Objectives: Insight into the underlying structure of ASD-symptoms in the general population may facilitate the development of screening-instruments and screening-procedures.

Methods: Data have been collected in the context of the ongoing prospective Autism-Birth-Cohort-Study, funded by the Norwegian Institute of Public Health, and National Institute of Health of the USA. Mothers of 13,346 18-months-old children completed a questionnaire including items about their child’s socio-emotional development and behaviour. The DSM-IV three-domain classification of ASD at 18 months was tested using exploratory and confirmatory factor analyses. Latent Class Analysis (LCA) was conducted to empirically identify homogeneous groups of children.

Results: An exploratory factor analysis extracted a three-factor solution with a satisfactory fit and meaningful interpretation. Confirmatory factor analysis demonstrated further support for this three-factor structure. LCA revealed four 4 classes. Class 1 (0.7% of  sample) has the highest scores of all classes on the domains of Social interaction and Communication, but moderate scores on the domain Stereotypies; Class 2 (18.6% of sample) scores in-between class 1 and 4 on Social interaction and Communication but scores low on Stereotypies; Class 3 (13.4%) scores low on the domains of Social interaction and Communication, but the highest of all classes on the domain of Stereotypies; Class 4 (67.4% of sample) represents the reference group which relatively low scores on all domains.

Conclusions: The underlying structure of ASD symptoms is generalizable to the population at 18 months. In the general population four classes can be identified with a clear distinction between the clusters Social interaction and Communication on one hand and the cluster Stereotypies on the other hand.