Your reporter Pat Thomas confuses two studies which we shall refer to as the Health Food Shoppers study (HFS) and the Oxford Vegetarian study (OV). Mortality data from HFS were reported in 1996 (BMJ, 1996; 313: 775-9) and from OV in 1994 (BMJ, 1994; 308: 1667-70). The 1996 analysis was not a "reanalysis of the same group", as these are two separate cohorts.

Pat Thomas also confuses results for bowel cancer mortality and incidence with results for IHD mortality in this analysis. Paul Appleby and Dr Tim Key, Imperial Cancer Research Fund, Oxford.....

Pat Thomas replies: Vast amounts of research went into the preparation of the vegetarian article. In putting it all together, two pieces of research were misquoted. The 1996 study by Margaret Thorogood and associates was not a reanalysis of the same cohort used in their 1994 study. But it was a reanalysis of a cohort first analysed by Michael L Burr (who also was involved in Thorogood's 1996 study): Am J Clin Nutri, 1982; 36: 873-7 and Am J Clin Nutri, 1988; 48: 830-2). Thorogood's reanalysis of Burr's original cohort did, as stated, show a narrowing of the difference in mortality between meat eaters and non meat eaters (from 30 per cent to 15 per cent) over time.

Under the heading of cancer I referred to a pooled analysis undertaken by Dr. Timothy Key and colleagues in 1998. The analysis showed no relationship between meat consumption and bowel cancer, as stated.

However the figures that followed, which showed that vegetarians were 35 per cent less likely to die from the disease, related to isehaemic heart disease and should have appeared under that heading.

While irritating, these errors do not detract from the broad conclusions of the piece namely that vegetarianism is unlikely to be a universally "best" diet; that it is the addition of vegetables and fruits, and not the omission of meats which is necessary to maintain good health and lower the risk of certain diseases; and that more research is needed before we can begin to understand fully the complex links which exist between diet, lifestyle and physical well being.

Editor's comment: A 670 page report, Food Nutrition and the Prevention of Cancer, the result of three years of work by a panel of more than 2,250

scientists evaluating 4500 research studies, implicated meat itself only in colon and rectal cancer (which the Oxford study did not). High levels of saturated animal fats which also appear in dairy produce, a staple of lacto vegetarian diets are implicated in lung, colorectal, breast, endometrial and prostate cancer. Nevertheless, other studies show that a high intake of vegetables, fruits and dietary fibre are associated with a reduced risk, independent of saturated fat intake (Am J Epidemiol, 1990; 132: 783).

The biggest problem with the current research is avoiding confounding factors. Jan Vandenbrouke of the Leiden University Hospital, The Netherlands, questioned the original 1994 Thorogood study because two thirds of the vegetarians were women, while half the meat eaters were slightly older and male (BMJ, 1994; 308: 1671). This alone could have accounted for the difference in heart disease mortality.