Evaluation of data imputation strategies in complex, deeply-phenotyped data sets: the case of the EU-AIMS Longitudinal European Autism Project.

An increasing number of large-scale multi-modal research initiatives has been conducted in the typically developing population, e.g. Dev. Cogn. Neur. 32:43-54, 2018; PLoS Med. 12(3):e1001779, 2015; Elam and Van Essen, Enc. Comp. Neur., 2013, as well as in psychiatric cohorts, e.g. Trans. Psych. 10(1):100, 2020; Mol. Psych. 19:659-667, 2014; Mol. Aut. 8:24, 2017; Eur. Child and Adol. Psych. 24(3):265-281, 2015. Missing data is a common problem in such datasets due to the difficulty of assessing multiple measures on a large number of participants. The consequences of missing data accumulate when researchers aim to integrate relationships across multiple measures. Here we aim to evaluate different imputation strategies to fill in missing values in clinical data from a large (total N = 764) and deeply phenotyped (i.e. range of clinical and cognitive instruments administered) sample of N = 453 autistic individuals and N = 311 control individuals recruited as part of the EU-AIMS Longitudinal European Autism Project (LEAP) consortium. In particular, we consider a total of 160 clinical measures divided in 15 overlapping subsets of participants. We use two simple but common univariate strategies-mean and median imputation-as well as a Round Robin regression approach involving four independent multivariate regression models including Bayesian Ridge regression, as well as several non-linear models: Decision Trees (Extra Trees., and Nearest Neighbours regression. We evaluate the models using the traditional mean square error towards removed available data, and also consider the Kullback-Leibler divergence between the observed and the imputed distributions. We show that all of the multivariate approaches tested provide a substantial improvement compared to typical univariate approaches. Further, our analyses reveal that across all 15 data-subsets tested, an Extra Trees regression approach provided the best global results. This not only allows the selection of a unique model to impute missing data for the LEAP project and delivers a fixed set of imputed clinical data to be used by researchers working with the LEAP dataset in the future, but provides more general guidelines for data imputation in large scale epidemiological studies.

Infant Effortful Control Mediates Relations Between Nondirective Parenting and Internalising-Related Child Behaviours in an Autism-Enriched Infant Cohort.

Internalising problems are common within Autism Spectrum Disorder (ASD); early intervention to support those with emerging signs may be warranted. One promising signal lies in how individual differences in temperament are shaped by parenting. Our longitudinal study of infants with and without an older sibling with ASD investigated how parenting associates with infant behavioural inhibition (8-14 months) and later effortful control (24 months) in relation to 3-year internalising symptoms. Mediation analyses suggest nondirective parenting (8 months) was related to fewer internalising problems through an increase in effortful control. Parenting did not moderate the stable predictive relation of behavioural inhibition on later internalising. We discuss the potential for parenting to strengthen protective factors against internalising in infants from an ASD-enriched cohort.

Anxious parents show higher physiological synchrony with their infants.

BACKGROUND: Interpersonal processes influence our physiological states and associated affect. Physiological arousal dysregulation, a core feature of anxiety disorders, has been identified in children of parents with elevated anxiety. However, little is understood about how parent-infant interpersonal regulatory processes differ when the dyad includes a more anxious parent. METHODS: We investigated moment-to-moment fluctuations in arousal within parent-infant dyads using miniaturised microphones and autonomic monitors. We continually recorded arousal and vocalisations in infants and parents in naturalistic home settings across day-long data segments. RESULTS: Our results indicated that physiological synchrony across the day was stronger in dyads including more rather than less anxious mothers. Across the whole recording epoch, less anxious mothers showed responsivity that was limited to 'peak' moments in their child's arousal. In contrast, more anxious mothers showed greater reactivity to small-scale fluctuations. Less anxious mothers also showed behaviours akin to 'stress buffering' - downregulating their arousal when the overall arousal level of the dyad was high. These behaviours were absent in more anxious mothers. CONCLUSION: Our findings have implications for understanding the differential processes of physiological co-regulation in partnerships where a partner is anxious, and for the use of this understanding in informing intervention strategies for dyads needing support for elevated levels of anxiety.

Preference for biological motion is reduced in ASD: implications for clinical trials and the search for biomarkers.

