All of the data were checked for completeness at the end of each interview. The data were coded and analyzed using the SPSS package (21) . To assess the differences between the control groups, tea consumption, demographic, lifestyle, physical activity, and dietary variables were initially compared between the two hospital control groups, and then compared between hospital and community controls. The consistency of findings among the three groups would suggest that hospital bias was minimal. To assess potential survival bias, data for the 127 patients (interviewed within 3 months from diagnosis) and data for all of the cases (interviewed within 3 years from diagnosis) were also analyzed separately.