Emmetropia Verifying Optical (EVO) formula in secondary scleral fixation using the Yamane technique: a retrospective comparison of intraocular lens power calculation formulas Authors: Yvonne Nowosielski, Roman Lischke, Teresa Rauchegger, Christoph Palme, Reinhard Angermann, Claus Zehetner, Martha Nowosielski, Eduard Schmid, Barbara Teuchner, Matus Rehak, Andreas Dimmer Published: 2026 ABSTRACT ObjectiveTo compare the accuracy of Emmetropia Verifying Optical (EVO) 2.0 with Sanders-Retzlaff-Kraff/Theoretical (SRK/T), Hoffer Q and Barrett Universal II (BUII) intraocular lens power calculation formulas in eyes that have undergone scleral fixation in Yamane technique.Methods and analysisIn this retrospective study, patients undergoing secondary intraocular lens fixation in Yamane technique between 03/2019 - 04/2024 were included. Prediction errors for EVO 2.0, BUII, SRK/T, and Hoffer Q formulas were calculated by subtracting the actual spherical equivalent from the predicted spherical equivalent. The performance of each method was evaluated by Eyetemis, an online tool for analyzing prediction errors. The tool includes robust t-tests to compare the trimmed means of the data.Results59 eyes met the inclusion criteria and were included in the analysis. All formulas showed myopic mean refractive prediction errors. With regard to spherical equivalent prediction error (SEQ-PE), precision of the SEQ-PE and absolute SEQ-PE, there were no significant differences between the formulas examined. The EVO 2.0 formula showed the lowest trimmed mean error with 0.62±0.58 D followed by BUII with 0.66±0.71 D. Within ± 0.50 D the EVO 2.0 formula achieved the highest accuracy with 54% whereas BUII and Hoffer Q achieved the highest accuracy within ±0.25 D with 28%.ConclusionThe EVO 2.0 formula represents a viable option for IOL calculation in patients undergoing scleral fixation with the Yamane technique. Although none of the formulas showed significant differences in performance, all consistently resulted in a myopic prediction error in the analyzed cohort. Therefore, when using the Yamane technique, targeting a slightly hyperopic refraction may help compensate for this myopic shift in similar patient populations. Introduction In pseudophakic patients with lost capsular support, secondary intrascleral fixation of the intraocular lens (IOL) has become a popular operating method. In 2017 Yamane et al. described a sutureless fixation technique where the haptics of a three-piece IOL are externalized and melted into a flange, thereby securing themselves intrasclerally [1]. Apart from mastering this complex surgical method, surgeons are also challenged by the preoperative assessment of the intraocular lens power. The prediction of the effective postoperative lens position is complicated by the lack of postoperative data on the actual lens position after Yamane technique. Additionally, patients who are in need of a secondary IOL implantation are often aphakic or suffer from an IOL subluxation. These conditions impede the preoperative optical measurement of the anterior chamber depth (ACD) which often results in a lack or only an approximation of the ACD. Previous reports indicate that inaccuracies in biometric measurements remain a major source of refractive surprise, with errors in anterior chamber depth, axial length, and keratometry contributing approximately 42%, 36%, and 22% of IOL power calculation error, respectively [2]. Due to the absence of reliable ACD measurements in patients undergoing Yamane fixation, surgeons must rely on IOL calculation formulas that do not require ACD measurement. Despite advances in contemporary formulas, refractive predictability remains imperfect, with only 73.7% of eyes achieving postoperative prediction error within ± 0.50 D of target refraction after cataract surgery [2]. In recent years new formulas such as emmetropia verifying optical (EVO 2.0) formula has been introduced which incorporates additional factors like anterior and posterior corneal curvatures and central corneal thickness in its algorithm to improve accuracy. This formula showed superior results in refractive prediction compared to many conventional IOL formulas in patients undergoing primary lens exchange [3, 4]. Furthermore it was noticed that the formula