Interobserver agreement using computed radiography in the adult intensive care unit

We determined whether computed radiographs in an adult intensive care unit produce greater agreement between readers in descriptions of radiologic findings and interval changes.

Fifty-six conventional radiographs and 56 computed radiographs were obtained on 28 patients. Computed and conventional radiographs were imaged simultaneously in a single exposure with the use of a sandwich cassette. The two sets of films per patient were taken less than 1 week apart.

Four readers evaluated the chest films and judged interval changes with the use of standardized confidence scales. Overall exposure and film quality also were evaluated. Computed radiographs were rated better in overall quality. Tubes and lines were detected with comparable accuracy. Interobserver agreement in the evaluation of radiologic descriptors was generally greater with computed radiographs (intraclass correlation). No differences were noted in interobserver agreement in the judgment of interval changes.

The diagnostic usefulness of computed radiography is comparable to conventional films in the intensive care unit. The higher interobserver agreement rate for radiologic descriptors with computed radiography is an added advantage of the system.