Objectives: To examine the association between paternal age, maternal age and autism in the offspring using robust and novel methodological approaches.
Methods: Multiple study methods were adopted. First, a meta-analysis included previously published, population-based, epidemiological studies. Second, a consortium of six birth cohorts from Australia, Denmark, Finland, Israel, Norway and Sweden with similar data on variables of interest was established. Linkage to national registers ascertained autism cases and associated perinatal, familial, and demographic data for all births in each cohort.
Results: Meta-analysis demonstrated an association of advancing paternal (>=40) and maternal age (>=35) with increased risk of autism across studies [OR=1.78 (95%CI:1.52-2.07); OR=1.38 (95%CI:1.28-1.49) for paternal and maternal age, respectively]. There was no evidence for publication bias; Heterogeneity across studies was substantially influenced by the covariates included in the analyses. Data from the six birth cohorts are currently being analyzed, first separately for each birth cohort, then followed by a pooled analysis. Informed by the meta-analysis, the separate analyses are aimed at characterizing the functional form of the relationship between parental age and autism and identifying confounders to be included in adjusted models. This will allow creating common scaling across birth cohorts, and reliable data pooling.
Conclusions: These findings, if corroborated in the additional analyses, will provide the strongest evidence to date of an association between advanced parental age and autism in the offspring. Previously proposed biological mechanisms for the relationship between parental age and autism, including de-novo aberrations, mutations and epigenetic alterations associated with aging, will be critically evaluated in relation to the findings from the pooled analysis.