Journal of Autism and Developmental Disorders

https://doi.org/10.1007/s10803-023-06184-3

ORIGINAL PAPER

The Developmental Autism Early Screening (DAES): A Novel Test

for Screening Autism Spectrum Disorder
Lara Cirnigliaro1 · Maria Stella Valle2 · Antonino Casabona2 · Martina Randazzo1 · Francesca La Bruna1 ·
Fabio Pettinato1 · Antonio Narzisi3 · Renata Rizzo1 · Rita Barone1,4

Accepted: 5 November 2023

© The Author(s) 2024, corrected publication 2024

Abstract
This study was undertaken to set a novel developmental screening test for autism spectrum disorder (ASD) using the Griffiths
Scales of Child Development (Griffith III) (Green et al., 2016; Stroud et al., 2016), in order to intercept the early atypical
developmental patterns indicating ASD risk in the first 3 years of age. An observational and interactive ASD screener, the
Developmental Autism Early Screening (DAES), was developed by detecting Griffiths III items differentiating toddlers with
ASD risk from those with global developmental delay (DD) or neurotypical development. The DAES was validated with
ASD-specific diagnostic instruments (ADOS-2) and the cut-off score based on sensitivity, specificity and positive predictive
value that best differentiates between ASD and non-ASD children was identified. We enrolled a total sample of 297 subjects,
including children at risk for ASD or DD and neurotypical children. At a cut-off score of 12.5, the DAES had a sensitivity of
93%, specificity of 98.4%, positive predictive value of 96.3% and negative predictive value of 96.9% for identifying children
at risk for ASD from non-ASD participants (DD/neurotypical children). The DAES total score correlated significantly with
the ADOS-2 calibrated severity scores (CSS) (R = 0.53, p < 0.001). Three ASD risk ranges were identified according to
DAES total and ADOS-2 CSS: Little-to-no risk (CSS: 1–3, DAES: 1–7); Mild-to-moderate risk (CSS: 4–5, DAES: 8–14);
Moderate-to-severe risk (CSS: 6–10, DAES ≥ 15). The DAES provides a direct approach based on developmental profiles
to stratify risk for ASD in early childhood ensuring at risk children the most appropriate diagnostic procedures and targeted
intervention.

Keywords Autism spectrum disorder · Early diagnosis · Developmental profile · Griffiths III

Abbreviations CTRL Neurotypical control group

DAES Developmental Autism Early Screening DA Developmental age
ASD Autism spectrum disorder CA Chronological age
DD Global developmental delay SA Social affect
RRB Restricted and repetitive behaviour
CSS Calibrated severity score
Antonino Casabona—Deceased. PPV Positive predictive value
NPV Negative predictive value
* Rita Barone
rbarone@unict.it
1
Child Neurology and Psychiatry Unit, Department Introduction
of Clinical and Experimental Medicine, University
of Catania, Policlinico Via Santa Sofia, 78, 95123 Catania, Autism spectrum disorder (ASD) is a neurodevelopmental
Italy
disorder characterized by social communication impairment
2
Laboratory of Neuro‑Biomechanics, Department and restricted/repetitive behaviors and interests. The preva-
of Biomedical and Biotechnological Sciences, School
of Medicine, University of Catania, Catania, Italy lence rate of ASD is 27.6 per 1.000 (one in 36) 8-year-old
3 children in the United States (US). The average age of diag-
IRCCS Stella Maris Foundation, Pisa, Italy
nosis is about 49 months but current estimates suggest that
4
Reseach Unit of Rare Diseases and Neurodevelopmental the median age of diagnosis is between 42 and 59 months in
Disorders, Oasi Research Institute-IRCCS, Troina, Italy

Vol.:(0123456789)
 Journal of Autism and Developmental Disorders

the US (Maenner et al., 2023). This is much later than pos- developmental disorders present a limitation in the assess-
sible even with only clinical evaluation and not biomarkers. ment of negative predictive value (NPV), due to the financial
In fact, there is much evidence that a sound diagnosis is costs involved in testing large numbers of negative screening
possible and in fact practiced at 18 to 24 months (McCarty children for final diagnosis and the long time period over
& Frye, 2020). According to a recent survey on ASD con- which the disorder could potentially develop (Robins, 2020).
ducted in 14 European countries (ASDEU project network), Taking into account the above mentioned limitations, using a
the average age of access to diagnostic services in Europe is sequence of primary and secondary screening tools is impor-
approximately 36.4 months and diagnoses occur between 36 tant to maximize the predictive value of screening.
and 42 months. However, the average age at which concerns Level 2 (secondary) screeners aim to identify children
were first raised about the child subsequently being diag- at risk of ASD either because they are already under obser-
nosed with ASD was 18.3 months, which suggests a need for vation for developmental concerns, or because they failed
earlier diagnosis (Bejarano-Martìn et al., 2020). Level 1 screening, or because they are siblings of children
The development of autism-specific screening has facili- with ASD (Petrocchi et al., 2020). Level 2 interactive screen-
tated the early identification of children at risk (Robins, ers need to be confrontational instruments where a trained
2008). This is consistent with the fact that evidence-based professional interacts with the child providing a quantita-
early interventions designed to harness experience-dependent tive score and limiting the risk of subjective judgements.
neuroplasticity improve developmental trajectories in ASD According to important criteria such as replication in multi-
(Dawson, 2008; Lord et al., 2015). The American Academy ple health-care settings and accuracy of classification, some
of Pediatrics (AAP) recommends screening all children for level 2 screeners warrant consideration for clinical appli-
symptoms of ASD through a combination of developmen- cation such as the Screening Tool for Autism in Two-Year
tal surveillance and standardized autism-specific screen- Olds: STAT (Stone et al., 2004); the Baby and Infant Screen
ing tests at 18 and 24 months of age (Hyman et al., 2020; for Children with aUtIsm Traits–Part 1: BISCUIT-Part 1
Johnson et al., 2007). The large majority of ASD screeners (Matson et al., 2010); the Autism Detection in Early Child-
designed for caregivers are based on a conceptual analysis of hood: ADEC (Nah et al., 2014); the Systematic Observa-
early communication development and identification of “red tion of Red Flags: SORF (Dow et al., 2017) and the Rapid
flags” commonly described as early ASD indicators (Magàn- Interactive screening Test for Autism in Toddlers: RITA-T
Maganto et al., 2017). The Modified Checklist for Autism (Choueiri & Wagner, 2015). In particular, the STAT and the
in Toddlers, Revised with Follow-Up (M-CHAT-R/F), is RITA-T represent the two secondary screening instruments
one of the most studied and implemented ASD screening with the most evidence, valid for children aged 36 months
tool worldwide. It is a 2-stage parent-report screener devel- or less. Limitations of level 2 screeners are related to the
oped to identify children from 16 to 30 months with ASD performances in discriminating ASD from different neurode-
risk, in the general population (primary, level 1 screener) velopmental disorders at certain ages, inadequacy of stud-
(Campbell et al., 2017; Marlow et al., 2019). Recently, Aish- ied sample sizes, consistency issues, limited involvement of
woriya et al (2023) performed a meta-analysis of the specific independent researchers, requirements for training in test
performances of M-CHAT-R/F that showed a pooled sensi- administration and test administration time (Brewer et al.,
tivity of 82.5%, higher than expected for a good screening 2020; Norris & Lecavalier, 2010; Petrocchi et al., 2020).
tool (70–80%). The probability of ASD diagnosis following Screening for normal development may unravel chil-
M-CHAT-R/F positivity (pooled positive predictive values, dren with potential developmental delays (DD; APA, 2013)
PPV) was estimated at 51.2% [95% CI 43.0–59.5] in low- including delays in social-communication. In this respect,
risk samples. This implies that a positive screening with the some instruments such as the Ages and Stages Question-
M-CHAT-R/F is predictive of an ASD diagnosis in approxi- naire, have been integrated with the M-CHAT-R/F to com-
mately 50% of children. However, as previously recognized bine screening for DD and ASD (Hardy et al., 2015).
(Robins et al., 2014; Weitlauf et al., 2015), the pooled PPV Recently, the importance of studying developmental
of the M-CHAT-R/F for the presence of any developmental trajectories and, consequently, developing new tools to
disorder rises to 89% (Aishworiya et al., 2023). It has been probe the atypical developmental trajectories of ASD in
pointed out that several factors, such as low ASD prevalence young children has been emerging. Developmental assess-
rates and false responses due to lack of parental awareness ment of children is instrumental for understanding the
of expected socio-communicative milestones, are associated child’s developmental level at the time of testing, reveal-
with lower PPV regardless of screening sensitivity (McCarty ing strengths and weakness in the different domains of
& Frye, 2020). On the other hand, parents may over-report learning. This is relevant in order to plan further investi-
the presence of developmental abilities, leading to false nega- gations and/or referrals for appropriate therapeutic inter-
tive screenings. Most studies examining screening tools for ventions. Systematic evaluation of early developmental
Journal of Autism and Developmental Disorders

