To assess a staged visual field review model (initially intensive, then downtitrated) for detecting cases of glaucoma progression (especially catastrophic and fast), in comparison to clinical standard, fixed review approaches. Computer simulation study. 100,000 subjects. 100,000 subjects were simulated to undergo four VF tests in one “eye” per visit, 3 monthly, over 30 years. The staged visual field model consisted of an initial intensive review (e.g. three tests per eye per visit, 3 monthly) that was then downtitrated as 90% of progressors within a rate bin (e.g. catastrophic <-2 dB/year) was detected. The test number and review cadence were systematically downtitrated with 90% of successive progression rate bins detected until the end of the 30 year follow up. Comparator models were fixed reviews: one test 12 monthly, one test 6 monthly, two tests 6 monthly (“frontloading”) and three tests 24 monthly (“wait-and-see”). To reduce false positives, an immediate subsequent confirmatory progression result was required for all conditions. Number of progressors detected and number of tests/visits required per case of progression detected. All staged review approaches detected 90% of catastrophic (21-36 months) and fast (36-60 months) progression bins sooner compared to all fixed methods (catastrophic: 66-108 months; fast: 84-132 months) except fixed frontloading. Detection of 50% of all progressors was also achieved sooner for the staged approaches (21-36 months) compared to all fixed methods (60-96 months), except for fixed frontloading (36 months). When using one test 12 monthly as a reference standard, the staged approaches were 66%-251% more likely to detect progression. The two tests per visit at 3 monthly and 6 monthly staged intervals were the most efficient with the lowest number of visits (10.6 and 8.6 at 5 years) and tests (15.8 and 12.9 at 5 years) per case detected over time. A staged visual field review model starting with intensive testing (such as two tests per visit at 3 or 6 monthly intervals) with downtitration at 30-36 months may increase the likelihood of detecting progressors and may provide long-term efficiencies in case detection, in comparison to conventional fixed approaches.
Phu et al. (Fri,) studied this question.