Abstract

ABSTRACT In a survey of the effects of exposure to asbestos in the UK Naval Dockyards, small- and large-film chest radiographs of 674 men have been examined. These films have been read under survey conditions by two readers using a simple screening classification, and also in a controlled trial by five readers using the full ILO U/C classification. Comparison between the reading methods showed a deficiency, independent of the size of film, of at least 30% in the detection of asbestos-related radiographic abnormalities when the screening classification was used. For adequate diagnostic sensitivity the ILO U/C classification appears to be essential. There was a deficiency of 43% in significant abnormalities observed by a majority of readers in the small films when directly compared with large film readings. This deficiency could be reduced to 7% by using readings of the small films at any level of abnormality by any of the five readers. When the ILO U/C readings were related to the clinical diagnoses, the only abnormality missed was a small pleural plaque. Films with previously agreed coding were inserted at intervals during the reading trial and helped to maintain the consistency of reading. Right oblique views were taken for 1884 men, in addition to the full-sized postero-anterior view, but the contribution provided by this view proved insufficient to justify its use in large surveys. The cost of a survey when small films are used as a screening method is reduced to between one-third and one-half of the cost when large films are used, assuming that the abnormality rate is not more than 5%. However, this cost advantage for small films is likely to be overtaken by the development of automatic large-film units. The radiation dose when small films are used is increased by a factor of about 20, but is within the prescribed safety level. It is concluded that at least three readers should be involved, using the full ILO U/C classification. Small films may be of particular use in a large-scale survey, in which the abnormality rate is expected to be low, and which might otherwise be too expensive. A sensitive reading method and a high standard of film quality are essential factors in the use of this technique.