BACKGROUND: The neurocognitive mechanisms underlying autism spectrum disorder (ASD) remain unclear. Progress has been largely hampered by small sample sizes, variable age ranges and resulting inconsistent findings. There is a pressing need for large definitive studies to delineate the nature and extent of key case/control differences to direct research towards fruitful areas for future investigation. Here we focus on perception of biological motion, a promising index of social brain function which may be altered in ASD. In a large sample ranging from childhood to adulthood, we assess whether biological motion preference differs in ASD compared to neurotypical participants (NT), how differences are modulated by age and sex and whether they are associated with dimensional variation in concurrent or later symptomatology. METHODS: Eye-tracking data were collected from 486 6-to-30-year-old autistic (N = 282) and non-autistic control (N = 204) participants whilst they viewed 28 trials pairing biological (BM) and control (non-biological, CTRL) motion. Preference for the biological motion stimulus was calculated as (1) proportion looking time difference (BM-CTRL) and (2) peak look duration difference (BM-CTRL). RESULTS: The ASD group showed a present but weaker preference for biological motion than the NT group. The nature of the control stimulus modulated preference for biological motion in both groups. Biological motion preference did not vary with age, gender, or concurrent or prospective social communicative skill within the ASD group, although a lack of clear preference for either stimulus was associated with higher social-communicative symptoms at baseline. LIMITATIONS: The paired visual preference we used may underestimate preference for a stimulus in younger and lower IQ individuals. Our ASD group had a lower average IQ by approximately seven points. 18% of our sample was not analysed for various technical and behavioural reasons. CONCLUSIONS: Biological motion preference elicits small-to-medium-sized case-control effects, but individual differences do not strongly relate to core social autism associated symptomatology. We interpret this as an autistic difference (as opposed to a deficit) likely manifest in social brain regions. The extent to which this is an innate difference present from birth and central to the autistic phenotype, or the consequence of a life lived with ASD, is unclear.

A white paper on a neurodevelopmental framework for drug discovery in autism and other neurodevelopmental disorders.

In the last decade there has been a revolution in terms of genetic findings in neurodevelopmental disorders (NDDs), with many discoveries critical for understanding their aetiology and pathophysiology. Clinical trials in single-gene disorders such as fragile X syndrome highlight the challenges of investigating new drug targets in NDDs. Incorporating a developmental perspective into the process of drug development for NDDs could help to overcome some of the current difficulties in identifying and testing new treatments. This paper provides a summary of the proceedings of the 'New Frontiers Meeting' on neurodevelopmental disorders organised by the European College of Neuropsychopharmacology in conjunction with the Innovative Medicines Initiative-sponsored AIMS-2-TRIALS consortium. It brought together experts in developmental genetics, autism, NDDs, and clinical trials from academia and industry, regulators, patient and family associations, and other stakeholders. The meeting sought to provide a platform for focused communication on scientific insights, challenges, and methodologies that might be applicable to the development of CNS treatments from a neurodevelopmental perspective. Multidisciplinary translational consortia to develop basic and clinical research in parallel could be pivotal to advance knowledge in the field. Although implementation of clinical trials for NDDs in paediatric populations is widely acknowledged as essential, safety concerns should guide each aspect of their design. Industry and academia should join forces to improve knowledge of the biology of brain development, identify the optimal timing of interventions, and translate these findings into new drugs, allowing for the needs of users and families, with support from regulatory agencies.

Parent-child interaction during the first year of life in infants at elevated likelihood of autism spectrum disorder.

Autism spectrum disorder (ASD) likely emerges from a complex interaction between pre-existing neurodevelopmental vulnerabilities and the environment. The interaction with parents forms a key aspect of an infant's social environment, but few prospective studies of infants at elevated likelihood (EL) for ASD (who have an older sibling with ASD) have examined parent-child interactions in the first year of life. As part of a European multisite network, parent-child dyads of free play were observed at 5 months (62 EL infants, 47 infants at typical likelihood (TL)) and 10 months (101 EL siblings, 77 TL siblings). The newly-developed Parent-Infant/Toddler Coding of Interaction (PInTCI) scheme was used, focusing on global characteristics of infant and parent behaviors. Coders were blind to participant information. Linear mixed model analyses showed no significant group differences in infant or parent behaviors at 5 or 10 months of age (all ps≥0.09, d≤0.36), controlling for infant's sex and age, and parental educational level. However, without adjustments, EL infants showed fewer and less clear initiations at 10 months than TL infants (p = 0.02, d = 0.44), but statistical significance was lost after controlling for parental education (p = 0.09, d = 0.36), which tended to be lower in the EL group. Consistent with previous literature focusing on parent-infant dyads, our findings suggest that differences between EL and TL dyads may only be subtle during the first year of life. We discuss possible explanations and implications for future developmental studies.

