Fruit and vegetables have been suggested to help prevent excessive weight gain during adult life. Their low Energy density and high content of water and fiber could enhance satiation signals. Previous studies investigating this relation have been inconsistent so far. Several prospective observational studies and intervention studies have shown an inverse association between fruit and vegetable consumption and weight change. However, individuals who eat higher quantities of fruit and vegetables also tend to eat less meat, especially processed meat, saturated fat, and refined carbohydrates, all of which have been positively associated with weight gain. Therefore, it remains unclear whether the associations previously observed were fully attributable to fruit and vegetable intake itself or to an underlying dietary pattern. In addition, most previous observational studies were performed in small- to medium-size national samples. Homogenous dietary intakes and limited sample sizes could have reduced their power to observe an association, especially when studying specific subgroups of participants. We aimed to investigate whether higher intakes of fruit and vegetables were related to subsequent midterm changes in body weight in 373,803 men and women participating to the EPIC project.

Overall, fruit and vegetable intakes measured at Baseline were not associated with weight change in both men and women after an average of five years of follow-up¹. Fruit intake did not include nuts, olives, and fruit juices and vegetable intake did not include legumes, potatoes, and other tubers. We excluded participants who were likely to have modified their diet before baseline or to have misreported energy intakes. We also took into account physical activity, dietary pattern and other life style factors in our analysis. In addition, measurement error of diets has been partially corrected thanks to a randomized dietary calibration study. This result was in agreement with a recent review concluding that fruit and nonstarchy vegetables were not associated with amounts of subsequent excess weight gain and obesity².

However, fruit and vegetable intakes were inversely associated with weight change in participants who quit smoking during follow-up. This association had previously been reported in a sub-sample of the EPIC cohort3 but no other prospective cohort studied this subgroup specifically. This finding will need further investigation. If confirmed in other populations, it may have important public health implications because weight gain after smoking cessation is a frequent reason for relapse.

Several comments must be made on the present study:

First, weight at follow-up was self-reported in most centers and was therefore corrected with the use of a prediction equation to improve its accuracy. Concordant results were observed in the two centers with measured weights at both baseline and follow-up, which indicated that the observed associations were unlikely to be due to misreporting of weight only. Second, we only measured diet at baseline and could not rule out that residual confounding related to changes in diet during follow-up remained in our analyses. The time sequence between the change in diet and change in weight is critical in observational studies, and reverse causation may remain even when diet is measured at both baseline and follow-up. Finally, despite a high response rate at follow-up (80%), nonresponse was more likely in participants who reported a poorer health and unhealthy lifestyle, especially a low BMI or high BMI. This selection bias could have limited the generalizability of our results.

We resolved some of the limits encountered by previous studies  investigating the association between fruit and vegetable consumption and weight change in free living subjects. The very large sample size and heterogeneity of both dietary behaviours and obesity prevalence increased our statistical power to detect small associations and allowed us to explore a variety of interaction factors including change of smoking status. We also took into account for the first time the potential confounding effect of underlying dietary patterns and our results were not modified. We concluded that higher baseline fruit and vegetable intakes, while maintaining total Energy intakes constant, did not substantially influence midterm weight change overall but could help to reduce risk of weight gain in persons who stop smoking.