By how much does fruit and vegetable consumption reduce the risk of ischaemic heart disease: response to commentary

In our review (Law & Morris, 1998), we presented analyses of data from 10 cohort studies yielding the summary estimate that the risk of ischaemic heart disease was 15% lower at the 90th than at the 10th centile of fruit and vegetable consumption. This 10th-90th centile difference in consumption is a realistic increase for an individual (Zino et al, 1997). The estimate of a 15% difference in heart disease mortality was similar to the expected difference in risk from the increase in potassium consumption (given the corresponding decrease in blood pressure) and the increase in folate consumption (given the corresponding decrease in plasma homocysteine) that would result from this specified increase in fruit and vegetable consumption. Ness and colleagues' own approach to such a review was to tabulate the studies with their methodological details and list the result of each study as showing 'no association' or 'protective effect' (Ness & Powles, 1997), when the evidence did not justify the implicit dichotomy. Associations were reproduced as published, in different formats for different studies, and confidence intervals commonly not reported. This left the reader with little impression of the average size of the association nor the degree of consistency between studies. Ness and colleagues express disapproval of our quantitative approach but provide no sound basis for rejecting it. We believe that our results are valid, and that the quantification of the effect is useful in establishing for the first time the moderate but important reduction in heart disease risk that results from a realistic increase in fruit and vegetable consumption. The main argument of Ness and colleagues is that estimates of effect derived from cohort studies are unreliable. We respond to this first, and then to four methodological issues that they raise.