King and Bearman Respond

We appreciate the interest that Reichenberg et al. have shown in our work. In our article, we showed that examining parental age effects by birth cohort can yield important insight into variability in risk across time and ameliorate risk inflation arising from pooling data across birth cohorts.1 A recent study published by Grether et al.2 further affirmed the value of careful attention to temporality in risk, rather than diminishing the importance of cohort analyses as Reichenberg et al. suggested. Using the same dataset employed in our analyses but extending it to several additional birth cohorts, Grether et al.'s analysis revealed a striking temporal pattern of maternal risk. Among early birth cohorts born between 1989 and 1994 the autism risk associated with maternal age was consistently and substantially greater than was the risk among later born cohorts. The risk associated with maternal age for cohorts born after 1994 never approached the levels of risk observed among earlier cohorts. This not only has implications for pooling data but, more importantly, may offer etiological clues. Simply controlling for birth cohort as Reichenberg et al. suggested is insufficient. Rather, it is necessary to allow the effect of age to vary across years either by including a parental age-cohort interaction3 or by stratifying by birth cohort as we did.1 We absolutely agree with Reichenberg et al. that the importance of parental age as a risk factor for autism has been well established and can be detected in most populations. However, we believe that it is time to move beyond this fact to understand how and why parental age is associated with autism risk. One important step toward understanding how parental age may be causally related to autism is to begin to disentangle the differential importance of maternal and paternal age. There are several possible age-related biologic pathways through which parental age could influence risk for autism. These are different for men and women. Untangling the relative importance of maternal and paternal age may yield important insight into which biologic mechanisms may be causally related to autism. We obviously agree that parental age matters for autism, but we think that considering the temporal patterning and parental distribution of risk may allow us to move from just asserting that parental age matters to understanding why it matters—a goal we know that Reichenberg et al. share with us

Autism · Cohort · Cohort effect · Cohort study · Psychiatry · Sociology · Autism Spectrum Disorder Research · Demography · Epigenetics and DNA Methylation · Genetic Associations and Epidemiology · Internal Medicine · Medicine

• Estimated Autism Risk and Older Reproductive Age

  Marissa King, Marissa D King et al.•American Journal of Public Health•2009

  Add to list