Identification of risk factors associated with non-arteritic anterior ischemic optic neuropathy and development of a risk stratification model

Abstract Objective To identify risk factors associated with non-arteritic anterior ischemic optic neuropathy (NAION) and develop a risk stratification model. Methods We conducted a retrospective analysis of clinical data from patients with NAION at Tianjin Medical University Baodi Hospital from 2015 to 2025 and enrolled an age- and gender-matched healthy control group. Through the LASSO algorithm and multivariable logistic regression analysis, we identified the risk factors related to NAION and constructed a risk-stratification model. Risk stratification was performed based on the total score of the model. The discriminatory performance and calibration were assessed using receiver operating characteristic (ROC) curves, confusion matrices, and calibration curves. Meanwhile, the accuracy of risk classification was evaluated. Internal validation was assessed by Bootstrap validation. Results Fasting blood glucose (FBG), total cholesterol (TC), carotid artery disease, a dangerous optic cup, sleep apnea hypopnea syndrome (SAHS), and the residential area differences were identified as risk factors associated with NAION. The original area under the ROC curve (AUC) was 0.928. After bootstrap validation, the corrected AUC was 0.921. Based on the Youden index, the optimal cutoff point was 0.375. At this threshold, the model had good classification performance. The overall trend of the calibration curve closely aligned with the ideal reference line. According to the experience distribution of total scores on the nomogram, the 25th and 75th percentiles (53.5 and 109 points, respectively) were identified as objective cutoff points for our dataset. Subsequently, subjects were classified into low, medium, and high risk groups. In the internal validation, satisfactory classification accuracy was demonstrated across all risk levels. Conclusion We identified risk factors associated with NAION and developed a risk stratification model, which is not intended to predict absolute probability but rather to serve as a tool for relative risk stratification. This model features robust risk factor weighting and powerful individual risk ranking capabilities. Relying solely on basic clinical data, it can effectively assist primary care physicians in recognizing high-risk populations for NAION, particularly in settings with limited medical resources. However, this model is primarily applicable to populations similar to ours. Furthermore, more validation is required through multicenter, large-scale prospective cohort studies.