The detection of thoracic abnormalities using posterior-anterior (PA) vs PA and oblique roentgenograms

A reading trial was conducted as part of the 1978 Bay Area Asbestos Screening Project to evaluate the utility of adding oblique-view roentgenograms to standard posterior-anterior (PA) views. Chest films from a sample of 555 workers with histories of long-term asbestos exposure were read twice as PA and twice as PA + oblique sets, providing a basis for assessing reliability through intrareader, interreader, and intermethod agreement. The ancillary use of oblique view films resulted in higher rates of detection of asbestos-related abnormalities than with PA films alone, but the increased rates were gained at the expense of reliability in the interpretation process. Depending on the source of this unreliability, which requires further study, different remedial actions might be indicated. A reader's lack of experience could indicate the need for additional training, the use of several readers, an average over several reader judgements, or the development of standards for evaluating obliques. If, on the other hand, no way is found to reduce the unreliability, the use of oblique films in routine screening programs ought to be discouraged on the basis of measurement theory, since reliability is a prerequisite to validity.