Correlations between variability of preoperative ocular biometric measurements and accuracy of intraocular lens power calculation

To investigate the relationship between variability of preoperative ocular biometric measurements and the accuracy of intraocular lens (IOL) power calculation.

This retrospective study included cataract patients who underwent surgery at the People's Hospital of Baoan Shenzhen from January 2022 to December 2024. Preoperative ocular biometric parameters were measured using the Lenstar LS-900, and the coefficients of variation (CV) of corneal curvature (K1-CV, K2-CV), axial length (AL-CV), anterior chamber depth (ACD-CV) were calculated. Intented residual refractive error (IRRE) was recorded. Postoperative spherical equivalent (SE), prediction error (PE = SE - IRRE), and absolute prediction error (APE = |PE|) were evaluated at one month postoperatively. The mean, median and interquartile range of PE and APE were calculated. Patients were divided into two groups based on the number of valid axial length measurements (Group 1: <3; Group 2: ≥3). Differences in age, sex, preoperative visual acuity (LogMAR VA), K1-CV, K2-CV, AL-CV, ACD-CV, PE, and APE were compared between the two groups. Spearman correlation was used to assess the relationship between biometric variability (K1-CV, K2-CV, AL-CV, ACD-CV) and refractive outcomes (PE and APE).

A total of 118 eyes were included (Group 1: 44 eyes; Group 2: 74 eyes). Preoperative LogMAR VA was significantly worse in Group 1 than those in Group 2 (0.84 vs. 0.46, p < 0.001), but no significant differences were observed in age, sex, ACD, AL, mean keratometry (Km), PE, or APE between the two groups. Group 1 showed a mean PE of -0.13 and mean APE of 0.61, with 27.03 %, 39.19 %, and 50.0 % of eyes within ±0.25 D, ±0.50 D, and ±1.00 D of target refraction, respectively. In comparison, Group 2 demonstrated a lower mean PE (-1.80) and APE (0.48), and higher proportions of eyes within ±1.00 D (89.19 %). Variability in biometric parameters did not differ between the two groups (p ≥ 0.05). A weak correlation was found between preoperative VA and APE in Group 1 (r = 0.314, p = 0.038), but no significant correlations were detected between variability in biometric parameters with PE or APE.

The number of valid axial length measurements did not significantly affect preoperative ocular biometric variability or IOL power calculation accuracy in patients with cataract. Patients with fewer valid measurements (<3) had worse preoperative VA, which was weakly correlated with APE.