ACCURACY OF BIOMETRIC FORMULAE FOR INTRAOCULAR LENS POWER CALCULATION IN EYES WITH HIGH MYOPIA

To evaluate the accuracy of the modern biometric formulae for intraocular lens power calculation in eyes with high myopia, using the ESCRS online calculator.

Refractive Surgery Unit, Ophthalmology Department, Centro Hospitalar Universitário do Porto (CHUPorto)

Retrospective analysis of consecutive eyes with high myopia submitted to cataract surgery in the Refractive Surgery Unit of CHUPorto (April 2021 - September 2022). Eyes submitted to keratorefractive surgery, with anterior segment co-morbidities or in which appropriate refraction was not possible were excluded. Biometric evaluation was performed with the IOL Master® 700 (Zeiss). The refractive outcome was evaluated 1 month after surgery. High myopia was defined as an axial length (AL) ≥26.5mm. The ESCRS online calculator was used to evaluate the following formulae: Barrett, Cooke K6, EVO, Hill-RBF, Hoffer® QST, Kane and Pearl DGS. The median error (ME) and median absolute error (MAE) were calculated to assess the accuracy of each formula. The percentage of eyes with a prediction error within ±0.25 D and ±0.50 D was assessed. The Pearl DGS could not be used with an MA60MA lens. Those values were treated as missing. 

We included 70 eyes of 47 patients. Mean AL was 29.2±2.3 mm and 52.9% had an AL≥29.0 mm. The MAE was 0.25 [0.01 - 0.86] D with the Hill-RBF, 0.26 [0.00 - 1.01] D with the EVO, 0.28 [0.00 - 1.13] D with the Cooke K6, 0.28 [0.00 - 0.90] D with the Kane, 0.32 [0.00 - 0.93] D with the Barrett, 0.35 [0.01 - 1.46] D with the Hoffer® QST and 0.36 [0.01 - 1.02] D with the Pearl DGS. Differences were significant between the Pearl DGS and all other formulae (p<0.008), between the Hoffer® QST and the Hill-RBF (p=0.002), the Kane (p=0.010) and the EVO (p=0.025) and between the Barrett and the Hill-RBF (p=0.012). The ME ranged from 0.08 [-0.86 - 0.81] D with the Hill-RBF to 0.36 [-0.25 - 1.02] D with the Pearl DGS. A MAE≤0.5 D was obtained in 70 to 80% of eyes with all formulae except the Hoffer® QST (64.3%) and the Pearl DGS (59.5%) and a MAE≤0.25 D was obtained in 40 to 50% of eyes with all formulae except the Hoffer® QST (37.1%) and the Pearl DGS (35.1%). The MAE increased in eyes with an AL ≥29.0 mm with the Cooke K6 (p=0.022), the EVO (p=0.034), the Hill-RBF (p=0.016) and the Hoffer® QST (p=0.009). For an AL<29.0 mm, the MAE ranged between 0.21 [0.00 – 0.72] D with the EVO and 0.32 [0.01 – 0.93] D with the Pearl DGS. For an AL≥29 mm, the MAE ranged between 0.31 [0.01-0.78] D with the Kane and 0.64 [0.03 – 1.02] D with the Pearl DGS. In both groups the Pearl DGS had a significantly superior MAE than all other formulae (p<0.028), except compared to the Hoffer® QST in eyes with an AL≥29 mm (p=0.110). 

The Hoffer® QST and the Pearl DGS seem to be the formulae with the worst accuracy.  All formulae had a positive ME, showing a tendency to more positive refractive outcomes than predicted. The Cooke K6, the EVO and the Hill-RBF had greater MAE with AL≥29.0 mm.