Regression to the Mean May Explain Visual Acuity Measurement Variations

Visual acuity measurements demonstrating high variability from mean values are often closer to those mean values during previous or subsequent measurements, according to a brief communication published in Eye. This regression to the mean may reveal greater visual benefits in eyes with worse baseline vision or support a need for early intervention in eyes with better baseline vision in clinical trials, the report suggests. 

Researchers retrospectively assessed visual acuity measurements in the untreated eyes of 1375 participants who underwent anti-vascular endothelial growth factor (VEGF) treatments. Visual acuities from the first 2 consecutive time points were included provided they occurred within a 90-day period. 

The team compared visual acuity values for the top and bottom 20% of participants at visit 1 with their values obtained during visit 2. They also compared visual acuity values for the top and bottom 20% at visit 2 with their values from visit 1. Overall, mean [SD] visual acuities were 74.3 [15.9] and 74.7 [16.1] Early Treatment Diabetic Retinopathy Study (ETDRS) letters at the first and second visits, respectively. 

Participants who scored in the top 20% (n=275) for visual acuity at visit 1 had worse mean visual acuity at visit 2 (mean difference, 1.0 letters; P <.0001). Individuals who scored in the bottom 20% at visit 1 had better visual acuity at visit 2 (mean difference, 2.3 letters; P <.0001). 

The report shows a similar pattern when comparing participants with the top and bottom 20% of scores from visit 2 with their initial measurements. For the top 20% at visit 2, mean visual acuity was worse at visit 1 (mean difference, 1.5 letters; P <.0001). For the bottom quintile at visit 2, the mean visual acuity was better at visit 1 (mean difference, 1.6 letters; P =.0039). 

“The phenomenon of regression to the mean was thus observed,” according to the study authors. “In eyes undergoing no intervention, those in the top quintile appeared to lose vision, while those in the lowest quintile appeared to gain vision. This was regardless of whether one looked forward or backward in time, and the difference achieved strong apparent statistical significance.”

Study limitations include a retrospective nature.