The Association of Paternal IQ with Autism Spectrum Disorders and Its Comorbidities

Background:  Both Leo Kanner and Hans Asperger noted that the parents of the autistic children they originally described tended to be highly educated and intelligent, sewing the seeds of the hypothesis that Autism Spectrum Disorders (ASDs) are a case of “brilliance gone awry”. The association between genetic risk for ASD and intelligence in the general population and the notable excess of ASD cases born to parents working in highly technical professions would seem to support this hypothesis.

However, results examining cognitive abilities amongst ASD cases and their family members have been mixed. One potential issue is that co-morbidities common to ASD, such ADHD and intellectual disability (ID) that are themselves heritable impairments of cognitive functions, are often not taken into account.

Objectives:  Our goal was to examine the association between paternal intelligence and ASD, accounting for co-morbidity with ADHD and ID, in a population-based register study.

Methods: The study included 360,151 children, born 1984-2008, resident in Stockholm, Sweden, with a father that was conscripted to the Swedish military. Paternal IQ was measured on the stanine scale (1-9) during conscription at age 18. Paternal IQ subscale values (Logic, Verbal, Spatial, and Technical Reasoning) were available for 309,803 children. Children’s diagnoses were recorded in Swedish health registers.

The association of paternal IQ with offspring risk for ASD, ADHD, ID and overlapping co-morbid groups was evaluated using restricted cubic spline models with paternal IQ=5 (average intelligence) as the referent, to allow flexible fitting of the relationship between paternal IQ and offspring risk for neurodevelopmental disorders.

Results: We observed a U-shaped relationship between paternal IQ and offspring risk of ASD with no co-morbidities (OR_IQ=1 1.20, 95% CI 1.00–1.45; OR_IQ=9 1.31, 1.14–1.50), after adjustment for socioeconomic status (SES). The association of low paternal IQ and ASD with co-morbid ID was stronger, though the overall relationship was similar (OR_IQ=11.81, 1.29–2.54; OR_IQ=9 1.30, 0.98–1.72). For ASD with co-morbid ADHD, only the association with low paternal IQ was apparent (OR_IQ=1 1.49, 1.24–1.81; OR_IQ=9 1.06, 0.88–1.28). The association between high paternal IQ and risk for offspring ASD alone was driven largely by the fathers’ technical subdomain score (OR_{Technical IQ=9} 1.51, 1.30–1.76). The association between paternal IQ and ID alone was striking (OR_IQ=1 4.53, 3.68–5.57; OR_IQ=9 0.66, 0.47–0.93), and a similar but weaker pattern was observed for ADHD alone (OR_IQ=11.65, 1.50–1.82; OR_IQ=9 0.71, 0.63–0.80). For all outcomes, similar results were observed before and after adjustment for the marked gradient in SES between paternal IQ levels. The associations of low paternal IQ with ASD or ADHD outcomes tended to be attenuated and the associations of high paternal IQ with ASD outcomes were strengthened when SES was considered.

Conclusions:  High paternal IQ was indeed associated with children’s risk of ASD, though this was most apparent in the group without co-morbidities. This risk was driven largely by the fathers’ score on the technical sub-scale. Overall, the pattern for risk of ASD was distinct compared to that for ID or ADHD alone.