Lens Thickness and Refractive Prediction Error after Cataract Surgery across Age Groups, Including Patients Aged 90 Years or Older.

PURPOSE: To investigate the impact of lens thickness (LT) on refractive prediction error (PE) and age-related ocular biometric changes after cataract surgery across age groups, including patients aged 90 years or older.

METHODS

This retrospective single-center study included one eligible eye from each of 372 patients who underwent phacoemulsification with a single intraocular lens model and were measured with the IOLMaster 700. Patients were classified into five age groups. PE and absolute prediction error (APE) were evaluated using the Haigis, SRK/T, Hoffer Q, and Barrett Universal II formulas. The association between LT and PE was assessed by dividing eyes at the cohort mean LT of 4.57 mm (an exploratory cutoff). Patients aged ≥90 years were compared with those aged <90 years, and a bilateral-eye analysis accounting for intereye correlation was performed.

RESULTS

Advancing age was associated with shorter axial length (AL) (p = 0.001), shallower anterior chamber depth (ACD), and greater LT (both p < 0.001). Age correlated negatively with AL (r = -0.202) and ACD (r = -0.311) and positively with LT (r = 0.430, all p < 0.001). Patients aged ≥90 years showed greater LT and shallower ACD than those aged <90 years (both p < 0.05), whereas APE did not differ between the two groups. Although APE did not differ among age groups, eyes with greater LT showed larger APEs than those with smaller LT across all formulas (all p < 0.05) and a relative hyperopic shift in adjusted analyses (all p ≤ 0.027). Barrett Universal II showed the lowest APE in the overall cohort.

CONCLUSIONS

Advancing age was associated with characteristic ocular biometric changes, particularly evident in patients aged ≥90 years. Greater LT, rather than chronological age, was associated with reduced refractive predictability after cataract surgery and may serve as a practical marker of refractive uncertainty. Barrett Universal II performed best overall.