The Hoffer Q formula: A comparison of theoretic and regression formulas

TLDR

The study introduces the Hoffer Q formula to predict pseudophakic anterior chamber depth for theoretic intraocular lens power calculations. The formula uses personalized ACD, axial length, and corneal curvature to compute IOL power. In 630 eyes, the Hoffer Q formula outperformed constant‑ACD and regression formulas, matched measured‑ACD accuracy, and showed superior precision—especially in eyes <22 mm—while reducing mean absolute error and error range compared to SRK and SRK II.

Abstract

A new formula, the Hoffer Q, was developed to predict the pseudophakic anterior chamber depth (ACD) for theoretic intraocular lens (IOL) power formulas. It relies on a personalized ACD, axial length, and corneal curvature. In 180 eyes, the Q formula proved more accurate than those using a constant ACD (P < .0001) and equal (P = .63) to those using the actual postoperative measured ACD (which is not possible clinically). In 450 eyes of one style IOL implanted by one surgeon, the Hoffer Q formula was equal to the Holladay (P =.65) and SRK/T (P =.63) and more accurate than the SRK (P <.0001) and SRK II (P =.004) regression formulas using optimized personalization constants. The Hoffer Q formula may be clinically more accurate than the Holladay and SRK/T formulas in eyes shorter than 22.0 mm. Even the original nonpersonalized constant ACD Hoffer formula compared with SRK I (using the most valid possible optimized personal A-constant) has a better mean absolute error (0.56 versus 0.59) and a significantly better range of IOL prediction error (3.44 diopters [D] versus 7.31 D). The range of error of the Hoffer Q formula (3.59 D) was half that of SRK 1(7.31 D). The highest IOL power errors in the 450 eyes were in the SRK II (3.14 D) and SRK I (6.14 D); the power error was 2.08 D using the Hoffer Q formula. The series using overall personalized ACD was more accurate than using an axial length subgroup personalized ACD in each axial length subgroup. The results strongly support replacing regression formulas with third-generation personalized theoretic formulas and carefully evaluating the Holladay, SRK/T, and Hoffer Q formulas.