Infertility and Risk of Autism Spectrum Disorder in Children

This cohort study evaluates the association between infertility and autism spectrum disorder in children in Canada.

e Not applicable, as "the [proportion-mediated measure] is problematic when the natural direct effect and natural indirect effect operate in different directions.
One can obtain a proportion mediated much larger than 100%, and the measure is no longer really meaningful.Even more dramatically, if the natural direct and indirect effects are of roughly the same magnitude but of opposite signs, then the total effect will be close to zero and the proportion-mediated measure then takes the natural indirect effect and divides it by a number close to zero which will result in an enormous (and again meaningless) proportion."Source: VanderWeele TJ.

eTable2. Risk of Autism Spectrum Disorder (ASD) by Mode of Conception. These analyses are limited to singleton live births. Mode of conception No. with ASD/ No. at risk Rate of ASD per 1000 person-years Unadjusted hazard ratio (95% CI) Adjusted hazard ratio (95% CI) a
These analyses are limited to live-born children of mothers aged < 45 years, including by mode of conception.
Analysis limited to 15,966 live-born children among individuals who underwent In vitro fertilization or intracytoplasmic sperm injection a Adjusted for maternal age, parity, income quintile, rurality, immigration status, smoking, obesity, any drug or alcohol use, maternal history of mental illness or ASD, pre-pregnancy diabetes mellitus or chronic hypertension, and infant sex.eTable 3. Risk of Autism Spectrum Disorder (ASD) by Mode of Conception.aAdjusted for maternal age, income quintile, rurality, immigration status, smoking, obesity, parity, any drug or alcohol use, maternal history of mental illness or ASD, pre-pregnancy diabetes mellitus or chronic hypertension, and infant sex.eTable 4.

Risk of Autism Spectrum Disorder (ASD) by Mode of Conception.
Analysis additionally adjusted for calendar year.Analysis limited to 23,519 live-born children of individuals who underwent in vitro fertilization or intracytoplasmic sperm injection a Adjusted for maternal age, parity, income quintile, rurality, immigration status, smoking, obesity, any drug or alcohol use, maternal history of mental illness or ASD, pre-pregnancy diabetes mellitus or chronic hypertension, infant sex, and calendar year eTable 5.

Mediation Analysis of the Effect of Selected Adverse Pregnancy Outcomes on the Association Between Mode of Conception and Associated Risk of Autism Spectrum Disorder (ASD). Each mode of conception is compared to a birth in a mother with unassisted conception (reference group).
This analysis is restricted to 1,322,578 singleton live births.
b Statistically significant natural indirect effect at a P-value < 0.05.c Among 135,058 pregnancies.d Among 135,139 pregnancies.© 2023 Velez MP et al.JAMA Network Open.

Mediation Analysis of the Effect of Selected Adverse Pregnancy Outcomes on the Association Between Mode of Conception and Associated Risk of Autism Spectrum Disorder (ASD) Among 1,370,152 Live-Born Children.
Explanation in Causal Inference: Methods for Mediation and Interaction.New York, NY: Oxford University Press, 2015, page 48.Each mode of conception is compared to a birth in a mother with subfertility (reference group).Not applicable, as "the [proportion-mediated measure] is problematic when the natural direct effect and natural indirect effect operate in different directions.One can obtain a proportion mediated much larger than 100%, and the measure is no longer really meaningful.Even more dramatically, if the natural direct and indirect effects are of roughly the same magnitude but of opposite signs, then the total effect will be close to zero and the proportion-mediated measure then takes the natural indirect effect and divides it by a number close to zero which will result in an enormous (and again meaningless) proportion."Source: VanderWeele TJ.Explanation in Causal Inference: Methods for Mediation and Interaction.New York, NY: Oxford University Press, 2015, page 48.
b Statistically significant natural indirect effect at a P-value < 0.05.© 2023 Velez MP et al.JAMA Network Open.c Among 1,176,963 pregnancies.e e Among 1,159,016 pregnancies.eFigure 1.