What Can State-Birth Records Contribute to Our Understanding of ASD Risk?

Background: Autism Spectrum Disorder (ASD) is thought to be the most heritable of all neurodevelopmental and psychiatric disorders (Abrahams & Geschwind, 2008; O’Roak & State, 2008).  However, identification of specific genetic risk variants has proved to be a challenging endeavor and many researchers are now suggesting that there may be multiple pathways to this disorder, including prenatal and postnatal insult (Geschwind & Levitt, 2007).  Further investigations into well phenotyped samples are needed in order to help elucidate medical and environmental factors that may contribute to ASD risk.   

Objectives: The purpose of this study was to link patients within a local, well phenotyped ASD registry (N=1647) at a large university in the mid-south to official state birth records in order to examine early factors potentially associated with ASD risk. These birth records contain approximately 140 variables pertaining to infant, parent and prenatal health practices.

Methods: Multiple personal identifiers (e.g. names, DOB’s, addresses) were used to match ASD registry with birth records. Birth records matched on DOB-year, DOB-month, gender and zip code were selected for the control (N=2200).

Results: Several significant (all p’s =< .05) univariate comparisons of child, parent, prenatal practice and reproductive history variables were identified.  The ASD group differed from the control group on the following variables:  higher APGAR scores  (8.8 v 8.7), shorter gestational age (38.1 v 38.4), increased interval between births (54.9 v 48.9 months), started prenatal care earlier (month 2.7 v 2.9), older maternal age (28.5 v 27.0), less use of antibiotics (25.5% v 24.1%), more birth complications (80.2% v 69.4%), fewer obstetrical procedures (26.9% v 37.2%), fewer vaginal deliveries (55.5% v 65.7%), more transfusions (.5% v .01%), more white (89.6% v 83.9%), fewer Hispanic (5.5% v 11.4%), more with advanced education (college or more) (68.4% v 55.3%), and more were married (73.7% v 69.4%).   Since the variables in the univariate comparisons are correlated to a greater or lesser degree, a logistic regression analysis comparing the ASD and the control group was completed.  The logistic regression confirmed the significant independent contribution of gestational age, inter-pregnancy interval, month care began, antibiotic administration, and maternal ethnicity. 

Conclusions:  Although several findings represent replications of previous findings (i.e., gestational age, maternal age, fewer vaginal deliveries) several findings are in conflict with previous reports (i.e., higher APGAR scores, increased interval between births, less use of antibiotics).  Associations between ASD and non-ASD characteristics documented within medical birth records may help us better understand factors potentially associated with ASD risk. However, such methodologies need to demonstrate replication across samples and be linked to scientific enterprise that can move from association to actual understanding of true risk and risk pathways.