was also capable of operating with just two parameters being axial length and K, in patients with aphakia or eyes with subluxated lenses [5]. To date, evidence regarding the performance of the EVO 2.0 formula in the setting of Yamane technique secondary IOL fixation remains limited [6]. The primary endpoint of this retrospective study was to compare the spherical equivalent prediction error (SEQ-PE) for EVO 2.0, Barrett Universal II (BUII), Sanders-Retzlaff-Kraff/Theoretical (SRK/T) and Hoffer Q formula in patients who underwent secondary IOL fixation in Yamane technique. Secondary endpoints were to evaluate absolute SEQ-PE, precision of SEQ-PE, and proportion of eyes within ± 0.25 D and ± 0.50 D of target refraction. Material Patients This retrospective analysis has been approved by the local ethics committee of the medical university of Innsbruck (EC Nr: 1079/2022) and was conducted in accordance with the tenets of Declaration of Helsinki. The requirement for informed consent was waived by the ethics committee due to the retrospective nature of the study. We retrospectively evaluated patients who underwent secondary intraocular lens (IOL) fixation using the Yamane technique between March 2019 and April 2024 at the Department of Ophthalmology, Medical University of Innsbruck. All patients who underwent sutureless scleral fixation using the Yamane technique during this period were included. Exclusion criteria comprised eyes with corneal sutures, corneal decompensation, loss to follow-up, combined refractive surgery associated with significant postoperative IOL tilt, prior cerclage, iridocorneal malformations, alternative surgical approaches, or retinal detachment. Measurements Postoperative refraction was assessed one month after surgery and compared with the preoperatively predicted refraction calculated by 4 different formulas ((EVO 2.0, Barrett Universal II, SRK/T and Hoffer Q). Measurements were initially obtained using an autorefractor and subsequently refined, when necessary, by subjective refraction in accordance with the standard clinical protocol of the Department of Ophthalmology, Medical University of Innsbruck. Due to lens subluxation or aphakia, reliable ACD measurements could not be obtained in our patient population. Therefore, we used IOL calculation formulas that function without ACD as an input, avoiding potential inaccuracies associated with estimated or unavailable ACD values. Preoperative lens status (aphakia, IOL subluxation, complete IOL dislocation, or IOL at high risk of dislocation) was documented for all patients. The interval between prior cataract surgery and secondary intraocular lens implantation was not analyzed, as the study specifically focused on refractive outcomes after Yamane scleral fixation, regardless of the underlying indication for secondary IOL placement. Preoperative biometric measurements were generated by the IOL Master700 (Carl Zeiss Meditec, Jena, Germany). Data were retrieved from the electronic medical records of the patients. For SRK/T and Hoffer Q, the IOL Master 700 was used for calculation. Results for the EVO 2.0 and the Barrett Universal II formula were calculated using freely available online calculators (evoiolcalculator.com and calc.apacrs.org). EVO 2.0 was run in its reduced-input mode, considering only keratometry and axial length measurements for all cases. The implanted IOL power was used to determine predicted refraction in all formulas retrospectively. In accordance with current recommendations for IOL power calculation studies [7, 8], no IOL constant optimization was performed within the study cohort. Instead, lens constants optimized in external reference populations and published on the User Group for Laser Interference Biometry (ULIB) website were used for all analyses (KOWA Avansee Preset PUAS: 119.0; Alcon Acrysof MA60AC: 119.2; Johnson & Johnson Vision Sensar AR40e: 118.7). SEQ-PE was calculated by subtracting the postoperatively measured from the predicted refractive spherical equivalent. A negative SEQ-PE implies a more myopic result, and a positive SEQ-PE represents a more hyperopic shift. SEQ-PE (trueness), precision of SEQ-PE (repeatability), and absolute SEQ-PE (accuracy) were calculated. The absolute SEQ-PE thresholds within ± 0.25 diopters (D), ± 0.5 