Influence of test reliability on the screening performance of frequency-doubling perimetry.

Abstract

To investigate the influence of test reliability on the screening performance of frequency-doubling perimetry (FDT). Cross-sectional study. FDT sensitivity and specificity were calculated three times using three different strategies for handling unreliable (that is, >0 reliability indices outside normal limits) and unfeasible (that is, could not be completed) tests in 452 glaucoma patients from our outpatient department and 237 healthy subjects recruited outside the hospital. Unreliable (P = .01) and unfeasible (P < .001) tests occurred more frequently in glaucoma patients when compared with healthy subjects. Best screening performance was obtained by interpreting unreliable tests as if they were reliable and considering unfeasible tests as positive. With this strategy, sensitivity was 0.90 (95% confidence interval 0.87-0.93) and specificity 0.81 (95% confidence interval 0.76-0.86). Unreliable and unfeasible FDT tests appear to contain useful information.