Plant-based food groups and cardiovascular disease: inverse associations with variable mechanistic support.

This review integrates recent evidence on the associations between major plant-based food groups and cardiovascular disease (CVD), the biological plausibility of the proposed mechanisms, certainty-of-evidence assessments, and recurrent methodological limitations. PubMed and Web of Science were searched for English-language publications from 2020 to July 2026, with earlier studies included when evidence was sparse. Association evidence was derived primarily from umbrella reviews and systematic reviews of observational studies, supplemented by prospective cohort studies where review-level evidence was limited. Mechanistic evidence from human intervention studies, animal models, and in vitro investigations was synthesized using a three-tier framework. Higher intakes of whole grains, legumes or pulses, fruits and vegetables, and nuts and seeds were generally associated with lower risks of CVD outcomes. However, the magnitude and consistency of associations varied across food groups and outcomes. Human trials most consistently showed reductions in low-density lipoprotein cholesterol with cereal bran, oat β-glucan, legumes, and nuts relative to comparators, together with modest improvements in lipid profiles following fruit and vegetable interventions. Evidence for other proposed effects was more limited or heterogeneous and, for several pathways, derived mainly from preclinical models. Certainty-of-evidence assessments varied and several limitations recurred across the literature. Higher consumption of major plant-based food groups is generally associated with lower risks of several cardiovascular outcomes. However, the observational nature of the epidemiological evidence limits causal inference. Moreover, heterogeneity in findings and recurrent methodological limitations across the included sources warrant caution in drawing firm conclusions. Future research should standardize food-group definitions and dietary assessment methods and strengthen the integration of epidemiological and mechanistic evidence.