DECEMBER 2023 | EYEWORLD | 23
on the five parameters (MAE, MedAE, RMSAE,
FPI, percentage of eyes within ±1 D of predic-
tive error), the Kane formula was found to the
most accurate formula for post-vitrectomy eyes
(MAE 0.49 D, MedAE 0.36 D, FPI 0.85, RMSAE
0.67 D). The Barrett Universal II formula was
found to be the most accurate in eyes with prior
scleral buckle (MAE 0.63 D, FPI 0.69, highest
percentage within ±1 D of PE 82.81%). In long
eyes (AL>26 mm), the Holladay 1 with the
nonlinear version of the Wang-Koch AL adjust-
ment (WKn) was most accurate (MAE 0.52 D,
RMSAE 0.69 D, FPI 0.75, highest percentage
within ±1 D of PE 88.14%). The Kane was also
found to be the most accurate in eyes with flat
corneas (K<42 D; MAE 0.44 D, FPI 0.84, high-
est percentage within ±1 D of PE 95.83%).
had a scleral buckle, and 33.9% of eyes had
silicone oil tamponade (with the remaining
eyes having balanced salt solution tamponade).
The majority of implanted IOLs were one-piece
(69.4%) hydrophobic acrylic (84.7%) IOLs. Of
the included eyes, 178 (18.3%) had a hyperopic
predictive error greater than 1 D. The most
significant risk factors for hyperopic refractive
predictive error were silicone oil tamponade
(OR 1.82) and higher axial length (OR 1.55 per
1 mm increase of axial length). In the study,
106 eyes (11.2%) were found to have a myopic
predictive error greater than 1 D, with the high-
est risk factors being sulcus IOL placement (OR
6.65), history of scleral buckle (OR 2.43), and
shallower anterior chamber depth (OR 1.79 per
1 mm decrease). When the performance of the
IOL calculation formulas was analyzed based
continued on page 24
