19885
Assisted Conception and Risk of Autism Spectrum Disorders in a Large Commercially Insured Population
(ASD) initiates during fetal development. Further, epigenetic modifications and de novo mutations appear to be important for autism etiology. Consequently, exposures that may affect gametes preconception and embryonic development soon after conception are of particular interest. Assisted conception (AC), which may include technical (i.e. IVF) and/or hormonal treatments has been hypothesized as a potential mechanism inducing epigenetic modifications and AC has been linked to a range of adverse birth outcomes. Previous epidemiologic studies have reported statistically significant positive unadjusted associations between AC and ASD risk, while in most instances adjusted associations have decreased in magnitude and no longer remained statistically significant. A few studies reported significantly increased risk among certain subgroups and one study found a significant negative association. Larger studies conducted in a variety of populations appear to be needed to determine if AC is associated with a small, but real increased risk of ASD.
Objectives: To assess the association between AC and ASD risk in a large commercially-insured US population.
Methods: Using the HealthCore Integrated Research Database, a cohort of 157,649 linked mother-child dyads meeting enrollment criteria (mother had enrollment for 12 months prior to delivery date and child had 1) enrollment at either the 3rd or 4thbirthday, and 2) a minimum of six months of continuous enrollment with at least one claim for a well child visit between ages 2.5 and 4.5) was identified. AC treatments were determined using the medical and pharmacy claims of mothers in the year prior to delivery. ASD diagnosis was determined by ≥1 ICD-9 diagnosis code in the medical claims of children between 18 months and 4.5 years of age. Demographic information (i.e. maternal age, child’s sex) was attained from enrollment files and other covariates were based on the claims of mothers and/or children. Crude and adjusted logistic regression was used to estimate odds-ratios. Potential confounders considered for inclusion in adjusted models were maternal age, year of birth, multiple birth, preterm delivery, low birth weight, baby’s sex, geographic region, and length of enrollment.
Results: 2,183 mothers (1.4%) had an AC technical treatment of in vitro fertilization or artificial insemination. 1,403 children (0.9%) had a diagnosis of ASD. The crude OR for AC compared to no AC was 2.17 (95% CI: 1.58, 2.96; p<0.0001). After adjustment, the OR was 1.34 (95% CI: 0.96, 1.85, p=0.082).
Conclusions: This study, which uses data from a large privately insured population, is consistent with much of the previous research from larger study populations in suggesting that AC effects on ASD risk are confounded by other ASD risk factors also associated with AC utilization. The magnitude of any real independent effect of AC is likely to be modest – here the adjusted effect was not statistically significant. Further work in this cohort will include a range of sensitivity analyses that will assess the robustness of the association estimate to measurement error and omitted confounders as well as exploration of the effect of individual AC treatments.