In autism spectrum disorder (ASD), medical conditions in infancy could be predictive markers for later ASD diagnosis. In this study, electronic medical records of 579 autistic individuals and 1897 matched controls prior to age 2 were analyzed for potential predictive conditions. Using a novel tool, the relative association of each condition in the autistic group was compared to the control group using logistic regressions across medical records. Generalized convulsive epilepsy, nystagmus, lack of normal physiological development, delayed milestones, and strabismus were more likely in those later diagnosed with ASD while perinatal jaundice was less likely to be associated. Lesser-known conditions, such as strabismus and nystagmus, may point to novel predictive co-occurring condition profiles which could improve screening practices for ASD.
Keywords Autism • Early identification • Electronic medical recordsIndividuals with autism spectrum disorders (ASD) experience a significant number of comorbid medical conditions, ranging from psychiatric, gastrointestinal, and sleep conditions to neurological conditions throughout their lifetimes (Alexeeff et al., 2017;Croen et al., 2015;Fombonne et al., 2020). Many of these conditions may coincide with an ASD diagnosis or appear prior to an ASD diagnosis. However, few studies to date have investigated the onset patterns of these comorbid conditions, and little remains known about the specific medical phenotypic profiles in infancy that may serve as predictive markers for later ASD diagnosis.Leveraging a large dataset derived from electronic medical records, the current study aimed to fill this gap in knowledge by retrospectively identifying early-onset conditions that may be predictive of a later ASD diagnosis. Identification of these risk markers could aid targeted screenings, improve diagnosis, and allow for earlier intervention. Using the novel tool, pyPheWAS, we hypothesized that there would be multiple comorbidity profiles in this population that would be associated with a later ASD diagnosis.
Methods
Patient SampleElectronic health record (EHR) data was pulled from an anonymized database at Vanderbilt University Medical Center for individuals with ASD and age-, sex-, and