Abstract

Aims: An association between paternal age and type 1 diabetes (IDDM) among their offspring was recently reported as well as transgenerational responses in humans. This paper aims to assess the association of markers for prenatal exposures with IDDM.

Methods: We analysed data from two birth cohorts in Great Britain on 5214 cohort members from the National Child Development Study (NCDS) and 6068 members of the 1970 British Birth Cohort Study (BCS70) with full information on IDDM and explanatory variables using multivariate logistic regression.

Results: IDDM prevalence was 0.7% (95% CI 0.5–1.0%; n=38) in the NCDS and 0.4% (95% CI 0.3–0.6%; n=27) in the BCS70 cohort. Paternal age was not associated with IDDM possibly due to lack of sample power. Unexpectedly, a lowered prevalence of IDDM was observed among offspring of smoking fathers in both cohorts, with a combined odds ratio of 0.44 (95% CI 0.25–0.75). This association could not be explained by maternal smoking prior to, during or after pregnancy, number of siblings, parental social class, maternal and paternal age, or cohort. Maternal smoking in pregnancy did not alter the IDDM prevalence among offspring.

Conclusions: This unexpected finding may be explained by germ-line mutations or other mechanisms associated with paternal smoking. This phenomenon should be investigated and these results should not be used as a justification for smoking. Paternal exposures may be important in determining IDDM risk.