Nomogram-based risk stratification for postoperative dry eye in diabetic cataract patients and its application in tiered nursing care.
Source: PubMed, NCBI / U.S. National Library of Medicine
To identify risk factors for postoperative dry eye in diabetic cataract patients, develop a nomogram prediction model, and establish a tiered nursing intervention protocol. A retrospective cohort study was conducted involving 678 diabetic patients who underwent phacoemulsification with intraocular lens implantation between January 2024 and January 2025. Patients were divided into a dry eye group (n=132) and a non-dry eye group (n=546). Least absolute shrinkage and selection operator (LASSO) regression and multivariate logistic regression were used to identify independent risk factors. A nomogram was constructed, and patients were stratified into low-, moderate-, and high-risk groups based on tertile cutoffs. Model performance was validated using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis. Postoperative dry eye occurred in 19.5% (132/678) of patients. Independent risk factors included: preoperative tear break-up time (odds ratio [OR]=0.715), Schirmer test value (OR=0.868), age ≥65 years (OR=1.975), diabetes duration ≥10 years (OR=8.511), glycated hemoglobin (HbA1c) ≥7% (OR=1.907), diabetic retinopathy (OR=0.090), glaucoma (OR=2.530), corneal comorbidities (OR=4.074), operative time ≥20 minutes (OR=2.327), and preservative-free artificial tear use (OR=0.407). The model demonstrated good discrimination (area under the curve [AUC]=0.877), calibration (Brier score = 0.1035), and clinical utility. The
Abstract
To identify risk factors for postoperative dry eye in diabetic cataract patients, develop a nomogram prediction model, and establish a tiered nursing intervention protocol. A retrospective cohort study was conducted involving 678 diabetic patients who underwent phacoemulsification with intraocular lens implantation between January 2024 and January 2025. Patients were divided into a dry eye group (n=132) and a non-dry eye group (n=546). Least absolute shrinkage and selection operator (LASSO) regression and multivariate logistic regression were used to identify independent risk factors. A nomogram was constructed, and patients were stratified into low-, moderate-, and high-risk groups based on tertile cutoffs. Model performance was validated using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis. Postoperative dry eye occurred in 19.5% (132/678) of patients. Independent risk factors included: preoperative tear break-up time (odds ratio [OR]=0.715), Schirmer test value (OR=0.868), age ≥65 years (OR=1.975), diabetes duration ≥10 years (OR=8.511), glycated hemoglobin (HbA1c) ≥7% (OR=1.907), diabetic retinopathy (OR=0.090), glaucoma (OR=2.530), corneal comorbidities (OR=4.074), operative time ≥20 minutes (OR=2.327), and preservative-free artificial tear use (OR=0.407). The model demonstrated good discrimination (area under the curve [AUC]=0.877), calibration (Brier score = 0.1035), and clinical utility. The nomogram effectively identifies high-risk patients. Risk stratification combined with tiered nursing enables rational resource allocation and targeted intervention for diabetic cataract patients.