D, ± 0.75 D, and ± 1.0 D were calculated. To rule out a potential influence of axial length on refractive outcomes, the included eyes were divided into three groups (22–24.5 mm, > 24.5–26 mm, and > 26 mm). Eyes were further categorized based on sclerotomy location (2 mm versus 2.5 mm limbal insertion) to evaluate differences in SEQ-PE across all four formulas using one-way ANOVA with Bonferroni post-hoc testing. Since the Sensar AR40e lens showed a different haptic angulation (10° haptic angulation) compared to the KOWA Avansee Preset PUAS and Alcon Acrysof MA60AC (both 5° haptic angulation), we also separated our patients into these three groups to rule out a statistical influencing factor. Furthermore, we separated groups into the different surgeons to rule out a possible bias. Analysis of variance of SEQ-PE of the mentioned subgroups with Bonferroni post-hoc testing was performed. All this information was retrieved from the patient’s electronic surgical report. Surgical technique Six advanced and high frequent cataract and retinal surgeons (AD, BT, ES, CZ, MK and MR) performed the sutureless scleral fixation in Yamane technique as described by Yamane et al. [1] The technique was passed on to the others by one surgeon, so that every surgeon was using the same technique. The only variation in implantation technique among the surgeons was the limbal distance of the sclerotomies, which was either 2.0–2.5 mm, depending on each surgeon’s established surgical approach. Surgeries were performed in either retrobulbar or subtenonal anaesthesia with monitored anaesthesian care. In every case undergoing Yamane surgery a 23 or 25 gauge vitrectomy was performed using DORC EVA ophthalmic surgical system (Dutch Ophthalmic Research Center, Zuidland, The Netherlands) through a three port entry. After vitrectomy the subluxated or luxated IOL was removed via the anterior chamber either by cutting it into 2 or 3 pieces and extruded from the 3 mm sclerocorneal incision or in case of PMMA material extruded from a 6.0-mm sclerocorneal incision. A three piece IOL was injected into the anterior chamber through the main incision in a way that the haptic insertions were at the 12 and 6 o’clock position. Three types of IOLs were used: Avansee Preset PUAS (Kowa Medical Care Europe GmbH, Japan) a monofocal three-piece intraocular lens, consisting of a hydrophobic soft acrylic aspheric optic and blue haptics made of polyvinylidene fluoride (PVDF), a soft and flexible polymer. Acrysof MA60AC (Alcon Pharma GmbH, Germany) a monofocal three-piece intraocular lens with a hydrophobic acrylic optic and blue haptics consisting of Polymethylmethacrylate (PMMA) monofilament and Sensar Amo AR40e (Johnson & Johnson Surgical Vision, Inc., Santa Ana, Kalifornien, USA) a monofocal hydrophobic soft acrylic three piece IOL with blue PMMA haptics. Two 30 gauge needles (TSK ultra-thin wall needle; Tochigi Seiko, Tochigi, Japan) were then bent with the needle holder and sclerotomies were performed at 30 degrees at 2–2.5 mm distance from the limbus. The haptics were then feeded into the 30 gauge needles, externalized and the end was cauterized as described by Yamane [1]. Statistical analysis Statistical analysis of the refractive datasets was performed with the Eyetemis online analysis tool (https://www.eyetemis.com), using the “paired SEQ-PE” option [8]. To correct for heteroscedasticity, the Eyetemis software applies robust t-tests based on trimmed means, in which 10% of the lowest and highest values are removed before calculating the mean, thereby reducing the influence of outliers [9, 10]. Further data analysis was done using SPSS Statistics software (version 29; IBM Corp., Armonk, NY, USA) and Microsoft Excel 2016 (Microsoft Corp, Redmond, WA, USA). The Kolmogorov–Smirnov test was used to test variables for normal distribution. A p-value less than 0.05 was considered statistically significant. Due to the retrospective design of the study, no a-priori sample size calculation was performed. A post-hoc sensitivity power analysis was conducted to estimate the ability of the study to detect clinically meaningful differences between IOL calculation formulas. This analysis should be interpreted as exploratory in nature. Results In this study 59 patients corresponding to 59 eyes met