Risk prediction model for cataract after vitrectomy surgery: a 2-year study on primary rhegmatogenous retinal detachment.

AIM: To establish a risk prediction model for secondary cataract within 2y after pars plana vitrectomy (PPV) in patients with primary rhegmatogenous retinal detachment (RRD). METHODS: Clinical data of patients with primary RRD treated at the Shenzhen Eye Hospital were retrospectively collected. Twenty-four potential influencing factors, including patient characteristics and surgical factors, were selected for analysis. Independent risk factors for secondary cataract were identified through univariate comparisons and multivariate logistic regression analysis. A risk prediction model was constructed and evaluated using receiver operating characteristic (ROC) curves, area under the ROC curve (AUC), calibration plots, and decision curve analysis (DCA) curves. RESULTS: The 386 cases (389 eyes) of patients who underwent PPV and had complete surgical records were ultimately included. Within a 2-year longitudinal observation, 41.39% of patients developed cataract secondary to PPV. Logistic regression results identified a history of hypertension [odds ratio (OR)=1.78, 95%CI: 1.002-3.163, P=0.049], silicone oil tamponade (OR=3.667, 95%CI: 2.373-5.667, P=0.000), and lens thickness (OR=1.978, 95%CI: 1.129-3.464, P=0.017) as independent risk factors for cataract secondary to PPV. The constructed nomogram achieved AUC=0.6974. Calibration plots indicated good agreement between predicted and observed outcomes, while DCA curves demonstrated the model's clinical utility. CONCLUSION: By incorporating a history of hypertension, vitreous substitute type, and lens thickness, this study constructs a prediction model with moderate discriminative ability. This model offers a valuable tool for clinicians to identify high-risk patients early, potentially allowing for more timely interventions and improved patient outcomes.