Background and aims There is increasing interest and speculation about the relationship between prematurity and autism spectrum disorders (ASD). The majority of studies have been retrospective with limited collateral information about the child. This study sought to determine how children designated by their parents’ ratings for ASD differed in premature morbidities, cognitive abilities, autism ratings by a psychologist, and executive functioning skills.

Method This prospective longitudinal study of 174 extremely premature infants born ≤30 weeks gestation were followed to age four. The BSID-III was administered at 6, 15 and 24 months with the Infant/Toddler Sensory Profile being completed at 6 months of age. At 4 years, the parents completed the Gilliam Autism Rating Scale (GARS) and the Behavioural Rating Inventory of Executive Function-Preschool Version (BRIEF-P). The psychologists completed the Children’s Autism Rating Scale (CARS) while administering the WPPSI-III. GARS and CARS scores were collapsed into autistic versus non-autistic categories and used to analyse the data.

Results Parent’s GARS identified a significantly higher number of children (38%) with ASD than did the examiner’s ratings (CARS, 10%). GARS’ groups did not differ significantly in NICU length of stay, incidence of sepsis, or severe IVH. At 6, 15, and 24 months the ASD group scored significantly lower on the BSID-III and at four years they had a lower mean IQ (89 v. 105, p < 0.0001). Executive functioning for the ASD group on the Brief-P was poor (70.03 v. 47.4, p < 0.0001).

Conclusion Parent’s perception of NICU morbidities and cognitive delays affect the identification of ASD in extremely preterm infants.