Prediction Error in Low Power Intraocular Lens in Highly Myopic Eyes

Abstract Purpose: To assess refractive prediction error (RPE) after cataract surgery in highly myopic eyes, focusing on the accuracy of intraocular lens (IOL) power calculations using the SRK/T formula without axial length adjustment. Setting: Tertiary ophthalmology department at Lariboisière Hospital, Paris, France. Retrospective observational study. Methods: We reviewed the records of eyes with axial length ≥ 26.5 mm that underwent cataract surgery between January 2019 and December 2021. RPE was defined as the difference between achieved and predicted spherical equivalent using SRK/T. We analyzed the absolute prediction error, proportion of eyes within ± 0.5D of target, and the direction of the error (hyperopic/myopic). Subgroup analyses were performed based on IOL power, especially plano and negative lenses, and the impact of implant-specific A-constant adjustments. Results were compared with predictions from the Barrett II and Pearl-DGS formulas. Results: A total of 102 eyes from 71 patients were included. Sixty-one percent achieved RPE within ± 0.5D. The mean absolute error was + 0.50 ± 0.49 D, with a 60% hyperopic shift. The plano/negative IOL subgroup (n = 26) showed a significantly higher RPE (+ 0.93 ± 0.68 D, p < 0.001), with a hyperopic shift in 85% of cases. This was strongly associated with IOLs lacking optimized A-constants for meniscus design at low/negative power. Barrett II (0.56 ± 0.36 D) and Pearl-DGS (0.57 ± 0.45 D) significantly reduced RPE and hyperopic shift in this subgroup (p < 0.05). Conclusions: SRK/T exhibited a frequent hyperopic bias in highly myopic eyes, driven primarily by the poor performance with plano and negative power meniscus IOLs. This highlights the clinical need for optimizing A-constants based on IOL geometry at low powers and mandates the routine use of modern formulas like Barrett II or Pearl-DGS in this specific population.