Background: Despite strictly defined criteria for visual field progres
sion in the ongoing Normal-tension Glaucoma Study, the authors noted a
surprisingly large number of patients reaching the endpoint. Traditio
nal methods could not be used to check the diagnostic accuracy of thei
r criteria, because no ''gold standard'' was established for distingui
shing true change from physiologic long-term fluctuation. Methods: The
authors developed a statistical method based on the results of duplic
ate tests for progression in their subjects. This method allowed the a
uthors to assess the sensitivity, specificity, and predictive values o
f their diagnostic criterion. It also estimated the true incidence of
progression and provided standard errors for the estimates. Results: T
he authors found that their original strict criteria for progression,
based on duplicate testing, produced false calls of progression 57% of
the time. By raising the requirement for deterioration and by repeati
ng the entire sequence of duplicate testing once more, the authors hav
e successfully reduced the rate of false calls to 2%. Conclusion: Accu
racy in recognizing progression is improved by not accepting small cha
nges as evidence of progression and by confirming the findings on repe
at testing.