Risk assessment of diet in a community-based study

A community-based screening for colorectal cancer targeted to all residents over 30 in Luh-Guu Town in Taiwan was conducted in mid-1986 and diet history was taken at the same time to examine the relationship between diet and screening positives of Hemoccult Test. Using dichotomous screening result as the outcome variable, we divided the exposure variable, diet history, into 3 categories: (A) meat less than vegetable, (B) meat = vegetable, and (C) meat greater than vegetable. The control variables in this study were age, sex, and upper G.-I bleeding history. The bivariate analysis with Chisquare test and the multivariable analysis with logistic regression were applied respectively. Although upper G-I bleeding was a protective factor for screening positive (OR = 0.5732 and 95% C.I = 0.34-0.96), it was not a confounder in this study. The main effect of age and sex were not correlated with screening positive. The interaction terms up to 3-factor were not significant. Based on the best fitted model in this study, those taking more meat than vegetable have a 1.74 times higher risk for screening positive of colorectal cancer than those taking less meat than vegetable (OR (3.1) = 1.74, 95% C.I. = 1.09-2.80). However the 95% C.I. for OR (2,1) and OR (3,2) included the null value OR = 1 [OR (2,1) = 1.30, 95% C.I = 0.93-1.82 and OR (3,2) = 1.34, 95% C.I = 0.84-2.08)]. High vegetable together with low meat diet is recommended.