Evaluating Time in Range for Visual Acuity and Macular Thickness as Outcome Measures for Randomized Clinical Trials.

PURPOSE: To compare performance of time in range (TIR) with other traditional longitudinal outcome measures (change from baseline and area under the curve [AUC] for change). METHODS: We used data from DRCR.net Randomized Clinical Trials (Protocol T, I and AC) for diabetic macular edema, with outcomes of visual acuity (VA) and optical coherence tomography (OCT) central subfield thickness (CST). Outcomes were calculated at one year, stratified by baseline VA (0-39, 40-59, 60-69, and ≥70 letters score). TIR-VA was defined as proportion of time with VA above threshold (64, 69, 74, 79, and 84), and TIR-CST as proportion of time with CST equal or less than threshold (250, 300, 350, and 400 µm). Performance was assessed based on effect size. RESULTS: In Protocol T, TIR-VA had largest effect size when baseline VA was poor: TIR-VA 69 (0.655), AUC-VA (0.607), and change VA (0.557). When baseline VA was moderate and good, TIR-VA effect sizes were similar to AUC-VA and change-VA. When baseline VA was excellent, effect size was largest for TIR-VA 74 versus AUC-VA or change-VA (0.199 vs. 0.179 vs. 0.083). For CST outcomes, TIR-CST was similar to or better than AUC-CST and change-CST across most baseline VA strata. In Protocols I and AC, TIR outcomes also had similar or larger effect size than AUC and change. CONCLUSIONS: TIR-VA and TIR-CST performed well across baseline VA strata and across RCTs. TRANSLATIONAL RELEVANCE: TIR-VA and TIR-CST may be complementary outcome measures for future RCTs.