Detection of lung cancer on the chest radiograph: impact of previous films, clinical information, double reading, and dual reading.

To study the impact of clinical information, previous chest radiographs, double reading (one pair of observers read the films independently) and dual reading (one pair of observers read the films simultaneously) on the sensitivity, specificity and odds ratios of observers in the detection of early lung cancer on the chest radiograph. The study was performed in 3 sessions. In the first session, 14 observers reviewed the chest radiographs of 100 cases (30 with early lung cancer, 35 with no abnormalities, 35 with other cardiopulmonary diseases). Sensitivity, specificity, odds ratios and the effect of double reading were calculated. After 4 months, a second session was held in which 7 observers reviewed all cases with the provision of all information. The other 7 observers reviewed all cases without information again. To determine the effect of dual reading, in the third session, 4 pairs of observers reviewed all cases in which they disagreed in the first independent reading session. The effects of information, double reading and dual reading on sensitivity, specificity and odds ratio were calculated. The sensitivity of the observers in the first session ranged from 0.20 to 0.60 and the specificity from 0.87 to 0.95. In the second session, these parameters changed only slightly, independent from the availability of clinical information and previous films. With double reading the sensitivity increased and specificity decreased. With dual reading sensitivity increased and specificity remained unchanged. The odds ratios were improved with double reading by 4% and with the dual reading by 14%. Additional clinical information, previous chest radiographs, double reading and dual reading have little impact on the detection of early lung cancer on the chest radiograph.