profiles illustrated some weakness in language, social and for widespread use due to its ease of administration and
communication skills in children with ASD when com- interpretation.
pared to peers with developmental and/or language delays
(Barbaro & Dissanayake, 2012; Delehanty et al., 2018;
Mitchell et al., 2011; Torrens & Ruiz, 2021). Methods
The Griffiths Scales of Child Development 3rd Edition
(Griffiths III) is the latest version of the Griffiths Develop- Study Design
mental Scales, a tool validated for developmental assess-
ment in children with ASD from birth to 6 years (Sandberg We conducted an instrumental, quantitative and descriptive
et al., 1993). Griffiths III was standardized in 2015 on a study divided into two phases: in phase I, the DAES was
representative sample from the UK and Ireland, thereafter developed by clinicians experienced in assessing children
it was published and adapted for other population sam- with neurodevelopmental concerns. In phase II, the reliabil-
ples, with a normative age range between 1 and 72 months ity and validity of the instrument were assessed.
(Green et al., 2016; Stroud et al., 2016). Previous research
embedded the use of Griffiths scales in describing dis-
tinct psychomotor profiles related to discrete diagnostic DAES Development
classes in pre-school aged children (Jansen et al., 2020; Li
et al., 2020). A recent study aimed to compare the Griffiths The DAES has been developed by comparing Griffiths III
III developmental profile of children with co-occurring scores obtained in children at risk of ASD, children at risk of
ASD + DD with that of a group of children with DD but DD and TD children. For this purpose, 78 subjects were con-
not ASD. The two diagnostic groups exhibited lower age secutively recruited between January 2019 and June 2021,
equivalent scores with respect to their chronological age including 'clinically referred' children with expressed con-
in all the considered developmental domains. However, cerns for either ASD or DD and neurotypical children. Par-
unlike the group with DD having an uniform decrease ticipants were assessed by clinical evaluation using DSM-5
of expected performances in all the domains, children criteria and the Griffiths III. They were matched for devel-
with ASD + DD showed an uneven profile with relative opmental age, as measured by the Griffith III A-subscale, in
failures in Language and Communication and Personal- order to compare children with the same non-verbal cogni-
Social–Emotional subscales (Taddei et al., 2023). tive level between groups (Table 1).
We aimed to timely understanding predictors of the The inclusion criteria were: A-subscale developmental age
atypical developmental trajectories associated with the (DA) of 12–36 months; clinical referral for ASD-risk or DD-
wide phenotypic variability in ASD-risk children. Identi- risk (first and second group, respectively). TD children were
fying specific risk signs in the period of maximum brain defined as having General Development Quotient ≥ 90 at Grif-
plasticity could facilitate an early therapeutic intervention fiths III (CTRL group). They were healthy children prospec-
and a more favourable outcome. The present study was tively controlled for transient neonatal jaundice, suspected
undertaken for developing and preliminarily validating a maternal infection, or late-preterm born infants (gestational
novel observational and interactive level 2 ASD screener, age > 34 weeks) with no signs of neonatal distress.
the Developmental Autism Early Screening (DAES). The Exclusion criteria included bilingualism; syndromes or
DAES was conceived from psychomotor developmental genetic abnormalities; diagnosis of other neurological dis-
figures stemmed from the Griffiths III. In particular, we orders (i.e. epilepsy, hearing and visual defects). A best-esti-
hypothesized that the Griffiths III might intercept the early mate clinical diagnosis of ASD or non-ASD was made by
recurrent atypical developmental patterns in children at experienced clinicians using all available information and all
ASD-risk in the first years of life. The DAES, based on testing measures, including the Autism Diagnostic Interview-
Griffiths III, was designed to detect significant differences Revised (ADI-R) and the Autism Diagnostic Observation
in the developmental patterns of children at risk of ASD, Schedule, 2nd edition (ADOS-2) when clinically indicated.
with developmental age 12 to 36 months, compared with
children with developmental delay (DD) and typically
developing (TD) children of the same developmental age. DAES Validation
We report on the development, validity and discriminative
properties of the DAES in differentiating young children at In phase two, we randomly recruited three experimental
true risk of ASD from those with DD/neurotypical develop- groups: two groups of 'clinically referred' children with
ment. Based on the present study, we foresee that the DAES parental or professional concerns about ASD and/or DD,
could complement current screeners and have the potential and one group without developmental concerns (CTRL).
 Journal of Autism and Developmental Disorders

Table 1  Demographic features Participants ASD DD CTRL F2 p value

and developmental profiles
(developmental ages) on the N. 78 (sex) 26 (M: 20) 26 (M: 22) 26 (M: 18)
Griffiths III of participants
CA (months) (mean ± SD) 39.46 ± 12.48 34.07 ± 7.88 26.38 ± 7.48 12.288 0.00002
recruited for test development
A-Scale
DA (months) (Mean ± SD) 24.38 ± 5.80 24.5 ± 6.28 25.23 ± 5.87 0.152 0.8588
B-Scale
DA (months) (mean ± SD) 14.61 ± 7.35 18.27 ± 7.47 25.46 ± 8.52 13.00 0.00001
C-Scale
DA (months) (mean ± SD) 22.92 ± 7.50 23.19 ± 6.89 25.23 ± 7.03 0.809 0.4488
D-Scale
DA (months) (Mean ± SD) 17.0 ± 6.21 22.15 ± 7.20 26.53 ± 6.64 12.345 0.00002
E-Scale
DA (months) (mean ± SD) 25.73 ± 6.60 26.07 ± 7.07 26 ± 7.08 0.017 0.9822

Bold indicates statistically significant values (p ≤ 0.05)

CA chronological age, DA developmental age, M males, ASD autism spectrum disorder, DD global devel-
opmental delay, CTRL neurotypical children

The CTRL group consisted of healthy toddlers prospectively (including attention, problem-solving abilities, sequential
screened for transient neonatal jaundice, suspected maternal reasoning, processing speed, visuospatial skills and mem-
infection or preterm birth without signs of neonatal distress ory); subscale B (Language and Communication) evaluates
(Apgar scores: 9/10). Participants were recruited at the study the development of both receptive and expressive language
site between July 2021 and June 2022 using the same inclu- and social communication abilities; subscale C (Eye and
sion and exclusion criteria as in the previous analysis. They Hand Coordination) assesses visual perception and fine
had a chronological age ranging from 18 to 48 months with motor skills; subscale D (Personal-Social-Emotional) evalu-
a Griffith III A-Subscale DA equal to 12–36 months. Partici- ates child’s ability to adapt, personal autonomy and early
pants were assessed by clinical evaluation and by the Grif- social and emotional development through items measuring
fiths III. The DAES was administered at the time of clinical imitation, joint attention, emotional recognition and empa-
evaluation by two research assistants at the referral univer- thy; subscale E (Gross Motor domains) refers to the child’s
sity hospital for Child Neuropsychiatry where this study was early development of postural control, gross body coordina-
based. The Cohen κ statistic was calculated for each rater tion, balance and visual-spatial coordination. Each item is
and varied between 0.7 and 1, indicating good to excellent scored as a pass or a fail, + 1 or 0 respectively. Subscale raw
agreement. Two senior board-certified researchers blind to scores and general development raw scores are calculated to
the DAES scores generated final diagnoses independently of determine the Developmental Age, Scaled Score and Devel-
the DAES based on a full assessment (history, observation, opment Quotient, according to the norm tables.
and all testing measures) (Fig. 1). Written informed consent DSM-5 criteria defined by the American Psychiatric
from both parents was acquired before the beginning of the Association were used for ASD and DD diagnosis (Ameri-
study. The current study was part of an overall larger study can Psychiatric Association, 2013). Symptoms of ASD
aimed at identifying markers, predictors and developmental were established using the gold-standard tools for ASD
trajectories of ASD. The larger overall study was approved diagnosis: Autism Diagnostic Interview-Revised (ADI-R)
by the local ethics committee at the University Hospital (Lord et al., 1994) and the Autism Diagnostic Observation
Referral Centre for ASD with number n° 759. All proce- Schedule, 2nd edition (ADOS-2) (Lord et al., 2012) The
dures performed in the present study were in accordance ADI-R is a structured interview to the parents of children
with the 1964 Declaration of Helsinki and its later amend- referred for evaluation of possible ASD. The ADOS-2 is
ments (2013). a semi-structured, standardized assessment of core deficits
in ASD. It contains five modules that are differentiated by
Measures children’s developmental and language levels. In the pre-
sent study participants with ASD risk completed the Toddler
Participants were assessed using the Griffiths III that pro- Module (designed specifically for children 12–30 months
vides an overall measure of child psychomotor develop- old with limited language), the Module 1 (used for children
ment across five subscales. (Green et al., 2016). Subscale A aged from 31 months who do not consistently use phrase
(Foundations of Learning) assesses the ability of learning speech) or the Module 2 in a minority of children using
Journal of Autism and Developmental Disorders

