ABSTRACT
There is evidence that diet and nutrition are modifiable risk factors for several cancers, but associations may be flawed due to inherent biases. Nutritional epidemiology studies have largely relied on a single assessment of diet using food frequency questionnaires. We conduct an umbrella review of meta-analyses of observational studies to evaluate the strength and validity of the evidence for the association between food/nutrient intake and risk of developing or dying from 11 primary cancers. It is estimated that only few single food/nutrient and cancer associations are supported by strong or highly suggestive meta-analytic evidence, and future similar research is unlikely to change this evidence. Alcohol consumption is positively associated with risk of postmenopausal breast, colorectal, esophageal, head & neck and liver cancer. Consumption of dairy products, milk, calcium and wholegrains are inversely associated with colorectal cancer risk. Coffee consumption is inversely associated with risk of liver cancer and skin basal cell carcinoma.