To compare the accuracy of sixteen intraocular lens (IOL) power calculation formulas incorporating targeted adjustments or regression-based modifications of axial length (AL) in eyes longer than 26.00 mm. Retrospective observational study. The data of myopic patients with cataract who underwent uneventful phacoemulsification with in-the-bag implantation of a PARTIAL−RoF narrow IOL between January 2020 and June 2025 were reviewed. Preoperative IOL power was calculated using the IOLMaster 700 with six formulas: Barrett Universal II (BU II), Haigis, Hoffer Q, Holladay 1, Holladay 2, and SRK/T. The implanted IOL power was selected from BU II or SRK/T recommendations. Refraction was measured three months postoperatively. Postoperative IOL power predictions were generated using the following formulas or formula variants: K6, PEARL-DGS, Castrop, Eom, VRF CMAL; Holladay 1 with Wang-Koch 2 center optimization (WK2), modified Wang-Koch (MWK), non-linear regression (NLR), and Fam-adjusted method (F2); SRK/T WK2, SRK/T MWK, SRK/T F2; Holladay 2 NLR; Hoffer Q WK2; Haigis WK2; and Barrett True AL. The primary outcome measures were root mean square absolute error (RMSAE) and the percentage of eyes with prediction error (PE) within ± 0.25 D, ± 0.50 D, ± 0.75 D, and ± 1.00 D. One hundred sixty-four eyes with ALs ranging from 26.04 mm to 29.72 mm were included. RMSAE values across the sixteen formulas ranged from 0.393 (Holladay 1 NLR and SRK/T WK2) to 0.803 (Haigis WK2). The percentage of eyes with PE within ±0.50 D ranged from 35.98% (Haigis WK2) to 81.1% (Holladay 1 NLR). Holladay 1 NLR—followed by SRK/T WK2, Holladay 1 MWK, VRF CMAL, PEARL-DGS, and Eom—demonstrated significantly higher accuracy than most other formulas. Haigis WK2 and Hoffer Q WK2 were the least accurate. Certain modified third-generation formulas (Holladay 1 NLR, SRK/T WK2) achieve accuracy comparable to that of new-generation formulas (PEARL-DGS, K6, VRF CMAL) in IOL power calculations for long eyes. However, some third- and fourth-generation formulas—even after AL-based modification (Hoffer Q WK2, Haigis WK2)—continue to yield suboptimal results in this anatomical range.
Stopyra et al. (Sun,) studied this question.