Background: Increased interest in the potential cardio-protective effects of fruit and vegetables is currently unsupported by systematic reviews of the reported associations of these foods with risk.
Method: All ecological, case-control, cohort studies and unconfounded trials in humans were eligible for inclusion. Eligible outcomes were symptomatic coronary heart disease, stroke and total circulatory disease. Only studies of diet that reported on fresh fruit and vegetables or a nutrient which could serve as a proxy (reversing the usual direction of inference) were included. MEDLINE (1966-1995) and EMBASE (1980-1995) were searched using the terms cerebrovascular disorder, coronary heart disease, fruit(s) and vegetable(s) as keywords. Personal bibliographies, books and reviews were also searched, as were citations in located reports.
Results: For coronary heart disease nine of ten ecological studies, two of three case-control studies and six of 16 cohort studies found a significant protective association with consumption of fruit and vegetables or surrogate nutrients. For stroke three of five ecological studies, none (of one) case-control study and six of eight cohort studies found a significant protective association with consumption of fruit and vegetables or surrogate nutrients. For total circulatory disease, one of two cohort studies reported a significant protective association. No attempt was made to arrive at a summary measure of the association because of the differences in study type, study quality and the different exposure measures used.
Conclusions: Although null findings may be underreported the results are consistent with a strong protective effect of fruit and vegetables for stroke and a weaker protective effect on coronary heart disease. Greater use of food-based hypotheses and analyses, would complement existing nutrient-based analyses and help guide the search for underlying causes.