Abstract
Research associating the increased prevalence of familial autoimmunity with neuropsychiatric disorders is reliant upon the ascertainment of history of autoimmune diseases from relatives. To characterize the accuracy of self-report, we compared self-reported diagnoses of 18 autoimmune diseases using an online self-report questionnaire to the electronic medical record (EMR) diagnoses in 1,013 adult (age 18-70 years) patients of a primary care clinic. For the 11 diseases meeting our threshold observed prevalence, we estimated sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for self-reported diagnoses under the assumption that EMR-based diagnoses were accurate. Six diseases out of 11 had either sensitivity or PPV below 50%, with the lowest PPV for dermatological and endocrinological diseases. Common errors included incorrectly self-reporting type 2 diabetes mellitus (DM), when type 1 DM was indicated by the EMR, and reporting rheumatoid arthritis when osteoarthritis was indicated by the EMR. Results suggest that ascertainment of familial autoimmunity through self-report contributes to inconsistencies and inaccuracies in studies of autoimmune disease history and that future studies would benefit from incorporating EMR review and biological measures.