Fig. 1  Participant flowchart

for DAES validation. ASD
Autism Spectrum Disorder, DD
Global Developmetal Delay,
CTRL control, ADOS-2 Autism
Diagnostic Observation Sched-
ule, ADI-R Autism Diagnostic
Interview-Revised, M males, m
months

phrase speech, but who were not verbally fluent. To allow predictive of ASD risk, were used to develop the instrument
comparisons among different modules, ADOS-2 scores (DAES).
(total score, Social Affect, SA, and Restricted and Repeti- In the second phase (tool validation), the one-way
tive Behaviour, RRB, scores) were converted to respective ANOVA test with pairwise comparisons, based on Bon-
calibrated severity scores (CSS 1–10 indicating absence to ferroni correction, was applied to compare chronological
severe autism) (Esler et al., 2015; Gotham et al., 2009; Hus age (CA) and developmental age (DA) among participants
et al., 2014). Raw SA domain scores were standardized using recruited for the test validation and including three groups
the same 10-point severity rating scale as total-raw scores. (ASD, DD and CTRL).
Instead, raw RRB domain scores were standardized in CSS Receiver operating characteristic (ROC) curve analysis
values from 5 to 10, due to the limited range of the RRB raw was performed to validate the capability of the DAES total
total (Hus et al., 2014). Thus, a RRB CSS of 5 represents score in discriminating ASD children vs children with typi-
raw scores in the mild-to-moderate concern range. cal development (CTRL) and ASD children vs children with
DD. Sensitivity, specificity, positive predictive value (PPV)
Statistical Analyses and negative predictive value (NPV) of the DAES total score
were calculated and the cut-off values, with the optimal sen-
In the first phase of tool development, we initially consid- sitivity and specificity, were determined by using Youden’s
ered and compared equivalent mental/developmental ages J index and Euclidean distance (Akobeng, 2007; Kruizinga
(months) as means (M) and standard deviations (SD) on et al., 2014).
the Griffiths III in the three groups (ASD, DD and CTRL). Linear correlations between DAES total score and
The one-way ANOVA statistical test was applied for each ADOS-2 total CSS, SA CSS and RRB CSS, were calculated
Griffiths III subscale (A-E) in order to find out possible sig- using the Spearman’s rank correlation and the strength of
nificant differences among groups. correlation was assessed by Spearman’s Rho coefficient (R).
A multiple comparison test was then performed (Tukey The statistical significance level α was established at 0.05.
HSD test) to understand how the impact of each group could All statistical tests were performed by using SPSS version
determine the statistical differences found. 27 (SPSS, Inc., Chicago, IL, USA, IBM, Somers, NY, USA).
Based on previous analyses, we continued focusing on
the Griffiths III B and D-subscales, in the first 3 years of
age, as language and social and emotional skills are mostly Results
impaired in ASD. For tool development all items of B- and
D-subscales were considered and grouped by year of age and DAES Development
by constructs. The scores obtained for each single item in the
three different groups were analysed by the Pearson’s Chi- For test development, the three groups (ASD, DD, CTRL)
square independence test. The Chi-square test was applied were compared for significant differences on each Griffith
to assess which items showed significant differences over the III subscale. Participants were matched on their develop-
three groups and between each pair of groups: ASD/CTRL, mental age (DA) on subscale A, expression of the learning
ASD/DD and DD/CTRL group. Items significantly different base: mean participant DA was not significantly different
between ASD and DD or CTRL groups, and therefore most across the three study groups (F = 0.152; p = 0.858). It was
thought that this might give more weight to any differential
 Journal of Autism and Developmental Disorders

elements in the achievement of the specific items provided and interaction, play, social cognition, joint attention,
in the other scales. A significant difference among groups self-awareness, emotional understanding and expression
was found for the B (F = 13; p = 0.00001) and D subscale DA and empathy. The tool is administered in approximately
means (F = 12.3; p = 0.00002). No differences were found in 20 min, starting with the first items (starting points)
the remaining Griffith III subscales (Table 1). regardless of the child's age. According to Griffith III, the
A multiple comparison test (Tukey HSD test) was then discontinue administration rule is established after six
carried out to understand how the impact of each group consecutive items not passed within each subscale. No
could determine the statistical differences found. DA sig- specific training is required for users trained in the use of
nificant differences were found between the ASD group the Griffith III. The standardised administration of items
and the CTRL group on the B subscale (p = 0.00001) and may refer to the Griffith III manual (Table S1, Supplemen-
between the ASD group and both the DD (p = 0.029) and tary Material).
CTRL (p = 0.0001) groups on the D subscale.
From this initial survey, we deduced that the Griffiths
III B and D subscales were the most sensitive in capturing DAES Scoring
differences between groups. We therefore focused on each
individual item of the B and D subscale constructs that dif- Based on statistical analyses, we assigned a score of + 2
fered significantly in the comparisons between ASD and the to the items that were significantly different in the ASD
remaining two groups, and this analysis was carried out for children compared to both the DD and CTRL groups, and
each year in the first 3 years of life (Table 2a, b). The items a score of + 1 to the items that were significantly different
significantly different between groups representing the most in the ASD children compared to the CTRL group.
predictive ones for ASD risk were used to develop the DAES For each item, 0 represents the skill being expressed.
(Table S1, Supplementary material). Therefore, if the child fails the item (skill not expressed),
As an example, among the skills expected in the first year a score of + 1/ + 2 is assigned during the observation, so
of age in the area of listening and attention, a significant that a higher total score indicates a higher risk of ASD.
difference between the three groups was found in items B1 For example, if the child fails items B1.1 or B1.6 (signifi-
(makes eye contact with the speaker) and B6 (responds when cantly different in ASD children compared to both CTRL
called—gets the child's attention in some way). These skills and DD groups), a score of + 2 is assigned, whereas if the
basically reflect the child's ability to establish an interaction child fails items B1.10 and B1.11 (significantly different
with the interlocutor, an indispensable prerequisite for the in ASD group compared to CTRL group), a score of + 1
acquisition of further communicative skills. Group compari- is assigned. If the child passes the items, a score of 0 is
son analyses using Pearson’s chi-squared test for independ- assigned.
ence revealed that children with ASD significantly failed
both items B1 and B6 when compared with either the DD
or CTRL groups (Table 2a). DAES Validation Analyses
As aforementioned, the items for years one to three, rep-
resenting the most predictive for ASD risk, were added to For preliminary validation analyses, the DAES was admin-
figure the novel screening tool based on differences in early istered to 219 children recruited over a 12-month period
developmental profiles measured on the Griffith III: Devel- (ASD/DD risk and CTRL). Mean total DAES scores were
opmental Autism Early Screening (DAES). higher in the ASD risk group (19 ± 5.4) than in the DD risk
(6 ± 3.7) and CTRL (4 ± 3.5) groups.
DAES Description and Administration According to DSM-5 criteria, Griffiths III, ADI-R and
ADOS-2 scores, 57 children were diagnosed with ASD and
The DAES is a 36-items observational and interactive 61 children were diagnosed with DD. 101 children were in
measure, developed by selecting those we found to rep- the CTRL group (Fig. 1).
resent the most predictive Griffiths III items for ASD risk TD children (CTRL group) were significantly younger
in the first 3 years of age. Items are organized accord- in comparison with ASD (p < 0.001) and DD participants
ing to the Griffiths III specific constructs of B- (Language (p < 0.001). DA (months) on Griffith III A subscale, was
and Communication) and D- (Personal-Social-Emotional) computed as measure of non-verbal mental development.
subscales, that are expected particularly impaired in chil- DA was not significantly different when comparing ASD
dren with ASD. In the DAES, items probe the follow- (20 ± 5.8 months) with CTRL groups (20 ± 8.4 months)
ing areas: listening, attention and communicative intent, (p = 1), considering the younger chronological age of the
expressive communication, receptive language develop- CTRL. DA was significantly lower in children with ASD
ment, social-emotional reciprocity, social communication than in children with DD (24 ± 6.3 months) (p = 0.005),t
Table 2  Griffiths III items with significant differences among groups (ASD/DD/CTRL) in subscales B (a) and D (b) constructs for year (Pearson’s Chi-square independence test)
(a)

