NIH Public Access

NIH Public Access
Author Manuscript
J Autism Dev Disord. Author manuscript; available in PMC 2015 October 01.
Published in final edited form as:
NIH-PA Author Manuscript
J Autism Dev Disord. 2014 October ; 44(10): 2400–2412. doi:10.1007/s10803-012-1719-1.
Standardizing ADOS Domain Scores: Separating Severity of

Social Affect and Restricted and Repetitive Behaviors
Vanessa Hus,
Department of Psychology, University of Michigan, 530 Church Street, Ann Arbor, MI 48109, USA
vhus@umich.edu
Katherine Gotham, and

Vanderbilt Kennedy Center, Vanderbilt University, Nashville, TN, USA
Catherine Lord
Center for Autism and the Developing Brain, Weill Cornell Medical College, White Plains, NY,
USA
Abstract
Standardized Autism Diagnostic Observation Schedule (ADOS) scores provide a measure of
autism severity that is less influenced by child characteristics than raw totals (Gotham et al. in
Journal of Autism and Developmental Disorders, 39(5), 693–705 2009). However, these scores
combine symptoms from the Social Affect (SA) and Restricted and Repetitive Behaviors (RRB)
domains. Separate calibrations of each domain would provide a clearer picture of ASD
dimensions. The current study separately calibrated raw totals from the ADOS SA and RRB
domains. Standardized domain scores were less influenced by child characteristics than raw
domain totals, thereby increasing their utility as indicators of Social-Communication and
Repetitive Behavior severity. Calibrated domain scores should facilitate efforts to examine
trajectories of ASD symptoms and links between neurobiological and behavioral dimensions.
Keywords
Autism spectrum disorders; Autism Diagnostic Observation Schedule; Severity; Social Affect;
Restricted and Repetitive Behaviors
Introduction
The search to elucidate underlying biological mechanisms which cause or increase risk for
autism spectrum disorders (ASD) has been made more complicated by the marked
phenotypic heterogeneity associated with this developmental disorder (State and Levitt
© Springer Science+Business Media New York 2012

At the start of this study, Vanessa Hus, Katherine Gotham, and Catherine Lord were at the University of Michigan, Department of
Psychology and University of Michigan Autism & Communication Disorders Center in Ann Arbor, Michigan.
Vanessa Hus has remained at the University of Michigan, Department of Psychology. Katherine Gotham is now at the the Vanderbilt
Kennedy Center, Nashville, Tennessee. Catherine Lord is now at the Center for Autism and the Developing Brain, Weill Cornell
Medical College, White Plains, New York.
Conflict of interest C. Lord receives royalties for the ADOS; profits from this study were donated to charity.
Hus et al. Page 2
2011). Diagnostic criteria focus on the presence or absence of specific behaviors or

impairments in three domains: Communication, Reciprocal Social Interaction, and
Restricted and Repetitive Stereotyped Behaviors and Interests (American Psychiatric
Association 2000; World Health Organization [WHO] 1992). However, ASD symptoms
within each domain vary considerably in type and severity, depending upon an individual’s
age, language level, and IQ.
Current nosology attempts to capture some of this variation through categorical diagnoses
(e.g., Autistic Disorder, Asperger’s Disorder and Pervasive Developmental Disorder, Not
Otherwise Specified; APA 2000). However, research has demonstrated that differentiations
made between ASD subgroups are often not reliable across different sites (Lord et al.2011).
In addition, in several studies, items reflecting social and communication impairments
comprised a single factor on ASD diagnostic instruments (e.g., Frazier et al. 2012; Gotham
et al. 2007). In light of these findings, proposals for DSM-5 and ICD-11 call for subgroups
to be subsumed into a single category of ASD defined by two behavioral domains: Social/
Communication Deficits and Fixated or Restricted Interests and Repetitive Behaviors (APA
2011, WHO 2012). Several initial studies support these proposed changes (Frazier et al.
2012; Huerta et al. in press, Mandy et al. 2012, though see Mattila et al. 2011 and
McPartland et al. 2012). To further capture the heterogeneity, criteria for assessing severity
within each domain are recommended.
As these changes are implemented, many of the currently used ASD diagnostic instruments
will need to be revised to more accurately reflect new DSM-5 and ICD-11 criteria, both to
inform diagnosis and to describe severity of symptoms within each behavioral domain. For
example, the diagnostic algorithm of the Autism Diagnostic Interview—Revised (ADI-R;
Rutter et al. 2003), a widely-used parent interview in autism research, is divided into three
domains reflecting the current DSM-IV and ICD-10 criteria for Autistic Disorder, whereas
the Social Responsiveness Scale (SRS; Constantino and Gruber 2005), a caregiver
questionnaire, relies on a single total score for diagnostic classification. In contrast, the
Autism Diagnostic Observation Schedule, Second Edition (ADOS-2; Lord et al. 2012), a
clinician-administered observational assessment, has recently revised diagnostic algorithms
that comprise two behavioral domains [referred to as Social Affect (SA) and Restricted and
Repetitive Behaviors (RRB)] and provide cut-offs for ASD classification (Gotham et al.
2007). In addition, total scores from the revised ADOS algorithms have been standardized to
provide a continuous measure of overall ASD symptom severity that is less influenced by
child characteristics, such as age and language skills, than raw totals [Calibrated
SeverityScores (CSS); Gotham et al. 2009]. These scores can be used to compare ASD
symptom severity across individuals of different developmental levels. As such, they
provide a “purer” metric of overall ASD severity than raw totals from the ADI-R and SRS,
for which studies have demonstrated strong influences of child characteristics, such as age,
language level, and non-ASD specific behavior problems (e.g., Constantino et al. 2003; Hus
et al. in press; Hus and Lord in press).
Although the ADOS-CSS may provide some advantages over these other measures of
general ASD severity, the nature of the symptoms underlying an individual’s CSS may vary
greatly. For example, an ADOS-CSS of 10, indicating the highest level of severity, may be
Hus et al. Page 3
assigned to a child with very significant social-communication impairments who exhibits

few repetitive behaviors during the ADOS. The same score may also be assigned to a child
who has moderate levels of impairments in both domains or very high levels of repetitive
behaviors and more subtle social-communication difficulties. Social-communication

