The Epidemiologic Necropsy

To the Editor.— The article by McFarlane et al 1 entitled "The `Epidemiological Necropsy'" presents the startling finding that there may be a substantial reservoir of undetected lung cancer cases. If this is verified, it has serious implications for the validity of case-finding by the Cancer Registry in Connecticut and by other registries that use the same methods. It therefore is important to be absolutely sure that no hidden factors are present that may have inadvertently introduced bias into this study. The authors point to a fear of demographic and clinical bias that has inhibited investigators from using necropsy data in the past. They were wise to do this: the claim is that demographic bias has been eliminated in their study by using a standard population with gender and age as adjustment variables and that clinical bias has been avoided by eliminating previously diagnosed cases of lung cancer. Despite the