualized as a communication disorder and is characterized by impairments in pragmatics. The proportion of children in this study who may have met criteria for SCD was not examined as the criteria for SCD have not yet been clearly operationalized. It could be that this group is, in fact, quite distinct from those who will meet formal ASD criteria in DSM-5 with meaningful differences in outcome and intervention. Alternatively, this group may actually be more closely aligned to those who meet the full DSM-5 criteria. Future research should aim to examine the similarities and differences between Autism spectrum disorder as defined in DSM-5 and the proposed Social Communication Disorder. It is important to examine this, as it would have implications for the development and implementation of effective, targeted intervention programs.
There are several limitations to this study that should be acknowledged. Firstly, there are a number of behaviors that have long been associated with ASDs that were not formally outlined in DSM-IV-TR and are now included in DSM-5, e.g. difficulties understanding non-verbal behaviors, hypersensitivity to sensory input, indifference to pain/heat/cold. The assessment tools utilized in this study are based on DSM-IV-TR criteria. Therefore, information relating to these newer behaviors may not have been elicited during the assessment process and may have resulted in an over-estimate of the numbers of children who would miss out on a DSM-5 diagnosis of ASD. However, clinical experience is that even without direct questioning, when these behaviors are evident in a child’s presentation, they are often volunteered by parents at some point during the assessment process. Nevertheless, future research into the effects of the proposed criteria should specifically probe for new behaviors included in DSM-5. Secondly, the nature of the sample may limit the generalizability of the results. That is, as a tertiary referral service many of the cases represent what may be considered subtler presentations and/or presentations that may be complicated by the presence of significant co-morbidities. It could be that the proportion of those who would not meet criteria for a DSM-5 diagnosis may be less if cases are taken across the wider community. Further research utilizing a larger sample size across a wider community sample would provide more information. Finally, consideration of a diagnosis under DSM-5 was always conducted after reaching a decision according to DSM-IV-TR in this study. Future studies could evaluate whether arriving at a decision according to DSM-5 prior to DSM-IV-TR would affect results.
In conclusion, our findings indicate that there will be some children who will no longer meet criteria for an ASD under DSM-5 despite meeting DSM-IV-TR criteria. This will most likely be those who would currently be diagnosed with PDD-NOS. As these children do have significant social communication difficulties, it is possible that they would meet criteria for SCD. Future research is required to determine the impact of changes to the diagnostic criteria on prevalence rates for ASD, funding and treatment implications, diagnostic practices and comparative outcomes for children with ASD and SCD.
This research was presented at the Asia Pacific Autism Conference (APAC) which was held in Perth, Australia from 8th to 10th of September 2011.
- American Psychiatric Association. (2011). A 09 Autism Spectrum Disorder. Retrieved November 23, 2011 from http://www.DSM-5.org/Documents/12-03%20Autism%20Spectrum%20Disorders%20-%20DSM-5.pdf.
- American Psychiatric Association. (2012). DSM-5 proposed criteria for autism spectrum disorder designed to provide more accurate diagnosis and treatment. Press release. American Psychiatric Association. http://www.DSM-5.org/Documents/12-03%20Autism%20Spectrum%20Disorders%20-%20DSM-5.pdf. Accessed 15 Feb 2012.
- First, M. B. (2008). Autism and other pervasive developmental disorders. Conference Paper. “The Future of Psychiatric Diagnosis: Refining the Research Agenda” administered by the American Psychiatric Institute for Research and Education. Sacramento, California, February 3–5, 2008. http://www.DSM-5.org/Research/Pages/AutismandOtherPervasiveDevelopmentalDisordersConference(February3-5,2008).aspx. Accessed 29 Nov 2011.
- Lord, C., Rutter, M., DiLavore, P. C., & Risi, S. (2002). Autism diagnostic observation schedule (ADOS). Los Angeles, CA: Western Psychological Services.Google Scholar
- Mattila, M., Kielinen, M., Linna, S., Jussila, K., Ebeling, H., Bloigu, R., et al. (2011). Autism spectrum disorders according to DSM-IV-TR and comparison with DSM-5 draft criteria: An epidemiological study. Journal of the American Academy of Child and Adolescent Psychiatry, 50, 583–592.PubMedCrossRefGoogle Scholar
- Mordre, M., Groholt, B., Knudsen, A., Sponheim, E., Mykletun, A., & Myhre, A. (2011). Is long term prognosis for pervasive developmental disorder not otherwise specified different from prognosis for autistic disorder? Findings from a 30 year follow up study. Journal of Autism and Developmental Disorders. doi:10.1007/s10803-011-1319-5.Google Scholar
- Rutter, M., Le Couteur, A., & Lord, C. (2003). Autism diagnostic interview-revised (ADI-R). Los Angeles, CA: Western Psychological Services.Google Scholar