Association Between ART Conception and Autism in California 1997-2007

Background: The estimated US prevalence of autism spectrum disorders has increased rapidly over the past two decades and is now estimated as 1 in 88 children. During the same period, the use of assisted reproductive technology (ART) has increased rapidly. ART and autism have many common correlates including older parental age, higher parental education, multiple births, pre-term delivery, low birth weight, and pregnancy and labor complications. Few studies have investigated the relationship between ART and autism, and existing results are mixed and inconclusive. Given the increasing use of ART and the uncertain etiology of autism it is important to know whether these phenomena are associated.

Objectives: To assess the association between ART and diagnosed Autistic Disorder in a population-based sample of US births using the largest existing source of information on ART and autism.

Methods: We linked records from the California Birth Master Files for 1997-2007, the California Department of Developmental Services (DDS) autism caseload for 1997-2011, and the CDC’s National ART Surveillance System for live births in 1997-2007. Of 5,926,251 births, 48,865 infants (0.8%) were conceived through ART and 32,922 (0.6%) were later diagnosed with autism and enrolled with the DDS. We calculated autism risk for ART and non-ART births. To reduce the likelihood of ascertainment bias for autism diagnosis, we created an analytic subsample restricted to births to women age >20 years who had obtained a high school diploma or higher education, adequate prenatal care, and whose prenatal care and delivery were covered by a non-public payment source.  We estimated crude and adjusted Hazard Risk Ratios for the subsample, considering several demographic and pregnancy-related factors as potential confounders.  We also estimated adjusted Hazard Risk Ratios stratified by infant sex, parental age, maternal education, and parity. Finally, nested adjusted models were estimated to examine potential causal pathways for the ART-autism association.

Results: In the full population, the prevalence of diagnosed autism is twice as high for ART as non-ART births (crude HRR=2.3 [2.1-2.5]). The association is diminished in the analytic subsample (crude HRR=1.8 [1.6-1.9]); adjusting for demographics, maternal age, and parity further attenuates the association although it remains significant (HRR=1.4 [1.3-1.5]). Stratified adjusted analyses show significantly elevated HRRs in all subgroups, with a significantly stronger effect among children born to mothers <35 years than >35 years of age (adjusted HRR 1.7 [1.5-2.0] vs. 1.4 [1.2-1.5]).  However, after inclusion of adverse prenatal and perinatal outcomes potentially in the causal pathways (multiple birth, preterm, small for gestational age, maternal diabetes and hypertension, and cesarean delivery) the association was reduced to the null for both maternal age groups.

Conclusions: Findings from this analysis suggest a potential association between ART and autism. However, it is uncertain if the pathway for the observed association is a direct effect of the ART procedure: when adverse prenatal and perinatal factors are accounted for, the independent association is reduced to null.