///2015 Abstract Details
2015 Abstract Details2019-08-02T16:54:43-05:00

Labor Induction and Offspring Risk of Autism Spectrum Disorder

Abstract Number: BP-03
Abstract Type: Original Research

Brian T Bateman MD, MSc1 ; Brian M D’Onofrio PhD2; Martin E Rickert PhD3; Sonia Hernandez-Diaz MD, DrPH4; Anna S Oberg MD, MPH, PhD5

Introduction: Recently, a large population-based study identified an association between induction/augmentation of labor and the risk of autism spectrum disorder (ASD) in the offspring [1]. However, the study has been criticized for being unable to control for all relevant confounding factors. We therefore sought to examine the association between labor induction and ASD in Swedish register data, with ability to include a sibling comparison. This approach controls for all confounders shared by full siblings, including shared genes, early environment, and stable maternal factors thus providing a more valid causal estimate of the effect of induction.

Methods: Linkage between population registers in Sweden facilitated a nation-wide cohort study of all live singleton births in Sweden (recorded in the Medical Births Register) from 1991 to 2001. Exposure was defined by the presence or absence of induction of labor prior to delivery. The outcome was identified by specialist diagnosis of ASD in offspring through the end of 2009. In the initial analysis, we assessed the association between labor induction and offspring ASD using Cox proportional hazard regression, with age as the underlying time scale and allowing censoring at age of migration, death or end of follow-up. We adjusted a baseline model for factors that always vary between pregnancies (mother’s age, parity and birth year), then added stable maternal characteristics and factors unique to each pregnancy. Finally, we accounted for all factors shared by full siblings using a fixed effects model in which the underlying hazard is allowed to vary between mothers (so that the contrast is made within siblings), while maintaining adjustment for unique pregnancy level covariates.

Results: The full cohort included 978,981 births, of which 10,329 were diagnosed with ASD (1.1%). In complete case analysis, labor induction was significantly associated with offspring ASD in the baseline model (Hazard ratio [HR], 1.27; 95% CI, 1.19 to 1.35). After adjustment for all measured factors, the association to both ASD was still present, albeit attenuated (HR, 1.17; 95% CI, 1.09 to 1.27). However, when further adjustment was made using a fixed-effects model (comparing outcomes in siblings discordant with respect to induction) to account for all factors shared by full siblings, labor induction was no longer associated with offspring ASD (HR, 1.04; 95%CI, 0.85 to 1.27). Analysis using a fixed-effects model in first cousins showed commensurate results.

Conclusion: In this nation-wide sample of live births we observed no meaningful relationship between induction of labor and the risk for ASD. Our findings suggest that concern about this outcome should not factor into the clinical decision about whether to induce labor.


[1] JAMA Pediatr. 2013 Oct;167(10):959-66.

SOAP 2015