Optimizing IOL Formula Performance in Atypical Eyes

Comparison of Intraocular Lens Formula Accuracy for Eyes With Prior Scleral Buckle Surgery

Kong AW, Jeon MJ, Lin SR, et al¹

Industry support for this study: None

Abstract Summary

This retrospective single-center study from an academic center in Los Angeles evaluated the predictive accuracy of multiple IOL formulas in eyes with a history of scleral buckle surgery. The study included 79 eyes of 77 patients who underwent cataract surgery after scleral buckle placement between 2014 and 2024.

The spherical equivalent prediction error (SEQ-PE) was assessed for the Barrett Universal II (BUII), Hoffer QST, Holladay 1, Kane, Radial Basis Function (RBF) 3.0, and SRK/T formulas. SEQ-PE was defined as the difference between the final postoperative SEQ refraction and the predicted SEQ refraction based on the IOL power implanted. Axial length was calculated using the Wang-Koch adjustment in eyes found to be longer than 25 mm with the Holladay 1 and SRK/T formulas.

Kong and colleagues evaluated the trueness of SEQ-PE (distance from zero), the precision of SEQ-PE (spread of data), and absolute SEQ-PE using the Eyetemis analysis tool (eyetemis.com). All six formulas demonstrated similar precision and comparable absolute SEQ-PE values. No statistically significant differences were found at any PE threshold. Across all formulas, at least 69% of eyes were within 0.50 D of the target refraction, and more than 92% were within 1.00 D.

Differences emerged in the trueness analysis. The BUII and Hoffer QST formulas demonstrated significantly more hyperopic outcomes, whereas the Holladay 1 and SRK/T formulas demonstrated significantly more myopic outcomes.

In eyes that had undergone scleral buckle surgery without a pars plana vitrectomy, the average SEQ-PE was significantly more hyperopic with the BUII (P = .005), Kane (P = .004), and RBF 3.0 (P = .006) formulas. In eyes that had undergone scleral buckle surgery combined with a vitrectomy, the average SEQ-PE was significantly more hyperopic with the BUII formula (P = .02) and more myopic with the Holladay 1 (P = .01) and SRK/T (P = .04) formulas.

Further analysis showed that each formula remained a strong predictor of the postoperative refractive outcome and that variables such as axial length did not significantly predict increased error. Additionally, no significant correlation was identified between intereye axial length differences and SEQ-PE for any formula.

Discussion

The study found that the BUII, Hoffer QST, Holladay 1, Kane, RBF 3.0, and SRK/T formulas were all capable of providing clinically acceptable refractive outcomes in eyes that had a history of scleral buckle surgery, particularly when the axial length was calculated using the Wang-Koch adjustment for eyes longer than 25 mm according to the Holladay 1 and SRK/T formulas.

Although no significant differences were identified in formula accuracy or precision, some trends were observed. The BUII and Hoffer QST formulas tended toward hyperopic outcomes, whereas the Holladay 1 and SRK/T formulas tended toward myopic outcomes. The study might have been underpowered to fully assess significant differences for the Kane and RBF 3.0 formulas.

Among eyes that underwent scleral buckle surgery alone, hyperopic shifts persisted with the BUII, Kane, and RBF 3.0 formulas, but larger studies are required to confirm these findings.

Differences between this study's results and prior reports evaluating outcomes after retinal surgery highlight the need for large multicenter studies. Because many patients in this cohort had also undergone a vitrectomy, the findings may be more applicable to eyes with a history of combined scleral buckle and vitrectomy surgery than those with prior scleral buckle procedures alone.

Intraocular Lens Power Calculation Formula Accuracy in 1178 Eyes With Short Axial Length: Systematic Review and Network Meta-Analysis

Zhang G, Ma Y, Liu Z, et al²

Industry support for this study: None

Abstract Summary

This systematic review and network meta-analysis evaluated the accuracy of 12 commonly used IOL power calculation formulas in eyes with short axial lengths. Zhang and colleagues reviewed studies published between 2003 and 2023 using PubMed, Embase, Web of Science, and the Cochrane Library databases.

Fifteen prospective and retrospective studies involving 1,178 eyes with an axial length of less than 22 mm were included. All eyes underwent cataract surgery with IOL implantation in the bag. Outcomes included the percentage of eyes within PE thresholds of 0.25 to 1.00 D, mean absolute error, and median absolute error. Formula performance was ranked using surface under the cumulative ranking curve (SUCRA) analysis. The risk of bias was assessed using the modified QUADAS-2 tool, and no publication bias was identified.

The meta-analysis demonstrated that the Kane formula achieved a higher percentage of eyes within 0.25 D of the target refraction compared with the Haigis formula, although the difference was not statistically significant. The Kane and Olsen standalone formulas achieved higher percentages of eyes within 0.50 and 1.00 D of the target refraction, but no statistically significant differences were identified among formulas at any PE threshold.

Based on SUCRA analysis, the Kane formula ranked the highest for prediction accuracy within 0.25 D, with a SUCRA value of 95.74%, followed by the Haigis formula at 94.79% and the Olsen formula at 84.04%. The Kane and Olsen standalone formulas demonstrated the lowest median absolute error values.

Discussion

Historically, the Hoffer Q formula has been preferred for short eyes, but recent comparative studies have shown that the BUII formula performs strongly in this subgroup.^3,4 The Kane formula has also emerged as a highly accurate formula, particularly for eyes with axial hyperopia.⁵

The network meta-analysis by Zhang et al² found that the Kane, Haigis, and Olsen formulas may provide more consistent refractive outcomes than the other nine evaluated formulas in eyes with short axial lengths. Despite these trends, no statistically significant differences were identified among formulas at PE thresholds from +/-0.25 to +/-1.00 D.

The Kane formula demonstrated the strongest overall performance based on SUCRA rankings, but these findings should be interpreted as a trend favoring this formula rather than definitive evidence of its superiority.

Zhang and colleagues noted several limitations of their meta-analysis,² including an inability to evaluate the effect of specific IOL models on formula performance. Some studies have reported on accuracy and outcomes with particular IOL types and specific formulas.^6,7 Further research stratified by IOL type might help address this limitation. As biometric technology and IOL calculation methods continue to evolve, ongoing evaluation of emerging formulas is essential to improve refractive accuracy, especially in challenging eyes.