1st year Listening, attention Communicative intent Preverbal expressive Preverbal receptive language development
communication

B1 B6 B13 B14 B11 B10 B16

3 groups
χ2 8.432 14.319 9.043 7.8 8.177 7.8 8.177
p 0.015 < 0.001 0.011 0.02 0.017 0.02 0.017
ASD/DD
χ2 4.333 6.584 5.318 2.364 0.591 2.364 3.519
p 0.037 0.01 0.021 0.124 0.442 0.124 0.061
ASD/CTRL
Journal of Autism and Developmental Disorders

χ2 4.333 9.455 5.318 6.783 5.532 6.783 6.933

p 0.037 0.002 0.021 0.009 0.019 0.009 0.008
DD/CTRL
χ2 0 1.02 0 2.08 3.184 2.08 0.754
p 0 0.313 1 0.149 0.074 0.149 0.385
2nd year Listening, attention Expressive language development Receptive language development

B2 B9 B10 B13 B3 B14

3 groups
χ2 5.318 9.743 19.808 20.748 6.797 9.77
p 0.021 0.008 < 0.001 < 0.001 0.033 0.008
ASD/DD
χ2 0.843 2.342 4.282 4.328 1.949 1.444
p 0.358 0.126 0.039 0.04 0.163 0.229
ASD/CTRL
χ2 5.65 9.665 19.692 20.17 6.718 9.433
p 0.017 0.002 < 0.001 < 0.001 0.01 0.002
DD/CTRL
χ2 2.364 2.769 6.718 7.076 1.564 3.775
p 0.124 0.096 0.01 0.008 0.211 0.052
Table 2  (continued)
(b)*

1st year Personal Social Emotional

D13 D3 D4 D14 D15 D18 D16

3 groups
χ2 8.488 9.797 9.73 21.589 9.797 14.585 7.8
p 0.014 0.007 0.008 < 0.001 0.007 < 0.001 0.02
ASD/DD
χ2 1.981 3.359 4.127 0 3.359 5.026 2.364
p 0.159 0.067 0.042 0.002 0.067 0.025 0.124
ASD/CTRL
χ2 8.089 8.089 6.786 15.6 8.089 12.381 6.783
p 0.004 0.004 0.009 < 0.001 0.004 < 0.001 0.009
DD/CTRL
χ2 3.184 2.08 1.02 2.08 2.08 3.184 2.08
p 0.074 0.149 0.313 0.149 0.149 0.074 0.149
2nd year Personal Social Emotional
D4 D8 D10 D14 D15 D2 D3 D7 D9 D2 D13

3 groups
χ2 3.914 21.775 6.424 11.209 6.413 10.079 10.079 7.682 8.203 13.689 9.742
p 0.048 < 0.001 0.040 0.004 0.041 0.006 0.006 0.021 0.017 < 0.001 0.008
ASD/DD
χ2 0.843 9.321 0.361 1.997 3.059 4.591 4.591 2.882 1.359 7.589 0.719
p 0.358 0.002 0.548 0.158 0.08 0.032 0.032 0.09 0.244 0.006 0.397
ASD/CTRL
χ2 3.9 20.17 5.65 11.143 4.282 8.308 6.564 6.584 8.308 9.774 9.774
p 0.048 < 0.001 0.017 < 0.001 0.039 0.004 0.004 0.01 0.004 0.002 0.002
DD/CTRL
χ2 1.209 2.882 3.359 2.882 0.115 0.754 0.754 1.083 3.359 0.221 2.882
p 0.271 0.09 0.067 0.09 0.734 0.385 0.385 0.298 0.067 0.638 0.09
3rd year Personal Social Emotional
D1 D2 D4 D7 D8 D5
3 groups
χ2 15.023 19.343 11.361 15.631 12.057 6.413
p < 0.001 < 0.001 0.003 < 0.001 0.002 0.041
Journal of Autism and Developmental Disorders
 Journal of Autism and Developmental Disorders

indicating that, on average, participants with ASD were

developmentally delayed. In fact, a minority of the ASD
group in the study (20%) had no difference in DA/CA.

Assessment of the Diagnostic Accuracy of the DAES

Emotional

Considering the clinical diagnosis and the DAES total score,

0.297

0.219
1.089

4.713

1.513
0.03
D5

ROC curve analyses were performed to validate the predic-

tive performance of DAES scores by determining its opti-
mal cut-off score in discriminating ASD group from non-
ASD participants (DD/CTRL groups). For these analyses,
we included 65 TD children assessed by Griffiths III and
DAES from the CTRL group (n = 101), in order to make
0.099

< 0.001

0.237
2.722

11.879

1.396

the compared samples numerically more homogeneous. We

Social

found the area under the curve (AUC) of DAES Total Score
D8

was 0.994 (95% CI 0.98–1), thus supporting the DAES

capacity in classifying the ASD group from the non-ASD
group (Fig. 2a). In this application, the DAES Total Score
cut-off of 12.5 showed higher optimal sensitivity (93%) and
specificity (98.4%), with a PPV of 96.3% and NPV of 96.9%
(Table 3).
0.046

< 0.001

0.024
3.972

15.084

5.103

A scatter plot with the distribution of the DAES total

D7

scores for each participant in the two groups (ASD/non-

ASD) showed a total of 4 out of 57 false negative and a total
of 2 out of 126 false positive; thus, 53 were true positive and
124 were true negative (Fig. 2b).
The ROC curve calculated for classifying ASD group
ASD autism spectrum disorder, DD global developmental delay, CTRL neurotypical control group

with respect to CTRL group indicated a DAES cut-off score

< 0.001

0.058
2.877

10.909

3.586

of 11.5 showing higher optimal sensitivity (94.7%) and

0.09

*Significant differences were not evident in the third year of age in B-subscale constructs

specificity (100%), with a PPV of 100% and NPV of 95.6%

D4

(Fig. S1A–D). A scatter plot with the distribution of the

DAES total scores for each participant in the two groups
(ASD/CTRL) showed a total of 3 out of 57 false negative
and a total of 0 out of 65 false positive; thus, 54 were true
positive and 65 were true negative (Fig. S1A–D). An addi-
tional ROC curve was calculated for comparing the ASD
0.009

< 0.001

0.052
6.799

19.461

3.775

group with the whole group of children with typical develop-

D2

ment (n = 101). This sample included also 36 children not

Bold indicates statistically significant values (p ≤ 0.05)

assessed by Griffiths III. The DAES cut-off score of 11.5

showed higher optimal sensitivity (94.7%) and specificity
(100%), with a PPV of 100% and NPV of 97%. A scatter
plot with the distribution of the DAES total scores for each
participant in the two groups (ASD/CTRL) showed a total
< 0.001

0.128
1.879

4.105
Personal

of 3 out of 57 false negative and a total of 0 out of 101 false

0.17

12.92

positive. These results are comparable to the previous ones,

D1

in which 65 children in the CTRL group were considered.

