Problems with the measurement of radiological change

From 1971-1978, chest X-rays were obtained from asbestos workers in British factories subject to the 1969 Asbestos Regulations. By the end of 1978, 36,592 X-rays from 26,471 workers, had been collected, at roughly two-yearly intervals and read using the 1971 ILO/UC classification of radiographs of pneumoconioses. Although consistent results had been obtained in a small pilot study, in which X-rays were read using the 'side-by'side' method by a small group of readers in close contact, serious inconsistencies were obtained in the main survey, in which X-rays were read shortly after they had been collected, rather than being accumulated and read together later, and different X-rays of the same worker were read independently, possibly by different readers. A general bias in classifying films to lower ILO/UC categories became evident as the survey progressed. It is concluded that, in order to obtain valid results, a set of films should be accumulated and then read together over as short a period as possible. It is intended that a representative sample of the films collected be re-read bearing these conclusions in mind.