difficulties often pertain to a “lack” of typical behaviors that are pervasive across social
contexts, such as reduced use of gestures or eye contact or reduced frequency of appropriate
social responses, making them more easily-observable during brief interactions. In
comparison, RRBs are often characterized by the presence of an abnormal behavior, such as
hand flapping, sensory examination of materials or excessive references to a particular topic.
Because RRBs may only occur in particular conditions (e.g., hand flapping when a child is
very excited or prolonged discussion of a topic only if it is raised), it is more difficult to
assess them in a short period of time. Therefore, it is important to acknowledge that, when
assessing and comparing symptom severity in different domains, the ADOS as a source of
information, particularly about RRBs, is limited by both time and context. While the
presence of RRBs during this brief observation may be clinically significant, the absence of
these behaviors in this time-limited, standardized context must be interpreted more
cautiously. Nevertheless, research has suggested that both social-communication and
repetitive behaviors measured by the ADOS are surprisingly good predictors of diagnosis
(e.g., Lord et al. 2006).
Separate calibration of these distinct domains is needed to provide a clearer picture of ASD
severity. For example, calibrated domain scores would allow for examination of two
dimensions (SA and RRB), which may have distinct developmental trajectories or respond
differently to intervention. In large samples, researchers could use estimates of social-
communication and repetitive behavior severity to increase phenotypic homogeneity by
clustering individuals according to similar levels of severity in each domain (e.g., high SA
and RRB; high SA and low RRB, etc.). In smaller studies that cannot afford the loss of
power resulting from sample stratification, researchers might use continuous scores to
statistically control for differences in one domain while focusing on the other. Separately
calibrated domain scores may also be useful in genetic and neurobiological studies seeking
to draw associations between biological mechanisms and specific behavioral domains, many
of which currently rely on raw domain totals (e.g., Dichter et al. 2011). While some studies
have controlled for effects of age or IQ in individual samples (e.g., Di Martino et al. 2011),
use of calibrated scores may facilitate comparisons across samples comprised of individuals
of varying developmental levels.
The goal of the current study was to separately calibrate raw totals from the ADOS SA and
RRB domains to reduce the effects of child characteristics and increase the utility of these
scores as continuous measures of social-communication and of repetitive behavior symptom
severity.
Methods
Participants
For comparability, the same sample used to standardize the overall ADOS total (see Gotham
et al. 2009) was also employed to calibrate separate severity metrics for the Social Affect
Hus et al. Page 4
(SA) and Restricted, Repetitive Behavior (RRB) domains. Briefly, this included data from
1,415 individuals ranging in age from 2 to 16 years. With repeated assessments for 25 % of
the sample, data from 2,195 ADOSes with contemporaneous best estimate clinical diagnoses
were available for analysis. Of these assessments, 1,786 cases were given an autism
spectrum disorder diagnosis (ASD; 1,187 Autistic Disorder, 599 Other-ASD) and 409 had a
Non-ASD diagnosis. Non-ASD diagnoses included language disorders (27 %), nonspecific
intellectual disability (20 %), Down syndrome (14 %), oppositional defiant disorder or
ADD/ADHD (13 %), mood or anxiety disorders (8 %), Fetal Alcohol Spectrum Disorders (7
%), other genetic or physical disabilities, such as Fragile X or mild cerebral palsy (6 %) and
early developmental delays (5 %).
Individuals were consecutive referrals to specialty clinics in Ann Arbor, Michigan and
Chicago, Illinois, and participants in research studies conducted through the University of
North Carolina—Chapel Hill, University of Chicago, and University of Michigan. All
participants provided informed consent and all procedures related to this project were
approved by institutional review boards at the University of Chicago or University of
Michigan. Sample characteristics are provided in Table 1.
Procedure
The ADOS was conducted as part of a clinical or research evaluation (see Gotham et al.
2009 for more detailed procedures). All ADOSes were administered and scored by a clinical
psychologist or trainee who met standard requirements for research reliability. The Pre-
Linguistic ADOS (PL-ADOS; DiLavore et al. 1995) was given in 418 (19 %) assessments
and a pilot version of the ADOS-Toddler (Luyster et al. 2009) was given in 82 assessments
(4 %). For both measures, scores from items identical to those in the Module 1 algorithms
were used. Verbal and/or nonverbal IQ scores were available for 2009 (92 %) assessments.
These were derived from a developmental hierarchy of cognitive measures (see Lord et al.
2006), most frequently the Mullen Scales of Early Learning (Mullen 1995) and the
Differential Ability Scales (Elliott 1990). Best estimate clinical diagnoses were made by a
supervising clinical psychologist and/or a child psychiatrist after review of all assessment
data (including, at a minimum, the ADOS and cognitive scores).
Standardization of Raw Totals

Calibration of each domain began by following a similar procedure to that described for
standardization of overall ADOS totals (Gotham et al. 2009). Only assessments from
individuals with ASD were used for raw domain total standardization. This included all
assessments with a corresponding best estimate clinical diagnosis of autism or Other-ASD,
as well as data from 13 individuals who had ADOS data with a contemporaneous Non-ASD
diagnosis but who were later diagnosed with ASD (total n = 1,807 assessments from 1,118
individuals). Participants were first divided into the 18 age/language groups used for the
calibration of the overall raw totals. SA and RRB scores were compared separately for each
1-year chronological age group within a given cell to ensure that distributions of the domain
scores were comparable. Some of the 18 cells were then collapsed due to comparable
distributions (likely due to the reduced range of scores in each domain compared to the
overall totals). This resulted in 12 age/language cells (See Fig. 1; note that the raw total-to-
Hus et al. Page 5
calibrated score mapping for the RRB domain could have been further collapsed into two
Module 2 cells, 2–3 year olds and above 4 years; however, these were left expanded across 4
cells so that both domains would have the same number of cells).
In the overall total calibration, ADOS diagnostic classifications were used to anchor raw
totals to ranges of severity scores. That is, raw totals corresponding to an ADOS
classification of “Autism” were mapped on to CSS of 6–10, “ASD” to CSS of 4–5 and
“Nonspectrum” to CSS of 1–3. This was done to make the metric more generalizable to
other samples, as we cannot assume that the datasets used for calibration in all
developmental cells were representative of the heterogeneous ASD population. Next, the
range of raw totals assigned to each point on the 10-point severity scale was determined by
the percentiles of available data within that classification range (Gotham et al. 2009).
Because there are not separate SA cut-offs for “Autism” and “ASD” classifications, the
same percentiles used for mapping raw ADOS totals (i.e. SA + RRB) to the 10-point scale
were used to inform the mapping of raw SA totals to SA-CSS within each of the 12 age/
language cells. Raw total-to-calibrated score mappings were then adjusted so that, for each
of the 5 diagnostic algorithm groups (Gotham et al. 2007), sensitivity for individuals
receiving an ADOS classification of “Autism” and an SA-CSS greater than or equal to 6
was, if possible, at or above 90 %. Within algorithm groups, the lowest individual cell
sensitivity was .89 for Module 2, 2–3 year olds. A goal of 80 % sensitivity across algorithm
groups was set for individuals with an “Autism Spectrum” ADOS classification and an SA-
CSS of 4 or higher. Sensitivity for individual developmental cells within algorithm groups
was sometimes lower in groups with few participants; however, considering cells with
greater than 20 participants, only Module 3, 3–5 year olds (n = 59) fell just below this
threshold, with a sensitivity of .78. Finally, adjustments were made to ensure that specificity
(individuals with a “Nonspectrum” ADOS classification and SA-CSS less than or equal to 3)
was, if possible, at least 80 % for each algorithm. Within algorithms, only the Module 2, 5–6
year old cell fell below this threshold, with a specificity of .76.
Because the RRB domain is limited to a range of 9 points (0–8), it was not possible to use
all 10 points in the severity metric for this domain. However, given concerns that SA- and
RRB-CSS scores may be misinterpreted if they are not on a comparable scale, it was
decided to maintain the full 10-point range and have some points on the severity scale for
which no raw scores were assigned. Thus, as with SA-CSS, percentiles from mapping of raw
overall totals were used to inform mapping of raw RRB totals to the calibrated metric. This
resulted in the raw RRB totals mapping on to CSS values of 5–10. These distributions were
skewed compared to the overall and SA-CSS scales and reflect the trade-off in using the
ADOS as a measure of RRBs: while a lack of RRBs is difficult to interpret, the presence of
RRBs during this brief observation is more meaningful as an indication of greater severity.
Given the lower sensitivity of repetitive behaviors in the limited context in which they may
be observed during the ADOS, a goal of 80 % sensitivity was set for individuals receiving
an ADOS classification of “Autism” and RRB calibrated scores of 6 or greater; Module 3,
2–5 year olds fell just below this threshold with a sensitivity of 77 %. No sensitivity
threshold was set for individuals with an “Autism Spectrum” classification. A goal of 80 %
specificity was set for scores less than or equal to 6. For individual cells with greater than 20
participants, the lowest specificity was 79 % for Module 3, 6–16 year olds.
Hus et al. Page 6
Table 2 shows the mappings of raw SA and RRB totals to the 10-point severity scale for
each of the 12 calibration cells.
Associations Between Participant Characteristics, Raw Domain Totals and Calibrated