Then we computed the DAES performance in classify-
ing correctly between ASD and DD. In this context, the
Table 2  (continued)

DAES cut-off score of 12.5 showed sensitivity (93%) and

specificity (96.7%), with a PPV of 96.3% and NPV of 93.7%
ASD/CTRL

DD/CTRL

(Fig. S1A–D). The scatter plot distribution of the DAES

ASD/DD
3rd year

total scores for each participant in the two groups (ASD/DD)

χ2

χ2

χ2

showed a total of 4 out of 57 false negative (53 were true

p

p
 Journal of Autism and Developmental Disorders

Fig. 2  a ROC curve of DAES

total score to evaluate sensitiv-
ity and specificity in separating
ASD (n = 57) from non-ASD
groups (DD + CTRL, n = 126).
DAES total score cut-off of 12.5
(red circle) shows higher opti-
mal sensitivity and specificity
based on Youden’s J index and
Euclidean distance. b Distri-
bution of DAES total score
for each participant of ASD
(n = 57) and non-ASD groups
(DD + CTRL, n = 126). With a
cut-off value of 12.5, a total of
4 out of 57 were False Negative
and a total of 2 out of 126 were
false positive. thus, 53 were
true positive and 124 were true
negative

positive) and a total of 2 out of 61 false positive (59 were Three risk ranges of DAES total score were identified in
true negative) (Fig. S1A–D). the ASD group according to the ADOS-2 total CSS: Little-
to-no risk (CSS: 1–3, DAES total score: 1–7); Mild-to-mod-
Relationships Among the Different Measures erate risk (CSS: 4–5, DAES total score: 8–14); Moderate-
to-severe risk (CSS: 6–10, DAES total score ≥ 15) (Fig. S2).
Linear regression was used to assess correlations between
DAES score and ADOS-2. To allow comparisons between
different modules, ADOS-2 CSS were used for correla- Discussion
tion analyses. The DAES total score was found to be sig-
nificantly correlated with the ADOS-2 total, SA and RRB We report a new level 2 ASD screening tool based on devel-
CSS (Fig. 3a–c), which provide a measure of ASD symptom opmental profiles obtained through a systematic analysis of
severity. all Griffiths III items in the first 3 years of age in young
Table 3  Sensitivity, specificity and other associated parameters of the ROC curve in separating ASD (n = 57) from non-ASD (DD + CTRL, n = 126) groups for different cut-off scores
Cut-off Sensitivity (%) Specificity (%) False negative False positive True positive True negative PPV (%) NPV (%) Youden index Euclidean
distance

0.5 100.0 4.0 0 121 57 5 32.0 100.0 0.040 0.960

1.5 100.0 21.4 0 99 57 27 36.5 100.0 0.214 0.786
2.5 100.0 27.8 0 91 57 35 38.5 100.0 0.278 0.722
3.5 100.0 35.7 0 81 57 45 41.3 100.0 0.357 0.643
4.5 100.0 44.4 0 70 57 56 44.9 100.0 0.444 0.556
5.5 100.0 54.0 0 58 57 68 49.6 100.0 0.540 0.460
6.5 100.0 61.1 0 49 57 77 53.8 100.0 0.611 0.389
7.5 98.2 65.1 1 44 56 82 56.0 98.8 0.633 0.349
Journal of Autism and Developmental Disorders

8.5 98.2 72.2 1 35 56 91 61.5 98.9 0.704 0.279

9.5 98.2 80.2 1 25 56 101 69.2 99.0 0.784 0.199
10.5 96.5 87.3 2 16 55 110 77.5 98.2 0.838 0.132
11.5 94.7 95.2 3 6 54 120 89.9 97.5 0.899 0.072
12.5 93.0 98.4 4 2 53 124 96.3 96.9 0.914 0.072
13.5 86.0 100.0 8 0 49 126 100.0 94.0 0.860 0.140
14.5 80.7 100.0 11 0 46 126 100.0 92.0 0.807 0.193
15.5 78.9 100.0 12 0 45 126 100.0 91.3 0.789 0.211
16.5 73.7 100.0 15 0 42 126 100.0 89.4 0.737 0.263
17.5 57.9 100.0 24 0 33 126 100.0 84.0 0.579 0.421
18.5 42.1 100.0 33 0 24 126 100.0 79.2 0.421 0.579
19.5 36.8 100.0 36 0 21 126 100.0 77.8 0.368 0.632
20.5 31.6 100.0 39 0 18 126 100.0 76.4 0.316 0.684
21.5 28.1 100.0 41 0 16 126 100.0 75.5 0.281 0.719
22.5 24.6 100.0 43 0 14 126 100.0 74.6 0.246 0.754
23.5 19.3 100.0 46 0 11 126 100.0 73.3 0.193 0.807
24.5 17.5 100.0 47 0 10 126 100.0 72.8 0.175 0.825
25.5 14.0 100.0 49 0 8 126 100.0 72.0 0.140 0.860
26.5 12.3 100.0 50 0 7 126 100.0 71.6 0.123 0.877
27.5 8.8 100.0 52 0 5 126 100.0 70.8 0.088 0.912
28.5 7.0 100.0 53 0 4 126 100.0 70.4 0.070 0.930
30 5.3 100.0 54 0 3 126 100.0 70.0 0.053 0.947
31.5 3.5 100.0 55 0 2 126 100.0 69.6 0.035 0.965
32.5 1.8 100.0 56 0 1 126 100.0 69.2 0.018 0.982

Bold indicates higher optimal sensitivity and specificity, with a PPV of 96.3% and NPV of 96.9% at a DAES total score cut-off of 12.5, determined by using Youden’s J index and Euclidean
distance
PPV positive predictive value; NPV negative predictive value
 Journal of Autism and Developmental Disorders

Fig. 3  Linear correlations

between DAES total score and
a ADOS-2 total CSS (R = 0.53,
p < 0.001), b SA CSS (R = 0.52,
p < 0.001), and c RRB CSS
(R = 0.35, p = 0.007), calcu-
lated using the Spearman’s
rank correlation. The strength
of correlation was assessed by
Spearman’s Rho coefficient
(R) and the level of statistical
significance was set at p < 0.05

children at risk of ASD, at risk of DD and in TD children. significantly low in a higher proportion of girls than boys
Previous studies have highlighted the importance of assess- while motor abilities measured on the locomotor subscale
ing neurodevelopment in children with ASD using the decreased with age at diagnosis.
Griffiths Scales in order to provide more targeted learn- Limitations of current level 2 screeners include difficul-
ing strategies. In fact, the Griffiths scales are increasingly ties in the exploration of potentially discriminating items
being used as part of test batteries to establish baselines, aid unravelling the developmental trajectories in ASD children.
diagnosis and monitor the development of children with a This, in turn, is functional for discriminating ASD from dif-
range of developmental disorders, including ASD (Jansen ferent neurodevelopmental disorders at certain ages. Thus,
et al., 2020; Li et al., 2020; Taddei et al., 2023). In this exploration of potentially discriminating items at the target
regard, peculiarities in longitudinal assessment of psycho- age is particularly envisaged. To date, little is known on the
motor development using the Griffiths scales were searched potential application of the Griffiths III for early identifica-
for to assess ASD risk and symptom severity at diagnosis. tion of children with ASD compared with both neurotypi-
Children with ASD showed lower scores of General Quo- cal children and children with developmental disorders dif-
tient and all sub-quotients measured on the Griffiths Mental ferent from ASD. Assuming that some peculiarities of the
Development Scale (GMDS) over time, with the exception development of children with ASD can orient towards an
of the performance sub-quotient. Interestingly, three sub- early diagnosis of this condition, we sought a new level 2
quotients (Personal-Social, Hearing and Language and Prac- screening tool through a systematic analysis of all Griffiths
tical Reasoning) were associated with the symptom severity III items in the first 3 years of age in order to differentiate
of ASD at the time of diagnosis (Pino et al., 2022). Moreo- children with ASD risk from children with DD and those
ver, Li et al. (2020) using the Chinese version of the GMDS with typical development.
explored the relationships among developmental levels and According with previous studies, we proved that B- and
ASD severity, sex and age of the child at ASD diagnosis. D-subscales are more sensitive in intercepting differences
Scores of sub-quotients were significantly lower in children between the groups (Jansen et al., 2020; Li et al., 2020; Pino
with more severe ASD. The performance sub-quotient was et al., 2022; Taddei et al., 2023). In particular, we pursued to
Journal of Autism and Developmental Disorders

