Abstract

oesophageal cancers was EMA positive. Even though the original sample appears large, the numbers with a presumptive diagnosis of coeliac disease are not large and the confidence intervals around the risk estimate for lymphoproliferative malig-nancy are wide. Overall though the risk estimate for any malignancy is not increased, with the upper 95 % confidence limit of being only 60 % above unity. So where does this leave us? What this study shows is that the risk of malignancy in undetected coeliac disease is not mark-edly increased over that of the rest of the population. This is not to say that there might not be some increase in risk of certain uncommon specific malignant conditions such as EATL, but the absolute level of risk is small.25 An increased risk of malignancy in undetected disease should no longer be used as argument in favour of screening or early detection of coeliac disease.