Domain Scores
Following procedures in Gotham et al. 2009, separate linear regression analyses were
conducted using the sample of participants with ASD who had contemporaneous
demographic data (N = 1,369) to examine the influences of child characteristics on raw
domain totals and calibrated domain scores. The child’s verbal and nonverbal IQs and
mental ages were entered into the first block, followed by child chronological age, gender,
maternal education and race in the second block. Only model R2 are reported because
interpretation of the meaning of these individual coefficients is limited by multicollinearity.
Next, significant predictors were entered into Forward Stepwise models to assess the relative
contributions of these variables in predicting raw domain totals and calibrated domain
scores. (Results from analyses including Non-ASD participants are available from authors.
Consistent with the results for the participants with ASD, when applied to the entire
clinically-referred sample, standardized severity scores were less influenced by participant
characteristics than were raw domain totals.)
Results
Comparison of Raw Domain Totals and Calibrated Domain Scores by Calibration Cell
As shown in Table 3 and Fig. 2a, c, distributions of raw SA and RRB domain totals varied
significantly by age/language group. Across algorithms reflecting different language levels,
individuals with less language had higher scores than those who were more verbally fluent.
Within algorithm groups, older children and adolescents tended to have higher scores than
toddlers and young children. In contrast, calibrated SA and RRB domain scores were more
comparable across calibration cells, though not uniform (see Table 3 and Fig. 2c, d).
Notably, children who were verbally fluent (i.e., Module 3) have a wider distribution of
RRB-CSS scores compared to children of other language levels. This reflects the somewhat
larger proportion of verbally fluent children (8.5–12.9 %) that did not have repetitive
behaviors during the ADOS (i.e., received a RRB-CSS of 1).
As noted above, ADOS classifications, which are based on raw overall totals (SA + RRB)
were used to anchor the raw total-to-overall severity score mappings for the domains to
specific calibrated score ranges (e.g., “Autism” to CSS of 6–10). Using percentiles from the
raw total-to-overall CSS mapping to inform raw domain totals-to-domain severity score
mappings, mean SA-CSS and RRB-CSS also distinguished between individuals grouped by
clinicians’ best estimate clinical diagnoses (i.e., Autism vs. Other-ASD vs. Non-ASD
diagnoses; SA-CSS: F(2,2192) = 974.43, p ≤ .001; RRB-CSS: F(2,2192) = 421.35, p ≤ .
001). Nonetheless, there was marked overlap in the distribution of scores across the three
diagnostic groups (see Fig. 3a, b).
Hus et al. Page 7
Correlations Between Domain Calibrations and Overall Calibrated Severity Score

In the ASD sample, associations between SA-CSS and RRB-CSS were significant, but weak
(r = .25; Cohen 1988). Although correlations between each of the domain calibrated scores
and the overall CSS were both strong, the association between SA-CSS and CSS (r = .89)
was greater than that observed for RRB-CSS and CSS (r = .57). This is a reflection that the
overall total from which the CSS is derived is comprised of a greater proportion of items
from the SA domain than the RRB domain.
Predictors of SA-Raw and SA-CSS

The final model including all predictors explained a total of 45 % of variance in the SA-Raw
total. Verbal IQ and maternal education (mothers with graduate/professional degrees vs. all
others) emerged as significant predictors of SA-Raw. In contrast, the same model accounted
for only 13 % of the variance in the SA-CSS, with verbal IQ and nonverbal IQ both making
small, but significant contributions to the calibrated SA score. Thus, although there is still a
significant association between SA-CSS and the child’s cognitive level, the calibrated SA
scores are markedly less influenced by child cognitive level than SA-Raw.
Next, verbal IQ, nonverbal IQ, and maternal education were entered into a Forward
Stepwise model to assess the relative contributions of each of these variables in predicting
SA-Raw. As shown in Table 4, verbal IQ accounted for the majority of variance (43 %) and
the contributions of nonverbal IQ and maternal education were minimal (0.3 and 0.2 %,
respectively). In the Forward model predicting SA-CSS, verbal IQ accounted for 10.5 % of
variance while nonverbal IQ explained an additional 0.4 %; maternal education was
excluded by the model, indicating that it was not significant (see Table 4). These results
reflect a reduction in the influence of verbal IQ from a large effect on SA-Raw (R = .66) to a
small-to-medium effect on SA-CSS (R = .33; Cohen 1988; McCarthy et al. 1991). It is
noteworthy that verbal and nonverbal IQ were highly correlated (r = .76) and when verbal
IQ was removed as a predictor, nonverbal IQ accounted for 21.8 % of variance in SA-Raw
and only 4.3 % in SA-CSS; both models excluded maternal education as a predictor.
Predictors of RRB-Raw and RRB-CSS