identify the items of Griffiths III B- and D-subscales which cut-off of 12.5 enabling differentiation between ASD group
showed a significant difference among the three groups from a non-ASD group (DD/CTRL), in the first 3 years of
(ASD, DD and CTRL) in the first 3 years of life. life, with high accuracy, showing a sensitivity of 93% and a
In the first year ASD children showed more difficulties in specificity of 98.4% (PPV: 96.3%, NPV: 96.9%).
acquiring communication skills required for age, compared Notably, the DAES correlated positively with the
to DD children. Significant differences among groups were ADOS-2 CSS scores and with its diagnostic assignment
found in the B-subscale constructs “listening and attention” by clinicians who were blinded to the DAES test results.
and “intentional communication”, focused on the ability of Because of its significant correlation with ADOS-2 scores,
children to build dyadic interactions. Children with ASD and the DAES has its value in stratifying those children at risk
DD specifically lacked social bases of language (i.e. use of for ASD. In fact, following the overall ADOS-CSS risk strat-
gestures, facial expressions, joint attention), with respect to ification, we further identified an ASD risk score divided
children with DD in the items of the first year of life. This is into three risk bands, with a total DAES score ≥ 15 indicat-
consistent with consolidated knowledge showing that chil- ing moderate to high risk.
dren in the DD group may increasingly attend to the social There is an important demand for psychometrically valid,
world through more advanced joint attention skills, which, interactive level 2 ASD screening tests that clinicians can
in turn, leads to better responsiveness to language (Barbaro easily learn and administer (Brewer et al., 2020). Some cur-
& Dissanayake, 2012). rent Level 2 interactive screening tests require significant
In the second year of age, the children with ASD showed training and can take a long time to administer, making them
greater deficits in the subscale B construct 'expressive lan- difficult to integrate into clinical settings.
guage development' than the DD children, whereas these A strength of the DAES is that it can be promptly used in
differences were no longer evident in the third year of age. In the clinical practice with “clinically referred” children who
particular, in children with ASD, significant language devel- have already been assessed with the Griffiths III, to measure
opment can occur between 24 and 48 months of age. There- ASD risk at 12–36 months of age.
fore, the level of language development at 48 months may In addition, the DAES can be administered indepen-
predict language outcome in ASD (Brignell et al., 2018). dently of the Griffith III in approximately 20 min as a
With regard to D-subscale constructs, in the first year front-line screener for ASD in 'clinically referred' toddlers
of life, ASD children, compared to DD children, showed aged ≤ 36 months with expressed concerns for either ASD
more deficits in acquiring skills of “social” construct, or DD. In both instances, no additional training is required
such as early referential understanding and joint atten- for Griffith III trained users.
tion. The analysis of the data in the second and third years
highlighted a progressive divergence between the group Study Limitations
of children with autism and the other two groups in the
D-subscale constructs “personal” and “social-emotional”. The present study has certain limitations. We found a lower
In particular children with ASD failed in imitation, dyadic significant correlation between the total DAES score and
interactions and self-processing skills (self-concept and the RRB CSS. This is probably related to the lack of items
self-awareness) compared to children with DD. concerning restricted and repetitive behaviours in the DAES,
Altogether these results are consistent with recent stud- derived from Griffiths III, that in turn doesn’t explore the
ies focused on the assessment of ASD children compared RRB domain. The present finding is consistent with other
with DD children through Griffiths scales, that highlighted studies on screening tools in children at risk for ASD (Row-
a more significant impairment in the personal and social berry et al., 2015). While it is clear that RRBs are present
domains in ASD children (Taddei et al., 2023; Wang et al., in young children, these studies highlighted that the RRB
2021). The presence of communicative and social deficits domain does not discriminate children with ASD from
in children with ASD confirms that the core features of children with Development Delay/Typical Development as
ASD translate into a specific profile of early psychomotor effectively as Social Communication symptoms when used
functioning. The DAES, developed by analysing the most in screening measures (Dow et al., 2017).
predictive Griffiths III items for ASD risk, may assist in One more limitation is that our sample of ASD-risk chil-
the identification of young children at risk of ASD in rela- dren with a DA of 12–36 months, ranged in age from 18 to
tion to DD, in order to target intervention prior to formal 48 months. Further studies are needed to extend the assess-
diagnosis. ment to ASD-risk children younger than 18 months of age.
For this purpose, we assessed the DAES ability to differ- On the other hand, the variability and non-linearity of the
entiate children with ASD from those non-ASD including ASD phenotype in early development defines a diagnostic
children with DD or with neurotypical development in a val- instability over time (“lost or later diagnosis”), with some
idation sample. Overall, we found a unique DAES total score
 Journal of Autism and Developmental Disorders

children meeting diagnosis at follow-up, and other children Author Contributions Lara Cirnigliaro, Rita Barone: conceptualization,
no longer meeting diagnostic criteria (Landa et al., 2013, methodology and study design, writing-original draft preparation. Lara
Cirnigliaro, Martina Randazzo, Francesca La Bruna, Fabio Pettinato:
2022). Therefore, further studies may well consider expand- investigation and data curation. Maria Stella Valle, Antonino Casabona:
ing the age range of the DAES, by analysing additional Grif- data curation and formal analysis. Rita Barone, Lara Cirnigliaro, Anto-
fiths III items, that may be predictive of ASD risk at older nio Narzisi, Renata Rizzo: validation, writing-review. All authors criti-
ages. cally reviewed the manuscript, participated in its revision and approved
the final manuscript.
Another possible limitation of this research is that the
majority of the children with ASD in the study had comorbid Funding This research received no external funding.
developmental delays in two or more domains of the Grif-
fiths III. Nonetheless, it is essential to recognise that level 2 Declarations
screeners' limitations are linked to their ability to differenti-
Conflict of interest The authors state no conflict of interests. This re-
ate between ASD and diverse neurodevelopmental disorders, search is in no way an approved extension of the Griffith scales, nor it
such as DD. In this regard, studying a sample of children is intended to be, and is simply a product of the authors’ own independ-
with ASD and DD allowed us to better identify predictive ent research.
profiles that could differentiate the two conditions. This is
Open Access This article is licensed under a Creative Commons Attri-
important for aiding the process of differential diagnosis bution 4.0 International License, which permits use, sharing, adapta-
and informing individualised interventions. Further studies tion, distribution and reproduction in any medium or format, as long
with larger samples of children with ASD without DD may as you give appropriate credit to the original author(s) and the source,
confirm that the specific social-communicative difficulties provide a link to the Creative Commons licence, and indicate if changes
were made. The images or other third party material in this article are
are caused by the presence of ASD alone, rather than by the included in the article's Creative Commons licence, unless indicated
combined effects of ASD and DD. otherwise in a credit line to the material. If material is not included in
Overall, the study findings suggest potential for a level 2 the article's Creative Commons licence and your intended use is not
ASD screening test, but further replication in independent permitted by statutory regulation or exceeds the permitted use, you will
need to obtain permission directly from the copyright holder. To view a
referral samples is necessary. copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/.

