Evidence Based Analysis Enhances Surgical Outcomes of Novice Resident Surgeons

TL;DR

Using evidence-based optimization improves the accuracy of cataract surgeries performed by novice surgeons.

Contribution

The study demonstrates that evidence-based IOL constant optimization significantly improves refractive outcomes in resident surgeries.

Findings

MAE improved from 0.44 D to 0.19 D with a0/a1/a2 optimization.

95% of eyes were within ±0.50 D of predicted refractive outcomes after full optimization.

Optimized results matched advanced prediction models like Barrett UII and Hill-RBF.

Abstract

Evidence based practice enhances healthcare delivery and prevents unsafe procedures. While competency based assessments of resident cataract surgery are standard, evidence based analysis of refractive outcomes remains underutilized in educational curricula. This retrospective single center study evaluated refractive outcomes from 21 novice ophthalmology resident surgeons. Three independent groups were compared based on formal constant optimization for intraocular lens (IOL) calculation: non-optimized Haigis (n = 216), a0-optimized (n = 94), and a0/a1/a2-optimized (n = 121). All surgeries were supervised by a single attending surgeon. Mean absolute error (MAE) and the percentage of eyes within ±0.25 D and ±0.50 D of predicted spherical equivalent (SEQ) were calculated. Also, systematic bias in effective lens position (ELP) was analyzed to update manufacturer IOL constants. MAE improved from 0.44 D (non-optimized) to 0.35 D (a0-optimized p = 0.009) and 0.19 D (a0/a1/a2-optimized p < 0.001). The percentage within ±0.50 D increased from 65.7% to 74.4% to 95.0%, respectively. With ELP bias correction, updated A constant and ACD were 119.266 and 5.755 mm. a0/a1/a2-optimized outcomes were comparable to ELP bias correction for the Barrett UII, Kane, and Hill-RBF formulas. Evidence based optimization of IOL constants significantly enhances novice resident surgical outcomes, achieving parity with prediction models. A formal curriculum on IOL calculation and optimization is warranted.