Child characteristics such as IQ explained much less variance in raw RRB totals (i.e., 15.3
%). Significant predictors included verbal IQ, nonverbal IQ, and race (African American vs.
all others). In the Forward Stepwise Model, verbal IQ, nonverbal IQ and race each remained
significant predictors of RRB-Raw (see Table 4). Verbal IQ accounted for the majority of
variance (11.7 %) and nonverbal IQ and race each made small contributions (1.4 and 1.1 %,
respectively). Again, if verbal IQ was excluded from the models, nonverbal IQ explained
11.4 % and race explained 0.8 % of variance in RRB-Raw.)
Calibrated RRB scores reduced the influence of child characteristics; in the end, child
characteristics explained only 5.5 % of the variance, with verbal IQ, nonverbal IQ and race
emerging as small, but significant predictors of RRB-CSS. In the Forward Model predicting
RRB-CSS, nonverbal IQ explained 3.5 % of the variance in RRB-CSS; verbal IQ and race
accounted for an additional 0.5 and 0.6 %, respectively.
Hus et al. Page 8
Case Summaries
Four children with ASD diagnoses were chosen to demonstrate the utility of the newly
calibrated domain scores for separately examining the severity of social and repetitive
behaviors over time (see Table 5 for child characteristics at first and last assessments). Each
child’s SA-CSS and RRB-CSS are plotted by age in Fig. 4. Overall CSS scores are also
provided; in many cases the overall CSS and SA-CSS follow similar, if not identical,
trajectories, again reflecting that the overall total from which the CSS is derived is
comprised of a greater proportion of items from the SA domain than the RRB domain.
Case 1—“Bianca,” a Caucasian female, was diagnosed with autism at 4 years of age when
she was first seen as a clinical referral (see Gotham et al. 2009). Her overall CSS suggests
that her symptom severity was relatively stable across early childhood, followed by gradual
a decrease in severity throughout late childhood and early adolescence. Her SA-CSS follows
a similar trajectory, reflecting persistent difficulties with eye contact and unusual social
overtures accompanied by an increase in use of gestures and shared enjoyment with the
examiner. In contrast, her RRB-CSS follows a quite different pattern, with a RRB-CSS of 10
at Bianca’s first assessment (reflecting her exhibition of sensory-seeking behaviors, delayed
echolalia, repetitive asking of questions and repeated lining up of toys). This was followed
by a considerable decrease in severity at age 5 and a year of relative stability, during which
time she demonstrated some repetitive speech and mild preoccupations with a particular
musician, but no hand and finger mannerisms. Although Bianca did not demonstrate
repetitive behaviors when she was assessed at 8 years old, in early adolescence, she again
exhibited clear hand and finger mannerisms and engaged in somewhat repetitive speech
(though recall that there is not a RRB-CSS of 2–4, so the fluctuation in severity later
childhood may appear greater than it actually was).
Case 2—“Joey,” a Caucasian male, was first seen as a clinical referral at 2 years, 10
months of age, at which time he received a diagnosis of PDD-NOS. When first seen, he
exhibited severe social-communication symptoms (i.e., an SA-CSS of 10 demonstrating
poor eye contact and very limited social overtures), but mild repetitive behaviors (RRB-CSS
of 5 reflecting very brief repetitive behaviors) during the ADOS. In his subsequent
assessments, there was an apparent increase in repetitive behaviors due to his use of
stereotyped language (e.g., “That’s all folks!”), accompanied by an improvement in the
social affect domain (i.e., improvements in eye contact and more frequent and appropriate
overtures). At age 7 years, 7 months, Joey’s SA-CSS of 3 and RRB-CSS of 7 suggested
milder severity of social-communication symptoms compared to repetitive behaviors. His
overall CSS followed a similar trajectory to his SA-CSS, showing a steady decrease in
severity across early childhood, and did not reflect the apparent increase in repetitive
behaviors during this same period.
Case 3—“Carolyn,” a Caucasian female, was first seen as part of a clinical research project
just after her second birthday. At this time, she received a diagnosis of PDD-NOS and her
SA-CSS of 4 suggested milder severity of social-communication impairments during the
ADOS (e.g., strengths in shared enjoyment and facial expressions, but difficulties using
coordinated eye gaze) compared to her RRB-CSS of 9 (reflecting hand and finger, as well as
Hus et al. Page 9
whole-body mannerisms, a preoccupation with cars and brief peering at objects). However,
over the next 8 years, there was a steady increase in deficits in SA, resulting in an SA-CSS
of 10 by the time she was 10 years old; while she continued to express some shared
enjoyment with the examiner, her use of facial expressions was more limited and deficits in
eye contact persisted. Her overall CSS also follows this pattern. In contrast, during the
period in which she had the most dramatic increases in SA-CSS, the severity of Carolyn’s
repetitive behaviors remained relatively stable. Over time, she continued to exhibit hand and
finger and whole-body mannerisms (e.g., twirling and jumping), and brief visual sensory
interests. She also demonstrated unusual preoccupations (e.g., with time), as well as
ritualistic behaviors, such as placing objects in toy trucks in a particular way.
Case 4—“Matthew,” an African American male, was seen at age 4 years as part of a
clinical research study, at which time he received a diagnosis of autism. During his first
ADOS, Matthew exhibited more severe social-communication symptoms (SA-CSS = 8) than
repetitive behaviors (RRB-CSS = 5). Separate examination of his SA-CSS and RRB-CSS
suggest relatively stable severity in both domains across early childhood, marked by
persistent difficulties in nonverbal social communication (e.g., facial expressions and eye
contact), initiation of overtures, brief sensory interests and possible hand and finger
mannerisms. At 11 years of age, Matthew showed an apparent decrease in severity of social-

communication symptoms (a greater range of facial expressions and more reciprocal social
communication) and a worsening of repetitive behaviors, including clear hand and finger
mannerisms, excessive references to Batman and wrestling, repetitive stereotyped questions,
and listing of his classmates when asked the names of his friends. In his case, the overall
CSS showed a gradual worsening of symptom severity between ages 4 and 11, failing to
account for the possible divergence of trajectories in social-communication skills and
repetitive behaviors in later childhood.
Discussion
ADOS calibrated domain totals achieved the goal of significantly reducing associations with
child characteristics compared to raw SA and RRB totals. For SA-Raw domain scores, 45 %
of variance was explained by child characteristics not specific to ASD, with verbal IQ and
maternal education emerging as significant predictors. For the SA-CSS, verbal IQ remained
the only significant predictor, accounting for just under 11 % of variance in the calibrated
SA score. Similarly, approximately 12 % of variance in RRB-Raw Total was explained by

