Comparison of Five Intraocular Lens Power Formulas to Predict Postoperative Refraction Accuracy

Abstract

The present study sought to investigate and compare the accuracy of two third-generation intraocular lens calculation formulas contrasted against three new-generation intraocular lens calculation formulas regarding their ability to predict postoperative refraction following cataract surgery. A retrospective case study following 172 patients (172 eyes) exhibiting age-related cataracts in their eyes who were subject to phacoemulsification between September 2017 and September 2018 at the Department of Cataracts, Tianjin Medical University Eye Hospital, was carried out. Based upon ocular axial length, the sampled patients were grouped into a short axis group (ocular axial length ≤ 22 mm; 17 cases; 17 eyes), a normal axis group (22 mm < ocular axial length ≥ 24.5 mm; 132 cases; 132 patients), and a long axis group (ocular axial length > 24.5 mm; 23 cases; 23 eyes); mean absolute prediction error (MAE) postoperative refraction in each group was determined using five formulas, and the percentage of eyes displaying postoperative myopic shift symptoms, postoperative hyperopic shift symptoms, alongside the percentage of eyes displaying postoperative refractive shift symptoms in the range of (−0.25 to 0.25 D, −0.50 to 0.50 D, −1.00 to 1.00 D), were all calculated following the procedures of the five selected formulas. The MAE of the 172 patient cases was compared within the five selected formulas, and SRK/T possessed the highest prediction accuracy, exhibiting a significant difference from the other four formulas (P < 0.05), with accuracy levels subsequently followed by the Holladay 1 and Barrett Universal II formulas-however, the two formulas lacked a significant difference between them (P > 0.05). In addition, the MAE of the normal axial group was compared and analyzed within the five formulas, with analysis revealing that the SRK/T, Holladay 1, and Barrett Universal II formulas exhibited strong prediction accuracy, with no significant difference present among these three formulas (P > 0.05), and also revealing a significantly difference between the aforementioned formulas and remaining two formulas (P < 0.05). For further analysis, the MAE of the short axis group was compared, and the SRK/T and Haigis (Holladay 1, and Barrett Universal II) demonstrated stronger prediction accuracy when compared to the Olsen formula (P < 0.05). Finally, the MAE of the long axis group was compared, and it was found that the SRK/T and Barrett Universal II formulas exhibits the best prediction accuracy, followed by the Haigis and Holladay 1 formulas, with no significant difference (P > 0.05) between the former two formulas or the latter two. The majority of patients exhibited hyperopic shift post-surgery. Of the five formulas studied, the SRK/T and Barrett Universal II formulas possessed strong accuracy capable of predicting postoperative refraction. However, more long-term observation, including large patient samples, is necessary in order to corroborate our result.