References
Conclusions
Aishworiya, R., Ma, V. K., Stewart, S., Hagerman, R., & Feldman, H.
Future priorities for level 2 screeners include the exploration M. (2023). Meta-analysis of the Modified Checklist for Autism
of potentially discriminating items at the target age range in Toddlers, Revised/follow-up for Screening. Pediatrics, 151(6),
e2022059393. https://​doi.​org/​10.​1542/​peds.​2022-​059393
while attempting to unravel the complexity of developmen- Akobeng, A. K. (2007). Understanding diagnostic tests 3: Receiver
tal trajectories in children with ASD. In this scenario, the operating characteristic curves. Acta Paediatrica, 96(5), 644–647.
present study aimed to develop a novel level 2 screening https://​doi.​org/​10.​1111/j.​1651-​2227.​2006.​00178.x
test for ASD based on differences in early developmen- American Psychiatric Association. (2013). Diagnostic and Statistical
Manual of Mental Disorders (DSM-5) (5th ed.). American Psy-
tal profiles on the Griffith III (DAES) that may predict an chiatric Association.
ASD-risk. The DAES includes the most predictive items Barbaro, J., & Dissanayake, C. (2012). Developmental profiles of
for ASD-risk in children aged 18–48 months with a DA of infants and toddlers with autism spectrum disorders identified
12–36 months, differentiating ASD-risk from DD-risk chil- prospectively in a community-based setting. Journal of Autism
and Developmental Disorders, 42(9), 1939–1948. https://​doi.​org/​
dren and TD peers. The tool was developed and preliminar- 10.​1007/​s10803-​012-​1441-z
ily validated in two phases, including two different sets of Bejarano-Martín, Á., Canal-Bedia, R., Magán-Maganto, M., Fernán-
participants respectively, supporting the test effectiveness dez-Álvarez, C., Cilleros-Martín, M. V., Sánchez-Gómez, M. C.,
and demonstrating moderate to high correlation rates among García-Primo, P., Rose-Sweeney, M., Boilson, A., Linertová, R.,
Roeyers, H., Van der Paelt, S., Schendel, D., Warberg, C., Cramer,
DAES total score and ADOS-2 CSS. If the present results S., Narzisi, A., Muratori, F., Scattoni, M. L., Moilanen, I., …
are replicated, the DAES has strong potential for adding to Posada de la Paz, M. (2020). Early detection, diagnosis and inter-
current screener efforts for early identification of children vention services for young children with Autism Spectrum Dis-
with ASD-risk enabling more informed referrals to the most order in the European Union (ASDEU): Family and professional
perspectives. Journal of Autism and Developmental Disorders,
appropriate diagnostic procedures and facilitating access to 50(9), 3380–3394. https://​doi.​org/​10.​1007/​s10803-​019-​04253-0
targeted intervention. Brewer, N., Young, R. L., & Lucas, C. A. (2020). Autism screening in
early childhood: Discriminating autism from other developmental
Supplementary Information The online version contains supplemen- concerns. Frontiers in Neurology, 11, 594381. https://​doi.​org/​10.​
tary material available at https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 10803-0​ 23-0​ 6184-3. 3389/​fneur.​2020.​594381
Brignell, A., Williams, K., Jachno, K., Prior, M., Reilly, S., & Morgan,
Acknowledgments The authors are grateful to all participants and their A. T. (2018). Patterns and predictors of language development
families for the collaboration.
Journal of Autism and Developmental Disorders

from 4 to 7 years in verbal children with and without autism spec- assessment. PLoS ONE, 9(5), e97630. https://​doi.​org/​10.​1371/​
trum disorder. Journal of Autism and Developmental Disorders, journ​al.​pone.​00976​30
48(10), 3282–3295. https://​doi.​org/​10.​1007/​s10803-​018-​3565-2 Landa, R. J., Gross, A. L., Stuart, E. A., & Faherty, A. (2013). Develop-
Campbell, K., Carpenter, K. L. H., Espinosa, S., Hashemi, J., Qiu, Q., mental trajectories in children with and without autism spectrum
Tepper, M., Calderbank, R., Sapiro, G., Egger, H. L., Baker, J. disorders: The first 3 years. Child Development, 84(2), 429–442.
P., & Dawson, G. (2017). Use of a digital modified checklist for https://​doi.​org/​10.​1111/j.​1467-​8624.​2012.​01870.x
autism in toddlers—revised with follow-up to improve quality of Landa, R. J., Reetzke, R., Holingue, C. B., Herman, D., & Hess, C. R.
screening for autism. The Journal of Pediatrics, 183, 133-139.e1. (2022). Diagnostic stability and phenotypic differences among
https://​doi.​org/​10.​1016/j.​jpeds.​2017.​01.​021 school-age children diagnosed with ASD before age 2. Frontiers
Choueiri, R., & Wagner, S. (2015). A New Interactive screening test for in Psychiatry, 13, 805686. https://​doi.​org/​10.​3389/​fpsyt.​2022.​
autism spectrum disorders in toddlers. The Journal of Pediatrics, 805686
167(2), 460–466. https://​doi.​org/​10.​1016/j.​jpeds.​2015.​05.​029 Li, H. H., Wang, C. X., Feng, J. Y., Wang, B., Li, C. L., & Jia, F. Y.
Dawson, G. (2008). Early behavioral intervention, brain plasticity, (2020). A developmental profile of children with autism spec-
and the prevention of autism spectrum disorder. Development trum disorder in china using the Griffiths Mental Development
and Psychopathology, 20(3), 775–803. https://​doi.​org/​10.​1017/​ Scales. Frontiers in Psychology, 11, 570923. https://​doi.​org/​10.​
S0954​57940​80003​70 3389/​fpsyg.​2020.​570923
Delehanty, A. D., Stronach, S., Guthrie, W., Slate, E., & Wetherby, Lord, C., Bishop, S., & Anderson, D. (2015). Developmental trajecto-
A. M. (2018). Verbal and nonverbal outcomes of toddlers with ries as autism phenotypes. American Journal of Medical Genetics.
and without autism spectrum disorder, language delay, and global Part C, Seminars in Medical Genetics, 169(2), 198–208. https://​
developmental delay. Autism & Developmental Language Impair- doi.​org/​10.​1002/​ajmg.c.​31440
ments. https://​doi.​org/​10.​1177/​23969​41518​764764 Lord, C., Rutter, M., Di Lavore, P. C., Risi, S., Gotham, K., & Bishop,
Dow, D., Guthrie, W., Stronach, S. T., & Wetherby, A. M. (2017). Psy- S. (2012). Autism Diagnostic Observation Schedule (2nd ed.).
chometric analysis of the Systematic Observation of Red Flags for Western Psychological Services.
autism spectrum disorder in toddlers. Autism: The International Lord, C., Rutter, M., & Le Couteur, A. (1994). Autism Diagnostic
Journal of Research and Practice, 21(3), 301–309. https://d​ oi.o​ rg/​ Interview-Revised: a revised version of a diagnostic interview for
10.​1177/​13623​61316​636760 caregivers of individuals with possible pervasive developmental
Esler, A. N., Bal, V. H., Guthrie, W., Wetherby, A., Ellis Weismer, S., disorders. Journal of Autism and Developmental Disorders, 24(5),
& Lord, C. (2015). The Autism Diagnostic Observation Schedule, 659–685. https://​doi.​org/​10.​1007/​BF021​72145
Toddler Module: Standardized severity scores. Journal of Autism Maenner, M. J., Warren, Z., Williams, A. R., Amoakohene, E., Bakian,
and Developmental Disorders, 45(9), 2704–2720. https://​doi.​org/​ A. V., Bilder, D. A., Durkin, M. S., Fitzgerald, R. T., Furnier, S.
10.​1007/​s10803-​015-​2432-7 M., Hughes, M. M., Ladd-Acosta, C. M., McArthur, D., Pas, E.
Gotham, K., Pickles, A., & Lord, C. (2009). Standardizing ADOS T., Salinas, A., Vehorn, A., Williams, S., Esler, A., Grzybowski,
scores for a measure of severity in autism spectrum disorders. A., Hall-Lande, J., … Shaw, K. A. (2023). Prevalence and char-
Journal of Autism and Developmental Disorders, 39(5), 693–705. acteristics of autism spectrum disorder among children aged 8
https://​doi.​org/​10.​1007/​s10803-​008-​0674-3 years—autism and developmental disabilities monitoring network,
Green, E., Stroud, L., O'Connell, R., Bloomfield, S., Cronje, J., Fox- 11 Sites, United States, 2020. Morbidity and Mortality Weekly
croft, C., Hurter, K., Lane, H., Marais, R., Marx, C., McAlin- Report. Surveillance Summaries (Washington, D.C.: 2002), 72(2),
den, P., Paradice, R., & Venter, D. (2016). Griffiths Scales of 1–14. https://​doi.​org/​10.​15585/​mmwr.​ss720​2a1
Child Development, 3rd ed. Part II: Administration and scoring. Magán-Maganto, M., Bejarano-Martín, Á., Fernández-Alvarez, C.,
Hogrefe. Narzisi, A., García-Primo, P., Kawa, R., Posada, M., & Canal-
Hardy, S., Haisley, L., Manning, C., & Fein, D. (2015). Can Screening Bedia, R. (2017). Early detection and intervention of ASD: A
with the ages and stages questionnaire detect autism? Journal of European overview. Brain Sciences, 7(12), 159. https://​doi.​org/​
Developmental and Behavioral Pediatrics: JDBP, 36(7), 536–543. 10.​3390/​brain​sci71​20159
https://​doi.​org/​10.​1097/​DBP.​00000​00000​000201 Marlow, M., Servili, C., & Tomlinson, M. (2019). A review of screen-
Hus, V., Gotham, K., & Lord, C. (2014). Standardizing ADOS domain ing tools for the identification of autism spectrum disorders and
scores: Separating severity of social affect and restricted and repet- developmental delay in infants and young children: Recommen-
itive behaviors. Journal of Autism and Developmental Disorders, dations for use in low- and middle-income countries. Autism
44(10), 2400–2412. https://​doi.​org/​10.​1007/​s10803-​012-​1719-1 Research: Official Journal of the International Society for Autism
Hyman, S. L., Levy, S. E., Myers, S. M., Council on Children with Research, 12(2), 176–199. https://​doi.​org/​10.​1002/​aur.​2033
Disabilities, Section on Developmental and Behavioral Pediatrics. Matson, J. L., Boisjoli, J. A., Hess, J. A., & Wilkins, J. (2010). Factor
(2020). Identification, evaluation, and management of children structure and diagnostic fidelity of the Baby and Infant Screen for
with autism spectrum disorder. Pediatrics, 145(1), e20193447. Children with aUtIsm Traits-Part 1 (BISCUIT-part 1). Develop-
https://​doi.​org/​10.​1542/​peds.​2019-​3447 mental Neurorehabilitation, 13(2), 72–79. https://d​ oi.o​ rg/1​ 0.3​ 109/​
Jansen, J. M., Green, E., Stroud, L. A., & Watson, M. B. (2020). Using 17518​42090​32135​76
the Griffiths III and Quartile Charts in assessing autism spectrum McCarty, P., & Frye, R. E. (2020). Early detection and diagnosis of
disorder: A case study. Journal of Education and Learning, 9(1), autism spectrum disorder: Why is it so difficult? Seminars in
30–40. https://​doi.​org/​10.​5539/​jel.​v9n1p​30 Pediatric Neurology, 35, 100831. https://​doi.​org/​10.​1016/j.​spen.​
Johnson, C. P., Myers, S. M., American Academy of Pediatrics 2020.​100831
Council on Children With Disabilities. (2007). Identification Mitchell, S., Cardy, J. O., & Zwaigenbaum, L. (2011). Differentiat-
and evaluation of children with autism spectrum disorders. ing autism spectrum disorder from other developmental delays
Pediatrics, 120(5), 1183–1215. https://​doi.​org/​10.​1542/​peds.​ in the first two years of life. Developmental Disabilities Research
2007-​2361 Reviews, 17(2), 130–140. https://​doi.​org/​10.​1002/​ddrr.​1107
Kruizinga, I., Visser, J. C., van Batenburg-Eddes, T., Carter, A. S., Nah, Y. H., Young, R. L., Brewer, N., & Berlingeri, G. (2014). Autism
Jansen, W., & Raat, H. (2014). Screening for autism spectrum detection in early childhood (ADEC): Reliability and validity data
disorders with the brief infant-toddler social and emotional for a Level 2 screening tool for autistic disorder. Psychological
Assessment, 26(1), 215–226. https://​doi.​org/​10.​1037/​a0034​472
 Journal of Autism and Developmental Disorders