verbal IQ, with nonverbal IQ and race collectively accounting for an additional 3 %. For the
RRB-CSS, nonverbal IQ, verbal IQ and race remained significant predictors, but explained
less than 5 % of variance. Thus, though the effects of child characteristics were not
completely eliminated, the calibrated domain scores provided a measure of ASD severity
that was significantly less influenced by child characteristics, particularly verbal IQ, than
were raw totals.
It is interesting to note that associations between IQ and RRB Raw were much smaller
compared to the relationship between IQ and SA-Raw. A similar difference in associations
with developmental level was noted for Social + Nonverbal Communication vs. Repetitive
Behavior raw domain totals on the Autism Diagnostic Interview-Revised (Hus and Lord in
Hus et al. Page 10
press). The restricted range of RRB-Raw scores may explain the weaker associations.
Nevertheless, in spite of relatively smaller influences of developmental level on RRB-Raw,
it is important to calibrate RRB-CSS in order to provide a comparable severity metric for
both ADOS domains. Most important, the RRB-CSS reduced the influence of
developmental level on RRB totals even further.
It is also noteworthy that there was marked overlap in the distributions of domain calibrated
scores across diagnostic groups. On one hand, the overlap of the Non-ASD group with the
Autism and Other-ASD groups may reflect recruitment biases in our Non-ASD sample,
some of whom were referred for assessment of ASD, but who received a clinical Non-ASD
diagnosis. On the other hand, the overlap between the Autism and Other-ASD group could
reflect that the calibrated scores are capturing the heterogeneity of symptom severity that
characterizes ASDs. Moreover, the overlap with the Non-ASD group highlights that some
social-communication and repetitive behaviors captured on the ADOS are not specific to
ASD.
The newly standardized SA-CSS and RRB-CSS provide useful measures of autism symptom
severity which are consistent with the two symptom domains defining ASD proposed for
DSM-5. As we move toward a single classification of “autism spectrum disorder” in

DSM-5, calibrated domain scores have the potential to play a role in the clinical
specification of ASD severity. When DSM-5 criteria are finalized, assessing the degree to
which the 10-point CSS scale indicating severity of ASD symptomatology relates to
different DSM-5 levels of severity for each behavioral domain (currently proposed as
“requiring support,” “requiring substantial support,” and “requiring very substantial
support”) will be an important step. If the scores can be mapped on to clinical levels of
severity, they may be useful to inform the level of impairment in each behavioral domain;
however, these scores will not be sufficient to make such clinical determinations, as they
provide information about behaviors in a limited context. Information collected from other
modalities of assessment, such as caregiver interview or observation in other settings, will
be needed to inform the appropriate level of severity to describe the level of support an
individual requires.
It is also hoped that the calibration of severity metrics for social-communication deficits
(SA-CSS) and repetitive behaviors (RRB-CSS) will bring us a step closer to parsing apart
the phenotypic heterogeneity in ASD. Current studies frequently rely on totals from
diagnostic instruments such as the SRS or ADI-R as estimates of ASD severity. Yet these
totals are known to be greatly influenced by child characteristics such as age, language level,
and non-ASD-specific behavioral problems (e.g., Constantino et al. 2003; Hus et al. in press;
Hus and Lord in press). Although the original ADOS calibrated severity metric was derived
to reduce the effects of non-specific child characteristics (Gotham et al. 2009), it yields an
estimate of overall severity that does not allow for separate examination of the variation in
behavioral domains underlying these scores. In comparison, the SA-CSS and RRB-CSS
provide more behavioral specificity than each of these general measures. Because potential
biomarkers are frequently postulated to be related to specific domains of behavior (e.g.,
severity of RRBs), separate calibrated domain scores offer an important advance.
Additionally, use of these calibrated domain scores in place of raw totals increases the
Hus et al. Page 11
likelihood that associations with genetic or neurobiological abnormalities are specific to

ASD symptoms rather than associated with general developmental factors, such as age, IQ
or language level.
Using these scores to separately examine distinct trajectories of social-communication and

repetitive behaviors may also provide a more sensitive measure of intervention response
over longer periods of time, enabling change in one domain to be detected, even when
behaviors in the second domain persist. Although children may become more familiar with
particular tasks (e.g., participating in the birthday party routine) if they are administered the
ADOS several times within a short period, because scores are based on spontaneous
initiations and responses, rather than performance on tasks, scores and ADOS classifications
do not demonstrate practice effects (Lord et al. 2012). Thus, the SA-CSS and RRB-CSS may
provide a way to measure more global changes in behaviors in response to intervention,
rather than improvements in very specific skills. Furthermore, different SA-CSS and RRB-
CSS trajectory profiles may provide an additional method of stratification to increase
phenotypic homogeneity in samples, which can be used to gain insight into biological
mechanisms underlying specific developmental patterns.
Limitations
Domain calibrations were based on the large “convenience” sample that was used to create
the overall ADOS CSS (Gotham et al. 2009). As these authors acknowledged, this sample is
likely to be representative of other samples ascertained through North American clinical
research centers over the past two decades. It is hoped that using ADOS classifications of
(i.e., “Autism,” “Autism Spectrum” and “NonSpectrum”), rather than clinical best estimate
diagnoses, to anchor overall severity scores and set thresholds for sensitivity and specificity
of domain calibrated scores would circumvent, to some extent, recruitment effects in this
sample (Gotham et al. 2009). However, it is possible that calibration using ADOSes from
population studies or more recently ascertained samples may result in different mappings of
raw totals and calibrated scores. Additionally, samples recruited outside of North America,
or from other clinical populations, may show a somewhat different distribution of scores.
Here, the effects of maternal education and race observed on both overall raw totals and
calibrated scores are likely to be an artifact of recruitment biases (Gotham et al. 2009),
though the significance of these predictors may also have been influenced by the large
sample size. Replication of the domain calibrations in independent samples is an important

