METHODS: Descriptive statistics were generated for demographics, smoking habits, histology, and EGFR mutation subtypes, stratified by status of brain metastasis at diagnosis. Chi-squared tests and t-tests were used for testing associations of categorical variables and continuous variables with brain metastasis at diagnosis, respectively. Logistic regression models were used to study the association between race and brain metastasis at diagnosis, with and without adjusting for age at initial diagnosis, gender, EGFR mutation type, smoking status, and histology. Odds ratio (OR) and corresponding 95% confidence intervals (CI) were obtained. All analyses were two sided and a p value <0.05 were considered significant.