Screening for Autism Spectrum Disorder at 18 Vs. 30 Months In Extremely Preterm Infants

Background: The AAP recommends screening for autism spectrum disorder (ASD) in all infants at 18-24 months. But accuracy of early screening in extremely preterm infants is unknown. High rates of positive screen for ASD in extremely preterm infants at 18 months are reported but the rate of ASD diagnosis is unknown. Screening at 30 months may result in fewer false positives than at 18 months.

Objectives: To determine the rate of positive screen for ASD at 18 and 30 months in extremely preterm infants using three ASD screens (one validated in high risk children and two that involve direct observation of the infant); to determine the relationship of positive ASD screen to cognitive and language delay.

Methods: All infants born <28 weeks gestational age at Women and Infants Hospital and seen for 18 and/or 30 months follow-up from 5/01/08-10/30/10 were screened using the Pervasive Developmental Disorders Screening Test, 2nd edition, Stage 2 (PDDST) and 2 items from the Autism Diagnostic Observation Schedule, response to name (RN) and response to joint attention (RJA).  Association between number of positive screens and cognitive or language composite score <70 on the Bayley Scales of Infant Development, 3^rd edition) were determined.  Diagnostic assessment was performed on infants with at least one positive screen at 30 months.

Results: At 18 months 28/152 (18%) had one or more positive screen.  At 30 months 12/116 (10%) had one or more positive screen.  Positive screen at 18 or 30 months was associated with cognitive and language delay.  There was little overlap between the screens at either time point (Table 1).  All infants who failed all 3 screens at 30 months were later diagnosed with ASD.

Conclusions:   The lower rate of screen positive for ASD at 30 months than 18 months in extremely preterm infants suggests that many positive 18 month screens may be false positives, possibly due to high rates of language and cognitive impairment in these infants.  Diagnostic confirmation is needed to determine true rate of ASD and accuracy of early ASD screening in extremely preterm infants.  However, these results indicate that later screening and/or multiple screening data points improve screening accuracy.

Table 1: Screen overlap

	18 months (n=152)	30 months (n=116)
Failed one screen        PDDST             RN         RJA	10 (7%)   3 (2%)   5 (3%)	4 (3%) 1 (1%) 0
Failed two screens        PDDST and RN             PDDST and RJA        RN and RJA	0   0   5 (3%)	1 (1%) 1 (1%) 1 (1%)
Failed all three screens	5 (3%)	4 (3%)