Gestational Age and the Risk of Autism Spectrum Disorders: Findings from the Stockholm Youth Cohort

Background: The rising prevalence of autism spectrum disorders (ASD) is mainly driven by the increase in high-functioning ASD (i.e. ASD without intellectual disability (ID)). Gestational age at birth (GA) has been identified as a risk factor of ASD. However, the question of how the specific length of gestation might increase the risk of ASD remains unanswered.

Objectives: We investigated the relationship between GA and high- vs. low-functioning ASD, by quantifying the risk of ASD by each week of gestation in individuals with and without co-occurring ID.

Methods: A sample of 483,843 non-adopted, singleton live births between 1984 and 2007 was derived from the Stockholm Youth Cohort, a register-based cohort in Stockholm County, Sweden. Case status was ascertained using national and regional registers covering all pathways to ASD diagnosis and care in Stockholm County. A total of 9,325 ASD (2,319 ASD with ID, 7,006 ASD without ID) cases were included in the study sample. Splines in generalized additive models were used to model the non-linear relation of GA with ASD, adjusting for sex, birth year, parity, parental ages, gestational diabetes, gestational hypertension, maternal BMI at first antenatal visit, maternal history of depression, family income, maternal educational attainment and maternal country of origin.

Results: The prevalence of both high- and low-functioning ASD was increased in pre- and post-term births. There were approximately 17 individuals with ASD per 1,000 live births at term (GA = 40 weeks). This measure was 43 per 1,000 among very preterm births at 28 weeks of GA, and 25 per 1,000 among post-term births at 43 weeks of GA. The risk of high-functioning ASD monotonically decreased as GA increased. Odds ratios (95% confidence interval) of ASD without ID among those born at 28 and 43 weeks of GA was 2.0 (1.6‒2.4) and 0.9 (0.9‒1.0), respectively, compared to the risk of ASD in term births. On the other hand, the risk of low-functioning ASD was elevated in both pre- and post-term births. Odds ratios for co-occurring ASD and ID were 4.2 (3.0‒5.9) and 1.3 (1.1‒1.6) in infants born at 28 and 43 weeks of GA, with respect to the same referent group. Causal mediation analyses are underway to determine what factors may explain these observed associations.

Conclusions: We confirmed previous reports that both shortened and prolonged gestation were associated with higher risk of ASD and demonstrated that this relationship may be more complex than previously thought. In high-functioning ASD, each additional week of gestation lowers the risk incrementally, yet in a non-linear fashion. In low-functioning ASD, the risk associated with gestational length exhibits a U-shaped pattern. Our mediation analyses will elucidate how gestational length may play distinct roles on the causal pathways to ASD in the presence or absence of co-occurring ID.