We first tested the effect of covariates on our phenotypes: SC and DSM-IV AD. As expected, gender was a highly significant predictor of SC and DSM-IV AD, and was included as a covariate in all analyses. We identified cohort effects and therefore divided subjects into 4 cohorts based on their year of birth (< 1930, 1930 – 1949, 1950 – 1969, and ≥ 1970). For SC, the age-squared parameter was still significant after cohort effect was included, and the final model therefore included gender, age, age-squared and cohort. For DSM-IV AD, the age-squared parameter was not a significant factor after considering cohort and was omitted. The first principal component from the EIGENSTRAT analysis (pc1), while not statistically significant, was still included in all analyses to reduce the risk of false positive associations due to population stratification.