Norris, M., & Lecavalier, L. (2010). Screening accuracy of Level 2 Stone, W. L., Coonrod, E. E., Turner, L. M., & Pozdol, S. L. (2004).
Autism Spectrum Disorder Rating Scales: A review of selected Psychometric properties of the STAT for early autism screening.
instruments. Autism: The International Journal of Research and Journal of Autism and Developmental Disorders, 34(6), 691–701.
Practice, 14(4), 263–284. https://​doi.​org/​10.​1177/​13623​61309​ https://​doi.​org/​10.​1007/​s10803-​004-​5289-8
348071 Stroud, L., Foxcroft, C., Green, E., Bloomfield, S., Cronje, J., Hurter,
Petrocchi, S., Levante, A., & Lecciso, F. (2020). Systematic review of K., Lane, H., Marais, R., Marx, C., McAlinden, P., O’Connell, R.,
Level 1 and Level 2 screening tools for autism spectrum disorders Paradice, R., & Venter, D. (2016). Griffiths Scales of Child Devel-
in toddlers. Brain Sciences, 10(3), 180. https://​doi.​org/​10.​3390/​ opment 3rd Ed. Part I: Overview, development, and psychometric
brain​sci10​030180 properties. Hogrefe.
Pino, M. C., Donne, I. L., Vagnetti, R., Tiberti, S., Valenti, M., & Taddei, M., Bulgheroni, S., Toffalini, E., Pantaleoni, C., & Lanfranchi,
Mazza, M. (2022). Using the Griffiths Mental Development Scales S. (2023). Developmental profiles of young children with autism
to evaluate a developmental profile of children with autism spec- spectrum disorder and global developmental delay: A study with
trum disorder and their symptomatologic severity. Child Psychia- the Griffiths III scales. Autism Research. https://​doi.​org/​10.​1002/​
try and Human Development. https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 10578-0​ 22-​ aur.​2953.​Advan​ceonl​inepu​blica​tion.​10.​1002/​aur.​2953
01390-z.A ​ dvanc​ eonli​ nepub​ licat​ ion.1​ 0.1​ 007/s​ 10578-0​ 22-0​ 1390-z Torrens, V., & Ruiz, C. (2021). Language and communication in pre-
Robins, D. L. (2008). Screening for autism spectrum disorders in pri- school children with autism and other developmental disorders.
mary care settings. Autism: The International Journal of Research Children (Basel, Switzerland), 8(3), 192. https://​doi.​org/​10.​3390/​
and Practice, 12(5), 537–556. https://​doi.​org/​10.​1177/​13623​ child​ren80​30192
61308​094502 Wang, H., Du, Y., Mao, Z., Che, Y., Li, H., Ding, L., & Jin, H. (2021).
Robins, D. L. (2020). How do we determine the utility of screening Use of the Griffiths mental development scale-Chinese in the
tools? Autism: The International Journal of Research and Prac- assessment of children with autism spectrum disorder and global
tice, 24(2), 271–273. https://​doi.​org/​10.​1177/​13623​61319​894170 developmental delay/intellectual disability. Medicine, 100(13),
Robins, D. L., Casagrande, K., Barton, M., Chen, C. M., Dumont- e25407. https://​doi.​org/​10.​1097/​MD.​00000​00000​025407
Mathieu, T., & Fein, D. (2014). Validation of the modified check- Weitlauf, A. S., Vehorn, A. C., Stone, W. L., Fein, D., & Warren,
list for Autism in toddlers, revised with follow-up (M-CHAT-R/F). Z. E. (2015). Using the M-CHAT-R/F to identify developmental
Pediatrics, 133(1), 37–45. https://d​ oi.o​ rg/1​ 0.1​ 542/p​ eds.2​ 013-1​ 813 concerns in a high-risk 18-month-old sibling sample. Journal of
Rowberry, J., Macari, S., Chen, G., Campbell, D., Leventhal, J. M., Developmental and Behavioral Pediatrics: JDBP, 36(7), 497–502.
Weitzman, C., & Chawarska, K. (2015). Screening for autism https://​doi.​org/​10.​1097/​DBP.​00000​00000​000194
spectrum disorders in 12-month-old high-risk siblings by parental
report. Journal of Autism and Developmental Disorders, 45(1), Publisher's Note Springer Nature remains neutral with regard to
221–229. https://​doi.​org/​10.​1007/​s10803-​014-​2211-x jurisdictional claims in published maps and institutional affiliations.
Sandberg, A. D., Nydén, A., Gillberg, C., & Hjelmquist, E. (1993). The
cognitive profile in infantile autism—a study of 70 children and
adolescents using the Griffiths Mental Development Scale. British
Journal of Psychology (London, England: 1953), 84(3), 365–373.
https://​doi.​org/​10.​1111/j.​2044-​8295.​1993.​tb024​88.x