Dissecting the phenotypic heterogeneity in sensory features in autism spectrum disorder: a factor mixture modelling approach.

BACKGROUND: Heterogeneity in the phenotypic presentation of autism spectrum disorder (ASD) is apparent in the profile and the severity of sensory features. Here, we applied factor mixture modelling (FMM) to test a multidimensional factor model of sensory processing in ASD. We aimed to identify homogeneous sensory subgroups in ASD that differ intrinsically in their severity along continuous factor scores. We also investigated sensory subgroups in relation to clinical variables: sex, age, IQ, social-communication symptoms, restricted and repetitive behaviours, adaptive functioning and symptoms of anxiety and attention-deficit/hyperactivity disorder. METHODS: Three hundred thirty-two children and adults with ASD between the ages of 6 and 30 years with IQs varying between 40 and 148 were included. First, three different confirmatory factor models were fit to the 38 items of the Short Sensory Profile (SSP). Then, latent class models (with two-to-six subgroups) were evaluated. The best performing factor model, the 7-factor structure, was subsequently used in two FMMs that varied in the number of subgroups: a two-subgroup, seven-factor model and a three-subgroup and seven-factor model. RESULTS: The 'three-subgroup/seven-factor' FMM was superior to all other models based on different fit criteria. Identified subgroups differed in sensory severity from severe, moderate to low. Accounting for the potential confounding effects of age and IQ, participants in these sensory subgroups had different levels of social-communicative symptoms, restricted and repetitive behaviours, adaptive functioning skills and symptoms of inattention and anxiety. LIMITATIONS: Results were derived using a single parent-report measure of sensory features, the SSP, which limits the generalisability of findings. CONCLUSION: Sensory features can be best described by three homogeneous sensory subgroups that differ in sensory severity gradients along seven continuous factor scores. Identified sensory subgroups were further differentiated by the severity of core and co-occurring symptoms, and level of adaptive functioning, providing novel evidence on the associated clinical correlates of sensory subgroups. These sensory subgroups provide a platform to further interrogate the neurobiological and genetic correlates of altered sensory processing in ASD.

Infant EEG theta modulation predicts childhood intelligence.

Intellectual functioning is a critical determinant of economic and personal productivity. Identifying early neural predictors of cognitive function in infancy will allow us to map the neurodevelopmental pathways that underpin individual differences in intellect. Here, in three different cohorts we investigate the association between a putative neurophysiological indicator of information encoding (change in frontal theta during a novel video) in infancy and later general cognitive outcome. In a discovery cohort of 12-month-old typically developing infants, we recorded EEG during presentation of dynamic movies of people and objects. Frontal theta power (3-6 Hz) significantly increased during the course of viewing each video. Critically, increase in frontal theta during viewing of a video was associated with a differential response to repetition of that specific video, confirming relation to learning. Further, individual differences in the magnitude of change in frontal theta power were related to concurrent nonverbal cognitive level. We then sought to extend this association in two independent samples enriched for variation in cognitive outcome due to the inclusion of infants at familial risk for autism. We observed similar patterns of theta EEG change at 12 months, and found a predictive relation to verbal and nonverbal cognitive skills measured at 2, 3 and 7 years of age. For the subset of high-risk infants later diagnosed with autism, infant theta EEG explained over 80% of the variance in nonverbal skills at age 3 years. We suggest that EEG theta change in infancy is an excellent candidate predictive biomarker that could yield substantial insight into the mechanisms that underlie individual differences in childhood intelligence, particularly in high risk populations.

Eurosibs: Towards robust measurement of infant neurocognitive predictors of autism across Europe.

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder that affects social communication skills and flexible behaviour. Developing new treatment approaches for ASD requires early identification of the factors that influence later behavioural outcomes. One fruitful research paradigm has been the prospective study of infants with a first degree relative with ASD, who have around a 20% likelihood of developing ASD themselves. Early findings have identified a range of candidate neurocognitive markers for later ASD such as delayed attention shifting or neural responses to faces, but given the early stage of the field most sample sizes are small and replication attempts remain rare. The Eurosibs consortium is a European multisite neurocognitive study of infants with an older sibling with ASD conducted across nine sites in five European countries. In this manuscript, we describe the selection and standardization of our common neurocognitive testing protocol. We report data quality assessments across sites, showing that neurocognitive measures hold great promise for cross-site consistency in diverse populations. We discuss our approach to ensuring robust data analysis pipelines and boosting future reproducibility. Finally, we summarise challenges and opportunities for future multi-site research efforts.

