Retrospective Comparison of 5 Modern IOL Calculation Formulas in Eyes With Various Axial Lengths

Monday, April 20, 2015: 8:21 AM
Room 5A (San Diego Convention Center)
Jay Ching Chieh Wang, MD
Xavier Campos-Möller, MD
Moness Masri, MD
Chris J. Rudnisky, MD MPH
Ike K. Ahmed, MD

To compare the accuracy and precision of the following modern IOL power formulas: SRK/T (S), Holladay 1 (H1), Holladay 2 (H2), Hoffer Q (HQ) and Haigis (H), in a single study population consisting of eyes of short (<22.50mm), average (22.50 – 25.00 mm) and long (>25.00mm) axial lengths (AL).

Under Institutional Review Board ethics approval, clinical data was retrospectively collected from charts of patients who underwent phacoemulsification and in-the-bag IOL implantation. One eye from each patient was used. Primary outcome measure was mean absolute error (MAE). Secondary outcome measures were: mean arithmetic error (MArE) and the percentage of eyes within ±0.25 diopter (D), ±0.50D and ±1.00D. MAE and MArE were compared with one-way ANOVA, while Pearson’s Chi Square tests were used to compare the percentage of eyes within dioptric ranges. Statistical significance was set at p<0.05.

We included an initial cohort of 245 patients. Mean ALs of short (n=61), average (n=161) and long eyes [JW1] (n=23) were 21.93±0.45mm, 23.43±0.59mm, and 26.32±1.15mm, respectively, and statistically different (p<0.0001). There was no statistically significant difference between MAE (short: p=0.19, average: p=0.99, long: p=0.42), MArE (short: p=0.15, average: p=0.99, long: p=0.47) and proportion of eyes within 0.25D for all five formulas. When analyzing all eyes, statistically significant difference was found for the proportion of eyes within 0.50D (S=62.86%, H1=57.14%, H2=51.43%, HQ=49.80% and H=57.96%, p=0.025) and 1.00D (S=88.57%, H1=87.35%, H2=70.61%, HQ=83.27%, and H=86.12%, p<0.0001).  Additional patient data is being collected to increase sample size and will be added to the analysis.

The MAE, MArE and proportion of eyes within 0.25D of refractive targets were not statistically different for any of the five formulas evaluated.  There is a difference between all formulas in the proportion of eyes within 0.50D in all eyes and within 1.0D in all, average and long eyes.