Experimental infection of man with pig whipworm.

metry please, hydramnios; and what constituted a positive radiological correlation in such cases? The method used for scoring agreements in this study (stated simply as "whether the radiological interpretation agreed with the clinical diagnosis as stated on the request card") is not clearly defined (except when nothing was detected on a routine-no pathology suspected-request). Perhaps the author had accurate and specific clinical diagnoses in all cases, but this is not stated or implied. The reverse is actually implied when he refers to such diagnoses as vertigo or headache. I was, in fact, impressed with the overall detection rate of 613% additional (unsuspected) significant chest lesions discovered (differently stated by the author, it is even more impressive -"more than one-fifth of all lesions detected in the chest. .. are not apparently suspected. . .") and almost 20% in the normal (routine) chests done. This is a high discovery rate in most present day surveys. Agreement with a specific clinical diagnosis (pulmonary tuberculosis, intestinal obstruction, fetal mal-formation, etc.) is quite different from radiological detection on examinations done for symptoms or signs (cough, abdominal pain, hydramnios, etc.). In the former, clear agreement or disagreement can be recorded and the contribution towards disagreement by each party can be known when the true lesion is discovered. The use of radiology for detection purposes when there is no firm clinical or pathological diagnosis must be appraised differently. Perhaps this difference helps to explain the wide agreement differences of 1.4% for skulls (rarely is a true clinical diagnosis given) and 54% for barium meals. (Frequently, a true diagnosis is given.) Other factors are undoubtedly important also as the author states. The author admits "this method has obvious limitations as there is no final pathological diagnosis". He then goes on to assume the initial clinical diagnosis (although headache or vertigo hardly seem to be diagnoses) is the correct one, and considers the radiological diagnosis (such as no abnormality seen) wrong, if it does not agree, except when the clinician expected a normal result and the radiologist discovered a significant lesion. This situation is simply counted as an agreement for the radiologist as it is added to his other agreements for his total score; it is not counted as a disagreement for the clinician who suspected no pathology. True, this method has the merit of simplicity he notes, but it has no scientific merit. To then suggest that a …