Temperament as an Early Risk Marker for Autism Spectrum Disorders? A Longitudinal Study of High-Risk and Low-Risk Infants.

To investigate temperament as an early risk marker for autism spectrum disorder (ASD), we examined parent-reported temperament for high-risk (HR, n = 170) and low-risk (LR, n = 77) siblings at 8, 14, and 24 months. Diagnostic assessment was performed at 36 months. Group-based analyses showed linear risk gradients, with more atypical temperament for HR-ASD, followed by HR-Atypical, HR-Typical, and LR siblings. Temperament differed significantly between outcome groups (0.03 ≤ ηp2 ≤ 0.34). Machine learning analyses showed that, at an individual level, HR-ASD siblings could not be identified accurately, whereas HR infants without ASD could. Our results emphasize the discrepancy between group-based and individual-based predictions and suggest that while temperament does not facilitate early identification of ASD individually, it may help identify HR infants who do not develop ASD.

Evaluating Sex and Age Differences in ADI-R and ADOS Scores in a Large European Multi-site Sample of Individuals with Autism Spectrum Disorder.

Research on sex-related differences in Autism Spectrum Disorder (ASD) has been impeded by small samples. We pooled 28 datasets from 18 sites across nine European countries to examine sex differences in the ASD phenotype on the ADI-R (376 females, 1763 males) and ADOS (233 females, 1187 males). On the ADI-R, early childhood restricted and repetitive behaviours were lower in females than males, alongside comparable levels of social interaction and communication difficulties in females and males. Current ADI-R and ADOS scores showed no sex differences for ASD severity. There were lower socio-communicative symptoms in older compared to younger individuals. This large European ASD sample adds to the literature on sex and age variations of ASD symptomatology.

Prediction of Autism at 3 Years from Behavioural and Developmental Measures in High-Risk Infants: A Longitudinal Cross-Domain Classifier Analysis.

We integrated multiple behavioural and developmental measures from multiple time-points using machine learning to improve early prediction of individual Autism Spectrum Disorder (ASD) outcome. We examined Mullen Scales of Early Learning, Vineland Adaptive Behavior Scales, and early ASD symptoms between 8 and 36 months in high-risk siblings (HR; n = 161) and low-risk controls (LR; n = 71). Longitudinally, LR and HR-Typical showed higher developmental level and functioning, and fewer ASD symptoms than HR-Atypical and HR-ASD. At 8 months, machine learning classified HR-ASD at chance level, and broader atypical development with 69.2% Area Under the Curve (AUC). At 14 months, ASD and broader atypical development were classified with approximately 71% AUC. Thus, prediction of ASD was only possible with moderate accuracy at 14 months.

Visual search performance in infants associates with later ASD diagnosis.

An enhanced ability to detect visual targets amongst distractors, known as visual search (VS), has often been documented in Autism Spectrum Disorders (ASD). Yet, it is unclear when this behaviour emerges in development and if it is specific to ASD. We followed up infants at high and low familial risk for ASD to investigate how early VS abilities links to later ASD diagnosis, the potential underlying mechanisms of this association and the specificity of superior VS to ASD. Clinical diagnosis of ASD as well as dimensional measures of ASD, attention-deficit/hyperactivity disorder (ADHD) and anxiety symptoms were ascertained at 3 years. At 9 and 15 months, but not at age 2 years, high-risk children who later met clinical criteria for ASD (HR-ASD) had better VS performance than those without later diagnosis and low-risk controls. Although HR-ASD children were also more attentive to the task at 9 months, this did not explain search performance. Superior VS specifically predicted 3 year-old ASD but not ADHD or anxiety symptoms. Our results demonstrate that atypical perception and core ASD symptoms of social interaction and communication are closely and selectively associated during early development, and suggest causal links between perceptual and social features of ASD.

Adaptive behavior in autism: Minimal clinically important differences on the Vineland-II.

