Identifying confounding factors in studies of the relationship between colorectal cancer and diet

Abstract

Introduction. Diet is the second preventable cause related to the development of cancer. Given its multi-causal nature, in studying the relationship between cancer and diet, other factors with potential confounding effect must be considered to avoid bias in risk estimates. Objectives: a) Identifying the confounding effect of physical activity level, smoking habits and socioeconomic status in the relationship between colorectal cancer and cooked red meat, fiber and alcohol intake; b) Assessing the effect of dietary factors on the occurrence of colorectal cancer, considering the confounding variables identified.

Methods. A case-control study was conducted (102 cases with colorectal cancer and 217 controls) in Cordoba, Argentina, over 2006-2011. A bivariate analysis, between food variables and the presence of colorectal cancer, and a Mantel-Heanzel analysis, stratifying by the potential confounders, were conducted. Finally, multiple logistic regression models were constructed, including the confounding variables.

Results. Confounding effect of the socioeconomic status related to cooked red meat, fiber and alcohol intake, and physical activity level was verified. There was no association between fiber intake and colorectal cancer (OR 0,71; IC95% 0,31-1,62), while a promoting effect of low socioeconomic status (OR 1,75; IC95% 0,95-2,60), and daily energy intake (OR 1,0003; IC95% 1,00008-1,0006) were found.

Conclusion. It is recommended to consider socioeconomic status and physical activity as adjusted factors when conducting food risk analysis in the study of the relationship between colorectal cancer and diet in Argentina.