A PRELIMINARY EVALUATION OF DIAGNOSTIC ODDS IN LUNG-SCAN REPORTING

Reporting of lung scans for pulmonary embolism (PE) using a descriptiv e probability notation is tried and tested. Subjectivity in interpreta tion of this jargon can be a problem for internists. Parallel descript ive and numerical probability reporting has been recommended, but the numerical probability scale is less precise than likelihood ratios exp ressed as odds. We therefore assessed internists' intuitive understand ing of lung scan reports in the odds format compared to the descriptiv e probability notation. A questionnaire was sent to Scotland's 217 int ernists to assess their intuitive understanding of odds reporting and to compare their management strategies when confronted by lung scan re ports in both an odds and a descriptive probability notation. There wa s a broad understanding of numerical odds. Internists used `normal' an d `100:1 against PE' identically; similarly, `low probability' and `10 : 1 against PE'. There was a statistically significant preference for the diagnosis of PE when internists were given the `1: 1 evens' report compared with the `indeterminate' report. There does appear to be a g reater awareness of the risk of PE when non-diagnostic lung scans are reported in numerical odds as compared with the descriptive probabilit y format.