One study, unusually among premenopausal women, found that the cancer rate among women on a diet with the highest animal fats was a third higher than that among women on a diet with the lowest animal-fat content.

Despite their findings, researchers doubt the oft-repeated hypothesis that a high-fat diet causes breast cancer; if that were so, they say, it would equally apply to vegetable fats, and it doesn't. Instead, they suspect that cooked red meat and natural, fat-soluble hormones in cows' milk may play a part.

A second study came up with contradictory results, which points to a basic flaw in the way that data have been collated all these years. Participants completed a food-frequency questionnaire - a simple form that has been used in most diet studies over the years - and a food diary, a complete and exhaustive report. The diary method discovered a significant relation between a diet high in saturated fats and breast cancer - and almost none when the food-frequency forms were analysed.

This finding is almost as significant as the main outcome, and throws into question some assumptions about diet and the role it has in maintaining health. Perhaps it plays a more significant role than medicine has realised after all, and that food really should be our medicine.

(Sources: Journal of the National Cancer Institute, 2003; 95: 1079-85; The Lancet, 2003; 362: 212).