Magnitude of Visual Acuity Change with ETDRS versus Snellen Testing in Clinical Trials

Purpose: To compare visual acuity (VA) changes using standardized ETDRS best-corrected visual acuity (BCVA) and nonstandardized Snellen VA among subjects enrolled in clinical trials.Design: Retrospective study.Participants: Patients enrolled in prospective clinical trials at 3 urban retina practices.Methods: Best available Snellen VA at the clinic visit before study entry and after exit were compared with the ETDRS BCVA at trial entry and exit. The correlation and discrepancies between Snellen VA and ETDRS methods as well as the VA changes from trial entry to exit were evaluated.Main Outcome Measures: The discrepancy between VA change from trial entry to exit using Snellen VA versus ETDRS BCVA methods.Results: A total of 273 eyes were included. The mean (standard deviation [SD]) Snellen VA was 58.1 (20) ETDRS-equivalent letters (Snellen 20/69) at the clinic visit before trial entry and 61.6 (21) ETDRS-equivalent letters (Snellen 20/59) at the visit after trial exit. The mean (SD) ETDRS BCVA was 65.5 (16) letters (Snellen 20/49) at trial entry and 70.5 (17) letters (Snellen 20/39) at trial exit. The mean VA change from trial entry to exit was not significantly different for ETDRS (5 letters of vision gain) compared with Snellen (3.6 letters of vision gain) methods (P = 0.061). Eyes with baseline Snellen VA 20/50 or worse gained significantly more letters using Snellen (9.3 +/- 22.3 letters) compared with ETDRS (5.2 +/- 18.7 letters; P = 0.012). Among eyes with baseline Snellen VA of > 20/50, VA gain was significantly greater with the ETDRS method (4.9 +/- 12.3 letters) compared with Snellen (-1.5 +/- 12.3 letters; P < 0.001).Conclusions: The mean VA change from clinical trial entry to exit was similar between the ETDRS and Snellen methods. However, among patients with worse baseline Snellen vision, the magnitude of VA change was greater with Snellen compared with ETDRS, whereas among those with better baseline vision, this magnitude was greater with the ETDRS method. Understanding the proportion of the study population with varying VA levels may have implications for interpreting VA outcomes from retrospective clinic-based studies compared with those reported in clinical trials.