Autism Spectrum Disorder (ASD) is associated with persistent impairments in adaptive abilities across multiple domains. These social, personal, and communicative impairments become increasingly pronounced with development, and are present regardless of IQ. The Vineland Adaptive Behavior Scales, Second Edition (Vineland-II) is the most commonly used instrument for quantifying these impairments, but minimal clinically important differences (MCIDs) on Vineland-II scores have not been rigorously established in ASD. We pooled data from several consortia/registries (EU-AIMS LEAP study, ABIDE-I, ABIDE-II, INFOR, Simons Simplex Collection and Autism Treatment Network [ATN]) and clinical investigations and trials (Stanford, Yale, Roche) resulting in a data set of over 9,000 individuals with ASD. Two approaches were used to estimate MCIDs: distribution-based methods and anchor-based methods. Distribution-based MCID [d-MCID] estimates included the standard error of the measurement, as well as one-fifth and one-half of the covariate-adjusted standard deviation (both cross-sectionally and longitudinally). Anchor-based MCID [a-MCID] estimates include the slope of linear regression of clinician ratings of severity on the Vineland-II score, the slope of linear regression of clinician ratings of longitudinal improvement category on Vineland-II change, the Vineland-II change score maximally differentiating clinical impressions of minimal versus no improvement, and equipercentile equating. Across strata, the Vineland-II Adaptive Behavior Composite standardized score MCID estimates range from 2.01 to 3.2 for distribution-based methods, and from 2.42 to 3.75 for sample-size-weighted anchor-based methods. Lower Vineland-II standardized score MCID estimates were observed for younger and more cognitively impaired populations. These MCID estimates enable users of Vineland-II to assess both the statistical and clinical significance of any observed change. Autism Res 2018, 11: 270-283. © 2017 International Society for Autism Research, Wiley Periodicals, Inc. LAY SUMMARY: The Vineland Adaptive Behavior Scales (2nd edition; Vineland-II) is the most widely used scale for assessing day-to-day "adaptive" skills. Yet, it is unknown how much Vineland-II scores must change for those changes to be regarded as clinically significant. We pooled data from over 9,000 individuals with ASD to show that changes of 2-3.75 points on the Vineland-II Composite score represent the "minimal clinically-important difference." These estimates will help evaluate the benefits of potential new treatments for ASD.

Cortical responses before 6 months of life associate with later autism.

Autism spectrum disorder (ASD) is a common, highly heritable, developmental disorder and later-born siblings of diagnosed children are at higher risk of developing ASD than the general population. Although the emergence of behavioural symptoms of ASD in toddlerhood is well characterized, far less is known about development during the first months of life of infants at familial risk. In a prospective longitudinal study of infants at familial risk followed to 36 months, we measured functional near-infrared spectroscopy (fNIRS) brain responses to social videos of people (i.e. peek-a-boo) compared to non-social images (vehicles) and human vocalizations compared to non-vocal sounds. At 4-6 months, infants who went on to develop ASD at 3 years (N = 5) evidenced-reduced activation to visual social stimuli relative to low-risk infants (N = 16) across inferior frontal (IFG) and posterior temporal (pSTS-TPJ) regions of the cortex. Furthermore, these infants also showed reduced activation to vocal sounds and enhanced activation to non-vocal sounds within left lateralized temporal (aMTG-STG/pSTS-TPJ) regions compared with low-risk infants and high-risk infants who did not develop ASD (N = 15). The degree of activation to both the visual and auditory stimuli correlated with parent-reported ASD symptomology in toddlerhood. These preliminary findings are consistent with later atypical social brain responses seen in children and adults with ASD, and highlight the need for further work interrogating atypical processing in early infancy and how it may relate to later social interaction and communication difficulties characteristic of ASD.

Profiles of children with Down syndrome who meet screening criteria for autism spectrum disorder (ASD): a comparison with children diagnosed with ASD attending specialist schools.