the inclusion criteria and were used for further analysis. Exclusion and inclusion criteria as well as individual clinical data of the included patients are summarized in Tables 1 and 2. Table 1Number of eyes and reasons for exclusionParameterEyes (n)Eyes undergoing operation121Postoperative exclusion62Corneal suture20Corneal decompensation9Lost of follow up19Combined refractive procedure1Obvious IOL tilt2St.p. refractive procedure2Rhegmatogenous retinaldetachment (macula off)3Iridocorneal malformation1St.p. cerclage1Choroidal ammotio1different surgical approach1duplicates2Number of eyes and reasons for exclusion. St.p.: status post, IOL: intraocular lens Table 1Number of eyes and reasons for exclusionParameterEyes (n)Eyes undergoing operation121Postoperative exclusion62Corneal suture20Corneal decompensation9Lost of follow up19Combined refractive procedure1Obvious IOL tilt2St.p. refractive procedure2Rhegmatogenous retinaldetachment (macula off)3Iridocorneal malformation1St.p. cerclage1Choroidal ammotio1different surgical approach1duplicates2Number of eyes and reasons for exclusion. St.p.: status post, IOL: intraocular lens Number of eyes and reasons for exclusion Rhegmatogenous retinal detachment (macula off) Number of eyes and reasons for exclusion. St.p.: status post, IOL: intraocular lens Table 2Patients characteristicsParameterValueNo. of eyes59Age (years) Mean±SD76±13 Range23-98Gender (female/male)29/30Lens status prior to surgery Aphakia4 Total luxation15 Subluxation38 in loco-but at risk2Cause of IOL luxation PEX syndrome36 Complicated cataract surgery8 Trauma12 Unknown3CDVA prior to surgery (logMAR) Mean±SD0.75±0.59 Range0.1-2.7CDVA post surgery (logMAR) Mean±SD0.37±0.42 Range-2Eye (OD/OS)27/32Axial length (mm) Mean±SD24.12±1.49 Range22.12-30.04Flat keratometry (K1) (D)42.40±1.53Steep keratometry (K2) (D)43.69±1.54Mean keratometry (Km) (D)43.04±1.47ACD (mm) Mean±SD4.6±0.89 Range1.17-6.79Estimated refraction (D)-0.56±0.58 Mean±SDPostoperative SE (D)-0.68±1.17 Mean±SDIridectomy (yes/no)37/22Distance from limbus (2mm/2.5mm)46/13Sclerotomy sutured (yes/no)25/34IOL type used KOWA Avansee Preset PU6AS46 Alcon Acrysof MA60AC10 Johnson&Johnson Amo Sensar AR40e3CDVA: corrected distance visual acuity, OS: left eye, OD: right eye, D: diopter, SD: standard deviation, IOL: intraocular lens, ACD: anterior chamber depth Table 2Patients characteristicsParameterValueNo. of eyes59Age (years) Mean±SD76±13 Range23-98Gender (female/male)29/30Lens status prior to surgery Aphakia4 Total luxation15 Subluxation38 in loco-but at risk2Cause of IOL luxation PEX syndrome36 Complicated cataract surgery8 Trauma12 Unknown3CDVA prior to surgery (logMAR) Mean±SD0.75±0.59 Range0.1-2.7CDVA post surgery (logMAR) Mean±SD0.37±0.42 Range-2Eye (OD/OS)27/32Axial length (mm) Mean±SD24.12±1.49 Range22.12-30.04Flat keratometry (K1) (D)42.40±1.53Steep keratometry (K2) (D)43.69±1.54Mean keratometry (Km) (D)43.04±1.47ACD (mm) Mean±SD4.6±0.89 Range1.17-6.79Estimated refraction (D)-0.56±0.58 Mean±SDPostoperative SE (D)-0.68±1.17 Mean±SDIridectomy (yes/no)37/22Distance from limbus (2mm/2.5mm)46/13Sclerotomy sutured (yes/no)25/34IOL type used KOWA Avansee Preset PU6AS46 Alcon Acrysof MA60AC10 Johnson&Johnson Amo Sensar AR40e3CDVA: corrected distance visual acuity, OS: left eye, OD: right eye, D: diopter, SD: standard deviation, IOL: intraocular lens, ACD: anterior chamber depth Patients characteristics CDVA: corrected distance visual acuity, OS: left eye, OD: right eye, D: diopter, SD: standard deviation, IOL: intraocular lens, ACD: anterior chamber depth The SEQ-PE of the various formulas are presented in Table 3; Fig. 1. Evaluating the trueness the trimmed mean SEQ-PE of the formulas ranged from + 0.02 to + 0.07 D and none were significantly different from zero (p ≥ 0.56) or from each other (p ≥ 0.06) (Fig. 1 Aii). Table 3Detailed statisticsFormulatr- meanmeanSDRMSmin10%25%Median (50%)75%90%Max SEQ-PE (D) BUII0.0571430.070677970.9312260.926002-1.84-1.392-0.410.070.61.2322.4EVO2.00.0732650.083559320.8979720.894242-1.78-1.372-0.3950.050.541.1542.25Hoffer Q0.0602040.049491530.9638310.956909-2.11-1.358-0.4350.040.671.2862.02SRK/T0.0157140.021186440.9523860.944518-1.88-1.324-0.510.030.7551.132.35Precision of SEQ-PE (D)BUII0.6573760.711404360.5937560.92340.0128570.0891430.230.5228571.081.5651432.342857EVO2.00.6228530.680505020.5791290.8903890.0067350.0799590.2217350.4732651.0751.5092652.176735Hoffer Q0.6710160.729162920.623090.9556880.0002040.0742040.2452040.5897961.1897961.7177962.170204SRK/T0.683790.736319610.5962840.9442960.0042860.0342860.2250.7242861.1657141.5197142.334286 Absolute SEQ-PE (D) BUII0.6585710.713389830.5954530.92600200.0860.2150.571.081.5322.4EVO2.00.6222450.680847460.5847320.8942420.010.0760.20.461.0751.4362.25Hoffer Q0.6746940.731864410.6217730.9569090.010.0760.2350.541.251.772.11SRK/T0.6834690.737118640.5956370.9445180.010.0480.2250.721.151.5042.35SEQ-PE: spherical equivalent prediction error, tr-mean: trimmed mean, SD: standard deviation, min =minimal SEQ-PE; RMS = root mean square; 25% = first quartile (Q1); 50% = median, second quartile (Q2); 75% = third quartile (Q3); D = diopter; max = maximal SEQ-PE Fig. 1Comprehensive statistical analysis and heatmaps. A: SEQ-PE (spherical equivalent prediction error), B: precision of SEQ-PE, C: absolute SEQ-PE, D: percentage of eyes falling within a specific diopter range, A(i): heatmap for SEQ-PE Table 3Detailed statisticsFormulatr- meanmeanSDRMSmin10%25%Median (50%)75%90%Max SEQ-PE (D) BUII0.0571430.070677970.9312260.926002-1.84-1.392-0.410.070.61.2322.4EVO2.00.0732650.083559320.8979720.894242-1.78-1.372-0.3950.050.541.1542.25Hoffer Q0.0602040.049491530.9638310.956909-2.11-1.358-0.4350.040.671.2862.02SRK/T0.0157140.021186440.9523860.944518-1.88-1.324-0.510.030.7551.132.35Precision of SEQ-PE (D)BUII0.6573760.711404360.5937560.92340.0128570.0891430.230.5228571.081.5651432.342857EVO2.00.6228530.680505020.5791290.8903890.0067350.0799590.2217350.4732651.0751.5092652.176735Hoffer Q0.6710160.729162920.623090.9556880.0002040.0742040.2452040.5897961.1897961.7177962.170204SRK/T0.683790.736319610.5962840.9442960.0042860.0342860.2250.7242861.1657141.5197142.334286 Absolute SEQ-PE (D) BUII0.6585710.713389830.5954530.92600200.0860.2150.571.081.5322.4EVO2.00.6222450.680847460.5847320.8942420.010.0760.20.461.0751.4362.25Hoffer Q0.6746940.731864410.6217730.9569090.010.0760.2350.541.251.772.11SRK/T0.6834690.737118640.5956370.9445180.010.0480.2250.721.151.5042.35SEQ-PE: spherical equivalent prediction error, tr-mean: trimmed mean, SD: standard deviation, min =minimal SEQ-PE; RMS = root mean square; 25% = first quartile (Q1); 50% = median, second quartile (Q2); 75% = third quartile (Q3); D = diopter; max = maximal SEQ-PE Detailed statistics SEQ-PE: spherical equivalent prediction error, tr-mean: trimmed mean, SD: standard deviation, min =minimal SEQ-PE; RMS = root mean square; 25% = first quartile (Q1); 50% = median, second quartile (Q2); 75% = third quartile (Q3); D = diopter; max = maximal SEQ-PE Comprehensive statistical analysis and heatmaps. A: SEQ-PE (spherical equivalent prediction error), B: precision of SEQ-PE, C: absolute SEQ-PE, D: percentage of eyes falling within a specific diopter range, A(i): heatmap for SEQ-PE The precision of SEQ-PE is presented in Table 3; Fig. 1B. The trimmed means ranged between 0.62 and 0.68 D. There were no significant differences between EVO 2.0, BUII, Hoffer Q and SRK/T formulas. The absolute SEQ-PEs are presented in Table 3; Fig. 1C. The trimmed mean ranged between 0.62 and 0.68 D. There were no significant differences between the formulas. The absolute SEQ-PE thresholds within ± 0.25D, ± 0.5D, ± 0.75D and ± 1.0D are presented in Fig. 1D. No significant differences were noted within the groups. To assess the impact of axial length on refractive outcomes, eyes were categorized into three groups: 22–24.5 mm (n = 46), > 24.5–26 mm (n = 9), and > 26 mm (n = 4). Descriptively, mean prediction errors varied across the groups for all formulas: Barrett Universal II showed 0.14, 0.001, and − 0.52; EVO 2.0, 0.14, 0.009, and − 0.36; SRK/T, 0.05, − 0.067, and − 0.115; and Hoffer Q, 0.077, − 0.018, and − 0.098 for the 22–24.5 mm, > 24.5–26 mm, and > 26 mm groups, respectively. One-way ANOVA showed no statistically significant differences among the groups for any formula (all p > 0.05). Comparison between eyes where the sclerotomies were placed 2 mm (n = 46) or 2.5 mm (n = 13) away from the corneal limbus showed no statistical difference in mean SEQ-PE within the IOL formulas (EVO 2.0 -0.01 ± 0.85 vs. 0.34 ± 1.04, p = 0.25, BU