Development And Validation Of A Regression-Based Nomogram For Correcting With-The-Rule Astigmatism In Small Incision Lenticule Extraction Surgery
Published 2025 - 43rd Congress of the ESCRS
Reference: FP14.07 | Type: Free paper | DOI: 10.82333/c40z-p724
Authors: Mateja Jagić* 1 , Ivan Gabrić 1 , Maja Bohač 1 , Sara Blazhevska 1 , Lucija Žerjav 1
1Refractive surgery department,University Eye Hospital Svjetlost,Zagreb,Croatia
Purpose
To evaluate a regression-derived nomogram for correcting moderate to high with-the-rule astigmatism in small incision lenticule extraction (SMILE).
Setting
Guangzhou Aier Eye Hospital, China
Methods
Data from 428 eyes with WTR astigmatism ≥1.00 diopters (D) after SMILE surgery (2020-2023) were grouped into the model group (376 eyes) and the control group (82 eyes) at a 4:1 ratio. The model group developed a personalized nomogram based on the linear regression between targeted induced astigmatism and surgically induced astigmatism, as well as between the original input SE and the achieved SE. These two formulas were derived as new input correction formulas for SE and astigmatism. The nomogram was applied to a validation group (86 eyes) with WTR astigmatism ≥1.00 D from January to June 2024. Safety, efficacy, predictability and vector analysis at 3 months postoperatively were compared between the control group and the validation group.
Results
At 3 months postoperatively, the mean postoperative SE was 0 ± 0.19 D in the validation group, better than the control group’s -0.09± 0.27 D (P = 0.015). Linear regression between TIA and SIA revealed 2.7% undercorrection (R² =0.8514) in the validation group, compared to 11.4% undercorrection in the control group (R² = 0.7154). In the validation group, 95.3% of the eyes (82/86) achieved postoperative astigmatism within 0.50 D, significantly higher than 85.4% (70/82) in the control group (P = 0.028). Correction index (CI: 1.00 ± 0.13) and magnitude of error (ME: 0 ± 0.15 D) were significantly better in the validation group compared to the control group (CI: 0.95 ± 0.17, ME: -0.07 ± 0.24 D) (P = 0.033, P = 0.022).
Conclusions
The nomogram based on linear regression between targeted induced astigmatism and surgically induced astigmatism, as well as between input SE and achieved SE, significantly improved the accuracy and predictability of astigmatism correction in SMILE, particularly for moderate to high WTR astigmatism.