BACKGROUND: Recent research suggests that around 16% to 18% of children with Down syndrome (DS) also meet diagnostic criteria for autism spectrum disorder (ASD). However, there are indications that profiles of autism symptoms in this group may vary from those typically described in children with ASD. METHOD: Rates of autism symptoms and emotional and behavioural problems among children with DS who screened positive for ASD on the Social Communication Questionnaire (SCQ) (n = 183) were compared with a group of children with clinical diagnoses of ASD (n = 189) attending specialist schools in the UK. Groups were matched for age and approximate language level (use of phrase speech). RESULTS: Profiles of autistic symptoms in the two groups were generally similar, but children with DS meeting ASD cut-off on the SCQ tended to show fewer problems in reciprocal social interaction than those in the ASD group. They also showed slightly lower rates of emotional and peer-related problems. The results mostly confirm findings from a previous study in which the original validation sample for the SCQ was used as a comparison group. CONCLUSION: Findings suggest that children with DS who meet screening criteria for ASD show similar profiles of communication and repetitive behaviours to those typically described in autism. However, they tend to have relatively milder social difficulties. It is important that clinicians are aware of this difference if children with DS and ASD are to be correctly diagnosed and eligible for specialist intervention and education services.

Predicting the diagnosis of autism in adults using the Autism-Spectrum Quotient (AQ) questionnaire.

BACKGROUND: Many adults with autism spectrum disorder (ASD) remain undiagnosed. Specialist assessment clinics enable the detection of these cases, but such services are often overstretched. It has been proposed that unnecessary referrals to these services could be reduced by prioritizing individuals who score highly on the Autism-Spectrum Quotient (AQ), a self-report questionnaire measure of autistic traits. However, the ability of the AQ to predict who will go on to receive a diagnosis of ASD in adults is unclear. METHOD: We studied 476 adults, seen consecutively at a national ASD diagnostic referral service for suspected ASD. We tested AQ scores as predictors of ASD diagnosis made by expert clinicians according to International Classification of Diseases (ICD)-10 criteria, informed by the Autism Diagnostic Observation Schedule-Generic (ADOS-G) and Autism Diagnostic Interview-Revised (ADI-R) assessments. RESULTS: Of the participants, 73% received a clinical diagnosis of ASD. Self-report AQ scores did not significantly predict receipt of a diagnosis. While AQ scores provided high sensitivity of 0.77 [95% confidence interval (CI) 0.72-0.82] and positive predictive value of 0.76 (95% CI 0.70-0.80), the specificity of 0.29 (95% CI 0.20-0.38) and negative predictive value of 0.36 (95% CI 0.22-0.40) were low. Thus, 64% of those who scored below the AQ cut-off were 'false negatives' who did in fact have ASD. Co-morbidity data revealed that generalized anxiety disorder may 'mimic' ASD and inflate AQ scores, leading to false positives. CONCLUSIONS: The AQ's utility for screening referrals was limited in this sample. Recommendations supporting the AQ's role in the assessment of adult ASD, e.g. UK NICE guidelines, may need to be reconsidered.

Elephants in Pyjamas: Testing the Weak Central Coherence Account of Autism Spectrum Disorders Using a Syntactic Disambiguation Task.

According to the weak central coherence (CC) account individuals with autism spectrum disorders (ASD) exhibit enhanced local processing and weak part-whole integration. CC was investigated in the verbal domain. Adolescents, recruited using a 2 (ASD status) by 2 (language impairment status) design, completed an aural forced choice comprehension task involving syntactically ambiguous sentences. Half the picture targets depicted the least plausible interpretation, resulting in longer RTs across groups. These were assumed to reflect local processing. There was no ASD by plausibility interaction and consequently little evidence for weak CC in the verbal domain when conceptualised as enhanced local processing. Furthermore, there was little evidence that the processing of syntactically ambiguous sentences differed as a function of ASD or language-impairment status.

Child's verbal ability and gender are associated with age at diagnosis in a sample of young children with ASD in Europe.

BACKGROUND: Autism spectrum disorder can in some cases be reliably diagnosed by age 2 years, but in community settings, the mean age at diagnosis is often considerably higher. Later diagnosis has been found to be associated with lower symptom severity, lower parental socioeconomic status and fewer parental concerns. Gender differences in age at diagnosis have been examined, with mixed evidence. METHODS: We examined the association of child's verbal ability and gender, and parental education, with age at diagnosis in a large sample of young children with autism spectrum disorder in 18 European countries (n = 1410). RESULTS: There was considerable variation in age at diagnosis across countries. Children with better communication skills were diagnosed significantly later than non-verbal and minimally verbal children. There was also a significant interaction of gender with verbal ability on age at diagnosis, in that female children with complex phrase speech were diagnosed later than male children with the same level of verbal ability. CONCLUSIONS: Our findings highlight the need to implement public awareness initiatives and training for professionals to promote early detection and, consequently, early intervention for autism spectrum disorder in Europe.