To the Editor.
—This letter is a response to information presented in a recent editorial1 and study2 regarding the predictability of radial keratotomy. I disagree with the statement that "nomograms and regression formulas that use the R2 value to demonstrate predictability of a regression formula do not permit prediction of an individual surgical outcome,"1 because the regression analysis approach used by two earlier studies3,4 and the Prospective Evaluation of Radial Keratotomy (PERK) Study2 demonstrated how to predict the outcome for an individual case. Although this editorial and study could lead one to conclude that the PERK "confidence interval" approach differs from the previously published "R2 " approach, the two approaches are quite similar and correct, in that a multiple regression model is fitted and R2 is computed. Arrowsmith and my first predictability publication3 included tables along with the R2 values, which
Marks RG. Predictability of Radial Keratotomy. Arch Ophthalmol. 1987;105(6):741. doi:10.1001/archopht.1987.01060060019004