II 0.01 ± 0.88 vs. 0.31 ± 1.09, p = 0.29, SRK/T -0.56 ± 0.88 vs. 0.26 ± 1.18, p = 0.31, Hoffer Q -0.04 ± 0.92 vs. 0.37 ± 1.1, p = 0.18). Furthermore neither the implanted IOL models nor the surgeons had an influencing factor on the reported results (data not shown). Discussion This retrospective study compared the accuracy of the EVO 2.0, Barrett Universal II (BUII), SRK/T, and Hoffer Q formulas in eyes undergoing secondary IOL implantation using the Yamane intrascleral fixation technique. The key finding is that no formula demonstrated statistically superior predictive performance. All formulas showed comparable overall accuracy, with a consistent myopic shift in mean refractive prediction error observed across all groups. These results have direct implications for preoperative planning in this challenging surgical setting. These findings are consistent with the existing literature. Khowaya et al. conducted a systematic review and meta-analysis examining IOL power calculation formulas in eyes undergoing Yamane scleral fixation and found no formula with superior predictive performance among SRK/T, BUII, Hoffer Q, Haigis, Holladay I and II, and Hill-RBF [11]. Similarly, Yamane et al. originally reported a mean refractive prediction error of − 0.21 ± 0.99 diopters using SRK/T in 97 eyes, a myopic tendency that is closely paralleled in our cohort [1]. The direction of refractive error in our study was uniformly myopic across all formulas, consistent with the systematic refractive bias reported in multiple prior studies of secondary scleral-fixated IOL implantation. Within the ± 0.5 D interval, EVO 2.0 achieved the highest precision (54%), demonstrating reliable performance across a broad range of axial lengths despite the inclusion of extreme values (22.12–30.04 mm). This is consistent with reports confirming that newer-generation formulas such as EVO 2.0 show no systematic bias against axial length, attributed to its emmetropization-based design, which uses corneal power as a reference for predicting pseudophakic lens position. However, EVO 2.0 did not outperform other formulas overall, which may be partly explained by the absence of lens thickness data in this cohort, given that EVO 2.0 derives axial length using Cooke’s modified axial length (CMAL), a value that incorporates lens thickness. Within the narrower ± 0.25 D interval, BUII and Hoffer Q each achieved the highest precision (29%). The performance of BUII aligns with a systematic review by Stopyra et al., which identified BUII as the most precise among vergence formulas [12]. It should be noted that not all implementations of the BUII online calculator permit calculations without entering anterior chamber depth (ACD). Since patients in this cohort were aphakic or had a subluxated lens, no reliable ACD could be obtained. A version of the calculator that omits the ACD input was therefore selected. Broader integration of such ACD-independent versions into multi-formula online calculators - such as the ESCRS calculator - would facilitate IOL planning for this patient population. The performance of Hoffer Q, a third-generation formula, was noteworthy. Despite its age, it matched BUII within ± 0.25 D and ranked second within ± 0.5 D (43%), suggesting underappreciated robustness in complicated or pre-operated eyes. Its integration into biometers such as the IOL Master further reduces the risk of transcription errors inherent in manual data entry into online calculators—a relevant practical advantage in complex clinical settings. The uniformly myopic prediction error observed across formulas is likely attributable to multiple factors. First, lens constants used in this study were derived from standard cataract surgery populations (ULIB) and were not optimized for Yamane fixation. The use of non-optimized constants is well established as a source of systematic refractive bias, particularly in small or heterogeneous cohorts. Second, the Yamane technique introduces inherent variability in effective lens position (ELP) that is not fully captured by current IOL formulas. Subtle differences in sclerotomy distance from the limbus, haptic fixation depth, flange size, cauterization technique, and intrascleral tunnel geometry can alter the postoperative ELP and thereby shift the refractive