What constitutes 'clinically significant' improvement in implant power calculation accuracy?

摘要

In their article, Sheard et al. state that their T2 formula "improves clinical accuracy by approximately 10% compared with the standard SRK/T formula." We find this extraordinary given that the results reported by these authors with the 2 formulas are so similar; the difference in prediction error between the SRK/T and T2 formulas was 0.003 diopter (D), the mean absolute error was different by less than 0.02 D, the proportion of cases within +-1.00 D was different by only 0.5 percentage points (98.0 % versus 98.5 %), and the proportion of cases within +- 0.50 D was different by only 2.2 percentage points (78.7% versus 80.9%). We do not believe these differences are clinically significant.