Comparison of Intraocular Lens Formulas in Patients With Postoperative Refractive Surprise.

OBJECTIVE

This study investigates the refractive accuracy of eight intraocular lens (IOL) power calculation formulas in patients with postoperative refractive surprise after phacoemulsification. It aims to determine if a different formula could result in better refractive outcomes in these eyes.

METHODS AND ANALYSIS

We retrospectively reviewed consecutive patients undergoing uncomplicated phacoemulsification as a sole procedure between March 2007 and September 2020 at the University of Washington by glaucoma subspecialists as part of a study investigating cataract surgery in normal eyes. The refractive surprise was defined as > 0.5 D difference between the zeroed-out predicted error (PE) using the Barrett Universal II (BUII) formula and postoperative refraction at four and 12 weeks. Mean absolute error (MAE) and zeroed out PE for Hoffer Q, Holladay 1 and 2, Sanders Retzlaff Kraff theoretical (SRK/T), radial basis function (RBF) 3.0, Kane, and Ladas super formula AI (LSFAI) formulas were calculated.

RESULTS

Of 440 eyes, 88 (20.0%) met inclusion criteria (MAE 0.77 ± 0.28). Of these, 35.2% and 64.8% had myopic (MAE 0.76 ± 0.23) and hyperopic (0.78 ± 0.31 D) surprise, respectively. The predicted SE significantly differed from postoperative refraction for all formulas (p≤ 0.035). The proportion of eyes with refractive surprise was not different comparing BUII with RBF, Kane, and LSFAI (range 19.5 to 20.2%; p ≥ 0.831), but was significantly higher for Holladay I, Holladay II, Hoffer Q, and SRK/T (29.5-31.8%; p < 0.001). For 13 eyes (3.0%) with refractive surprise ≥ 1.0 D (MAE 1.34 ± 0.25), no formula reduced the MAE below 1.22 D. Logistic regression revealed shorter axial length (AL) to be a risk factor for both myopic and hyperopic refractive surprise; total astigmatism and biometric ratios (keratometry(K)/AL and anterior chamber depth(ACD)/AL) were novel risk factors for hyperopic surprise.  Conclusion: Four more recent IOL calculation formulas were statistically equivalent in accuracy, indicating that these eyes are outliers across different formulas. Shorter eyes had a higher risk for refractive surprise. Novel biometric parameters warrant further investigation to improve refractive outcomes.