Introduction The association between fruit and vegetable intake and colorectal cancer risk has been investigated by many studies, but remains a controversial issue because of inconsistent results and weak observed associations. We summarise the evidence from cohort studies in categorical, linear and non-linear dose-response meta-analyses.

Methods We searched PubMed for studies of fruit and vegetable intake and colorectal cancer risk up to May 2010. Prospective studies that reported RR estimates and 95% CIs of colorectal cancer associated with fruit and vegetable intake were included. Random effects models were used to estimate summary RRs.

Results Nineteen cohort studies were included in the meta-analysis. The summary RR for the highest vs the lowest intake was 0.92 (95% CI 0.86 to 0.99) for fruit and vegetables combined, 0.90 (95% CI 0.83 to 0.98) for fruit and 0.91 (95% CI 0.86 to 0.96) for vegetables. The inverse associations appear to be restricted to colon cancer. In linear dose-response analysis only intake of vegetables was significantly associated with colorectal cancer risk, summary RR 0.98 (95% CI 0.97 to 0.99) per 100 g per day. However, significant inverse associations emerged in non-linear models for fruits (p_{non-linearity} <0.001) and vegetables (p_{non-linearity} =0.001). The greatest risk reduction was observed when increasing intake from very low levels of intake. There was generally little evidence of heterogeneity in the analyses and there was no evidence of small-study bias.

Conclusion This meta-analysis indicates that there is a weak, but statistically significant non-linear inverse association between fruit and vegetable intake and colorectal cancer risk.