Abstract
Diagnosing chest radiographs is a complex task that is difficult to master for medical students. Many textbooks in radiology thus recommend a systematic search strategy for detecting abnormalities in a radiograph. This systematic search strategy entails that anatomic regions are always inspected in the same order over radiographs. This is expected to lead to a complete coverage of the radiograph, and by that it should lead to a lower number of missed abnormalities. There is, however, very little evidence for the recommendation to use a systematic search strategy. We therefore investigated the assumed relationship between systematic search, coverage and number of misses, as well as the effect of training on these variables. Seventy-five 2nd year medical students underwent training in a systematic, a full-coverage (no focus on systematicity) or a non-systematic search strategy. Eye tracking was used to investigate the amount of systematic search (using Levenshtein Distance) and average percentage coverage of images. A more systematic search (lower Levenshtein Distance) was correlated with increased coverage, r = -.35, p < .01. Neither Levenshtein distance nor coverage were related with number of misses (respectively, r =.05, p =.73 and r = -.13, p =.31). Participants who underwent systematic search training were more systematic than the two other groups, K(2) = 16.58, p < .01. Participants in the systematic search group, and the full-coverage group covered more of the image than the non-systematic search group, K(2) = 7.42, p =.03. Participants in the full-coverage group missed most of the abnormalities, while the systematic-viewing group and non-systematic viewing group did not differ significantly F(2, 71) = 3.95, p =.02. The eye tracking data show that the training influenced viewing behaviour as predicted. However, performance did not increase with increased coverage. The data question the effectiveness of teaching students to systematically search a radiograph.
Meeting abstract presented at VSS 2016