next step.
Given the restricted range of raw RRB totals, the RRB-CSS is not a full 10-point severity
metric. Nonetheless, scores were mapped onto the 10-point scale to avoid confusion when
using the calibrated domain scores together. That is, there was concern that a reduced RRB-
CSS scale (e.g., of 1–6) may result in confusion when interpreting the meaning of an RRB-
CSS score in comparison to a score on the overall-CSS scale (i.e., an assumption that an
RRB-CSS of 6 would be equal to an overall-CSS of 6, when it actually would be more
meaningful to interpret as similar to an overall-CSS of 10). The method of using the overall-
CSS percentiles to inform mapping of domain raw scores to the 10-point calibrated scale
allows comparability across the three scales, such that a given value on the overall-CSS, SA-
Hus et al. Page 12
CSS, and RRB-CSS correspond to approximately the same percentile of raw score (for a
child of that language level and age) for each. Such comparability also increases the clinical
utility of this metric; for example, a child who has a high overall-CSS comprised of an SA-
CSS of ‘10′ and an RRB-CSS of ‘6′ may need a different treatment approach than another
child with the same overall-CSS reflecting an RRB-CSS of ‘10′ and an SA-CSS of ‘6′.
When using scores to monitor change over time or in response to intervention, researchers
and clinicians must bear in mind that there are not RRB-CSS values of 2, 3 or 4. Thus,
changing from a score from RRB-CSS of 1, indicating that no repetitive behaviors were
observed during the ADOS, to 5 (reflecting mild severity), is not the same as a change in
severity from an RRB score of 6–10. This distribution of scores reflects that, given the
limited timeframe of the ADOS, the presence of repetitive behaviors is likely to be more
meaningful than the absence of such. In order to ensure that a change is CSS for either
domain is meaningful, the lower (or higher) score should be observed across several time
points. In contrast, a significant increase or decrease during one particular session may
suggest that other factors were influencing the child’s behavior on that particular day.
Conclusions
ADOS domain calibrations provide separate estimates of severity of ASD-related social-

communication deficits and repetitive behaviors that are relatively independent of child
characteristics, such as age and language skills, compared to their respective raw totals. This
improves their utility as continuous measures of ASD symptom severity that can be used to
increase homogeneity of samples and identify links between specific behavioral domains
and biological mechanisms, as well as to examine different trajectories of ASD symptoms
over time.
Acknowledgments
This research was supported by a Dennis Weatherstone Predoctoral Fellowship to VH and National Institute of
Mental Health grants T32-MH18921 to KG and R01MH081873 and RC1MH089721 to CL. We gratefully
acknowledge Drs. Andrew Pickles, Christopher Gruber and Sheri Stegall for their consultation in preparation of this
manuscript, as well as all of the families who participated in this research.
References
American Psychiatric Association. Diagnostic and statistical manual of mental disorders. Fourth
Edition. American Psychiatric Association; Washington, DC: 2000. Text Revision: DSM-IV-TR
American Psychiatric Association. [Accessed 04 March 2012] DSM-5 Development: A 09 Autism
Spectrum Disorder. 2011. http://www.dsm5.org/Proposed Revisions/Pages/proposedrevision.aspx?
rid=94
Cohen, J. Statistical power analysis for the behavioral sciences. Psychology Press; 1988.
Constantino, JN.; Gruber, C. The Social Responsiveness Scale. Western Psychological Services; Los
Angeles: 2005.
Constantino JN, Hudziak JJ, Todd RD. Deficits in reciprocal social behavior in male twins: evidence
for a genetically independent domain of psychopathology. Journal of the American Academy of
Child and Adolescent Psychiatry. 2003; 42(4):458–467. doi:10.1097/01.CHI.0000046811.95464.21.
[PubMed: 12649633]
Di Martino A, Kelly C, Grzadzinski R, Zuo X-N, Mennes M, Mairena MA, et al. Aberrant striatal
functional connectivity in children with autism. Biological Psychiatry. 2011; 69(9):847–856. doi:
10.1016/j.biopsych.2010.10.029. [PubMed: 21195388]
Hus et al. Page 13
Dichter GS, Richey JA, Rittenberg AM, Sabatino A, Bodfish JW. Reward circuitry function in autism
during face anticipation and outcomes. Journal of Autism and Developmental Disorders. 2011 doi:
10.1007/s10803-011-1221-1.
DiLavore P, Lord C, Rutter M. The Pre-Linguistic Autism Diagnostic Observation Schedule. Journal
of Autism and Developmental Disorders. 1995; 25(4):355–379. doi:10.1007/BF02179373.
[PubMed: 7592249]
Elliott, CD. Differential Ability Scales. Psychological Corporation; San Antonio: 1990.
Frazier TW, Youngstrom EA, Speer L, Embacher R, Law P, Constantino J, et al. Validation of
proposed DSM-5 criteria for Autism Spectrum Disorder. Journal of the American Academy of
Child and Adolescent Psychiatry. 2012; 51(1):28–40. doi: 10.1016/j.jaac.2011.09.021. [PubMed:
22176937]
Gotham K, Risi S, Pickles A, Lord C. The Autism Diagnostic Observation Schedule: Revised
algorithms for improved diagnostic validity. Journal of Autism and Developmental Disorders.
2007; 37(4):613–627. doi:10.1007/s10803-006-0280-1. [PubMed: 17180459]
Gotham K, Pickles A, Lord C. Standardizing ADOS scores for a measure of severity in autism
spectrum disorders. Journal of Autism and Developmental Disorders. 2009; 39(5):693–705. doi:
10.1007/s10803-008-0674-3. [PubMed: 19082876]
Huerta M, Bishop SL, Duncan A, Hus V, Lord C. How many will we miss? Application of DSM 5
criteria to 3 samples of children with DSM-IV diagnoses of ASD. American Journal of Psychiatry.
(in press).
Hus V, Lord C. Effects of child characteristics on the Autism Diagnostic Interview-Revised:
Implications for use of scores as a measure of ASD severity. Journal of Autism and
Developmental Disorders. (in press). doi:10.1007/s10803-012-1576-y.
Hus V, Bishop S, Gotham K, Huerta M, Pickles A, Lord C. Factors influencing scores on the Social
Responsiveness Scale. Journal of Child Psychology and Psychiatry. (in press).
Lord C, Petkova E, Hus V, Gan W, Lu F, Martin DM, et al. A multisite study of the clinical diagnosis
of different autism spectrum disorders. Archives of General Psychiatry. 2011 doi:10.1001/
archgenpsychiatry.2011.148.
Lord C, Risi S, DiLavore PS, Shulman C, Thurm A, Pickles A. Autism from 2 to 9 years of age.
Archives of General Psychiatry. 2006; 63:694–701. doi:10.1001/archpsyc.63.6.694. [PubMed:
16754843]
Lord, C.; Rutter, M.; DiLavore, PC.; Risi, S.; Gotham, K.; Bishop, S. Autism Diagnostic Observation
Schedule. 2nd Edition. Western Psychological Services; Torrance: 2012. (ADOS-2)
Luyster R, Gotham K, Guthrie W, Coffing M, Petrak R, Pierce K, et al. The autism diagnostic
observation schedule—toddler module: a new module of a standardized diagnostic measure for
autism spectrum disorders. Journal of Autism and Developmental Disorders. 2009; 39(9):1305–
1320. doi:10.1007/s10803-009-0746-z. [PubMed: 19415479]
Mandy WPL, Charman T, Skuse DH. Testing the construct validity of proposed criteria for DSM-5
Autism Spectrum Disorder. Journal of the American Academy of Child and Adolescent
Psychiatry. 2012; 51(1):41–50. doi:10.1016/j.jaac.2011.10.013. [PubMed: 22176938]
Mattila M-L, Kielinen M, Linna S-L, Jussila K, Ebeling H, Bloigu R, Joseph RM, et al. 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 & Adolescent Psychiatry.
2011; 50(6):583–592.e11. [PubMed: 21621142]
McCarthy PL, Cicchetti DV, Sznajderman SD, Forsyth BC, Baron MA, Fink HD, et al. Demographic,
clinical, and psychosocial predictors of the reliability of mothers’ clinical judgments. Pediatrics.
1991; 88(5):1041–1046. [PubMed: 1945609]
McPartland JC, Reichow B, Volkmar FR. Sensitivity and specificity of proposed DSM-5 diagnostic
criteria for Autism Spectrum Disorder. Journal of the American Academy of Child and Adolescent
Psychiatry. 2012; 51(4):368–383. doi:10.1016/j.jaac.2012.01.007. [PubMed: 22449643]
Mullen, E. The Mullen Scales of Early Learning. American Guidance Service, Inc; Circle Pines: 1995.
Rutter, M.; Le Couteur, A.; Lord, C. Autism diagnostic interview-revised. Western Psychological
Services; Los Angeles: 2003.
Hus et al. Page 14
State MW, Levitt P. The conundrums of understanding genetic risks for autism spectrum disorders.
Nature Neuroscience. 2011; 14(12):1499–1506. doi:10.1038/nn.2924.
World Health Organization. The ICD-10 classification of mental and behavioral disorder: Clinical
descriptions and diagnostic guidelines. World Health Organization; Geneva: 1992.

