Abstract
Purpose :
During recruitment for retina trials, screen failure commonly results from discrepancies between inclusion criteria, real-world Snellen VA, and ETDRS VA verified at the study visit. This simulation study evaluated how failure rate is affected by different inclusion criteria and VA tests used for screening: Snellen, ETDRS and quantitative VA (Zhao et al. 2021).
Methods :
To screen a population typical for retina (Fig.1 Kaiser 2009), we approximated a general patient (GP) distribution with piecewise uniform distribution of VA thresholds that varied from -0.2 to 1.8 logMAR, and VA range that scaled proportionally: VA range = 0.253 + 0.185 VA threshold (Zhao et al. 2021). We considered four inclusion criteria, bounded by 20/X and 20/320 (X= 25, 32, 40, or 50). To simulate recruitment, patients were randomly sampled from the GP, screened with a VA test (61-letter Snellen, 70-letter ETDRS, and 15-letter qVA), and verified with ETDRS. For each test and criterion, 1000 simulations that each comprised 1000 patient screenings, were used to calculate failure rate as the % of ETDRS scores verified outside of 20/X - 20/320.
Results :
Fig. 2 shows that, across criteria, screen failure rates were highest for Snellen (12-15%), and lower for ETDRS (7-9%), and qVA (4-6%). Across tests, most failures were observed when screening for VA (20/50 – 20/320) known to exhibit the highest variability and worst Snellen-ETDRS discrepancy.
Conclusions :
In simulations, the well-known deficiencies of Snellen VA lead to the highest screen failure rates (13%), which can be reduced but not eliminated using ETDRS (9%) and qVA (6%). The qVA’s intelligent algorithm with 15-letter testing (5 rows of 3 letters) exhibits potential for feasible real-world screening that can improve patient recruitment for retina.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.