Abstract

A recent review concluded that about 9% of all cancers in the UK are attributable to four dietary factors: low intake of fruits and vegetables, any consumption of meat, low intake of fibers, and high intake of salt. Other authors have reached different conclusions, from less than 1% to over 30%. Such a wide range reflects the uncertainties in the understanding of the relationship between dietary factors and human cancer, and the variability in the methods used to derive the estimates (e.g., choice of counterfactual, ‘optimal' exposure distribution). A review of the evidence linking low fruit and vegetable intake to stomach cancer illustrates how differences in the criteria used by review panels have led to different conclusions on the causal nature of the association and its strength. Aspects such as opportunity for bias, exposure misclassification, residual confounding, biologically relevant timing, and statistical power are at the core of the interpretation of the evidence from observational studies. Furthermore, the association between diet and cancer, and its implications for cancer prevention, can be addressed at multiple levels, from micro- and macro-nutrients, to food and food group, to dietary patterns.