World Health Organization. [Accessed 22 September 2012] International Classification of Diseases
11th Revision: Beta draft. 2012. http://apps.who.int/classifications/icd11/browse/f/en
Hus et al. Page 15
Fig. 1.
Age by language level calibration cells. Note. Ns denote the number of ASD assessments
within each cell
Hus et al. Page 16
Fig. 2.
a (top, left) Distributions of raw Social Affect domain totals by age/language cells. b (top,
right) Distributions of calibrated Social Affect domain scores by age/language cells. c
(bottom, left) Distributions of raw Restricted and Repetitive Behavior domain totals by age/
language cells. d (bottom, right) Distributions of calibrated Restricted and Repetitive
Behavior domain scores by age/language cells
Hus et al. Page 17
Fig. 3.
a (left) Distributions of calibrated Social Affect domain scores by best estimate clinical
diagnosis. b (right) Distributions of calibrated Restricted and Repetitive Behavior domain
scores by best estimate clinical diagnosis
Hus et al. Page 18
Fig. 4.
Case summaries of longitudinal domain severity scores
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
Table 1
Sample descriptives
Module 2, 5
Hus et al.
Module 1, no words Module 1, some words Module 2, younger than 5 Module 3

or older
N Mean SD N Mean SD N Mean SD N Mean SD N Mean SD
ASD
Age 551 4.22 2.21 395 4.41 1.99 197 3.78 .78 215 7.82 2.54 428 8.54 2.54
VIQ 522 26.84 14.71 361 52.63 21.80 164 80.80 20.93 199 55.14 19.49 386 95.53 22.97
NVIQ 515 53.16 21.40 358 69.74 21.67 161 92.53 22.82 201 76.60 23.43 383 96.22 22.32
VMA 528 .97 1.49 355 2.32 3.01 163 4.45 6.05 202 4.63 4.60 377 8.46 5.36
NVMA 516 1.98 .83 359 3.05 2.40 158 3.69 1.35 190 5.74 2.25 357 8.23 2.88
SA Raw 551 16.79 2.95 395 13.23 4.44 197 10.44 4.30 215 13.20 4.29 428 9.26 4.37
RRB Raw 551 4.67 2.11 395 4.07 2.07 197 3.90 2.11 215 4.68 2.10 428 2.71 1.87
Non-ASD
Age 60 3.30 1.61 107 3.51 1.60 57 3.67 .62 44 8.00 2.55 141 8.95 2.47
VIQ 57 40.96 18.72 90 68.08 23.74 51 85.33 21.83 44 58.09 19.06 135 91.70 22.29
NVIQ 55 58.80 28.73 89 70.52 23.75 49 92.04 20.46 44 61.93 24.13 136 89.85 22.23
VMA 57 1.15 .47 87 2.30 .69 50 4.72 6.29 43 4.18 1.16 134 8.60 5.13
NVMA 55 1.72 .72 86 2.44 .71 46 3.47 .83 44 4.72 1.39 132 7.92 2.71
SA Raw 60 8.37 5.83 107 4.71 3.91 57 3.56 2.77 44 4.16 3.14 141 3.90 2.95
RRB Raw 60 1.88 1.88 107 1.40 1.49 57 1.49 1.43 44 1.64 1.64 141 .99 1.15
ASD autism spectrum disorder (Autistic Disorder, Aspergers, PDD-NOS); VIQ verbal IQ; NVIQ nonverbal IQ, VMA nonverbal mental age, NVMA nonverbal mental age, SA Raw Social Affect raw total,
RRB Raw Restricted, Repetitive Behaviors raw total, Non-ASD non autism spectrum disorder diagnosis
Page 19
Table 2
Mapping of ADOS raw domain totals onto calibrated severity scores
Domain Calibrated Raw domain totals

Hus et al.
severity score
Module 1; no words Module 1; some words Module 2 Module 3
2 years 3 years 4–14 years 2–3 years 4 years 5–14 years 2–3 years 4 years 5–6 years 7–16 years 3–5 years 6–16 years
Social affect domain 1 0–3 0–3 0–2 0–1 0–1 0–1 0–1 0–1 0–1 0–1 0–2 0–1
2 4–5 4–5 3–5 2–4 2–3 2–3 2–3 2 2–3 2 3 2
3 6–8 6–9 6–9 5 4–5 4–5 4 3–4 4–5 3–4 4 3–4
4 9 10 10 6–7 6–7 6–7 5 5–6 6 5 5 5
5 10–13 11–12 11–12 8 8–9 8–9 6 7 7 6–7 6 6
6 14–16 13–16 13–14 9–11 10–12 10–13 7–8 8–9 8–9 8–10 7–8 7
7 17 17 15–16 12–13 13 14–15 9–10 10–11 10–11 11–13 9–10 8–9
8 18 18 17–18 14–15 14–15 16 11 12–13 12–15 14–15 11–12 10–11
9 19 19 19 16–17 16–17 17–18 12–14 14–15 16 16–17 13–14 12–14
10 20 20 20 18–20 18–20 19–20 15–20 16–20 17–20 18–20 15–20 15–20
Restricted and Repetitive 1 0 0 0 0 0 0 0 0 0 0 0 0
behaviors domain
2
3
4 1 1 1–2 1 1 1 1 1 1 1 1 1
5
6 2 2–3 3 2 2 2–3 2 2–3 2–3 2–3
7 3 4 4 3 3–4 4 3 4 4 4 2 2
8 4 5 5–6 4 5 5 4 5 5 5 3 3
9 5 6 7 5 6 6 5–6 6 6 6 4 4–5
10 6–8 7–8 8 6–8 7–8 7–8 7–8 7–8 7–8 7–8 5–8 6–8
Page 20
Table 3
Domain raw totals and calibrated severity score means and standard deviations by age/language cell (ASD assessments only)
Module Age (years) N SA-Raw SA-CSS RRB-Raw RRB-CSS

