Abstract
Background: We have shown that mothers of children with autism spectrum disorder (ASD) have considerably higher burden of disease around pregnancy than mothers of unaffected children. Nevertheless, the full spectrum of maternal medical diagnoses has never been comprehensively assessed for their association with offspring ASD.
Objectives: The goals of our study were (i) to identify all maternal medical diagnoses around pregnancy associated with ASD risk in offspring, and (ii) to distinguish the effects of maternal health before, during and after pregnancy on the risk of ASD.
Methods: We used medical information on Israeli children (N=89,483) born 1998-2007 and their parents, to identify diagnoses more common around the pregnancies (pregnancy + preceding 1 year) of mothers of children with ASD, than of mothers of control children. In all analyses, we used logistic regression analysis and controlled for child’s year of birth, maternal age and family’s socioeconomic status. In order to eliminate the associations arising due to a correlation between diagnoses, we tested the effects of all diagnoses significantly associated with ASD after false discovery rate adjustment for multiple testing (q < 0.05) in a joint, penalized regression model. In order to resolve whether these associations are underlain by shared genetic factors, or by disruption of fetal development, we compared the ASD risk associated with maternal number of diagnoses during pregnancy, with the risk associated with diagnoses that are unlikely to directly affect the fetus, i.e. those recorded before or after pregnancy.
Results: Twenty eight ICD-9 codes recorded in mothers 21 months prior to child’s birth were associated with ASD in offspring in the initial, single-diagnosis models. Entering these items into a single penalized regression model eliminated all but 13 of these associations, suggesting the issue of phenotype correlation in epidemiological studies. Maternal diagnoses most strongly associated with ASD included psychiatric, metabolic and genitourinary diagnoses (e.g. depression: OR=1.25 (1.04-1.48); non-inflammatory disorders of cervix: OR=1.22 (1.06-1.41); disorders of adrenal glands: OR=1.71 (1.03-2.86)). Comparison of the effects of maternal disease burden before, during and after pregnancy revealed that only the number of diagnoses recorded during and after, but not before, were significantly associated with ASD risk in children (during: OR=1.11 (1.07-1.14), after OR=1.05 (1.01-1.08)).
Conclusions: Our analyses suggest pervasive effects of maternal health on the risk of ASD in offspring. We have identified both known and potentially novel early-life risk factors for ASD, highlight the utility of unbiased, data-driven approaches in epidemiology of ASD.
Objectives: The goals of our study were (i) to identify all maternal medical diagnoses around pregnancy associated with ASD risk in offspring, and (ii) to distinguish the effects of maternal health before, during and after pregnancy on the risk of ASD.
Methods: We used medical information on Israeli children (N=89,483) born 1998-2007 and their parents, to identify diagnoses more common around the pregnancies (pregnancy + preceding 1 year) of mothers of children with ASD, than of mothers of control children. In all analyses, we used logistic regression analysis and controlled for child’s year of birth, maternal age and family’s socioeconomic status. In order to eliminate the associations arising due to a correlation between diagnoses, we tested the effects of all diagnoses significantly associated with ASD after false discovery rate adjustment for multiple testing (q < 0.05) in a joint, penalized regression model. In order to resolve whether these associations are underlain by shared genetic factors, or by disruption of fetal development, we compared the ASD risk associated with maternal number of diagnoses during pregnancy, with the risk associated with diagnoses that are unlikely to directly affect the fetus, i.e. those recorded before or after pregnancy.
Results: Twenty eight ICD-9 codes recorded in mothers 21 months prior to child’s birth were associated with ASD in offspring in the initial, single-diagnosis models. Entering these items into a single penalized regression model eliminated all but 13 of these associations, suggesting the issue of phenotype correlation in epidemiological studies. Maternal diagnoses most strongly associated with ASD included psychiatric, metabolic and genitourinary diagnoses (e.g. depression: OR=1.25 (1.04-1.48); non-inflammatory disorders of cervix: OR=1.22 (1.06-1.41); disorders of adrenal glands: OR=1.71 (1.03-2.86)). Comparison of the effects of maternal disease burden before, during and after pregnancy revealed that only the number of diagnoses recorded during and after, but not before, were significantly associated with ASD risk in children (during: OR=1.11 (1.07-1.14), after OR=1.05 (1.01-1.08)).
Conclusions: Our analyses suggest pervasive effects of maternal health on the risk of ASD in offspring. We have identified both known and potentially novel early-life risk factors for ASD, highlight the utility of unbiased, data-driven approaches in epidemiology of ASD.
Original language | English |
---|---|
State | Published - 3 Jun 2020 |