Gestational age at birth and risk of autism spectrum disorders in Alberta, Canada

To examine the association between autism spectrum disorders (ASD) and each completed week of gestation using a graphical method of presenting results at all possible categorizations of gestational age (GA).

The risk of ASD in a total of 218110 singleton live births with complete data from Alberta, Canada between 1998 and 2004 was examined through linkage to health insurance records. The relative risk of developing ASD according to the 21 dichotomizations of shorter gestation (GA ≤ 23 weeks vs >23 weeks to ≤ 43 weeks vs >43 weeks, in 1-week increments) was calculated using log-binomial regression and adjusted for fetal sex, socioeconomic status, and birth year.

We observed a gradual increased risk of ASD with shorter gestation. Cutoffs only between 29 and 40 weeks clearly denoted an elevated risk of developing ASD compared with longer gestation, and the risk increased with earlier GA cutoff. The results were not affected by sex or measures of fetal growth.

Our data confirm the role of shortened gestation in ASD risk. We warn against the use of prespecified or a data-driven GA cutoff, however; instead, we recommend systematically examining all plausible cutoffs for GA to avoid overstating the homogeneity of risk in children on either side of a given cutoff, as well as to increase the comparability of studies.