Hus et al.
Mean SD Mean SD Mean SD Mean SD

Module 1, no words 2 203 16.38 3.85 7.36 2.11 3.75 2.01 7.49 2.21
3 141 16.88 2.88 7.45 1.75 4.76 2.00 7.77 1.75
4–14 216 16.76 2.59 7.75 1.46 5.36 2.03 7.82 1.65
Module 1, some words 2–3 214 12.10 4.73 6.81 2.33 3.66 1.94 7.44 2.10
4 82 13.01 4.75 7.16 2.39 4.12 2.34 7.30 2.25
5–14 108 14.85 3.65 7.57 1.78 4.67 2.01 7.56 2.05
Module 2 2–3 106 10.03 4.02 7.08 2.18 4.02 2.02 7.59 1.94
4 94 10.69 4.69 6.88 2.37 3.74 2.18 6.87 2.22
5–6 103 12.25 4.62 7.49 2.05 4.59 2.09 7.59 1.93
7–16 112 14.07 3.79 7.99 1.59 4.77 2.11 7.67 2.02
Module 3 3–5 71 9.52 4.06 6.68 2.48 2.65 1.83 6.94 2.43
6–16 357 9.21 4.43 6.77 2.52 2.73 1.88 6.86 2.68
All modules, all ages 1807 12.98 4.99 7.21 2.17 3.96 2.18 7.39 2.19
Page 21
Table 4
Forward stepwise linear regression models for domain raw totals and calibrated domain scores
SA-Raw SA-CSS
Hus et al.
R2 ΔF df B SE B β R2 ΔF df B SE B β
Step 1 .430 1079.07 1430 Step 1 .105 167.72 1430

Constant 18.75 .20 Constant 8.45 .11
Verbal IQ −.10 .00 −.66 Verbal IQ −.02 .00 −.32
Step 2 .433 7.41 1429 Step 2 .109 5.97 1429
Verbal IQ −.11 .00 −.72 Verbal IQ −.03 .00 −.40
Nonverbal IQ .02 .01 .08 Nonverbal IQ .01 .00 .09
Step 3 .435 5.51 1428 Step 3
Constant 18.15 .29 Constant
Verbal IQ −.11 .00 −.73 Verbal IQ
Nonverbal IQ .01 .01 .08 Nonverbal IQ
Mat Ed .56 .24 .05 Mat Educ
RRB-Raw RRB-CSS
R2 ΔF df B SE B β R2 ΔF df B SE B β
Step 1 .117 208.86 1573 Step 1 .035 56.62 1573

Verbal IQ −.02 .00 −.34 Nonverbal IQ −.02 .00 −.19
Step 2 .131 25.64 1572 Step 2 .041 10.12 1572
Nonverbal IQ −.01 .00 −.18 Race −.50 .16 −.08
Step 3 .143 20.40 1571 Step 3 .045 7.47 1571
Nonverbal IQ −.02 .00 −.19 Race −.56 .16 −.09
Race −.67 .15 −.11 Verbal IQ −.01 .00 −.11
Page 22
Table 5
Case summary characteristics
Demographics First assessment Last assessment

Hus et al.
Gender Race Diagnosis Age VIQ NVIQ ADOS module Age VIQ NVIQ ADOS module
Biancaa Female White Autism 4.0 108 80 2 11.0 126 107 3
Joey Male White PDD-NOS 2.8 69 74 2 5.1 105 119 3

Carolyn Female White PDD-NOS 2.3 33 72 1 10.2 42 51 2
Matthew Male Af. Amer. Autism 4.0 31 63 1 11.0 58 88 3
All ages in years; VIQ verbal IQ, NVIQ nonverbal IQ

a
Cognitive assessment was not completed at last assessment; IQs are from previous assessment at age 10
Page 23

NIH Public Access

Uploaded by

Copyright:

Available Formats

NIH Public Access

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

NIH Public Access

Uploaded by

Copyright:

Available Formats

NIH Public Access

J Autism Dev Disord. 2014 October ; 44(10): 2400–2412. doi:10.1007/s10803-012-1719-1.

Standardizing ADOS Domain Scores: Separating Severity of

Katherine Gotham, and

© Springer Science+Business Media New York 2012

2011). Diagnostic criteria focus on the presence or absence of specific behaviors or

assigned to a child with very significant social-communication impairments who exhibits

behaviors and more subtle social-communication difficulties. Social-communication

(e.g., Lord et al. 2006).

Standardization of Raw Totals

Associations Between Participant Characteristics, Raw Domain Totals and Calibrated

Correlations Between Domain Calibrations and Overall Calibrated Severity Score

Predictors of SA-Raw and SA-CSS

Predictors of RRB-Raw and RRB-CSS

mannerisms. At 11 years of age, Matthew showed an apparent decrease in severity of social-

SA score. Similarly, approximately 12 % of variance in RRB-Raw Total was explained by

DSM-5. As we move toward a single classification of “autism spectrum disorder” in

likelihood that associations with genetic or neurobiological abnormalities are specific to

Using these scores to separately examine distinct trajectories of social-communication and

sample size. Replication of the domain calibrations in independent samples is an important

ADOS domain calibrations provide separate estimates of severity of ASD-related social-

descriptions and diagnostic guidelines. World Health Organization; Geneva: 1992.

Module 1, no words Module 1, some words Module 2, younger than 5 Module 3

Mapping of ADOS raw domain totals onto calibrated severity scores

Domain Calibrated Raw domain totals

Module Age (years) N SA-Raw SA-CSS RRB-Raw RRB-CSS

Mean SD Mean SD Mean SD Mean SD

Step 1 .430 1079.07 1430 Step 1 .105 167.72 1430

Step 1 .117 208.86 1573 Step 1 .035 56.62 1573

Case summary characteristics

Demographics First assessment Last assessment

Biancaa Female White Autism 4.0 108 80 2 11.0 126 107 3

Joey Male White PDD-NOS 2.8 69 74 2 5.1 105 119 3

All ages in years; VIQ verbal IQ, NVIQ nonverbal IQ

You might also like