Studies on the Value of Serial Films in Estimating the Progress of Pulmonary Disease

It has been generally accepted that serial chest roentgenograms made at intervals of two or three months permit valid conclusions as to the presence of change or activity in a majority of cases of chronic pulmonary disease. In the 1940 issue of “Diagnostic Standards” published by the National Tuberculosis Association, the following pertinent sentences appear: “In clinical cases requiring a period of observation for accurate judgment, serial roentgenograms of the chest are most important…. The value of serial roentgenograms to help determine the seriousness of the lesions and the kind of treatment or observation required can hardly be over-emphasized.” The recently published 1950 edition of that manual carries the statement: “Serial roentgenograms over a period of weeks to determine the degree of instability of lesions… have value in determining whether a lesion is active and in need of treatment.” The importance of testing the apparent usefulness of such serial films is therefore quite evident. The present paper is a report of a study of serial roentgenograms in estimating the progress of pulmonary disease. The investigation was designed to determine how consistent were the interpretations of two postero-anterior roentgenograms of the chest made three months apart in a series of patients with established pulmonary tuberculosis. The interpretations were made on the roentgen findings alone, without reference to clinical or laboratory data. A consecutive pair of films for each of 150 patients was chosen by independent observers for this work, thereby providing 300 standard 14 × 17-inch films for review. All the markings on the films were covered or removed, and a new numbering system was used to identify the sequence of the films in any given test. The six readers who undertook this study knew only which was the first and which was the second film of any pair. After an interval of several weeks, and without the record of the results of the first readings, the same films were re-interpreted by all six readers, thus providing a total of 12 independent interpretations on each set of the entire group. This permitted examination of the consistency of interpretation of each reader with himself (intra-individual consistency or variation), and of each reader with the other five readers (inter-individual consistency or variation). The interpreters were a group of three radiologists and three chest specialists, all of relatively extensive experience in chest roentgenology. The three radiologists had previously had considerable experience in group study on the “error” problem; one of the three chest specialists also had taken part in such research, while the remaining chest specialists were men of recognized ability in their field.2 The entire study was made over a period of approximately twelve months.