outcome. These technique-specific factors likely contribute to the myopic tendency independently of formula selection. From a clinical standpoint, the absence of a clearly superior formula and the consistent myopic shift across all evaluated formulas support a practical recommendation: when performing Yamane scleral fixation, surgeons should target a slightly hyperopic refraction to compensate for the expected myopic shift. Given the complex anatomical context of aphakic and pseudophakic eyes with lens subluxation, small numerical differences in mean prediction error between formulas are unlikely to be clinically decisive, as variability in ELP may outweigh formula-specific effects. A pragmatic approach - using multiple formulas in parallel and interpreting results in the context of each patient’s biometric profile - is therefore reasonable in this setting. Axial length subgroup analysis using one-way ANOVA revealed no statistically significant differences in mean prediction error between groups. However, the subgroup with axial lengths greater than 26 mm included only four eyes, substantially limiting the robustness of this finding. Similarly, comparison between sclerotomy distances of 2.0 mm and 2.5 mm from the limbus showed no significant difference, although the 2.5 mm group comprised only 13 eyes. No significant differences were found between IOL subtypes or between surgeons. All subgroup analyses should be regarded as exploratory given the limited sample sizes involved and should not be used to draw firm conclusions. Limitations Several limitations of this study warrant careful consideration. First, the retrospective design introduces the potential for selection bias and limits the ability to draw causal inferences. Second, the overall sample size of 59 eyes, and particularly the small numbers in several subgroups, reduces statistical power and limits the robustness of subgroup comparisons. Third, the inclusion of multiple IOL models with differing haptic characteristics introduces heterogeneity that may influence ELP and thereby affect refractive outcomes, even though no statistically significant differences between IOL subtypes were identified. Fourth, lens constants were not formally optimized for this cohort. The ULIB constants used were derived from routine cataract surgery populations and are not specifically validated for secondary scleral-fixated IOL implantation. This likely contributed to the observed systematic myopic shift and represents a significant methodological limitation; the refractive bias should therefore be interpreted in this context rather than as an inherent property of the formulas evaluated. However, as identical constants were applied across all cases, this bias is expected to be systematic rather than differential, thereby preserving the validity of relative comparisons between formulas. Technique-specific constant optimization in future, larger prospective studies is strongly warranted. Fifth, anterior segment OCT was not performed systematically, which limits the ability to detect subtle IOL tilt or decentration that may have influenced refractive outcomes. No clinical signs suggestive of severe malposition—such as iris chafing, pigment dispersion, or unexplained reduction in visual acuity - were observed; however, minor positional deviations cannot be fully excluded. Finally, the Yamane technique is inherently variable, with differences in needle entry angle, scleral tunnel preparation, haptic externalization, and cauterization potentially influencing ELP both between and within surgeons, even when analyses are stratified by operator. Conclusion In summary, EVO 2.0, BUII, Hoffer Q, and SRK/T demonstrated comparable accuracy in eyes undergoing secondary IOL implantation using the Yamane technique. No formula showed statistically superior performance. A consistent myopic shift was observed across all formulas, most likely reflecting a combination of non-optimized lens constants and the inherent variability of ELP in Yamane fixation. Clinically, targeting a slightly hyperopic refraction and using multiple formulas in parallel represents a pragmatic approach in this complex setting. These findings should be confirmed in larger prospective studies with technique-specific lens constant optimization.