Samii VF, Kyles AE, Long CD, et al.
OBJECTIVE: To determine interoperator variance in shunt fraction calculation. DESIGN: Case series. SAMPLE POPULATION: 101 transrectal portoscintigraphic studies. PROCEDURE: Results of dynamic portoscintigraphic studies were reviewed by 4 radiologists without knowledge of signalment, history, or medical profile. Results were judged to be negative or positive on the basis of the dynamic scan. Composite images were formulated, and hand-drawn regions of interest were determined for the heart and liver. Time-activity curves were generated, time-zero points were selected, curves were integrated during a 10-second interval, and shunt fractions were calculated. RESULTS: Radiologists were in agreement regarding positive versus negative results for 99 of 101 studies. Interoperator variance in shunt fraction calculation ranged from 0.4 to 59.6%. For 51 studies with positive results, variance ranged from 2.5 to 59.6% (mean +/- SD, 22.8 +/- 14.5%); differences among reviewers were significant. For 48 studies with negative results, variance in shunt fraction ranged from 0.4 to 25.9% (mean, 5.3 +/- 5.8%); significant differences among reviewers were not detected. Shunt fraction calculations were not exactly reproducible among radiologists in 94 and 100% of studies with negative or positive results, respectively. CONCLUSIONS AND CLINICAL RELEVANCE: Results suggest that shunt fraction values are not reproducible among operators. Range in variability was greater in studies with positive results. This factor may be of particular clinical importance in reassessment of patients after incomplete shunt ligation.